Ultrafine particles at workplaces of a primary aluminium smelter

Abstract

The number concentration and size distribution of ultrafine particles in a Søderberg and a prebake potroom of an aluminium primary smelter have been measured using a scanning mobility particle spectrometer. The particle morphology was studied by transmission electron microscopy (TEM). The study shows the existence of elevated number concentrations of ultrafine particles in both potrooms. The main source of these particles is likely to be the process of anode changing. The ultrafine particles were measured directly at the source but could also be identified as episodes of high number concentrations in the general background air. Unlike the larger particles belonging to the 50–100 nm mode, the nanoparticle mode could not be detected in the TEM indicating that they may not be stable under the applied sampling conditions and/or the high vacuum in the instrument.

---

Yngvar Thomassen

Yngvar Thomassen was born in Norway, in year 1947. Thomassen’s academic training was obtained in the Department of Analytical Chemistry at the University of Oslo from where he graduated in 1973. He spent one year at the Norwegian Defence Institute before taking a postdoctoral research associate position at the Department of Analytical Chemistry at the University of Oslo, where he conducted research for 2½ years. In 1978 he visited The University of Toronto, Department of Environmental Studies and Geology, for one year with a Royal Norwegian grant (visiting scientist). He is currently Research Director in Department of Occupational Hygiene, National Institute of Occupational Health in Oslo where he has spent 29 years of his professional life. He is also appointed as a professor in environmental chemistry at the Norwegian University of Life Sciences outside Oslo. His current research interests are: analytical chemistry, atomic spectroscopy, speciation, occupational and environmental chemistry.

Introduction

Potroom asthma is still frequently occurring among workers in primary aluminium smelters, but the pathophysiological mechanisms and the causative agent(s) involved are presently unknown.^1–5 Assessment of possible adverse health effects associated with inhalation of potroom fumes requires a detailed chemical and physical characterization of the work room atmosphere. The air quality in the potrooms of primary aluminium smelters has traditionally been monitored by measuring workers’ exposures to “total” dust and gases including particulate fluoride, hydrogen fluoride, sulfur dioxide and coal tar pitch volatiles (e.g. PAH’s).¹ We have recently initiated a large industry wide study to sample and measure the mass concentration and the chemical composition of the “conventional” health related size fractions of the airborne dust, i.e. the respirable fraction (particles penetrating into the alveolar region of the lung), the thoracic fraction (particle passing the head airways and entering the thorax) and the inhalable fraction (all particles entering the respiratory tract of humans via the mouth or the nose). As part of this initiative a number of single respirable and thoracic particles has been collected in potrooms and analysed by scanning and transmission electron microscopy. The most important findings in this study were (a) the frequently observable coating on aluminium oxide and cryolite particles exhibiting a nanostructure of ultrafine particles with an enormous specific surface, and (b) that the agglomerates identified often consisted of small fluoride-containing primary particles with diameters in the range of 0.1 to 1 μm.⁶

In the recent past, special attention has been addressed to so called ultrafine particles covering the size range below 100 nm.⁷ Special hazard has been attributed to the numbers of these particles with their large specific surface area that deposits per unit surface of the epithelial lining of the human respiratory system.^8,9 It is believed that the number concentrations, even at ambient levels, may be a significantly more important determinant of the health risk than inhaled mass measures.¹⁰ In the course of the discussion, dose metrics other than the mass have been proposed for health hazard analysis such as the number and the surface area concentration.¹¹

Ultrafine particles are mainly formed from gaseous precursors by gas to particle conversion. High-temperature welding operations, e.g., generates substantial number concentrations of ultrafine particles which are formed primarily through the nucleation of metal vapour followed by competing growth mechanisms as coagulation and condensation.¹² Due to the high temperatures involved in the Hall–Héroult aluminium reduction process, ultrafine particles are therefore likely to be formed and emitted when the molten mass is exposed to the colder environment. This is the case for example when the reduction cells are opened for anode changing. It is suggested that in addition to entrained fluorides by anode gas, direct evaporation of fluorides into the anode gas stream with their simultaneous heterogeneous condensation on drops of electrolyte generated in the process of bubble burst is the main source of fluoride losses from the electrolyte.¹³ Vapour condensation under high supersaturation conditions may explain the formation of ultrafine particles from the electrolyte.¹³

Workers helping with the anode change directly at the cell may thus be exposed to freshly generated fume containing ultrafines. Personnel not involved in the anode change (e.g. crane operators) or working in other areas of the potroom are likely exposed to more or less aged fumes.

