Characteristics of Airborne Particulates in Urban Areas

Dissertation for the Degree of PhD

By

RuiKai Xie

DEPARTMENT OF CHEMISTRY

FACULTY OF MATHEMATICS AND NATURAL SCIENCES

UNIVERSITY OF OSLO OCTOBER 2007

© RuiKai Xie, 2007

Series of dissertations submitted to the Faculty of Mathematics and Natural Sciences, University of Oslo Nr. 678

ISSN 1501-7710

All rights reserved. No part of this publication may be reproduced or transmitted, in any form or by any means, without permission.

Cover: Inger Sandved Anfinsen. Printed in Norway: AiT e-dit AS, Oslo, 2007.

Produced in co-operation with Unipub AS. The thesis is produced by Unipub AS merely in connection with the thesis defence. Kindly direct all inquiries regarding the thesis to the copyright holder or the unit which grants the doctorate.

Unipub AS is owned by The University Foundation for Student Life (SiO)

ACKNOWLEDGEMENTS

PhD is often teased as Permanent Head Damage, or Piled Higher and Deeper, or Patiently Hoping for a Degree, so on and so forth. Well, some of these may overstate the hardship; nonetheless, it has never been an easy job.

So, finally, I can call it a finish! For years, I have been richly blessed by numerous people on and off campus. I wish I could pop the champagne with the presence of all those who have aided me in the ways they did or did not know. It is truly a luxury to have so many fabulous people to thank for their help in midwiving this dissertation.

My advisor, Hans Martin Seip, has been a mentor, role mode, and critic in the best of senses. His uncompromised insistence on quality balanced by granting freedom to explore will be beneficial to me for the years to come. He is the very person who can turn water into wine with quick wit and incredible imagination, manifested by the refreshed looks of the draft manuscripts after his artistic input of his own thoughts. “Thank you” does not seem sufficient, but it is said with sincerity and respect!

My co-advisor, Grethe Wibetoe, has been always there, willing to listen and to give advice. I am deeply grateful to her for discussions. I also thank her for correcting the grammatical errors no matter how big or how trivial they might be in the manuscripts.

I owe a lot to my other co-advisor, Jan Erik Hanssen. I only regret that this dissertation did not turn out earlier. I was stunned to learn his passing away. He is to be remembered as a dedicative, easygoing, and genuine man.

Rolf David Vogt and Thorjørn Larssen, both have influenced me to shape this study. I have always been fascinated by Rolf’s tough yet probing questions and trenchant and insightful comments during seminars. Sharing an office and discussing about environmental issues with Thorjørn were something I enjoyed very much. I also need to say a big “THANK YOU” to both of you for providing some extra funding, through which I could see generosity flowing so naturally.

Gratitude goes to the lab technicians, Anne-Marie Skramstad, Hege Lynne, and many others. They have helped me behind the scene without ever receiving due credits for their impact of their assistance.

Appreciation is extended to the library staff at the University of Oslo. I am always amazed by their excellent and prompt services and the power of electronic communication in information retrieval.

I am also thankful to my friends and colleagues in England. Cameron William McLeod hosted me at the University of Sheffield for ten months. Interacting with the students and academics there only did me good. Thanks to David Anderson and Mario-Jo Schofield, I could do some research work in Corus RD&T, Swinden Technology Centre for six months. Financial support from both institutions is profoundly appreciated.

ii

Thanks are due to my colleagues in the environmental group. People come and people go, yet the spirit stands still. I treasure the thought-provoking discussions and communications, either open or personal. Listing all the names would cost pages. You are always memorized and friendship is always cherished!

Many Chinese friends in Oslo have helped me stay sane throughout these harsh years. Their warm support and ubiquitous care helped me overcome setbacks and stay focused on my PhD study. To name a few of them will only throw a large party behind. What I can say is “xie xie”, the Chinese version of “thank you”, out of fear that the English version “thank you” would dilute what it truly means.

My immediate family has been a constant source of love, affection, and strength all these years. My wife, Min Zeng, has in many ways contributed to the present form of the dissertation. She has undertaken all the logistical tasks, acted as a model mother and wife, and served as a stabilizing force in the family. No single complaint has ever been filed even during the last months of write-up when every possible funding source drained dry. For all of these, I feel indebted. I cordially thank my sons: Tian and Di. I feel sorry for not giving them love and time much needed for them to prosper during their young ages.

Finally, I sincerely acknowledge the financial support from Quota Program and Worldwide University Network that funded parts of this study.

Ruikai Xie

Oslo, October 2007

iii

ABSTRACTS

The characterization of particulate matter (PM) provides information important to the understanding of our environment and associated health risks. In this research project, experimental methods have been developed for the collection and characterization of urban PM. Samples have been collected in three cities: Guiyang and Taiyuan in China, and Sheffield in the UK. Three technical approaches have been used: microscopic analysis, bulk sample analysis, and size-resolved analysis. The results produced add to the understanding of airborne PM in the selected areas.

1. Microscopic study: samples were collected on Teflon filters. The sampling was set to get sufficient particles distributed on the surface while avoiding overloading. Individual particles collected on the filter were analyzed by Scanning Electron Microscope with Energy-Dispersive Spectrometer (SEM-EDS). This technique was applied to two sets of samples from two cities, which produced one paper and one manuscript: 1), Chemical characterization of individual particles (PM10) from ambient air in Guiyang City, China (Science of the Total Environment, 2005, 343, 261-72), and 2), Individual particles (PM10) characterization in Taiyuan City, China. Distinctly different features were found. Although anthropogenic sources dominated in both cities, we concluded that Guiyang’s particulate matter was due largely to metallurgical industries, manganese smelting in particular; and in Taiyuan it was mainly from coal combustion.

2. Bulk analysis: samples were collected on Teflon filters. The sampling duration was from a few hours to one day to get optimal loading. Samples were first digested with combination of acids and afterwards analysed by Inductively Coupled Plasma Atomic Emission Spectrometry (ICP-AES) and inductively coupled plasma-mass spectrometry (ICP-MS). The high concentrations of arsenic and selenium, and their close correlations with lithophile elements are caused by coal combustion. The results are presented in the published article: Heavy coal combustion as the dominant source of particulate pollution in Taiyuan, China confirmed by high concentrations of arsenic and selenium in PM10 (Science of the Total Environment, 2006, 370, 409-415). In addition, the bulk analysis also revealed the high mass concentration and enrichment of manganese in Guiyang, which was consistent with the individual particle analyses.

3. Size-resolved analysis: samples were collected by Electric Low Pressure Impactor (ELPI). Water soluble inorganic components were analyzed by ion chromatography (IC) and ICP-MS. The results are summarized in the manuscript: Characteristics of water-soluble inorganic chemical components in size-resolved airborne particulate matters in Sheffield, UK. It concludes that, in Sheffield airborne particulate matters, water-soluble components are a mixture of secondary aerosols and sea-salt aerosols, the extent of which depends on the origins of the air mass.

The PM10 concentrations in Taiyuan and Guiyang were many times the guideline given by WHO (50 g m-3 for 24-hour mean) while the concentrations in Sheffield were far below this value and close to the guideline for annual mean (20 g m-3).

iv

LIST OF PAPERS

Paper 1: Chemical characterization of individual particles (PM10) from ambient air in Guiyang City, China. RuiKai Xie, Hans Martin Seip, John Rasmus Leinum, Turid Winje and Jing Song Xiao Science of the Total Environment. 2005, Vol. 343 (1-3), p. 261-72.

Paper 2: Individual particles (PM10) characterization in Taiyuan City, China.RuiKai Xie, Hans Martin Seip, Daisheng Zhang, Li Liu.

Paper 3: Heavy coal combustion as the dominant source of particulate pollution in Taiyuan, China confirmed by high concentrations of arsenic and selenium in PM10.RuiKai Xie, Hans Martin Seip, Grethe Wibetoe, Showan Nori, Cameron William Mcleod.Science of the Total Environment. 2006, Vol. 370, p. 409-415.

Paper 4: Characteristics of water-soluble inorganic chemical components in size-resolved airborne particulate matters - Sheffield, UK RuiKai Xie, Cameron William Mcleod, Marie Jo Schofield, David Anderson, Hans Martin Seip, Grethe Wibetoe, Kevi A. Jackson, Jan Erik Hanssen. Submitted to Atmospheric Environment.

v

ABBREVIATIONS

AQG: Air Quality Guideline EC: Elemental Carbon EDS: Energy-Dispersive Spectrometer EF: Enrichment factor ELPI: Electrical Low-Pressure Impactor IC: Ion Chromatography ICP-AES: Inductively Coupled Plasma-Atomic Emission Spectrometry ICP-MS: Inductively Coupled Plasma-Mass Spectrometry NAAQS: National Ambient Air Quality Standards OC: Organic Carbon PAH: Polycyclic Aromatic Hydrocarbon PBA: Primary Biogenic Aerosol PMs: Particulate Matters SEM-EDS: Scanning Electron Microscope-Energy-Dispersive Spectrometer SEPA: State Environmental Protection Administration SOA: Secondary Organic Aerosol TSP: Total Suspended Particulates US-EPA: US-Environmental Protection Agency VOCs: Volatile Organic Compounds WHO: World Health Organization

vi

Table of Contents

ACKNOWLEDGEMENTSABSTRACTSLIST OF PAPERS ABREVIATIONS

1. Introduction…….……………………………………………………..12. Review of Literature……………………………………….…………2

2.1 Characteristics of urban PMs…………………………………………….…..2 2.1.1 Size and size fractions of PMs……………………………………….2 2.1.2 Chemical composition and sources of urban PMs……………………...4

2.1.2.1 Primary PMs…………………………………………………...4 2.1.2.2 Secondary PMs………………………………………………...7

2.2 Concerns of PMs……………………………………………………………....9 2.2.1 Effects on Health……………………………………………………….9 2.2.2 Effects on climate change……………………………………………..11 2.2.3 Effects on vegetation………………………………………………….12 2.2.4 Effects on materials…………………………………………………...13

3. PM regulations………………………………………………………14 3.1. USA standards……………………………………………………………….14 3.2 WHO guidelines……………………………………………………………...15

4. Objectives of Study………………………………………………….17 5. Measurements…………………….....................................................18

5.1 Study sites and sampling….………………………………………………....18 5.2 Sample analyses………………………………………………………………19

5.2.1 Individual particulate analyses………………………………………..19 5.2.2 Bulk sample analyses………………………………………………….20

5.2.2.1 ICP-AES……………………………………………………...20 5.2.2.2 ICP-MS……………………………………………………….22

5.2.3 Size-fractionated sample analyses…………………………………….22 6. Results………………………………………………………………..25

6.1 Individual particulate analyses……………………………………………...25 6.2 Bulk sample analyses ………………………………………………………..27 6.3 Size-fractionated sample analyses ………………………………………….29

7. Discussion……………………………………………………………32 8. Summary and conclusions…………………………………………..369. Suggested research activities..………………………………………38 10. Appendices………………………………………………………….39 11. References…………………………………………………………..41 Papers………………..…………………………………………………53

vii

1. Introduction

Particulate matters (PMs), alternatively in the text referred to as particulates, aerosols, or particles, are a suspension of solid, liquid or a combination of solid and liquid particles in the air. PM originates from both anthropogenic and natural sources, and may be classified as primary and secondary pollutants. Primary particulates are emitted by sources as particles and dispersed in the atmosphere without any major chemical transformation. They are of both natural origins and anthropogenic origins. Secondary particulates are those formed in the atmosphere from gaseous pollutants. The great interest in atmospheric particles at present is due to their effects on health, climate, and to a lesser extent, materials and vegetation. Primary particles generally have effects on local scales, whereas secondary particles affect regional and much broader areas.

In health and exposure studies, particles are often classified by size fraction, or aerodynamic diameter (da), which can range from a few nanometres up to 100 micrometres (μm). Size fraction is a term that aids in the classification of particles and refers more to the physical behaviour of particles rather than their actual size. Commonly used parameters are PM10 (da < 10 μm), PM2.5 (da < 2.5 μm), PM1 (da < μm), also called fine particles, and ultrafine particles (da < 0.1 μm). Ultrafine particles have almost no mass, but the number of particles in this size category is very large. Total suspended particles (TSP), is used less frequently today, and represents particles da <100 μm. Adverse health effects have been associated with all inhalable particles, PM10,as well as the PM2.5 and the ultrafine sub-fraction of PM10. The potential for particulate matter to induce adverse health effects relates to particle size. Particles of 10 microns or less in aerodynamic diameter can be inhaled deep into the lungs where they can induce tissue damage and various adverse health effects. Particles larger than 10 microns in diameter are generally filtered out in the nasal passages, and do not enter the lungs to any great extent. The World Health Organization gives 20 g/m3 as annual guideline for health effects for PM10 (WHO, 2006). In many cities especially in developing countries, the concentrations may be much higher.

Coarse particulates are produced by mechanical processes and the fraction consists mainly of finely divided minerals and sea salt particles. Pollen and spores also inhabit the coarse particle range. Since the size of these particles is normally >2.5 μm their retention time in the air parcel is shorter than for the fine particle fraction. Fine particulates are found to consist of six major categories (U.S. EPA, 2004): sulphates, nitrates, organic carbon, elemental carbon, trace metals, and water. Sulphates and nitrates are mainly secondary particles. Organic carbon is found in both primary and secondary forms. Elemental carbon and trace metals are usually emitted as primary particles.

In this study, PM10 is chosen as the target because at present the majority of monitoring data is based on measurement of PM10, and the most recent and extensive epidemiological evidence is largely based on studies using PM10 as the exposure indicator. As an indicator, PM10 comprises the particle mass that enters the respiratory tract and includes both the coarse (PM10-2.5) and fine (PM2.5) particles considered to contribute the health effects observed in urban environments.

1

2. Review of Literature

2.1 Characteristics of urban PMs

Urban PMs comprise those from both natural and anthropogenic sources. The natural sources include suspended dust, sea sprays, volcanoes and forest fires. These natural emissions result in a relatively low background of PM concentration around the world. The major sources of anthropogenic PMs are from transport, stationary combustion, space heating, biomass burning, industries, and human-induced resuspension of dust. Urban PMs from anthropogenic sources are very complex, including both primary particles and secondary particles formed from precursor gases. The major anthropogenic PMs are often emitted within relatively small urban or industrial areas, causing hotspots of elevated concentrations of PMs and some other air pollutants. As anthropogenic PMs are often fine in size, they can be airborne longer than coarse natural PMs and travel up to thousands of kilometres thereby affecting regional background concentrations.

2.1.1 Size and size fractions of PMs

PM size is the most important determinant of the properties of particles and it has implications on formation, physical and chemical properties, transformation, transport, and removal of PMs in the atmosphere. PM size is normally given as the aerodynamic diameter, which is defined as the diameter of a spherical particle with an equal gravitational setting velocity but a material density of 1 g cm-3 (U.S. EPA, 2004). Thedefinitions of coarse and fine may vary in literatures (Willeke and Baron, 1993; Seinfeld and Pandis, 2006). Most often the divisions between ultrafine, fine, and coarse particles are < 0.1 μm, < 1 μm, and > 1 μm. The rationale behind this classification is that 1 μm constitutes a natural division between particles from combustion processes and particles from mechanical processes (Morawska and Zhang, 2002). Still this definition is somewhat arbitrary, as nature does not provide a perfect division.

The distribution of particles with respect to size is an important physical parameter governing particle behaviour. Different approaches or conventions are used to classify the size of particles: (1) modes, based on the observed size distributions and formation mechanisms; (2) dosimetry or occupational health sizes, based on the entrance into various compartments of the respiratory systems; and (3) cut points, usually based on the 50% cut point of the specific sampling device, including legally specified, regulatory cut points for air quality standards (U.S. EPA, 2004).

Modes: The mass size distribution of PMs in the urban area is conventionally characterised by three modes: nucleation mode, accumulation mode, and coarse mode (Meng, 1994). Typical size distributions for an urban area and a freeway-influenced urban area are presented in Fig.1. The smallest of these modes, below 0.1 μm in diameter, is called the nucleation mode (or Aitken mode) and is formed by condensation of hot vapour from combustion sources and from chemical conversion of gases to particles in the atmosphere. Particles of this size have a high chance of deposition in the gas-exchanging (alveolar) part of the lung; they are relatively short-lived and grow into larger particles. This size range can be detected only when fresh emission sources are close to a measurement site or when new particles have been recently formed in the atmosphere (Whitby, 1978; Chow, 1995). Conversely, a

2

nucleation mode in the measured atmospheric particle distribution indicates local sources of several particle components (Pakkanen et al., 2001).

Fig.1. Volume (also mass) size distribution, measured in an urban and a freeway-influenced urban area (U.S. EPA, 2004).

The accumulation range particles consist of those with diameters between 0.1 and about 1 μm. They result from the coagulation of smaller particles from combustion sources, the condensation of volatile species, gas-to-particle conversion. These particles remain suspended for up to several weeks in the air, and are not readily removed by rain. The nucleation and accumulation ranges constitute the “fine particle size fraction”, and the majority of sulphuric acid, ammonium bisulphate, ammonium sulphate, ammonium nitrate, and organic and elemental carbon is found in this size range (Temesi et al., 2001; Bardouki et al., 2003).

The coarse mode comprises particles greater than about 1 μm in diameter. These PMs are generally formed by mechanical break-up, including wind-blown dust and soil, particles from construction and sea spray. The size means they remain in the air for relatively short periods. They make a disproportionate contribution to PM10 mass (relative to their numbers) when they are measured close to their sources. Note that a particle of 1 μm has a mass 1 million times heavier than does a particle of 10 nm (assuming the same density)!

3

Dosimetry or occupational health sizes: The occupational health community has defined size fractions in terms of their entrance into various compartments of the respiratory system. This convention classifies particles into inhalable, thoracic, and respirable particles according to their upper size cuts. Inhalable particles enter the respiratory tract, starting with the head airways. Thoracic particles travel past the larynx and reach the lung airways and the gas-exchange regions of the lung. Respirable particles are a subset of thoracic particles that are more likely to reach the gas-exchange regions of the lung.

Cut points: Another size fraction is usually specified by the 50% cut point size; e.g., PM2.5 refers to particles collected by a sampling device that collects 50% of 2.5 μm particles and rejects 50% of 2.5 μm particles. Size-selective sampling are used to measure particle size fractions with some special significance (e.g., health, source apportionment, etc.), to measure mass size distributions, or to collect size-segregated particles for chemical analysis. Dichotomous samplers split particles into smaller and larger fractions that may be collected on separate filters. Cascade impactors use multiple size cuts to obtain a distribution of size cuts for mass or chemical composition measurements. One-filter samplers with a variety of upper size cuts are also used, e.g., PM2.5, PM10.

2.1.2 Chemical composition and sources of urban PMs

2.1.2.1 Primary PMs

Primary particulate matter is generated by a variety of physical and chemical processes. It is emitted to the atmosphere through combustion, industrial processes, fugitive emissions and natural sources. These PMs include mineral dust, sea salt, biogenic particles, and combustion sources.

Soil dust: Soil-derived PMs are produced by aeolian erosion mainly in the arid and semi-arid regions. Natural or human-induced climatic changes, which alter wind velocity and precipitation, influence emission strength and locations. Soil PMs play an important role in tropospheric chemical reactions. Soil particles interact via heterogeneous chemical reactions, with sulphuric acid (Wall et al., 1988; McInnes et al., 1994; Pakkanen et al., 1996) and nitric acid (Harrison and Kitto, 1990) and thus influence particle acidity and gas phase composition.

The size distribution is typically bimodal in nature, with maxima reported at 1-10 μm and 50 μm (Gillette, 1980). More than 90% of the mass is greater than 1 μm in diameter (Gillette, 1980; Houck et al., 1989b). Large particles with a diameter greater than 30 μm settle down near the source regions, while relatively smaller particles have a lifetime of several days to weeks and can be transported up to several thousand kilometres (Zhang and Carmichael, 1999). Soil dust is a major source of coarse aerosol particles. Its annual global emission amounts to ~1,000-2,150 Tg (IPCC, 2007). Soil dust is distinct to other sources by the crustal signatures such as silicon, aluminium, iron, sodium, potassium, calcium and magnesium. Actual composition may vary from site to site; however, silicon, aluminium, iron and calcium are typically used as key indicators of soil-derived materials.

4

Urban areas can be affected by the natural sources, the strength of which depends on the sources, the proximity, and meteorological conditions. In addition, soil particles can also be resuspended by human activities, such as transport, construction, and industry. Resuspension of road dust is a major source of PMs in Europe (Kukkonen et al., 1999; Harrison, 2004). The composition of resuspended dust may be altered by enriched pollution-induced elements in surface soil (Swietlicki et al., 1996).

Sea salt: Sea salt particles are produced at the ocean surface by bursting of air bubbles. The emission of sea-salt particles is about 5900 Tg yr-1, more than 20 times the combined emissions of organics, black carbon, sulphate, nitrate and ammonium into the atmosphere (Raes et al., 2000). Sea salt aerosols are largely comprised of sodium chloride and hence very hygroscopic and readily change size with change of ambient relative humidity. The fresh and aged sea-salt particles play an important role in affecting the global climate. Sea-salt can react with acidic gases to form sodium sulphate and soldium nitrate and release chloride to the gas phase (Pakkanen, 1996; Zhuang et al., 1999a).

Sea salt is the dominant aerosol mass component in the remote marine surface air and occasionally a significant one over the continents. Sea salt concentrations in marine regions depend strongly on the state of the sea surface, which is in turn a function of meteorological conditions, especially wind speed. Sea salt particles are usually characterised by three modes: the Aitken, the accumulation mode and coarse mode (Fitzgerald, 1991). The sea salt aerosols size distribution in urban areas, though, is very variable depending on the distance to the sea. Sea salt particles can travel hundreds of kilometres. Air masses transported over continental areas are effectively depleted in relatively large sea salt particles (Kerminen et al., 1999). In urban particulates, an important feature is the so called “chloride depletion” in which chloride is lost in the sea salt particles due to the interaction between sea salt particles and acidic components such as sulphuric and nitric acids. This phenomenon has been observed in many studies (e.g. Ten Harkel, 1997; Yao et al., 2003; Fu and Watanabe, 2004).

Biogenic particles: Biogenic aerosols are ubiquitous in the Earth’s atmosphere, where they influence atmospheric chemistry and physics, the biosphere, climate, and public health. Biogenic aerosols can be from primary and secondary sources. Primary biogenic aerosols (PBA) are emitted directly from the biosphere to the atmosphere. They consist of many different particles, including pollens, bacteria, spores, virus, algae, protozoa, fungi, fragments of leaves, excrement and fragments of insects (Simoneit and Mazurek, 1982; Artaxo, 1995). PBA particles range from millimetres down to tens of nanometres in size. The secondary sources are particles formed by gas-to-particle conversion of gaseous organic matter released from biosphere. These particles are often sub-micrometre in size. The biogenic aerosol particles, even in urban areas, sometimes can be as much as 50% in mass.

Forest is a major natural global source of aerosols (Artaxo, 1995, and reference therein). Volatile organic compounds (VOCs) react with atmospheric oxidizing agents (O3, OH and NO3 radicals) to form secondary organic aerosols (SOAs). Over 90% of the total VOCs entering the atmosphere are biogenic (Greenberg et al., 1999). Biogenic aerosols of marine origin also contribute to the global biogenic aerosol loading. O’Dowd et al. (2004) reported biogenic organic matters enriched in the oceanic surface layer being transferred to the atmosphere by bubble-bursting process.

5

Characterisation of biogenic aerosols and differentiation of biogenic aerosols from other organic aerosols have been conducted by Lin and Lee (2004) by analysing n-alkanes in Kaohsiung, China. They found that biogenic aerosols accounted for 13% and 17% for PM1 and PM1-10 respectively. In another study, Tanner et al. (2004) made 14C measurements in Tennessee. They found 50-90% of organic fraction of PM10 orTSP aerosols was derived from biogenic sources.

Combustion particles: Combustion processes result in generation of large number of particles and gaseous products. Combustion sources can be classified into stationary (industrial plants, power plants, households etc) and mobile sources (mainly motor vehicles). Primary particles emitted directly from combustion processes consist mainly of carbonaceous and fly ash particles.

Combustion PMs size distribution is multimodal. The finest particles are produced by gas-to-particle conversions and form the nuclei or nanoparticles. These PMs grow by coagulation and grow into the “accumulation” mode. The larger supermicron particles are produced from the inorganic material in the fuel. The emissions depend on the composition of the fuels, the combustion conditions, and effectiveness of cleaning devices (Lighty et al., 2000).

Diesel exhaust typically consists of particles in the 5-50 nm diameter range. This mode usually consists of volatile organic and sulfur compounds that form during exhaust dilution and cooling, and may also contain solid carbon and metal compounds. The nuclei mode typically contains 1-20% of the particle mass and more than 90% of the particle number (Kittelson, 1998). Fig. 2 shows idealized diesel exhaust particle number and mass weighted size distributions. The distributions shown are trimodal,

Fig. 2. Typical engine exhaust size distribution for mass and number (Kittelson 1998).

lognormal in form. Most of the particle mass exists in the so-called accumulation mode in the 0.1-0.3 μm diameter range. This is where the carbonaceous agglomerates and associated adsorbed materials reside. Unleaded gasoline powered vehicles emit particles containing carbonaceous spherical submicrometre agglomerates smaller than diesel particles (Concawe Report, 1998).

6

Fly ash is the finely divided mineral residue resulting from the combustion of coal. It consists of inorganic, incombustible matter present in the coal that has been fused during combustion into a glassy, amorphous structure. It consists mostly of (in the decreasing order) oxygen, silicon, aluminum, calcium, iron, carbon, potassium, magnesium, sulfur, and some other minor and trace elements (Vassilev and Vassileva, 2005, and references therein). Combustion processes also result in emission of precursors (SO2, NOx, and hydrocarbons) for secondary aerosols (sulfate, nitrate, and organic aerosols). Coal combustion can also produce acid aerosols such as H2SO4depending on the partitioning of sufur between SO2 and SO3 and on the temperature and humidity (Lighty et al., 2000).

Trace elements: Trace elements compose a very small fraction in terms of mass concentration, but concern for toxicological and environmental effects has raised interest among the scientific community. Trace metals, such as V, Cu, Fe, and Pt, can catalyze the generation of reactive oxygen species that have been associated with direct molecular damage (Halliwell and Gutteridge, 1999). Trace metals can also catalyse secondary aerosol formation (Grgi and Ber i , 2001). Trace metals are released to the atmosphere during combustion of fossil fuels and wood, as well as during high temperature industrial processes and waste incineration. Natural emissions result from a variety of processes acting on crustal minerals, including volcanism, erosion and surface winds, as well as from forest fires and oceans (Allen et al., 2001).

Combustion of fossil fuels is the principal anthropogenic source of Be, Co, Hg, Mo, Ni, Sb, Se, Sn and V, as well as a large contributor of As, Cr, Cu, Mn and Zn, while industrial metallurgical processes produce the largest emissions of As, Cd, Cu, Ni, and Zn (Pacyna, 1998, and references therein). Trace elements are found in almost all aerosol size fractions, and their concentrations and size distributions are controlled by the nature of emissions to the atmosphere (Wang et al., 2005, and references therein). Combustion and industrial sources of trace metals are mainly in ultrafine mode, while resuspended dust and mechanical wear sources are mainly in coarse mode (Allen et al., 2001). During coal combustion, trace elements are partially or fully vaporized, and tend to condense on the submicron particles with a large surface-to-volume ratio. This can explain the enrichment of trace elements on submicron fly ash particles (Xu et al., 2003, and references therein).

Elevated trace element concentrations are found in many industrial and urban areas (e.g. Bilos et al., 2001; Balasubramanian and Qian, 2004; Wang et al., 2005; Karanasiou et al., 2007). When it comes to the identification of natural and anthropogenic sources, it is often useful to consider the enrichment factor (EF), i.e. the double ratio between a specific trace element and a typical crustal element (as Ti or Al) in the particles and in the earth crust (see p413, Paper 3). Elements with EF values significantly higher than unity may suggest anthropogenic sources. Some elements are often used as tracers for some typical fuel types. V and Ni are often deemed to be from residual oil combustion (Samet et al., 1996; Dreher et al., 1997; Watkinson et al., 1998). As and Se are fingerprints of coal combustion (Nriagu, 1989; Harrison et al., 1996; Park and Kim, 2004).

2.1.2.2 Secondary PMs

7

Secondary PMs are formed in the atmosphere by mass transfer to the aerosol phase of low vapour pressure products of oxidation of gases. The precursor gases include sulphur dioxide, nitrogen oxides, and organic vapours, and the oxidation products are sulphate, nitrate, and organic matters. In the presence of ammonia, sulphate and nitrate will be in the forms of ammonium sulphate and ammonium nitrate.

Sulfate: Sulfate in ambient air mainly comes from oxidation of sulphur-containing precursors such as SO2, H2S, CS2, COS, DMS (Xiu et al., 2004), among which SO2 isthe largest contributor (Khoder, 2002). SO2 can be oxidized to H2SO4 by gas-, aqueous-, or multi-phase reactions with OH, H2O2, and O3, followed by condensation or nucleation of H2SO4 both onto pre-existing particles and into new particles with partial or full neutralization by NH3.

Sulfate is known to be a typical long-range transported component having similar concentration over large areas (Harrison and Jones, 1995). Sulphate is often found in the fine mode (<1 μm) in atmosphere (Zhuang et al., 1999b; Parmar et al., 2001; Temesi et al., 2001). The atmospheric life time of fine sulphate is in the order of 5 days (Langner and Rodhe, 1991), during which it interacts with ambient NH3. It is first partially neutralized as NH4HSO4, which is further neutralized as (NH4)2SO4. Some minor fraction of sulphate may be found in coarse mode (>1 μm) where it is present in soil and/or sea-salt particles (Wall et al., 1988; Zhuang et al., 1999b).

Nitrate: Fine nitrate particulate is formed by homogeneous gas-phase oxidation of nitrogen oxides to gaseous nitric acid, which is followed by the reaction with gaseous ammonia to form highly volatile NH4NO3 (Ocskay et al., 2006). The emission of nitrogen oxides from traffic is the most important precursor for nitrate aerosol. Like sulphate, nitrate is normally distributed in fine mode in the form of ammonium nitrate (Zhuang et al., 1999b; Parmar et al., 2001; Temesi et al., 2001). However, in areas unfavourable for ammonium nitrate formation, such as regions of high SO2 to NH3emission ratios over land and marine environment, nitrate can be found mainly in coarse mode (>1 μm), combined with soil and/or sea-salt particles (Savoie and Prospero, 1982; Wolff, 1984; Pakkanen, 1996).

Ammonium: Ammonium is often accompanied by sulphate and nitrate as the result of neutralization in the atmosphere. Sulphate and nitrate are rivals competing for ammonia to form the corresponding salts. Ammonium nitrate emerges only after sulphate is nearly fully consumed (Seinfeld and Pandis, 2006). Among the inorganic secondary aerosols, ammonium sulphate is the most stable species. The seasonal variation of ammonia emissions differs regionally depending on farming practice and climate conditions. In Europe, 80-95% is from agriculture (Van der Hoek, 1998). When combined with sulphate and nitrate, it resides with the accompanying pairs in the fine mode. It may occur in coarse mode, but the mechanism is different from that in fine mode. Zhuang et al. (1999b) suggested that coarse mode ammonium was the product of reactions of ammonia gas with sulphuric or nitric acid pre-existing on the coarse particles.

