Inadvertent climate modification due to anthropogenic lead

Transcription

1 LETTERS PUBLISHED ONLINE: 19 APRIL 2009 DOI: /NGEO499 Inadvertent climate modification due to anthropogenic lead Daniel J. Cziczo 1,2 *, Olaf Stetzer 2, Annette Worringen 3, Martin Ebert 3, Stephan Weinbruch 3, Michael Kamphus 4, Stephane J. Gallavardin 2,4, Joachim Curtius 4,5, Stephan Borrmann 4,6, Karl D. Froyd 7, Stephan Mertes 8, Ottmar Möhler 9 and Ulrike Lohmann 2 Aerosol particles can interact with water vapour in the atmosphere, facilitating the condensation of water and the formation of clouds. At temperatures below 273 K, a fraction of atmospheric particles act as sites for ice-crystal formation. Atmospheric ice crystals which are incorporated into clouds that cover more than a third of the globe 1 are thought to initiate most of the terrestrial precipitation 2. Before the switch to unleaded fuel last century, the atmosphere contained substantial quantities of particulate lead; whether this influenced ice-crystal formation is not clear. Here, we combine field observations of ice-crystal residues with laboratory measurements of artificial clouds, to show that anthropogenic lead-containing particles are among the most efficient ice-forming substances commonly found in the atmosphere 3. Using a global climate model, we estimate that up to 0.8 W m 2 more long-wave radiation is emitted when 100% of ice-forming particles contain lead, compared with when no particles contain lead. We suggest that post-industrial emissions of particulate lead may have offset a proportion of the warming attributed to greenhouse gases. It is highly certain that the anthropogenic addition of greenhouse gases to the atmosphere has caused global warming 4. Conversely, the addition of small aerosol particles has caused regional solar dimming phenomena 5 that counteract some greenhouse-gas warming in a so-called direct effect. Aerosol particles can also interact with atmospheric water vapour by acting as sites of condensation and can thereby lead to the formation of clouds. Clouds, in turn, affect the global radiative balance by reflecting solar energy or trapping terrestrial radiation 6 ; this is termed an aerosol indirect effect. Net warming or cooling is dependent on cloud properties such as altitude and thickness. Depending on the ambient temperature and saturation, warm liquid water clouds, cold ice clouds or intermediate-temperature mixed-phase clouds can form. Owing to the typically high altitude and often remote location, the nucleation of ice is the less wellunderstood process. Despite this uncertainty, cirrus ice clouds can influence the Earth s radiative budget owing to their large global coverage 7. For example, it has been shown 8 that indirect aerosol perturbations to ice clouds can cause radiative changes of the same magnitude as the direct radiative impact of all anthropogenic particles. Ice can nucleate homogeneously from the aqueous droplets commonly found in the atmosphere, but this requires a supercooling of 40 K below the equilibrium freezing temperature 9,10. Atmospheric ice formation at higher temperatures requires the presence of a special solid particle that acts as an ice nucleus 10. Owing to the presence of a surface that enhances the stability of an ice embryo, this process is known as heterogeneous ice nucleation. An ice nucleus can cause ice nucleation by several modes, for example from within a liquid water droplet or by acting as a site for the direct deposition of water vapour. Not all solid aerosol particles enhance ice nucleation, but several materials have been shown to act as ice nuclei. These include, but are not limited to, mineral dust, anthropogenic metal oxides, pollen and bacteria 9. Each aerosol type shows a characteristic saturation and temperature, which varies depending on the mode by which freezing occurs. This freezing onset point is dependent on the surface area of the particle and some researchers have theorized that there are specific active sites that stabilize an ice embryo 10. Larger particles and materials that inherently contain more active sites thus exhibit a higher nucleation temperature 11. The exact nature of these sites remains unknown, but features such as surface defects are one candidate. Another atmospherically important property of ice nuclei is their abundance. For example, silver iodide and some biological materials are known to be effective ice nuclei, but it is uncertain if their abundance is large enough to induce a global radiative impact 12. Here, we have determined the nature of ice nuclei using three complementary methods: (1) creating artificial clouds by exposing ambient aerosol to controlled supersaturation and temperature, thereby mimicking atmospheric cloud formation, (2) examining the residue of ice crystals from naturally occurring clouds and (3) creating artificial clouds on laboratory-prepared aerosol particles (Supplementary Information contains a full discussion of the methods and data statistics in Supplementary Table S1). For all methods, mass spectrometry was used as the analytical technique and the percentage represents a number fraction of analysed particles. Using the first method, we find that mineral dust is the most common atmospheric aerosol type that acts as an 1 Atmospheric Sciences and Global Change Division, Pacific Northwest National Laboratory, 902 Battelle Blvd, Richland, Washington 99354, USA, 2 Institute for Atmospheric and Climate Science, ETH Zurich, Universitätstrasse 16, CH-8092 Zürich, Switzerland, 3 Institute for Applied Geosciences, Technical University Darmstadt, Schnittspahnstraße 9, D Darmstadt, Germany, 4 Institute for Atmospheric Physics, Johannes Gutenberg-University of Mainz, Joh.-Joachim-Becher-Weg 21, D Mainz, Germany, 5 Institute for Atmospheric and Environmental Sciences, Goethe-University Frankfurt am Main, Altenhöferallee 1, D Frankfurt am Main, Germany, 6 Particle Chemistry Department, Max Planck-Institute for Chemistry, D Mainz, Germany, 7 Chemical Sciences Division, National Oceanic and Atmospheric Administration, 325 Broadway Ave., Boulder, Colorado 80305, USA, 8 Leibniz-Institute for Tropospheric Research, D Leipzig, Germany, 9 Institute for Meteorology and Climate Research, Forschungszentrum Karlsruhe, Postfach 3640, D Karlsruhe, Germany. * daniel.cziczo@pnl.gov. NATURE GEOSCIENCE ADVANCE ONLINE PUBLICATION Macmillan Publishers Limited. All rights reserved.