The present study was aimed at measuring number concentration, number size distribution and to study particle morphology of ultrafine particles directly at the source during the process of anode change, and at locations representing typical mean exposure situations to aged smoke in the workroom air of a primary Al smelter.

Experimental

Site and process description

The measuring campaign was conducted in December 2002 in two potrooms of a primary smelter located at the west coast of Norway.

Two production areas were selected: (a) the prebake area where prebaked anodes are used in the electrolytic reduction cells. Since the prebake anodes had to be changed regularly the cell covers had to be removed, (b) the Søderberg area with a technological process where the mass loss of the anodes of the reduction cells was continuously replaced by adding a paste (petroleum coke and pitch) during the production process.¹⁴

Materials and methods

Number concentration and particle size distribution were measured continuously using commercially available instrumentation. For the total number concentration a hand-held condensation nucleus counter (CNC), (Model, 3007, TSI-Deutschland, Aachen, Germany) was used. In this instrument, ultrafine particles are enlarged in size by vapour condensation and subsequently counted by light scattering. The instrument samples the aerosol at a flow rate of 0.7 l min⁻¹ and counts all particles in the size range between 0.01 and approximately 1 μm. The measuring concentration range is up to 10⁵ particles cm⁻³. For higher concentrations, coincidence correction has to be taken into account. The portable counter is battery operated and can be used in a flexible way for a fast inspection of the concentration at various locations of interest.

The number size distribution was measured using a scanning mobility particle spectrometer (SMPS) (Model 3936L10, TSI-Deutschland, Aachen, Germany). This instrument combines particle size classification according to the particles’ mobility diameter and concentration measurement by counting in a CNC.¹⁵ The mobility diameter of a particle is the diameter of a sphere with the same air resistance and, hence, the same Brownian diffusion coefficient as the particle under consideration. Since the deposition of ultrafine particles in the respiratory tract is controlled by Brownian diffusion, the mobility diameter is the correct equivalent diameter to be measured.¹⁶

In the classification unit of the SMPS, the trajectories of single charged particles entering at the periphery of a concentric tube are deflected towards the central rod depending on the particles’ mobility diameter and the voltage applied between the outer and inner cylinder. Particles whose trajectories end at the slit in the bottom part of the central electrode, i.e. those with a given mobility diameter, will be sucked out of the system and will be measured. By scanning the voltage, a certain mobility diameter range will be scanned. This measuring range is determined by the air-flow passing through the cylinder and the classification length. The TSI-system was adjusted to a size range between 7.2 and to 294 nm by setting the sheath air flow to 10 l min⁻¹ and the aerosol flow rate to 0.9 l min⁻¹. The aerosol flow was established by a critical orifice inside a CNC (Model 3010,TSI-Deutschland, Aachen, Germany).

The count rates were collected by a computer that performs data analysis, presentation and storage. In this study, the size distribution was represented as a distribution density (q₀) with the logarithm of the mobility diameter as independent variable:

where d_mob is the mobility diameter and dN determines the concentration of particles in the size range between log(d_mob) and log(d_mob) + d log(d_mob). The total number concentration, given by:
is also calculated by the computer as well as distribution parameters such as median and mean particle diameter as well as the geometric standard deviation. Case measurements and long term overnight measurements with the SMPS were also carried out.

The particle morphology was analysed by sampling particles on carbon film covered copper grids (300 mesh, Plano GmbH, Wetzlar, Germany) that could be analysed in a transmission electron microscope (TEM) (LEO 910, Zeiss, Oberkochen, Germany) with no further sample treatment required. In total 4 electron microscope samples were collected throughout the measurement campaign; two long-term samples taken during the SMPS measurement periods in the Søderberg and prebake potrooms and two in the workers breathing zone during the anode change operations.

The deposition of the particles on the grids was by means of thermophoresis, i.e., the migration of particles in a temperature gradient. This gradient is established by heating the aerosol drawn into a tube (sampling flow rate 3 l min⁻¹) and cooling (by a Peltier cooler) the grid surface attached to a punch (see Fig. 1). The thermophoretic sampling unit has been developed at Fraunhofer ITEM-Hannover. Thermophoretic sampling was used because deposition of ultrafine particles by thermophoresis is nearly independent of particle size. In addition, the migration velocities are quite low (<0.1 ms⁻¹) so that no large forces are applied to the particles upon deposition onto surfaces. Hence, the sample provides a realistic picture of the size distribution and the morphology of the particles in the workroom air passing through the thermophoretic sampling unit.