Other inorganic secondary aerosols may be encountered, such as ammonium chloride in industrial areas where chloride is emitted from incinerators (Kaneyasu et al., 1999) and coal combustion (Xiu et al., 2004), and where chloride is produced from the reaction of acids with sea-salt (Jonson et al., 2000).

8

Secondary organic aerosol (SOA): SOA is formed in the ambient atmosphere through oxidation of volatile organic compounds (VOCs) and subsequent partitioning of some oxidation products into the aerosol phase. The pathways for SOA formation can be gas-phase oxidation, in which VOCs are oxidized by species such as the OH, O3,and NO3 (Seinfeld and Pandis, 2006), and/or aqueous-phase oxidation in which SOA is formed in cloud and fog droplets (Blando and Turpin, 2000).

The contributions of primary and secondary components of aerosol organic carbon can be quantified. Tracer compounds for either of them have been used (Gray et al., 1986; Turpin and Huntzicker, 1991), as have models describing the formation of SOA (Pandis et al., 1992; Bowman et al., 1997). Worth mentioning is the use of elemental carbon (EC) tracer. EC, predominantly coming from combustion, has often been used as a tracer for primary organic carbon. The underlying hypothesis is that because EC and primary OC often have the same sources, there exists a representative ratio of primary OC/EC for a given area. By measuring the total OC and EC, the SOA can be calculated. Of critical importance is the choice of the OC/EC ratio. This issue is addressed by Strader et al. (1999).

2.2 Concerns of PMs

2.2.1 Effects on health

Epidemiological evidence

Scientific evidence of the health effects of particulate air pollution has been shown in studies of short-term exposure. In several cities throughout the world, consistent associations have emerged between daily mortality and ambient concentration of PM during the same or the previous few days. The results of two collaborative projects conducted in 90 cities in the USA (National Morbidity Mortality Air Pollution Study; NMMAPS) and in 29 cities in Europe (Air Pollution Health Effects Approach; APHEA-II) have been reported. In the American cities, where annual average concentrations of PM10 ranged 23–46 μg m-3, a 0.27% increase in total mortality and a 0.69% increase in cardiorespiratory mortality were detected for a 10 μg m-3 increase in PM10 (Dominici and Burnett, 2003). In the European study, based on the most extensive database available in Europe and covering a large range of PM10concentrations, the risk estimate for overall mortality was 0.6% per increase of 10 μg m-3 in PM10 (Katsouyanni et al., 2001) and was 0.76% per 10 μg m-3 PM10 for cardiovascular mortality.

The evidence that mortality increases due to long-term, low-level exposure to PM was provided by the Harvard Six Cities Study (Dockery et al., 1993). These findings were confirmed in the long-term follow-up (1982–1998) of the American Cancer Society (ACS) II cohort, consisting of ca. 500,000 adults from metropolitan areas throughout the USA (Pope et al., 2002). In the latter study, each 10 μg m-3 elevation in PM2.5 was associated with approximately a 6, 9 and 14% increased risk of all-cause, cardiopulmonary and lung cancer mortality, respectively. A recent report from Los Angeles (Jerrett et al., 2005), which included a large proportion of the ACS II cohort from that area, indicated that a more refined method for assessing exposure produces a higher risk estimate of mortality increase (17% increase; 95% confidence interval

9

5–30%) for an increase of 10 μg m-3 in PM2.5. The evidence of chronic effects has also accumulated for morbidity data, which indicate that the respiratory effects of long-term exposure include a decrease in lung function and signs of atherosclerosis progression (Janssen et al., 2003; Kunzli et al., 2005).

Epidemiological studies can assess real exposure, and have been given the greatest weight in setting standards for airborne particles, but have so far contributed little to the understanding of what PM constituents that may be detrimental to human health. The studies must be complemented by another approach: toxicological studies. In a toxicological study it is possible to evaluate directly the influence of various factors on the response without the complication of confounding variables inherent in the epidemiological study. By allowing precise control over critical variables, it can establish cause-effect relationship for specific particle characteristics and materials, and can be used as a basis to develop actual dose-response profiles.

Toxicological plausibility

Toxicological studies on PM provide critical information for PM health effects. Toxicological evidence is complementary to the observational findings of epidemiological studies, providing the framework for assessing the biological plausibility of observed associations. Studies designed to address the dose–response relationships can also facilitate the interpretation of exposure–response modeling of epidemiological data. Much toxicological research is now directed at identifying those characteristics of PM that determine toxicity, so that the most important PM sources for control can be identified.

Size mode: Epidemiological studies revealed that coarse particles are not associated with mortalities (Pope et al., 1999; Schwartz et al., 1999), but are associated with fine particles (Dockery et al., 1993). The reason behind this is that fine particles are deposited in the lung rather than in the upper airway; fine particles have greater surface area per unit mass, which enhances solubility; and fine particles can enter cells more readily and can be transported from lung to other organs (Lighty et al., 2000).

Carbonaceous materials: A large fraction of PM10 is derived from carbonaceous materials: EC and OC. Either or both may contribute to health effects. Both components were associated with changes in brachial artery diameter in young healthy adults (Brook et al., 2002; Urch et al., 2004). Polycyclic Aromatic Hydrocarbon (PAH) in urban PM has been implicated in PM-related mutagenicity (Somers et al., 2004).

Sulphate and nitrate: Toxicological studies in both human and animals demonstrated that exposure to strong acidic sulphate particles caused an alteration in lung function and particle clearance rates (Leikauf et al., 1984; Bauer et al., 1988). However, these effects were not observed following exposure to the weakly acidic ammonium sulphate particles (Utell et al., 1982; Schlesinger, 1989). The absence of health effects following exposure to ammonium sulphate leads to the hypothesis that the association of sulphate with health effects in the epidemiological studies was misleading, and sulphate merely served as a surrogate of H+ (Lippmann, 1989). Similarly, inhaled nitrate salts (sodium nitrate and ammonium nitrate) showed no health effects (Loscutoff et al., 1985; Cassee et al., 2002). However, acidic nitrate (nitric acid vapour) exposure did produce various effects on pulmonary functional and

10

lung defence parameters (Aris et al., 1993; Wong et al., 1996). The conclusion is that acidity, rather than sulphate or nitrate is causing the health effects.

Trace metals: The role of transitional metals in producing adverse health effects is based on their potential for oxidative activity and the production of reactive oxygen species. Sorensen et al. (2005) discovered a correlation between V and Cr of fine particulate and oxidative damage to DNA. The role of specific metal content or metal valence state to the ranking of transitional metal toxicity is still unknown. Zn in PM may be responsible for various pulmonary effects, such as inflammation, necrosis, and airway hyperreactivity (Adammson et al., 2000; Dye et al., 2001). However, Harrison and Yin (2000) argue that trace metal exposure in urban PM is insufficient to cause toxic effects through classic mechanisms of toxicity, but rather the non-classic mechanisms such as Fenton reaction in which trace metals play a catalytic role, contribute to the production of hydroxyl radicals.

Crustal materials: Studies using volcanic ash, isolated from organic materials associated with soil dust normally found in ambient PM, indicate that such particles are relatively inert and generally non-inflammatory (U.S. EPA, 1996). Kobzik et al. (2001) found that the degree of brochoconstriction in mice was associated with the concentration of Si. However, this may not necessarily mean that Si per se is toxic, but that components of ambient PM, whose concentrations vary with that of Si, may be toxic. Veranth et al. (2004) observed that crustal-related PM, such as kaolin clay and aluminium oxide, showed no sign in inducing proinflammatory cytokine in vitro.

Biogenic materials: PM contains pollens, molds, spores, and biological toxins, such as bacterial endotoxine. A large fraction of biogenic PM is from biogenic-derived OC. Positive correlations between mortality rates and pollen concentration were found, suggesting that pollen-associated acute exacerbation of allergic inflammation may cause death among some compromised individuals (Brunekreef et al., 2000). Increases in hospitalizations for asthma have also been reported to be correlated with pollen exposure in Mexico City (Rosas et al., 1998) and London (Celenza et al., 1996). Studies have found relationships between exposure to fungi and their byproducts in respiratory illnesses and immune pathology (Tuomi et al., 2000; Yang and Johanning, 2002).

Somewhat surprisingly, epidemiological studies have revealed that for every 10 μg m-3 increase in PM10 concentration, there is a ~ 0.5% increase of daily mortality, irrespective of geographic locations where the physicochemical properties of ambient PM might differ (Samet et al., 2000; Katsouyanni et al., 2001; Cohen et al., 2004; HEI International Oversight Committee, 2004). The consistency of this finding across different communities suggests that it is mainly the mass concentration of the particle mix, rather than specific chemical species within the mix, that governs the association between particulate pollution and health effects. However, there still remain some unanswered questions when it comes to the relative contribution to health effects of specific properties of particles and, therefore, the importance of specific sources.

2.2.2 Effects on climate change

PMs have a direct radiative forcing. They scatter and absorb solar and infrared radiation in the atmosphere. Specifically, sulphate, nitrate, organic carbon, and mineral dust have cooling effects, with sulphate presenting a most negative direct radiative forcing; black

11

carbon has a positive radiative forcing (IPCC, 2007). PMs also alter cloud formation processes by increasing droplet number concentrations and ice particle concentrations. Particles decrease the precipitation efficiency of warm clouds and thereby cause an indirect radiative forcing associated with these changes in cloud properties, and the total net effect of all aerosols is negative (see Fig. 3). PM has significantly offset the warming effect caused by greenhouse gases.

Fig. 3. Principal components of radiative forcing of climate change. The values represent the forcing in 2005 relative to the start of the industrial era (about 1750). Positive forcings lead to warming of climate and negative forcings lead to a cooling (IPCC, 2007).

2.2.3 Effects on vegetation

PM deposited from atmosphere to foliar surfaces may reside on the leaf, or bark surface for extended periods. Any PM deposited on above-ground plant parts may exert physical or chemical effects. The effects of “inert” PM are mainly physical; whereas those of toxic particles are both chemical and physical. The effects of dust deposited on vegetation are more likely to be associated with their chemistry, rather than their mass and chemical effects may be more important than any physical effects (Farmer, 1993).

PM can cause physical and chemical effects. Deposition of inert PM on above-ground plant organs sufficient to coat them with a layer of dust may result in changes in radiation received, a rise in leaf temperature, and the blockage of stomata. Deposition of particles increased leaf temperature and contributed to heat stress,

12

reduced net photosynthesis, and caused leaf chlorosis, necrosis, and abscission (Guderian, 1985). Limestone dust showed detrimental signs on plant photosynthesis, which could be explained by both reduced stomatal conductance and increased pH caused by limestone PM (Vardaka et al., 1995). Crust formation reduced photosynthesis and the formation of carbohydrates needed for normal growth, induced premature leaf-fall, damaged leaf tissues, inhibited growth of new tissue, and reduced starch storage. Permeability of leaves to ammonia increased with increasing dust concentrations and decreasing particle size (Farmer, 1993). Stomatal clogging by PM from automobiles, stone quarries, and cement plants was also studied by Abdullah and Iqbal (1991).

The chemical composition of PM is usually the key phytotoxic factor leading to plant injury. In experimental studies, applications of cement-kiln dust of known composition for 2 to 3 days yielded dose-response curves between net photosynthetic inhibition or foliar injury and dust application rate (Darley, 1966). Alkalinity was probably the essential phytotoxic property of the applied dusts. Sea-salt particles contain significant amounts of sulfate, sodium, chloride, and trace elements (as well as living material) in the atmospheric aerosol that impact coastal vegetation. Foliar accumulation of airborne salt particles may lead to foliar injury, affecting the species composition in coastal environments (Smith, 1984).

2.2.4 Effects on materials

PMs have impacts on materials such as metals, wood, stone, painted surfaces, electronics and fabrics. PMs may exert a corrosive effect due to acidic aerosols and to the catalytic roles of heavy metals and carbonaceous particles in formation of sulphuric and nitric acid (Zappia et al., 1993; Sabbioni, 1996). The deposition of PM on historical buildings and materials can cause soiling and discolouration (Mansfield et al., 1991), thus reducing their aesthetic appeals. With the decreasing levels of SO2 in urban areas the relative importance of NOx and airborne particles will increase.

13

3. PM regulations

3.1 USA standards

In 1970, the enactment of the Clean Air Act in the United States set the stage for the promulgation of the first National Ambient Air Quality Standards (NAAQS) (U.S. Code, 1970). This act called on the then newly established U.S. EPA to review the science on public health and air pollution and to establish NAAQS sufficient to “protect the public health with an adequate margin of safety”. In 1971, the U.S. EPA promulgated the first NAAQS for PM using TSP as the measure. EPA set a 24-h average standard of 260 g m-3, allowing 1 exceedance per year and an annual average standard of 75 g m-3.

While implementation of these standards was proceeding during the 1970s, scientists addressed continuing questions and made progress toward better understanding the health effects of PM. At the same time, the 1970s witnessed major scientific advances with the initiation of the Harvard Six Cities Study (the first prospective cohort study designed to assess the relationship between air pollution and health). Also, emerging dosimetry studies began to clarify that not all forms of total suspended particles (TSP) were likely to penetrate the lower reaches of the respiratory system and bringing attention to PM10, the “inhalable fraction” of TSP (Greenbaum et al., 2001). Because of these emerging studies, the 1980s became a period during which the health and regulatory focus in the United States shifted from TSP to PM10. In 1984, the U.S. EPA proposed the first PM10 NAAQS. The U.S. EPA promulgated the first PM10 standards in 1987, setting the 24-h average standard at 150 g m-3 with 1 exceedance allowed per year, and the annual average standard at 50 g m-3 (U.S. EPA, 1986).

In the 1990s, numerous epidemiological studies taking advantage of two new research designs were conducted. The first, so-called time-series studies, compared daily changes in measures of PM with daily changes in mortality and morbidity; a large number of studies conducted by different investigators in North America and Europe found a relatively consistent, albeit small, increase in relative risk associated with elevated levels of ambient PM (U.S. EPA, 1996). The second type, cohort studies, made use of extensive longer term health and socioeconomic information about large numbers of study subjects; this enabled risks of mortality to be compared with ecological data on the levels of air pollution in the city where the subjects lived. These studies, the Harvard Six Cities Study (Dockery et al., 1993) and the American Cancer Society Study (Pope et al., 1995), both found significant differences in relative risk of premature mortality between those living in the most and least polluted cities. After a contentious public review process for proposed new standards, the U.S. EPA in July 1997 retained the 1987 PM10 standards and promulgated a new set of NAAQS for fine particles, measured as PM2.5 (annual average of 15 g m-3 with 7 exceedances per year and 24-h average of 65 g m-3) (U.S. EPA, 1997).

In September 2006, U.S. EPA revised the 1997 standards (U.S. EPA, 2006). The revised 2006 standards address two categories of particle pollution: fine particles (PM2.5), which are 2.5 micrometers in diameter and smaller; and inhalable coarse particles (PM10) which are smaller than 10 μm and larger than 2.5 μm. The 2006 standards tightened the 24-hour fine particle standard from 65 g m-3 to 35 g m-3, and

14

retained the current annual fine particle standard at 15 g m-3. EPA has decided to retain the existing 24-hour PM10 standard of 150 g m-3. Due to a lack of evidence linking health problems to long-term exposure to coarse particle pollution, the Agency has revoked the annual PM10 standard. The Agency selected the levels for the final standards after reviewing thousands of peer-reviewed scientific studies about the effects of particle pollution on public health and welfare.

3.2 WHO guidelines

The first edition of the WHO Air quality guidelines (AQGs) for Europe was published in 1987. In the 1987 edition (WHO, 1987), sulfur dioxide and PM were treated jointly. Short-term (24-hour average) guideline values were derived for combined exposure to sulfur dioxide and particulate matter, expressed in “black smoke”, “total suspended particulates” and “thoracic particles”. Long-term (one-year average) guideline values were derived only for sulfur dioxide and black smoke. At the time, published studies were inadequate to develop a guideline for thoracic particles per se. Therefore, the guideline value for thoracic particles was based on site-specific ratios for total suspended particulates to thoracic particles, and on a single study that also involved exposure to sulfur dioxide.

In 1990s, a number of studies were published that permitted direct evaluation of the health effects of thoracic particles, because PM10, essentially equivalent to thoracic particles, was actually measured in such studies. Also, studies were published that permitted evaluation of the health effects of PM10 alone, either because exposure to other pollutants was low or because adequate adjustment was possible. In 2000, the European Centre for Environment and Health of the WHO concluded its review of the health data and published the European AQGs (WHO, 2000). In these guidelines, the WHO suggested specific limits for all pollutants except PM; the apparent lack of a threshold for effects of PM led to the WHO to put forward a concentration-response curve for PM. In its PM chapter, the WHO stated “…The available information does not allow a judgment to be made of concentrations below which no effects would be expected. Effects on mortality, respiratory and cardiovascular hospital admissions as well as other health variables have been observed at levels well below 100 g m-3,expressed as a daily average PM10 concentration. For this reason, no guideline value for short-term average concentrations is recommended either. Risk managers are referred to the risk estimates provided in the tables for guidance in decision-making regarding standards to be set for particulate matter…”.

In 2006, the WHO published its new version of AQGs (WHO, 2006). The guidelines for annual mean concentrations of PM10 and PM2.5 are 20 and 10 μg m-3,respectively; 24-hour concentrations for PM10 and PM2.5 are 50 and 25 μg m-3,respectively. The WHO new AQGs for PM are based on the studies using PM2.5 as an indicator. The PM2.5 guidelines values are converted to the corresponding PM10guideline values by application of PM2.5/PM10 ratio of 0.5. Adoption of a guideline value at 10 μg m-3 for annual mean places significant weight on the long-term exposure studies that use the American Cancer Study and the Harvard Six-Cities data (Dockery et al., 1993; Pope et al., 1995; HEI, 2000; Pope et al., 2002; Jerrett, 2005). In all of these studies, robust associations were reported between long-term exposure to PM2.5 andmortality. Adoption of a guideline value at 25 μg m-3 will protect against peaks of pollution that would otherwise lead to substantial excess morbidity or mortality.

15

Besides the guideline values, three interim targets are defined for PM2.5.These values have been shown to be achievable with successive and sustained abatement measures.

16

4. Objectives of study

The overall objective of this study is to gain more knowledge of the characteristics and sources of urban ambient particulate matters and to evaluate different techniques to achieve this knowledge.

Specifically, the objectives are

1. to reveal the microscopic structure, morphology, and chemistry by looking into the individual particles using Scanning Electron Microscopy in conjunction with Energy-Dispersive Spectrometer (SEM-EDS) and evaluate if some of the characteristic particles can be treated as physiochemical markers to relate them to the pollution sources and transformation processes.

2. to find size distributions which are characteristic for certain components by analysing size-fractionated particulate samples. The mode or modes appearing in the size distribution can indicate different mechanisms of formation and transformation for different species.

3. to measure mass concentration and composition of bulk samples collected on filters with focus on elements important for air quality assessment and epidemiological studies.

4. to obtain experience in applying state-of-the art methodologies and measurement techniques used in research of airborne particulates.

5. to contribute to better characterization of aerosol amounts and composition in the chosen cities.

17

5. Measurements

5.1 Study sites and sampling

The sampling sites are shown in Fig.4, and sampling time and study approaches are listed in Table 1.

Guiyang

Guiyang, the capital of Guizhou Province, with a population of 1.56 million, is a key industrial base in southwest China. It is located in the middle of Guizhou Province, on the eastern slopes of the Yungui Plateau. The urban area of Guiyang is a basin surrounded by high mountains. Guiyang’s economy relies on its heavy industries, such as electricity, steel, chemicals, and non-ferrous metals. The heavy industrial structure of Guiyang is the main reason for severe air pollution in Guiyang. For many years, Guiyang was labelled as one of the most polluted cities in world in terms of sulphur dioxide concentration (ca. 420 μg m-3, 8 times of the WHO 24-hour guideline of 50 μg m-3) as the result of large amount of coal consumption (World Resources Institute, 1999). The sampling site was on the roof the five-story office building of Guizhou Institute of Environmental Science in south-eastern part of the city.

Altogether 5 samples were collected in Guiyang. For detailed description of the sampling, please see Paper 1.

Meanwhile, particles were also collected for bulk sample analysis. The 90 mm diameter, 3 μm pore size Fluoropore® membrane filters were used as the substrates. 3 samples were collected.

Table 1 Sampling time and study approaches.

Sampling sites Sampling time Approaches

Guiyang, China 24-28, July 2003 Individual particle analysis

Taiyuan, China 1. 14-19, July 2003 2. 2-16, March 2004

1. Individual particle analysis 2. Bulk sample analysis

Sheffield, United Kingdom 30 10-13 November, 2006 Size-fractionated sample

analysis

Taiyuan

Taiyuan, the capital of the Shanxi Province (see map in Paper 3), with a population of 2.67 million and an area of 1,500 km2, sits in a valley, surrounded by mountains on three sides. The annual coal consumption is around 25 million tons. The huge amount of coal consumption and the characteristic topography result high concentrations of PM and SO2 (annual averages in 2005, 175 μg m-3 and 87 μg m-3, respectively) (NBS, 2006).

18

The samples were collected on a four-story building in Taiyuan University of Science and Technology. The first sampling campaign lasted from 14 to 19 July 2003. The sampling method is described in Paper 2.

The second campaign was from 2 to 16 March 2004, see Paper 3.

Fig. 4. Sampling sites. (a) Guiyang and Taiyuan in China, and (b) Sheffield in England.

Sheffield

Sheffield is a cosmopolitan city situated in central England. The city is located in a confined valley of the Rivers Don, Sheaf and Porter which drain the eastern moor land margins of the Pennies Hills, some 8 km to the west. The sampling site was on the roof of the building of the Department of Chemistry, the University of Sheffield, about 1.6 km to the west of the city centre.

Sampling details are described in Paper 4.

5.2 Sample analyses

5.2.1 Individual particulate analyses

Principles of SEM-EDS

In a typical SEM, electrons are thermionically emitted from a tungsten or lanthanumhexaboride (LaB6) cathode and are accelerated towards an anode. The electron beam, which typically has an energy ranging from a few hundred eV to 50 keV, is focused by one or two condenser lenses into a beam with a very fine focal spot sized 1 nm to 5 nm. As the primary electrons strike the surface they are inelastically scattered by atoms in the sample. Through these scattering events, the primary electron beam effectively spreads and fills a teardrop-shaped volume, known as the interaction volume, extending from less than 100 nm to around 5 μm into the surface. Interactions in this region lead to the subsequent emission of electrons and X-rays.

19

Secondary electrons are emitted from the atoms occupying the top surface and produce a readily interpretable image of the surface. The contrast in the image is determined by the sample morphology. A high resolution image can be obtained because of the small diameter of the primary electron beam.

Interaction of the primary beam with atoms in the sample causes shell transitions which result in the emission of an X-ray. The emitted X-ray has an energy characteristic of the parent element. The X-ray can be detected by energy-dispersive spectrometers (EDX). The signals detected are transformed into electronic pulses and, after amplification, stored in a multi-channel device according to the corresponding energy. X-rays may also be used to form maps or line profiles, showing the elemental distribution in a sample surface.

Sample Preparation

Samples need to be mounted on the stub before they can be analyzed. Since the SEM-EDS uses electrons to produce X-rays, the SEM-EDS system requires that the samples are electrically conductive. Teflon filter is non-conductive, and hence some measures need to be taken to remove the electrons accumulated during the bombardment of electron beam. This involves carbon tape mounting and surface carbon coating. For the detailed description of the sample preparation, readers are referred to the Paper 1.

Instrumentation and procedures

Single-particle analysis was performed using a Hitachi S-4300SE-Shotky Field Emitter Scan Electron Microscope. The S-4300 is a computer controlled high resolution field emission scanning electron microscope which produces exceptional images at both normal and low operating voltages. Energy dispersive X-ray microanalysis may be carried out on a sample at both normal and low operating voltages to determine the elemental composition. This instrument is equipped with an Oxford 6853 energy dispersive spectroscopic detector, and is run by Oxford Instruments INCA Energy and Feature software. The SEM image analysis was manually performed on a JEOL JSM-6460LV high-performance SEM. The detailed parameters’ settings for the instruments are presented in the Paper 1.

5.2.2 Bulk elemental analyses

5.2.2.1 ICP-AES

Principles of ICP-AES

ICP is designed to generate plasma, which is a gas in which atoms are present in an ionized state. The formation of the plasma is dependent upon an adequate magnetic field strength and the pattern of the gas streams follows a particular rotationally symmetrically pattern. The plasma is maintained by inductive heating of the flowing gases.

In ICP-AES, a sample solution is introduced into the core of inductively coupled argon plasma (ICP) at a temperature of approximately 8000°C. At this temperature all

20

elements become thermally excited and emit lights at their characteristic wavelengths. These lights are collected by the spectrometer and pass through a diffraction grating that serves to resolve the lights into a spectrum of its constituent wavelengths. Within the spectrometer, these diffracted lights are then collected by wavelengths and amplified to yield intensity measurements that can be converted to elemental concentrations by comparison with calibration standards.

Sample preparation

For both ICP-AES and ICP-MS, complete sample dissolution is required prior to analysis. Previously, dry ashing and conventional wet digestion at atmospheric pressure were employed. But the drawbacks are obvious – the method is time consuming and can often result in the loss of volatile analyte species.

Complete dissolution of airborne particles is achieved by digestion with a HNO3+HF mixture in a microwave oven. Generally, the addition of nitric acid is intended to destroy organic compounds and to oxidise metals, but for some silicon bound elements, such as Al, Si, Ti, Zr, HF must be added. By this treatment carbonaceous as well as siliceous materials are dissolved. The closed PTFE bomb digestion procedure is chosen since losses of volatile components are reported to be low and no separation of insoluble residues is required.

Validation of method

The SRM1648 urban particulate matter is intended primarily for use as a control material and in the evaluation of methods used in the analysis of atmospheric particulate matter. It consists of natural atmospheric particulate matter collected at an urban location. The certified values are based on measurements of 6 to 30 samples by each of the modern analytical methods. To determine the accuracy and precision of our measurement, recovery experiment was conducted by comparing the observed results with the certified or reference values.

Microwave digestion procedures

All sample digestions were accomplished by using a Milestone microwave sample preparation system, equipped with temperature and pressure sensors and controlled by a personal computer. The whole digestion procedures are presented in Paper 3.

Instrumentation and procedures

Major and minor elements such as Al, Ca, Fe, K, Mg, Na, Pb, Ti, and Zn were measured by using ICP-AES (Varian Vista AX CCD, Simultaneous ICP-AES). The operating conditions were presented in Appendix .

ICP-AES instrument was calibrated by the external calibration method. A multi-element standard solution was used to prepare three standard solutions. The multi-element standard solution consists of the following elements: Al, B, Ca, Cr, Co, Cu, Fe, Pb, Mg, Ni, P, K, Si, Na, Sn, Ti, V, Zn and Zr, with each element’s concentration at 50 ppm. Three standard solutions were prepared with elemental concentrations of 4.545 ppm, 8.333 ppm and 16.667 ppm, and 3% HNO3 acid was used during standard preparation.

21

For each element a proper analytical wavelength was chosen to determine the concentration of the element. The principle of choosing a wavelength is based on its sensitivity and more importantly its absence or the least of spectra interferences. The chosen wavelength for each element is in Appendix II.

5.2.2.2 ICP-MS

Principles of ICP-MS (Quadrupole)

ICP technology was built upon the same principles used in atomic emission spectrometry. Samples are decomposed in high temperature argon plasma and analyzed based on their mass to charge ratios. An ICP-MS can be thought of as four main processes, including sample introduction and aerosol generation, ionization by an argon plasma source, mass discrimination, and the detection system.

Once the sample passes through the nebuliser, aerosol is formed in the nebuliser chamber. The aerosol moves into the torch body and is mixed with more argon gas. A coupling coil is used to transmit radio frequency to the heated argon gas, producing an argon plasma "flame" located at the torch. The hot plasma removes any remaining solvent and causes sample atomization followed by ionization. The formed ions enter a quadrupole mass spectrometer, where the positive ions are separated from electrons and molecular species by an applied negative potential, then they are accelerated towards a quadrupole filter which allows only ions with one particular mass-to-charge ratio to reach the detector. By varying the applied potential, ions of different masses are selected to reach the detector. The detector converts the beam of selected ions to an electrical signal proportional to the concentration. Quantization is performed using calibration curves derived from known standards.

Sample preparation

Sample preparation is the same as that of ICP-AES.

Instrumentation and procedures

Trace elements such as As, Co, Mn, Ni, Sb, Se, Sn, and V were measured by ICP-MS (Perkin-Elmer SCIEX ELAN 5000). The operating conditions were set as in Appendix

. The elemental isotopes and the ratio of mass to charge (m/z) were chosen as specified in Appendix IV. The choices of the ratio were based on the isotope abundance and the spectra interferences as well. ICP-MS was also calibrated by external method. A multi-element standard solution was used to prepare three standard solutions. The multi-element standard solution consists of the following elements: As, Bi, Ga, Ge, In, Pb, Sb, Se, Sn, Te, Ti and V. The concentration of each element is 50 mg l-1. Two standard solutions were prepared with elemental concentrations of 50 μg l-1, 5 μg l-1, and 3% HNO 3 acid solution was used during standards preparation.

5.2.3 Size-fractionated sample analyses

Theory of cascade sampler

22

ELPI (Dekati Electrical Low Pressure Impactor) is a cascade low pressure impactor combined with electrical charging to classify airborne particles into different size fractions. The size classification is made from 30 nm up to 10 μm with evenly distributed impactor stages. When ELPI is used without electrical charging, it becomes LPI. The LPI utilizes principles of inertia to separate particles of different sizes. Larger particles in the air stream will impact the surface of collection stage where they can adhere, while lighter particles are carried on to later stages. LPI has 13-stages so particles are size classified into 13 size fraction inside the unit (see Fig. 5). The particles are collected on 25 mm collection substrates that are weighed before and after the measurement to obtain gravimetric size distribution of the particles.

Fig. 5. Schematic of ELPI cascade impactor sampler.

Theory of IC analysis

Ion Chromatography (IC) is the separation of inorganic and organic ionic species by ion exchange chromatography followed by suppressed conductivity detection. Anion species, such as fluoride, chloride, sulfate and phosphate, are separated on an analytical column packed with a quaternary ammonium base as the exchange functional group. Potassium, sodium hydroxide and methanol or sodium carbonate and bicarbonate would be used as effluents. Cations species, such as ammonium, potassium, sodium, calcium, magnesium, are separated on an analytical column packed with resin functionalized with carboxylate, phosphate cation-exchange sites and crown ether groups. Sulfuric acid or methanesulfonic acid is used as mobile phase.

Sample preparation

The particle loaded polycarbonate substrates were extracted in an ultrasonication bath with 15 ml Millipore water (18.2 M .cm). The supernates were transferred to scintillation bottles, ready for analyses.

