UNIVERSITY OF GHANA, LEGON
CHARACTERIZATION AND SOURCES OF 
AIR PARTICULATE MATTER AT 
KWABENYA, NEAR ACCRA, GHANA
BY
This thesis is submitted to the University of Ghana, 
Legon in partial fulfillment of the requirement for the 
award o f PhD Physics degree.
MAY 2009

DECLARATION
Candidate’s Declaration
I hereby declare that except for the references to other peoples work, which have been 
duly cited, this thesis is the result of my own research and that it has neither in part nor 
whole been presented for the awar^-of any degree elsewhere.
CANDIDATE: INNOCENT JOY KWAME ABOH
DATE:
Supervisors’ Declaration
We hereby declare that the preparation and presentation of the thesis were supervised in 
accordance with guidelines on supervision of thesis laid down by the University of Ghana.
PROF. G. K. TETTEH
DATE:
EVANG. PROF. E. K. OSAE
DATE:
DR. V. C. K. KAKANE
DATE :
THESIS RELATED PUBLICATIONS
Part of the work presented in this thesis appears in the following publications and 
presented as posters at International Conferences.
Publications
1. Identification of Aerosol Particle sources in semi-rural area of Kwabenya, near Accra, 
Ghana by EDXRF - Innocent Joy Kwame Aboh, Dag Henriksson, Jens Laursen, 
Magnus Lundin, Francis Gormon Ofosu, Niels Pind, Eva Selin Lindgren and Tomas 
Wahnstrom.X-rays Spectrometry, Vol38, pp. 348, (2009)
2. Possible Sources of Atmospheric Aerosol during 2005/06 Harmattan Season at 
Kwabenya, Ghana - I. J. K. Aboh and F. G. Ofosu, Journal of Applied Science and 
Technology (JAST), Vol 13 No. 1 & 2, pp. 55, 2008
3. Levels and sources o f particulate lead in air at Kwabenya, near Accra - F. G. Ofosu 
and I. J. Kwame Aboh, Journal of Ghana Science Association, Vol. 10 No. 1, pp. 1, 
(2008)
Poster Presentations
1. Identification of Aerosol Particle Source in Semi-rural area of Kwabenya, near Accra, 
Ghana - Innocent Joy Kwame Aboh, Dag Henrikssonl, Jens Laursen, Magnus 
Lundin, Francis Gormon Ofosu, Niels Pind, Eva Selin Lindgren and Tomas 
Wahnstrom (Poster presented at EXRS2008, Cavtat, Dubrovnik, Croatia, June 16-20, 
2008)
2. Characteristics and source assignment of aerosol particles in a semi-urban area in 
Ghana during the Harmattan season using EDXRF analysis - Innocent Joy Kwame
Aboh, Dag Henrikssonl, Jens Laursen, Magnus Lundin, Francis Gormon Ofosu, 
Niels Pind, Eva Selin Lindgren and Tomas Wahnstrom (Poster presented at 
EUROanalysisXIVConference, Antwerp, Belgium, 9-14 September 2007)
Articles submitted for publication
1. Determination of mass, element and black carbon concentrations in 2005/06 
Harmattan aerosol at Kwabenya -  (near Accra ) -  Ghana -  I. J. Kwame Aboh and F. 
G. Ofosu (Paper presented to Journal Ghana Science Association - 2008)
2. Seasonal Variation o f Suspended Particulate Matter at Kwabenya, near Accra -  F. G. 
Ofosu and I. J. Kwame Aboh (Paper presented to Journal of Applied Science and 
Technology - 2008)
ABSTRACT
Gravimetric, reflectometric and elemental analyses have been carried out on airborne 
particulate matter sampled in a semi-rural area of Kwabenya, near Accra-Ghana. The 
PM 10 aerosols were sampled using a Gent sampler, size segregating the aerosol into 
coarse (PM10.2.5) and fine (PM2.5) fractions. The data and derived information were 
generated from 216 days of sampling spanning a period o f about 14 months, 28th 
December 2005 to 12th February 2007. The particulate matter (PM) at Kwabenya was 
dominated by the coarse particulates and showed low levels during the Rainy season and 
high levels during the Harmattan period. The levels measured during the 2006/07 
Harmattan were very high. The mass concentration for the measuring period were in the 
following ranges; coarse (PM 10-2 .5) fraction (0.16 - 1794.01 ^ig/m3); PM2 .s(fme) fraction 
(0.50 - 430.23 ng/m3) and PM 10 (0.87 ng/m3 to 2064.89 |ig/m3). Additional information 
about the ambient air was obtained through the subsequent determination of elemental 
concentration using energy dispersive x-ray fluorescence (EDXRF) analygia^and black 
carbon (BC) concentration through the “black smoke method”. The/^lpfnents iafe^ti|led 
and quantified with the Quantitative X-ray Analysis System (QXAS^ package software 
were: Al, Si, S, Cl, K, Ca, Ti, V, Cr, Mn, Fe, Ni, Cu, Zn, Br, Rb, Sr an^ i^ itL ^ i^coa rse  
fraction. The following elements were identified and quantified in the fine fraction: Al, Si, 
S, K, Ca, Ti, Mn, Fe, Cu, Zn, Br, Rb, Sr and Pb. Validation of the quantitative methods 
with the standard reference filter SRM2783 gave very good agreement (within ±15%) for 
most elements analysed except for Ni (±43%)which was very close to the detection limit. 
The elemental concentrations in the two fractions vary from season to season. Using 
simple correlation analysis some elements correlate, the elemental correlations also vary
iv
from season to season, for example during the Harmattan S, Cl, V, Br and Sr correlated 
very well but during the Rainy season S did not correlate with V and Br. This could serve 
as possible source indicators. The BC concentration in the fine fraction (ranging from
data from some developed countries. A receptor model using principal component and 
regression analysis was used to identify sources contributing to the air particulate matter 
at Kwabenya. The species used in the model were mass, BC and elemental concentrations. 
The following major sources were identified in the coarse aerosol: Soil/Dust, 
Biomass/LDT and Sea aerosol. In the fine aerosol the following sources were identified: 
Soil/Dust, Biomass/LDT and some industrial sources. The contribution of the sources to 
the PM load varied from season to season, There was very good agreement between the 
experimental and model data (mass, BC and elemental concentrations). Comparing the 
data with WHO limit (50 ^grrf3 for 24-hour mean) and Ghana EPA guideline limit (70 
Hgm' 3 for 24 hours) for PMio, a total of 185 and 130 days respectively out of 216 days 
had values above these limits. For PM 2 5 a total of 60 days had values exceeding the 
WHO limit (25 (agm' 3 for 24-hour mean). The levels of S, Ni and Pb were also 
comparable to industralised countries. There is the need for some mitigation measures to 
curb the emission of these elements and fine BC.
0.01 to 5.97 figm"3) was generally higher than in the coarse fraction and comparable to
v
ACKNOWLEDGEMENTS
I would like to thank my IAEA “sandwich” supervisors Professors Eva Selin Lindgen 
and Dag Henrikson of the University of Boras, Sweden, for their advice, guidance, moral 
and material support during my stay in Sweden. I also wish to thank Prof. G. K. Tetteh, 
Evang. Prof. E. K. Osae and Dr. V. C. K. Kakane, my supervisors at the University of 
Ghana, Legon, for their encouragement and guidance. I am very grateful to Prof. E. H. K. 
Akaho, the Director General o f the Ghana Atomic Energy Commission (GAEC) for
My sincere thanks goes to the International Atomic Energy Agency (IAEA) for granting 
me the “sandwich” programme at the University of Boras, Sweden. I cannot forget the 
useful suggestions and advice of Dr. Andre Markowicz, Rev. Dr. S. Akoto Bamford (now 
at SNAS, Legon) and the personnel at the X-ray laboratory at the Seibersdorf Laboratory 
of the IAEA. The following officials of the IAEA also deserve special mention for their 
assistance -  Dr. Neil Jarvis, the country officer for Ghana, Ms. Marie-Pierre Bakhoum 
and Thoko Mueller who administered the “sandwich” programme.
The equipment and facilities used for this work were located at different places. I would 
therefore like to acknowledge the assistance and collaboration of the following people 
and their institutes:
• Prof. Jens Laursen of the Department of Natural Sciences, University of 
Copenhagen, Denmark for the use of their EDXRF Spectrometer.
VI
• Prof. Niels Pind of the Department of Chemistry, University of Arhus, Denmark 
for the model development and spectral fitting.
• The air pollution study group of the University o f Boras -  Prof. Eva Selin 
Lindgren, Prof. Dag Henriksson, Dr. Magnus Lundin and Dr. Thomas Wahnstrom 
for their advice, guidance and assistance.
• Mr. Francis G. Ofosu, my colleague at GAEC, for his assistance in sampling and 
spectral fitting and encouragement. This work would not have been possible 
without your dedication and effort.
I cannot forget the love and care shown me by Mrs. Eva 
made my stay in Boras very memorable and stress free.
There are a lot of people outside my “scientific cycle" that need special mention. Dr. 
Peter Atadja for his support and encouragement during the programme. To Messers. 
George Hanyo and Foster Agotse, I say a big thanks for being there for me. To my family, 
friends and colleagues whom I was inaccessible to during this programme, I owe a debt 
of gratitude and apology.
At this point I cannot forget the unwavering support and prayers of my wife -  Joyce and 
my children -  Sedem, Makafui, Dzidefo, Junior and Eli. Though they bore the blunt of 
my travels they never gave up.
To God be the Glory for all the great things He has done!
DEDICATION
This thesis is dedicated to my wife -  Mrs. Joyce A. S. Aboh 
and my children -  Dzidefo, Makafui and Sedem 
To God be the Glory!
TABLE OF CONTENT
TITLE PAGE 
DECLARATION
THESIS RELATED PUBLICATIONS 
ABSTRACT
ACKNOWLEDGEMENTS 
DEDICATION 
LIST OF FIGURES 
LIST OF TABLES
CHAPTER 1 -  INTRODUCTION
1.1. Background
1.1.3. Effect on Climate
1.1.4. Effect on Forest and Vegetation
1.1.5. Measures for Addressing Air Pollution in Ghana
1.2. Current Air Pollution Status in Ghana
1.2.1. Some activities contributing to air pollution
1.2.1.1. Introduction
1.2.1.2. Energy and industrialisation
1.2.1.3. Mining
1.2.1.4. Vehicular Emission
1.1.2. Health Effects
1.1.1. Overview
1.2.1.5. Street dust, soil and desert soil ■••12
1.2.1.6. Sea spray •■■13
1.2.2. Issues ...14
1.2.2.1. Urbanisation ...14
1.2.2.2. Climate ...15
1.2.2.3. Increased Public Environmental awareness ...16
1.2.3. Trends in Monitoring ...17
1.2.3.1. Air quality monitoring and management at the
Takoradi thermal power plant ... 17
1.2.3.2. Biomonitoring of air pollution using lichens ...19
1.2.3.3. Air pollution monitoring of Kpone and Tema
Oil Refinery ... 19
1.2.4. Challenges ...25
1.3. Objectives of the Thesis ...26
1.3.1. General objective ...27
1.3.2. Specific objectives ...27
CHAPTER 2 -  AEROSOL PARTICLES AND SAMPLING INSTRUMENTS
2.1. Particle Formation ...28
2.1.1. Nucleation mode particles ...29
2.1.2. Accumulation mode particles ...33
2.1.3. Coarse mode particles ...34
2.2. Particle Transport ...34
x
2.2.1. Residence time ...34
2.2.2. Particle motion ...35
2.3. Particle Deposition Mechanisms ...37
2.3.1. Dry deposition ...38
2.3.2. Wet deposition ...39
2.4. Aerosol Sampling Instruments ...40
2.4.1. General considerations ...40
2.4.2. Integrated (discontinuous) sampling instruments ...44
2.4.2.1. Impactors ...44
2.4.2.2. Cyclones ...48
2.4.3. Direct-reading instruments ...51
2.4.4. Sampling o f gases ...52
2.5. Filters ...54
2.5.1. Fibrous filters ...55
2.5.2. Porous membrane filter ...56
CHAPTER 3 -  ANALYTICAL TECHNIQUES AS APPLIED TO AEROSOLS
3.1. Introduction ...57
3.2. X-rays ...60
3.2.1. X-ray fluorescence analysis ...60
3.2.2. EDXRF Spectrometry ...64
3.3. The Black Carbon Reflectometer ...69
3.4. Other Analytical Techniques ...71
CHAPTER 4 -  EXPERIMENTAL METHODS, ANALYSIS AND RESULTS
4.1. Selection of Study Area ■ ■ ■ 73
4.2. Experimental Procedures • • - 76
4.2.1. Sampling and gravimetry ••■76
4.2.2. Black carbon determination ■ • • 80
4.2.3. Energy dispersive x-ray fluorescence (EDXRF) analysis ...87
4.3. Meteorological Data ...92
CHAPTER 5 -  AIR POLLUTION AND RECEPTOR MODELING
5.1. Overview o f Air Pollution Modeling ... 96
5.1.1. Dispersion models ...97
5.1.2. Statistical models ...98
5.1.3. Receptor models ... 99
5.2. Receptor modeling used in this work ... 101
5.3. Results o f the Receptor Model ... 104
CHAPTER 6 -  QUALITY CONTROL AND DISCUSSION OF RESULTS
6.1. Quality Assurance and Quality Control as applied to aerosol analysis ... 125
6.2. Discussion of Results ...132
6.2.1. Mass concentration ...132
6.2.1.1. Coarse mass concentration ... 132
6.2.1.2. Fine mass concentration ... 133
xii
6.2.1.3. Comparison of this work with some selected Works and
International standards ... 134
6.2.2. Black carbon concentration ... 135
6.2.2.1. Coarse black carbon concentration ... 136
6.2.2.2. Fine black carbon concentration ... 137
6.2.2.3. Comparison of BC values with some selected Works ...137
6.2.3. Elemental concentrations ... 139
6.2.4. Source profiles ...141
6.2.4.1. Coarse fraction particulate matter ...142
6.2.4.2. Fine fraction particulate matter ... 148
6.2.5. Comparison between model and experimental data ... 153
CHAPTER 7 -  SUMMARY, CONCLUSIONS AND RECOMMENDATION
7.1. Summary ...154
7.2. Conclusions ... 157
7.3. Recommendations ...161
REFERENCES ...163
xiii
Figure
1-1 PMio Concentration in ambient air at some roadsides in Accra ...20
1-2 Ambient NOx concentration/ppm ...21
1-3 Ambient S 0 2 concentration/ppm ...22
1-4 PIXE spectra of some selected aerosol-loaded filter sample .. .24
2-1 Size range of aerosol particles ... 31
2-2 Idealised trimodal size distribution of fresh particles, showing general
relationships between the three common size ranges and their formation 
mechanisms in the atmosphere ... 32
2-3 Diagram illustrating the concept of particle inertia .. .37
2-4 Schematic Diagram outlining Emission, Transport, Transformation and
Sampling of airborne pollutants .. .41
2-5 Isokinetic Sampling ...41
2-6 Anisokinetic Sampling ...43
2-7 Cross-sectional view o f an impactor ...45
2-8 Dichotomous particle sampler for separating airborne particulate matter
into two size fractions .. .46
2-9 A Schematic diagram of a Cascade Impactor .. .47
2-10 Schematic diagram o f a cyclone ...50
4-1 Schematic diagram o f the Gent sampler used ...78
LIST OF FIGURES
xiv
4-2 Two stage stack filter cassette unit (SFU) loaded with filters and capped ...19
4-3 Schematic diagram o f SFU ... 79
4-4 Coarse mass concentration during the investigation period • • • 81
4-5 Fine mass concentration during the investigation period .. .82
4-6 PMio mass concentration during the investigation period ...83
4-7 Mean monthly Fine, Coarse and PM jo concentration ...84
4-8 Black carbon concentration during the investigation period ... 85
4-9 Percentage black carbon concentration during the investigation period ... 86
4-10 A sketch of the EDXRF spectrometer ...88
4-11 Daily variation of some selected coarse elements during the
investigation period ... 93
4-12 Daily variation of some selected fine elements during the
investigation period ... 94
4-13 Monthly variation of PM with total precipitation ... 95
5-1 Coarse mass -  model compared with experimental ...114
5-2 Coarse BC -  model compared with experimental ...115
5-3 Coarse Fe -  model compared with experimental ... 116
5-4 Coarse Cl -  model compared with experimental ... 117
5-5 Coarse mass -  comparison of model and experimental ... 118
5-6 Coarse BC -  comparison of model and experimental ...119
5-7 Coarse Fe -  comparison of model and experimental ...120
5-8 Coarse Cl -  comparison of model and experimental ... 121
6-1 Stages in the aerosol analysis used in this thesis work ...127
xv
6-2 Variation o f volume o f air sampled with sampling time o f the 
Gent sampler
6-3 Comparison of initial blank coarse filter mass with final blank 
coarse mass
. . .129
...130
xvi
LIST OF TABLES
Table
1-1 Mean concentrations of elements in the two lichen species
1 -2 Average concentration of elements in air particulate matter at Kpone 
and Tema Oil Refinery in Ghana
2-1 Control efficiency ranges for the three cyclone classification and type 
of aerosol
4-1 Validation of EDXRF spectrometer using SRM2783
4-2 Minimum detection limit for particulate matter on nuclepore filters
with EDXRF spectrometer
4-3 Concentration of coarse PM -  elements, BC and mass 
(28th December 2005 -  12 February 2007)
4-4 Concentration of fine PM - elements, BC and mass 
(28th December 2005 -  12 February 2007)
5-1 Concentration o f Coarse PM - Elements, BC and Mass 
(28th December 2005 -  3 1st March 2006)
5-2 Concentration of Fine PM - Elements, BC and Mass 
(28th December 2005 -  3 1st March 2006)
5-3 Concentration of Coarse PM - Elements, BC and Mass 
(4th April -  31st October 2006)
5-4 Concentration of Fine PM - Elements, BC and Mass 
(4th April -  31st October 2006)
5-5 Concentration of Coarse PM - Elements, BC and Mass
xvii
...25
...51
...89
...90
...91
...92
...105
...106
...107
...23
...107
(2nd November 2006 -  15th February 2007)
5-6 Concentration of Fine PM - Elements, BC and Mass
(2nd November 2006 -  15th February 2007)
5-7 PCA Factor Scores for coarse data
(2nd November 2006 -  15th February 2007)
5-8 PCA Factor Scores less tracer value (C000) for coarse data 
(2nd November 2006 -  15th February 2007)
5-9 Computed source contribution to the various coarse filter mass 
(2nd November 2006 -  15th February 2007)
5-10 Computed source profile values for coarse data
(2nd November 2006 -  15th February 2007)
5-11 Coarse PM Concentration Model and Experimental data fit - Element,
BC and Mass (28th December 2005 -  31st March 2006)
5-12 Fine PM Concentration Model and Experimental data fit - Element,
BC and Mass (28th December 2005 -  31st March 2006)
5-13 Coarse PM Concentration Model and Experimental data fit - Element,
BC and Mass (4th April -  31st October 2006)
5-14 Fine PM Concentration Model and Experimental data fit - Element,
BC and Mass (4th April -  3 1st October 2006)
5-15 Coarse PM Concentration Model and Experimental data fit - Element,
BC and Mass (2nd November 2006 -  15th February 2007)
5-16 Fine PM Concentration Model and Experimental data fit - Element, 
BC and Mass (2nd November 2006 -  15th February 2007)
xviii
108
108
110
111
112
113
122
122
123
123
124
124
6-1 Some output voltage measurement for black carbon determination ... 131
6-2 Coarse mass concentration ...133
6-3 Fine mass concentration ... 133
6-4 Comparison o f PM with results from literature ...134
6-5 Coarse black carbon concentration ...137
6-6 Fine black carbon concentration ... 137
6-7 Comparison of BC with results from literature ... 137
6-8 Coarse particle source profile (28th December 2005 -  31st March 2006) ... 144
6-9 Coarse particle source profile (4th April -  31st October 2006) ... 146
6-10 Coarse particle source profile (2nd November 2006 -  15th February 2007) ... 147
6-11 Fine particle source profile (28th December 2005 -  3 1st March 2006) ... 149
6-9 Fine particle source profile (4th April -  31st October 2006) ... 150
6-10 Fine particle source profile (2nd November 2006 -  15th February 2007) ... 152
xix
CHAPTER 1 
INTRODUCTION
1.1 BACKGROUND
1.1.1 OVERVIEW
Particles exist in the environment as suspensions in air and are commonly called aerosol. 