Fig. 1 Schematic drawing of the thermophoretic sampling unit developed at Fraunhofer ITEM-Hannover. The diameter of the grid to be placed on the holder is 3 mm.

A sampling averaging of the particle concentration was performed by drawing the entire sample flow of 3.9 l min⁻¹ through a 10 l volume flask. The mean residence time of the sample in the buffer volume was of the order of the scan time (30 s) of the SMPS to measure a complete size distribution.

The SMPS, the computer and the thermophoretic sampler were all placed inside a steel box made of 5 mm black steel serving to shield the instrumentation against the high magnetic fields existing close to the electrolytic cells (Fig. 2). Experiences obtained during the practical work showed that the steel box had to be completely covered with the steel lid when the instrumentation was operated next to the pots.

Fig. 2 The shielding box and the instrument rig inside (prebake area).

Results and discussion

Total number concentration

The portable CNC was used to obtain an overview on the total number concentration in the prebake area, the Søderberg area as well as at selected specific work places: inside the crane cabin operating in the prebake area during the anode shift and inside the transport vehicle in the Söderberg area. Fig. 3 shows a concentration pattern as recorded when bringing the CNC into different work environments. During the time period I the instrument was in the prebake area next to the transfer door to the Søderberg area. The door was opened from time to time and the higher concentrations result from air entering from the Søderberg area which was visually much more polluted than the prebake area. The time period II represent measurements very close to cells in the prebake area where repair and anode change work was going on. Surprisingly the lowest concentrations in the prebake potroom were measured in between the (closed) cells indicated by time period III. These were significantly lower than the mean concentration in the hall. The time period (IV) represent typical concentrations in the Søderberg area not specifically taken close to strong emitters but collected in the walkways beside the electrolytic cells.

Fig. 3 The total number concentration in different work environments in the prebake potroom. Time period I: area next to a transfer door to Søderberg potroom; time period II: close to cells in with repair and anode change operations; time period III: between closed cells; time period IV: walkways beside the electrolytic cells in Søderberg potroom.

The exposure concentration inside the cabins of cranes and vehicles are significantly reduced likely due to the on-board air filtration system as can be seen from Fig. 4 (crane only). During the time periods indicated by the double arrow the hand held CNC was continuously measuring in the cabin. The reduction in the number concentration compared to the concentration outside the cabin is more than 95%. The exposure concentrations inside the vehicles are generally quite low (<10⁴ [cm⁻³]) except when the cabin is engulfed by the plume emitted from the open cells in the prebake area during e.g., anode changing (time period 30–80 min in Fig. 4) or when the driver opens the door and leaves the cabin. For comparison, the typical outdoor number concentration in polluted urban areas is 10⁵–4*10⁶ [cm⁻³].¹⁷

Fig. 4 The total number concentration inside the crane cabin operating above the cells in the prebake area during the time period indicated by the double arrow).

Mobility size distributions

Two long-term measurements were carried out in the prebake and the Søderberg potrooms, respectively. The air sample was taken at a height of approximately 2 m and was passed through a tube of 4 m length to the instruments. To verify that there was no wall deposition of particles in the tube, the total number concentration was compared with direct measurement with the portable CNC. A 10–20% lower total particle number concentration could be detected in air after passing through the 4 m length plastic tube. Since no coagulation of the aerosol is expected during the short time period required to pass the air through this tube, the particle size-distribution of the aerosol should not be influenced. The sampling positions were at the driveways between the cells. The temporal pattern of total number concentration and median diameter of these measurements are presented in Fig. 5. In the prebake area the influence of the activities during the working shift (from t = 150 to t = 900 min) on the particle exposure is clearly to be identified. Particle concentration peaks occurred periodically and the median particle size is smaller during the shift. This is likely to be caused by high fraction of nanoparticles in the size distribution that presumably emerge from the electrolytic cells during anode change and repair work. There is, as expected, a negative correlation between number concentration and median particle diameter. The concentration level of the ventilation air (minimum concentrations measured) is about 10000 [cm⁻³]. This level is higher than the atmospheric background since this ventilation air entering the building is already contaminated by emissions from other parts of the plant.

Fig. 5 Total number concentration, median diameter and 3D and contour plots of the mobility size distribution in the prebake and Søderberg potrooms.