23

Instrumentation and procedures

The extracts were analysed by ion chromatography (IC) for major water-soluble inorganic ions using a DIONEX 2000 instrument. For the detailed procedures for determination of anions (Cl-, NO3

-, and SO42-), see Paper 4.

24

6. Results

6.1 Individual particulate analyses

Guiyang campaign, summer, 2003 (Paper 1)

Five samples were collected in Guiyang City. Totally 2500 particles were analyzed by SEM-EDS. Cluster analysis applied to the data set yielded 17 different particle types. Manual inspection was also conducted to reveal particles’ size and morphology.

Particle morphology:

Normally, spherical particles are derived from fluid melts due to high-temperature processes. The spherical shapes of some of the particles suggest that they were emitted from smelters or from furnaces. These particles include fly ash, silicomanganese, and iron particles. However, some of the particles from high-temperature processes were not well-shaped, such as silicomanganese slag, coal burning, and quartz particles.

Particle chemistry:

Chemical composition analyses revealed high abundance of particle types associated with metallurgical industry. These particle types included silicomanganese slag particles (23.6%), silicomanganese particles (10.3%), S-bearing iron particles (5.9%), sphalerite particles (3.8%), iron (2.6%), alloy particles (1.8%), lead sulfate particles (1%), Zinc rich particles (0.8%), and aluminum manufacturing dust (0.3%). They accounted for about half of the particle numbers, in contrast to the situation in Taiyuan, where coal combustion-related particles were dominant.

Guiyang is known for its ferrous and nonferrous metallurgical industry. Manganese is essential to iron and steel production due to its sulfur-fixing, deoxidizing, and alloying properties. Iron and steel industry accounts for most manganese demand. The appetite for huge quantity of iron and steel has stimulated the ferromanganese and silicomanganese industry.

The individual particle analyses are presented in Paper 1: Chemicalcharacterization of individual particles (PM10) from ambient air in Guiyang City, China.

Taiyuan campaign, summer, 2003 (Paper 2)

Totally six samples were collected, around 2500 particles were automatically analyzed by SEM-EDS. Particles with zero counting and particles with number abundance less than 0.1% were excluded for further data analysis. Meanwhile SEM photomicrographs of some typical particles were obtained manually.

Particle Size and morphology:

Particles from natural sources are generally larger than particles from anthropogenic processes. Soot particles from fuel combustion are generally ultrafine particles with spherical shape. However, with the aging of the particles, the soot particles tend to

25

aggregate and coagulate during transport, and the size grows larger (Fig. 6). The relative size of soot particles can be an indication for “young” and “old” soot particles (Ruellan et al., 1999).

Fig. 6. Soot aggregate.

Morphology can aid to identify processes under which the particles are produced. For example, EDS result shows that two particle types have the same chemical composition (mainly silica). Morphological analysis shows that the particle in Fig. 7a is angular and sharp-edged, indicating a product of mechanical abrasion, from a natural source (McCrone and Delly, 1973); whereas the particles in Fig. 7b are the agglomerates of fine quartz particles as a result of coal combustion, where silica is first reduced in fuel-rich condition and re-oxidised later (Buhre et al., 2007).

Fig. 7. (a) A natural silica particle. (b) Aggregate silica particles from coal combustion.

Particle Chemistry:

26

On the basis of SEM-EDS and cluster analysis, 2329 particles were classified into 20 different types of particles. Soil and fly ash particles were the most abundant (23.3%). These two different types of particles were chemically identical and cannot be resolved solely by chemical composition. Theoretically they can be distinguished by morphology or shape factor. However the large number of particles makes the manual identification by morphology almost impossible. Shape factor, on the other hand, may provide some clues. Generally, soil particles are irregular, with larger shape factor. Fly-ash particles are, on the other hand, mostly spherical, with a shape factor close to unity. However, the existence of amorphous fly-ash particles, resulting from short residence time in the combustion zone, with a shape factor larger than unity, makes differentiation by shape factor difficult.

The second largest fraction is denoted “coal burning particles” (16.3%). The chemistry of these particles, mainly consisting of Ca, S, Al, and Si, is different to that of fly ash, which is dominated by Al and Si. The formation mechanism behind these particles is that the formed gypsum from sulfation of Ca added to or inherent in the coal is adhered to or surrounded by melted Ca Al-silicates (Yoshihiko et al., 2001).

There are some other particles, such as gypsum (7.4%), syngenite (5.1%), and silicon sulphide (3.7%), which can be traced to coal combustion, making the total around 56 percent. This is plausible considering the huge quantity of coal consumption consumed in this area.

The particle occurrence of syngenite and silicon sulphide in the ambient has never been reported as far as we know. They are likely to be found in areas where coal consumption is extensive.

Normally, trace elements cannot be determined by SEM-EDS due to high detection limits (~ 0.1%). However, some typical trace elements, such as Cu, Cr, Mo, Pb, and Zn, occurred in particles with high mass percentage. Each type of the particle may represent an industry or industries: Cu may originate from a copper smelter six km away; Cr, Mo, and Zn may stem from metallurgical industry; and Pb may be from metallurgical smelters, heavy machinery manufacturing, and coal combustion.

The detailed results are presented in the manuscript (Paper 2): Individual Particles (PM10) Characterisation in Taiyuan City, China.

6.2 Bulk sample analyses

Taiyuan campaign, March, 2004 (Paper3)

High mass concentration of PM10 was found. The average concentration was 553 μg m-3, which was far above the Class II level of the National Ambient Air Quality Standard of China (150 μg m-3). This is partly due to the fact that the sampling took place in winter time when space heating from coal combustion gave extra burden for PM10 loading.

Arsenic (As) and selenium (Se) are characteristic trace elements from coal combustion. Comparison was made between different studies from different places (see

27

Table 3 in Paper 3). We found the concentration levels of these two elements are higher in Taiyuan than any other study we are aware of. Attempts were made to explore their origins. Concentrations of As and Se are highly correlated, with r = 0.997. This strong correlation suggests an overwhelmingly dominant source that dictates the relationship between these two, the only reasonable one being coal combustion.

The correlation between distinctly different groups of elements provides additional evidence that the source is coal combustion. Taking aluminium (Al) as the representative of lithophile elements (representing Ca, Fe, K, Mg, Mn, Na, Ti, and V), and As representing chalcophile elements, the correlation between Al and As was significant, with r = 0.75. This coexistence can be explained by the hypothesis that both lithophile and chalcophile elements are the results of coal combustion.

The crustal matter mass was calculated by using Al as the crustal indicator, assuming 8.23% abundance (Taylor, 1964). The average fraction of crustal matter is about 59%, with a standard deviation of 6.4%. That means PM10 mass is mainly composed of crustal matter which is strongly associated with coal combustion. Furthermore, PM10 mass was highly correlated with Al, As, and Se, with r 0.95, 0.90, and 0.88 respectively.

Fig. 8. Categorization of trace elements in coal based on volatility behaviour (Ratafia-Brown, 1994).

Enrichment factor (EF) calculation indicates that some major and trace elements are not enriched. Al, Ca, Fe, K, Mg, Na, Mn, V, and Ni, all have EF less than 5, while As, Zn, Sb, Se, Pb, Bi, and Sn, with EF larger than 5, are considered to be enriched.

A volatilization-condensation mechanism has been invoked to explain the observed varying EFs of different elements during coal combustion. Fig. 8 shows the classification of trace elements in coal according to their volatilities. For the enriched elements, the mechanism involves vaporization of volatile Class II elements during combustion, followed by condensation onto particle surfaces in the flue gas stream during cooling. Class elements do not volatilise easily at the temperature encountered

28

during coal combustion and form bottom ash and fly ash particles. Fly ash particles provide surfaces for volatile elements to condense on due to their large surface-to- volume ratio.

In summary, the bulk sample analysis by ICP-AES/MS in conjunction with strong acids digestion provided reliable concentration data for 16 different metals contained in the PM10 samples, ranging from major to trace level. The high concentrations of trace metals relative to other areas and their strong interrelation with each other and their correlations to crustal metals linked them to coal combustion. The huge amount of coal consumed in the city and the volatilization-condensation mechanism accounted for this.

The detailed results of this study are presented in Paper 3: Heavy coal combustion as the dominant source of particulate pollution in Taiyuan, China, corroborated by high concentrations of arsenic and selenium in PM10.

Taiyuan campaign, summer, 2003 (not published)

Bulk sample analyses of three samples concurrently collected with the individual samples confirmed the enrichment of the trace elements in the samples. The average enrichment factors (EFs) were 206 ( = 185), 555 ( = 135), and 277 ( = 86) for Cu, Pb, and Zn, respectively (Cr and Mo not determined). However, some other trace elements, such as As and Se, also displayed high EFs, with 158 ( = 57) and 9221( = 2585), respectively, even though they did not occur in some particles as major elements, but rather appeared as minor elements in many particles (not detected by SEM-EDS). It can be explained by the volatilization and condensation mechanism during coal combustion process which was fully addressed in the bulk sample campaign.

Guiyang campaign, summer, 2003 (not published)

The high abundance of Mn-containing particle numbers revealed in the individual particle analyses might lead to Mn enrichment, which was verified by the concurrent bulk sample analyses. EF calculation showed that Mn was enriched 100 times ( = 80, n = 3). Manganese compounds are well known neurotoxic substances which may cause manganism, a severe neurological disorder characterized by disturbances of movements (McMillan, 1999). The average concentration was 1.79 μg m-3 ( = 1.53, n = 3), much higher than WHO guideline value (0.15 μg m-3 for annual average) (WHO, 2000).

Trace elements which occurred as major components in a small number of particles also showed enrichment, with EF 785 ( = 153) and EF 499 ( = 300) for Pb and Zn. Other trace elements existed as minor elements (hence could not be detected by SEM-EDS) had EFs 11 ( = 7.3) for Cu, 432 ( = 90) for As, 1905 ( = 752) for Sb, and 11966 ( = 802) for Se.

6.3 Size-fractionated sample analyses (Paper 4)

Cascade impactors are widely used to measure particle size distributions in aerosol studies. Size-resolved measurements and characteristic size distribution have been used to provide information on formation pathways of specific components and to identify particulate sources (Allen et al., 2001; Singh et al., 2002).

29

30

The study in Sheffield found that particle chemical composition and size distribution are strongly influenced by air mass origins (see Fig. 9 below and Table 1 in Paper 4). Air masses belonging to maritime regime bring more sea-salt ingredients such as Na+ and Cl-, and air masses belonging to the terrestrial regime carry more secondary components like SO4

2-, NO3-, and NH4

+.

In the maritime regime, SO42- is bimodal, with one peak in the fine size, and another

one in coarse size. The fine mode SO42- is the product of oxidation of

sulphur-containing precursors, while the coarse mode SO42- is from sea-salt. NO3

- is unimodal, with only one peak in the coarse range, resulting from the replacement of chloride in the sea-salt (the so-called chloride depletion phenomenon). NH4

+ is found in very low concentrations, mainly in the fine mode, associated with SO4

2-. Na+, Cl-, and Mg2+, on the other hand, are all distributed in the coarse mode, indicating their origins of sea-salt.

In the terrestrial regime, SO42- peaks in the fine mode. NO3

- behaves differently, with a major peak in the fine mode, and a small peak in the coarse mode. The former peak stems from homogeneous gas-phase oxidation of nitrogen oxides to gaseous nitric acid, followed by the reaction with gaseous ammonium. Calculation reveals that the later peak is the result of replacement of chloride. NH4

+ is found in the same range as fine SO4

2- and NO3-. Ammonium is in the forms of ammonium sulphate and ammonium

nitrate. Mg2+ occurs in very low concentrations. Na+ shows a small peak in the coarse size, obviously from sea-salt. Chloride, peaks both in the fine and coarse size ranges, with the fine particles from industry and the coarse ones from sea-salt.

Ca2+ and K+ are not presented in the Fig.9 due to their low concentrations. Ca+ is an indicator of soil dust, the low Ca2+-concentrations in the samples suggest either the rare presence or the low solubility of calcium compound(s) in soil in the area. K+ generally comes from sea-salt and biomass burning. Higher concentrations are found on three days. A statistical t-test shows that the K+ concentrations on these days are significantly different from those of other days. They are attributed to fireworks burning during “bonfire”. However, attempts to find the abnormity of accompanying anions failed. This might be due to the much higher background concentrations of the anions, such as SO4

2- and NO3-, relative to K+.

None of the components demonstrate a nucleation mode, probably due to the minor importance of local pollutants. This is understandable, as the measured water-soluble components are, by and large, long-range transported. However, the limited sensitivity of analytical methods must be kept in mind.

The details are presented in the manuscript (Paper 4): Characteristics of water-soluble inorganic chemical components in size-resolved airborne particulate matters in Sheffield, UK.

b. T

errs

tria

l reg

ime

sam

ple,

03-

04

Nov

. 200

6

0

1000

2000

3000

4000

5000

6000

0,0

0,1

1,0

10,0

Dp

(μm

)

dm/dlogDp(ng/m3)

Cl

NO

3SO

4N

H4

Na

Mg

a. M

ariti

me

regi

me

sam

ple,

11-

12 N

ov.

2006

0

1000

2000

3000

4000

5000

6000

7000

0,0

0,1

1,0

10,0

Dp

(μm

)

dm/dlogDp(ng/m3)

Cl

NO

3SO

4N

H4

Na

Mg

31

Fig.

9. C

hem

ical

com

pone

nts

mas

s siz

e di

strib

utio

n. a

. A m

arin

e re

gim

e sa

mpl

e. b

. A te

rres

tria

regi

me

sam

ple

7. Discussion

Methods

Individual particle analysis carried out in cities of Taiyuan and Guiyang is a step towards a comprehensive characterization of aerosols of major urban areas. It has been the primary goal of these studies to find particle types in the selected cities. Different and ever-changing sources, combined with complex atmospheric reactions and interactions, can produce an almost infinitive variety of particles. The numbers of particle types did reflect the diversity of pollution sources. Taiyuan ambient particles were characterized as dominated by coal combustion sources, followed by metallurgical industries. Conversely, Guiyang was dominated by metallurgical industries, followed by coal combustion. Because of the relatively large number of particles analyzed, we are confident that the major particle types, detectable by the method, in the study areas have been identified, and likely sources have been pinpointed.

However, our study has some limitations. Firstly, results are confined to particles larger than 0.19 μm. The diameters of a significant fraction of some types of particles, such as sulphate particles and combustion particles are below this threshold, impairing sample representativeness. Secondly, since the light elements (lighter than Na) cannot be determined by EDS, some important particle types are lost in the analysis. These particles include carbonaceous particles, ammonium nitrate, and fluoride salt. Thirdly, the choice of substrate may introduce uncertainties. The PTFE membrane filter has a web-like structure (see left of Fig. 10). This structure may trap small particles below the surface of the filter, which makes the SEM-EDS detection difficult. Polycarbonate membrane filters (see right of Fig. 10) are better in this respect. They have a smother surface texture making individual particles much easier to discern. However the incompatibility of the polycarbonate filter (high flow resistance) with the median flow sampler made the PTFE our choice. Lastly, high detection limit for trace elements is another disadvantage. Obtaining values for these elements may bring some more information regarding particles’ sources and formation mechanisms.

Fig. 10. Left, a blank PTFE membrane filter, pore size 1.0 μm; right, a blank polycarbonate membrane filter, pore size 1.2 μm.

32

SEM has been used in many studies to characterize ambient particles for about two decades. The ability to provide detailed information on the morphological and chemical characteristics of individual particles has made SEM the method of choice in studies where bulk analytical methods have insufficient resolution to identify source(s) affecting ambient air quality. Automated SEM-EDS can minimize the biases by otherwise manually selected particle analysis.

Bulk sampling in conjunction with ICP-AES/MS allows for elemental analysis. The power of multi-element analysis and low detection limits makes the ICP-AES/MS a good choice for elemental determination. Some elements are useful tracers for source identification as indicated in our study in Taiyuan, where trace levels of As and Se were used for attribution of coal combustion. However, as all the particles collected in a period of time are lumped in one single sample, information, such as elemental associations, may be masked. Relatively large number of samples is needed to reveal these associations either by regression, multivariate statistics, or principal component analyses. Moreover, stringent sample preparation procedures are needed in order to make the sample suitable for ICP-AES/MS analyses.

Size-resolved analysis can provide chemical composition as a function of particle size distribution. The chemical component distribution provides information about sources and physical and chemical processes affecting aerosols as they are transported in the atmosphere, which is demonstrated in our study in Sheffield. However, there are some trade-offs to be considered during the sampling and analysis procedures. Shorter sampling duration will provide better time resolution of the particles, but will challenge the analytical capability of instruments applied. High loading of samples facilitates easy instrumental analysis, but will probably cause particle bouncing, leading to size misclassification. A further complication is the uneven mass distribution on different stages making an appropriate sampling interval hard to decide.

Comparison with other studies

Guiyang city

No similar study has ever been done in Guiyang. However, it may make sense to make a comparison with studies elsewhere in China. In a similar work in Beijing (Lu et al., 2006), clay mineral particles were the most abundant species, accounting for 30%, followed by quartz (13.5%), and carbon (10%). The percentage of clay mineral particles is high in Beijing due to dry climate and low level of forestation (He et al., 2002), which leads to the high level of locally resuspended dust (Duan et al., 2007). Dust storm, originated from the desert regions in China and Mongolia (Xie et al., 2005; Zhao et al., 2007) also makes a contribution in springtime.

Source apportionment study in Shanghai by synchrotron radiation micro-beam induced X-ray fluorescence (Yue et al., 2006) found that automobile exhaust, metallurgical industry, soil dust, and coal combustion accounted for 22%, 18%, 11%, and 9%, respectively, of the PM2.5 numbers (the number percentage in PM2.5 iscomparable to that in PM10, since most particles are found in the fine mode (Seinfeld and Pandis, 2006)). The high contribution of automobile exhaust to the total analysed particles is attributed to the large fleet of automobiles. Shanghai is an industrial city,

33

with the largest metallurgical base of China, which is reflected in the relatively large number percentage of metallurgical particles.

Taiyuan city

Like in many other cities in China, PM is a major pollutant in Taiyuan. PM has been monitored for many years, but it has been mainly focused on mass concentration. Very little, if anything, has been done on the physiochemical characteristics of PM in Taiyuan. To our knowledge, our study was the first one to be published in an international journal. In the following years there emerged some studies carried out by Nankai University, China.

Source apportionment of PM10 in Taiyuan was carried out by Nankai University (Bi et al., 2007), in which the chemical mass balance (CMB) model was used. The samples were collected between April 2001 and January 2002. Their results showed that resuspended dust was the no. 1 contributor, accounting for 32% of the PM10 mass concentration, while coal fly ash came the second, contributing 18%. In their previous study (Zhao et al., 2006), they applied the same method to the resuspended dust, and attributed soil as the no. 1 contributor (47% of mass concentration), and coal combustion as the no. 2 (24% of mass concentration). If the sources are recalculated according to their chemical signatures in the PM10, irrespective of their transport paths, then coal combustion occupies the largest fraction of the PM10 mass concentration, with a value of ca. 26%. The coal combustion contribution to the PM10 is large, but smaller than in our study. There are many factors that may contribute to this. Firstly, sampling periods were different and therefore the sources strength, emission patterns, and meteorological conditions were different. Secondly, different methodologies must be expected to differ somewhat in classification of coal combustion particles and introduce different uncertainties, which makes it hard to compare.

Sheffield city

The major finding in this study is the dependence of particles’ chemical composition and size distribution on the air mass origins. No such study has been undertaken in this city before. However, another study carried out in Harwell, England (Abdalmogith and Harrison, 2005) might be comparable. Harwell is ca. 200 km to the south of Sheffield. An important difference is that Harwell is, unlike Sheffield, surrounded by agricultural land.

As in our study, back trajectories were used in the study to examine the long-rang transport of secondary inorganic aerosol in Harwell. In Harwell, the sampling was conducted for one year (2002-2003). Back trajectories were computed and clustered. Five major clusters of back trajectories were found: north, northwest, west, southwest, and east. The east flow, originating from continental Europe, was enriched with secondary components, which was never found in our study.

Maritime trajectories showed the lowest concentrations of sulphate and nitrate, and high concentration of chloride. Northerly air passing over UK before arriving at Harwell was associated with higher concentrations of sulphate and nitrate compared to maritime airflow. Our results agree well with those found in Harwell study. The SO4

2-/NO3- is also comparable in both studies (for our results see Table 1 in Paper 4).

34

In both studies, the highest SO42-/NO3

- ratios were generally associated with maritime trajectories and low-sulphate and –nitrate concentrations. This may be explained in terms of the expected longer lifetime of accumulation mode sulphate than nitrate, with the latter often in the coarse mode in maritime air (Huang et al., 2004). The contribution of sulphate derived from sea-salt and from DMS oxidation also tends to increase the ratio. The ratios are also consistent with different prevailing chemical processes in low- and high-NOx photochemical regimes (Stein and Lamb, 2003). Back trajectories analysis in Harwell showed very small urban effect for secondary components, which was also the case in our study.

35

8. Summary and conclusions

This dissertation presented three distinct analytic approaches and measurement results in four campaigns in three cities. The measurements include individual particle analysis, bulk sample analysis, and size-resolved analysis.

We focused on the following aspects:

Morphological features and chemical composition as relative mass percentage in individual particles;

Water-soluble components concentrations, their size distributions, and species interrelations in the size-fractionated samples;

Metallic elements concentrations and their interrelations in PM10.

The conclusions are:

Individual particle analysis by SEM/EDS is an excellent method for characterizing urban particles. It is intuitive and direct, although it is semi-quantitative and information of some important particles may be lost;

Individual particle analysis revealed that in Taiyuan the majority of particles are from coal burning, while in Guiyang the majority of particles are from metallurgic industry;

Size-fractionated sample collection and analysis are another method to characterize airborne particles. Comparison of the observed species distribution patterns in the sampling area with knowledge of PM from various sources can uncover their origin(s);

Results in the City of Sheffield show that size-fractionated water-soluble inorganic components are a combination of secondary aerosol and sea-salt particles. The degree to which the two mix depends on the origins of the air masses;

Bulk samples analysis by ICP-AES/MS, with emphasis on trace metals, provides another useful method for characterizing urban ambient particles;

Investigation into the occurrence of As and Se in Taiyuan air shed points to unusual high concentrations of the two elements. Their close correlations with lithophile elements (taking Al as their representative) illustrate their common source(s). A vitalization-condensation mechanism explains this phenomenon;

Results by individual particle analysis by SEM-EDS and bulk sample analysis by ICP-AES/MS are consistent in characterizing the pollution pattern in Taiyuan, although the samples were collected at different times.

We deem that the results of these studies at different sites, with variable approaches, add to the scientific understanding of urban ambient particles. However, we need to

36

stress that, due to time and cost constraints, the observations were made in quite limited periods. The environmental status will change over time, yet the methodology, the experience, and the know-how we gained from the research process will benefit us for the years to come.

37

9. Suggested research activities

As the studies in this thesis were carried out in some short campaigns, the results obtained just represent the pollution status at the specific conditions under which samples were collected. With the ever-changing conditions, such as industrial activities and meteorological parameters, the types and strength of PM sources will undoubtedly change. Long-term monitoring is necessary to get a satisfactory profile of PM pollution.

The three approaches we have developed are complementary to each other. However, restrained by funding, capacity, and time, not all of them have been applied at the three sampling sites. Consequently, the information gained in our studies is fragmented and the usefulness is to some extent limited. A thorough study applying all the methods can provide a panorama of exceptional scientific value of PM10.

Amounts of organic and elemental carbon (OC and EC) are important characteristrics of PM not included in this study. Some preliminary studies have been carried out at this institute in collaboration with Norwegian Institute for Air Research (Nori, 2005; Grodzinski, 2007). Continuation of studies on EC and OC may give important information in addition to that obtained in this study.

During the study with SEM/EDS, we discovered some characteristic particle types. However, it was not without difficulties to link them to some industrial installations and processes. For the easy and precise attribution of the particles, compilation of an atlas listing as comprehensive as possible individual particles from various pollution sources, both natural and anthropogenic, will definitely benefit researchers in this field. This atlas, containing both morphological features and chemical signatures, should include not just the normally encountered particles but also some less abundant particles characteristic of various industrial and environmental processes.

In the study on size-fractionated samples, attention was focused on the water-soluble components. The results unveiled that the pollutant sources were long-range transported secondary aerosols and sea-salts. Little information could be found about the local pollution sources. The main reason is the slow rates at which the secondary inorganic particles are formed through chemical reactions of gas phase precursors. A study of the total composition of primarily emitted particles, especially the characteristic trace metals, can signal the natural and industrial sources. The differences in PM downwind and upwind of a site may reveal the contribution of local pollution source(s) more directly.

38

10. Appendices

Appendix

ICP-AES Operating Conditions. RF power 1250 W Plasma flow 15 l min-1

Auxiliary flow 1 l min-1

Nebuliser flow 0.7 l min-1

Replicate read time 4 s Instrument stabilization delay 15 s Sample uptake delay 15 s Pump flow rate 1 ml min-1

Rinse time 15 s Replicates 3Calibration method ExternalType of detector CCDType of nebuliser V-groove

Appendix II

ICP-AES: analytical wavelength used for each element Element Wavelength (nm)

Al 396.152Ca 317.933Fe 259.940K 766.491

Mg 280.270Na 588.995Pb 220.353Ti 334.941Zn 213.857

39

Appendix

ICP-MS operating conditionsInstrument Perkin-Elmer SCIEX Elan Model

5000Forward power 1000W Plasma flow Ar 15 l min-1

Nebulizer flow ~ 0.91 min (variable) 1

Sample uptake rate 1 ml min-1

Scanning mode Peak hop Dwell time 100 ms Replicate time 200 ms Ion lenses Optimized on Mg, 103 Rh and 208 Pb24

Sweeps/Reading 2Readings/Replicate 1Number of replicates 3Points across peak 1Baseline time 0 ms Polarity +Calibration method ExternalType of mass spectrometer Quadrupole

Appendix IV

ICP-MS: masses of selected elemental isotopes and the analytical ratio (m/z) used. Element Isotope mass m/z

As 75 75Co 59 59Cu 63 63Mn 55 55Ni 58 58Sb 121 121Se 80 80V 51 51Sn 120 120Pb 208 208

40

11. References

Abdalmogith SS, Harrison RM. 2005. The use of trajectory cluster analysis to examine the long-range transport of secondary inorganic aerosol in the UK. Atmospheric Environment 39, 6686-6695.

Abdullah UM, Iqbal MZ. 1991. Response of automobile, stone and cement particulate matters on stomatal clogging of plants. Geobios (Jodhpur, India) 18, 196-202.

Adamson IYR, Prieditis H, Hedgecock C, Vincent R. 2000. Zinc is the toxic factor in the lung response to an atmospheric particulate sample. Toxicology and Applied Pharmacology 166, 111-119.

Allen AG, Nemitz E, Shi JP, Harrison RM, Greenwood JC. 2001. Size distributions of trace metals in the atmospheric aerosols in the United Kingdom. Atmospheric Environment 35, 4581-4591.

Aris R, Christian D, Tager I, Ngo L, Finkbeinger WE, Balmes JR. 1993. Effects of nitric acid gas alone or in combination with ozone on healthy volunteers. American Review of Respiratory Disease 148, 965-973.

Artaxo P. 1995. Size distribution of biogenic aerosol particles from the amazon basin. Atmospheric Environment 29, 393-402.

Balasubramanian R, Qian WB. 2004. Characterisation and source identification of airborne trace metals in Singapore. Journal of Environmental Monitoring, 6, 813-818.

Bardouki H, Liakakou H, Economou C, Sciare J, Smolík J, Ždímal V, Eleftheriadis K, Lazaridis M, Dye C, Mihalopoulos N. 2003. Chemical composition of size-resolved atmospheric aerosols in the eastern Mediterranean during summer and winter. Atmospheric Environment 37, 195-208.

Bauer MA, Utell MJ, Speers DM, Gibb FR, Morrow PE. 1988. Effects of near ambient levels of sulfuric acid aerosol on lung function in exercising subjects with asthma and COPD. American Review of Respiratory Disease 137, A167.

Bi XH, Feng YC, Wu JH, Wang YQ, Zhu T. 2007. Source apportionment of PM10 in six cities of northern China. Atmospheric Environment 41, 903-912.

Bilos C, Colombo JC, Skorupka CN, Rodriguez Presa MJ. 2001. Sources, distribution and variability of airborne trace metals in La Plata City area, Argentina. Environmental Pollution 111, 149-158.

Blando JD, Turpin BJ. 2000. Secondary organic aerosol formation in cloud and fog droplets: a literature evaluation of plausibility. Atmospheric Environment 34, 1623-1632.

41

Bowman F, Odum J, Pandis SN, Seinfeld JH. 1997. Mathematical model for gas-particle partitioning of secondary organic aerosol. Atmospheric Environment 31, 3921-3931.

Brook RD, Brook JR, Urch B, Vincent R, Rajagopalan S, Silverman F. 2002. Inhalation of fine particulate air pollution and ozone cause acute arterial vacoconstriction in healthy adults. Circulation 105, 1534-1536.

Brunekreef B, Hoek G, Fischer P, Spieksma FTM. 2000. Relation between airborne pollen concentrations and daily cardiovascular and respiratory-disease mortality. Lancet 355, 1517-1518.

Buhre B, Hinkley J, Gupta R, Nelson P, Wall T. 2007. Factors affecting the vaporisation of silica during coal combustion. Fuel Processing Technology 88, 157-164.

Cassee FR, Arts JHE, Fokkens PHB, Spoor SM, Boere AJF, Van Bree L, Dormans JAMA. 2002. Pulmonary effects of ultrafine and fine ammonium salts aerosols in healthy and monocrotaline-treated rats following short-term exposure. lnhalation Toxicology 14, 1215-1229.

Celenza A, Fothergill J, Kupek E, Shaw RJ. 1996. Thunderstorm associated asthma: a detailed analysis of environmental factors. British Medical Journal 312, 604-607.

Chow JC. 1995. Critical Review: Measurement Methods to Determine Compliance with Ambient Air Quality Standards for Suspended Particles. Journal of the Air & Waste Management Association 45, 320-382.

Cohen AJ, Anderson HR, Ostro B, Pandey KD, Krzyzanowski M, Kuenzli N, Gutschmidt K, Pope CA, Romieu I, Samet JM, Smith KR. 2004. Mortality impacts of urban air pollution. In: Ezzati M, Lopez AD, Rodgers A, Murray CJL eds. Comparative quantification of health risks: global and regional burden of disease attributable to selected major risk factors. Geneva, World Health Organization, 1353–1434.

Concawe report 98/51, 1998. A study of the number, size and mass of exhaust particles emitted from European diesel and gasoline vehicles under steady-state and European driving cycle conditions, Brussels.

Darley EF. 1966. Studies on the effect of cement-kiln dust on vegetation. Journal of the Air Pollution Control Association 16, 145-150.

Dockery DW, Pope CA, Xu XP, Spengler JD, Ware JH, Fay ME, Ferris BG, Speizer FE. 1993. An association between air pollution and mortality in six US cities. The New England Journal of Medicine 329, 1753-1759.