2 LETTERS a b c Detector signal (normalized) Detector signal (normalized) Detector signal (normalized) C + Mg + Si + K + Ca + Zn C + Al + Pb Ion mass/charge Na + Al + K + Fe + Fe + I + Ion mass/charge BaO + Pb Ion mass/charge Figure 1 Positive ion mass spectra of single ice-crystal residual particles. a, Spectrum obtained at the Storm Peak Laboratory (3,200 m above sea level) in the Colorado Rocky Mountains when ambient aerosol was exposed to cirrus cloud conditions within a CFDC. b, Spectrum obtained at the JRS (3,600 m above sea level) in the Swiss Alps from the residue of an ice crystal from a mixed-phase cloud. c, Spectrum obtained during experiments when Arizona Test Dust was exposed to cirrus cloud conditions within the AIDA chamber. effective ice nucleus, representing 49% of all analysed particles 13. Figure 1a is the mass spectrum of one exemplary particle. The single component found with the highest frequency was lead, present in 32% of ice nuclei (an example of the isotopic identification of lead is given in Supplementary Fig. S1). The second method, the separation of ice crystals from naturally occurring clouds, yielded similar results for ice residue. The distinction between ice nuclei Pb + a b NATURE GEOSCIENCE DOI: /NGEO499 1 µm 1 µm Pb inclusions Figure 2 Secondary electron images of ice-crystal residual particles from mixed-phase clouds at the JRS. a, Lead inclusions (black dots), visible from an overlay of a scanning electron image and lead mapping by energy dispersive X-ray microanalysis of this residual silicate particle. b, A carbon-, chlorine- and lead-containing residual particle where lead is in the form of organic lead and/or lead halides. and ice residue is that the latter contains the ice nuclei and any gas and particle-phase material that adhered to the ice crystal during its atmospheric lifetime. As an example, Fig. 1b shows that the most common class of ice residue was again mineral dust, observed for 67% of the residue. The most common component was lead, observed in 42% of this residue. The frequency of mineral dust found using the first and second methods is in agreement with past studies, including identification of the central aerosol particle embedded within snowflakes 9. The correlation between lead and ice nucleus concentration was observed in a previous study of bulk ice nucleus properties 14, although this was attributed to both also being correlated to particle size. The authors suggested studies at the single-particle level to verify this 14 as have been carried out in this study. The results using the first two methods motivated the third set of studies, the formation of artificial clouds on collected samples of mineral dust prepared in a laboratory environment. This allowed control of conditions with greater precision than was possible in a field setting; previous studies on idealized mineral dust samples 2 NATURE GEOSCIENCE ADVANCE ONLINE PUBLICATION Macmillan Publishers Limited. All rights reserved.