H inds' and Colbeck2 define aerosol as suspension o f  solid or liquid particles in a gaseous 
medium (air). The aerosol particles have a very wide size range from molecular clusters 
o f  0.00 l|rm  to fog and dust particles as large as a few hundred m icrom eters.2"4
Pollution is defined in the Tenth Report o f  the Royal Commission on Environmental 
Pollution as: “The introduction by Man into the environment o f  substances or energy 
liable to cause hazard  to human health, harm to living resources and  ecological systems, 
damage to structure or am enity or interference with legitimate use o f  the environment”.5 
Air pollution is the em itting o f  solid, liquid and gaseous material into the air environment. 
The em issions can originate from stationary or mobile sources and can sometimes involve 
some chemical or/and physical transformations before eventually being returned to
surfaces such as the soil, plants, trees, monuments, buildings, etc.
It is generally perceived that air pollution is one o f  the most vexing problems facing 
industrialized and developing countries because these m icroscopic particles are present 
everywhere in our environment. All things, both living and non-living, are exposed in 
varying degrees to air pollution. The degree o f  exposure depends on the location and 
activities (industrial, domestic, vehicular, etc) taking place.
The ubiquitous nature o f  aerosol particles exerts an important influence on the
hydrological and climatic system. For example, aerosol particles are found in both the
troposphere and the stratosphere and affect climate through scattering, transm ission and 
absorption o f  radiation .6 Studies have shown that a layer o f  small particles is always 
located in the stratosphere at altitude centered around 25 km at the equator and 17 km at 
the poles.7 Aerosol particles provide surfaces for heterogeneous chemical reactions which 
can influence gas-phase chem istry in the troposphere.8 A ircraft exhaust particles in the 
upper atmosphere are source o f  ice and cloud nuclei. B iomass burning, especially in the 
tropics, leads to significant perturbations o f tropospheric aerosol loading in these regions, 
and could be leading to changes in cloud behaviour.
The origin o f  the particles could be from natural processes (e.g. sea spray, dust, etc) or 
from  man-made processes known as anthropogenic processes (vehicle em issions, 
em ission from industries, waste incineration, etc). From the definition o f  air pollution 
above, chem icals such as sulphur dioxide from volcanoes or methane from the decay o f  
natural vegetation are not air pollutants, but sulphur dioxide from coal burning or 
methane from rice grow ing are air pollutants. However, human activities disturb natural 
systems hence the line distinguishing between the natural and anthropogenic processes 
become increasingly blurred. For example, radon -  a radioactive gas is a hazard in some 
types o f  geographical formation and the radon exposure by m ining in those formations 
increases exposure levels, which is air pollution. It is estimated that anthropogenic 
sources account for more than 30%  o f the aerosol particles measured by mass and are 
mainly small size or “fine” (PM< 2.5pm) particles.9 For example the anthropogenic 
component o f  sulphate, which is essential for cloud formation, exceeds 60% o f total 
sulphate particles over urban areas.10 Natural sources are usually o f  the larger 
aerodynam ic diameter (coarse particle mode) and include dust storm, sea spray, bush and 
forest burning.
2
Aerosol particles in the atmosphere are caused by a wide range o f  sources and have 
diverse compositions and characteristics. As a result, the effects o f  aerosol particles on 
the environment and humans are many and varied. Some o f  the effect could ju s t be a 
nuisance like dust deposited on a clean surface to serious climatic effect like 
environmental degradation, acid rains, impact on climate, impact on human and animal 
health.
1.1.2 HEALTH EFFECTS
For many centuries aerosol particles have been recognised for their potentially negative 
impact on human health and ecosystems. A lready in the H ippocratic Corpus (c. 400 B.C) 
medical links between health and air quality were m entioned." The Romans complained 
o f  foul air in ancient Rome. W ith the introduction o f  coal in London in the 13th century, 
regulations were introduced to reduce smoke problems. In the 17th century, John Graunt, 
a fellow  o f  the Royal Society concluded that the high death rate in London was partly as a 
result o f  coal burning. Awareness o f  the effects soot deposited on buildings had been 
noticed over 200 years ago and it was speculated that it had the same effect on the 
respiratory organs. In Italy, Bernardino Ramazzini catalogued many occupational risks o f 
air pollution in De Morbis A rtific ium .'2 In 1713, Linne and his colleagues described 
silicosis in Swedish m iners. '3 However, it was not until the second ha lf o f  the 20 th century 
that more typical air pollution as experienced today have forced scientists, politicians and 
decision makers to take note and put in place the necessary remedial actions.
Since the famous smog incident in 1952 in London, resulting in about 4000 excess 
deaths, it has been recognised that the negative influence on human health caused by 
aerosol particles had been grossly under-estimated. These negative effects o f  airborne 
particles have also been observed in the work environment. Epidem iological studies in
3
several countries have shown conclusively that there is a direct link between particulate 
air pollution and adverse health effects. Both indoor and outdoor particulate air pollution 
are responsible for these adverse health effects. The physical properties o f  aerosols affect 
the transport and deposition o f  the particles in the human respiratory system , whilst the 
chemical composition and the physical properties determ ine their impact on health. For 
two to three decades environmental authorities in developed countries have been active in 
sampling and analysing particles with (aerodynamic) diameters sm aller than 10 
micrometers, PMio. Studies during the last ten years indicate, however that the very 
smallest particles are even more hazardous to health, and smaller particles PM2.5, PMU 
and even nanoparticles have been more extensively studied.14_'9 Fine carbon or fly ash 
particles have been suspected to increase the respiratory toxicity o f  coexisting acidic air 
pollutants, by concentrating acid on their surfaces and so delivering it efficiently to the 
lower respiratory trac t.20 The number o f  death from air pollution is estimated to be more 
than 3 m illion/year, w ith about 1 m illion/year in cities and 2 m illion/year in rural settings. 
The largest contributor to the rural death toll is biomass fuel burning, as a source o f 
domestic energy, from indoor pollution.9'21 The causes o f  death are not only related to 
lung injuries but also to cancers and cardiovascular diseases. The increased awareness has 
lead to regulatory legislation regarding emissions and concentration levels o f  particulate 
matter (PM ) in many countries all over the world. In general, the “finer" or smaller the 
aerosol particle the more hazardous it is due to the size and morphology. Thus small 
particles:
•  can easily be transported over long distances far away from their sources
• can easily be inhaled and deposited in the lower part o f  the human respiratory 
tract
4
•  have larger surface area per unit mass and hence higher ability to absorb gas 
molecules and transport them to any part o f  the respiratory system to catalyze 
chemical and biochem ical reactions
•  consist o f  a large soluble fraction.9
In fact, the American Academy o f  Pediatrics in a statement issued before the Clean Air 
Scientific Advisory Comm ittee o f  the United States Environmental Agency (USEPA) on 
April 7th 2005 stated "Research has firm ly  established that exposure to high levels o f  
particula te matter impacts the ability o f  children's lungs to grow. The adverse effects o f  
air pollu tion  on the development o f  lung function  is seen in boys and  girls, regardless o f  
history o f  asthma, suggesting  that most children are susceptible to chronic effect o f  
breathing particulate air pollutants. When this damage occurs, it is irreversible, and  
reduced lung func tion  is a strong risk fa c to r  fo r  fu tu re  health consequences as an adult. 
Particulate matter air pollu tion  is also linked to other adverse respiratory health effects 
in infants and  children, such as asthma exacerbations, chronic cough, and  bronchitis 
symptoms " .22 O ther researchers had confirmed these assertions and had gone on to show 
that low-birth weight, preterm  births and infant mortality are increased in communities 
with high levels o f  particulate air pollutions.23"25
Ingestion o f  aerosol particles that have initially deposited on crop or soil and inhalation o f 
air are the two main ways in which aerosol particles enter the human body. It has been 
shown that aerosol particles w ith diameter less that 2.5 |jm  have more serious influence 
on the occurrence o f  respiratory diseases.26'29 Lead in aerosol has been implicated in the 
impairment o f  the development o f  children.30 Some o f  the aerosol inhaled may contain 
radioactive nuclides, attached to very small particles, which are quickly transported to the
5
lower lung where the attached radionuclides could cause cancer.3' Increased morbidity 
and mortality due to chronic diseases have also been attributed to aerosol particles. 32-34
1.1.3 EFFECT ON CLIMATE
Aerosol particles can affect the atmospheric radiation budget by either absorbing and/or 
scattering o f  solar radiation. For example, soot particles are known to have a high 
absorption for solar radiation. This absorption o f  solar radiation by aerosol particles leads 
to heating o f  the atmosphere while scattering o f  solar radiation results in cooling o f  the 
atmosphere. The main processes that determine the overall state o f  the climate system  are 
heating by incoming solar radiation and cooling by outgoing long-wave (infrared) 
terrestrial radiation.35'37 This heating or cooling o f  the atmosphere can result in 
perturbation o f  the radiation balance o f the troposphere known as radiative forcing.38,39 
Radiative forcing is the change in the balance between radiation com ing into the 
atmosphere and radiation going out. A positive radiative forcing tends on average to 
warm the surface o f  the Earth, and negative forcing tends on average to cool the surface. 
Aerosol particles contribute significantly to radiative forcing o f  clim ate.35,40"43 Black 
particles are most effective in absorbing solar radiation, hence black carbon (BC) in 
aerosol particles has been suggested to be the second most important component o f  global 
warm ing.44 The 1997-1998 severe El Nino induced droughts in Indonesia, M exico and 
Central America have been linked to high biomass burning around the w orld .45
1.1.4 EFFECT ON FORESTS AND VEGETATION
It is a well known fact that forests and vegetation play an important role in the 
environment. For example, through photosynthesis forest and vegetation are sources o f 
oxygen, help in preventing soil erosion, and provide food and shelter for animals. 
Therefore, any damage to the forest and vegetation could pose serious problems to human
6
beings, animals and the whole environment. Heavy metals in aerosol particles have been 
suspected to cause damage, since the beginning o f  the twentieth  century, to  the forest and 
vegetation. In 1912 Hedgecock documented that heavy metal aerosol particles from a 
nearby copper smelter were partially responsible for devastation o f  a large area 
(approximately 47,000 acres) in Tennessee.46 The partial destruction o f  forest and 
vegetation in other parts o f  the world by particle em ission has also been reported by other 
researchers.47"49 This damage ranges from damage to leaves and other parts o f  the tree to 
the more catastrophic destruction o f  the whole vegetation.
1.1.5 MEASURES FOR ADDRESSING  AIR  POLLUTION  IN GHANA
Most developed countries have recognized the adverse effects posed by aerosol particles, 
especially on human health, and this has resulted in increased routine monitoring o f 
atmospheric particles.50 In contrast however, there is almost no routine monitoring o f 
aerosols in developing countries, especially in A frica (except for South Africa). Hence 
data on concentrations as well as characteristics o f  particulate matter are almost non­
existent in developing countries most o f which are in the Southern hem isphere.51,52 To 
understand fully how aerosol particles behave under different climatic conditions there is 
the need to characterize aerosol particles in many different places including the 
developing countries, especially those near the equator. Such characterization would not 
only assist the developing countries to comply with international conventions and 
standards but would form an informed-basis for setting appropriate standards for the 
developing countries them selves as pertains in the developed countries.
7
1.2 CURRENT AIR  POLLUTION STATUS IN GHANA
1.2.1 SOME ACTIVITIES CONTRIBUTING TO AIR  POLLUTION
1.2.1.1 INTRODUCTION
In 1980’s Ghana undertook major economic reforms with the assistance o f  the World 
Bank and the International Monetary Fund (IMF) which resulted in the expansion and 
growth o f  the economy. Follow ing the economic recovery programme there has been an 
increase in industrial activities in the country. The activities that have seen growth are in 
the areas o f  m ining, road and housing Construction, plastic manufacturing, transport, 
metallurgical (steel and alum inium) reprocessing and fabrication, agricultural and other 
industrial concerns that depend mainly on chemicals (fertilizers, insecticides, pesticides, 
pharmaceutical, textiles, etc). The main industrial activities in the country are mining, 
agriculture, and manufacturing listed in order o f  capacity.
All these industrial activities contribute towards atmospheric pollution in the form o f 
gaseous pollutants, coarse and fine airborne particulate matter and solid waste. In addition 
there are the annual bush fires, sea spray, automobile exhaust gases, etc. These pollutants 
have their effect on human health, physical materials, the ecosystem  and climate but 
unfortunately no known data is available in terms o f  their social and economic cost.
The Government has made it a general requirement by law, for new  businesses to provide 
an environmental impact assessment (EIA) and for the existing ones to provide 
environmental management plan (EMP). National environmental standards to serve as a 
guideline for compliance have also been introduced. For some specific sources however, 
additional legislation have been put in place to control to some extent their emissions.
1.2.1.2 ENERGY AND INDUSTRIALIZATION
Ghana Poverty Reduction Strategy (GPRS) document is the country 's developmental plan 
aimed at the attainment o f  m iddle-income level (GDP > US$1000) by the year 2015. 
This vision envisages the almost tripling o f  the current GDP, which currently is US$420 
(2004), in ten years. To achieve this will require the establishment o f  more industries with 
electrical power consumption expected to increase as much as six to nine times the 
current installed capacity o f  1600 MW. The Power crisis o f  1998 has also shifted power 
generation from hydro to thermal power, which involves the combustion o f  high volumes 
o f  crude oil. Hence to be able to meet this enormous anticipated increase, more thermal 
plants would have to be built. Ghana has at present two thermal plants supporting the 
hydroelectric-generation, at Aboadze near Takoradi.
Except for m ining all other industries are located in the urban areas and lack o f  proper 
zoning procedures give rise to the situation whereby industries are inter-m ixed with 
residential areas. W ith the growth o f  industries, one can envisage high pollution levels 
beyond the already existing levels o f  which little efforts are being made towards its 
mitigation. A great number o f  the manufacturing industries are located at Tema near 
Accra and more than 90%  o f  the country’s industries are in the Southern part o f  the 
country.
Most o f  the energy for everyday domestic use in Ghana and the rest o f  sub-Saharan 
Africa are derived from biomass burning (fuel wood and charcoal) and in places in the 
urban areas some people go to the extreme o f  using worn-out vehicle tires as fuel. 
Biomass burning accounts for about 70% o f  total energy consumption in A frica .53'54 The 
annual bush fires that occur during the dry season, when temperatures are low, also
9
coincide w ith high demand o f  fuel wood and charcoal period contributing large amounts 
o f  soot and ash particles into the atm osphere.55
The industrial processes, the energy processes and the bush fires all release toxic 
elements and gases, some o f  which interacts w ith rain w ater to cause acid precipitation 
which has its far reaching ecological effects.
1.2.1.3 M INING
Arsenopyrite gold ore deposits are the predom inant gold deposits in the country .56 The 
extraction processes require high temperature roasting releasing arsenic, antimony and 
other heavy metals into the atmosphere. Since these activities occur in the forest zones 
the pollutants, carried down by rainwater affect the forest and hence crop cultivation. 
Vast areas have been affected by acid rain. The acid rains also pollute waters that flow  
into rivers, which are used for human consumption.
1.2.1.4 VEHICULAR  EMISSION
Ghana, it is currently estimated, has vehicular population growth rate o f  more than 70,000 
vehicles per annum w ith more than 65% concentrated in the A ccra-Tema Metropolitan 
Area. Like most A frican countries, the vehicles are concentrated on only a small fraction 
o f  the road network.57 M ost o f  the vehicles are second-hand w ithout catalytic converters. 
Automobile em ission is becom ing a serious problem, especially in the cities and is the 
main focus o f  the country’s environmental authorities.
The Government o f  Ghana has approved a Legislation Instrument L.I. 1732 [Petroleum 
(amendment) Regulation 2003] banning the production, importation, storage, sale and use 
o f  leaded gasoline in the country with effect from 1st January 2004. Consequently, 
M ethylcyclopentadienyl Manganese Tricarbonyl (MMT) - a manganese-based gasoline
10
anti-knock additive, has been in use in Ghana since January 2004 in unleaded Gasoline 
production. MMT is used in some countries including the United States, Canada and 
China. But there is a serious controversy about MMT use in some other countries 
especially in the European Union. The addition o f  MMT to gasoline supply has raised 
concern about public health risks associated with the inevitable increase in the 
environmental levels o f  manganese. In fact, several studies have reported that manganese 
causes significant health hazard in heavily air polluted areas.58'60
Combustion o f  unleaded gasoline in internal combustion engines causes the em ission o f  
various compounds such as carbon monoxide, carbon dioxide, hydrocarbons and other 
organic compounds. When MMT-containing fuel is combusted, small quantities o f 
manganese compounds are emitted in the automotive exhaust along w ith carbon 
monoxide, carbon dioxide and other gaseous compounds. The amount o f  manganese 
emitted, the m anufacturer’s claim , is very small because only a few parts per million o f  
manganese are added to the unleaded fuel and only a small percentage o f  this is emitted.
Manganese is an element that is essential to maintaining good health. It is also a natural 
component o f  the soil and is found in food, water and air. Adverse health effects o f 
manganese have been associated w ith organic Manganese compounds (pesticides) and 
inorganic m anganese.61'63 However, exposure to high concentrations o f  airborne 
manganese for prolonged periods o f  time can produce adverse health effects affecting 
primarily the nervous system  leading to slower visual reaction time, poorer hand 
steadiness, and impaired eye-hand coordination.64,65 Some researchers have long 
recognized a relationship between manganese intoxication and Parkinsonism .66"68
1.2.1.5 STREET DUST, SOIL AND DESERT SAND
A lot o f dust, about 900 -  1500 m illion tons, enters the atmosphere each year from natural 
soil dust. Dust is easily blown away by the wind and can travel over thousands 
kilometers. For example, Sahara dust has been found by satellite images to reach the 
Western Herm isphere and even the Arctic. The Sahara desert is regarded as the largest 
dust source in the w orld .69 Also dusts originating from China have caused haze in 
California in the USA. When dust passes over polluted areas they can be coated with 
sulphur due to chem ical process on their surfaces.70 These particles can then serve as 
giant cloud condensation nuclei (CCN), which may enhance the collision and coalescence 
o f  droplets and therefore increase warm precipitation formation and decrease the clouds 
albedo.71 This in effect means enhanced rainfall but this has been disproved by some 
researchers show ing that it rather reduces rainfall.72 For a country like Ghana, which is 
under threat from the downward movement o f  the Sahara, this could pose a lot o f 
problem for rainfall patterns which have been decreasing over the years.
In the urban areas o f  Ghana, a large quantity o f  construction sand and gravel is needed to 
support the rapid expansion o f  the towns and construction o f  roads and other 
infrastructure. These are brought in from the surrounding rural areas in haulage trucks 
moving on unpaved roads trapping a lot o f  mud which are then deposited on the paved 
roads. In addition, the status o f  the road network whereby several roads are unpaved 
plays a significant role in the entrainment o f  dust; particles, which are suspended by 
vehicular movement on paved and unpaved roads. This is a major contributor to fugitive 
dust em issions.73 Most o f  the “sand w inning” as it is termed is done in unplanned and 
illegal manner w ith serious damage to the environment and the surrounding air quality. 
There is also a lot o f  construction sand and gravel processing plants whose processes
12
involve gravel transportation, crushing, screening, storage and hauling. The operations o f 
these plants also generate high concentration o f  dust into the atmosphere.
The harmattan, a dry desert w ind from the Sahara, blows from the northeast from 
December to March, lowering the humidity and creating hot days and cool nights in the 
north. In the southern part o f  the country, the effects o f  the harmattan are felt in January 
and February. It has been shown by Baumbach et al53 that in Lagos-N igeria the particle 
concentration more than doubled during the dry season when the northly harmattan wind 
from the Sahara are prevalent.