Dominici F, Burnett RT. 2003. Risk models for particulate air pollution. Journal of Toxicology and Environmental Health, part A 66, 1883–1889.

42

Dreher KL, Jaskot RH, Lehmann JR, Richards JH, McGee JK. 1997. Soluble transition metals mediate residual oil fly ash-induced acute lung injury. Journal of Toxicology and Environmental Health, 50, 285-305.

Duan FK, Liu XD, He KB, Li YW, Dong SP. 2007. Characteristics and Source Identification of Particulate Matter in Wintertime in Beijing. Water, Air, & Soil Pollution 180, 171-183.

Dye JA, Lehmann JR, McGee JK, Winsett DW, Ledbetter AD, Everitt JI, Ghio AJ, Costa DL. 2001. Acute pulmonary toxicity of particulate matter filter extracts in rats: Coherence with epidemiological studies in Utah Valley residents. Environmental Health Perspectives 109, 395-403.

Environmental Monitoring Center of Taiyuan: 2004, Environmental Quality Report for Taiyuan, 2003. Taiyuan, Shanxi, China.

Farmer AM. 1993. The effects of dust on vegetation-a review. Environmental Pollution 79, 63-75.

Fitzgerald JM. 1991. Marine aerosols: A review. Atmospheric Environment, Part A, 25, 533-545.

Fu FF, Watanabe K. 2004. Seasonal characteristics of chemical compositions of the atmospheric aerosols collected in urban seaside area at Tokaimura, eastern central Japan. Journal of Geophysical Research 109, D20212/1-D20212/9.

Gillette D. 1980. Major contributions of natural primary continental aerosols-source mechanisms. Annals of the New York Academy of Sciences 338, 348-358.

Gray HA, Cass GR, Huntzicker JJ, Heyerdahl EK, Rau JA. 1986. Characterisation of atmospheric organic and elemental carbon particle concentrations in Los Angeles. Environmental Science and Technology 20, 580-589.

Greenbaum DS, Bachmann JD, Krewski D, Samet JM, White R, Wyzga RE. 2001. Particulate air pollution standards and morbidity and mortality: Case study. American Journal of Epidemiology 154, S78–90.

Greenberg JP, Guenther A, Zimmermann P, Baugh W, Geron C, Davis K, Helmig D, Klinger LF. 1999. Tethered balloon measurements of biogenic VOCs in the atmospheric boundary layer. Atmospheric Environment 33, 855-867.

Grgi I, Ber i G. 2001. A simple kinetic model for autoxidation of S(IV) oxides catalyzed by ion and/or manganese ions. Journal of Atmospheric Chemistry 39, 155-170.

Grodzinski J. 2007. Determination of major particulate matter (PM10) components in urban aerosol from Taiyuan City, China. Master thesis. Department of Chemistry, University of Oslo.

Guderian R. ed. 1985. Air pollution by photochemical oxidants: formation, transport, control, and effects on plants. New York, NY: Springer-Verlag. (Billings WD,

43

Golley F, Lange OL, Olson JS, Remmert H. eds. Ecological studies: analysis and synthesis: v. 52).

Halliwell B, Gutteridge JMC. 1999. Free radicals in biology and medicine. Oxford University Press, Oxford, UK.

Harrison RM, Kitto A-MN. 1990. Field intercomparison of filter pack and denuder sampling methods for reactive gaseous and particulate pollutants. Atmospheric Environment 24A, 2633-2640.

Harrison RM, Jones MR. 1995. The chemical composition of airborne particles in the UK atmosphere. Science of the Total Environment 168, 195-214.

Harrison, RM, Smith DJT, Luhana L. 1996. Source apportionment of atmospheric polycyclic aromatic hydrocarbons collected from an urban location in Birmingham, U.K. Environmental Science and Technology 30, 825-32.

Harrison RM, Yin J. 2000. Particulate matter in the atmosphere: which particle properties are important for its effects on health? The Science of Total Environment 249, 85-101.

Harrison RM. 2004. Key pollutants-airborne particles. Science of the Total Environment 334-335, 3-8.

He KB, Huo H, Zhang Q. 2002. Urban air pollution in China: current status, characteristics, and progress. Annual Review of Energy and Environment 27, 397-431.

HEI. 2000. Reanalysis of the Harvard Six-Cities study and the American Cancer Society study of particulate air pollution and mortality. A special report of the Institute’s Particle Epidemiology reanalysis Project. Cambridge, MA, Health Effects Institute.

HEI International Oversight Committee. 2004. Health effects of outdoor air pollution in developing countries of Asia: a literature review. Boston, MA, Health Effects Institute (Special Report No. 15).

Houck JE, Chow JC, Ahuja MS. 1989b. The chemical and size characterization of particulate material originating from geological sources in California. In: Watson JG (ed.), Receptor models in air resources management. Transactions of an APCA Speciality Conference. Air & Waste Management Association, Pittsburgh, PA.

Huang Z, Harrison RM, Allen AG, James JD, Tilling RM, Yin J. 2004. Field intercomparison of filter pack and impactor sampling for aerosol nitrate, ammonium, and sulphate at coastal and inland sites. Atmospheric Research 71, 215-232.

IPCC: Changes in Atmospheric Constituents and in Radiative Forcing. In: Climate Change 2007: The Physical Science Basis. Contribution of Working Group I to the Fourth Assessment Report of the Intergovernmental Panel on Climate. Forster P, Ramaswamy V, Artaxo P, Berntsen T, Betts R, Fahey DW, Haywoo J, Lean J,

44

Lowe DC, Myhre G, Nganga J, Prinn R, Raga G, Schulz M, Van Dorland R. 2007. [Solomon S, Qin D, Manning M, Chen Z, Marquis M, Averyt KB, Tignor M, Miller HL (eds.)]. Cambridge University Press, Cambridge, United Kingdom and New York, NY, USA.

Janssen NA, Brunekreef B, van VP, Aarts F, Meliefste K, Harssema H, Fischer P. 2003. The relationship between air pollution from heavy traffic and allergic sensitization, bronchial hyperresponsiveness, and respiratory symptoms in Dutch schoolchildren. Environmental Health Perspectives 111, 1512–1518.

Jerrett M. 2005. Spatial analysis of air pollution and mortality in Los Angeles. Epidemiology, 16, 727–736.

Jonson JE, Semb A, Barrett K, Grini K, Tarrason L. 2000. On the distribution of sea salt and sodium nitrate particles in Europe. Transport and Chemical Transpotation in the Troposphere, Proceeding of the EUROTRAC Symposium, Sixth , Gaimisch-Partenkirchen, Germany, 27-31 March 2000, pp. 695-699.

Kaneyasu N, Yoshikado H, Mizuno T, Sakamoto K, Soufuku M. 1999. Chemical forms and sources of extremely high nitrate and chloride in winter aerosol pollution in the Kanto Plain of Japan. Atmospheric Environment 33, 1745-1756.

Karanasiou AA, Sitaras IE, Siskos PA, Eleftheriadis K. 2007. Size distribution and sources of trace metals and n-alkananes in the Athens urban aerosol during summer. Atmospheric Environment 41, 2368-2381.

Kerminen V-M, Teinilä K, Hillamo R. 1999. Chemistry of sea salt particles in the summer Antarctic atmosphere. Atmospheric Environment 34, 2817-2825.

Katsouyanni K, Touloumi G, Samoli E, Gryparis A, Le Tertre A, Monopolis Y, Rossi G, Zmirou D, Ballester F, Boumghar A, Anderson HR, Wojtyniak B, Paldy A, Braunstein R, Pekkanen J, Schindler C, Schwartz J. 2001. Confounding and effect modification in the short-term effects of ambient particles on total mortality: results from 29 European cities within the APHEA2 project. Epidemiology 12, 521–531.

Khoder MI. 2002. Atmospheric conversion of sulphur dioxide to particulate sulphate and nitrogen nitric acid in an urban area. Chemosphere 49, 675-684.

Kittelson DB. 1998. Engines and nanoparticles: a review. Journal of Aerosol Science 29, 575-588.

Kobzik L, Goldsmith CW, Ning YY, Qin G, Morgan B, Imrich A. 2001. Effects of combined ozone and air pollution particle exposure in mice. Health Effects Institute Research Report 106. Health Effects Institute, Boston.

Kukkonen J, Salmi T, Saari H, Konttinen M, Kartastenpää R. 1999. Review of urban air quality in Finland. Boreal Environmental Research 4, 55-65.

45

Kunzli N, Jerrett M, Mack WJ, Beckerman B, LaBree L, Gilliland F, Tomas D, Peters J, Hodis HN. 2005. Ambient air pollution and atherosclerosis in Los Angeles. Environmental Health Perspectives 113, 201–206.

Langner J, Rodhe H. 1991. A global three-dimensional model of the tropospheric sulphur cycle. Journal of Atmospheric Chemistry 13, 255–263.

Leikauf GD, Spector DM, Albert RE, Lippmann M. 1984. Dose-dependent effects of submicrometer sulfuric acid aerosol on particle clearance from ciliated human lung airways. American Industrial Hygiene Association Journal 45, 285–292.

Lin, JJ, Lee LC. 2004. Characterization of n-alkanes in urban submicron aerosol particles (PM1). Atmospheric Environment 38, 2983-2991.

Lighty JS, Veranth JM, Sarofim AF. 2000. Combustion aerosols: factors governing the size and composition and implications to human health. Journal of Air & Waste Management Association 50, 1565-1618.

Lippmann M. 1989. Background on health effects of acid aerosols. Environmental Health Perspectives 79, 3–6.

Loscutoff SM, Cannon WC, Buschbom RL, Busch RH, Killand BW. 1985. Pulmonary function in elastase-treated guinea pigs exposed to ammonium sulfate or ammonium nitrate aerosols. Environmental Research 36, 170-180.

Mansfield T, Hamilton R, Ellis B, Newby P. 1991. Diesel particulate emissions and the implications for the soiling of buildings. The Environmentalist 11, 243-254.

McCrone WC, Delly JG. The particle atlas. Edition two. Michigan, USA: Ann Arbor Science Publishers; 1973.

McInnes LM, Covert DS, Quinn PK, Germani MS. 1994. Measurement of chloride depletion and sulphur enrichment in individual sea-salt particle collected from the remote marine boundary layer. Journal of Geophysical Research 99, 8257-8268.

McMillan DE. 1999. A brief history of the neurobehavioral toxicity of manganese: some unanswered questions. Neurotoxicology, 20, 499–507.

Meng ZY, 1994. On the sources of the submicrometer droplet mode of urban and regional aerosols. Aerosol Science and Technology 20, 253-265.

Morawska L, Zhang JF. 2002. Combustion sources of particles. 1. Health relevance and source signatures. Chemosphere 49, 1045-1058.

NBS (National Bureau of Statistics of the People’s Republic of China), 2006. China Statistics 2005. China Statistics Press.

Nori SJ, 2005. Study of particulate matter (PM10) in air in Taiyuan, Shanxi, China. Master thesis. Department of Chemistry, University of Oslo.

Nriagu JO. “Natural versus anthropogenic emissions of trace metals to the atmosphere”

46

in : J. M. Pacyna and B. Ottar (eds), Control and Fate of Atmospheric Trace Metals. Kluwer Academic Publishers, Dordrecht, The Nertherlands, pp. 3-13.

Ocskay R, Salma I, Wang W, Maenhaut W. 2006.Characterisation and diurnal variation of size-resolved inorganic water-soluble ions at a rural background site. Journal of Environmental Monitoring 8, 300-306.

O'Dowd CD, Facchini MC, Cavalli F, Ceburnis D, Mircea M, Decesari S, Fuzzi S, Yoon YJ, Putaud J-P. 2004. Biogenically driven organic contribution to marine aerosol. Nature 431, 676-80.

Pacyna JM. 1998. Source inventories for atmospheric trace metals. In: Harrison, R. M., van Grieken, R. E. (Eds.) Atmospheric Particles, IUPAC Series on Analytical and Physical Chemistry of Environmental Systems, Vol. 5. Wiley, Chichester, UK, pp. 385-423.

Pakkanen TA. 1996. Study of formation of coarse particle nitrate aerosol. Atmospheric Environment 30, 2475-2482.

Pakkanen TA, Kerminen V-M, Korhonen CH, Hillamo RE, Aarnio P, Koskentalo T, Maenhaunt W. 2001. Urban and rural ultrafine (PM0.1) particles in the Helsinki area. Atmospheric Environment 35, 4593-4607.

Pandis SN, Harley RH, Cass GR, Seinfeld JH. 1992. Secondary organic aerosol formation and transport. Atmospheric Environment 26A, 2269-2282.

Park SS, kimYJ. 2004. PM2.5 particles and size-segregated ionic species measured during fall season in three urban sites in Korea. Atmospheric Environment 38, 1459-1471.

Parmar RS, Satsangi GS, Kumari M, Lakhani A, Srivastava SS, Prakash S. 2001. Study of size distribution of atmospheric aerosol at Agra. Atmospheric Environment 35, 693-702.

Pope CA III, Thun MJ, Namboodiri MM, Dockery DW, Evans JS, Speizer FE, Heath CW Jr. 1995. Particulate air pollution as a predictor of mortality in a prospective study of US adults. American Journal of Respiratory and Critical Care Medicine 151, 669–674.

Pope CA III, Hill RW, Villegas GM. 1999. Particulate air pollution and daily mortality on Utah’s Wasatch Front. Environmental Health Perspectives, 107, 567-573.

Pope CA III, Burnett RT, Thun MJ, Calle EE, Krewski D, Ito K, Thurston GD. 2002. Lung cancer, cardiopulmonary mortality, and long-term exposure to fine particulate air pollution. Journal of the American Medical Association 287, 1132–1141.

Raes F, Dingenen RV, Vignati E, Wilson J, Putaud J-P, Seinfield JH, Admas P. 2000. Formation and cycling of aerosols in the global troposphere. Atmospheric Environment 34, 4215-4240.

47

Ratafia-Brown JA. 1994. Overview of trace elements partitioning in flames and furnaces of utility coal-fired boilers, Fuel Processing Technology 39, 139–157.

Rosas I, McCartney HA, Payne RW, Calderón C, Lacey J, Chapela R, Ruiz Velazco S. 1998. Analysis of the relationships between environmental factors (aeroallergens, air pollution, and weather) and asthma emergency admissions to a hospital in Mexico City. Allergy 53, 394-401.

Ruellan S, Cachier H, Gaudichet A, Masclet P, Lacaux J-P. 1999. Airborne aerosols over central Africa during the Experiment for Regional Sources and Sinks of Oxidants (EXPRESSO). Journal of Geophysical Research 104, 30673-30690.

Sabbioni C, Zappia G, Gobbi G. 1996. Carbonaceous particles and stone damage in a laboratory exposure system. Journal of Geophysical Research (Atmospheres) 101, 19624-19627.

Samet JM, Reed W, Ghio AJ, Devlin RB, Carter JD, Dailey LA, Bromberg PA, Madden MC. 1996. Induction of prostaglandin H Synthase 2 in human airway epithelial cells exposed to residual oil fly ash. Toxicology and Applied Pharmacology 141, 159-168.

Samet JM, Zeger SL, Dominici F, Curriero F, Coursac I, Dockery DW, Schwartz J, Zanobetti A. 2000. The National Morbidity, Mortality, and Air Pollution Study. Part II: Morbidity and mortality from air pollution in the United States. Research Reports of the Health Effects Institute 94, 5–70.

Savoie DL, Prospero JM. 1982. Particle size distribution of nitrate and sulfate in the marine atmosphere. Geophysical Research Letter 9, 1207-1210.

Schlesinger RB. 1989. Factors affecting the response of lung clearance system to acid: aerosols role of exposure concentration, exposure time, and relative acidity. Environmental Health Perspectives 79, 121–126.

Schwartz J, Dockery DW. 1992. Particulate air pollution and daily mortality in Steubenville, Ohio. American Journal of Epidemiology 135, 12-19.

Schwartz J, Norris G, Larson T, Shepphard L, Claiborne C, Koenig J. 1999. Episodes of high coarse particle concentrations are not associated with increased mortality. Environmental health perspectives 107, 339-342.

Seinfeld JH, Pandis SN. 2006. Atmospheric Chemistry and Physics. From Air Pollution to Climate Change. 2nd ed. John Wiley and Sons, New York.

Simoneit BRT, Mazurek MA. 1982. Organic- matter of the troposphere. 2. Natural back-ground of biogenic lipid matter in aerosols over the rural western United-States. Atmospheric Environment 16, 2139-2159.

Singh M, Jaques PA, Sioutas C. 2002. Size distribution and diurnal characteristics of particle-bound metals in source and receptor sites of the Los Angeles Basin. Atmospheric Environment 36, 1675-1689.

48

Smith WH. 1984. Pollutant uptake by plants. In: Treshow M. ed. Air pollution and plant life. New York, NY: John Wiley & Sons, Ltd. pp. 417-450.

Somers CM, McCarry BE, Malek F, Quinn JS. 2004. Reduction of particulate air pollution lowers the risk of heritable mutations in mice. Science 304, 1008-1010.

Sorensen M, Schins RPF, Hertel O, Loft S. 2005. Transition metals in personal samples of PM2.5 and oxidative stress in human volunteers. Cancer Epidemiology, Biomarkers & Prevention 14, 1340-1343.

Stein AF, Lamb D. 2003. Empirical evidence for the low- and high-NOx photochemical regimes of sulphate and nitrate formation. Atmospheric Environment 37, 3615-3625.

Strader R, Lurmann F, Pandis SN. 1999. Evaluation of secondary organic aerosol formation in winter. Atmospheric Environment 33, 4849-4863.

Swietlicki E, Puri S, Hansson H, Edner H. 1996. Urban air pollution source apportionment using a combination of aerosol and gas monitoring techniques. Atmospheric Environment 30, 2795-2809.

Tanner RL, Parkhurst WJ, Mcnichol AP. 2004. Fossil sources of ambient aerosol carbon based on 14C measurements. Aerosol Science and Technology 38, 133-139.

Taylor SR. 1964. Abundance of chemical elements in the continental crust: a new table. Geochimica ET Cosmochimica Acta, 28, 1273-1285.

Temesi D, Molnár A, Mészáros E, Feczkó T, Gelencsér A, Kiss G, Krivácsy Z. 2001. Size resolved chemical mass balance of aerosol particles over rural Hungary. Atmospheric Environment 35, 4347-4355.

Ten Harkel MJ. 1997. The effects of particle-size distribution and chloride depletion of sea-salt aerosols on estimating atmospheric deposition at a coastal site. Atmospheric Environment 31, 417-427.

Tuomi T, Engström B, Niemelä R, Svinhufvud J, Reijula K. 2000. Emission of ozone and organic volatiles from a selection of laser printers and photocopiers. Applied Occupational and Environmental Hygiene 15, 629-634.

Turpin BJ, Huntzicker JJ. 1991. Secondary formation of organic aerosol in the Los Angeles Basin: a descriptive analysis of organic and elemental carbon concentrations. Atmospheric Environment 25A, 207-215.

Urch B, Brook JR, Wasserstein D, Brook RD, Rajagopalan S, Corey P, Silverman F. 2004. Relative contributions of PM2.5 chemical constituents to acute arterial vasoconstriction in humans. Inhalation Toxicology 16, 345-352.

U.S. Code. 1970. Clean Air Act. USC42 PL 91-604.

49

U.S. EPA. 1986. Second addendum to air quality criteria for particulate matter and sulfur oxides (1982). Assessment of newly available health effects information. EPA-600/8-86-020-F. Research Triangle Park, NC: Office of Research and Development.

U.S. EPA. 1996. Air quality criteria for particulate matter. EPA-600/P-95/001aF. Research Triangle Park, NC: Office of Research and Development.

U.S. EPA. 1997. National ambient air quality standards for particulate matter. Final rule. Federal Registry 62:38652–38701.

U.S. EPA. 2004. Air quality criteria for particulate matter. http://cfpub.epa.gov/ncea/cfm/recordisplay.cfm?deid=87903

U.S. EPA. 2006. http://www.epa.gov/pm/naaqsrev2006.html.

Utell MJ, Morrow PE, Hyde RW. 1982. Comparison of normal and asthmatic subjects’ response to sulphate pollutant aerosols. Annals of Occupational Hygiene 26, 691–697.

Van der Hoek KW. 1998. Estimating ammonia emission factors in Europe: Summary of the work of the UNECE ammonia expert panel, Atmospheric Environment 32, 315-316.

Vardaka E, Cook CM, Lanaras T, Sgardelis SP, Pantis JD. 1995. Effect of dust from a limestone quarry on the photosynthesis of Quercus coccifera, an evergreen schlerphyllous shrub. Bulletin of Environmental Contamination and Toxicology 54, 414-419.

Vassilev SV, Vassileva CG. 2005. Methods for characterization of composition of fly ashes from coal-fired power stations: a critical overview. Energy & Fuels 19, 1084-1098.

Veranth JM, Reilly CA, Veranth MM, Moss TA, Lanagelier CR, Lanza DL, Yost GS. 2004. Inflammatory cytokines and cell death in BEAS-2B lung cells treated with soil dust, lipopolysaccharide, and surface-modified particles. Toxicological Sciences 82, 88-96.

Wall SM, John W, Ondo JL. 1988. Measurement of aerosol size distribution for nitrate and major ionic species. Atmospheric Environment 22, 1649-1656.

Wang XL, Sato T, Xing BS, Tamamura S, Tao S. 2005. Source identification, size distribution and indicator screening of airborne trace metals in Kanazawa, Japan. Journal of Aerosol Science 36, 197-210.

Watkinson WP, Campen MJ, Costa DL. 1998. Cardiac arrhythmia induction after exposure to residual oil fly ash particles in a rodent model of pulmonary hypertension. Toxicological Sciences 41, 209-216.

50

Whitby KT. 1978. The physical characteristics of sulphur aerosols. Atmospheric Environment 12, 135-159.

Willeke K, Baron PA. 1993. Aerosol measurement: Principals, techniques and applications, Van Nostrand Reinhold, New York.

WHO. 1987. Air quality guidelines for Europe. WHO Regional Office for Europe, Copenhagen (WHO Regional Publications, European Series, No. 23).

WHO, 2000. Air quality guidelines for Europe. 2nd ed. World health Organization, Regional Office for Europe, Copenhagen.

WHO. 2006. WHO Air quality guidelines for particulate matter, ozone, nitrogendioxide and sulfur dioxide, Global update 2005.

Wolff GT. 1984. On the nature of nitrate in coarse continental aerosols. Atmospheric Environment 18, 977-981.

Wong CG, Bonakdar M, Mautz WJ, Kleinman MT. 1996. Chronic inhalation exposure to ozone and nitric acid elevates stress-inducible heat shock protein 70 in the rat lung. Toxicology 107, 111-119.

World Resources Institute, 1999. Urban air pollution risks to children: A global environmental health indicator. World Resources Institute, Washington, DC 20002.

Xie SD, Zhang YH, Qi L, Tang XY. 2005. Characteristics of air pollution in Beijing during sand storm periods. Water, Air, and Soil Pollution 5, 217-229.

Xiu GL, Zhang DN, Chen JZ, Huang XJ, Chen ZX, Guo HL, Pan JF. 2004. Characterisation of major water-soluble inorganic ions in size-fractionated particulate matters in Shanghai compus ambient air. Atmospheric Environment 38, 227-236.

Xu MH, Yan R, Zheng CG, Qiao Y, Han J, Sheng CD. 2003. Status of trace element emission in a coal combustion process: a review. Fuel Processing Technology 85, 215-237.

Yang CS, Johanning E. 2002. Airborne fungi and mycotoxins. In: Hurst CJ, Crawford RL, Knudsen GR, McInerney MJ, Stetzenbach LD eds. Manual of environmental microbiology. 2nd ed. Washington, DC: American Society for Microbiology pp. 839-852.

Yao XH, Fang M, Chan CK. 2003. The size dependence of chloride depletion in fine and coarse sea-salt particles. Atmospheric Environment 37, 743-751.

Yoshihiko N, Zhang L, Atsushi S. Simultaneous removal of SOx and halogens under pulverized coal combustion added fine limestone. Proceedings-Annual International Pittsburgh Coal Conference. 2001; 18th, 763-774.

51

Zappia G, Sabbioni C, Gobbi G. 1993. Non-carbonate carbon content on black and white areas of damaged stone monuments. Atmospheric Environment 27, 1117-1122.

Zhang Y, Carmichael GR. 1999. The role of mineral aerosol in tropospheric chemistry in East Asia-a model study. Journal of Applied Meteorology 38, 353-366.

Zhao PS, Feng YC, Zhu T, Wu JH. 2006. Characterizations of resuspended dust in six cities of North China. Atmospheric Environment 40, 5807-5814.

Zhao XJ, Zhuang GS, Wang ZF, Sun YL, Wang Y, Yuan H. 2007. Variation of sources and mixing mechanism of mineral dust with pollution aerosol-revealed by the two peaks of a super dust storm in Beijing. Atmospheric Research 84, 265-279.

Zhuang H, Chan CK, Fang M, Wexler AS. 1999a. Formation of nitrate and non-sea-salt sulfate on coarse particles. Atmospheric Environment 33, 4223-4233.

Zhuang H, Chan CK, Fang M, Wexler AS. 1999b. Size distribution of particulate sulfate, nitrate, and ammonium at a coastal site in Hongkong. Atmospheric Environment 33, 843-853.

52

Papers

53

I

Chemical characterization of individual particles (PM10) from

ambient air in Guiyang City, China

R.K. Xiea,*, H.M. Seipa, J.R. Leinumb, T. Winjec, J.S. Xiaod

aDepartment of Chemistry, University of Oslo, Blindern, P.B.1033, Oslo, NorwaybDepartment of Materials Technology, Norwegian University of Science and Technology, N-7491 Trondheim, Norway

cDepartment of Geosciences, University of Oslo, Blindern P. B.1047, Oslo, NorwaydGuizhou Institute of Environmental Science, P. C. 550002, Guiyang, Guizhou, China

Received 25 June 2004; accepted 27 October 2004

Abstract

PM10 samples were collected during 5 days in Guiyang, China in July 2003. A total of about 2300 particles was analyzed by

an automated Scanning Electron Microscope with Energy-Dispersive Spectrometer (SEM–EDS). Hierarchical cluster analysis

(HCA) was used to identify different particle types that occurred in the aerosol. Seventeen particle types were identified and

presented in the order of decreasing number abundance as: silicomanganese slag, soil and fly ash, coal burning,

silicomanganese, quartz, syngenite, S-bearing iron, calcium rich, gypsum, sphalerite, dolomite, iron, alloy, lead sulfate, zinc

rich, sulfur-rich particles and aluminum manufacturing dust. The majority of the particles in the studied size range are of

anthropogenic origin, especially from metallurgical industry. The study illustrates the complexity of particle pollution in air of

an industrial Chinese city and the results should be useful in planning mitigation measures.

D 2004 Elsevier B.V. All rights reserved.

Keywords: Air pollution; Individual particle analysis; Hierarchical cluster analysis; PM10; SEM–EDS

1. Introduction

Particulate matter with aerodynamic diameters less

than 10 Am (PM10) has been found to cause health

problems. It can trigger or exacerbate conditions, such

as asthma, emphysema, bronchitis, silicosis and lung

cancer (e.g., Anderson et al., 1992; Dockery and

Pope, 1994; Aunan, 1996). Possible causes of these

effects include particle composition–for example, they

may contain soluble transition metals such as copper,

iron, vanadium, nickel or zinc–their acidity and their

ultrafine size (Lighty et al., 2000). PM10 has also been

related to problems such as reduced visibility and

unpleasant odour.

Particulate air pollution is a serious problem

around the world, especially in less-industrialized

countries where the early stages of industrial growth

0048-9697/$ - see front matter D 2004 Elsevier B.V. All rights reserved.

doi:10.1016/j.scitotenv.2004.10.012

* Corresponding author. Tel.: +47 22855412; fax: +47 22855441.

E-mail addresses: xieruik@yahoo.com, ruikaixi@kjemi.uio.no

(R.K. Xie).

Science of the Total Environment 343 (2005) 261–272

www.elsevier.com/locate/scitotenv

are often pursued without much investment in

environmental protection, leading to heavy air

pollution in urban areas (Florig et al., 2002).

China’s rapid industrialization, urbanization and

economic growth are contributing to respiratory

diseases and chronic illnesses. Levels of particulate

air pollution from energy and industrial production

in several of China’s megacities, such as Shanghai

and Shenyang, are among the highest in the world

(World Resources Institute, 1998). The increasing

number of motor vehicles will inevitably contribute

to the complexity of the particulate pollution, which

has been largely emitted from the energy and

industrial sectors.

Chemical characterization of airborne particles in

Chinese cities has been largely limited to analysis of

bulk samples (Wei et al., 1999; He et al., 2001) and

analysis of a small number of individual particles

(Yang et al., 1996; Zhang et al., 2003). Compared

with bulk sample analysis, individual particle anal-

ysis has several advantages. It can give definitive

information about surface coatings, elemental asso-

ciations, agglomerations and detailed variation of

composition with particle size (Post and Buseck,

1984). For example, using individual particle anal-

ysis, it is possible to reveal whether an element

occurs as a major component in a small fraction of

the particles or as a minor element in many particles.

In bulk sample analysis, these characteristics are

largely masked. It is generally necessary to analyze a

large number of bulk samples in order to deduce the

pollution sources by multivariate techniques, while

by individual particle analysis, a limited number of

representative samples are usually enough to extract

a fairly well resolved source profile. Scanning

Electron Microscope with Energy-Dispersive Spec-

trometer (SEM–EDS) or similar techniques, such as

electron microprobe X-ray analysis, has been suc-

cessfully applied in many aerosol studies of individ-

ual particles to identify different particles and link

them to possible sources (Katrinak et al., 1995;

Paoletti et al., 1999; Ebert et al., 1999).

Guiyang has long been listed as one of the most

seriously polluted cities in China. Mass concen-

trations of total suspended particles (TSP), together

with some other parameters, have been routinely

monitored at five locations for many years

(Guiyang Monitoring Station, 2001). Yet, little is

known about the airborne particulates in terms of

physical and chemical characteristics and origins.

Lack of such information hinders human health risk

assessments and prevents development of a rational

abatement strategy. Therefore, the objectives of this

study are: (a) by using the automated SEM–EDS,

to determine the composition of different particles;

and (b) by applying hierarchical cluster analysis

(HCA) to distinguish different types of particles

and, as far as possible, to identify the pollution

sources.

2. Site descriptions

Guiyang, the capital of Guizhou Province (Fig. 1),

is located nearly 1100 m above sea level. The total

urban area is 8084 km2 and the population 1.56

million. It features an undulating topography, with

high mountains and plateaus from north to south, hills

and river valleys in the centre. The soil in Guiyang

area is developed on widely distributed bedrocks of

dolomite and limestone. It has a subtropical monsoon

climate with an annual average temperature of 15.6

8C, relative humidity 77%, and an average annual

rainfall of 1177 mm (Guiyang Monitoring Station,

2001). The prevailing wind direction is northeasterly

in winter, and southerly in summer. The forest and

bush coverage exceeds 20%. Due to the topography of

Guiyang city, thermal inversion occurs at a frequency

of 11.9% at the 0–100 m level, during which air

pollution increases dramatically as a mass of cold air

is trapped below warmer air (Guiyang Monitoring

Station, 2001).