3 NATURE GEOSCIENCE DOI: /NGEO499 RHi (%) Kaolinite Kaolinite + 1% PbSO 4 PbSO Temperature (K) Figure 3 Temperature and relative humidity required for the onset of ice nucleation using a cloud chamber. Conditions required for ice nucleation by kaolinite mineral dust and kaolinite doped with lead sulphate are shown. Below 240 K, the lead-doped kaolinite particles, which mimic atmospheric aerosol that coagulated with lead, induce ice nucleation at a lower saturation for a given temperature than undoped kaolinite. Pure lead sulphate is, in contrast, an inefficient nucleus. The solid line represents the saturation vapour pressure of liquid water. a b Outgoing long wave radiation (W m 2 ) Ice water path (g m 2 ) ECHAM5 Pb0 ECHAM5 Pb1 ECHAM5 Pb10 ECHAM5 Pb100 Latitude LETTERS 80 S 60 S 40 S 20 S 0 20 N 40 N 60 N 80 N 80 S 60 S 40 S 20 S 0 20 N 40 N 60 N 80 N Latitude have been used to determine specific ice nucleation threshold conditions 15. For the collected samples of mineral dust, lead was present in 8% of the aerosol particles, but in the subset that caused ice nucleation, lead was found in 44%. Figure 1c is an example of a lead-containing particle that nucleated ice under controlled laboratory conditions. By analysis of Antarctic ice cores, Boutron and Patterson 16 showed that lead from mineral dust and volcanic emissions were the main atmospheric sources during pre-industrial times. Anthropogenic activities now dominate, accounting for >99% in the mid-1980s. At that time, most of the lead introduced to the atmosphere was from tetraethyl lead (TEL) added to automotive petrol. With regulation of TEL, total lead has dropped significantly, with a 20-fold decrease reported in the continental United States in the two decades since 1980 (ref. 17). TEL in light aviation fuel is now thought to be the dominant new source, although the uncertain and increasing emissions from coal combustion and smelting could be equivalent or greater 3,17. In a study of atmospheric particles from 50 nm to 2 µm diameter, Murphy et al. 3 showed that lead was detectable in 5 10%. The size dependence of the lead concentration and the mixing state supported the coagulation of small (<50 nm) primary lead particles with preexisting atmospheric aerosol as the source of most of the particles 3. Since the mid-1940s it has been known that artificial ice nuclei could be produced 18. Two materials found to nucleate ice as high as a few degrees below 273 K were silver and lead iodide 19, although the high toxicity of the latter limited its use. Lead oxides and mixtures with ammonium iodide were subsequently found to be similar, if not better, ice nuclei than silver and lead iodide (ref. 20). Baklanov et al. showed that pure lead-containing materials were not required for ice nucleation; instead, lead need only be present as a surface inclusion on an inert core so long as its size remains at or above that of an active site, assumed to be nm in diameter 21. Figure 2 contains electron microscope images of lead-containing ice residuals from the Jungfraujoch Research Station (JRS). Using energy-dispersive X-ray microanalysis, lead was identified in 15% of the ice residuals. Surface inclusions, detected on 80% of the lead-containing particles, are denoted as black dots. Chemical Figure 4 Atmospheric properties with and without the influence of anthropogenic lead. a,b, Ice water path (a) and outgoing long-wave radiation (b) difference between the present-day and pre-industrial times. The four sensitivity conditions are 0, 1, 10 and 100% of mineral dust containing lead inclusions. analysis shows lead bound in several forms including lead sulphide, lead halides, organic lead and lead oxide. Individual inclusions are of the order of 10 nm diameter, consistent with coagulation of small primary lead particles onto pre-existing atmospheric aerosol as it passes through lead source regions 3. This is the method recommended by Baklanov et al. for maximizing the ice nucleating potential of an inert aerosol, unintentionally carried out, by adding surface inclusions of ice nucleus material 21. We therefore contend that anthropogenic emissions of lead to the atmosphere have had the effect of supercharging pre-existing particles, making them highly efficient ice nuclei. Furthermore, we note that the altitude at which light aircraft fly is above the planetary boundary layer and directly places the emitted lead at an altitude where ice and mixed-phase clouds form. The ice nucleation potential of atmospheric aerosol that coagulated with small primary lead particles has not previously been investigated. Previous research has instead focused on the production of artificial ice nuclei for weather modification The form of atmospheric lead may vary, depending on the specific emission, with lead sulphate and oxide being two common forms 3. The former has not been investigated as an ice nucleus. Laboratory studies were carried out by atomization of the precipitate of lead sulphate crystals in a water slurry containing kaolinite mineral dust. The results are illustrated in Fig. 3. Doping was at an atmospherically relevant level of 1% lead sulphate to kaolinite by mass 3 and this was compared with undoped particles. The presence of lead decreased the saturation required for the nucleation of ice when compared with pure kaolinite by 10 20% relative humidity with respect to water ice at temperatures below 240 K. To estimate the climatic effect of lead-containing ice nuclei, global climate model simulations were carried out with homogeneous and heterogeneous nucleation in cirrus clouds with NATURE GEOSCIENCE ADVANCE ONLINE PUBLICATION Macmillan Publishers Limited. All rights reserved.