1.2.1.6 SEA SPRAY
Ghana has 537-kilometer (334-m i.) coastline which is mostly a low, sandy shore backed 
by plains and scrub and intersected by several rivers and streams. Common salt is 
produced in commercial quantities along the coast o f  Ghana near A ccra (at Ada). H a lf o f  
the country lies less than 152 meters (500 ft.) above sea level, and the highest point is 883 
meters (2,900 ft.). The general w ind direction is North-East in the Southern part o f  the 
country and depth o f  penetration o f  the coastal w ind is high, sometimes covering more 
than two-thirds o f  the country.
The marine aerosol comprises two distinct aerosol types:
The first, being the primary sea-salt aerosol produced by mechanical disruption o f  the 
ocean surface. The sea-salt aerosol is produced at the ocean surface by the bursting o f  air 
bubbles resulting from entrapment o f  air induced by wind stress that are subsequently 
dechlorinated by acidic materials, when the Cl' is replaced by the S 0 42' or N 0 3'. This 
process is especially evident in the coastal atmosphere and has been regarded as the major 
global source for gaseous chlorine in the atmosphere.74,75 This dechlorination process is
13
highly corrosive and has had a major impact on the materials used in Ghana for 
infrastructure development. Metallic materials such as vehicles, electronic appliances, 
and telephone and streetlight poles have not been spared by this corrosive nature o f  the 
sea spray.
The second is the formation o f  secondary aerosol, primary in the form o f  non-sea salt 
sulphate and organic species formed by gas-to-particle conversion processes.
1.2.2. ISSUES
1.2.2.1 URBANISATION
The w orld 's urban population is growing very rapidly and by the United Nations 
estimations the total population growth between 2000 and 2030 w ill take place in the 
cities o f  the less developed countries.76 In Ghana, from the 2000 Housing and Population 
Census more than 20%  (about 4 million people) o f  the countries population live in the 
Accra-Tema Metropolitan area and its environs.
The urban areas are the commercial hub o f the country, more and more young people are 
drawn to the cities looking for new  jobs and new opportunities. The physical properties o f 
the urban areas in Ghana, like in the rest o f  the world, are divided between highly 
developed affluent areas and less developed slump areas.77,78 The city lim its/boundary are 
continuously being moved as a result o f  the creation o f  satellite settlements, w ithout the 
most basic o f  infrastructure, that are being absorbed. The economic crises o f  the African 
region have also fueled the migration o f  people to the urban areas since the 1970s. This 
demographic change has confronted city authorities with the task o f  providing the urban 
population with the necessary infrastructure such as housing, portable water, efficient 
solid and liquid waste disposal systems, etc.79
14
The increasing urbanization o f  developing countries, like Ghana, has in it attendant 
problems that lead to poor health and the degeneration o f  the urban environment.80 
Construction boom and other drastic changes to both land use and the atmosphere above 
compared to the surrounding rural areas are brought about by urbanisation.81 This 
inadvertently modifies both the climate and air quality as the urban area grows. Cities by 
nature are concentrations o f  human activity; hence they are subjected to the highest levels 
o f  pollution and pollution impact.82
Though most developed countries have long placed the improvement o f  air quality on 
their development agenda, the opposite is the case in sub-Saharan A frica where most city 
authorities are unable to keep pace w ith development and urbanization, and therefore lack 
pollution control measures. The lack o f  capacity, both capital and human, in sub-Saharan 
Africa together w ith expected population growth means that the situation can only get 
worse.83
1.2.2.2 CLIMATE
The climate o f  a place influences the type o f  activities that take place and hence have an 
impact on the levels o f  particulate pollution. The climate o f  Ghana is tropical and humid 
(relative humidity 55% - 85%), but temperatures vary w ith season and elevation. There 
are two main raining seasons in Ghana (April to July and from September to November) 
except in the north o f  the country where there is only one (starts April and lasts until 
September). Annual rainfall ranges from about 1,100 mm in the north to about 2,500 mm 
in the southwest.
During the harmattan period, there is an increase in particulate matter levels which are 
closely linked to the increased atmospheric stability that usually coincide w ith the dry
15
season and the absence o f  precipitation and effective ventilation .84'86 This is also enhanced 
by the absence o f  leaves and vegetation which acts as filters during the rainy seasons, 
coupled with the miniinal w ashout effect o f  rains during this period. These seasonal 
variations have been shown to be more visible in tropical climate than temperate 
regions.87
In most areas the highest temperatures occur in March (average high about 32 ”C) and the 
lowest in August (average low about 21 ”C). The variation in both annually and daily 
temperature are quite small. It is comparatively dry along southeast coast, hot and humid 
in southwest, and hot and dry in the north.
1.2.2.3 INCREASED PUBLIC ENVIRONMENTAL AWARENESS
Environmental awareness is on the increase in the country and leading to frequent social 
agitations on environmental issues. In 1999 the Accra M etropolitan Authority with the 
assistance o f  the British Government started the construction o f  an ultra modern landfill 
site for domestic and industrial waste at Kwabenya, near Accra after a thorough 
environmental impact assessment. This landfill project has been brought to a standstill 
because o f  environmental pollution concerns raised by residents around the area. 
Recently, residents o f  Juapong and Aflao in the Volta Region took to the streets in 
demonstrations against the Juapong Textile Company and the W est African Cement 
Company respectively for polluting their air environment. The people living in the 
M ining areas have had series o f  demonstrations against surface m ining as a result o f  the 
"perceived” environmental (including air) pollution. On June 15, 2005 a demonstration 
by the Chiefs and people o f  Prestea, a mining town in the Western Region, against 
surface m ining in the area, resulted in the Police firing live bullets into the crowd injuring 
eight seriously including an eleven year old boy who had his lip blown off.88 These
16
agitations do not augur well for G hana’s vision o f  accelerated industrial growth and a 
prime destination for foreign investments. The best way these agitations can be stemmed 
is to do systematic and continuous air quality monitoring and generate the necessary 
scientific data to allay the concerns o f  the public.
1.2.3 TRENDS IN MONITORING
Not much air pollution studies have been done in Ghana in particular and A frica in 
general.51,52 The follow ing modest activities have however been carried out in the country 
in the field o f  air pollution:
• Air Quality M anagement in Takoradi Thermal Power Station
• Biomonitoring o f  A ir Pollution using lichens
• Air pollution monitoring o f  Kpone and Tema Oil Refinery
In addition to these projects the Environmental Protection Agency (EPA) has a number o f 
mobile stations, which they use to measure NOx, SOx and TSP but not on regular and 
sustained basis. The Fig. 1-1 shows PMio concentration in roadside ambient air monitored 
by the EPA.
1.2.3.1 AIR  QUALITY  MONITORING AND MANAGEMENT AT THE  
TAKORADI THERMAL POWER STATION
The Volta River Authority (VRA) is in charge o f  the generation, transm ission, 
distribution o f power for industrial and domestic use in Ghana and selling o f  electricity to 
Togo and Benin. Until 1997 power generation was 95% hydro from the Akosombo dam 
built in 1961. Increased demand coupled with the decreasing water level in the lake has 
compelled VRA to produce only 45% o f Ghana's power requirement from hydro and 55% 
from the Takoradi Thermal Plant.
17
The thermal plant is 330 MW combined cycle plant, consisting o f  two 100 MW 
combustion turbine generator with associated heat recovery steam  generators. Currently, 
the plant is using light crude oil. The combustion o f  this fuel results in the emission o f 
various air pollutants, such as:
•  Oxides o f  N itrogen (NOx)
• Sulphur D ioxide (S 0 2)
•  Carbon D ioxide (C 0 2)
• Carbon Monoxide (CO)
• Particulate M atter (PMio and PM2 5)
CO:. CO, and the particulate matter are the direct products o f  the combustion whereas 
NOx, and S 0 2 are dependent on the nitrogen and sulphur contents o f  the oil. To m itigate 
the harmful effect o f  these atmospheric emissions the plant design incorporated a 
modelled stack height that improves the dilution and dispersion o f  the flue gas. This help 
to distribute the impact o f  the emitted pollutant over a w ide ground surface area and 
reduces the average ground level concentration and the impact on vegetation.
The concentration o f  fuel bound nitrogen content for the light crude oil imported is 
limited to 120mg/l to m inim ize NOx emission. S 0 2 is controlled by lim iting the fuel total 
sulphur content to 0.2%  by weight to ensure that S 0 2 em issions fall w ithin acceptable 
guidelines set by Ghana EPA.
Ambient air is monitored at three locations:
• Aboadze, the nearest community in the prevalent w ind direction
• Beposo, the point o f  maximum out fall o f  airborne plant emissions
• The Western fence o f  the plant as control or background monitoring station 
Results o f  some o f  the monitoring work done is given in Figures 1-2 and 1-3.
18
1.2.3.2 BIOMONITORING OF AIR  POLLUTION USING LICHENS
Two species o f  Lichens (Paramedia  and Lecanora ) in the Crutose  fam ily were identified 
and used in the arseno-pyrites gold m ining area o f  Ghana (Obuasi and Prestea) and the 
Tema/Doryim  area, which is the hub o f  Ghana's industries to m onitor the deposition o f  
heavy metals from  air pollution. Akim Oda and Sameraboi, two rural farm ing 
communities where the same lichens were identified were also used as the control. The 
results o f  this work, as shown in Table 1-1, showed that the m ining and the industrial 
areas had higher elemental concentration values than the farm ing communities suggesting 
clearly that these two places were polluted. S9'90
1.2.3.3 A IR  POLLUTION  MONITORING OF KPONE AND TEMA OIL  
REFINERY
The Tema Oil Refinery in 2003, as a result o f  persistent accusations by the citizens o f 
Kpone, a fishing community east o f  the refinery and the Tema industrial area, requested 
the Ghana Atomic Energy Commission to undertake air monitoring at both Kpone and its 
plants. Fig. 1-4 shows the spectra o f  selected aerosol filters from Kpone and the Tema Oil 
Refinery. The result is shown in Table 1-2. The elemental concentrations o f  most 
elements in the PM ]0 samples were consistently higher for Kpone than at the Tema Oil 
Refinery. However, the values for PM2.5 showed consistently lower values for Kpone than 
the Refinery. Since most o f  the elements identified were not precursors from fuel 
combustion it was concluded that the operations o f  the refinery were not responsible for 
the air pollution complains o f  Kpone, but other sources.
19
FIGURE 1-1: PMio CONCENTRATION IN AMBIENT AIR AT SOME ROADSIDES IN ACCRA
PA R T IC U L A T E  M A T T E R  (PM  10)  C O N C E N T R A T IO N  IN R O A D S ID E  A M B IE N T  A IR  ( 2 0 0 2 -0 3 )
□  PM  10 (ug /rr.3 ) W ERA  am b ien t (ug /rn3 )
3 5 0 .0 0
3 0 0 .0 0
2 5 0 .0 0
200.00 
1 5 0 .0 0  
100.00
50 .0 0  
0.00
20
0.05 
0.045 
0.04 
0.035 
0.03 
0.025 
0.02 
0.015 
0.01 
0.005 
0
1996 1997 1998 1999 2000
FIGURE 1-2 : AMBIENT NOx CONCENTRATION(ppm)
□ Beposo 
■ Aboadze
□ EPA max limit
21
FIGURE 1-3: AMBIENT S 0 2 CONCENTRATION (PPM)
■  Beposo
■  Aboadze 
□  EPA max lim it
22
TABLE 1-1: MEAN CONCENTRATIONS OF ELEMENTS IN THE TWO
LICHEN SPECIES
Mean Value (ng/g)
Element Obuasi Prestea Tema/Doryim Samreboi Akim  Tafo
A1 17260 7510 8010 1350 2070
As 110 73.2 0.8 1.84 0.8
Ba 650 110 156 204 190
Br 8.2 5.7 10.4 5.6 5.2
Ca 6860 14380 17190 26090 21640
Ce 9.6 3.2 10.4 0.5 2.6
Co 4.6 3.5 4.4 3.9 5.0
Cr 68.9 19.7 28.6 6.3 4.3
Cl 390 461 592 390 303
Dy 1.2 0.8 0.8 0.7 -
Fe 18730 4410 7530 1340 1150
Ga 66.6 45.0 - - -
H f 2.0 0.8 2.4 0.3 1.1
Hg 1.5 6.1 - 0.9 2.2
I 305 13.2 19.0 4.5 8.6
K 3926 4750 3840 2800 3740
La 5.5 2.3 5.8 2.0 1.5
Mg 5750 6000 4780 5230 6025
Mn 106 198 275 440 635
Na 1960 1248 765 360 170
Rb 15.1 13.7 13.7 8.1 11.3
Sb 4.5 1.9 0.7 0.1 0.4
Sc 3.4 1.5 1.8 0.8 0.4
Sm 1.1 0.4 1.1 0.1 0.2
Th 1.6 0.6 1.2 0.2 0.4
Ti 2270 715 1187 235 222
U 0.8 0.2 - - -
V 42.3 12.6 18.5 1.9 1.9
W 3.9 1.9 - - -
Zn 157 118 115 112 62
23
Co
un
ts 
Co
un
ts
FIGURE 1-4: PIXE SPECTRA OF SOME SELECTED 
AEROSOL-LOADED FILTER SAMPLE
E ( K e V )
P IX E  S P E C T R A  OF  A IR  F IL T E R  S A M P L E  A T  S IT E  T O R  SS22
J  i i_
PM10  
PM2.5
E(keV)
24
TABLE 1-2: AVERAGE CONCENTRATION OF  
ELEMENTS IN AIR PARTICULATE MATTER AT KPONE  
AND TEMA OIL REFINERY IN GHANA
KPONE Tema Oil Refinery
PM10 PM2.5 PM10 PM2.5
Element (ng/cm2) (ng/cm2)
Al 1541 464 434' 2171
Si 2913 512 977 5669
P 111 351 88 107
S 416 466 313 470
Cl 2482 231 1384 3242
K 323 171 488 541
Ca 974 54 351 1424
Ti 121 - 51 168
Cr 6 5 8 -
Mn 24 6 14 28
Fe 1919 58 696 2239
Ni - - 8 -
Cu 205 2 46 -
Zn 84 47 124 16
Br - 18 - -
1.2.4 CHALLENGES
A great deal o f  research and studies have been carried out on soil, water and 
vegetation pollution studies in Ghana. Air pollution however has had little attention 
and is only now being looked at as a result o f  the shift from hydro power generation 
(95% in 1997 to 45%  in 2003). As the country begins to industrialize on a large scale, 
it is in a unique position to implement regulations and thus avoid many o f  the 
problems that have occurred in the northern hemisphere.
Current activities are mostly based on gravimetric analysis for TSP, very little o f 
elemental characterization, and virtually no post data evaluation (source identification 
or source apportionment). Very little data available on background levels due
25
particularly to lack o f  access to mains supply for powering the samplers in rural 
“unpolluted” areas. Use o f  petrol-driven generators introduces additional on-site 
pollution that is sometimes picked up when wind direction changes. This has also 
contributed to data generated so far not put to optimum use.
The current national capacity to monitor and enforce environmental standards, 
particularly in air quality standard, is rather weak. This is due essentially to 
inadequate number o f  well-trained personnel in atmospheric sciences, and lack o f  
appropriate equipment for monitoring and assessment.
It will be more appropriate for proper studies to be carried out at this time when 
Ghana has sw itched from  leaded petrol to unleaded petrol (from  January 2004). 
Currently, non-vehicular air pollution levels are low in most parts o f  the country, 
except in the m ining areas, and much can be done to control the situation. W ith the 
expected growth o f  industries and human population in the future, the conditions w ill 
be so critical that a substantial portion o f  the national budget will be required to fight 
and control air pollution effects.
1.3 OBJECTIVE OF THE THESIS
Despite the measures taken to address these serious environmental challenges, there 
exist gaps regarding adequate baseline data upon which impact can be assessed. These 
gaps which are very pronounced due to lack o f  scientific data in air quality in the 
country can only be filled by field measurements, laboratory analysis and air pollution 
modelling. Secondly, the datu generated would also assist the industries and the 
regulatory authorities to take the necessary action, to repair or forestall any pollution 
threat by their operations.
26
1.3.1 GENERAL OBJECTIVE
The overall objective o f  the thesis is to carry out research and monitor the relevant 
variables used to describe air quality, w ith a view  to establishing a baseline data upon 
which impact may occur. It also seeks to establish the origin and characteristics o f  
size-segregated aerosols for PM2.s (fine) and PM 10-2.5 (coarse) over Kwabenya, near 
Accra w ith special emphasis placed on the influence o f  major sources like traffic 
emissions, bush fires and biomass burning, soil, sea spray and major industrial 
emissions. This w ill be assessed by a combination o f  field measurements, 
chem ical/physical analysis using mainly nuclear analytical techniques, and receptor 
modelling.
1.3.2 SPECIFIC  OBJECTIVES
The Specific objectives are:
• To establish mass concentrations o f  PM2.5, PM IH.5 and PM ,0 over K swabenya, near 
Accra.
• To analyse heavy metals and other elements in the particulate matter using 
EDXRF and other methods for chemical characterisation o f  the aerosol, e.g. black 
carbon.
•  To establish source signatures and relate variations in particle concentrations to 
the influence o f  strong sources
• To determ ine anthropogenic (man made) and natural contributions to air quality 
over Kwabenya
• To relate particle concentrations to available climatic parameters (temperature, 
wind direction and speed, ham idity/rainfall and air pressure)
• To model atmospheric pollution (receptor model) using data generated.
27
CHAPTER 2 
AEROSOL PARTICLES AND SAMPLING INSTRUMENTS
2.1. PARTICLE FORMATION
One o f  the most important physical parameters in the analysis o f  aerosol particles is 
the particle size. The others are particle shape and density. Many aerosol particle 
properties depend on the size, since it influences the transportation, deposition and 
migration o f  the aerosol through the environment. W hereas coarse particles result 
mainly from mechanical processes, fine particles are formed mainly by chemical 
reactions and by the coagulation o f  even smaller species including molecules in the 
vapour state.
While small liquid particles and some solid particles formed by condensation are 
almost always spherical, most solid aerosol particles usually have complex shapes.1 
For example, soil particles may be flaky and metallurgical fume may be aggregate 
chains formed from  condensed droplets. They have varying densities, hence for 
comparison to be made between air particulates in aerosol studies the concept o f  an 
“equivalent d iam eter” is used. This is the diameter o f  the sphere that would have the 
same value o f  a particular property as that o f  an irregular shaped particle. The most 
commonly used equivalent diameter is the aerodynamic diameter, defined as the 
diameter o f  a sphere o f  unit density ( lg /cm 3) which has the same terminal settling 
velocity in air as the particle under consideration.1,91
For this work, unless otherw ise stated, the diameter refers to the aerodynamic 
diameter. A irborne particles cover a wide size range (Fig. 2-1). Hence atmospheric 
aerosol particles size distribution varies from place to place depending on the sources 
emitting them and the meteorological parameters prevailing at that place. However,
28
the most important aerosol particles which have serious health, climatic and 
environmental effects are those smaller than 10|im in d iam eter.'1,92
Aerosol particles once in the atmosphere can have their size, number and chemical 
composition changed by several mechanisms until they are ultimately removed from 
the atmosphere. The atmospheric aerosol size distribution is normally multi-modal in 
shape reflecting the d ifferent formation mechanisms o f  primary and secondary 
(formed through gas-to-particle conversion processes) aerosol and have a w ide size 
range.94 There are three different types o f  particle size distributions, namely - number, 
diameter and mass. Normally, fresh particle size distribution shows at least three 
groups o f  particles. These are nucleation mode, accumulation mode and coarse mode 
(Fig. 2-2). Their respective mass median aerodynam ic diameter (MMAD) size 
distributions are: <0.1 |im , 0.1-2 pm and >2 |rm. In general, the nucleation and 
accumulation mode particles constitute the fine particles and there is comparatively 
little mass exchange between the fine and coarse particle modes, hence they exist 
together in the atmosphere as two chemically distinct aerosols. They have different 
chemical compositions, sources and lifetime in the atmosphere.