Samples were collected on the roof of the five-

storey office building of the Guizhou Institute of

Environmental Science in the southeastern part of

Guiyang city. The site is situated in a residential

area. There are several big point sources around this

site (see Fig. 1): an iron and steel plant about 1.8

km and a silicomanganese smelter about 28 km

away, both to the southeast; a coal-fired power plant

and a cement plant, 4.2 and 9 km away, respec-

tively, to the southwest; an aluminum work 15 km

to the northwest; some other small-scale silico-

manganese smelters are scattered in the north. In

addition, some small industrial point sources and

some small industrial coal-fired boilers can also

R.K. Xie et al. / Science of the Total Environment 343 (2005) 261–272262

contribute to the pollution at the site depending on

factors, such as emission intensity and meteorolog-

ical conditions. It is estimated that the number of

motor vehicles will be approximately 170,000 in

Guiyang in 2004, and the number is still on the rise

(Guiyang Municipal Government, 2004). The five-

year averages (1996–2000) of TSP, SO2 and NOx

were 0.209, 0.177 and 0.034 mg/m3, respectively,

TSP and SO2 showed a declining trend (Guiyang

Monitoring Station, 2001).

3. Experimental

3.1. Sampling

A medium volume (100 l min�1) sampler (Wuhan

Tianhong Intelligent Instrument Plant) was used with

an impactor cutoff of 10 Am (aerodynamic diameter).

Five daily samples were collected. Individual particles

were sampled on 90 mm diameter, 1 Am pore size

FluoroporeR (PTFE) membrane filters. The sampler

Fig. 1. Location of sampling site. (1) The zinc smelter 3 km to the southeast. (2) The silicomanganese plant 28 km to the southeast. (3) Guizhou

Cement Plant 9 km to the southwest. (4) The aluminum plant 15 km to the northwest.

R.K. Xie et al. / Science of the Total Environment 343 (2005) 261–272 263

was programmed to collect particles for 10 min every

hour, with a total of 4 h each day. In this way, there

were enough particles distributed on the surface while

undesirable overloading was avoided. To measure

particulate mass concentration, 24-h PM10 samples

were simultaneously collected on Millipore AP40

glass fiber filters with a diameter of 90 mm. Daily

PM10 mass concentration was obtained by calculating

the ratio of the net mass of particulates to the standard

air volume. The sampling conditions and mass

concentration are shown in Table 1.

3.2. SEM–EDS analysis

Single-particle analysis was performed using a

Hitachi S-4300SE-Shotky Field Emitter Scan Electron

Microscope. This instrument is equipped with an

Oxford 6853 energy dispersive spectroscopic detector,

and is run by Oxford Instruments INCA Energy and

Feature software.

During analysis, the electron beam scans over a

preset area. When the backscattered electron inten-

sity of the particle contour exceeds a predefined

threshold value, the considered object is detected.

An X-ray spectrum from the centre of the particle

can be accumulated and subsequently the spectrum

is processed automatically with a built-in XPP

correction procedure. The XPP program is a single

universal correction procedure that outperforms ZAF

and phi–rho–z correction procedures required for

quantitative electron microprobe analysis (Pouchou

et al., 1990). Elements with atomic numbers less

than 11 are not determined due to insufficient

accuracy and the presence of fluorine and carbon

in the substrate. The equivalent circular diameter

(ECD), defined as the diameter of a circle with the

same area as the projected particle, and the shape

factor, defined as the ratio of the square of the

perimeter to 4p times the area, are also measured. In

the case of a spherical particle, the two-dimensional

projection is a circle, and the shape factor is 1. The

more irregular the particle, the higher is the shape

factor. For each sample, a 1-cm2 portion was cut

from the centre of the Teflon filter and mounted with

a thin film of colloidal carbon directly on a smooth

aluminium stub. For better conductivity and there-

fore reduction of electron charge, the sample was

painted on the edge and coated with a carbon layer

about 20 nm thick. About 500 particles were

automatically analyzed for each sample (i.e., par-

ticles collected during 24 h sampling). The magni-

fication was set to 2000, which made each detection

field an area of 50�60 Am2. About 20 fields were

analyzed with 20–30 particles on each. Particles with

equivalent circular diameter smaller than 0.19 Amwere ignored. The X-ray spectrum was accumulated

for 15 s with a beam current of 0.3 nA and

acceleration voltage of 15 kV. Analysis of one

sample takes less than 3 h. The relative percentage

of the elements was normalized so that the sum of

the included elements is 100. The analysis’ precision

and accuracy were checked by analyzing laboratory-

prepared fine particles: NaCl and KCl solutions were

deposited on the polycarbonate filters, representing a

wide size range from submicrons to microns. The

95% confidence interval for atomic fraction for Na

in NaCl and K in KCl are 49.3–51.9 and 47.4–48.2,

respectively, indicating a slight bias for K.

In addition, SEM photomicrographs of some of

the particle types (see Fig. 2) were manually

obtained by a JEOL JSM-6460LV high-performance

SEM–EDS. The naming of the particles was based

on both SEM morphological features and EDS

chemical composition.

Table 1

Meteorological and PM10 mass concentration data in Guiyang during sampling days

Sampling date Precipitation

(mm)

Average

temperature

(8 C)

Average relative

humidity (%)

Prevailing wind

direction

Average wind

speed (m/s)

Mass

concentration

(Ag/Nm3)

24, 07, 2003 11 21 80 Southeasterly 3.9 45

25, 07, 2003 16 21 80 Southeasterly 5.0 45

26, 07, 2003 1.4 22 80 Southeasterly 2.9 58

27, 07, 2003 0 23 80 Southeasterly 3.4 84

28, 07, 2003 0 24 75 Southeasterly 3.6 64

R.K. Xie et al. / Science of the Total Environment 343 (2005) 261–272264

Fig. 2. SEM photomicrographs of airborne particles (on the rough side of 3 Am Teflon filter, sampled for 24 h) in Guiyang. (a) Silicomanganese

slag. (b) Soil-derived kaolinite agglomerates. (c) Soil-derived illite, mainly Si, Al and K. (d) A fly ash particle. (e) Coal burning particles mainly

composed of Ca, S and Si. (f) A silicomanganese particle, mainly Si and Mn. (g) A quartz particle from coal combustion (SiO2). (h) Syngenite

agglomerates. (i) An iron particle. (j) Aragonite aggregates (CaCO3). (k) Agglomerates of hydrated lime (Ca(OH)2). (l) Gypsum aggregates

(CaSO4. 2H2O). (m) Dolomite (CaMg(CO3). (n) Soot aggregates. (o) A biological particle, a pollen or a spore.

R.K. Xie et al. / Science of the Total Environment 343 (2005) 261–272 265

3.3. Data reduction strategy

The five daily samples, each contributing about

500 particles, were considered as a lumped sample

(altogether about 2300 particles and 14 elements,

about 200 particles of zero counting were omitted,

possibly carbonaceous, biological particles and sub-

strate artefacts). With such a relatively large data set,

some form of data reduction is needed. Hierarchical

cluster analysis (HCA) can greatly reduce the data

while retaining the most important information

embedded in the raw data.

Shattuck et al. (1985) applied hierarchical cluster

analysis to identify atmospheric particle types in the

urban aerosol of Phoenix, AZ. HCA was also used by

Bernard et al. (1986) to classify a big set of airborne

particle data from an estuarine area. Other applications

to aerosol data were presented by Xhoffer et al.

(1991), and Van Malderen et al. (1992, 1996, 2001).

The technique has been proved to be very useful.

The purpose of cluster analysis is to sort particles

into groups, or clusters, so that the degree of

association is strong between members of the same

cluster and weak between members of different

clusters. We use hierarchical cluster analysis to

identify the particle types according to their chemical

composition. The MINITAB statistics software is a

powerful tool (http://www.minitab.com/). A data

matrix (as large as several thousand observations by

14 variables) can be executed easily within few

minutes. The squared Euclidean distance is used to

measure the similarity of the particles. Each particle

(i) is represented by an object vector with the

components (Xi ,1, Xi ,2,. . .Xi ,n) representing the

weight percentage of the n chemical elements. The

squared Euclidean distance of particle i and particle j

is defined as:

d2ij ¼Xn

k¼1

ðXik � XjkÞ2 ð1Þ

where the index k represents the considered element.

Ward’s method (Van Born and Adams, 1988) was

applied on the relative percentage of the elements

based on the covariance matrix.

A major problem with this method is to obtain the

right number of clusters to reflect the btrueQ particletypes. A too low number will lead to clusters with

particles belonging to different groups, while a too

high number may cause split of inherently bidenticalQparticles. This can be overcome by setting a bcutoffQto the dendrogram based on Akaike’s criterion used

by Bondarenko et al. (1994) on Lake Baikal aerosol

particles. However, we found it satisfactory to

perform a stepwise hierarchical cluster analysis and

judge the best number of clusters by comparing intra-

and intercluster bdistancesQ and the possibility of

interpreting the cluster pattern in terms of sources.

4. Results and discussion

The hierarchical cluster analysis applied to about

2300 samples (14 variables) yielded to 17 different

particle types (Table 2). Since light elements, such as

C, N and O, were not included, some important

particle types, such as soot, organic carbon and

particles of biological origin, could not be identified.

The nature and origins of other particles are addressed

in detail below.

4.1. Silicomanganese slag particles (abundance

23.6%)

Silicomanganese is present as an alloying element

in the steel industry as well as raw material for the

production of medium and low carbon ferromanga-

nese. It is produced by direct reduction of manganese

ore and quartzite by coke and coal with the aid of flux

materials. Particles produced in this process contain

predominantly silicon and have moderate amounts of

sulfur, manganese and potassium. They are identified

as silicomanganese slag particles (Fig. 2a). It is the

most abundant particle type in the air of Guiyang.

They originate from the silicomanganese metallurgi-

cal plants, the largest one being 28 km away to the

southeast (Fig. 1).

4.2. Soil and fly ash particles (abundance 12.6%)

This particle type is the second most abundant.

These particles contain mainly aluminum and silicon,

with varying amounts of magnesium, potassium,

calcium and iron. The composition of single particles

shows that some of them contain sulfur. Airborne soil

particles originate from windblown soil dust and

R.K. Xie et al. / Science of the Total Environment 343 (2005) 261–272266

resuspension of dust from roads. Fly ash particles

originate from various kinds of combustion processes

(coal-fired power plants, boilers, metallurgical plants,

traffic and other urban sources). We put these two

types of particles in the same class since they are

chemically similar and hence cannot be distinguished

by composition only. However, they can sometimes

be identified by shape factor analysis (Kindratenko et

al., 1994). The soil particles normally have pro-

nounced irregular shapes (Fig. 2b,c). The fly ash

particles are, in most cases, nearly spherical (Fig. 2d)

and according to the definition, the shape factor

should be close to 1. But there are some exceptions, as

observed through optical microscope by Mamane et

al. (1986). They found that fly ash particles could be

classified into 11 shape classes. Over 95% of the fly

ash particles were smoothly spherical. The particle

shape of fly ash depends much on the combustion

condition. Normally, spherical particles are derived

from fluid melts due to high-temperature combustion.

The precise distinction of soil and fly ash using the

shape factor may also be hindered by coating of small

particles. As a rough approximation from the shape

factor and visual inspection, the fly ash amounts to

about 10% of the particles in this cluster.

4.3. Coal burning particles (abundance 10.4%)

Particles in this cluster contain mainly Si, S and

Ca, but also minor amounts of Mg, Al, K, Fe, Mn

and some other elements (Fig. 2e). This combination

of elements (gypsum/aluminosilicate) may result

from sulfation of limestone in coal combustion.

The particles may be formed either in the slag

containing Al, Si and Ca, or gypsum may be formed

first which is then adhered to or surrounded by

melted Ca–Al–silicates (Yoshihiko et al., 2001). The

origin is most likely coal combustion in local boilers

and/or a coal-fired power plant 4.2 km to the

southwest.

4.4. Silicomanganese particles (abundance 10.3%)

Particles mainly consisting of silicon and manga-

nese with nearly the same percentage are classified as

silicomanganese particles (Fig. 2f). These particles are

most probably emitted from some ferroalloy plants,

such as silicomanganese smelters (Gunst et al., 2000),

or from steel making in the nearby iron and steel plant

when silicomanganese is added to the steel as

deoxidizer to lower oxygen content.

Table 2

Particle types

Identity Particle abun.% ECDa

(Am)

Shape

factor

Elemental weight percentage (normalized to 100%)

Na Mg Al Si P S Cl K Ca Mn Fe Zn Mo Pb

Silicomanganese

slag particles

23.6 0.81 1.61 1.5 0.9 1.5 47.9 0.3 20.1 0.4 11.5 1.3 14.1 0.4 0.4 0.1 0

Soil and fly ash 12.6 0.94 2.05 0.7 1.9 17.5 45.1 0.9 8 0.2 8.9 4.5 2.6 9 0.4 0.4 0

Coal burning particles 10.4 1.33 1.99 0.4 2.3 6.2 22.9 1.3 16.3 1.2 6.2 35.3 3.7 3.6 0.4 0.3 0

Silicomanganese 10.3 0.87 1.54 0.8 1.6 2 36.5 0.1 7.5 0.3 7.8 1.5 40.9 0.2 1 0 0

Quartz 7.7 0.71 1.59 0 0 0.2 99.3 0 0.3 0 0.2 0 0 0 0 0 0

Syngenite 6.6 1.06 1.99 1.4 0.4 0.6 8.3 0.1 32.7 0.6 36.4 16.4 2.7 0.2 0.1 0 0

S-bearing iron 5.9 0.77 1.6 2.2 1.7 0.9 8.8 0.5 14.8 0.4 5.2 0.9 0.5 58.5 5.1 0.3 0

Calcium-rich 4.3 1.34 2.01 0 1.4 1.8 9.6 0.3 4.8 0.1 1.3 79.4 0.5 0.3 0 0.4 0

Gypsum 4.2 0.61 2.21 0.4 0.6 1.8 8.3 2.4 38.3 0.1 5.5 42.2 0.2 0 0.3 0 0

Sphalerite particles 3.8 0.64 1.78 0 0.6 0.3 6.3 1.7 22.2 2.3 10.8 1.2 0.8 17.7 35.5 0.6 0

Dolomite 3.3 1.18 2.23 0.1 26.9 2.5 10.7 0.6 8.9 0.4 2.9 44 0.9 1.2 0.9 0 0

Iron 2.6 0.69 1.36 0.1 0.6 0.1 2.7 0.1 1.8 0.1 0.6 0 0.1 93.7 0 0.2 0

Alloy particles 1.8 0.57 1.27 0.8 0.3 1.4 36.9 0.1 0 0 6.1 1.4 15.5 1 0 36.5 0

Lead sulfate 1 0.49 1.25 1.2 0 0 2.3 1.6 20.9 0 4.8 0.4 0 1 2.3 0 65.5

Zinc-rich particles 0.8 1.09 1.42 0 0 0 0.8 0.2 3.7 3.1 0.4 0.2 0 5.4 86.2 0 0

Sulfur-rich particles 0.8 0.94 2.12 0 0 0 0 0 100 0 0 0 0 0 0 0 0

Aluminum

manufacturing dust

0.3 0.52 2.26 1.4 0 90.2 2 0.7 4.4 0 1.4 0 0 0 0 0 0

a ECD: equivalent circular diameter.

R.K. Xie et al. / Science of the Total Environment 343 (2005) 261–272 267

4.5. Quartz particles (abundance 7.7%)

Particles containing predominantly silicon are

classified as quartz. As an important constituent of

many rock types, quartz is almost ubiquitous on land

areas. It can be induced by resuspension of road

particles or from construction sites; coal combustion is

another important source. A mechanism for particle

formation from combustion was suggested by Li and

Winchester (1993). They assumed reduction of SiO2

at high temperature inside the furnace followed by

evaporation, oxidation and nucleation processes. Our

SEM photograph (Fig. 2g) shows agglomerates of

small SiO2 particles differing from natural particles

(McCrone and Delly, 1973, p648, p658), which are in

the form of sharp, angular fragments, chips and flakes.

The particles in this cluster are very pure (on average

99.3%, oxygen excluded), consistent with the above

formation mechanism.

4.6. Syngenite particles (abundance 6.6%)

This cluster consists of particles containing a

mixture of calcium, potassium and sulfate. It is

morphologically well-formed crystals, looking like

tablets (Fig. 2h). The composition is close to that of

syngenite (K2SO4d CaSO4d H2O). To our knowledge,

the presence of syngenite-like particles in ambient

air has never been reported. The precise origins are

not known. It may be formed in coal combustion,

similar to the formation of gypsum, when K- and

Ca-containing coal is burned. Another possible

source may be cement production, when cement is

used to make mortar at construction sites, the

syngenite, formed during its storage, may be

released to the air during unloading, unpacking

and blending processes.

4.7. Iron and S-bearing iron particles (abundance

2.6% and 5.9%, respectively)

Iron particles have natural and anthropogenic

origins. However, in our study, iron particles (Fig.

2i) from ferrous metallurgic industry are likely to

dominate completely since there is a big iron and

steel plant only 1.8 km away from the sampling

site. S-bearing particles, on the other hand, may be

partially emitted from ferrous metallurgical indus-

try, and partially from high S-containing coal

combustion, in the form of pyrite and/or iron

sulfate.

4.8. Calcium-rich particles (abundance 4.3%)

These particles are probably limestone (CaCO3)

with some impurities, mainly silicon, sulfur and

aluminum. The sulfur content is assumed to be the

result of reactions between limestone particles and

sulfur dioxide and sulfuric acid. Limestone is widely

distributed as a crustal mineral and is found in soil

dust and road dust. The Guiyang area is very rich in

limestone. Limestone particles are generated through

its application in many areas, e.g., as building

material, in cement manufacturing and in metallurgic

industry. Quarrying, crushing and mixing processes

can also generate airborne particles. Some spherulitic

aggregates of elongated prismatic crystals (Fig. 2j)

contain almost only Ca of the elements determined

by SEM. These particles are clearly aragonite, one of

the polymorphs of calcium carbonate. Other particles

with similar composition but different in structure

from both calcite and aragonite, were also found

(Fig. 2k). They are most likely hydrated lime

particles. Their origins can be tracked to limekilns,

in which lime is produced for application in metal-

lurgical industry and building construction, and to

coal combustion.

4.9. Gypsum (abundance 4.2%)

The presence of gypsum particles (Fig. 2l) is

common and can be traced to many sources. The

particles can originate from weathering of gypsum-

bearing rocks in nature or from buildings made of

such material. Transformation of airborne limestone

by sulfuric attack is also an option. However, in

Guiyang, the largest fraction is presumably from

coal combustion. Three mechanisms may be

proposed regarding the gypsum emission in our

case. The first is direct emission of gypsum

contained in coal when coal is pulverized and

burnt. The second is the so-called bdesulfurizationQreaction between the sulfur and the limestone in

the combustion process. The third is reaction

between the emitted SO2 and airborne limestone

particles.

R.K. Xie et al. / Science of the Total Environment 343 (2005) 261–272268

4.10. Sphalerite particles (abundance 3.8%)

Particles containing mainly sulfur, zinc and iron are

classified as sphalerite (Zn, Fe)S. The particle shape

resembles the crystal habit of sphalerite. It is most

probably emitted from the nearby zinc smelter, which

is 3 km away to the southeast.

4.11. Dolomite (abundance 3.3%)

It is no surprise to find dolomite particles for this

mineral is abundant in the Guiyang area. It is mainly

composed of Ca and Mg as seen from the chemical

formula CaMg(CO3)2 (Fig. 2m). However, it may

contain some impurities, such as Al, Si, Fe, S, etc.

Dolomite finds many applications. It has been used in

metallurgical, chemical and glass industry, in fertil-

izers, ceramics and as filler in paper and plastic

industry. It can also be used in building materials,

such as cement and refractory bricks.

4.12. Alloy particles (abundance 1.8%)

This cluster is characterized by a combination of

Si, Mn and Mo. In steel making industry, these

metals are often added as deoxidizers. Individual

particle data inspection shows that some of the

particles have a shape factor near unity, indicating

that they are molten products from high-temperature

furnaces. They are most probably emitted from ladle

refining and stirring process in the iron and steel

plant.

4.13. Lead sulfate (abundance 1.0%)

Tetraethyl lead was often added to petrol to

increase octane number. 1,2 Dibromoethane was

also added to remove the lead from the cylinder as

PbBr2, a vapour at the cylinder temperature. Thus,

lead is mainly emitted from leaded fuels as PbBr2.

Since we do not find Br (or Cl) in these particles, it

is not likely that the origin of Pb particles is motor

vehicles which is in agreement with the fact that the

use of leaded petrol is being phased out in China.

The lead sulfate particles are most likely from the

zinc smelter (Fig. 1), as lead is often associated

with zinc. They are also possibly from some

foundries.

4.14. Zinc-rich particles (abundance 0.8%)

Anthropogenic zinc particles may be released to the

atmosphere as dust and fumes from mining, zinc

production facilities, processing of zinc-bearing raw

materials (e.g., lead smelters), iron and steel production,

coal combustion and refuse combustion. The zinc smelter,

which has been mentioned previously, is probably

responsible for the presence of this particle type.

4.15. Sulfur-rich particles (abundance 0.8%)

In this particle type, only S was detected by the

SEM–EDS analysis. These particles are probably

mainly (NH4)2SO4. Anthropogenic emissions, like

combustion of fossil fuel, present the main sources of

SO2 which is oxidized to SO42�. Ammonia is often

from agricultural sources (animal husbandry, soil

disturbance and fertilizer application), biomass burn-

ing, industry, refrigeration units and coal combustion.

H2SO4 is then neutralized or partially neutralized by

ammonia present in the atmosphere. Xhoffer et al.

(1991) suggested that S-rich compounds have prob-

ably condensed onto or reacted with existing carbona-

ceous particles that have acted as condensation nuclei.

4.16. Aluminum manufacturing dust (abundance

0.3%)

These particles contain mainly Al, with some

minor elements (Na, Si, P, S and K). Its main

composition is alumina (Al2O3), which is a refined

bauxite mineral. It mostly comes from the aluminum

smelter located in the northwest, 15 km away. In the

manufacture process, alumina particulates can be

emitted from alumina handling and processing, when

bauxite is digested by a sodium hydroxide and further

calcined in rotary kilns, and from the smeltering

process as well, when alumina is dissolved in a bbathQof molten cryolite (sodium aluminum fluoride) and

reduced to metallic aluminum.

Owing to the restriction of the instrument, some

other important particles may have been missed. SEM

inspection revealed quite a few soot aggregates (Fig.

2n) and some biological particles (Fig. 2o). Soot

consists of very fine particles of black carbon. Soot is

generated from traffic, industry, outdoor fires and

R.K. Xie et al. / Science of the Total Environment 343 (2005) 261–272 269

household burning of coal and biomass fuels. The size

of soot particles depends on temperature and other

burning conditions. Airborne biological particles

consist of viruses, bacteria, fungi, pollen, dander and

other material of animal or plant origin, or their

components or products.

Si and S were found in more than 1 wt.% in 15 of

the 17 clusters. Of the metals, K and Ca were found in

most clusters, Al, Mn and Fe also occurred frequently.

The high occurrence frequency is probably related to

the fact that all these elements are among the 15 most

abundant elements in the Earth’s crust. The especially

high percentage of S occurrence may be explained by

the use of sulphur-containing coal or coke as the

major energy source for most of the industries.

The largest percentage of the particles are from

industrial processes. These particles are silicomanaga-

nese slag particles (23.6%), silicomanganese particles

(10.3%), S-bearing iron particles (5.9%), sphalerite

particles (3.8%), iron particles (2.6%), alloy particles

(1.8%), lead sulfate paricles (1%), zinc rich particles

(0.8%) and aluminium manufacturing particles

(0.3%). The total is 50.1%, even neglecting that

particles in some of the other groups are partially of

industrial origins. The second largest percentage of

the particles is from coal combustion. Coal burning

(10.4%), gypsum (4.2%) and fly ash (1.3%) constitute

15.9% of all particles. Also some of the particles in

other clusters, such as quartz and syngenite, may

partly be from coal combustion. It is different from

findings in some studies elsewhere in China. Qiu et al.

(2001) used the micro-PIXE spectra of single aerosol

particles to identify their origins in Shanghai, a

megacity in east China. They found that the major

contributors to the atmospheric aerosol were soil dust

(31.6%), building dust (30.8%), followed by vehicle

exhaust (13.7%), the metallurgical industry (5.6%), oil

combustion (5%) and coal combustion (2.3%). Yang

and Chen (2002) applied factor analysis to identify the

dust sources in Lanzhou, the capital of Gansu

Province and the largest industrial city in the north-

west. They concluded that about 41.0% of the dust

came from coal combustion, 23.0% from windblown

sand and soil, while motor vehicle emissions and

building materials accounted for 18.7% and 12.8%,

respectively, and 4.5% of the dust was from other

sources. In Guiyang, neither soil dust nor coal

combustion gets the lion’s share. The differences

and the large fraction of particles of industrial origin

in Guiyang may be explained by the following

factors:

(1) In the north, soil particulate matter is generally a

large part of the total suspended particles (TSP),

with an average of 40% to 50% due to the dry

climate and low level of forestation (He et al.,

2002). In the dusty season in spring, the TSP

level can be several hundred times higher than

normal (Wang et al., 1982; Ta et al., 2003). In

Guiyang, a subtropical region remote from

sandstorm areas with humid climate and high

vegetation coverage, the soil contribution to

airborne particles is relatively low. In our case,

the soil particle number amounts to only 11.3%

of the total (obtained by subtracting fly ash

particles).

(2) The fraction of coal combustion particles

(15.9%) is fairly low considering that coal

combustion was once listed as the primary

contributor to atmospheric pollution in China

(Zhao and Sun, 1986), and as a result, Guiyang

was named in 1998 by the World Health

Organization (WHO) as one of the 10 worst

polluted cities in the world. However, things are

changing for the better now. The average SO2

ambient concentration has been reduced from

0.300 mg/Nm3 in 1996 (Guiyang Monitoring

Station, 2001) to 0.103 mg/Nm3 in 2002

(Guiyang Municipal Government, 2003); in

1996, the average TSP concentration was

0.253 mg/Nm3 (Guiyang Monitoring Station,

2001), while the PM10 concentration in 2002

was 0.106 mg/Nm3 (Guiyang Municipal Gov-

ernment, 2003). PM10 makes up about 60% of

TSP in China (Peng et al., 2002). This positive

development came as the result of years of

unremitting struggle against the air pollution.

Several important prevention and abatement

strategies have been adopted, including indus-

trial structural adjustment, energy restructuring

and enhancement of abatement capacity at end-

of-pipe. The most effective measures are the

ban of coal-fired boilers of less than 2 tons

steam per hour, the demolishing of two sets of

50,000-kW generator units in the power plant

and the shift to cleaner energy, such as natural

R.K. Xie et al. / Science of the Total Environment 343 (2005) 261–272270

gas, coal gas and electricity. Although these

were all primarily aimed at reducing emission

of sulfur dioxide, the particulate emission was

brought down as a cobenefit of the sulfur

control. The integrated preventive and mitigat-

ing measures have led to a considerable

reduction of coal combustion particles, hence

resulted in reduction of TSP and PM10.

(3) The sampling site in Guiyang is close to many

industrial installations and several of them are

located to the southeast so the meteorological

conditions during the experiment favoured

transport of particulates from these sources.

Guiyang is rich in diversified mineral and

energy sources; it is an important industrial

base in China, prestigious for its ferrous and

nonferrous metallurgy. These industries are all

energy-material-waste intensive, generating rel-

atively large amounts of ambient particles,

which pose a potential threat to human health.

(4) The absolute mass concentration of PM10 is

rather low during this sampling campaign

(Table 1). It is far below the EPA 24-h standard

of PM10, 150 Ag/m3 (US EPA, 1997). In

drawing conclusions from the study, it must be

taken into consideration that the sampling took

place over a short period in the rainy season.

The rain that occurred within the first days had a

scavenging effect on particles in the air prob-

ably removing a larger fraction of natural

particles than of anthropogenic ones since the

former are generally larger. Furthermore, wet

ground reduced suspension of soil particles.

The trace metals concentration could not be

determined by the SEM–EDS technique. The study

will therefore be supplemented with determination of

heavy metals by Inductively Coupled Plasma–Mass

Spectrometry (ICP–MS).

5. Conclusions

Automated SEM–EDS combined with cluster

analysis proved to be very useful for studying particles

in ambient air in Guiyang. The mix of particles was

very complex with a large number of sources.

However, during the sampling period, anthropogenic

sources clearly dominated, contributing to well above

66% of the particles (industrial processes+coal com-

bustion origin). The share of natural airborne particles,

such as soil particulates, may have been affected by

scavenging by rain and reduced resuspension due to

ground wetting. However, considering the extensive

ferrous and nonferrous industries in the area, the

humid climate and high forest coverage, it is reason-

able that the number abundance is dominated by

anthropogenic sources. The results give useful back-

ground information for effective countermeasures, but

ought to be supplemented with studies during other

seasons and meteorological conditions and with other

analytical methods.

Acknowledgement

We are grateful to Professor Tom Andersen from

the Department of Geosciences, University of Oslo,

for the identification of some of the particles.

References

Anderson KR, Avol EL, Edwards SA, Shamoo DA, Peng RC, Linn

WS, et al. Controlled exposure of volunteers to respirable

carbon and sulfuric acid aerosols. J Air Waste Manage 1992;42:

770–6.

Aunan K. Exposure–response functions for health effects of air

pollutants based on epidemiological findings. Risk Anal

1996;16:693–709.

Bernard PC, Van Grieken R, Eisma D. Classification of estuarine

particles using automated electron microprobe analysis and

multivariate techniques. Environ Sci Technol 1986;20:467–73.

Bondarenko I, Van Malderen H, Treiger B, Van Espen P, Van

Grieken R. Hierarchical cluster analysis with stopping rules built

on Akaike’s information criterion for aerosol particle classi-

fication based on electron probe X-ray microanalysis. Chemo-

metr Intell Lab Syst 1994;22:87–95.

Dockery DW, Pope CA. Acute respiratory effects of particulate air

pollution. Annu Rev Public Health 1994;15:107–32.

Ebert M, Weinbruch S, Hoffmann P, Ortner HM. Chemical

characterization of North Sea aerosol particles. J Aerosol Sci

1999;31:613–32.

Florig HK, Sun GD, Song GJ. Evolution of particulate regulation in

China—prospects and challenges of exposure-based control.

Chemosphere 2002;49:1163–74.

Guiyang Monitoring Station. Environmental Quality Report-

Guiyang, 1996–2000. Guiyang, Guizhou, China; 2001.

Guiyang Municipal Government. Guiyang environmental status

bulletin. Guiyang, Guizhou, China7 The Environmental Protec-

tion Bureau of Guiyang Municipality; 2003.