4 LETTERS temperatures below 238 K (refs 22, 23). We assume that 0, 1, 10 or 100% of the internally mixed mineral dust particles in a present-day climate simulation contain lead and that lead-containing particles initiate freezing at 110% relative humidity with respect to ice (RHi). The first case is a modern world with no lead emissions, the second and third are estimates for the present day and the fourth may be similar to conditions if TEL was unregulated. The remaining immersed dust particles initiate freezing at 130% RHi and all other supercooled solution droplets freeze homogeneously. As shown in Fig. 4, the change in the ice water path, the vertically integrated ice water content in mixed-phase and cirrus clouds, since pre-industrial times is largest in the mid-latitudes of the Northern Hemisphere. The increase is similar in the four sensitivity simulations despite the fact that the increase in lead-containing mineral dust particles causes more particles to initiative freezing at lower RHi. This increase in lead-containing particles corresponds to warmer temperatures and lower cloud altitudes, and thus, more emitted long-wave radiation. As compared with the case where mineral dust particles contain no lead in the present day, up to 0.8 W m 2 more long-wave radiation is emitted when 100% of the mineral dust aerosols contain lead. Methods Three types of experiment were undertaken to determine the nature of ice nuclei and ice residue. First, ice crystals were formed from ambient aerosol within a litre-sized continuous flow diffusion chamber 24 (CFDC). Also known as cloud chambers, CFDCs are used to mimic the atmospheric temperature and relative humidity at which clouds form to facilitate experiments in a controlled laboratory or field setting. Ice crystals formed within the CFDC were isolated for analysis using counterflow virtual impaction 25. Second, crystals from naturally occurring mixed-phase clouds at the Jungfraujoch Research Station (JRS) were separated for analysis using an ice counterflow virtual impaction 26. Third, experiments were conducted within the Aerosol Interactions and Dynamics in the Atmosphere (AIDA) cloud chamber, an 84 m 3 actively cooled vessel 27, as well as a CFDC on collected samples of mineral dust. Subsequent experiments were conducted on collected samples doped with lead. Size and compositional analysis of ice nuclei and ambient aerosol was carried out using single-particle mass spectrometry 28,29 and electron microscopy coupled to energy-dispersive X-ray microanalysis 30. Supplementary Information contains a full discussion of the methods. Data statistics are given in Supplementary Table S1. Received 4 December 2008; accepted 19 March 2009; published online 19 April 2009 References 1. Wylie, D. P. & Menzel, W. P. Eight years of high cloud statistics using HIRS. J. Clim. 12, (1999). 2. Lohmann, U. et al. Cloud microphysics and aerosol indirect effects in the global climate model ECHAM5-HAM. Atmos. Chem. Phys. 7, (2007). 3. Murphy, D. M. et al. Distribution of lead in single atmospheric particles. Atmos. Chem. Phys. 7, (2007). 4. Solomon, S. et al. (eds) Climate Change 2007: The Physical Science Basis. Contribution of Working Group I to the Fourth Assessment Report of the IPCC (Cambridge Univ. Press, 2007). 5. Liepert, B. G. Observed reductions of surface solar radiation at sites in the United States and worldwide from 1961 to Geophys. Res. Lett. 29, (2002). 6. Yau, M. K. & Rogers, R. R. Short Course in Cloud Physics 3rd edn (Butterworth-Heinemann, 1996). 7. Liou, K.-N. Influence of cirrus clouds on weather and climate processes: A global perspective. Mon. Weath. Rev. 114, (1986). 8. Lohmann, U. A glaciation indirect aerosol effect caused by soot aerosols. Geophys. Res. Lett. 29, (2002). 9. DeMott, P. J. in Cirrus (eds Lynch, D. K., Sassen, K., Starr, D. C. & Stephens, G.) (Oxford Univ. Press, 2002). 10. Pruppacher, H. R. & Klett, J. D. Microphysics of Clouds and Precipitation 2nd edn (Kluwer, 1997). 11. Archuleta, C. M., DeMott, P. J. & Kreidenweis, S. M. Ice nucleation by surrogates for atmospheric mineral dust and mineral dust/sulfate particles at cirrus temperatures. Atmos. Chem. Phys. 5, (2005). NATURE GEOSCIENCE DOI: /NGEO Möhler, O., DeMott, P. J., Vali, G. & Levin, Z. Microbiology and atmospheric processes: The role of biological particles in cloud physics. Biogeosciences 4, (2007). 13. DeMott, P. J. et al. Measurements of the concentration and composition of nuclei for cirrus formation. Proc. Natl Acad. Sci. 100, (2003). 14. Borys, R. D. & Duce, R. A. Relationships among lead, iodine, trace metals and ice nuclei in a coastal urban atmosphere. J. Appl. Meteorol. 18, (1979). 15. Möhler, O. et al. Efficiency of the deposition mode ice nucleation on mineral dust particles. Atmos. Chem. Phys. 6, (2006). 16. Boutron, C. F. & Patterson, C. C. Lead concentration changes in Antarctic ice during the Wisconsin/Holocene transition. Nature 323, (1986). 17. Environmental Protection Agency. Latest findings on National Air Quality, 2002 Status and Trends (US Environmental Protection Agency, 2003); available at < 18. Vonnegut, B. The nucleation of ice formation by silver iodide. J. Appl. Phys. 18, (1947). 19. Schaefer, V. J. Silver and lead iodides as ice-crystal nuclei. J. Meteorol. 11, (1954). 20. Reischel, M. T. Lead nuclei from reactions involving lead, ammonia, and iodine. Atmos. Environ. 9, (1975). 21. Baklanov, A. M. et al. The influence of lead iodide aerosol dispersity on its ice-forming activity. J. Aerosol. Sci. 22, 9 14 (1991). 22. Lohmann, U., Spichtinger, P., Jess, S., Peter, T. & Smit, H. Cirrus cloud formation and ice supersaturated regions in a global climate model. Environ. Res. Lett. 3, (2008). 23. Stier, P. et al. The aerosol-climate model ECHAM5-HAM. Atmos. Chem. Phys. 5, (2005). 24. Stetzer, O., Bascheck, B., Lüönd, F. & Lohmann, U. The Zurich ice nucleation chamber (ZINC) a new instrument to investigate atmospheric ice formation. Aerosol. Sci. Technol. 42, (2008). 25. Ogren, J. A., Heintzenberg, J. & Charlson, R. J. In situ sampling of clouds with a droplet to aerosol converter. Geophys. Res. Lett. 12, (1985). 26. Mertes, S. et al. Counterflow virtual impactor based collection of small ice particles in mixed-phase clouds for the physico-chemical characterization of tropospheric ice nuclei: Sampler description and first case study. Aerosol. Sci. Technol. 41, (2007). 27. Möhler, O. et al. Experimental investigation of homogenous freezing of sulphuric acid particles in the aerosol chamber AIDA. Atmos. Chem. Phys. 3, (2003). 28. Gallavardin, S. J. et al. Single particle laser mass spectrometry applied to differential ice nucleation experiments at the AIDA chamber. Aerosol. Sci. Technol. 42, (2008). 29. Murphy, D. M. The design of single particle laser mass spectrometers. Mass Spectrom. Rev. 26, (2007). 30. Zimmermann, F., Ebert, M., Worringen, A., Schütz, L. & Weinbruch, S. Environmental scanning electron microscopy (ESEM) as a new technique to determine the ice nucleation capability of individual atmospheric aerosol particles. Atmos. Environ. 41, (2007). Acknowledgements We thank P. J. DeMott, D. M. Murphy and D. S. Thomson for their assistance with the measurements. We also acknowledge the effort of all of the participants of the INSPECT and CLACE field studies, the support of the High Altitude Research Foundation Gornergrat and Jungfraujoch and the experimental group at AIDA. This research was supported by the Atmospheric Composition Change the European Network for Excellence, ETH Zurich, the German Research Foundation and Pacific Northwest National Laboratory directed research funding. Author contributions D.J.C., single-particle mass spectrometry, data analysis and paper writing; O.S., ice nucleation experiments and data analysis; A.W., M.E. and S.W., sample acquisition, electron microscopy and data interpretation and analysis; M.K., S.J.G., J.C. and S.B., mass spectrometer development, sample acquisition for single-particle mass spectrometry and data analysis; S.M., ice crystal sample acquisition and data analysis; O.M. and K.D.F., conducted AIDA experiments and data analysis; U.L., GCM programming and data analysis. Additional information Supplementary information accompanies this paper on Reprints and permissions information is available online at reprintsandpermissions. Correspondence and requests for materials should be addressed to D.J.C. 4 NATURE GEOSCIENCE ADVANCE ONLINE PUBLICATION Macmillan Publishers Limited. All rights reserved.