2.1.1 NUCLEATION MODE PARTICLES
Nuclei mode (also known as A itken mode) particles, are particles w ith a mass median 
aerodynamic diameter less than 0.1 m icrometer (MMAD <0.1 |im). These are formed 
by the condensation o f  hot vapours from combustion processes and by nucleation 
processes in the atm osphere .1096 Studies o f  atmospheric Aitken particles suggest that 
carbon (C ) and sulphur (S) are the main components, w ith a mass ratio o f  C/S o f 
approximately 3:1. This links the formation o f  this mode o f  particles mainly to the 
cycles o f  S and hydrocarbons.97 When a vapour becomes increasingly supersaturated.
29
molecules accumulate into clusters (embryo); formed due to the attractive Van der 
W aals’ forces w ithout the assistance o f  condensation nuclei or ions.98 This type o f 
nucleation is known as homogeneous nucleation or self-nucleation.
The formation o f  new  particles via homogeneous nucleation, though rare, is 
commonly believed to happen in relatively clean environment, where the 
condensation o f  nucleating species onto pre-existing aerosol particles is highly 
inefficient.99 But this is not a limiting factor, since significant production o f  H 2SO j- 
H20  particles is also possible in more polluted air, but it requires both cool and humid 
conditions combined w ith relatively high concentration o f  sulfuric acid vapour. When 
a vapour becomes increasingly supersaturated, leading to continuous formation o f 
molecular clusters, these clusters are unstable and continuously disintegrate. In super 
saturated vapour, the cluster concentration increases to the point where they collide 
with one another frequently in a process sim ilar to coagulation except that here the 
"agglomerates" disintegrate immediately they are formed.
In photochem ical smog, certain gas phase reactions are promoted by ultraviolet light 
and form low -vapour-pressure reaction products. These products exist at high super­
saturation, because o f  their low vapour pressure, and can form  particles by 
homogeneous nucleation. Th*. homogeneous nucleation rate increases w ith the 
following:
i. decreasing ambient temperature
ii. higher relative humidity
iii. decreasing surface tension, and
iv. lower equilibrium  vapour pressure o f  the nucleating species (or gases).
30
smog clouds and fog mist drizzle
   X  x  X ------------
metallurgical dusts and fumes____________ _______________________
combustion nuclei nebulizer drops hydraulic nozzle drops
<1— —  ' >  ■<  ' ■ <  - ——  ' ■  ............
collodial silica insecticide dusts<  >- 11 ■">
oil smoke fly ash<  ■ • ■ ................................. x ------- -------------- ---------------------------------- >
tobacco smoke coal dust< ------------------------------------------------------x ----------------------------------------------------- >
^ zinc oxide fume _ cement dust
paint pigments pollens
'  ——> ■ .......
^ carbon black ^  ^ ________ pulverized coal
spray dried milk__________<r-
milled flour
viruses ^  ^ __________bacteria
°*001 °-01 0.1 — 1 , . 10 100 1000Particle diameter (jjm)
FIGURE 2-1: SIZE RANGE OF AEROSOL PARTICLES (FROM COLBECK2)
31
FIGURE 2-2: IDEALISED TRIMODAL SIZE DISTRIBUTION  OF
FRESH PARTICLES, SHOWING GENERAL RELATIONSHIPS  
BETWEEN THE THREE COMMON SIZE RANGES AND THEIR  
FORMATION MECHANISMS IN THE ATMOSPHERE. (FROM  
COLLS95 W ITH SLIGHT MODIFICATIONS)
In contrast, heterogeneous nucleation occurs in areas where other nuclei, particles or
ions from pre-existing aerosol, are present. Here the process o f  particle formation
(nucleation) is complemented by heterogeneous condensation on other nuclei
(particles or ions) present in the condensable vapour system. The best known example
o f  this is the formation o f  cloud droplets by water condensation onto hygroscopic
aerosol particles (C loud Condensation Nuclei (CCN)). In the atmosphere the
following are the major components o f fine particles: sulphates, nitrates and
carbonaceous compounds, indicating that the fine particle material is mainly derived
from atmospheric chem ical reactions. The precursors which aid the formation o f  these
components o f  the fine particles are S 0 2 and other S-containing gases, NO* and other
N -containing gases and organic vapours. The fine particles formed by gas-to-particle
32
conversion (GPC) are known as secondary aerosols.8,100 In the presence o f  solar 
radiation GPC is highly enhanced. A lthough some directly em itted particles are found 
in the fine particle mode, secondary particles formed from gases dominate the fine 
particle mass.
2.1.2. ACCUMULATION MODE PARTICLES
Accumulation mode particles, w ith an MMAD in the range 0.1 — 1 |im , are formed 
from coagulation o f  nuclei mode particles, because o f  their small size and high 
concentration especially near the source. Hence nuclei mode particles have relatively 
short lifetime. As the name suggests, particles accumulate in this mode because the 
removal mechanisms for the nuclei mode particles are weak. The particles in the 
accumulation mode also coagulate too slowly to reach the coarse-particle mode where 
they can be removed by sedimentation. They can however be removed by rainout or 
washout. It has been found that carbonaceous particles exist mainly in the 
accumulation mode.101 Accumulation mode particles typically (though not generally 
true in all cases) have longer atmospheric life time (i.e. days to weeks) than coarse 
mode particles, and tend to be rather uniform ly dispersed across an urban area or large 
geographical regions. The size range o f  the accumulation mode includes the 
wavelengths o f  visible light, and these particles account for most o f  the visibility 
effects o f atmospheric aerosol.
Fog and cloud droplets play an important role in the accumulation mode formation 
processes and under conditions o f  high humidity, there exist two sub-modes: a 
condensation mode w ith MMAD o f 0.2 -  0.3|iin and a droplet mode w ith MMAD o f 
0.5 -  0.8|im . The droplets are formed by the growth o f  hygroscopic condensation- 
mode particles. This process may be facilitated by chemical reactions in the droplet.
33
2.1.3 COARSE MODE PARTICLES
Coarse mode aerosol particles originate chiefly from the disintegration o f  larger 
pieces o f  matter. M ineral pollutants constitute one source o f  coarse particulates in air 
and many o f  the large partic 'es in atmospheric dust, particularly in rural areas, 
originate as soil or rock and consequently their elemental concentrations o f  Al, Ca, Si 
and O are sim ilar to the earth crust. Sources o f  large coarse particles include natural 
ones such as sand storms, large salt particles from sea spray and human activities such 
as from agriculture, surface m ining, stone grinding and crushing in quarries. Human 
activities result in particles o f  the topsoil and rocks being picked up by the w ind. The 
particles readily settle out or impact on surfaces, because o f  their large size. Hence 
their lifetime in the atmosphere is only a few hours or days depending on the 
prevailing meteorological conditions.
2.2 PARTICLE TRANSPORT
In between em ission and removal, aerosol particles are transported in the atmosphere. 
The source-transport-sink sequence determines the residence (life) time o f  the particle 
in the atmosphere. This motion o f  particles in air is dependent on the aerodynamic 
diameter o f the particles.102
2.2.1 RESIDENCE TIME
Particles at the sm aller end o f  the size range undergo a continuous change in size and 
composition and are constantly loosing their entity. Nucleation mode particles have 
residence time less than one day because they undergo rapid coagulation. Particles 
larger than 1 Ojrin diameter are efficiently scavenged by sedimentation and washout, 
lntermediate-size particles arc mainly incorporated into cloud droplet. For this type o f 
particles a great difference between residence time and altitude is observed. In the 
lower troposphere the residence time is about a week and in the stratosphere the order
34
o f months to even years.58 Due to the absence o f  clouds in the stratosphere the 
removal by precipitation is not possible.
2.2.2 PARTICLE MOTION
In steady conditions the motion o f  aerosol particle is typically a result o f  the action o f 
two forces, a constant external force such as gravity and the resistance o f  the gas to 
particle motion. For example in still air, a particle undergoing gravitational settling 
can be shown to be settling w ith a velocity proportional to the square o f  the diameter. 
Hence gravitational settling is very important for large or coarse particles. Due to this 
rapid depletion by gravitational sedimentation, coarse particles (and particles larger 
than 10|iin) are mainly present in the atmosphere close to their sources.2
One parameter that is very useful for the analysis o f  complex particle motion is the 
particle relaxation time (x). The relaxation time is proportional to the square o f  the 
diameter and inversely proportional to the viscosity o f  the gaseous m edium .1,101 It 
characterizes the time required for a particle to adjust or “relax" its velocity to new 
conditions in the atmosphere. The relaxation time is analogous to the characteristic 
acceleration for a car, such as the “0 to 100” (km/h) time. It depends on the mass and 
mobility o f  the particle and is not affected by the nature and magnitude o f  the external 
forces acting on it. Hence large particles have long (large) relaxation times.
A particle injected into a moving gas with an initial velocity Uxo relative to the gas, in 
the absence o f  external force, would be progressively pulled by resistance force 
exerted by the gas medium until the particle has the same velocity as the medium. The 
distance traveled by the particle relative to the gas medium before it catches up with
the flow is known as the particle stopping distance, S or inertial range. The stopping
35
distance, Sx, is related to the relaxation time,x, and the particle initial velocity relative 
to the gas by
S x = V x O:  21
On a displacement scale, the stopping distance represents a measure o f  a particle’s 
effective initial momentum , which is dim inished to zero by air friction over a distance 
equal to the stopping distance.
Relaxation time t = mB, where m  is the particle mass and B is the particle mobility, 
hence the stopping distance can be expressed in terms o f  particle mobility and its 
initial momentum by
sx= Bmvx0 2 2
The stopping distance represents the distance a particle w ill travel in still air if  an 
external force acting on the particle were turned off. It is very important when 
considering how particles behave within moving air that is changing direction, e.g. 
when a particle moves in an airstream  that is abruptly turned 90*. Eqn. 2.1 shows that, 
for the same initial velocity, a larger particle will have a larger stopping distance. The 
velocity o f  the air and the size o f  the particle influences the stopping distance in 
changing w ind direction, and would determine whether the particle remains airborne 
(follow the airstream) or not (exit the airstream trajectory). As illustrated in Fig. 2-3, 
larger particles and particles w ith higher air velocity are not able to follow  the 
movement o f  airstream  as it changes direction. Hence they are removed from the 
airstream.
36
FIGURE 2-3: DIAGRAM  ILLUSTRATING THE CONCEPT OF  
PARTICLE INERTIA  (FROM MARK102 W ITH SLIGHT  
MODIFICATIONS)
2.3. PARTICLE DEPOSITION  MECHANISMS
Atmospheric aerosols are complex w ith respect to particle size, shape, chemical 
composition, concentration, etc. Once the aerosol particles are emitted into the 
atmosphere, their size distribution is modified by physical and chem ical mechanisms. 
Dry and wet removal are such physical mechanisms. Coagulation, caused by 
Brownian motion o f  particles, or by external forces acting between particles, is 
another. Condensation and evaporation processes change both the size and the 
composition o f  the particles depending on the local temperature and relative humidity. 
Heterogeneous chemical reactions on and inside aerosol particles result primarily in 
changes o f  the chem ical composition and may lead to change in particle size. 
Coagulation, condensation/evaporation and heterogeneous reactions are not 
considered as deposition mechanisms since they mainly remove particles from  certain 
size ranges to others. This could however aid the deposition mechanisms as the 
particle size grows (as the partible size becomes larger).
37
Aerosol particles emitted into the atmosphere are ultim ately removed from the 
atmosphere by deposition onto some surfaces. Surfaces like buildings, water, human 
body and respiratory system, soil and vegetation are some examples. A mass flux 
from the atmosphere to the earth 's  surface occurs by two processes: dry deposition 
and wet deposition. These processes transfer pollutants to the ecosystem  and are 
responsible for the many negative environmental effects.
2.3.1. DRY DEPOSITION
Dry deposition is defined as the continuous transport o f  particulate and gaseous matter 
from the atmosphere directly onto surfaces, not associated w ith precipitation. The 
mechanisms involved are:
1. G ravitational settling (o f predom inantly coarse particles) -  this is the process 
by which the particles are transported to a surface by the force o f  gravity, all 
particles are subjected 10 this process but in varying degree due to their mass.
2. Turbulent exchange o f  (predominantly) fine particles and gases to the surface.
3. D iffusion -  particles are deposited on a surface after being brought in contact 
with the surface by Brownian diffusion. Since Brownian diffusion increases 
w ith decreasing size, deposition by diffusion is mainly important for gases and 
sub-m icron particles.103
4. Interception -  this takes place when a particle, follow ing the stream lines o f  a 
flow  around an obstacle, is o f  a size sufficiently large such that its surface and 
that o f  the obstacle come into contact (geometricical effect o f  large particles).
5. Inertial Impaction -  this takes place when a particle because o f  its inertia, is 
unable to adjust to the abruptly changing stream lines near an object.
38
6. Biological mechanism s -  biological cell/organism  can attract and absorb onto 
their surfaces some particulates and gases. For example, gases can be 
transferred to surfaces o f  vegetation through stomata and lea f cuticles.
Though all the mechanism  listed above play an important role in dry deposition, the 
most important mechanism s are sedimentation, impaction, interception and diffusion. 
Diffusion is an important removal mechanism  for nucleation mode particles while 
inertial impaction, interception, and gravitational settling are important removal 
mechanisms for coarse mode particles. These removal mechanism s are least efficient 
for accumulation mode particles, thus the particles are caused to accumulate in this 
mode, hence the name.10'4
The factors that govern dry deposition are the airborne concentrations, the 
atmospheric movements and the ability o f  the underlying surface to absorb and/or 
capture the species that come in contact with it. For trace gases the uptake at the 
surface is dependent on the reactivity and solubility o f  the species and the type o f 
surface. Generally, the nature o f  the surface does not have a strong influence on the 
dry deposition flux o f  coarse particles but is more complex for fine particles.
Dry deposition has a strong diurnal cycle, because the mechanisms involved are 
meteorological and sometimes biological. The dry deposition o f  fine particles (and 
gases) is more significant during the daytime because it is mainly controlled by 
turbulent exchange (with the surface) which is more efficient during the day.
2.3.2 WET DEPOSITION
Wet deposition is the process by which particles and gases are removed from the 
atmosphere and are subsequently delivered to the earth’s surface during precipitation
39
events. It involves uptake into the cloud droplets (rainout) and uptake into 
precipitation elements (washout). It is important to note that only rainout leading to 
subsequent precipitation is regarded as wet deposition. Most o f  the uptake into clouds 
do not produce wet deposition, since most o f  the cloud never precipitate. Wet 
deposition is not receptor surface dependent.
There are many mechanism s that contribute to wet deposition, among these are the 
capture o f  small particles by cloud droplets (coalescence) and w ater condensation 
onto particles (nucleation scavenging). I f  the particles contain water soluble salts, 
their ability to form droplet is enhanced. In the washout process, only a fraction o f  the 
particles in the path o f  a falling rain drop will collide w ith it. The remainder, and 
particularly the fine particles, may be swept in the stream line around the rain drop. 
For this reason large particles are known to be more efficiently removed than small 
particles.
2.4. AEROSOL SAMPLING  INSTRUMENTS
2.4.1. GENERAL CONSIDERATIONS
The choice o f  an aerosol sampling instrument depends on the substance to be 
measured, the properties it haj, and the information to be gained from the measured 
values. Measurements can be carried out where the particles are formed to determine 
emissions or at receptor location to determine air quality (Fig. 2-4). The primary aim 
o f measurements is to determ ine concentrations o f  air pollutants but the mass flow 
rates are also very important to determine the dilution effect o f  the emissions.
The main aim o f  all aerosol sampling is to collect representative samples for analysis. 
These samples must accurately reflect the airborne particles in both concentration and 
size distribution. Isokinetic sampling is a procedure to ensure that a representative
40
sample o f  aerosol enters the inlet o f  a sampling tube when sampling from a moving 
aerosol stream. Sampling is said to be isokenetic when the inlet axis o f  the sampler, is 
aligned parallel to the gas stream lines and the gas velocity (U) entering the probe head 
is equal to the free-stream  velocity (U0)approaching the inlet (Fig. 2-5). In this type o f  
scenario there is no particle loss at the inlet, regardless o f  particle size or inertia. But 
there is no guarantee that there will not be particle loss between inlet and the collector 
(e.g. filter), it guarantees only the concentration and size distribution o f  the aerosol 
entering the tube is the same as that in the flowing stream.
FIGURE 2-4: SCHEMATIC  DIAGRAM  OUTLINING EMISSION, 
TRANSPORT, TRANSFORMATION  AND SAMPLING OF AIRBORNE  
POLLUTANTS
Sam p ling  p robe
A -
Gas stream Im os
U  -  Ur,
FIGURE 2-5 -  ISOKINETIC SAMPLING (FROM H INDS1)
41
Anisokinetic sampling is a failure to sample isokinetically, and may result in the 
distortion o f  particle size distribution and a biased estimate o f  the concentration. 
These effects arise because o f  particle inertia in the region o f  curved stream lines near 
the inlet. The sample may contain excess or deficiency o f  large particles depending on 
the conditions.
Fig. 2-6 below shows the three possible scenarios o f  anisokinetic sampling. In Fig. 2- 
6(a) the probe is not aligned to the gas flow stream lines; here the concentration will 
be underestimated because most o f  the heavy particles would be lost because o f  
inertia. In Fig. 2-6(b) the velocity in the probe exceeds the stream  velocity, this is 
known as superisokinetic sampling, here some particles w ith high inertia in the 
volume sampled cannot follow  the converging stream lines to enter the probe and are 
lost from the sample. Thus the concentration would be underestimated. In Fig 2-6(c), 
the stream  velocity exceeds the velocity in the probe, subisokinetic sampling. Here 
some heavy particles which were not initially in the flow  w ill enter because o f  the 
slow  velocity in the probe hence the concentration would be over-estimated.
In aerosol sampling technology it is very important to distinguish between integrated 
(discontinuous) and continuous measuring methods. The continuous sampling 
instruments measure the temporal profile o f  the measured quantity. Integrated or 
discontinuous sampling instruments provide the measured quantities as a mean value 
over a sampling period. It is advisable therefore that in the use o f  integrated sampling 
instruments to have some knowledge o f  the profile o f  the air pollutants to be 
measured. This would lead to setting o f  sensible sampling periods.
42
------------------ ..................—
-----
1 .
....
t<r>
FIGURE 2-6 -  ANISOKINETIC  SAMPLING (FROM  H INDS1),
(a) M ISALIGNMENT
(b) SUPERISOKINETIC  SAMPLING, U>U„,
(c) SUBISOKINETIC SAMPLING, U<U0
Aerosol sampling instruments often require the draw ing o f  ambient air through some 
type o f  filter material. This is achieved using a pump or a known volume o f  ambient 
air is pulled through a filter so as to collect the particles on the filter. The filter is then 
removed and taken to the laboratory to determine the gain in the filter mass due to 
particle collection. Ambient PM is calculated on the basis o f  the gain in filter mass, 
divided by the product o f  sampling period and sampling flowrate. U sing appropriate 
analytical techniques, its chem ical components can also be determ ined. The ambient 
air stream is processed on its way to the filter so that only particles o f  some 
predeterm ined aerodynam ic characteristics are allowed to reach the filter. This is one 
o f the most highly developed measurement principles that apply to the quantification 
o f  the wide variety o f  chem ical components in suspended particles.92,105 When the 
filter is impregnated with an absorbing solution, or when the filter material has
43
specific gas-absorbing properties, quantitative measures o f  gases as a well as particle 
phases are possible.3 The filter characteristics will govern the efficiency o f  particle 
collection. The chemical analysis to be done on the filter cannot be separated from the 
filter material used and the methods used to obtain the samples. D ifferent methods are 
available to fractionate the sample stream on the basis o f  particle size .92 They include 
direct impaction, virtual impaction and cyclonic flow.
2.4.2. INTEGRATED (DISCONTINUOUS) SAMPLING  INSTRUMENTS  
2.4.2.1. IMPACTORS
Impaction is one o f  the removal modes that is extensively applied in collection and 
measurement o f  aerosol particles. It has been thoroughly analysed, both 
experimentally and theoretically .1 It was first used in the early fifties for collecting 
dust for the evaluation o f  the occupational environment.