R.K. Xie et al. / Science of the Total Environment 343 (2005) 261–272 271

Guiyang Municipal Government, 2004. http://www.gygov.gov.cn/

jumpnews/JRGY/04-02-17f.htm.

Gunst S, Weinbruch S, Wentzel M, Ortner HM, Skogstad A,

Hetland S, et al. Chemical composition of individual aerosol

particles in workplace air during production of manganese

alloys. J Environ Monit 2000;2:65–71.

He KB, Yang FM, Ma YL, Zhang Q, Yao XH, Chan CK, et al. The

characteristics of PM2.5 in Beijing, China. Atmos Environ

2001;35:4959–70.

He KB, Huo H, Zhang Q. Urban air pollution in China: current

status, characteristics, and progress. Annu Rev Energy Environ

2002;27:397–431.

Katrinak KA, Anderson JR, Buseck PR. Individual particle types in

the aerosol of Phoenix, Arizona. Environ Sci Technol

1995;29:321–9.

Kindratenko VV, Van Espen PJM, Trelger BA, Van Grieken R.

Fractal dimensional classification of aerosol particles by

computer-controlled scanning electron microscopy. Environ

Sci Technol 1994;28:2197–202.

Li SM, Winchester JW. Aerosol silicon and associated elements in

the Arctic high and mid-troposphere. Atmos Environ, A Gen

Topics 1993;27A:2907–12.

Lighty JS, Veranth JM, Sarofim AF. Combustion aerosols: factors

governing their size and composition and implications to human

health. J Air Waste Manage 2000;50:1565–618.

Mamane Y, Miller JL, Dzubay TG. Characterization of individual

fly ash particles emitted from coal- and oil-fired power plants.

Atmos Environ 1986;20:2125–35.

McCrone WC, Delly JG. The particle atlas. edition two. Michigan,

USA7 Ann Arbor Science Publishers; 1973.

Paoletti L, Diociaiuti M, Berardis BD, Santucci S, Lozzi L, Picozzi

P. Characterization of aerosol individual particles in a controlled

underground area. Atmos Environ 1999;33:3603–11.

Peng CY, Wu XD, Liu G, Johnson T, Shah J, Guttikunda S. Urban

air quality and human health in China. Urban Stud

2002;39:2283–99.

Post JE, Buseck PR. Characterization of individual particles in the

Phoenix urban aerosol, using electron beam instruments.

Environ Sci Technol 1984;18:35–42.

Pouchou JL, Pichoir F, Boivin D. The XPP procedure applied to

quantitative EDS X-ray analysis in the SEM. Microbeam Anal

1990;25:120–6 [San Francisco].

Qiu ZJ, Lu RR, Guo PL, Wang JQ, Qiu HY, Li XL, et al. Source

apportionment of single aerosol particles in the atmosphere of

Shanghai city. Nucl Sci Tech 2001;12:215–23.

Shattuck TW, Germani MS, Buseck PR. Cluster analysis of

chemical composition of individual atmospheric particle data.

Am Chem Soc Symp 1985;292:118–29 [Environmental appli-

cations of chemometrics].

Ta WQ, Xiao Z, Qu JJ, Yang GS, Wang T. Characteristics of dust

particles from the desert/Gobi area of northwestern China during

dust-storm periods. Environ Geol 2003;43:667–79.

US EPA, National Ambient Air Quality Standards (NAAQS), 1997.

http://www.epa.gov/air/criteria.html.

Van Born WA, Adams FC. Cluster analysis of electron microprobe

analysis data of individual particles for source apportionment of

air particulate matter. Atmos Environ 1988;22:229–37.

Van Malderen H, Rojas C, Van Griken R. Characterization of

individual giant aerosol particles above the North Sea. Environ

Sci Technol 1992;26:750–6.

Van Malderen H, Van Grieken R, Khodzher T, Obolkin V, Potemkin

V. Composition of individual aerosol particles above Lake

Baikal, Siberia. Atmos Environ 1996;30:1453–65.

Van Malderen H, Hoornaert S, Injuk J, Przybylowicz WJ, Pineda

CA, Prozesky VM, et al. Individual particle characterization of

Siberian aerosols by micro-PIXE and backscattering spectrom-

etry. X-ray Spectrom 2001;30:320–9.

Wang MX, Winchester JW, Cahill TA, Ren LX. Chemical

composition and size distribution of a dust storm in Beijing.

Kexue Tongbao 1982;27:419–22 [In Chinese, with English

abstract].

Wei F, Teng E, Wu G, Hu W, Wilson WE, Chapman RS, et al.

Ambient concentration and elemental composition of PM10 and

PM2.5 in four Chinese cities. Environ Sci Technol 1999;33:

4188–93.

World Resources Institute. World resources 1998–99. Environ-

mental change and human health. Washington, DC7 Oxford

University Press; 1998. p. 116.

Xhoffer C, Bernard P, Van Grieken R, Van der Auwera L. Chemical

characterization and source apportionment of individual aerosol

particles over the North Sea and the English Channel using

multivariate techniques. Environ Sci Technol 1991;25:1470–8.

Yang LP, Chen FH. Study on the source apportionment of

atmospheric dust pollutants in Lanzhou. Huanjing Kexue

Xuebao 2002;22:499–502 [In Chinese, with English abstract].

Yang WM, Wu BY, Sheehen C. Studies on microscopic character-

ization of atmospheric particles. Zhonghua Yufang Yixue Zazhi

1996;30:292–5 [In Chinese, with English abstract].

Yoshihiko N, Zhang L, Atsushi S. Simultaneous removal of SOx

and halogens under pulverized coal combustion added fine

limestone. Proceedings—Annual International Pittsburgh Coal

Conference; 2001. p. 763–74 [18th].

Zhang DZ, Zang JY, Shi GY, Iwasaka Y, Matsuki A, Trochkine D.

Mixture state of individual Asian dust particles at a costal site of

Qingdao, China. Atmos Environ 2003;35:3145–53.

Zhao DW, Sun BZ. Atmospheric pollution from coal combustion in

China. JAPCA J Air Waste Manage 1986;36:371–4.

R.K. Xie et al. / Science of the Total Environment 343 (2005) 261–272272

II

Individual Particles (PM10) Characterization in Taiyuan City, China

R. K. XIE1, , H. M. SEIP1, L. LIU2 and D. S. ZHANG1,3

1Department of Chemistry, University of Oslo, Blindern, P.B.1033, Oslo, Norway 2Department of Geosciences, University of Oslo, Blindern P. B.1022, Oslo, Norway

3Shanxi Environmental Protection Bureau, P. C. 030024, Taiyuan, Shanxi, China

Abstract.Taiyuan, the capital of Shanxi province, China, is one of the most polluted cities in the

world. To characterize the ambient particulate pollution, samples of particulates with aerodynamic diameter less than 10 μm (PM10) were collected during a six-day campaign. Individual particles were analyzed by Scanning Electron Microscope with Energy-Dispersive Spectrometer (SEM-EDS). 17 elements were determined and their relative abundances are given. Hierarchical cluster analysis (HCA) was applied to the lumped data set to identify particle types in the Taiyuan airshed. Twenty different particle types were identified, namely, soil-fly ash, coal burning particles, sulfur-rich, iron-rich, gypsum, syngenite, quartz, cement, silicon sulfide, siliconferro alloy, calcium-rich, ferrochromium alloy, ammonium sulfate and chloride, iron-zinc, ammonium chloride, molybdenum-rich, potassium sulfate, dolomite, lead sulfate, and copper-rich. Meanwhile their possible origins and pathways are suggested. The majority of the particles seem to originate from coal combustion, which conforms to Taiyuan’s industrial structure.

Keywords: air pollution, individual particle analysis, hierarchical cluster analysis, PM10,SEM-EDS

1. Introduction

Airborne particulate matter is the sum of all solid and liquid particles suspended in air. This complex mixture includes both organic and inorganic particles with varying size, composition, and origins. It has been well understood that particulate matter frequently contains toxic components and that they influence global climate and visibility of atmosphere.

Information about the size, morphology and composition of PM10 in the atmosphere can provide clues about its source. For example, coarse particles that contain distinctive compounds, such as silica, iron, calcium, potassium, sodium and magnesium, are likely from wind-blown soil, but finer particles containing elemental and organic carbon are likely from combustion sources. The size, morphology and composition of PM10 can also provide clues about process performance. For example, a spherical, fine particle with chemical composition mainly of silicon and aluminium, may suggest it be a particle from coal combustion process (Xie et al., 2005). Characterizing PM10 can also help to understand its potential health effects (HEI, 2002).

SEM coupled with EDS or electron microprobe technique has been used by many aerosol researchers in identifying different particles and linking them to possible sources (Anderson et al., 1988; Van Borm and Adams, 1988; Van Malderen et al., 1996). Analysis of individual

Corresponding author. Tel.: +47 22855412; fax: +47 22855441 E-mail address: ruikaixi@kjemi.uio.no

1

particles is the method of choice; detailed knowledge of the compositions of the particles from the aerosol can lead to recognition of previously unknown sources (Post and Buseck, 1984).

Taiyuan is a centre for energy production, and chemical and metallurgical industries. The production of raw coal reached about 34 million tons in 2003, which was 2.5 percent of the total raw coal production of China (Shanxi Statistics Bureau, 2004). The annual coal consumption is around 25 million tons, of which 9.6 million tons are for energy use, accounting for 95% of the total energy production (Environmental Monitoring Center of Taiyuan, 2004). Main industrial activities include production of raw coal, coke, electricity, steel, chemical fertilizers and building materials, some of which are energy and/or pollution intensive. Taiyuan is one of the most polluted cities in the world (World Bank, 1997; Mestl et al., 2005). It is a heavily industrialized area in central China that still greatly relies on uncontrolled coal combustion. This reliance, combined with unfavorable topography, has resulted in extremely high ambient SO2 and particulate levels in the city. The five-year (1996-2000) average SO2 and TSP (total suspended particles) concentrations are 241- and 467 μg/m3, respectively, which are 4.0 and 2.3 times the second level values of the National Ambient Air Quality standards (Environmental Monitoring Center of Taiyuan, 2001). Several studies have shown that the pollution levels in Taiyuan are detrimental to human health (Zhang and Zhang, 2003; Zhao, 2003; Hu et al., 2004). Many point- and area sources may contribute to the build-up of airborne particulates in Taiyuan, but their relative contribution to a given area will vary depending on their locations, meteorological conditions, etc. This study focuses on chemical characterization of the particulate matters and attempts to identify possible sources.

2. Site descriptions

Taiyuan covers an area of 1500 km2 and has a population of 2.67 million. Taiyuan is encircled by mountains to the east, west and north. Taiyuan has a marked continental monsoonal climate. North-westerly and northerly winds prevail in winter, while southerly and south-easterly winds reign in summer. Average annual precipitation is 456 mm and most of the rain falls in the warm season. Air quality is affected by the frequent occurrence of an inversion layer. In winter, the frequency is 80%, with an average inversion depth of 490 m; while in summer it is 60%, with a thinner inversion of 247 m (Environmental Monitoring Center of Taiyuan, 2001).

The sampling site was at the campus of Taiyuan University of Science and Technology (see Fig. 1) on the roof of a four-story office building. Some of the major point and area sources are shown in Fig. 1. As indicated on the map, Power plant 1#, Power plant 2#, and the Iron and & Steel Group Company are the largest industrial polluters in Taiyuan representing 50% of industrial TSP (total suspended particulates) emissions (Mestl and Fang, 2003). Xishan Coal and Power Group Company Ltd and Taiyuan Heavy Machinery Making Group Company Ltd are also important polluters. There is a cement plant in the vicinity of the sampling site and another two to the northwest. There are areas where households still use raw coal or briquettes for heating and cooking, the most important one is shown in Fig. 1, although the share of the households using raw coal or briquettes is on the decline. Due to low emission heights these sources are likely to have large effects on the pollution concentrations close to the surface. Many intermediate and small particulate emitting sources are omitted for clarity purpose.

2

Figure 1. Sampling site and some important PM sources.

3. Experimental

3.1 Sampling

A medium flow (100 l min-1) sampler (Wuhan Tianhong Intelligent Instrument Plant) was used along with an impactor cutoff of 10 μm (aerodynamic diameter). Six samples were collected. Particles (PM10) were sampled on 90 mm diameter, 1 m pore size Fluoropore(PTFE) membrane filters. The sampler was programmed to collect particles for 5 to 10 minutes every hour, with varying duration each day, depending on the pollution level. The program was set to get sufficient particles distributed on the surface while undesirable overloading was avoided. Typical meteorological data is given in Table 1.

3

Table 1. Sampling time and conditions.

Sample Time AverageTemperature (°C) Episode(s) Precipitation (mm)

1 14, 07, 2003 18.0 Cloudy, Light fog 0

2 15, 07, 2003 24.8 Sunny 0

3 16, 07, 2003 21.6 Light fog 0.1

4 17, 07, 2003 19.3 Light fog, Light rain 0.7

5 18, 07, 2003 18.7 Light rain 6.4

6 19, 07, 2003 24.0 Light fog 0.1

3.2 SEM-EDS analysis

Single-particle analysis was performed using a Hitachi S-4300SE-Shotky Field Emitter Scan Electron Microscope. This instrument is equipped with an Oxford 6853 energy dispersive spectroscopic detector, and is automated by Oxford Instruments INCA Energy and Feature software. The detailed description of the analysis procedures can be found elsewhere (Xie et al., 2005). Essentially, during analysis, the electron beam scans over a preset area. When the backscattered electron intensity of the particle contour exceeds a predefined threshold value, the object is considered detected. An X-ray spectrum from the centre of the particle can be accumulated and subsequently the spectrum is processed automatically with a built-in XPP correction procedure. Elements lighter than Na are not determined with sufficient accuracy and have been excluded. The equivalent circular diameter (ECD), defined as the diameter of a circle with the same area as the projected particle, and the shape factor, defined as the ratio of the square of the perimeter to 4 times the area, are also measured. About 400 particles were automatically analyzed for each sample. The threshold for diameter was 0.19 m. The X-ray spectrum was accumulated for 15 s with a beam current of 0.3 nA and acceleration voltage of 15 KV. The relative percentage of the elements was normalized so that the sum of the included elements is 100. The analysis’ precision and accuracy were checked by analyzing laboratory prepared fine particles of NaCl and KCl. The 95% confidence interval of the atomic fraction for Na in NaCl and K in KCl were 49.3-51.9 and 47.4-48.2.

3.3 Data handling

Cluster analysis was used to identify the particle types according to their chemical composition. Hierarchical cluster analysis (HCA) was employed. The squared Euclidean distance was used to measure the similarity of the particles. Ward’s method (error sum of squares method) was applied on the relative percentage of the elements based on the covariance matrix. The software MINITAB was used for statistical analysis. For details of the HCA, please refer to Xie et al. (2005). Prior to cluster analysis, elements with occurrence rate less then 0.2% in the whole data set and particles only associated with such elements were excluded for simplicity reasons.

4

5

4. Results and discussion

The HCA applied to 2329 particles (17 variables) yielded 20 different particle types as presented in Table 2. This analysis was done by a stepwise approach, and the number of the particle types was based on the intercluster distances and chemical explainability. Since important light elements, such as C, N and O, were excluded in the analyses, some important particle types, such as ammonium nitrate, soot, organic carbon and particles of biological origin, could not be identified. The weight percentages given in Table 2 are based only on the 17 elements determined. The relative abundance of particle types in each sample is also listed.

Soil-fly ash particles. Soil particles (Fig. 2a) and fly ash particles (Fig. 2b), dominated by Si and Al, cannot be distinguished by their chemistry, and are put in the same cluster by the HCA. Together these two particle types constitute the largest cluster. The shape factor close to unity may suggest that most particles are formed at high temperatures and therefore most of them are fly ash rather than soil. This interpretation is consistent with the prevailing weather conditions at the time since immediately before and during part of the sampling period it was raining, reducing the amounts of soil particles whirled into the air by wind or resuspended by traffic.

Coal burning particles. Coal burning particles are produced in coal-fired power stations, boilers, kilns (e.g. for making building materials) and by domestic cooking appliances. This particle type mainly contains the elements Si, Al, S, Ca and Fe (Fig. 2c). Raw coal contains various amounts of ash, mainly kaolinite, quartz, siderite, calcite, dolomite, and sulfur, mainly as organic sulfur, pyrite (FeS2. nH2O) and sulfate (mainly CaSO4). When coal is burned at high temperature, all major minerals in coal will undergo extensive coalescence forming gypsum/Al-silicates (Xie et al., 2005, and reference therein).

Sulfur-rich particles. This particle type is characterized by high amounts of sulfur, ranging from 51% to 100%. Particles with close to 100% of sulfur likely represent secondary aerosols resulting from gas-to-particle conversion of sulfur compounds, in the forms of sulfuric acid, and/or ammonium sulfate and bisulfate. Fig. 2d is an ammonium sulfate particle. Ammonia can be emitted from fertilizer manufacturing plants and agricultural activities. But in Taiyuan, coal combustion, coal gasification, and coke making may be more important sources. For every ton of coke produced, 0.1kg of ammonia may be released to the atmosphere if there is no vapor recovery system (World Bank Group, 1998). Sulfur dioxide, precursor of sulfate, is mostly from coal combustion.

Iron-rich particles. Iron can originate from to both natural and anthropogenic origins. The paths can be from soil, mining, and ferrous metallurgic industry. It seems reasonable that some of the particles originated from the Iron & Steel Group Company, which is situated in the north of the city (Fig. 1). But contribution from coal combustion is also possible, where the particles were formed as an oxidation product of pyrite (FeS2). The iron-rich particle in Fig. 2e was a product of a high temperature process which was either from the steel industry and/or from coal combustion.

Gypsum. Particles in this cluster, containing mainly S and Ca in nearly equal atomic amounts, are most likely gypsum with some other impurities, like silicates (Fig. 2f). The presence of gypsum particles is common and can be traced to many sources. In Taiyuan the largest fraction is presumably from coal combustion. During coal combustion, a part of the total sulfur is retained as a result of reactions between feed or indigenous limestone and sulfur.

6

Tab

le 2

. Par

ticle

type

s in

Taiy

uan.

El

emen

tal w

eigh

t per

cent

age

(nor

mal

ized

to 1

00%

) Id

entit

yN

oA

bua

%EC

Db

SFc

Na

Mg

Al

SiP

SC

lK

Ca

TiC

rM

nFe

Cu

ZnM

oPb

Soil-

fly a

sh

542

23.3

1.

0 1.

3 1.

0 1.

6 25

.0

50.1

0.

0 7.

6 2.

0 4.

6 2.

7 0.

8 0.

0 0.

1 3.

7 0.

0 0.

3 0.

5 0.

0

Coa

l bur

ning

37

916

.3

1.2

1.6

1.1

2.6

9.3

18.6

0.

0 26

.5

4.2

4.1

29.1

0.

6 0.

0 0.

0 3.

2 0.

1 0.

7 0.

0 0.

0

Sulfu

r-ric

h24

910

.70.

71.

31.

61.

40.

73.

20.

086

.9

1.5

3.0

0.8

0.0

0.0

0.3

0.0

0.0

0.6

0.0

0.0

Iron

-ric

h 19

88.

5 0.

9 1.

2 0.

6 1.

2 1.

4 6.

7 0.

0 6.

6 0.

8 0.

5 0.

7 0.

3 0.

9 0.

9 77

.9

0.0

1.2

0.3

0.0

Gyp

sum

17

37.

4 1.

3 1.

4 0.

1 0.

5 0.

9 2.

6 0.

0 44

.2

0.3

0.8

50.6

0.

0 0.

0 0.

0 0.

0 0.

0 0.

0 0.

0 0.

0

Syng

enite

11

95.

1 1.

1 1.

6 0.

8 1.

4 2.

3 7.

0 0.

2 42

.5

1.2

14.5

29

.0

0.0

0.0

0.0

0.8

0.0

0.4

0.0

0.0

Qua

rtz

104

4.5

1.0

1.3

0.2

0.2

3.3

89.4

0.

0 4.

1 0.

6 1.

0 0.

6 0.

2 0.

0 0.

0 0.

4 0.

0 0.

0 0.

0 0.

0

Cem

ent

934.

0 1.

4 1.

8 0.

7 2.

6 5.

3 13

.5

0.0

18.5

2.

8 2.

5 52

.2

0.0

0.0

0.0

1.3

0.0

0.5

0.0

0.0

Silic

on su

lfide

86

3.7

0.9

1.3

3.3

2.8

6.6

28.8

0.1

34.1

8.3

8.3

5.4

0.3

0.2

0.2

0.1

0.8

0.7

0.0

0.0

Silic

onfe

rro

allo

y 82

3.5

0.9

1.3

1.5

1.7

11.1

19.4

0.1

14.8

2.9

2.9

5.7

0.0

1.1

0.1

38.2

0.0

0.5

0.0

0.0

Cal

cium

-ric

h 66

2.8

1.4

1.5

0.1

0.7

3.7

6.2

0.0

5.9

0.7

0.5

81.9

0.

0 0.

0 0.

0 0.

0 0.

0 0.

0 0.

4 0.

0

Ferr

ochr

omiu

m a

lloy

parti

cles

55

2.4

0.7

1.1

1.4

2.8

1.0

7.2

0.0

6.5

0.5

1.1

0.7

0.0

38.3

4.

0 34

.1

0.0

1.7

0.7

0.0

Am

mon

ium

Sul

fate

and

C

hlor

ide

421.

8 1.

0 1.

4 3.

3 0.

9 0.

5 5.

9 0.

0 34

.9

53.0

1.

3 0.

2 0.

0 0.

0 0.

0 0.

0 0.

0 0.

0 0.

0 0.

0

Iron

-Zin

c pa

rticl

es

411.

8 0.

7 1.

2 0.

4 3.

1 2.

2 12

.6

0.2

12.1

13

.2

3.2

1.9

0.0

4.1

1.0

18.0

0.

0 27

.4

0.0

0.7

Am

mon

ium

Chl

orid

e 25

1.1

0.6

1.3

0.0

0.0

0.0

0.0

0.0

0.0

100

0.0

0.0

0.0

0.0

0.0

0.0

0.0

0.0

0.0

0.0

Mol

ybde

num

-ric

h pa

rticl

es

210.

9 0.

8 1.

4 2.

2 0.

4 0.

7 3.

8 0.

0 0.

0 5.

0 0.

0 0.

0 0.

0 0.

0 0.

0 0.

0 0.

0 0.

0 87

.8

0.0

Pota

ssiu

m su

fate

170.

7 0.

8 1.

2 0.

6 0.

7 1.

0 6.

6 0.

0 30

.2

1.4

59.0

0.

0 0.

0 0.

0 0.

0 0.

0 0.

0 0.

6 0.

0 0.

0

Dol

omite

16

0.7

1.2

1.7

0.0

34.2

1.

7 6.

4 0.

0 2.

8 0.

9 0.

2 53

.8

0.0

0.0

0.0

0.0

0.0

0.0

0.0

0.0

Lead

sulfa

te

140.

6 0.

6 1.

1 0.

0 0.

7 1.

6 3.

6 0.

0 24

.7

2.8

9.9

2.3

0.0

0.0

0.0

0.0

0.0

0.7

0.0

53.9

Cop

per-

rich

70.

30.

81.

10.

00.

00.

00.

00.

0 8.

44.

50.

00.

70.

00.

00.

00.

086

.40.

00.

00.

0

Abu

a : Par

ticle

num

ber a

bund

ance

. EC

Db : e

quiv

alen

t circ

ular

dia

met

er. S

Fc : sha

pe fa

ctor

- th

e ra

tio o

f the

squa

re o

f the

per

imet

er to

4 ti

mes

the

area

7

Figure 2. SEM photomicrographs and EDS spectra of particles (sampled for 24 hours). Some overlapping or aggregation of particles is seen. (a) Soil particles; (b) A fly ash particle; (c) Aggregated coal burning particles; (d) A sulphur-rich particle, possibly an ammonium sulphate particle; (e) An iron particle; (f) Gypsum particles; (g) Synegnite particles; (h) A natural quartz particle; (i) Quartz particle agglomerate from coal combustion; (j) Cement particle aggregates; (k) CaCO3 particles; (l) A hydrated lime particle; (m) A ferrochromium alloy particles attached by some other small particles; (n) Dolomite aggregate.

Syngenite. This particle type contains mainly S, K, and Ca with some minor amounts of other elements (fig. 2g). It was first reported in urban atmosphere in our previous study (Xie et al., 2005), in which we suggested coal combustion and/or the cement industry were possible sources.

Quartz particles. Particles in this cluster contain mainly silicon (nearly 90% on average) and are classified as quartz (SiO2). There are several different sources to explain the ambient presence of silicate. One is in natural form (Fig. 2h), which may originate from soil suspension or from industries where quartz is utilized. Another pathway is from coal combustion (Fig. 2i). Silicon, the most abundant element in coal ash, exists in coal both in the pure state, as quartz, and in many other compounds such as clays and feldspars. SiO2 isformed by reduction of silicon-containing compounds at high temperature followed by evaporation, oxidation and nucleation processes (Xie et al., 2005).

Cement particles. These particles are characterized by high content of Ca, moderate amounts of S and Si, and minor amounts of Mg, Al, Cl, K and Fe (Fig. 2j). They are most

8

probably emitted during cement manufacturing, notably the cement installations marked in Fig. 1.

Silicon sulfide particles. This particle type resembles silicon sulfide stoichiometrically. As far as we know, its presence in the ambient air has never been reported. This is another form of coal combustion product (so called CCP). The formation mechanism was reported by Nagelberg et al. (1985), who concluded that, in the presence of carbon, the fuel-borne sulfur can readily react with SiO2 to form silicon sulfide vapor.

Silicoferro alloy particles. In this particle type iron and silicon are found in nearly equal atomic amounts, with somewhat less sulfur and aluminum. These particles are most probably from metallurgical industry, where silicon-iron alloy is produced.

Calcium-rich particles. This particle type is strongly dominated by Ca with some contributions of S, Si, and Al. Some of these particles are limestone (CaCO3) with some impurities (Fig. 2k). Limestone particles are generated through its application in many areas, e.g. as building material, in cement manufacturing, and in metallurgic industry. Some are hydrated lime from limekilns (Fig. 2l), in which lime is produced for application in metallurgical industry and building construction.

Ferrochromium alloy particles. The high contents of chromium and iron characterize this particle type (Fig. 2m). The near unity shape factor indicates that they originate from high temperature processes. They may stem from the iron and steel works in the northeast. Ferrochromium alloy is used to manufacture high-grade stainless alloy steel.

Ammonium chloride, sulfate. Individual particle chemistry inspection shows that particles in this cluster actually contain both chloride and sulfate, presumably associated with ammonium, and that the cluster is not a mixture of two different particle types misclassified by the statistical approach. The origin of these particles is most likely coal combustion and coke making just as sulfur-rich particles mentioned above and ammonium chloride to be discussed (see below).

Iron-zinc particles. These particles contain mainly Zn and Fe, but also on average 12-13% of each of the elements Si, S, and Cl. They are most likely of the same origin as silicoferro alloy particles, i.e. from metallurgical industry.

Ammonium chloride. Ammonium chloride is thought to be the product of reaction between gaseous ammonia and hydrogen chloride. Hydrogen chloride is most likely from coal combustion and the coke industry.

Molybdenum-rich particles. Molybdenum is primarily used as an alloying element in steel, cast iron, and nonferrous metals.

Potassium sulfate. The formation mechanism of potassium sulfate may be analogous to that of gypsum. Generally, the potassium content in raw coal is much lower than that of calcium, and it is associated with aluminosilicates and feldspars, which are less likely to absorb sulfur. This may explain its lower abundance relative to calcium sulfate (gypsum).

Dolomite. Dolomite is a common mineral in Taiyuan (Fig. 2n). Airborne dolomite particles can be both natural and from industrial process. Dolomite is applied in many fields; it may also be emitted during cement manufacturing.

Lead sulfate. The source is not likely to be leaded petrol since the particles contain litle or no Cl and Br. This is consistent with lead in petrol being phased out in China. In Taiyuan, the lead containing particles are most likely from metallurgical smelters, heavy machinery manufacturing, and coal combustion.

9

Copper-rich particles. This particle type contains mainly copper, but also some S and Cl. The shape factor is near unity, consistent with the high temperature under which copper is melted and particles emitted. Quite possibly the paths originate from a copper smelter six kilometers to the southeast.

Statistics show that quite a large number of particles are induced by coal combustion: fly ash particles (23.3%, if soil is neglected as discussed previously), coal burning particles (16.3%), sulfur-rich particles (10.7%), gypsum particles (7.4%), syngenite particles (5.1%), and silicon sulfide particles (3.7%). Our samples were collected in summer, when coal consumption is not at its peak. An even larger contribution of particles from coal burning is expected in winter when coal consumed for space heating aggravates the burden of particulate pollution (Environmental Monitoring Center of Taiyuan, 2004). The nature of the particulate matter pollution in Taiyuan is also corroborated by our bulk sample analysis by using Inductively Coupled Plasma Atomic Emission Spectrometry (ICP-AES) and Inductively Coupled Plasma Mass Spectrometry (ICP-MS), where the occurrence of high trace elements such as arsenic and selenium could only be explained by the source of coal combustion (Xie et al., 2006).

The composition of particulates at a given site is governed by many factors such as location and emission height of nearby sources, topography, meteorological conditions etc. Mestl et al. (2005) modeled pollutant dispersion in Taiyuan and showed that low level emission sources were responsible for most of the particulate pollution close to the ground, and thus for most of the health damage.

Normally, trace heavy metals in airborne particles cannot be detected by SEM-EDS due to technique limitation. But in Taiyuan several heavy metals, such as chromium, zinc, molybdenum, lead, and copper were present as major elements in some particle types. The presence of zinc, lead and copper (chromium and molybdenum were not analyzed) was also confirmed by their elevated concentrations and high enrichment factors in our bulk sample analyses.

6. Conclusions

SEM-EDS analysis is a method of choice for characterizing airborne particulate, especially for individual particles. The variety of particle types reveals the particulate pollution complexity and reflects the multiple industrial activities in Taiyuan. Some important sources and their origins as well as their possible transformation processes of the particles in Taiyuan airshed have been elucidated, which we hope it can contribute to a better understanding of the particulate pollution nature and a clearer strategy of pollution abatement. Coal combustion is the main source of particles in Taiyuan air, as evidenced by the abundance of the characteristic coal burning related particles, which is consistent with the industrial structure and the large use of coal, as well as with the bulk sample analyses by ICP-AES/MS. Correspondingly, in Taiyuan the reduction of coal combustion emissions will have large beneficial effects on residents’ health. Also particles high in heavy metals from the metallurgical industry may have serious effects and deserve attention.

Acknowledgement

Advice and sampling assistance from Professor Jinghua Fang and Master student Hongge Li at Taiyuan University of Science and Technology, are greatly appreciated.

10

References

Environmental Monitoring Center of Taiyuan, 2001: Environmental Quality Report for Taiyuan, 1996-2000, Taiyuan, Shanxi, China.

Environmental Monitoring Center of Taiyuan, 2004: Environmental Quality Report for Taiyuan, 2003, Taiyuan, Shanxi, China.

HEI (Health Effects Institute), 2002: Understanding the health effects of components of particulate matter mix: progress and next steps, HEI perspectives.