5 SUPPLEMENTARY INFORMATION doi: /ngeo499 Supplementary Information Methods Three types of experiments were undertaken to determine the nature of ice nuclei and ice residue. First, ice crystals were formed from ambient aerosol within a liter-sized continuous flow diffusion chamber (CFDC) 1. Ice crystals were then separated for analysis using a counterflow virtual impactor (CVI) 2. Second, ice crystals in naturally occurring mixed-phase clouds were separated for analysis using an Ice-CVI specifically designed for mixed-phase conditions 3. Third, experiments were conducted within the aerosol interactions and dynamics in the atmosphere (AIDA) cloud chamber 4 on collected samples of mineral dust. Experiments were also performed on collected samples of mineral dust doped with lead with a CFDC. Analysis was in all cases performed using single particle mass spectrometry (SPMS) 5,6. Electron microscopy (EM) coupled to energy dispersive X-ray microanalysis 7 was also used for ambient ice residue characterization. CFDC A diffusion chamber functions by generating an environment with a defined temperature below the melting point of water and a supersaturated relative humidity with respect to ice. These conditions are produced in the space between two ice coated walls held at different temperatures and mimic those at which atmospheric mixed-phase and ice clouds form. Diffusion leads to a linear gradient in temperature and absolute humidity between the walls but, because of the exponential relation between temperature and saturation vapor pressure, a supersaturation is achieved close to the center. A sample aerosol flow layered between two particle-free sheath flows constrains the conditions to which the nature geoscience 1