All inertial impactors operate on basically the same principle, i.e. an aerosol is passed 
through a nozzle and the output airstream directed against a flat plate (Fig. 2-7). The 
plate called an impaction plate deflects the flow  to form an abrupt 90 ' bend in the 
streamlines. Particles whose inertia exceeds a certain value are unable to follow  the 
streamlines and impact (collide) on the flat plate. The sm aller particles, w ith lower 
inertia, continue to follow  the streamlines and avoid the impaction plates. They 
therefore remain airborne and flow  out o f  the impactor. Hence the impactor separates 
aerosol particles into two size ranges; particles larger than a certain aerodynam ic size 
are removed from the air stream, and those smaller than that size remain airborne and 
pass through the impactor.
44
FIGURE 2-7: CROSS-SECTIONAL VIEW  OF AN  IMPACTOR  
(FROM  H INDS1)
Dichotomous Virtual Impactor
The dichotomous virtual impactor is based on the principle o f  inertial impaction. It is 
called “virtual” because the impaction plate is replaced by a collector probe so as to 
prevent the particles from bouncing o ff  and overloading o f  the collectors. Here 
aerosol particles below  a specified aerodynam ic diameter are drawn through a size- 
selective inlet and accelerated through an impactor nozzle (Fig. 2-8). After passing 
through the nozzle, the airflow  is divided into minor flow and major flow. The large 
particles have sufficient inertia that when the airstream  in which they are being 
transported is rapidly forced to change direction, the particles w ill separate from the 
airstream. The separated particles impact on a “virtual impaction surface”. They are 
then carried by the m inor flow onto a filter at the surface. This creates the divergence 
o f the flow. The sm aller particles move with the diverted stream. These are carried 
radially away from the je t axis by the major air flow. The separation is not caused by 
the interaction o f  the collected particles with the impaction surface. The separated 
particle streams are available for collection by standard filter media. In most cases, the 
cu t-off between the large and small particles is such that the particles are divided into 
coarse and fine particles.
45
FROM AEROSOL INLET  INHALABLE  
PARTICLES SMALLER  THAN 15 MICRONS
INLET TUBE
FINE PARTICLES ,
LE S S  THAN 2 .5  MICRONS •
COARSE PART ICLES ,
GREATER THAN 2 .5  MICRONS -
VIRTUAL
IMPACTOR
ACCELERATION
NOZZLE
- VIRTUAL 
IMPACTOR 
COLLECTION 
 .PROBE
AEROSOL
INLET
PLUNGER
I
0 .9  CMH 0.1 CMH
7.5 m TUBING TO CONTROL MODULE
FIGURE 2-8: DICHOTOMOUS PARTICLE SAMPLER FOR SEPARATING 
AIRBORNE PARTICULATE MATTER INTO TWO SIZE FRACTIONS
(FROM IAEA107)
The initial problems o f  the inlet size cu t-off being dependent on w ind speed106 and the 
failure o f  nozzle concentricity leading to loss o f  large particles (> 3(rm )107 have been 
resolved. Improvements have been made in the design o f  new  inlets to improve the 
separation and also address the w ind dependency o f  the previous design.108
Cascade Impactors
This is also based on the principle o f  impaction like the dichotomous virtual 
impactors. However, here the virtual impactors are replaced by a series o f  real 
impaction surfaces and there is only one air flow  stream through all the impaction 
surfaces making it look like operating a number o f  impactors in series. The concept o f 
the cascade impactor is shown in Fig. 2-9. The impactor is designed so that the 
airstream carrying the entrained aerosol particles are forced to change direction
46
rapidly. The particles w ith mass and/or velocity above a certain threshold would 
detach from  the stream line and impact on the collection surface.
FIGURE 2-9 -  A  SCHEMATIC  DIAGRAM  OF A  CASCADE IMPACTOR
The impactor plates are arranged in order o f  decreasing cu t-o ff size with the largest 
cut-off size impactor at the top and each separate impactor is referred to as an 
impactor stage. To achieve the different cut-offs w ith the same flow through all the 
stages, the nozzle size is decreased at each stage. By the principle o f  continuity o f 
flow, decreasing the nozzle size while the flow rate is constant results in an increase 
in the flow velocity through the nozzle. The increase in the flow  velocity results in a 
lower cu t-off diameter.
With the casacade impactor it is possible to obtain detailed information on the size 
distribution o f  the airborne particles and their constituents. There has been 
considerable use o f  cascade impactors that provide multiple samples in various size 
ranges with particles collected in each stage o f the im pactor.107 A lthough usually only
47
small amounts o f  mass are collected because o f  the large number o t stages, the 
samples are suitable for instrumental analysis such as NAA  and EDXRF. The 
samplers can be single or multiple je t devices that collect air over a w ide range o f  
flow rates.
One major practical difficulty w ith impactors is the bounce o f  particles after collision 
with the collector surface. It is therefore necessary to exert care and effort towards 
m inim izing the problems reviewed by some researchers in order to yield proper size- 
segregated samples.109 Coatings have often been applied to lower bounce in 
conventional multistage im pactor.110
2.4 2.2. CYCLONES
The cyclone also known as cyclone collector or cyclone separator has been in use 
over 100 years and is still one o f  the most w idely used o f  all the industrial gas- 
cleaning devices. The main reason is that it is inexpensive, have no moving parts and 
it can be constructed to w ithstand very harsh operating conditions depending 011 the 
material used in the construction. m -,,s Typically, an aerosol enters tangentially near 
the top (Fig. 2-10) and is forced into a downward spiral simply because o f  the 
cyclone’s shape and the tangential entry. Cyclones use inertia to remove particles 
from the aerosol. The cyclone imparts centrifugal force on the aerosol stream, usually 
within a conical shaped chamber. Cyclones operate by creating a double vortex inside 
the cyclone body. The incom ing gas is forced into circular motion down the cyclone 
near the inner surface o f  the lube. A t the bottom o f  the cyclone, the gas turns and 
spirals up through the center o f  the tube and out o f  the top o f  the cyclone.116
Particles in the aerosol are forced towards the cyclone walls by the centrifugal force 
o f the spinning gas but are opposed by the fluid drag force o f  the gas traveling 
through and out o f  the cyclone. For large particles, inertial momentum  overcomes the 
fluid drag force so that the particles reach the cyclone walls and are collected. For 
small particles, the fluid drag force overwhelms the inertial momentum  and causes 
these particles to leave the cyclone walls with the existing gas. G ravity also causes the 
larger particles that reach the cyclone walls to travel down into a bottom  hopper. 
While they rely on the same separation mechanism  as momentum  separators, cyclones 
are more effective because they have a more complex gas flow  pattern .116
This type o f  technology is part o f  the group o f  air pollution controls collectively 
referred to as “precleaners” . They are oftentimes used to reduce the inlet loading o f 
particulate matter (PM ) to downstream  collection devices by removing larger and 
abrasive particles. The partially clean airstream is then passed to a more expensive 
final control devices such as fabric filters or electrostatic precipitators (ESP). Hence 
they are used to control PM, and primarily PM greater than 10 m icrometers (|im ) in 
aerodynamic diameter. However, there are high efficiency cyclones designed to be 
effective for PM < 10 (im or < 2.5 |im  (PMio and PM2.5).
The collection efficiency o f  cyclones varies as a function o f  particles size and cyclone 
design. Cyclone efficiency generally increases with:
• particle size and/or density
• inlet duct velocity
• cyclone body length
• number o f  gas revolutions in the cyclone
• ratio o f  cyclone body diameter to gas exit diameter
49
•  dust loading
• smoothness o f  the cyclone inner wall
C le a n ed  nu i
V o r tn x - lin d u f tu&6
la n g u r iW r i
iiitet du&l
Dusty 
30s in
t
O u s t am
FIGURE 2-10: SCHEMATIC DIAGRAM  OF A  CYCLONE
Cyclone efficiency w ill decrease w ith increases in:
• gas velocity
• body diameter
• gas exit diameter
• gas inlet duct area
• gas density
50
A common factor contributing to decreased control efficiencies in cyclones is leakage 
o f air into dust outlet.117
There are three classifications o f  cyclone control efficiencies, i.e., conventional, high 
efficiency, and high-throughput. H igher efficiency range cyclones are designed to 
achieve higher control o f  smaller particles than conventional cyclones. H igher 
efficiency cyclones come with higher pressure drops, which is a lim iting factor in 
cyclone design. ln ’112’118 H igh throughput cyclones are only guaranteed to remove 
particles greater than 20 |im , although collection o f  sm aller particles does occur to 
some extent. The control efficiency o f  the various classifications is given in Table 2-1.
TABLE 2-1: CONTROL EFFICIENCY RANGES FOR THE THREE  
CYCLONE CLASSIFICATION AND TYPE OF AEROSOL119
% Efficiency
Type of 
Aerosol Conventional High Efficiency High Throughput
TSP 70 - 90 8 0 - 9 0 8 0 - 9 0
PM10 30 - 90 6 0 - 9 5 1 0 - 4 0
PM2.5 0-40 20 - 70 0 -  10
2.4.3. DIRECT-READING INSTRUMENTS
The direct reading instruments (usually referred to as survey instruments) are those 
that can measure the mass concentration in situ. Generally, the direct-reading 
instruments are less accurate than the integrated samplers, but more convenient 
because their speed o f  operation allows monitoring and assessment, in space or time, 
o f changing concentrations. Because o f  their time-sensitive data, they are very useful 
in source identification and control, short term compliance monitoring, emergency 
response, forensic investigations and numerical modeling. Some o f  the direct-reading
51
mass concentration measuring instruments are Tapered E lement Oscillating 
M icrobalance (TEOM ) , Piezobalance and Beta gauge methods among others.
In addition to direct-reading mass concentration measuring methods there are other 
instruments such as optical particle counters, electric mobility analyzers, condensation 
nuclei counters and photometers. These are all based on some characteristics o f  the 
aerosol which is measured and related to either the number o f  particles or the mass. 
O f interest to this thesis are the direct-reading instruments measuring mass 
concentration.
The direct-reading instruments require a sensitive means o f  determ ining the mass o f 
the particles collected in situ. The main problem associated with the direct-reading 
instrument is how to remove free water and standardize the sampling procedure. This 
is achieved by heating the air stream and filter. This heating procedure is likely to 
volatise some particles from the air stream and filter. To overcome this potential for 
volatisation, some researchers have reduced the operating temperatures from 50"C to 
40' C and further consideration is being given reduce the temperature to 30 ' C .120 In 
addition to the above lim itations o f  the direct-reading instrument methods, there are 
performance limitations due to some or all o f  the follow ing factors: humidity, gas 
adsorption, and particle collection efficiency.
2.4.4. SAMPLING OF GASES
There are gas analyzers w ith sufficient sensitivity and speed o f  response to give a 
continuous measurement o f  the pollutant concentration as the sample is delivered to 
them. Many other techniques simply capture the pollutant, or a chem ical derived from 
it, from the sample for later quantitative analysis in the laboratory by conventional 
methods such as titration and spectrophotometry. The technique for gas pollutant
52
measurement is that a specific chemical reaction should occur between the gas and the 
captive absorbed molecules retain all the gas from the sample air. There are different 
methods; among them are bubblers, impregnated filters, diffusion tubes and denuder 
tubes.
There are two distinct group o f  gas air pollution samplers -  Passive samplers 
(automatic samplers) and active samplers (non-automatic samplers). The non-passive 
samplers are generally cheaper and easier to install and maintain but do not give as 
good accuracy or resolution as passive methods.
The most commonly used passive sampler is the diffusion tube. This provides a 
simple and inexpensive method o f  screening air quality in an area, to give a general 
indication o f  average pollution concentration over a period o f  weeks or months. The 
sampler consists o f  a small plastic tube open at one-end and an absorbent packed at 
the other. The absorbent used depends on the pollutant gas to be monitored; nitrogen 
dioxide being the most common, then benzene, sulphur dioxide and ozone. Tubes are 
usually exposed for two to four weeks then sent to the laboratory for analysis. The 
main advantage is the low cost which permits sampling at a number o f  points in the 
area o f interest; this is very useful in highlighting “hotspots” o f  high concentration 
where detailed studies may be needed.
Active samplers can be divided into two groups; namely, the non- or sem i-automatic 
and the automatic (continuous analyzers). Non -  or sem i-automatic sampler methods 
collect pollutants either by physical or chemical means for subsequent analysis in a 
laboratory. Typically, a known volume o f  air is pumped through a collector such as a 
filter or a chemical solution for a known period o f  time, which is then removed for
53
analysis. Examples are sulphur dioxide or smoke bubblers. On the other hand the 
automatic samplers produce high-resolution measurements for pollutants such as 
ozone, oxides o f  nitrogen, sulphur dioxide and carbon monoxide. Hydrocarbons can 
also be measured, but are limited by cost. Because the data are analyzed on-line and in 
real-time, it is especially useful for monitoring pollution episodes (e.g. heavy traffic 
emissions). In order to ensure that the data produced are accurate and reliable, strict 
maintenance, operational and quality assurance/control procedures are required and 
must be rigidly followed if  the results are to be relied on.
There are other methods that involve adsorption o f  the gas molecules by 
intermolecular forces to the surfaces o f  a solid collecting substrate and subsequently 
stripping them o ff by heating to appropriate temperatures or solvent extraction and 
delivered into a gas chromatograph or an equivalent analysis system. Also there is 
condensation trapping at liquid oxygen temperature (-183" C) so that volatiles 
condense out but the air itself passes through unchanged.
2.5. FILTERS
The material on which the aerosol sample is collected is very important if  any 
meaningful results are to be obtained from the analysis. The material should be able to 
retain the particles, w ithout com ing o ff very easily, but m ust perm it air flow readily 
through it. The sample collected should be provided in a manner that would make it 
easy to analyze both for mass concentration and chemical composition. Fibrous and 
porous membrane filters are the most commonly and important filter types used in 
aerosol sampling.
54
Fibrous filters consist o f  a mat o f  randomly positioned fibers that have been pressed 
together, and often w ith a binding material, so that most are perpendicular to the 
direction o f  the airflow. Aerosol particles are removed by these types o f  filters when 
they collide and attach to the surface o f  the fibers and not as m icroscopic sieves in 
which only particles smaller than the pore size can get through, as in the case o f  liquid 
filtration.
These types o f  filters have porosity from 70 to greater than 99%  and range in size 
from subm icrometer to 100 urn. G lass fiber, quartz and cellulose (paper) are the most 
common types. They have lower pressure drops across them  and are well suited for 
high volume sampling. H igh-efficiency filters have a low air velocity  through them. 
Thus to obtain a large filter area in an element o f  convenient size the filter materials 
are pleated. The glass and quartz fiber filters have a high retention o f  particles with 
sizes above 0.3 urn.
Glass fiber filters have very high capacity but one major draw  back is its ability to 
convert sulphur dioxide to sulphate hence increase the PM load in situ. This is not a 
problem with quartz fiber or paper filters. The glass fiber filters have very high blank 
values for a lot o f  elements and therefore they are unsuitable for trace element 
analysis especially w ith nuclear analytical techniques. The quartz fiber filter, though 
better than glass fiber for blanks, still have high blank values and therefore are equally 
unsuitable for trace element analysis. The penetration o f  particles into the filter mat o f 
fibrous filters is not uniform thus they are not useful for EDXRF analysis. Though 
cellulose filters can be used for ambient monitoring, there are problems associated
2.5.1 FIBROUS FILTERS
55
with maintaining the same humidity for the exposed and unexposed filters. This has 
led to very limited use.
2.5.2 POROUS MEMBRANE FILTER
Porous membrane filters are made from cellulose esters, sintered metals, polyvinyl 
chloride and/or Teflon and other plastics. They have a  different structure from the 
fibrous filters, w ith less porosity, between 50 -  90%, than the fibrous filters. Gas flow  
through the filter follows an irregular path through the complex pore structure. When 
aerosol is drawn through the filter, particles are deposited on the structural elements 
that form the pores. Membrane filters have high efficiency, even o f  particles smaller 
than the m anufacturer's stated pore sizes which are based on liquid filtration, and a 
greater resistance to airflow  and higher pressure drop than other types o f  filters.
For trace element analysis using nuclear analytical techniques, as in this work, the 
membrane filters are better suited. The filters commonly used are polycarbonate (e.g. 
Nuclepore) and polytetrafluoroehylene (PTFE) popularly known as Teflon. A number 
o f tests have been carried out on the efficiency o f  membrane filters and it has been 
found that the collection efficiency depends on pore size, especially w ith the 
Nuclepore filters.121-123 For example, it was observed that the 0.8 |im Nuclepore filter 
had only 72% efficiency for subm icron particles observed w ith condensation nuclei 
counter.37
One problem that is associated with these filters is the loss o f  coarse particles from the 
filter in handling and transport from sampling site to the laboratory which has been 
noticed by a number o f  researchers. Dzubay and Barbour124 have suggested the use o f  
oil coating on the filters to prevent such shake-off effects. The coating should also be 
applied to the blank samples that are analysed as part o f  the analytical process.
56
CHAPTER 3 
ANALYTICAL TECHNIQUES AS APPLIED TO AEROSOLS
3.1 INTRODUCTION
Historically, people detected aerosol particulates w ith their lungs, their noses and their 
eyes. D iscom fort from smoke inhalation near cooking fires surely resulted in people 
moving upwind. The most common indicator o f  pollution, however, has been smoke 
and/or visible clouds due to the ability o f  the clouds to scatter light as a result o f  
pollutant gases and particulates.125 High correlation observed among black smoke 
from chimneys, reduced visibility, black deposits on buildings and clothing, and 
respiratory distress in the 14th century England led to the first recorded air pollution 
regulation by a royal decree.126 Though the measurement method was crude, and 
many o f the health consequences may have been caused by invisible S 0 2 gas, the 
regulatory decision was correct.
One o f the most common approaches for determ ining the composition o f  atmospheric 
aerosols involves the analysis o f  deposits collected on filter substrate. While filter 
sampling is inexpensive compared to online measurements, it requires manual 
operations and the number o f  filters that must be analyzed before any meaningful 
deductions can be made is large (usually > 30). G ravimetric analysis is used almost 
exclusively to obtain mass measurements o f  filters in a laboratory environment. 
Gravimetry determ ines the net mass by weighing the filter before and after sampling 
with a microgram sensitive balance in a temperature- and relative humidity-controlled 
environment. The main problem  that arises here is interference from electrostatic 
charges which induce non-gravimetric forces between the filter and the balance.127 
The charge is usually removed from most filters by exposing it to low-level
57
radioactive sources prior to weighing. Accurate gravimetric analysis requires the use 
o f  filters w ith low dielectric constants, high filter integrity, and inertness w ith respect 
to absorbing water vapour and other gases. L iquid w ater associated w ith particle 
deposits is effectively removed by equilibration at low temperature and relative 
humidity. Some particles may however volatize if  they are exposed to ambient air for 
more than a day or tw o .128,129
No matter how much air is drawn through a filter, and despite occasionally high 
particle loadings in the atmosphere, the amount o f  sample available for chem ical 
analysis is very small. The typical mass loading on a low- to medium  volume sampler 
is less than 5 mg and many o f  the chemical species o f  interest that must be measured 
are less than 1 |ig  in the deposit.
While most o f  the current air particulate standards are based on mass concentration, 
there is need to know  the chem ical composition so as to  determ ine the biological 
effect as well as the sources em itting these particles. The essence o f  any analytical 
method in environmental science is to make measurements in order to attain a 
specified goal. Hence the sampling and analytical methodology must be appropriate. 
Therefore in this work and for any analysis, there is the need to find the qualitative 
nature o f  the sample and the quantity or concentrations present so as to make 
informed decisions. There are three main factors to consider in any analytical 
technique and this is especially critical in analysis o f  aerosol particles, where the mass 
o f the deposits collected is very small:
• Specificity: In determ ining the presence o f an element, with what certainty is 
the presence determ ined?
58
•  Sensitivity: G iven the presence o f  an element, what is the change in the 
detected element per unit change in concentration?
• Detection limit: What is the smallest amount o f  concentration or amount that 
can be ascertained?
In addition, how will the sample be presented; is there a need for sample preparation?