Hu, E. B., Wu, Q. H., Wang, H, Ji, S. W. and Xin, C. T., 2004: Study on the environment risk through inhalation pathway in Taiyuan City. Huanjing Kexue Xuebao 24(1), 116-120 (In Chinese, with English abstract).

Mestl, H. E. S. and Fang, J. H., 2003: Air quality estimates in Taiyuan, Shanxi Province, China--Application of a multiple-source dispersion model, CICERO Report, Oslo, Norway.

Mestl, H. E. S., Aunan, K., Fang, J. H., Seip, H. M., Skjelvik, J. M. and Vennemo, H., 2005: Cleaner production as climate investment - integrated assessment in Taiyuan City, China. J. Clearner Production, 13, 57-70.

Nagelberg, A. S., 1985: Vaporization of silica under coal combustion conditions. Proc.- Int. Conf. Coal Sci., 385-8. Publisher: Pergamon, Sydney, Australia.

Post, J. E. and Buseck, P.R., 1984: Characterization of individual particles in the Phoenix urban aerosol, using electron beam instruments. Environ. Sci. Technol., 18, 35-42.

Shanxi Environmental Protection Bureau, 2004: Shanxi Environmental Statistics, 2003, Taiyuan, Shanxi, China.

Shanxi Statistics Bureau, 2004: Shanxi Statistical Yearly book, 2003, China Statistics Press.

Van Born, W. A. and Adams, F.C., 1988: Cluster analysis of electron microprobe analysis data of individual particles for source apportionment of air particulate matter. Atmos.Environ., 22, 229-237.

Van Malderen, H., Van Grieken, R., Khodzher, T., Obolkin, V., and Potemkin, V., 1996: Composition of individual aerosol particles above Lake Baikal, Siberia. Atmos. Environ., 30, 1453-65.

World Bank, 1997: China 2020: Development Challenges in the New Century, WashingtonD. C.

World Bank Group, 1998: Pollution, Prevention and Abatement Handbook 1998: Toward Cleaner Production, 286-290, Washington D. C.

Xie, R. K., Seip, H. M., Leinum, J. R., Winje, T. and Xiao, J. S., 2005: Chemical characterization of individual particles (PM10) from ambient air in Guiyang City, China. Sci. Total Environ., 343, 261-272.

Xie, R. K., Seip, H. M., Wibetoe, G., Nori, S. and Mcleod, C. W., 2006: Heavy coal combustion as the dominant source of particulate pollution in Taiyuan, China, corroborated by high concentrations of arsenic and selenium in PM10. Sci. Total Environ., 370, 409-415.

11

Zhang, Y. P. and Zhang, X. P., 2003: Association between air pollution and hospital daily mortality in Taiyuan. Huanjing Yu Zhiye Yixue, 20, 198-202 (In Chinese, with English abstract).

Zhao, B. X., 2003: Study on the effect of PM10 and SO2 air pollution on children's pulmonary function in Taiyuan City, China. Huanjing Yu Zhiye Yixue, 20, 203-206 (In Chinese, with English abstract).

12

III

Heavy coal combustion as the dominant source of particulatepollution in Taiyuan, China, corroborated by high concentrations of

arsenic and selenium in PM10

RuiKai Xie a,⁎, Hans Martin Seip a, Grethe Wibetoe a,Showan Nori a, Cameron William McLeod b

a Department of Chemistry, University of Oslo, Blindern, P. B. 1033, Oslo, Norwayb Department of Chemistry, The University of Sheffield, Sheffield S3 7HF, UK

Received 2 March 2006; received in revised form 16 May 2006; accepted 2 July 2006Available online 8 August 2006

Abstract

Coal burning generates toxic elements, some of which are characteristic of coal combustion such as arsenic and selenium,besides conventional coal combustion products. Airborne particulate samples with aerodynamic diameter less than 10 μm (PM10)were collected in Taiyuan, China, and multi-element analyses were performed by inductively coupled plasma atomic emissionspectrometry (ICP-AES) and inductively coupled plasma mass spectrometry (ICP-MS). Concentrations of arsenic and seleniumfrom ambient air in Taiyuan (average 43 and 58 ng m−3, respectively) were relatively high compared to what is reportedelsewhere. Arsenic and selenium were found to be highly correlated (r=0.997), indicating an overwhelmingly dominant source.Correlation between these two chalcophile elements and the lithophile element Al is high (r is 0.75 and 0.72 for As and Se,respectively). This prompted the hypothesis that the particles were from coal combustion. The enrichment of the trace elementscould be explained by the volatilization–condensation mechanism during coal combustion process. Even higher correlations ofarsenic and selenium with PM10 (r=0.90 and 0.88) give further support that airborne particulate pollution in Taiyuan is mainly adirect result of heavy coal consumption. This conclusion agrees with the results from our previous study of individual airborneparticles in Taiyuan.© 2006 Elsevier B.V. All rights reserved.

Keywords: Arsenic; Selenium; PM10; Pollution; Coal consumption

1. Introduction

Worldwide coal, an abundant fossil fuel, is exten-sively used to produce energy. Coal burning is thegreatest anthropogenic source of toxic air pollution anda large contributor to global warming (Kunstler, 2005).

Elements bound in coal are mobilized during coalcombustion and may be released associated withparticles or as vapours. Some of these elements areessential for healthy plant and animal life, while someare toxic if present in sufficient quantities.

Heavy reliance on coal combustion as a source of heatand power has created serious environmental problems inTaiyuan, China. Particulate matter (PM) and sulphurdioxide (SO2) are the major pollutants of concern.Pollution levels of these two components and their origins

Science of the Total Environment 370 (2006) 409–415www.elsevier.com/locate/scitotenv

⁎ Corresponding author. Tel.: +47 22855412; fax: +47 22855441.E-mail addresses: ruikaixi@kjemi.uio.no, xieruik@yahoo.com

(R. Xie).

0048-9697/$ - see front matter © 2006 Elsevier B.V. All rights reserved.doi:10.1016/j.scitotenv.2006.07.004

have been reviewed by Mestl et al. (2005). As coalcontains traces of nearly all elements, coal burning resultsin an increased release of trace metals into the environ-ment, some of which, e.g. As, Cd, Cr, Hg and Se, are ofgreat concern and labelled as potentially hazardous traceelements (Swaine and Goodarzi, 1995).

Taiyuan, the capital of Shanxi province, whichproduces one quarter of China's raw coal, has apopulation of 2.67 million and covers an area of1500 km2. The city is surrounded by mountains onthree sides, which favours the build-up of pollutants.The annual coal consumption is around 25 million tons,of which 9.6 million tons are for energy use, accountingfor 95% of the total energy production (EnvironmentalMonitoring Center of Taiyuan, 2003). Although com-pared with world averages most of Shanxi coals containsimilar concentrations of potentially hazardous traceelements (Zhang et al., 2003), the magnitude of coalconsumed within such a small area is likely to causehighly elevated levels of some pollutants.

This short study has been carried out to gain animproved quantitative estimate of the role of coalconsumption in airborne particulate pollution. Particu-lates (PM10) were sampled in Taiyuan and multi-

element analyses were performed by using ICP-AESand ICP-MS. The results revealed that certain traceelements, particularly arsenic and selenium, are unu-sually high compared to studies elsewhere. These twoelements are among those of most environmentalconcerns. Arsenic is both a carcinogen and mutagen,and leads to dangerous dermatological, respiratory anddigestive system diseases. Selenium in excessiveamounts has been cited as a cause of periodontal diseaseand dental caries in human and as a possible carcinogen(NAS, 1974). An endemic disease in China has beenattributed to selenium intoxication (Yang et al., 1983).Selenium was also reported as the cause of massive fishkills in Martin Lake, Texas, and in North Carolina(Shepard, 1987). Both of these elements can be releasedfrom coal during coal mining, beneficiation andcombustion (He et al., 2002 and reference therein).

2. Experimental

2.1. Sampling and sample preparation

The sampling site was on the roof of a four-storybuilding in Taiyuan University of Science and Tech-

Fig. 1. Map of Taiyuan. Jurisdictionally, Taiyuan comprises three suburban counties, one satellite city and six districts. The six districts (in grey)constitute the actual provincial capital city “Taiyuan”.

410 R. Xie et al. / Science of the Total Environment 370 (2006) 409–415

nology (see Fig. 1). A medium flow (100 l min−1)sampler (Wuhan Tianhong Intelligent InstrumentPlant) was used along with an impactor cutoff of10 μm (aerodynamic diameter). Particles (PM10) weresampled on 90 mm diameter, 3 μm pore sizeFluoropore® (PTFE) membrane filters with a collec-tion efficiency of 98% for particles larger than0.035 μm. Samples were collected from 2 March2004 to 16 March 2004, with sampling durationvarying from 4 to 24 h. Nine samples were collected.Filters loaded with samples were sealed off in cleanplastic bags and stored frozen until analysis.

The Teflon filters were weighed before and aftersampling on an electronic balance (OHAUS® Analy-tical Plus, readability 0.01 mg). Before weighing, thefilters or samples were dried at room temperature in adesiccator with silica gel for 24 h. Mass concentration ofPM10 (μg m−3) was calculated as the ratio of the netmass of particulates to the standard air volume.

Prior to chemical analysis, samples were digestedusing a microwave oven (ETHOS 1600 AdvancedMicrowave Labstation from Milestone, Sorisole,Italy) equipped with temperature sensor and con-trolled by a personal computer. Filters loaded withairborne particles (6–41 mg) or equivalent standardreference material (SRM 1648) were placed in closed100 ml PTFE vessels with 5 ml of a mixture ofHNO3 (65%) and HF (40%) (5+1 v/v) and digestedin the microwave oven according to the followingprogramme:

Step Temperature (°C) Time (min) Power (W)1. Ramp 172 15 10002. Dwell 172 15 10003. Ventilate 10 0

After cooling, the vessels were opened and thedigested solutions were diluted with 50 ml water. Thenthe vessels were closed again and kept at roomtemperature overnight to dissolve completely anyfluorides that might be present.

2.2. Instrumentation and analytical procedures

Measurements of major and minor elements, suchas Al, Ca, Fe, K, Mg, Na, Pb, Ti and Zn, wereperformed on ICP-AES (Varian Vista AX CCD,Simultaneous ICP-AES); measurements of As, Co,Mn, Ni, Sb, Se, Sn and V were conducted on a ICP-MS (Perkin-Elmer SCIEX Elan 5000). Both ICP-AESand ICP-MS instruments were calibrated by anexternal calibration method. Calibrants were prepared

from multi-element standard solution (50 μg l−1,Teknolab A/S, Norway).

Standard Reference Material, SRM 1648 ‘UrbanParticulate Matter’, from the National Institute ofStandards and Technology (Gaitherburg, MD, USA)was used to validate the methods. The SRM was treatedin the same manner as the samples. Reagent blanks andfilter blanks were also routinely analyzed in betweensamples to check for contamination.

3. Results and discussion

3.1. PM10 mass concentrations

Bulk mass concentration of PM10 varied from dayto day depending on the source emissions andmeteorological conditions. It ranged from 188 to979 μg m−3. The average was 553 μg m−3, far abovethe Grade Π (150 μg m−3) of the National AmbientAir Quality Standard of China (SEPA, 1996). Con-sidering this was close to a peak time for coal con-sumption (extra burden from space heating) of theyear, it seems plausible to record such high concen-tration of PM10.

3.2. Elemental concentrations

The reliability of elemental determinations wasassessed through analysis of the Standard ReferenceMaterial analysis. Of the 16 elements determined, Caand Sn have no certified or reference values. The valuesfor Al, Fe, K, Mg, Na, Pb, Ti, Zn, As, Mn, Ni, Sb, Se and

Table 1Method validation with SRM 1648 (n=7, σ is the standard deviation)

Element Unit Certified orreference value

Determined valuein this study

Average Uncertainty a Average σ

Al % 3.42 0.11 3.41 0.05Fe % 3.91 0.10 4.17 0.06K % 1.05 0.01 0.95 0.05Mg % 0.8 0.89 0.02Na % 0.425 0.002 0.427 0.007Pb % 0.655 0.008 0.717 0.026Ti % 0.40 0.40 0.02Zn % 0.476 0.014 0.501 0.017As mg kg−1 115 10 118 10Mn mg kg−1 786 17 809 50Ni mg kg−1 82 3 76 8.2Sb mg kg−1 45 39 5.5Se mg kg−1 27 1 30 3.8V mg kg−1 127 7 126 7a For details of the uncertainty, see the NIST Certificate of Analysis

Standard Reference Material® 1648.

411R. Xie et al. / Science of the Total Environment 370 (2006) 409–415

Vare shown in Table 1. Recoveries range between 87%and 111%.

The results for the elemental concentration in thePM10 samples are listed in Table 2. Of the trace elementsAs and Se are of particular interest. As concentrationsranged from 12 to 83 ng m−3, with the arithmeticaverage of 43 ng m−3, far exceeding the WHOguidelines for Europe (WHO, 2001), which is 0.7 ngm−3. Arsenic is one of the most toxic trace elements.Devastating arsenic pollution by copper smelting plantswas reported in the USA during 1940–1964, whereconcentrations of 0.9–259 μg m−3 were observed(Borgono and Greiber, 1997). The most recentlyreported high concentration of As in urban areas, i.e.73 ng m−3 in Chile, was also due to the impact of coppersmelter emissions (Kavouras et al., 2001). Highconcentrations of arsenic are expected to occur in theTaiyuan airshed due to the large quantity of coalconsumed in this area.

Selenium concentrations were from 9.5 to 126 ngm−3, with an arithmetic average of 58 ng m−3. Ingeneral, the concentrations of urban particulate seleniumare in the range of 1 to 10 ng m−3 (Ihnat et al., 1989).Extremely high concentrations of over 100 ng m−3 werereported in Ankara, Turkey (Ölmez and Aras, 1977).These samples were collected during the fall and winterin an area where peat was the major residence fuel andno emission control was in place.

The concentrations of As and Se found in our studyare compared to results from PM10 studies in otherurban areas in Table 3. The results of this study indicatethat levels are higher in Taiyuan than in any other studywe are aware of, except for mean As concentration

measured in Shanghai in winter 2001 and the excep-tional cases mentioned above.

3.3. Correlations between elements

To explore possible sources of the trace elements,correlations were calculated. Al is a typical lithophileelement. Very strong correlations of Ca, Fe, K, Mg,Na, Ti and V with Al are observed (r>0.92 in allcases). Thus, these elements are probably from thesame source(s). As, Mn, Se and Zn are alsosignificantly correlated with Al, with r=0.75 (Fig.2A), 0.81, 0.72 and 0.82, respectively, which alsosuggests a similar origin. That both lithophile elements(Ca, Fe, K, Mg, Mn, Na, Ti and V) and chalcophileelements (As, Se and Zn) are highly correlated withAl may be explained by the following hypothesis:both lithophile and chalcophile elements are the resultsof coal combustion. The lithophile elements aremainly from particles emitted from coal combustion,rather than soil or road dust, as was also illustrated inour previous study of individual particles (Xie et al.,submitted for publication). For chalcophile elements,the volatilisation–condensation mechanism may pro-vide an answer, i.e. these elements are first volatilizedduring coal combustion and condensed later on finefly ash particles (Clarke and Sloss, 1992). Se and Asare very highly correlated (see Fig. 2B), withr=0.997. This perfect correlation in the bulk samplesstrongly indicates that there is an overwhelminglydominant source that dictates the relationship betweenthese two elements. As both As and Se arefingerprints of coal combustion (Nriagu, 1989;Harrison et al., 1996; Ölmez et al., 2004; Park and

Table 2Elemental concentrations in air for PM10 collected in Taiyuan City inMarch 2004 (n=9)

Element Unit Minimum Median Maximum Mean σ

Al μg m−3 9.94 26.28 55.13 26.69 13.12Ca μg m−3 15.11 38.60 76.93 41.92 19.81Fe μg m−3 7.43 17.55 34.43 19.13 8.47K μg m−3 2.52 8.45 17.11 8.45 4.64Mg μg m−3 3.21 6.62 17.74 8.60 4.70Na μg m−3 1.46 4.60 11.17 5.08 3.12Mn μg m−3 0.27 0.69 1.21 0.69 0.37Pb μg m−3 0.17 0.63 1.34 0.63 0.38Ti μg m−3 0.59 1.43 2.85 1.55 0.69Zn μg m−3 0.35 1.02 2.44 1.30 0.86As ng m−3 11.98 31.57 82.55 43.36 27.61Ni ng m−3 6.96 41.59 123.90 43.62 41.64Sb ng m−3 2.85 9.57 19.57 11.08 9.57Se ng m−3 9.50 44.41 126.36 58.21 44.41Sn ng m−3 2.41 6.96 33.75 11.77 9.49V ng m−3 13.59 33.78 62.72 36.90 16.13

Table 3Comparison of As and Se concentration (all in PM10) from differentsites

Location Mean As(ng m−3)

Mean Se(ng m−3)

Time andremarks

Reference

Downtown,Los Angeles,USA

6.9 8.1 1987 Chow et al.(1994)

Birmingham, UK 6.7 4.9 Winter, 1992 Harrisonet al. (1996)

Birmingham, UK 5.65 3.05 Winter andAugust 1992

Harrisonet al. (1997)

Coimbra, Portugal 2.35 0.79 1992–1993Macao, China 17 21 2001,

backgroundWu et al.(2003)

Shanghai, China 42.1 19.5 Winter, 2001 Zheng et al.(2004)

Taiyuan, China 43 58 March 2004 This study

412 R. Xie et al. / Science of the Total Environment 370 (2006) 409–415

Kim, 2004), this source can only be coal combustionin Taiyuan.

Correlations between the concentrations of elementsand PM10 mass concentrations were also calculated.For Al the correlation coefficient is 0.95 (Fig. 3A),while the value for As is 0.90 (Fig. 3B) and for Se0.88. The high correlations of both representativelithophile and chalcophile elements with PM10 massconcentration give further support to the abovehypothesis. It is worth noting that there may bedifferent pathways for coal combustion particulatesreaching the sampling site: either directly fromemission stacks, or from resuspension from thepavements or roads by wind or traffic.

3.4. Enrichment factors (EF)

Enrichment factors have long been applied inenvironmental studies to differentiate sources, espe-cially in particulate matter studies. Enrichment factorswere calculated to get information about the contribu-tions to the ambient levels from sources other thancrustal matter. Ti was chosen as the reference elementdue to its accurate quantitative determination and

relatively few pollution sources. The global averagecomposition of the earth's crust (Taylor, 1964) is used asthe reference value owing to the lack of localrepresentative soil samples. EF is calculated from:

EF ¼ ðX=TiÞPM10=ðX=TiÞCrust

where X is the concentration of the element of interest,Ti is the concentration of element Ti, (X/Ti)PM10

represents the ratio derived from the concentrations ofelement X and Ti in PM10, and (X/Ti)Crust represents theratio derived from the concentrations of element X andTi in background crustal matter. Due to the variation ofcrustal element concentration, elements with EF lessthan 5 are not regarded as significantly enriched.

Enrichment factors based on average values for allthe elements determined are presented in Fig. 4.Important features are:

1. The EFs of Al and Fe are close to one, indicating thatAl and Fe can also be used as reference element inthis study;

2. The EFs of Al, Ca, Fe, K, Mg, Na, Mn, V and Ni areall less than 5, suggesting these elements are notsignificantly enriched relative to Ti;

Fig. 2. Linear correlations between elements: Al–As (A) and As–Se (B).

Fig. 3. Linear correlations between PM10 and Al (A) and As (B).

413R. Xie et al. / Science of the Total Environment 370 (2006) 409–415

3. As, Sb, Se, Sn, Pb and Zn are enriched by more thanone order of magnitude.

Often EF is used as a criterion to pinpoint whichelements are from natural processes and which are fromanthropogenic processes. However, if natural andanthropogenic sources have nearly the same chemicalcompositions, EF cannot be used for differentiation. Wereported in our previous paper that soil particles and flyash particles are chemically very similar and cannot bedistinguished by chemical composition alone (Xie et al.,2005). In a similar study in Taiyuan (Xie et al.,submitted for publication), we found a cluster ofparticles containing mainly aluminium, silicon andvarying amounts of magnesium, potassium, calciumand iron, which were identified mostly as fly ashparticles based on shape factors. Therefore, in coal-polluted areas, enrichment factors for elementsapproaching unity do not necessarily indicate thatthese elements are just of natural origin, e.g. soil-derived dust, they may also be of anthropogenic origin,e.g. coal combustion.

Different elements behave differently during the coalcombustion process. The non-enriched elements havelow volatility and thus are evenly distributed in bottomash and fly ash, and exhibit neither depletion norenrichment. The enrichment of trace elements can beexplained by the volatilisation–condensation mechan-ism described earlier. This mechanism and the fact that alarge fraction of PM10 is from coal combustion explainswhy the trace elements As, Se, Sb, Pb and Zn areenriched in the PM10 samples investigated.

4. Conclusions

Large coal consumption in Taiyuan resulted inelevated concentrations of trace elements, especiallyvolatile elements, such as arsenic and selenium.Enrichment factor calculations revealed that they were

enriched relative to crustal matter. The reason is thatthey are easily released into the gas phase at hightemperatures and then condense onto fly ash particles.Arsenic and selenium have long been recognized as coalcombustion markers. Correlation analysis revealed thatthese two elements are highly correlated, whichindicated a dominant source. Fairly high correlationsof As and Se with Al, a lithophile element, suggestedthey were from the same source: coal combustion.Furthermore, high correlations of Al and Se with PM10

signified coal combustion as a dominant source ofparticulate pollution in Taiyuan.

5. Future work

The results presented here are just for a short periodand a fraction of the Taiyuan airshed. Further monitor-ing for a longer period and larger area will definitelyprovide more useful information for effective particulatematter pollution controls. A Sino-Norwegian project isunder way to work out the best control scenarios againstair pollution. All these efforts will contribute to thebetter understanding of the air pollution and ultimatelyto a cleaner Taiyuan atmosphere.

Acknowledgement

Ruikai Xie is grateful to the Worldwide UniversityNetwork for the award of an exchange fellowship.

References

Borgono JM, Greiber R. Trace substance in environmental health-V.In: Hemphil DD, editor. Proceedings of the University ofMissouri's Annual Conference on Trace Substance in EnvironmentAnalytical Health, Columbia; 1997. p. 13–4. February.

Chow JC, Watson JG, Fujita EM, Lu ZQ, Lawson DR, Ashbaugh LL.Temporal and spatial variations of PM2.5 and PM10 aerosol in theSouthern California Air Quality Study. Atmos Environ 1994;28(12):2061–80.

Clarke LB, Sloss LL. Trace elements-emissions from coal combustionand gasification, IEACR/49. London: IEA Coal Research; 1992.

Environmental Monitoring Center of Taiyuan. Environmental QualityReport for Taiyuan, 2003; 2004. Taiyuan, Shanxi, China.

Harrison RM, Smith DJT, Luhana L. Source apportionment ofatmospheric polycyclic aromatic hydrocarbons collected from anurban location in Birmingham, U.K. Environ Sci Technol 1996;30(3):825–32.

Harrison RM, Smith DJT, Pio CA, Castro LM. Comparative receptormodeling study of airborne particulate pollutants in Birmingham(United Kingdom), Coimbra (Portugal) and Lahore (Pakistan).Atmos Environ 1997;31(20):3309–21.

He B, Liang LN, Jiang GB. Distribution of arsenic and selenium inselected Chinese coal mines. Sci Total Environ 2002;296:19–26.

Ihnat M, Edidin M, Szollosi J, Tron L. Occurrence and distribution ofselenium. CRC Press; 1989. p. 295–326.

Fig. 4. Enrichment factors of elements (Ti as reference).

414 R. Xie et al. / Science of the Total Environment 370 (2006) 409–415

Kavouras IG, Koutrakis P, Cereceda-Balic F, Oyola P. Sourceapportionment of PM10 and PM2.5 in five cities using factoranalysis. J Air Waste Manage Assoc 2001;51:451–64.

Kunstler JH. Long emergency: surviving the converging catastrophesof the twenty-first century. Atlantic Monthly Press; 2005. p. 118.

Mestl HES, Aunan K, Fang JH, Seip HM, Skjelvik JM, Vennemo H.Cleaner production as climate investment—integrated assessmentin Taiyuan City, China. J Clean Prod 2005;13:57–70.

NAS (National Academy of Sciences). The relation of selected traceelements to health and disease. Geochemistry and the Environ-ment, vol. 2. Washington, DC: National Academy of Sciences;1974. p. 113.

Nriagu JO. Natural versus anthropogenic emissions of trace metals tothe atmosphere. In: Pacyna JM, Ottar B, editors. Control and Fateof Atmospheric Trace Metals. Dordrecht, The Nertherlands:Kluwer Academic Publishers; 1989. p. 3–13.

Ölmez EI, Kut D, Bilge AN, Ölmez I. Regional elemental signaturesrelated to combustion of lignites. J Radioanal Chem 2004;259(2):227–31.

Ölmez I, Aras NK. Trace elements in the atmosphere determined bynuclear activation analysis and their interpretation. J RadioanalChem 1977;37(2):671–7.

Park SS, Kim YJ. PM2.5 particles and size-segregated ionic speciesmeasured during fall season in three urban sites in Korea. AtmosEnviron 2004;38:1459–71.

SEPA (State Environmental Protection Administration of China).Ambient Air Quality Standard of People's Republic of China(GB3095-1996); 1996.

Shepard M. Toxic research and the real world. EPRI J 1987(September):17–21.

Swaine DJ, Goodarzi F. Environmental aspects of trace elements incoal. Dordrecht: Kluwer Academic Publishers; 1995. p. 1–4.

Taylor SR. Abundance of chemical elements in the continental crust: anew table. Geochim Cosmochim Acta 1964;28(8):1273–85.

WHO. Air Quality Guidelines for Europe. 2nd ed. Copenhagen,Denmark: WHO Regional Publications, Regional Office forEurope; 2001.

Wu Y, Hao JM, Fu LX, Hu JN, Wang ZS, Tang UW. Chemicalcharacteristics of airborne particulate matter near major roads andat background locations in Macao, China. Sci Total Environ2003;317(1–3):159–72.

Xie RK, Seip HM, Leinum JR, Winje T, Xiao JS. Chemicalcharacterization of individual particles (PM10) from ambient airin Guiyang City, China. Sci Total Environ 2005;343:261–72.

Xie RK, Seip HM, Liu L, Zhang DS. Individual particles (PM10)characterization in Taiyuan City, China. [submitted, 2006].

Yang G, Wang S, Zhou R, Sun S. Endemic selenium intoxication ofhumans in China. Am J Clin Nutr 1983;37:872–81.

Zhang JY, Zheng CG, Ren DY, Chou CL, Liu J, Zeng RS, et al.Distribution of potentially hazardous trace elements in coals fromShanxi province. China Fuel 2003;83:129–35.

Zheng J, Tan MG, Shibata Y, Tanaka A, Li Y, Zhang GL, et al.Characteristics of lead isotope ratios and elemental concentrationsin PM10 fraction of airborne particulate matter in Shanghai after thephase-out of leaded gasoline. Atmos Environ 2004;38:1191–200.

415R. Xie et al. / Science of the Total Environment 370 (2006) 409–415

IV

1

Characteristics of water-soluble inorganic chemical components in size-resolved airborne particulate matters - Sheffield, UK

Ruikai Xiea, *, Cameron W. McLeodb, , Marie J. Schofieldc, David Andersonc,Hans Martin Seipa, Grethe Wibetoea, Kevi A. Jacksonb, Jan Erik Hanssend

aDepartment of Chemistry, University of Oslo, Blindern, P.B. 1033, Oslo, Norway bDepartment of Chemistry, The University of Sheffield, Sheffield, S3, 7HF, UK

cCorus RD&T, Swinden Technology Centre, Moor Gate, Rotherham, S60 3AR, UK dNorwegian Institute for Air Research, Kjeller, P.B 100, 2027, Norway

Abstract

To characterise the water-soluble inorganic components of PM10 in the urban area of Sheffield, size-resolved aerosol samples were collected using Electric Low Pressure Impactor during a 13-day sampling campaign in October and November 2006. Cl-, NO3

-, SO42-, and NH4

+ were analysed by ion chromatography, and Na+, K+, Mg2+, and Ca2+ were analysed by inductively coupled plasma-mass spectrometry. Back trajectories analyses showed that the air masses could be classified into two main groups. In the “maritime regime”, the air masses moved mostly over the sea and just a short distance over land; in “the terrestrial regime”, the air masses had moved long distances over land before reaching the sampling site. The particles chemical composition was strongly influenced by the origins of the air masses. Air mass belonging to the maritime regime brought more sea-salt ingredients such as Na+, Cl-,and Mg2+. Air mass belonging to the terrestrial regime carried more K+ and secondary components like SO4

2-, NO3-, and NH4

+. Sulphate showed bimodal distribution with a submicron mode and a supermicron mode. Nitrate exhibited a bimodal distribution in the terrestrial regime but only a supermicron mode in maritime regime. The submicron mode of both sulphate and nitrate were ammonium salts. The supermicron mode sulphate was from sea spray and reaction with ammonia on surface of coarse particles; whereas the supermicron mode nitrate was the result of reaction with sea-salt and reaction with ammonia on the surface of coarse particles. Ammonium displayed a unimodal size distribution in the submicron size range, mainly bound to sulphate and nitrate. Sodium and magnesium showed similar distribution patterns, suggesting a sea-salt source. In the maritime regime, chloride exhibited a unimodal distribution, peaking in the supermicron size range. In the terrestrial regime chloride appeared to be bimodal with one peak in the accumulation mode, reflecting the presence of chloride sources from industries, and another one in the coarse mode, mainly from sea spray. Air masses from the terrestrial regime suffered minor “chloride loss” (26%) for the supermicron particles, while chloride loss for fine particles was swamped by emitted chloride from industrial sources. Essentially all the chloride loss seemed to be caused by nitrate.

Keywords: Particulate matter; ionic components; air pollution; Mass size distribution

1. Introduction

Airborne particulate matter (PM) consists of microscopic pieces of solid or liquid material suspended in the air. Particles have a number of important environmental effects. The size distribution as well as the composition of PM is critical to all climate influences. PM causes both direct and indirect radiative forcing. Sub-micrometre particulates scatter more light per unit mass and have a longer atmospheric lifetime than larger particulates. The size and chemical composition of the initial nuclei (e.g., anthropogenic sulphates, nitrates, dust, organic carbon and black carbon), particularly the water-soluble fraction and presence of compounds that affect surface tension, are important in the activation and early growth of the cloud droplets (McFiggans et al., 2006 and references therein). Particular matter causes

* Corresponding author. Tel.: +47 22855693; fax: +47 22855441 E-mail address: ruikaixi@kjemi.uio.no, xieruik@yahoo.com

2

severe health damage, especially in developing countries (WHO, 2002). Particle size and/or chemical composition are important factors with respect to adverse health effects (Schwartz et al., 1996).

The major inorganic water-soluble species comprise Cl-, NO3-, SO4

2-, NH4+, Na+, Mg2+, K+ and Ca2+.