6 supplementary information doi: /ngeo499 sample is exposed. CFDC volumes were a few liters with a residence time on the order of seconds 1. CVI A conventional CVI inertially separates particles by directing a dry, particle-free, flow of gas in a direction opposite that of the sample. The rate of this counterflow can be adjusted to create a specific inertial cutpoint 2. A conventional CVI was used for ice crystal separation during droplet-free conditions in the CFDC and AIDA experiments 8. One limitation is that a conventional CVI can not be used to distinguish frozen and liquid hydrometeors of the same inertia when sampling in mixed-phase conditions. A custom built Ice-CVI 3 was used in order to overcome this limitation for field studies. A vertically oriented inlet horn in combination with a virtual impactor (VI) was used to remove precipitating ice particles larger than 20 µm diameter. The upper limit of 20 µm assured an aspiration efficiency of ~1 for smaller ice particles. Downstream of the VI a preimpactor (PI) removed supercooled drops by contact freezing on cold impaction plates. Ice particles bounced and passed the impaction plates. A conventional CVI was located downstream of the PI to reject interstitial particles smaller than 5 µm. Ice particles in the 5-20 µm diameter range were thus passed by the Ice-CVI. As with a conventional CVI these small ice crystals were injected into a particle-free and dry carrier gas which led to evaporation and allowed for the analysis of the residual material. AIDA The ice nucleation properties of laboratory-prepared aerosol, including re-dispersed mineral dust samples, can be investigated at simulated conditions of natural clouds in the AIDA chamber at the Forschungszentrum Karlsruhe 4,5. Experiments typically begin at ice 2 nature geoscience

7 doi: /ngeo499 supplementary information saturated conditions in the 84 m 3 actively-cooled cloud chamber due to a thin ice coating on the chamber walls. The formation of a cloud is induced by expansion cooling via vacuum pumping which mimics an updrafting atmospheric air parcel. Droplets and ice particles are detected by optical particle counters, in situ scattering and depolarization measurements, and Fourier transform infrared spectroscopy. Clouds typically exist within the AIDA chamber for several minutes. SPMS The single particle mass spectrometers used in this study draw aerosol into the instrument through an aerodynamic inlet which imparts a size dependent velocity. Light scattering as particles pass through two visible lasers beams set a known distance apart indicates the presence of a sample and, when combined with the size dependant velocity, provides the aerodynamic diameter. A near ultraviolet laser is triggered by the light scattering event to desorb the aerosol material and ionize it. The chemical composition can then be inferred from ions detected using a time of flight mass spectrometer. Analysis is performed in situ and in real time on a particle by particle basis. A recent review of SPMS was conducted by Murphy 6. The lower size limit of a SPMS is set by the detection limit of scattered light; this is normally between nm aerodynamic diameter. The upper limit is set by the aerodynamic properties of the inlet and this is normally between 2 and 3 m diameter 6. It is noteworthy that previous studies have shown that atmospheric ice nuclei are found within this range 9,10. Without extensive laboratory testing particle matrix effects and the disparate ionization efficiencies of common aerosol materials single particle mass spectrometers produce nature geoscience 3