Elemental concentrations in atmospheric aerosol reflect the processes generating 
them. Therefore, elemental and/or chemical composition analysis o f  atmospheric 
aerosols is very useful to detect the sources o f  aerosols. The ultim ate goal o f  any 
aerosol measurement and analysis is to determ ine the particle sources and where 
appropriate study physical and chemical processes occurring in the particles during 
their life time (atmospheric transportation).
There are many analytical techniques available for the determ ination o f  elemental 
composition o f  materials (i.e. aerosol particulates). Some o f  the methods used in 
aerosol chemical (or species) determ ination includes among others; inductively 
coupled plasma mass spectroscopy (ICP-MS), inductively coupled plasma auger 
electron spectroscopy (ICP-AES), inductively coupled plasma optical emission 
spectrometry (ICP-OES), gas chromatography -  mass spectrometer (GC-MS), 
instrumental neutron activation analysis (INAA), energy dispersive x-ray fluorescence 
(EDXRF), m icro-X -ray energy dispersive fluorescence analysis (p.-EDXRF), proton 
induced x-ray emission (PIXE), total reflection x-ray fluorescence (TXRF), thermo- 
optical and light scattering methods (for total, organic and black carbon [BC ]).130'137 In 
particular, advances in electronics have led to the development o f  a w ide variety o f 
instrumental methods to be used with the older classical chem ical procedures. Most o f 
the instrumental techniques use “comparator” instead o f  “absolute methods” . In
59
general, the equipment required may be rather expensive, but their convenience, 
sensitivity, accuracy and precision make them attractive. These instruments have 
changed the face o f  both qualitative and quantitative analysis.
A variety o f  spectrometric techniques depend on the electromagnetic spectrum  
emitted or absorbed by the sample being analysed. The spectra are characteristic o f 
the elements or chemical compounds present in the sample. The analysis is made by 
comparing results from samples w ith those obtained from  calibration standards under 
identical conditions. The key component o f  these types o f  instrumentation is devices 
to isolate a chosen discrete energy or a range o f  energies (or w avelengths) for 
measurement.
The energies (or w avelengths) emitted or absorbed by elements and chemical 
compounds o f analytical interest range from gamma- and x-ray regions (<100 A or 10' 
8m) through the ultraviolet and visible regions to the infrared (7000 A). There is no 
single spectrometer that can span this wide range which is about 1 -  150,000 eV. 
Hence instruments that are versatile in terms o f  good performance over a wide range 
o f applications are preferred. X-ray spectrometers, especially X-ray Fluorescence 
(XRF) and electron probe analysis systems, belong to this category and are currently 
used world wide.
3.2. X-RAYS
3.2.1 X-RAY FLUORESCENCE ANALYSIS
X-rays are part o f  the electromagnetic spectrum with wavelength ~10‘5 to -100A  
produced by the deceleration o f  high energy electrons and/or by electron transitions in 
the inner orbits o f  atoms. X-rays have the follow ing general properties:
•  Occur as continuous spectra
60
• Occur as characteristic line spectra
•  Occur as characteristic band spectra
• Produce characteristic absorption spectra
•  Propagate w ithout transfer o f  matter
• Are unaffected by electric and magnetic fields
• Are invisible and hence undetected by the human senses
There are three main interactions o f  x-rays w ith matter, namely -  Photoelectric 
absorption, Compton scattering and Pair production. Depending on the material, when 
X-rays interact w ith matter a fraction might pass through the sample, a fraction might 
be absorbed in the sample and produces fluorescent radiation, and a fraction is 
scattered. Scattering can occur with or w ithout loss o f  energy. W hen there is loss o f 
energy it is known as Compton scatter, and w ithout loss o f  energy is known as 
Rayleigh scatter.
When x-rays o f  intensity, I0, pass through a material o f  finite thickness, d, and density, 
p, the transm itted intensity o f  photons which have not suffered interactions, I, in the 
material is given by the Beer-Lambert law
j    r ~ U p d  i  i
/ - y o e  3-1
where
|a is the mass attenuation coefficient, which is energy dependent. Thus |r is 
composed o f  three major components:
H ( E ) = t ( E ) +  <j  c o h ( E )  + G inc ( E )  . . .3 - 2
where
I (E )  is the photoelectric mass absorption coefficient
61
Gcoh(E) is total coherent mass scattering coefficient 
<Ji„c(E) is total incoherent mass scattering coefficient
Pair production is only possible when the energy o f  the photons is greater than 1.02 
MeV and as such not important for x-ray photon interactions. These three main 
processes depend on the thickness o f  the sample (d ), density (p) and composition o f  
the material and the incident x-rays energy (E„).
Photoelectric absorption occurs when the atom (or sample) being irradiated 
completely absorbs the x-ray photons. Electrons in the inner shell o f  the atom 
(sample) can be expelled by x-ray photons or electrons w ith sufficient energy. For 
example if the ejection takes place in the K-shell, a hole is created in the K-shell 
making the atom unstable w ith a higher energy. An electron from  a higher orbit w ill 
fill the hole created in the K-shell. The excess energy is em itted as x-rays and when 
this happens, it is seen as a line in a spectrum. The energy o f  the emitted x-rays 
depends on the difference in energy o f  the shell w ith the initial hole and the energy o f  
the electron that fills the hole. Each atom has its specific energy levels, so the emitted 
radiation is characteristic for that atom. D ifferent holes can be produced in the atom, 
hence an atom emits more than ju s t one energy (or line). The emitted line(s) is/are 
characteristic and serve as a fingerprint o f the element.
For an electron to be removed from an atom, the x-ray must have a higher energy than 
the binding energy o f  the electron. I f  an electron is removed, the incom ing radiation is 
absorbed, the higher the absorption the higher the fluorescence. Hence the process is 
known as X -ray Fluorescence (XRF). For very high energies very few  photons are
62
able to interact w ith the atom to eject electrons as most o f  the energy ju s t passes 
through.
X-ray fluorescence (XRF) is an analytical technique used to determ ine chemical 
composition o f  all kinds o f  material and have been in use around 1910.138 The 
sample/material could be solid, liquid, powder, filters and in other forms. XRF also 
used to measure thickness and composition o f  layers and coatings. The main 
advantages are that it is fast, accurate and non-destructive and usually requires little or 
no sample preparation. Currently the application o f  XRF is very w ide and diverse and 
includes among others the food, plastic, metal, cement, oil and m ining industries. It is 
also applied in m ineralogy, geology, pharmacy, medicine and research. It is also used 
extensively in environmental analysis o f  soil, water, air, vegetation and waste.
X-ray fluorescence has been used to quantify elements from  atom ic number 11 (Na) 
to 92 (U). In aerosols, it has been used to detect and quantify elements from N a to La, 
Au, Hg, Tl, Pb and U. In addition to its multi-elemental nature, another advantage is 
that sample preparation is generally not necessary. Most importantly, the filters 
remain intact after analysis and can be analyzed using other techniques to give 
additional information. To achieve greater sensitivity and efficiency for low Z 
elements like Na, Mg, Al, and Si the filter is placed in a vacuum . This can also result 
in the evaporation o f  volatile compounds but can be m inim ized by the introduction o f  
helium atmosphere.
Basically, XRF spectrometer systems consist o f  a source (o f x-ray or generator), a 
sample holder and a detection system (detector). The spectrometer systems are 
generally divided into two main groups: energy dispersive (EDXRF) and wavelength
63
dispersive (WDXRF). The difference between the EDXRF and WDXRF is found 
mainly in the detection system. The EDXRF detector detects and measures different 
characteristic energies em itted from the sample. On the other hand, the WDXRF 
detector uses an analyzing crystal to disperse the various w avelengths sim ilar to a 
prism dispersing different colours in different directions. A lthough WDXRF, 
generally, displays a higher sensitivity for low-atomic number elements than EDXRF 
but it requires higher excitation power. This higher excitation energy may produce 
excessive sample heating, causing the filter to brittle and potential degradation o f  the 
filter material. EDXRF applies 100 to 1000 times lower excitation energy and hence 
avoids sample damages.
In this work only the EDXRF system  has been used. Generally, the source irradiates a 
sample and a detector measures the radiation coming from  the sample. X -ray tubes 
and Si(Li) detectors was used in this work as the source and detector respectively.
3.2.2 EDXRF SPECTROMETRY
When characteristic x-rays pioduced from a sample are incident on the detector, in 
this work a sem iconductor detector, it interacts w ith the detector material. The x-ray 
energies produce ionization in the detector. The total ionization produced by each 
energy photon entering the detector is converted to voltage signals w ith amplitude 
proportional to the energy o f  the incident photons.139 The detector signals can then be 
processed by a pulse processor unit and a multi channel analyzer to obtain a spectrum 
o f the sample. The position o f  the peak in the spectrum  corresponds to the energy o f 
the x-ray photon from which it is coming and the net area under each peak in the 
spectrum is used in the elemental quantification o f  the sample.
64
There are th ree very  im portan t p roperties o f  de tec to rs that a re  taken  into considera tion
before they  are used in x -ray  fluo rescence  analysis nam ely;
•  R eso lu tion  - The ab ility  o f  the d e tec to r to  d iffe ren tia te  betw een  d iffe ren t
photon  energ ies. H igher reso lu tion  de tec to r is one tha t has the best ab ility  to
d ifferen tia te  betw een  tw o  very  close pho ton  energ ies.
•  Sensitiv ity  - H ow  effic ien tly  the incom ing  pho tons are counted . T herefo re  
h igh  sensitiv ity  im plies tha t the num ber o f  pu lses being  recorded  aga in s t the 
num ber o f  incom ing  pho tons is very  high.
•  D ispersion  - The ability  o f  the de tec to r to  separate  x -rays w ith  d iffe ren t
energ ies, a h igh d ispersion  im plies tha t d iffe ren t energ ies are w ell separated .
The fluo resced  characteristic  x -rays p roduced  by the sam ple is used  fo r both  the 
qualitative and quan tita tive  analysis. T hough  the fluo resced  characteristic  x -rays are 
generated  as a resu lt o f  pho toe lec tric  in terac tions, the sca ttering  (both  coheren t and 
incoherent) o f  the excita tion  energy  genera tes a background  tha t in terferes w ith  the 
analysis o f  the characteristic  x -rays. The in tensity  o f  the m easu red  signal depends on a 
number o f  factors; the
•  E nergy  o f  inciden t x -rays for m onochrom atic  excita tion  o r energ ies o f  incident 
x-rays for po lych rom atic  excita tion
• D etector effic iency
• F luorescence y ield  o f  sam ple
• A ttenuation  o f  pho tons by sam ple (depends on sam ple th ickness)
•  A ttenuation  o f  pho tons in a ir (depends on the sou rce-sam p le-de tec to r
d is tances)
65
•  Inter-element effects (the measurement o f  the element o f  interest may not 
depend only on its concentration but also on the concentration o f  other 
elements present)
• The number o f  atoms o f  the element o f  interest in the sample.
In this work monochromatic excitation was used and will be the only excitation 
discussed.
For monochromatic excitation, the characteristic (fluorescent) x-rays which are 
emitted by an element i in a thin layer dx  at a depth w ithin a sample is given by140
d lr - 70Gcsc?1T,-(prf)f{l-  (1 / J ^ f f  e nxp CSCV ' XP CSCp2^
...3-3
where
Io, I  -  intensity o f  primary and fluorescent radiation respectively 
(photons/second)
x, -  photoelectric mass absorption coefficient at the incident energy, E„, 
(cn r/g )
J, -  the ratio (jump ratio) between the photoelectric mass absorption 
coefficient at the top and bottom o f  the absorption edge o f  the element o f 
interest
W, -  fluorescent yield o f  the element for interest
/ - f r a c t i o n  o f  the energy E, with respect to total x-rays energy emitted 
Ho, Hi -  mass absorption coefficients o f  the sample to primary, E0, and 
fluorescent, E„ radiation respectively 
£, -  efficiency o f  the detector for x-rays o f  energy E,
66
(fit, q>2 -  angle between sample surface and primary/fluorescent radiation
respectively
G  -  geometric factor
p, -  density o f  element i w ithin sample
p -  density o f  the sample
By integrating Equation 3-3 for a specimen o f  thickness d (i.e. between x  = 0 and x 
= d ), and multiplying the numerator and the denom inator by d, the follow ing general 
equation is obtained:
Ij = I 0G C S C (1 /  J i ')}wif ie i {pd) i {pd'){\-  e ~a:xpdl} / a i s pd
...3 -4
where a  — (ioCSC Cpi +  |U iCSC  (p2 ... 3-5
For very thick samples, the exponential term in Eqn 3-4 decreases rapidly with 
increasing thickness d  o f  the sample layer. This infinitely thick sample layer is often 
reached in practice where solids and bulky materials are analyzed.56 Under such 
conditions, Eqn 4 can be written as
j  =  G 0  K j ( p d ) j  3_6
1 a
where
a  = |i0 + (a; for cpi = (p2 = 90°
Go =  Io G
and K j  =  { 1  — ( 1 /  J i  )  } \ V i f j £  j
a constant that involves all the fundamental parameters 
Eqn 3-2 can be further w ritten as
67
1 =  G o K t P ^ i  Ac.rr • ■ ■3-7
where A Co r r =  {1 — e ~ a '.*P d ' } /  a i s  p d  ... 3-8
For a sufficiently thin sample, Eqn 3-4 can be approximated by the expansion o f  the 
exponential term, as:
exp(-x) ~  1 — x, and becomes
I  /  =  . . .3 - 9
But
S i  =  G o K  j ...3-io
where S, is the Sensitivity o f  the spectrometer.
The sensitivity value depends on a number o f  experimental factors and the 
characteristics o f  the sample and is independent o f  the physical nature o f  the sample. 
Hence the sensitivity can be obtained by simply calibrating the equipment with 
standards samples o f  known mass per unit area.
From Eqns 3-9 and 3-10 we get
I  i  =  S i i p d ^ i  ...3 -1 1
Hence for “th in” samples there is a linear relationship between the fluorescent x-ray 
intensity (/,) and the concentration (pd),.
Aerosol samples collected on membrane filters can be regarded as thin samples, 
because the x-ray attenuation by the filter material can be neglected, since the 
particulates are deposited on the surface o f  the filter. For very low filter loadings, e.g.
68
less than 100 |ig cm '\  the net measured intensity is proportional to the mass per unit 
area o f  the analyte.
Aerosol particulates however show significant effects o f  the sizes and compositions o f  
individual particles. These effects result from the absorption o f  both the incident and 
fluorescence x-ray radiation within the particles. This self-absorption effect can be 
very significant in x-rays from  light (low  Z) elements. It has been shown that this can 
occur w ithin the individual coarse particles, which are collected on the surface o f  the 
filter or in fine particles collected in absorbing layer which is at least partly w ithin the 
volume o f  the filter.141 This can be reduced if  the aerosol is size segregated. For 
example, the fine particle mode aerosol can be the result o f  gas to particle conversion 
and coagulation, so that a reasonable amount o f  sim ilarity in composition among 
particles is expected. But this is not the case for the coarse particulates which are 
generated from mechanical processes and can be diverse in composition.
3.3 THE BLACK CARBON REFLECTOMETER
Black carbon (also known as elemental carbon) and organic carbon (OC) have been 
found to be a significant cause o f  light extinction142"145 and a major chemical 
constituent o f  atmospheric aerosol (TSP, PM ;S, PMio).143'146,147 Three classes o f  carbon 
are commonly measured in aerosol samples collected on quartz-fiber filters. These are 
organic (volatilized or non-light absorbing carbon), elemental or light absorbing 
carbon and carbonate carbon. Carbon dioxide (C 0 2) measurement evolved upon 
acidification however can be used to determine carbonate carbon (K 2C 0 3, N a2C 0 3, 
M gC 03, C aC 03).148
Elemental or black carbon (EC or BC) generally refers to particles that appear black 
and are sometimes called “soot” or “graphitic carbon” . BC is more useful when
69
consideration is being given to visibility reduction or light absorbing carbon, though 
other organic carbon also absorbs light (e.g. motor oil, asphalt, tar, coffee). For source 
apportionment by receptor models, several consistent but distinct fractions o f  carbon 
in both source and receptor samples are preferred regardless o f  their light absorbing or 
chemical properties.149
Carbon is very abundant in suspended particles, hence simple and reliable methods 
are needed to quantity it in aerosols. Methods that are able to distinguish between BC 
and OC are preferable because these species differ in origin, atmospheric chemistry 
and optical properties. One such method was developed by G agel150 and is known as 
the “black smoke method” . A light source (high-performance LED with maximum 
emission at 650 nm) shines on the aerosol filter and the reflected light is measured by 
photo-diodes located in a black housing. The Reflectometer is calibrated by the 
manufacturer. M easuring the reflected light emitted by a white filter and a totally 
black filter, enables the calibration parameter o f  the m anufacturer’s to be used.
The output voltage is converted to a measure o f  blackness known as “black smoke 
number”, RZ. This is determ ined from the three output voltages obtained (from the 
aerosol filter to be evaluated, the totally white filter and the totally black filter). The 
equation relating the output voltage to the black smoke number is:
R 2 ,= RZmax(URZ0 — U rz) / (U rzo “  URZmax)) ...3 -12
where
Urzo =  output voltage with blank (white) filter (which is set to 8.0 V according 
to the instructions manual)
URz,„ax = output voltage with totally black filter (set to 0.4 V)
Urz = output voltage with actual filter to be evaluated
70
Provided that a thin layer o f  aerosol particles is collected on the filter (“single dust 
layer”), there is a simple relationship between the mass concentration o f  BC collected 
on the filter, CR and the RZ. This relation is given by:
CR = - (RMi/V) In (l -  (RZ -  RZ0)/(kRZmax)) ...3 -13
where
CR = the black carbon concentration 
V =  the sampled air volume
RMi = the black carbon mass in a single dust layer on the filter 
RZo = the black smoke number for a white (blank) filter 
RZ = the black smoke number for the actual filter 
RZmax = the black smoke number for a black filter 
k is calibration constant.
3.4 OTHER ANALYTICAL TECHNIQUES
The techniques mentioned above are commonly applied methods to aerosols and they 
are used to generate very important results in terms o f  mass, elements and BC. 
However, other aerosol analytical techniques such as neutron activation analysis are 
also applied to aerosol. N eutron Activation Analysis (NAA) is a very powerful 
technique for the non-destructive multi-elemental determ ination o f  many trace 
elements and has been applied to aerosol by many o f  researchers.,51‘15S It was first 
proposed as an analytical tool in the late 1930’s .,5<’' 1S9 It was discovered that samples 
containing certain rare earth elements become highly radioactive after exposure to 
neutrons. In general, the technique involves the bombardment o f  a sample with 
particles (such as neutrons) or radiations. A nuclear reaction will occur if  the energy 
o f  the radiations or particles exceeds the threshold energy. Stable isotopes in the
71
sample may be converted to radionuclides, which then undergo radioactive decay, a 
process accompanied by the emission o f  gamma-radiation (y-rays). The detection o f 
radiation can be used to identify and quantify the elements present in the sample.
New analytical techniques are also being continuously created that provide specific 
measurements and for example, organic compounds, m ineral compounds, particle 
shape and isotopic abundance in particles sampled onto filter. Some o f  these 
measurements have also been used to relate aerosols concentrations to their sources.
Electron microscopy has been used, for example, to study the size, morphology and 
concentration data .160,161 Elements have been identified in single particles using 
Electron probe m icroanalysis (EPMA).162 X-ray diffraction (XRD) has been used to 
identify compounds in crystalline structures including silica, m ineral dusts and 
asbestos.163-'66
Isotope Dilution Mass Spectrometry and Accelerator Mass Spectrometry have been 
used to measure isotopic ra tios167"175and isotopic concentrations176_179respectively for 
pollution source identification. For example, l4C is present in em issions from 
vegetative burning but absent in emissions from fossil fuel combustion, hence by 
using accelerator mass spectrometry the isotopic concentration o f  |,4C can be measured 
with a very high sensitivity.
Other analytical techniques such as Raman Spectroscopy, Laser M icroprobe Mass 
Spectrometry (LMMS), Fourier Transform  Infrared (FTIR) Spectrometry, Electron 
Spectroscopy for Chemical Analysis (ESCA), Scanning E lectron M icroscopy (SEM), 
Light M icroscopy, etc have all been applied to examine, identify and/or quantify 
aerosol samples.