These species can be used as indicators of sources such as industrial activities, resuspended street dust, or some natural sources such as wind-blown soils, and sea sprays. The interrelationship and size distributions of these species can provide information on formation mechanisms of particles.

Sheffield (53° 23 N, 1° 28 W) is a cosmopolitan city situated in central UK, with ca. 100 km to the eastern and western coasts, ca. 600 km to the northern and 300 km to the southern coasts (see Fig.1). Its geographic location may suggest a diverse atmospheric chemistry of airborne particulates as the results of the interactions between particles and gases of both oceanic and terrestrial sources. However, published data on the actual airborne particulate concentration and nature are very scarce (Moreno et al., 2004). In this work, urban airborne particulate were collected by an Electric Low Pressure Impactor (ELPI), and subsequently mass size distributions of water-soluble inorganic chemical components were measured. The main purposes were to determine mass concentrations of inorganic components and their mass size distribution, and to investigate the possible sources and plausible formation mechanisms. A detailed discussion of water-soluble inorganic components in size-resolved PM10 collected during 2006 is presented.

2. Experimental

2.1. Sampling

The sampling was carried out in the period from 30 October to 13 November 2006 in Sheffield. The sampling site was on the roof of the building of the Department of Chemistry, ca. 15 m above ground, within the University of Sheffield, about 1.6 kilometres to the west of the city centre. Particulate matter was collected using low-pressure 13-stage ELPI impactors (DEKATI Inc.) with a sampling flow rate of 29.26 l min-1. The collecting substrates were polycarbonate membrane filters (IsoporeTM) with diameter 25 mm and pore size 0.4 μm. The particle equivalent aerodynamic 50% cut-off diameters of each impaction stage were 0.029, 0.056, 0.094, 0.157, 0.264, 0.386, 0.619, 0.956, 1.61, 2.41, 4.03, 6.6, and 10 μm. Sampling started at 11:00 a.m. and extended to 10:30 a.m. the next day. Only particles between 0.029-10 μm were analysed, particles with diameters larger than 10 μm (stage no. 13) were not used because of the undefined upper size limit. Since particles less than 29 nm contribute very little to the total mass, the sum of the samples of the 12 stages approximates closely that of PM10. In Total 13 sets of samples were collected, however 4 sets of PM10 mass data were lost due to a computer crash, and hence they were not included in the mass calculation.

2.2. Gravimetric and chemical analyses

The mass size distribution of the particles collected on the ELPI stages was determined by weighing the substrates before and after sampling using a microbalance (Mettler MT5) with an accuracy of 0.001 mg. The ELPI filters were weighed in a conditioned atmosphere with controlled relative humidity (45-55%) and temperature (20±5ºC). Prior to weighing, the filters were equilibrated for 24 h inside the room. All the samples were stored frozen in Petri dishes until analysis.

The samples were extracted in an ultrasonication bath with 15 ml Millipore water (18.2 M .cm). The extracts were analysed by ion chromatography (IC) for major water-soluble inorganic ions using a DIONEX 2000 instrument. For the determination of anions (Cl-, NO3

-, and SO42-), IonPac® AG18 guard

and IonPac® AS18 analytical columns with ASRS-Ultra auto-suppressor were used. For NH4+, IonPac®

CG16 guard and IonPac® CS16 analytical columns with CSRS-Ultra auto-suppressor were used. For anions, the elution was performed with 35 μM KOH, with current 87 mA and flow rate of 1.0 ml min-1;and for NH4

+, the elution was carried out with 35 μM MSA, using 41 mA current and flow rate of 0.4 ml min-1. The temperatures for columns and detectors were all controlled at 35ºC. Sample injection was

automated with an auto-sampler AS40. The IC instrument was regularly calibrated and quality control was routinely conducted. Other cations (Na+, K+, Mg2+, and Ca2+) were analysed by inductively coupled plasma-mass spectrometry (ICP-MS) (Agilent 7500). Sampling and analyses of blanks were performed the same way as real samples except that no air was drawn during sampling. 6 blanks were used. Detection limits were taken as 3 times the standard deviation of the blank concentrations. All concentrations of the samples in the paper have been corrected for blank values. The detection limits for the species in the aqueous solution were 10 μg/L for Cl-, 15 μg/L for NO3

-, 4 μg/L for SO42-, 5 μg/L for

NH4+, 0.3 μg/L for Na+, 0.05 μg/L for Mg2+, 0.8 μg/L for K+, and 0.7 μg/L for Ca2+.

2.3. Meteorological conditions and air mass back trajectories

Generally, the weather was relatively cool and moderately humid during the sampling campaign. The hour-averaged daily temperature was between 5 and 13 ºC, with an average of 9.4 ºC, and the hour-averaged daily relative humidity was between 40 and 60%, with an average of 53%. To estimate the effect of long-range air mass transport on PM mass concentration and ionic components, air mass back-trajectories were computed by using the US National Oceanic and Atmospheric Administration’s Hybrid Single-Particle Lagrangian Integrated Trajectory model (HYSPLIT version May 2007) utilizing the Final Model Run (FNL) meteorological data set. Four-day 6-h interval backward trajectories were computed for heights of 200 m above ground level. Three characteristic air mass regimes were found: 1) the westerly (including south-westerly and north-westerly) regime (Fig. 1a): air masses started from the Atlantic Ocean landed in west England, bringing sea-salt enriched air parcels to the sampling site. This regime, denoted the maritime air regime, occurred for four sets of samples collected on the days of 30 October to 1 November and 11-13 November; 2) the northerly regime (Fig. 1b): air masses originated from the North Atlantic, reached Scotland and became mixed with terrestrial air, moved to the south reaching the sampling site with chemical modifications. This regime encompassed 3 sets of samples collected on the days of 01-04 November; 3) the southern air regime (Fig. 1c): air masses originated from the southern England circulated to the west, reaching Ireland, and turning to the east all the way to Sheffield. This regime occurred for 6 sets of samples collected from 5 to 11 November. The regimes of 2 and 3 were both characterised by high concentration of secondary aerosols mixed with some minor to moderate concentrations of sea-salts. The samples from these two regimes are grouped together in subsequent discussions (called terrestrial regime) as opposed to the maritime regime. These terms just reflect the main influence, as air masses over England are always to some extent affected by the proximity to the ocean. No air mass from continental Europe was found during this period.

Fig. 1. Back-trajectories of air mass reaching the sampling site. a. Maritime regime. b. Terrestrial regime (northerly air mass).c. Terrestrial regime (southern air mass). Note, in Fig.1a and b, trajectories start outside the figures.

3. Results and discussions

3.1. Mass concentration and size-resolved mass distribution

3

The daily PM10 mass (the sum of the 12 stages) concentrations of 9 measurements ranged from 8.6 to 33.7 μg m-3, with an average of 22.3 μg m-3, and a standard deviation of 8.8 μg m-3. The mass concentrations are well below the World Health Organization 24-hour mean air quality guidelines for particulate matter (50 μg m-3) (WHO, 2006) and close to the guideline for annual mean (20 μg m-3). 60% of the average mass was in the submicron particle size range (<0.956 μm. Hereinafter, particles with diameters less than this value are called submicron or fine particles; those with larger diameters are called supermicron or coarse particles).

0

5

10

15

20

25

30

35

0.0 0.1 1.0 10.0Dp (μm)

dm/d

logD

p (μg

m-3

)

30-31, Oct.01-02, Nov.02-03, Nov.03-04, Nov.04-05, Nov.09-10, Nov.10-11, Nov.11-12, Nov.12-13, Nov.

Fig. 2. PM10 Mass size distributions of 9 sets of samples

Size distributions of particulate mass were obtained by plotting dm/dlogDp as a function of Dp (particle diameter), where dm is the mass concentration found on the stage, and dlogDp is the difference of logarithms of stage size boundaries. The geometric average diameters for the stage size boundaries are used. The result is a discrete size distribution. Atmospheric aerosols are typically described as consisting of three modes, the nucleation mode (0.01-0.1 μm in diameter), the accumulation mode (0.1-1.0 μm), and the coarse mode (> 1 μm) (Meng, 1994). There appeared to be a nucleation mode for most of the samples (see Fig. 2), although more or less overlapping with the accumulation mode. It can be inferred that there existed some freshly produced particles, probably carbonaceous particles from nearby industries or car exhaust. It was not surprising to find such particles at this sampling site, since it is surrounded by highways and there was considerable traffic during sampling. Accumulation and coarse mode size ranges were found for all samples, but relative peak heights varied considerably. Samples collected on 2-3 and 3-4 November had a dominant accumulation mode (see Fig. 2), peaking around 0.32-0.49 μm, while the contribution from the coarse mode was very small, indicating dominance of secondary aerosol sources. The sample collected on 11-12 November had a very pronounced coarse mode (see Fig. 2), suggesting largely soil or sea-salt sources.

3.2. Ionic species

3.2.1. Ion balance and ion mass concentrations

4

The equivalent sums of anions (Cl-, NO3-, and SO4

2-) and cations (NH4+, Na+, K+, Mg2+, and Ca2+)

were calculated for each size fraction of each sample. Total anions and cations are strongly correlated, with a correlation coefficient of 0.98 and a slope of 1.12 (Fig. 3). The 12% discrepancy might have been caused by the incomplete inclusion of some other water-soluble components and/or by uncertainties in analyses. Among the 9 sets of samples with mass data, the total ion mass concentrations ranged from 2.95 to 9.81μg m-3, with an average of 5.76 μg m-3, accounting for 26% on average of the PM10 mass concentration. Cl-, NO3

-, SO42-, Na+, and NH4

+ were the main ions, accounting for 24%, 24%, 21%, 14%, and 10% of total ion mass, respectively.

y = 1.12x - 0.51R2 = 0.98

0

20

40

60

80

100

0 20 40 60 80 100Total Cations (nequ.m-3)

Tota

l Ani

ons (

nequ

.m-3

)

Fig. 3. Ionic balance for 13 sets of samples (n=156)

3.2.2. Mass size distributions of NH4+, SO4

2-and NO3- ions

The mass distributions of the anions and cations are presented in Fig. 4a and Fig. 4b for terrestrial and maritime regimes, respectively. The distribution of SO4

2- was bimodal with one peak in the submicron size range (maximum at 0.32-0.49 μm) and a secondary peak in the supermicron size range (maximum at 1.97-3.11μm), albeit the latter was minimal for SO4

2- in the “terrestrial regime” distribution. Nitrate showed bimodal distribution in the terrestrial regime but unimodal distribution in the maritime regime. Ammonium was unimodal, peaking at 0.32-0.49 μm. No nucleation mode was found for these species, prompting the hypothesis of negligible impact by local pollution sources.

Table 1 Concentrations of ions (in ng m-3) from different air mass origins in 13 sets of samples collected in Sheffield City, 2006.

NH4+ Na+ Mg2+ K+ Ca2+ Cl- NO3

- SO42-

1. Maritime regime Average 193 1140 143 74 29 1951 278 807

Standard deviation 116 724 98 26 11 1400 47 282Median 165 937 116 75 29 1517 279 766

2. Terrestrial regimeAverage 1099 535 91 353 41 1002 2481 2033

Standard deviation 758 281 48 230 27 433 1662 1116Median 794 508 64 341 27 821 1626 2221

Fine mode NH4+ may originate by reaction of NH3 vapours with acidic gases such as H2SO4, HNO3,

and HCl, or NH3 vapour may react with or condense on an acidic particle surface of anthropogenic origin (Zhuang et al., 1999a; Parmar et al., 2001). The major source of NH3 in England is agriculture, especially

5

the decomposition of excrement and urine produced by grazing animals (Harrison and Pio, 1983). NH4+

was also found in the supermicron fraction, albeit in much smaller amounts (see Fig. 5). Coarse NH4+

may be formed by the reaction of NH3 gas on sulphate and nitrate enriched sea-salt and soil particles when excess NH3 is available, in the presence of moisture (Wall et al., 1988; Zhuang et al., 1999a; Chandra Mouli et al., 2002). NH4

+ concentrations varied considerably with the origins of the air masses, for example the average concentration in samples belonging to the “maritime regime” was 193 ng m-3

compared to 1099 ng m-3 in samples from the so-called “terrestrial regime” (Table 1). This is plausible as air masses from the “maritime regime” travelled shorter distances and hence had shorter residence times over the land, where the ammonium sources were.

(a)

0

1000

2000

3000

0,0 0,1 1,0 10,0Dp (μm)

dm/d

logD

p (n

g m-3

)

Cl

NO3

SO4

NH4

Na

Mg

K

(b)

0

1000

2000

3000

0,0 0,1 1,0 10,0Dp (μm)

dm/d

logD

p (n

g m

-3)

Cl

NO3

SO4

NH4

Na

Mg

K

Fig. 4. Water-soluble components size distributions. (a) Terrestrial regime (9 sets of samples averaged). (b) Maritime regime (4 sets of samples averaged).

Sulphate is known to be a typical long-range transported (LRT) component having similar concentration over large areas (Harrison and Jones, 1995). Sulphate in ambient air mainly comes from oxidation of sulphur-containing precursors such as SO2, H2S, CS2, COS, and DMS (Xiu et al., 2004). Among the sulphur-containing compounds, sulphur dioxide is the largest contributor (Khoder, 2002). Sulphur dioxide can be oxidized to H2SO4 by gas-phase or multi-phase reactions with OH, H2O2, and O3,followed by condensation or nucleation of H2SO4 both onto pre-existing particles and into new particles. Reactions with NH3 imply first that sulphuric acid is partly neutralized and strongly acidic (NH4)HSO4 isformed, which can further react with NH3, producing the weakly acidic (NH4)2SO4. Particulates containing mainly (NH4)HSO4 can be considered as moderately aged, while particulates containing mainly (NH4)2SO4 can be viewed as highly aged (Ocskay et al., 2006 and reference therein). Fine sulphate particles can be transported far away from their sources since their atmospheric lifetime is in the order of several days. Sulphate exhibited higher concentrations during “terrestrial regime” sampling (see Table 1) and was mainly found in the fine mode (Fig. 5), which is consistent with literature (Zhuang et al., 1999a, b; Temesi et al., 2001). Coarse sulphate particles, found in less amounts, may originate from sea-salt and/or sulphate formed on coarse sea-salt or soil particles. The marine source of sulphate was investigated using Na+ as a reference element assuming all Na+ to be of marine origin. The supermicron fraction in samples from the “maritime regime” had a SO4

2-/Na+ mass ratio (0.29) near that of sea-salt (0.25) indicating sea-salt as a major source. (Seawater composition ratio values were calculated from Seinfeld and Pandis, 2006), All the rest of the samples exhibited higher SO4

2-/Na+ ratios, with 0.74 for the “terrestrial regime” supermicron fraction, 33 for “terrestrial regime” submicron particles, and 5.6 for “maritime regime” submicron fraction. These results suggested much stronger sulphate terrestrial sources than maritime sources. It also indicated that sulphate variation was mainly determined by LRT rather than by local pollution.

6

Fine nitrate particles are formed by homogeneous gas-phase oxidation of nitrogen oxides to gaseous nitric acid, which is followed by the reaction with gaseous ammonia to form highly volatile NH4NO3(Ocskay et al., 2006). NH4NO3 in the atmosphere exists in reversible phase equilibrium with gaseous HNO3 and NH3:

NH4NO3 (s) NH3 (g) + HNO3 (g) (1)

The NH4NO3 dissociation depends on temperature and relative humidity (Seinfeld and Pandis, 2006), which may result an underestimation of ammonium and nitrate in the aerosols. It also implies a possible nitrate size shift. The size shift lies in the fact that NH4NO3 (often fine particles) releases HNO3, that willbe available for other chemical reactions in the atmosphere, such as reactions with coarse soil and sea-salt particles. The reaction between nitrate and sea-salt is described as:

NaCl (aq, s) + HNO3 (aq) NaNO3 (aq, s) + HCl (g) (2)

As a result of reactions (1) and (2), nitrate is transferred from fine particles to coarse particles. In our samples 64% of the nitrate was in the fine mode. Since gaseous HCl is formed, equation (2) is an important mechanism explaining the so-called “chloride loss”, or “chloride depletion”, or “chloride deficiency”, terms that are used to describe the excess of Na+ relative to Cl-. Equation (2) may also imply a route for chloride size redistribution: the chloride released may combine with ammonia to form finer ammonium chloride particles when the temperature is favourable.

(a)Cl

7 %

NO333 %SO4

33 %

NH420 %

K6 %

Na1 %

(b)

NO333 %

SO414 %

NH45 %

Na19 %

K3 %

Cl26 %

(c)

NO39 %

SO450 %

NH417 %

Na9 %

K3 %

Cl12 %

(d)Na

31 %

Cl53 %

K1 %

NO35 %

NH41 %

SO49 %

Fig. 5. Average, relative mass contributions of ionic species. (a) Terrestrial regime submicron size range. (b) Terrestrial regime supermicron size range. (c) Maritime regime submicron size range. (d) Maritime supermicron size range.

3.2.3.Mass size distribution of Na+, Mg2+ and Cl- ions

7

The size distributions of Na+ and Mg2+ are similar (Fig. 4), suggesting their possible coexistence in the atmosphere as sea-salt particles. The modal peaks were all between 1.97 and 3.12 μm, which were smaller than reported at costal sites of Terra Nova Bay, Antarctica (Hillamo et al., 1998), and Tokaimura,

a seaside city in Japan (Fu and Watanabe, 2004). This is likely due to Sheffield City’s longer distance to the coasts, with the nearest ca. 100 km away. Air masses transported to the site appeared to be effectively depleted for relatively larger sea-salt particles. Cl-, besides the peak in the supermicron range, on some days had a second peak in the submicron size around 0.49 μm, and the reason for which will be addressed in the following discussion.

y = 0.13x + 1R2 = 0.92

Seawater ratio 0.12

020406080

100120

0 200 400 600 800Na+ (ng m-3)

Mg2+

(ng

m-3

)

Fig.6. Correlation between Na+ and Mg2+ (13 sets of samples, 12 size fractions, n=156)

Statistically significant correlations were found between Na+ and Mg2+ in all samples (Fig. 6). Since the slope of the regression (0.13) was close to that of seawater (mass ratio 0.12), the dominant source is most likely to be sea-salt (road salt can be excluded at the time of the year). The average Ca2+/Na+ ratio (0.039) for all samples was close to its seawater value (0.038), which implied a negligible amount of non-sea-salt-Ca2+ (nss-Ca2+). This indicated that water-soluble components from soil particles constituted a negligible fraction, as nss-Ca2+ is considered as a soil tracer (Zhuang et al., 1999b).

(a)

y = 1.34x - 11.63R2 = 0.90

0

200

400

600

800

1000

1200

1400

0 200 400 600 800

Na+ (ng m-3)

Cl- (n

g m

-3)

(b)

y = 7.73x + 30.21R2 = 0.18

0

50

100

150

200

250

300

0 50 100 150

Na+ (ng m-3)

Cl- (n

g m

-3)

(c)y = 1.87x - 23.38

R2 = 0.99

0200400600800

100012001400

0 200 400 600 800Na+ (ng m-3)

Cl- (n

g m

-3)

(d)

y = 1.32x - 1.199R2 = 0.94

0

50

100

150

200

250

300

0 50 100 150Na+ (ng m-3)

Cl- (n

g m

-3)

Fig.7. Stage to stage correlation between Cl- and Na+. (a) Terrestrial regime supermicron (n=45). (b) Terrestrial regime submicron (n=63). (c) Maritime regime supermicron (n=20). (d) Maritime regime submicron (n=28).

8

9

The correlation between Na+ and Cl- varied for different source regimes (Fig. 7). Fig. 7a shows that the mass ratio of Cl- to Na+ in the terrestrial regime supermicron size was 1.34, less than that in seawater (1.80), which implies the so called “chloride depletion” did exist. Assuming that essentially all Na is from seawater and using the formula:

Cl-depletion=100% (Cl-

in sea-salt-Cl-in aerosol )/ Cl-

in sea-salt

=100% (1.80Na+in aerosol-Cl-

in aerosol) /1.80Na+in aerosol

The chloride loss was found to be 26% ( =19, n=9), which is not as pronounced as reported elsewhere (Zhuang et al., 1999b; Kerminen et al., 2000). This was expected at the time of the year when temperature was mostly below 10ºC. However, in the terrestrial regime submicron fraction Na+ and Cl- were poorly correlated, with R2 = 0.18 (n=63). It is often expected that fine sea-salt particles will lose more chloride, as they have much larger surface-to-volume ratio and longer atmospheric residence time (Zhuang et al., 1999b), which favours the chemical reaction (2). However, it was not the case in this study. On the contrary, excess chloride relative to sodium was found (see Fig. 7b). As discussed earlier, in the supermicron fraction, Cl- was highly correlated with Na+ and Mg2+, but in the submicron fraction Cl- wasmore closely associated with NO3

-, SO42-, NH4

+, and K+ (see Table 2). This suggests a source or sources other than sea-salt in the submicron size. This phenomenon has also been observed in some other studies (Harrison and Pio, 1983; Willison et al., 1989; Kaneyasu et al., 1999; Yao et al., 2002; Xiu et al., 2004). We attribute the excess chloride to emissions from coal combustion, and/or smelters. These sources of chloride may mask the chloride depletion in the fine particles.

Table 2. Pearson correlations between Cl- and other species

in terrestrial regime samples

Submicron Supermicron Cl-/NO3

- 0.70 0.01Cl-/SO4

2- 0.79 0.27Cl-/NH4

+ 0.76 -0.08Cl-/Na+ 0.42 0.98

Cl-/Mg2+ 0.62 0.96Cl-/K+ 0.75 0.22

The chemistry of air masses belonging to the maritime regime arriving at Sheffield also showed different patterns for different size ranges. The supermicron size particles showed no chloride loss, indicated by the slope value close to that of sea-water (Fig. 7c), while submicron particles suffered a chloride loss with an average of 27% ( =15%, n=4) (Fig. 7d). These observations can be explained by the fact that the sea-salts in this regime were relatively fresh and hence the coarse sea-salts were intact. However, the fine particles, due to their larger surface-to-volume ratio, were partially modified.

We also investigated the roles of nitrate and sulphate in chloride depletion. As there is no known major source directly emitting nitrate, the coarse nitrate mode is assumed to be from the reaction of HNO3or NOx with coarse particles, or reaction with ammonia on the surface of pre-existing coarse particles, such as sea-salt and soil. However, as the soil contribution for water-soluble components is negligible, it is most likely that the coarse nitrate particles must be formed through reaction with sea-salt particles (eqn. 2), or reaction with ammonia. For sulphate, it is much more complicated. Coarse sulphate could originate from sea-salt, from terrestrial based gypsum emissions, from the reaction with ammonium on the surface of coarse particles, and from the replacement of chloride. To simplify the calculation, gypsum was neglected, as nss-Ca2+ was minimal. The sulphate available for chloride depletion is then equal to total sulphate - (sulphate in sea-salt + sulphate associated with ammonium). In the terrestrial regime supermicron fraction, there was an excess of ammonium relative to sulphate, indicating no contribution from sulphate in chloride replacement. We conclude that the major player for chloride depletion was

10

nitrate. The excess of ammonium relative to sulphate also suggested that part of the coarse nitrate was bound to ammonium, and the rest served as a replacement for chloride. In the maritime submicron size fraction, calculation revealed that, sulphate contributed an average of 3.6% (ranging from 0 to 14%,

=5.8%) to the chloride depletion, with the rest from nitrate. Although the relative amount of sulphate was far larger than of nitrate (see Fig. 5c), the availability of abundant ammonium resulted in minor sulphate contribution to chloride depletion. Thus we can reasonably assert that coarse modal sulphate mainly consisted of sea-salt sulphate and sulphate combined with ammonium, and coarse modal nitrate was due to the replacement of chloride in the sea-salt and nitrate bound with ammonium.

3.2.4. Mass size distribution of potassium

The mass size distributions of potassium were mostly unimodal peaking around 0.49 μm (see Fig. 4a), with average ratio of K+/Na+ (0.31), much larger than in seawater (0.036), implying source(s) of K+ other than sea-salt. Only four days displayed a minor peak in coarse size range (see Fig. 4b). Correlation of K+

with Na+ in the coarse size of these four samples was quite good, with a correlation coefficient of 0.92 and a K+/Na+ ratio of 0.036 (seawater 0.037), which was a strong indication of maritime origin. Apotassium peak in the coarse mode can also be expected if there is a significant soil-derived source. The observed patterns support our assertion that water-soluble components from soil were negligible. Fine mode K+ may be released into the atmosphere by burning of plant materials and emission through vegetative respiration (Parmar et al., 2001 and reference therein). Three samples collected on 3-6 November displayed higher K+ concentrations (3.8-fold) than the rest of the samples. Statistics T-test showed that the differences between these two sample groups were significant (P=0.05). The unusual occurrences of K+ were most probably attributable to customary “bonfire” events in which fireworks were burned.

4. Conclusions

The measured size distribution and chemical composition in fractionated fraction in PM10 have provided important information to the understanding of the sources, behaviour and mechanism of formation of particles in the urban atmosphere of Sheffield. Trajectory computations showed that there were two main regimes during sampling, denoted terrestrial and maritime regimes. The water-soluble composition revealed a mixture of secondary aerosols and sea-salt aerosols, the contribution of which depended on the air mass sources. The terrestrial regime air masses were higher in K+ and secondary aerosols containing SO4

2-, NO3-, NH4

+, and the maritime regime air masses carried more sea-salt related components, such as Na+, Cl-, and Mg2+.

Acknowledgements

The authors are grateful to N. Bramall, J. Seuma, and Department of Chemistry, The University of Sheffield, for assistance in sampling, and P. Collins and Corus UK Ltd for providing financial support and facilities for sample analysis.

References

Chandra Mouli, P., Venkata Mohan, S., Jayarama Reddy, S., 2002. A study on major inorganic ion composition of atmospheric aerosols at Tirupati. Journal of Hazardous Materials 96, 217-228.

Fu, F.F., Watanabe, K., 2004. Seasonal characteristics of chemical compositions of the atmospheric aerosols collected in urban seaside area at Tokaimura, eastern central Japan. Journal of Geophysical Research 109, D20212/1-D20212/9.

Harrison, R.M., Pio, C.A., 1983. Major ion composition and chemical associations of inorganic atmospheric aerosols. Environmental Science and Technology 17, 169-174.

11

Harrison, R.M., Jones, M.R., 1995. The chemical composition of airborne particles in the UK atmosphere. Science of the Total Environment 168, 195-214.

Hillamo, R., Allegrini, I., Sparapani, R., Kerminen, V.-M., 1998. Mass size distributions and precursor gas concentrations of major inorganic ions in Antarctic aerosol. International Journal of Environmental Analytical Chemistry 71, 353-372.

John, W., Wall, S.M., Ondo, J.L., Winklmayr, W., 1990. Modes in the size distributions of atmospheric inorganic aerosol. Atmospheric Environment 24A, 2349-2359.

Kaneyasu, N., Yoshikado, H., Mizuno, T., Sakamoto, K., Soufuku, M., 1999. Chemical forms and sources of extremely high nitrate and chloride in winter aerosol pollution in the Kanto Plain of Japan. Atmospheric Environment 33, 1745-1756.

Kerminen, V.M., Teinilä., K., Hillamo., R., 2000. Chemistry of sea-salt particles in the summer Antarctic atmosphere. Atmospheric Environment 34, 2817-2825.

Khoder, M.I., 2002. Atmospheric conversion of sulphur dioxide to particulate sulphate and nitrogen nitric acid in an urban area. Chemosphere 49, 675-684.

McFiggans, G., Artaxo, P., Baltensperger, U., Coe, H., Facchini, M.C., Feingold, G., Fuzzi, S., Gysel, M., Laaksonen, A., Lohmann, U., Mentel, T.F., Murphy, D.M., O’Dowd, C.D., Sinder, J.R., Weingartner, E., 2006. The effect of aerosol composition and properties on warm cloud droplet activation. Atmospheric Chemistry and Physics 6, 2593–2649.

Meng, Z.Y., 1994. On the sources of the submicrometer droplet mode of urban and regional aerosols. Aerosol Science and Technology 20, 253-265.

Moreno, T., Gibbons, W., Jones, T., Richards, R., 2004. Geochemical and size variations in inhalable UK airborne particles: The limitations of mass measurements. Journal of the Geological Society (London, United Kingdom).

Ocskay, R., Salma, I., Wang, W., Maenhaut, W., 2006. Characterisation and diurnal variation of size-resolved inorganic water-soluble ions at a rural background site. Journal of Environmental Monitoring 8, 300-306.

Pakkanen, T.A., Kerminen, V.M., Hillamo, R.E., Mákinen, M., Mákelá, T., Virkkula, A., 1996. Distribution of Nitrate over sea-salt and soil derived particles – implications from a field study. Journal of Atmospheric Chemistry 24, 189-205.

Parmar, R.S., Satsangi, G.S., Kumari, M., Lakhani, A., Srivastava, S.S., Prakash, S., 2001. Study of size distribution of atmospheric aerosol at Agra. Atmospheric Environment 35, 693-702.

Schwartz, J., Dockery, D.W., Neas, L.M., 1996. Is daily mortality associated specifically with fine particles? Journal of Air and Waste Management Association 46, 927-939.

Seinfeld, J.H., Pandis, S.N., 2006. Atmospheric Chemistry and Physics. From Air Pollution to Climate Change. John Wiley and Sons, New York.

Temesi D, Molnár A, Mészáros E, Feczkó T, Gelencsér A, Kiss G, Krivácsy Z., 2001. Size resolved chemical mass balance of aerosol particles over rural Hungary. Atmospheric Environment 35, 4347-4355.

12

Wall, S.M., John, W., Ondo, J.L., 1988. Measurement of aerosol size distributions for nitrate and major ionic species. Atmospheric Environment 18, 977-981.

WHO. “The World Health Report 2002.” World Health Organization. 2002; http://www.who.int/whr/2002/en/

WHO. 2006. “WHO air quality guidelines global update 2005.” World Health Organization. http://www.euro.who.int/Document/E90038.pdf

Willison, M.J., Clarke, A.G., Zeki, E.M., 1989. Chloride aerosols in central northern England. Atmospheric Environment 23, 2231-2239.

Xiu, G.L., Zhang, D.N., Chen, J.Z., Huang, X.J., Chen, Z.X., Guo, H.L., Pan, J.F., 2004. Characterisation of major water-soluble inorganic ions in size-fractionated particulate matters in Shanghai campus ambient air. Atmospheric Environment 38, 227-236.

Yao, X.H., Chan, C.K., Fang, M., Cadle, S., Chan, T., Mulawa, P., He, K.B., Ye, B.M., 2002. The water-soluble ionic composition of PM2.5 in Shanghai and Beijing, China. Atmospheric Environment 36, 4223-4234.

Zhuang, H., Chan, C.K., Fang, M., Wexler, A.S., 1999a. Size distribution of particulate sulphate, nitrate, and ammonium at a coastal site in Hong Kong. Atmospheric Environment 33, 843-853.

Zhuang, H., Chan, C.K., Fang, M., Wexler, A.S., 1999b. Formation of nitrate and non-sea-salt sulphate on coarse particles. Atmospheric Environment 33, 4223-4233.