8 supplementary information doi: /ngeo499 qualitative, not quantitative, data 11. Isotopic abundance can be used to identify specific elements, as is shown in Figure S1 for lead in an ice residual from a naturally occurring mixed-phase cloud. Gallavardin et al. showed that specific mineralogical information was difficult to ascertain on the single particle level, however, and we therefore use the broad classifications mineral dust and lead containing in this study 12. A list of the number of spectra analyzed for each study is given in Table S1. EM Size, morphology, and elemental composition of individual ice residue and interstitial particles (i.e., those which did not form droplets or ice) were analyzed during the field studies using environmental scanning EM and transmission electron microscopy (TEM), each combined with energy dispersive X-ray microanalysis 7. Samples from the Ice-CVI were impacted onto EM grids which were analyzed at the Institute for Applied Geosciences at the Technical University Darmstadt. Lead inclusions of <100 nm diameter were studied with High Resolution-TEM. Field Sites Storm Peak Laboratory (SPL) Field experiments were conducted on ambient aerosol at SPL located at 3200 meters above sea level (m.s.l.) in the Colorado Rocky Mountains. The high altitude and absence of local sources results in access to free tropospheric conditions, typically during overnight periods when subsidence of the planetary boundary layer occurs 13. Experiments at SPL were performed during the fall of 2001 and spring of 2004 by exposing free 4 nature geoscience

9 doi: /ngeo499 supplementary information tropospheric aerosol to cirrus cloud formation conditions with a CFDC. Ice crystals were then separated with a CVI and analyzed using SPMS 14. Jungfraujoch Research Station (JRS) The JRS is located at 3600 m.s.l. in the Swiss Alps in a region with minimal local particle sources. Mixed-phase clouds occur with a frequency of 40% during late winter and early spring at this site 15. Ice crystals were separated when mixed-phase clouds were present using the Ice-CVI. Residue was analyzed with SPMS and samples were collected for offline EM during the late winter and early spring of 2006 and Laboratory Particle Preparation Laboratory studies at the AIDA chamber were conducted on Arizona Test Dust (Powder Technology, Inc.). The mineral samples were dispersed into a particle-free flow of dry synthetic air with a small scale powder disperser (TSI, Inc.). Large dust particles were removed in an inertial impactor with a cutoff diameter of 1 m diameter before being added to the aerosol chamber. Laboratory CFDC experiments were performed on kaolinite clay (Fluka, Inc.). The kaolinite was dispersed in distilled deionized water and aerosolized with a custom build atomizer. Aerosol both in an un-doped state and associated with 1% lead sulfate by mass was produced and size selected at 200 nm diameter with a differential mobility analyzer (TSI, Inc.). The internal mixing state of mineral dust and lead was ascertained with SPMS. nature geoscience 5

10 supplementary information doi: /ngeo499 Global Climate Model Simulations For the global climate simulations the ECHAM5 general circulation model was used. ECHAM5 includes a two-moment cloud microphysics scheme 16 coupled to the ECHAM5-HAM two-moment aerosol scheme which predicts the aerosol mixing state in addition to the aerosol mass and number concentrations 17. The size-distribution is represented by a superposition of log-normal modes including the major global aerosol compounds sulfate, black carbon, organic carbon, sea salt and mineral dust. In order to avoid changes in meteorology, both the pre-industrial and present-day simulations were nudged to the ECMWF ERA40 reanalysis data for the year 2000 after an initial spin-up of 3 months 17. The aerosol emissions of the pre-industrial simulations are representative of the year The simulations were conducted in T42 horizontal resolution (~2.8º x 2.8º) with 19 vertical layers. 6 nature geoscience

11 doi: /ngeo499 supplementary information Supplementary Figure intensity (a.u.) mass/charge Supplementary Figure 1: Identification of lead from isotopic abundances. This exemplary ice residual, shown in arbitrary units (a.u.) of signal intensity versus positive ion mass to charge, is from a mixed-phase cloud at the Jungfraujoch Research Station. The isotopic abundance of lead at mass 204, 206, 207, and 208 is 1.4, 24.1, 22.1, and 52.4 %, respectively. This pattern, discernable using SPMS, was used to identify lead. nature geoscience 7