72
CHAPTER 4 
EXPERIMENTAL METHODS, ANALYSIS AND RESULTS
4.1 SELECTION OF STUDY AREA
Monitoring o f  pollution levels, and hence air particulates, in the atmosphere is o f 
fundamental importance because it enables us to measure the extent to which 
pollution is actually occurring and how efforts to mitigate it are working, if  any. This 
is not an easy task in urban and sem i-urban areas because the air quality is the result 
o f a complex interaction between natural and anthropogenic environmental 
conditions.180 The urban and sem i-urban air pollution is a serious environmental 
problem because o f  its varying and irregular distribution o f  sources. Urban areas are 
small densely populated area and over this small area are mixed development 
comprising industries, real estate and in some cases, like in Ghana, agricultural 
industry.181 Site selection in any monitoring programme is a very important and 
crucial and a lot o f  consideration has to be given to it. For a strong source or few 
strong sources, air monitoring sites are not determ ined qualitatively but quantitatively 
using atmospheric dispersion m odels.181"186 In Ghana, as in this work, we do not know 
where the sources are and other parameters using dispersion model for site selection 
was not feasible. It was therefore critical that whatever site that was selected can 
produce samples that are representative o f  conditions prevailing in that environment at 
the time o f  sampling. Site selection requires the need to:
• Identify the purpose to be served by monitoring
• Identify the monitoring site type that will best serve the purpose
• Identify the general location where the site is placed
• Identify specific monitoring sites
The sampling position at the sampling site is also very important.
73
There are several functions that a monitoring station can serve and hence there are a 
number o f  criteria for the location o f  a sam pler.182 For example two important criteria 
are:
•  Determ ination o f  the effect o f  source emission changes on air quality
•  Assessment o f  the effective dose level to the population.
These two reasons w ill certainly not lead to the same sampling location and hence it is 
the object o f  the study that provides a rational and systematic means o f  sighting the 
monitoring station. A number o f  issues must be taken into account in site selection 
since it is not always possible to optim ise measurements for all air pollutants at any 
one location.
Monitoring sites are usually classified according to the type o f  environment in which 
they are located or the pollutant source. The site description reflects the influence o f  
either the particular pollutant source or the overall land use. Typical monitoring 
location type includes:
• Urban centre -  an urban location representative o f  typical population exposure 
in towns and city centre
•  Urban background -  an urban location distanced from sources and therefore
broadly representative o f  city-w ide background conditions (e.g. urban
residential area)
• Suburban -  a location type situated in a residential area on the outskirt o f  a 
town or city
• Roadside -  a site sampling within 1 -  5 m o f  a busy road
• Kerbside — a site sampling within 1 m o f  a busy road
74
•  Industrial -  an area where industrial sources make an important contribution to 
the total pollution burden
• Rural -  an open countryside location, in an area o f  low  population density 
distance as far as possible from roads, polluted and industrial areas
• O ther -  any special source-oriented or location category covering monitoring 
undertaken in relation to specific em ission sources such as power stations, 
carparks, airports, tunnels, etc.
In Ghana, no sustained air particulate monitoring has been undertaken to guide this 
work. Lack o f  adequate equipment to do multi-site sampling and the need to have a 
general view o f  the air quality were the main guiding principles that influenced the 
site selection. The need to locate the site away from any major source coupled with 
the need to avoid local anthropogenic and soil derived contam inations as a result o f 
re-suspension o f  dust were also considered. The sampling was done in a sem i-urban 
area and the follow ing are some o f  the additional issues considered:
•  Site accessibility -  the sampling site must be accessible for site visits but must 
be free from public interference
•  Security o f  the sampler — the sampler was placed in an environment where 
nobody could tem per or vandalise it.
•  Access to utilities -  the sampler was located at a point where it could be 
connected to power source. This is to prevent the use o f  generators which will 
contribute to the sample.
•  Site visibility -  The site was very visible alerting people o f  the measurement 
to prevent any mishap.
75
• A erodynam ic clearance/sheltering — the site chosen allows for free airflow 
around the sampling inlet to ensure representative sampling thus preventing 
the sampling o f  stagnant air or highly sheltered m icroenvironment.
• Local air trajectories -  the location so selected was about 20 km from Accra 
City Centre in the North-East direction which is on a m ajor air trajectory from 
the G u lf o f  Guinea.
In order for future health impact assessment, the sampling height selected is 1.6 m; 
this height is in the breathing level o f  the average Ghanaian. The sampling was done 
at the Ghana Atomic Energy Commission, precisely at the Radiation Technology 
Centre. The coordinates o f  the sampling site are 5°40'35.0 N , 0°11'12.0 W. This is in 
the North-East o f  the Accra City Centre w ith coordinates o f  5”35'00 N , 0"13'08.0 W.
4.2 EXPERIMENTAL PROCEDURES
4.2.1 SAMPLING AND GRAVIMETRY
The aerosol particles were collected using the GENT sampler. The GENT sampler 
consists o f  a compact vacuum  pump system that is controlled by a tim er and 
connected to the stacked filter head unit (SFU) as shown in Fig. 4-1. The filter head 
unit is based on sequential filtration through two filters with different pore sizes. The 
SFU has a pre-impaction stage that acts as a PMn> inlet. The SFU is connected to the 
pumping system via flexible Poly-flow  tubing which was less than 20 m (the 
manufacturer’s recommendation is that it should be less than 100 m). The GENT 
sampler is used to collect particles in the two size fractions, fine (aerodynamic 
diameter, da<2.5 |am) and coarse (2.5<da< l0 |im ) on Nuclepore track etched 
polycarbonate filter. The air inlet is at a height o f  1.6 m above the ground and the total
76
airflow  through the filters was 1 m3lr'(16-17  litres per m inute (lpm)). A sampling time 
o f approximately 24 hours was used in this work.
A filter diameter o f  47 mm was used; the coarse fraction has a nom inal pore size o f  8 
pm and the fine fraction 0.4 pm. The coarse filters are pre-coated (by the 
manufacturer) w ith Apiezon Type L grease to provide a tacky surface to prevent 
particle rebound and consequently prevent sample loss. The fine filters are however 
not coated with Apiezon.
Loading o f  stacked filter unit w ith the pre-weighed Nuclepore filters is done in a dust- 
free environment. Fig. 4-2 show  the SFU cassette unit loaded w ith filters and capped 
to be sent to the sampling site. More than two stacked filter units were used; hence 
when one SFU with the exposed filters was changed, it was replaced w ith another 
preloaded SFU.
The coarse filter is placed in the top part o f  the filter cassette so that the air flow will 
pass through it before getting to the fine filter as shown in Fig. 4-3. The filter holders 
are tightened w ith a special PVC wrench set attached to the sampler. This is very 
important since any air leak will affect the loaded mass on the filter.
The samples were collected during the period 27th December 2005 -  12"’ February 
2007. The sampling time was approximately 24 hours. The filters were conditioned 
for 72 hours, before and after sampling in a temperature and humidity control 
weighing room. Unannounced and planned load shedding made the sampling very 
irregular creating gaps in the data.
77
w o o d e n  p o i c  
w o u u c n  b o a r d  
R y  I o n  c o n n c c m t  
i r a m p a r c m  t v i b i n g  ( s h o r t )  
b r j a s s  c o n n c c t c i i  
s i c e !  c k m p i  
p i  a s i  i c  c t a m p s
r a i n  p r o i c e i i o n  c o v c r  <o r u > c )  
p l a s t i c  c o n t a i n c T  f h l n c k )
w i t h  s t a c k e d  f i l t e r  c a s s e t t e  ( S F U )  i n s i d e
f l e x i b l e  P O L Y  F L O  t u b i n g  ( l o n g )
b r « M  c o n n c c t o r
( p um p  s e t u p
FIGURE 4-1: SCHEMATIC  DIAGRAM OF THE GENT SAMPLER USED
78
vo
lu
m
e 
ro
ei
ar
FIGURE  4-2 TW O  STAGE  STACK  F ILTER  CASSETTE  UN IT  (SFU ) 
LOADED  W ITH  F ILTERS  AND  CAPPED
O -R ings
P lastic  ho lder 
F ine  F ilte r 
P lastic  M esh  
O -R ings
SFU  Base
P lastic  ho lder 
C oarse F ilte r 
P las tic  M esh
O -R ings
FIGURE  4-3 SCHEMATIC  D IAGRAM  OF  SFU
79
Gravimetric analysis determ ines the net mass by weighing the filter before and after 
sampling. The measured mass concentration was calculated in |igm '3as follows:
,  „  L K 4  — I M
M C  =  -------------------------------------------------  . . .  4-1
V T
where
MC = mass concentration (|igm -3)
LM = loaded filter mass (^g )
IM  = initial filter mass ([ig)
V = volume flow  rate in m3h"'
T = sampling time (h)
For practical purposes the PM™ mass concentration was calculated from
PMio =  Coarse fraction (PM io_2.s) + Fine fraction (PM2.s) ... 4-2
A Sartorious MC-5 m icrobalance w ith readability o f  0.001 mg was used for the mass 
measurement after static on the filter was removed using a Po-210 source. The total 
mass o f particulate matter collected on the filters varied from  day to day as shown by 
Fig. 4-4 and 4-5. The calculated daily PM10 mass concentration is presented in Fig. 
4-6. Mean monthly mass concentration for Coarse (PM 10-2.5), Fine (PM2.5) and PM ]0 
aerosol are shown in Fig. 4-7.
4.2.2 BLACK CARBON DETERMINATION
The black carbon analyzer used in this work has been described in Chapter 3. The 
Gagel m ethod150 was used for the analysis. The daily black carbon (BC) concentration 
in the coarse and fine size fractions are shown in Fig. 4-8. The percentage BC to Mass 
Concentration is given in Fig. 4-9.
80
M
as
s 
C
o
n
ce
n
tr
at
io
n
 
(p
g
m
' 
)
FIGURE 4- 4 : COARSE MASS CONCENTRATION  DURING  THE INVESTIGATION  PERIOD
2000
1800
1600
1400
1200
1000
800
600
400
200
IMass Time Series
^  ^  ^  c&  r&  r&  r&  r&
D a t e  E n d e d
81
M
as
s 
C
on
ce
nt
ra
ti
on
 
(n
gm
 
)
FIGURE 4-5: FINE MASS CONCENTRATION  DURING  THE INVESTIGATION  PERIOD
450
400
350
300
250
200
150
100
50
0
500
■  Expt Mass
I
| [  1 I 1
l u l l 1 ll
no°fe ^  ^  ^  ^  ^  ^  c # 5 c #  < #  c #
D a t e  E n d e d
82
FIGURE 4-6: PM 10 MASS CONCENTRATION  DURING  THE INVESTIGATION  PERIOD
2200
Si>
D a t e  E n d e d
83
M
as
s 
C
o
n
ce
n
tr
at
io
n
 
((
jg
m
')
FIGURE 4-7: MEAN MONTHLY  FINE, COARSE AND PM 10 MASS CONCENTRATION
10000
1000
100
10
1
M o n t h
84
U
O
I
FIGURE 4-8: BLACK CARBON CONCENTRATION DURING THE INVESTIGATION PERIOD
20000
0<#> 0<£ 0C#> 0<#> 0<£ <£> (#> #  <#> 0c£ ^  ^  0<£ ^J> 0c£ jd *  X ? jrf*  J $  _<#> ^  ^  ^  ^
^  nnX #  O® CP nQ C? nA 0N ^  $  <i> $  T? &  cf* N% OnN N*3 ■# <* ^  nON ^  qA
D a t e  E n d e d
85
% 
B
la
ck
FIGURE  4-9: PERCENTAGE  BLACK  CARBON  CONCENTRAT IO N  DUR ING  THE  IN V E ST IG AT IO N
PER IOD
35
/ / / / / / / / / / / / / / / / / / / / / / / / / / / / / /
D a t e  E n d e d
86
4.2.3 ENERGY DISPERSIVE X-RAY FLUORESCENCE ANALYSIS (EDXRF)
The general principle o f  EDXRF analysis has been described in Chapter 3. The 
EDXRF spectrometer at the Royal Veterinary and Agricultural University o f 
Denmark was used in the study.’87 The spectrometer is a compact, flexible and 
sensitive unit, using a high power Mo X-ray tube. The primary beam was 
monochromatized by a H ighly Oriented Pyrolytic G raphite (HOPG) crystal o f 
dimension 5x25x75 mm. Using 90° irradiation geometry a vertical/horizontal 
reference system  is maintained. The detector is a Kevex SuperD ry  Peltier cooled 
Si(Li) detector. The detector has an active area 20 mm2 and a 5 |im  Be window. 
The multi-stage peltier module cools the detector crystal and its assembly below 
-95 °C to maintain the energy resolution at 146 eV (FWHM  o f  Mn K a ). The 
vacuum in the cryostat is maintained by an ion pump w ith an internal water-cooled 
heat exchanger to transfer the heat to the ambient air.
The filters to be analysed were mounted onto a standard Spectro  cup and placed in 
the evacuated cylindrical alum inium  irradiation chamber. A PC-controlled conical 
shaped exchangeable eight position sample holder for the mounted filters. The X- 
ray tube was operated at a voltage o f  40 kV and a current o f  40 mA in the 
measurements. The live time o f  each spectrum was 2000 s. The geometry o f  the 
spectrometer is shown in Figure 4-10. Since the irradiation chamber is evacuated, 
elements down to A1 can be detected, analysed and quantified. The spectrometer 
was calibrated using thin film material from N IST  (NBS SRM 1832) and the X-ray 
fluorescence spectra were fitted with AXIL softw are.188,189 The evaluation model 
used for quantification o f  the elements was the fundamental parameters approach19" 
in the QXAS package.191
87
Figure 4-10: A SKETCH OF THE EDXRF SPECTROMETER
When the fundamental parameter approach is applied to air filters, the EDXRF 
intensity Eqn 3-4 in Chapter 3 reduces to
I i = G o £  ( E i ' ) K i ( E o ' ) C i  . . .4 .3
where
I,= measured net intensity o f  analyte i 
C, = concentration o f  the analyte i 
G0 = the instrumental constant 
£ = detector efficiency
K ,=  product o f  all the fundamental parameters o f  the analyte i 
E0, E, = excitation energy and analyte energy respectively
This equation neglects matrix correction because the air filters can be assumed to 
be thin samples. Hence a linear relationship is expected and equation 4.3 can be re­
arranged to obtain
C i = M i I i  . . .4 .4
where M, = Calibration factor
The quantification system  was validated using N IST  Standard Reference Material 
SRM 2783. Table 4.1 shows analysis o f  the standard reference material.
TABLE 4-1: VALIDATION  OF EDXRF SPECTROMETER USING SRM2783
ng/cm 2
Experimental Certified
Ratio
Expt/CertifiedElement
Al 1793  ± 300 2330  ± 53 0 .77
Si 6 5 0 6  ± 4 2 0 5884  ± 160 1.11
K 496  ± 3 1 530  ± 52 0 .94
Ca 1274  ± 70 1325  ± 170 0 .96
Ti 140  ± 10 150  ± 24 0 .93
Cr 13 ± 3 14 ± 3 0 .93
Mn 35  ± 3 32 ± 2 1 .09
Fe 2 5 6 4 ± 138 2661 ± 160 0 .96
Ni 10 ± 2 7 ± 2 1 .43
Cu 41 ± 3 41 ± 4 1 .00
Zn 135 ± 7 180 ± 13 0 .75
Rb 2 .0  ± 0 .6 2 .0  ± 0 .6 1 .00
Pb 34 ± 2 32  ± 6 1 .06
The elemental concentrations determ ined by the AXIL programme are given in 
ng/cm2. Eqn 4-1 was used to express the value in ng/m 3. The m inimum  detection 
limits for the spectrometer are shown in Table 4-2.
89
TABLE 4-2: M INIMUM  DETECTION LIMIT FOR PARTICULATE  
MATTER ON NUCLEPORE FILTERS W ITH THE EDXRF  
SPECTROMETER
Element DL (ng/cm2)a DL (ng/m3)1’
Al 436 228
Si 2 18 1 1 4
S 80 42
Cl 53 28
K 20 1 1
Ca 13 6.8
Ti 7.7 4
V 6 3.2
Cr 4.5 2.4
Mn 3.6 1.9
Fe 2.3 1.2
Co 1.9 1
Ni 1.3 0.7
Cu 1.2 0.6
Zn 1 . 1 0.6
Br 0.7 0.3
Rb 0.7 0.4
Sr 1 0.5
Pb 1.7 0.9
(a)DL is calculated as 3 times the square root o f  background concentration (3a). 
Mo Ka: 17.44 keV. V=40 kV, I=40mA, collection time 2000 s.
(b )DL for particle concentration is calculated fo r  a sampling o f  24 m 3.
The statistical analysis for the measurements is given in Tables 4-3 and 4-4 for coarse 
and fine fractions respectively . The number o f  filters contributing to the data set is 
given as counts together w ith the mean, standard deviation (Std Dev.), median, 
minimum and maximum values. It should be noted that the standard deviations listed 
in the tables, (Std Dev), are not "true" deviations which express fluctuations in 
experimental conditions for the analytical methods. Instead, they are combinations o f 
these and the variations that occur due to changing weather conditions and human 
activities from one day to another. The median, minimum, maximum and St. Dev may
90
give better information on the extent o f  influence o f  these extreme conditions. The 
true Standard deviations for measurements on the same standard sample on this 
instrument are in the order o f  about 10 % (Selin Lindgren; 2006).190
Fig. 4-11 and 4-12 show  the daily variations in the elemental concentration o f  some 
selected elements for coarse and fine fractions respectively.
TABLE 4-3: CONCENTRATION COARSE PM  - ELEMENTS, BC AND MASS  
(28th D ecember 2005 -  12 February 2007)
Element/ nq/m3
# o f
Days Mean StDev Median Min Max
13 Al 201 3828.56 8707.43 1405.39 363.96 52309.42
14 Si 213 13540.01 33808.86 4286.64 118.63 233707.50
16 S 211 522.36 555.15 346.20 42.29 3425.45
17 Cl 213 1458.22 1210.71 1064.24 26.75 6981.94
19 K 213 1414.46 3290.09 467.99 12.08 24135.81
20 Ca 213 2203.30 5992.61 588.88 23.62 49899.12
22 Ti 213 398.75 971.19 128.18 4.21 7413.49
23 V 101 22.17 37.38 10.84 3.11 233.23
24 Cr 171 7.46 8.52 4.94 1.65 70.18
25 Mn 211 76.09 194.58 23.13 2.63 1644.17
26 Fe 216 3604.38 8521.80 1307.18 2.52 67143.01
27 Co 67 12.74 19.29 7.72 2.04 120.46
28 Ni 173 6.85 10.79 3.24 1.55 64.08
29 Cu 198 8.01 12.56 4.37 0.50 89.71
30 Zn 213 18.63 33.81 8.10 1.10 210.55
34 Se 23 0.54 0.13 0.53 0.27 1.06
35 Br 215 8.77 12.26 5.72 0.70 84.58
37 Rb 209 6.79 17.11 1.84 0.49 137.12
38 Sr 211 20.10 56.38 5.07 0.57 495.94
82 Pb 204 6.72 11.89 2.77 0.53 78.94
BC( (jq/m3) 216 1.647 3.235 0.708 0.002 20.028
Mass (jjq/m3) 216 151.142 308.035 57.040 0.160 1794.006
91
Table 4-4: CONCENTRATION OF FINE PM  - ELEMENTS, BC and MASS  
(28th D ecember 2005 -  12 February 2007)
Element/ nq/m3
# of 
Days Mean StDev Median Min Max
Al 83 780.32 1031.55 374.31 49.16 6601.87
Si 177 1370.86 2843.47 374.54 70.27 22310.09
S 214 395.92 292.07 325.43 32.62 1252.09
K 214 329.82 305.84 230.56 17.44 2283.64
Ca 205 119.87 287.92 34.80 7.02 2702.69
Ti 181 32.63 59.55 11.86 2.67 465.19
Mn 185 6.81 10.63 3.24 0.99 84.58
Fe 216 246.27 456.51 96.20 2.07 3909.44
Ni 171 3.24 2.76 2.26 1.09 18.80
Cu 173 4.03 3.88 2.91 0.25 37.22
Zn 204 5.69 5.42 4.59 0.52 45.61
Br 215 5.07 J 3.56 4.13 0.79 17.13
Rb 130 1.38 1.35 1.06 0.18 9.40
Sr 84 2.81 4.19 1.08 0.33 23.49
Pb 169 2.38 1.66 2.00 0.49 10.02
BC (uq/m3) 216 1.646 1.133 1.681 0.010 5.968
Mass (|jq/m3) 216 30.267 48.968 18.029 0.495 430.23
4.3 METEOROLOGICAL DATA
About 100 m from the sampling site, the Ghana A tom ic Energy Commission has a 
mini weather station. The station measures the follow ing parameters; maximum 
relative humidity, solar radiation, precipitation, m inimum and maximum temperature 
with a resolution o f  24 hours (daily). Unfortunately the station did not have the 
capability to measure w ind speed and direction which is critical in determ ining the air 
pollution source directions. Fig. 4-13 shows the monthly variation o f  particulate mass 
with total precipitation.