In the Søderberg environment the general pollution level is higher and the particles are on average larger than in the prebake potroom. The minimum concentration level of 50000 [cm⁻³] is a factor of 5 higher than in the prebake area, the nanoparticle events are rare and the size distribution function peaks at around 40 nm most of the time compared with below 20 nm in the prebake potroom.

The specific sources of particles in the two potrooms are obviously different. In prebake, the anode change operations resulting in large open bath surfaces are most likely a significant source of nanoparticles. These operations were considered important since workers could be exposed to high number concentrations of nanoparticles. Due to experimental problems only one 1 h measurement campaign could be conducted where the size distribution close to the open cells during the anode change was monitored by the SMPS. For this purpose the instrument box was placed in the walkway beside the cells and was moved forward as the anode change progressed from cell to cell. The sample was drawn through a 4 m long sampling tube positioned close to the worker who cleaned the bath surface during the anode change. However, due to the strong convection flow, most of the emitted plume was going upwards, thus the direct plume was not accessible to size distribution measurements. Fig. 6 shows the time dependence of distribution parameters during this measurement. It started with monitoring of the background air entering the area through the large openings in the wall next to the cells. The first cell was opened 30 minutes after start and the core was broken by the crane operator (chiselling and hammering). The actual anode change work started at about t = 45 min and was continued cell after cell until t = 98 minutes. During this time period the anodes of 6 cells were exchanged. The concentration increased by an order of magnitude and the median particle size decreased to values significantly smaller than 20 nm. In order not to interfere with the anode change operation we were not able to sample directly in the plume from the open bath. We expect a large variation in the number concentration of particles in the breathing zone around the operators. The particles emitted were also so small that the measured values are already significantly influenced by the small particle detection limit of the CNC being in the range of 10 nm. This means that the concentration of smaller particles is underestimated due to the decreasing counting efficiency of the detector. It is evident that the aerosol emerging from the electrolytic cells is characterized by a number size distribution with its maximum below 10 nm i.e. a strong nanoparticle mode. This type of distribution is not untypical for high temperature and combustion related aerosol sources. The exposure concentration close to the cells is of the order of 10⁶ or higher dependent on the minimum particle size detected. After being released into the air, the nanoparticle mode is subject to ageing leading to a shift in the size distribution towards larger particles. This may explain the differences in the average size distribution close to the cells and in the background air in the prebake potroom.

Fig. 6 Total number concentration, median diameter and 3D and contour plots of the mobility size distribution during anode change operations in the prebake potroom.

Particle morphology

In Fig. 7 examples of the variety of particle morphologies present in the emissions are shown. These examples may indicate that different particle formation and crystallization processes are existing.

Fig. 7 Electron micrographs on samples taken parallel to the long term SMPS measurements in the prebake and Søderberg area. The size information indicated on top of each photo is the total width of the respective photo.

The general emission sampled in the prebake potroom is characterized by singlet particles, (carbon) agglomerates, as well as, to a minor extent, small fibres. All the (black) particles with strong electron absorption shown here are larger in size than 30 nm. Smaller nanoparticles as indicated by the SMPS size distribution should be present at much higher numbers than the larger particles. However, they are rarely to be identified in the TEM pictures. They either show a very small electron absorption or are composed of volatile substances undergoing evaporation during sampling when heated up to 180 °C in the thermophoretic sampling tube or upon evacuation of the sample chamber of the TEM. By having a closer look at the pictures taken at high magnification one could argue to see some grey spots resulting from the remainder of evaporated materials.

Particles collected in the Søderberg production hall show an even larger variety of morphologies. The fraction of fibres or needles is high (appr. 30% of total number). The presence of fibres/needles are consistent with previous studies performed by Gylseth et al.¹⁸ and Höflich et al.⁶ There are chain-like agglomerates with “small” and “large” primary particles, particles showing crystalline transformation, restructuring and evaporation when exposed to the electron beam. Also here, the number of (non-volatile) nanoparticles that can clearly be identified in the TEM pictures is relatively low when compared with the number size distribution as measured using the SMPS. The same is true for the samples taken directly at the source. Although the sampling time is an order of magnitude smaller here, the number of particles is expected to be comparable to the other two samples since the airborne concentration is a factor of 10 higher.

No further information can be extracted from simply looking at the particles’ morphology. A detailed single particle microanalysis would be required for this purpose.