12 supplementary information doi: /ngeo499 Supplementary Table 1 Experiment Cloud Source Location Ice Nuclei Ice Residue Aerosol SPMS Spectra Mineral Dust Lead 1 CFDC SPL X Ambient ~2500 total 49% 32% (cloud chamber) ~250 IN 2 Natural clouds JRS X Ambient ~1750 total 67% 42% (mixed-phase) ~500 residue 3 AIDA AIDA X Resuspended ~ % 44% (cloud chamber) mineral dust Supplementary Table 1: Source, location, ice crystal and aerosol origin, and spectral statistics for the 3 experimental studies. Experiments 1 and 3 utilized ice chambers while 2 was performed from within naturally occurring clouds. Experiments 1 and 2 were conducted on ambient aerosol while 3 was performed on resuspended mineral dust. In total ~3150 ice-forming aerosols were analyzed of which 42% contained lead. Note that lead is present at the 5-10% level in ambient aerosol 18. References 1. Stetzer, O., Bascheck, B., Lüönd, F. & Lohmann, U. The Zurich ice nucleation chamber (ZINC)-a new instrument to investigate atmospheric ice formation. Aero. Sci. Tech. 42, (2008). 2. Ogren, J. A., Heintzenberg, J. & Charlson, R. J. In situ sampling of clouds with a droplet to aerosol converter. Geophys. Res. Lett. 12, (1985). 3. Mertes, S. et al. Counterflow virtual impactor based collection of small ice particles in mixed-phase clouds for the physico-chemical characterization of tropospheric ice nuclei: sampler description and first case study. Aero. Sci. Tech. 41, (2007). 8 nature geoscience

13 doi: /ngeo499 supplementary information 4. Möhler, O. et al. Experimental investigation of homogenous freezing of sulphuric acid particles in the aerosol chamber AIDA. Atmos. Chem. Phys. 3, (2003). 5. Gallavardin, S. J., Froyd, K., Lohmann, U., Möhler, O., Murphy, D. M. & Cziczo, D. J. Single particle laser mass spectrometry applied to differential ice nucleation experiments at the AIDA chamber. Aero. Sci. Tech. 42, (2008). 6. Murphy, D. M. The design of single particle laser mass spectrometers. Mass Spec. Rev. 26, (2007). 7. Zimmermann, F., Ebert, M., Worringen, A., Schütz, L. & Weinbruch, S. Environmental scanning electron microscopy (ESEM) as a new technique to determine the ice nucleation capability of individual atmospheric aerosol particles. Atmos. Environ. 41, (2007). 8. Boulter, J. E., Cziczo, D. J., Middlebrook, A. M., Thomson, D. S. & Murphy, D. M. Design and performance of a pumped counterflow virtual impactor. Aero. Sci. Tech. 40, (2006). 9. Chen, Y. L., Kreidenweis, S. M., McInnes, L. M., Rogers, D. C. & DeMott, P. J. Single particle analyses of ice nucleating aerosols in the upper troposphere and lower stratosphere. Geophys. Res. Lett. 25, (1998). nature geoscience 9

14 supplementary information doi: /ngeo Cziczo, D. J., Thomson, D. S., Thompson, T. L., DeMott, P. D., & Murphy, D. M. Particle analysis by laser mass spectrometry (PALMS) studies of ice nuclei and other low number density particles. Inter. J. Mass Spec. 258, (2006). 11. Cziczo, D. J., Thomson, D. S. & Murphy, D. S. Ablation, flux, and atmospheric implications of meteors inferred from stratospheric aerosol. Science 291, (2001). 12. Gallavardin, S. J., Lohmann, U., & Cziczo, D. J. Analysis and differentiation of mineral dust by single particle laser mass spectrometry. Inter. J. Mass Spec. 274, (2008). 13. Borys, R. D. & Wetzel, M. A. Storm Peak Laboratory: A research, teaching and service facility for the atmospheric sciences. Bulletin. Amer. Meteorol. Soc. 78, (1997). 14. DeMott, P. J. et al. Measurements of the concentration and composition of nuclei for cirrus formation. Proc. Nat. Acad. Sci. 100, (2003). 15. Cozic, J. et al. Black carbon enrichment in atmospheric ice particle residuals observed in lower tropospheric mixed phase clouds. J. Geophys. Res. 113, D15209 (2008). 10 nature geoscience

15 doi: /ngeo499 supplementary information 16. Lohmann, U., Spichtinger, P., Jess, S., Peter, T. & Smit, H. Cirrus cloud formation and ice supersaturated regions in a global climate model. Environ. Res. Lett. 3, (2008). 17. Stier, P., et al. The aerosol-climate model ECHAM5-HAM. Atmos. Chem. Phys., 5, (2005). 18. Murphy, D. M. et al. Distribution of lead in single atmospheric particles. Atmos. Chem. Phys. 7, (2007). nature geoscience 11