Conclusion

This study shows the existence of elevated number concentrations of ultrafine particles at workplaces both in Søderberg and prebake potrooms of a primary aluminium smelter. The main source of this particle size fraction is the process of anode change in the prebake potroom. The ultrafine particles were measured directly at the source but could also be identified as episodes of high number concentrations in the general “background” air both in the Søderberg and prebake potrooms. The data on particle size and number concentration are not sufficient for toxicological assessment of the nanoparticles. Unlike the larger particles belonging to the 50–100 nm mode, the nanoparticle mode could not be detected in the TEM indicating that they may not be stable under the sampling conditions or the high vacuum in the instrument.

Further systematic investigations are required to improve the understanding on the properties of the nanoparticles.

References

1. G. Benke, M. Abramson and M. Sim, Ann. Occup. Hyg., 1998, 42(3), 173.
2. B. Sorgdrager, A. J. A. de Looff, J. G. R. de Monchy, T. M. Pal, A. E. J. Dubois and B. Rijcken, Int. Arch. Occup. Environ. Health, 1998, 71(1), 53.
3. B. Sorgdrager, A. J. A. de Looff, T. M. Pal, F. J. H. van Dijk and J. G. R. de Monchy, Int. Arch. Occup. Environ. Health, 2001, 74(1), 55.
4. K. Radon, D. Nowak, R. Heinrich-Ramm and D. Szadkowski, Int. Arch. Occup. Environ. Health, 1999, 72(5), 297.
5. J. Kongerud, J. Boe, V. Søyseth, A. Naalsund and P. Magnus, Eur. Respir. J., 1994, 7(1), 165.
6. B. L. W. Höflich, S. Weinbruch, R. Theissmann, H. Gorzawski, M. Ebert, H. M. Ortner, A. Skogstad, D. G. Ellingsen, P. A. Drabløs and Y. Thomassen, J. Environ. Monit., 2005, 7, 419.
7. W. G. Kreyling, M. Semmler and W. Moller, J. Aerosol Med., 2004, 17(2), 140.
8. B. S. Cohen, W. Li, J. Q. Xiong and M. Lippmann, Appl. Occup. Environ. Hyg., 2000, 15(1), 80.
9. A. P. Pietropaoli, M. W. Frampton, R. W. Hyde, P. E. Morrow, G. Oberdorster, C. Cox, D. M. Speers, L. M. Frasier, D. C. Chalupa, L. S. Huang and M. J. Utell, Inhal. Toxicol., 2004, 16(Suppl. 1), 59.
10. J. J. de Hartog, G. Hoek, A. Peters, K. L. Timonen, A. Ibald-Mulli, B. Brunekreef, J. Heinrich, P. Tiittanen, J. H. Wijnen, W. Kreyling, M. Kulmala and J. Pekkanen, Am. J. Epidemiol., 2003, 157, 613.
11. A. Peters, D. W. Dockery, J. Heinrich and H. E. Wichmann, Environ. Health Perspect., 1997, 105(4), 430.
12. A. T. Zimmer, J. Environ. Monit., 2002, 4(5), 628.
13. B. V. L’vov, L. K. Polzik, S. Weinbruch, D. G. Ellingsen and Y. Thomassen, J. Environ. Monit., 2005, 7, 425.
14. J. Thonstad, P. Fellner, G. M. Haarberg, J. Hives, H. Kvande and Å. Sterten, Aluminium Electrolysis, Aluminium-Verlag, Düsseldorf, 3nd edn, 2001.
15. Aerosol Measurements, ed. K. Willeke and P. A. Baron, Van Nostrand, New York, 1993, p. 418.
16. C. S. Kim and P. A. Jaques, Respiratory Dose of Inhaled Ultrafine Particles in Healthy Adults, in Ultrafine Particles in the Atmosphere, ed. L. M. Brown, N. Collings, R. M. Harrison, A. D. Maynard and R. L. Maynard, The Royal Society, London, 2000, ch. 10, pp. 169, ISBN 1-86094-358-6.
17. E. Weingartner, C. Keller, W. A. Stahel, H. Burtscher and U. Baltensperger, Atmos. Environ., 1997, 31, 451.
18. B. Gylseth, O. Børseth, Ø. Dugstad and J. Gjønnes, Scand. J. Work Environ. Health, 1984, 10, 189.

---

Footnote

† Presented at the Fifth International Symposium on Modern Principles of Air Monitoring & Biomonitoring, June 12–16 2005, Norway.

---