92
(n
gr
ri3
)
FIGURE 4-11: DA ILY  VARIATION  OF SOME SELECTED  COARSE PM  ELEMENTS DURING  THE
INVESTIGATION  PERIOD
1000000
100000
10000
1000
100
cJp CpA
$  <vx Nt»x ,£> <!> ( $  <$■ $  <$■ rf$ d* &  oN <$ <$ <$ N° T? $
Date Ended
93
rati 
cxi 
(ng
rri3
)
FIGURE 4-12: DA ILY  VARIATION  OF SOME SELECTED  FINE PM  ELEMENTS DURING
THE INVESTIGATION  PERIOD
100000
10000
1000
100
f t  nn ' f t  f t  f t  f t  f t  f t  &  f t  <\X f t  ^  f t  NT> f t  f t  f t  f t  f t  f t  ^  f t  N4  f t  f t  f t  f t  f t  f t 3
Date Ended
94
Ma
ss
 
C
on
ce
nt
ra
tio
n 
((jg
m 
3) 
or
 
R
ai
nf
al
l 
(m
m
)
FIGURE 4-13: MONTHLY VARIATION OF PM  WITH TOTAL PRECIPITATION
300
250
200
150
100
50
0
Month
NB: January 2007 coarse and PM 10 values have been truncated to show the other monthly variations in detail, the real values are 
1302J  and 147L4 fjgm-3 respective
95
CHAPTER 5 
AIR POLLUTION AND RECEPTOR MODELING
5.1 OVERVIEW OF AIR POLLUTION MODELLING
Modeling o f  a ir po llu tion  has becom e a very  im portan t too l fo r regu la to ry  purposes, 
po licy-m aking and research  app lica tions. D ue to the h igh cost o f  a ir po llu tion  m onito ring , 
m easurem ents com bined  w ith  m odeling  have becom e a very  e ffic ien t m ethod  in air 
pollution research . F o r e ffec tive  m anagem en t stra teg ies to  be d eve loped  fo r im proving  
air quality, an unders tand ing  o f  the re la tionsh ip  betw een  po llu tan t sources and their 
impact at a recep to r site is requ ired . H ence the need  to  iden tify  the sources em itting  the 
air pollutants, the am oun t o f  po llu tan ts they  are em itting  and  th e  physica l and  chem ical 
transform ations tak ing  p lace  during  the d ispersion .
Z annetti1’2 classified  a ir po llu tion  m odels accord ing  to  the basic  charac te ristics o f  the 
model. This c lassifica tion  includes:
•  Eulerian  m odels -  so lve  num erica lly  atm ospheric  d iffu sion  equation
•  G aussian  m odels -  in w h ich  the concen tra tion  d is tribu tion  is G aussian  in both the 
horizontal and  vertical d irections
•  L agrangian  m odels — w hich  e ither consider p rocesses in m ov ing  air m ass to 
stim ulate d ispersion  p rocess
•  S em i-em pirical m odels -  w hich  are based on sem i-em p irica l o r sta tistical m ethods 
and seek to analyse  the re la tionsh ips o f  air quality  and a tm ospheric  m easurem ents 
or to fo recast a ir po llu tion  episodes
96
•  R ecep to r m odels -  w h ich  con sid er the observed  concen tra tions a t a  recep to r po in t 
and attem pt to  appo rtion  the con tribu tions from  various sources.
Broadly, there are th ree m ain  a ir quality  m odels, nam ely  d ispersion  m odels, statistical 
models and recep to r m odels.
5.1.1 DISPERSION MODELS
Atmospheric d ispersion  m odels describe  the tu rbu len t d iffu sion  p rocesses in the 
atmosphere and are m ain ly  u sed  fo r a  w ide  varie ty  o f  pu rposes such  a s : '93
•  E stab lish ing  a sou rce-recep to r relationsh ips;
• Evaluating  the con tribu tion  to  concen tra tions from  various sources
•  E stim ating  the d is tribu tion  o f  spatial concen tra tion  and  popu la tion  exposure  to 
pollution
• O ptim izing  em ission  reduc tion  stra teg ies and analyz ing  em ission  scenarios
• Pred icting  the concen tra tions over tim e
•  A nalyzing  the rep resen ta tiv ity  m easurem ent o f  stations
•  As tools fo r research
All the above canno t be accom plished  using only  a ir quality  m easu rem en ts. H ence these 
models require m eteo ro log ica l and geographical in fo rm ation  in add ition  to  source and
emission data. There  are som e lim itations such as inaccuracy  in estim ation  o f  the input
data, deficiencies in m odeling  the physical and chem ical phenom ena . To m in im ize  these, 
the models have to be sub jec ted  to  continual quality  contro l and assu rance  p rocedures. In 
addition, a high quality  da ta  base  and  con tinuous evalua tion  o f  the m odel is needed.
The dispersion m odels are usually  c lassified  as:
97
•  L ocal -  tim e scale less than  a few  m inutes
•  Local to  R eg ional -  several hours
•  R egional to  C on tinen ta l -  several days
•  Continen tal to G lobal -  w eeks o r more
D ispersion m odels on the local scale are usually  used fo r quan tifica tion  o f  the 
concentrations o f  po llu tan ts tha t can cause adverse health  effec ts  and  deposition  o f  a ir 
pollutants and its influence. M odels based  on G aussian  concen tra tion  are w ide ly  used and 
mainly fo r regu la to ry  purpose.
Regional scale m odels are m ain ly  used fo r po licy -m ak ing  or research  pu rposes. They  are 
used for quan tify ing  d eposition  and concen tra tions o f  e lem en ts  such as su lfur, n itrogen  
compound and o ther com pound  such as pho to -ox idan ts (e.g. ozone). T hey  are also used 
for heavy m etal, p e rs is ten t o rgan ic  po llu tan ts (POP) and  rad ioac tive  and  hazardous 
materials air quality  ep isodes.
Continental scale m odels, also  know n as long-range tran spo rt has the sam e aim  and 
purpose as the reg ional scale  m odels. In the con tinen ta l scale  how ever, it is very  
important to  take into accoun t param eters o f  the a tm ospheric  boundary  layer and  the 
relevant w eather cond itions over the m odeling  period.
5.1.2. STATISTICAL MODELS
These types o f  m odels do no t exp lic itly  cover the d ispersion  and chem ical transfo rm ation  
o f  air pollu tants in the atm osphere . S tatistical m odel can be o f  several types, including
98
rapid a ssessm en t and  em pirica l m odels. The m ain aim  o f  these  types o f  m odels is to 
summarise the key  resu lts  from  deta iled  m odels o f  a ir po llu tan ts  in the fo rm  o f  g raphs 
and look-up chart. Som e sta tistica l m odels also  use th e  re la tion sh ip  betw een  m easured  
quality and param eters re la ted  to  w ea ther and em issions. W hen  these  are in tegrated  w ith 
inventory techn iques these  m odels could be u sed  to  study  the fo llow ing  types o f  
situations:
•  The im pact o f  em issions from  an ind iv idual po in t source  on sho rt-term  a ir quality  
at a critical recep to r and /o r at a know n d istance from  the  source
•  The im pact o f  the em issions from  an ind iv idual p o in t source on the long-term  
average air quality  in the v ic in ity
•  The im pact o f  area source em ission , such as road tra ffic  and bush  bu rn ing  related 
em issions, on the long-term  average a ir quality  o f  an u rban  area.
These types o f  m odels requ ire  ex tens ive  database o f  h is to rica l a ir qua lity  m easu rem en t 
and m eteoro log ical data  to  fo rm ula te  the m odel. In the N e th e rla n d s194 such a m odel for 
ozone m onitoring  is in use. The m odel requires the m ax im um  concen tra tion  o f  the 
measuring sites o f  the p rev ious day, statistics from  th e  p as t and  the m axim um  
temperature o f  the p rev ious day  and the fo recast tem pera tu re , bo th  as averages fo r the 
N etherlands to fo recast the ozone level fo r the nex t th ree  days.
5.1.3 RECEPTOR MODELS
Receptor m odels focus on the behav iou r o f  the am bien t env ironm en t at site  (norm ally  
called a recep to r) as opposed  to source-orien ted  m odels w h ich  focus on transport, 
d ilution and tran sfo rm ations from  the source to the sam pling  o r recep to r site. The
99
fundam ental p rinc ip le  o f  recep to r m odeling  is tha t m ass conserva tion  can be assum ed  and 
a mass balance analysis can  be used to  identify  and appo rtion  sources o f  con tam inan ts  in 
the a tm osphere .19S A irbo rne  po llu tan ts in the atm osphere  fo rm  a very  com plex  system , 
hence m athem atical o r sta tistical m ethods are used to iden tify  and appo rtion  the sources.
Chem ical characteristics such  as e lem ental concen tra tion  o r shape o f  partic les in a series 
o f particle sam ples at the recep to r can  be used to reso lve the m ain  sources. To obtain  a 
data set for recep to r analysis, the  norm al approach  is to  analyse  a large num ber o f  
chem ical constituen ts such  as e lem en ta l, o rgan ic  o r gaseous concen tra tion  in th e  air 
samples. E lem ental tracers  are u sed  in m ost recep to r m odeling  bu t e lem en ts  a lone  are not 
always su ffic ien t to  d is tingu ish  em itting  source. For exam ple , w hen  leaded  m oto r fuel 
was in use, lead and b rom ine w ere  the m ain m arkers o f  road traffic  po llu tion , bu t w ith  the 
ban on leaded fuel these  e lem en ts  are d isappearing  as m akers o f  road  traffic  po llu tion . 
Though curren t recep to r m odels are focusing  on chem ical com positions, e lem ental 
concentrations are still be ing  used. The elem ental concen tra tion  if  com b ined  w ith  o ther 
measurements such as BC, m ass and chem ical com positions are be tter in reso lv ing  the 
sources.
Several approaches have been  successfu lly  app lied  to recep to r m od e lin g .196'199 R ecep to r 
models can be d iv ided  into tw o  m ain  groups; chem ical m ass balance  and m ultivariate  
methods. Data from  m easu ring  site and potential sources are used in chem ical mass 
balance m ethod thus allow ing , in princip le , ca lcu la tion  o f  p ropo rtions o f  various known 
sources sam pled  at the recep to r (m easuring) site. The m ultivaria te  m ethods, e.g. target 
transformation fac to r analysis and principal com ponen t analysis w ith  m ultip le  linear
100
regression  analysis, u sua lly  use on ly  chem ical com position  data  to  ascerta in  the num ber 
o f  sources, the chem ical com position  o f  th e ir em issions and th e ir re la tive  con tribu tions to 
the m easured  concen tra tions. B ecause the source appo rtionm en t is based  on statistical 
methods, these m ethods requ ire  a large sam ple dataset, the  m ore the better.
Chem ical mass balance  requ ires  th a t the com position  o f  all th e  con tribu ting  sources be 
known, but th is is no t o ften  the case since it is no t p rac ticab le  since  the em issions are 
difficult to recogn ize  and sam ple. Secondly  the analy tica l techn ique  canno t m easure  all 
the species in the source o r the source com position  fail to  con ta in  all the  species observed  
in the samples. U sing  the m u ltivaria te  m odels these p rob lem s can  be overcom e. It is also 
the only m ethod  availab le  w hen  there  is no source in form ation .
5.2 RECEPTOR MODELING USED IN THIS WORK
Most m ultivariate  recep to r m odels, as in th is w ork , first app ly  p rincipal com ponen t 
analysis (PCA ) to the data  set. PCA  describes a system  by  d e te rm in ing  a m in im um  se t o f  
vectors that span the data  space to  be in terpreted . A  new  set o f  variab les are found  as 
linear com binations o f  the m easured  variab les so th a t the observed  varia tions in the 
system can be rep roduced  by a sm aller num ber o f  these  causal fac to rs. PCA  has been 
w idely used  in stud ies o f  a irbo rne  particu la te  m atter com position  data  since the early  
1980s.200 PCA  is genera lly  availab le  in m ost com pu ter p ackage fo r sta tistica l analysis for 
examples SPSS and NCSS .
The model assum es a linear m odel rela ting  experim en ta l m easu red  variab les and the 
source profile  m atrix
C = P S  ...5 -1
101
where
C is the data  m atrix  o f  d im ension  i,k  and units are ng /m 3
P is the source p ro file  m atrix  o f  d is tinc t sources o f  d im ension  i,j and  un its are ng/|XgS0Urce 
S is the source con tribu tion  m atrix  o f  d im ension  j,k  and  units n g s/m 3
Equation 5-1 can be genera lised  as:
C m ,n  ^ 1  P m , l  S l ,n  5-2
where
Cm.n is the m easured  concen tra tion  o f  variab le  m in obse rva tion  num ber n
p m,l is the m odelled  concen tra tion  o f  variab le  m in source  1
Si,n is the m odelled  con tribu tion  to source 1 in observa tion  num ber n
PCA can be perfo rm ed  on ly  on se t o f  sam ples in w hich th e  v arious sources con tribu te  
different am ounts o f  partic les to  each sam ple; the m ass balance  is a m atrix  equation  o f  
the form;
Z = L F  5 3
where
Z  is sam pling  m atrix
L  is fac to r load ing  m atrix  and
F  is fac to r sco re  m atrix  respectively .
102
In a PCA  analysis, the  data  are norm alised  by sub trac ting  a m ean  value  <Cm>  and 
div id ing by the standard  dev ia tion , <jm
Zm ,n  (C m ,n  <' C m ' > ) / O m  5 - 4
where a row , m , in Z  co rresponds to the au toscaled  values o f  the v a riab le , m , in C .
In this case each standard ised  value  has a m ean value  o f  zero  and a standard  dev ia tion  o f  
1. H ence the values o f  Z  are au to scaled  concen tra tions and L gives the con tribu tions to 
the m easured variab les from  the fac to rs identified. The fac to r scores F  show  the daily  
contributions from  the d iffe ren t fac to rs (sources).
To rescale L and F to  the physica l m ean ingfu l m atrices P and S, a ‘tra c e r’ sam ple w ith 
sample num ber n+1, hav ing  all the variab les set to  zero  is inc luded  in the data  s e t .129,190 
PCA is used to  d eterm ine  the score m atrix  F in w hich  the row s are trea ted  as au toscaled  
values o f  the row  in the source m atrix .
The sample m ass w h ich  w as determ ined  by grav im etric  analysis  is then  used in a m ass 
balance ca lcu lation  to  tran sfo rm  the scaled scores into unsealed  source m atrix . This was 
achieved by reg ression  o f  the tran sfo rm ed  values on the m ass variab le  o f  the coarse. The 
source m atrix  o f  the p articu la te  m atte r (i.e coarse partic les) va lues m ust be related  to  the
experimental m ass value , C csby the relation:
C c s , n  ^ m S m . n  ^ m ( f m , n  “  fm , n + l )  . . . 5 - 5
where Sm,„ and fm,„ are the source, S and autoscaled  variab le  in F  respective ly  for 
variable m in observation  n..
103
The coeffic ien ts a ,„  are found  by reg ression  o f  (fm,n - fm,n+i) on C cs,n- The e lem en ts in  the 
source m atrix  describe  the daily  varia tion  o f  the PM  (coarse  PM  as in th is exam ple) m ass- 
variable o f  the source in n g /m 3. The source pro file  m atrix  is then  ca lcu la ted  from :
P =  CST(SST) 1 . . .5 -6
This was repeated for the fine fraction (PM2.s)-
There exists a  set o f  na tu ra l physica l constra in ts in o rde r to  ob ta in  physica l m ean ingfu l 
results:
•  The pred ic ted  source  com positions m ust be non -nega tive ; a source  canno t have  a 
negative percen tage  o f  an elem ent. H ence all nega tive  values o f  S and L  must be 
truncated  to  zero.
•  The p red ic ted  source  com positions to  the aeroso l m ust all be non -negative ; source 
cannot em it a nega tive  m ass.
•  The sum  o f  the p red ic ted  e lem en ta l m ass con tribu tions fo r each  source m ust be 
less o r equal to  th e  to ta l m easu red  m ass for each elem ent.
The accuracy o f  the m odel is determ ined  by how  good the experim en ta l data  can be 
reproduced by the m odel.
5.3 RESULTS OF  THE  R ECEPTOR  MODEL
The species used  in recep to r m odel are mass, BC, e lem ental concen tra tions. In the PCA  
analysis, several runs w ere  m ade. The num ber o f  fac to rs w as varied  and varim ax  as well 
as prom ax ro ta tions w ere  perfo rm ed . It w as assum ed tha t the po llu tion  sources w ill be 
independent o f  each o th er and because  varim ax  gave the m ost con sis ten t resu lts it w as the
104
rotation th a t w as used  in the fac to r analysis. It w as found  th a t th e  sam e fac to rs appeared  
in the analysis even  i f  som e variab les w ere om itted , a lthough  th e  fac to r load ings fo r the 
d ifferent e lem en ts varied  slightly .
TABLE  5-1: CONCENTRAT ION  OF  COARSE  PM  - E LEM ENTS , BC  AND  
MASS  (28th D ECEM BER  2005 -  31 M ARCH  2006)
Element/
(ng/m3)
# of 
Days Mean StDev Median Min Max
Al 68 1672.25 936.16 1510.35 363.96 7044.23
Si 69 4771.18 2847.20 4449.88 819.19 22834.64
S 68 265.58 137.69 251.98 42.29 684.55
Cl 69 573.55 340.60 511.87 26.75 1861.36
K 69 480.89 243.34 439.50 72.91 1657.06
Ca 69 647.50 415.33 615.47 102.76 3444.44
Ti 69 151.03 79.10 140.13 23.54 574.79
V 51 11.83 4.67 10.91 3.11 24.27
Cr 55 6.29 2.94 6.11 1.79 14.87
Mn 69 27.53 14.69 27.37 4.02 106.16
Fe 69 1377.56 700.99 1277.82 211.83 5177.45
Co 48 8.32 3.09 8.07 2.19 17.35
Ni 26 3.28 1.13 3.15 1.82 7.77
Cu 51 3.18 1.42 3.17 0.50 10.36
Zn 69 8.29 6.86 6.86 1.72 53.94
Br 68 2.90 1.35 2.86 0.70 9.49
Rb 69 2.22 1.15 2.13 0.53 7.77
Sr 69 5.66 3.47 5.65 0.57 28.48
Pb 65 2.80 1.34 2.63 0.53 6.45
BC (|jg/m3) 69 0.57 0.33 0.50 0.01 1.69
Mass (ug/m3) 69 88.66 48.25 89.59 8.64 228.17
When the w hole data  spann ing  14 m onths w ere u sed  in the m odel, the fit betw een  the 
experimental and the m odel w as no t very  good. T h is could  be attribu ted  to  the varying 
sources and source streng th  w hich  are seasonal dependent. The data  w as therefore 
divided into th ree, acco rd ing  to the seasons -  H arm attan  (D ecem ber 2005 -  M arch 2006), 
Raining (April 2006— O ctober 2006) and H arm attan  (N ovem ber 2006  -  February  2007).
105
Tables 5-1 to 5-6 give the sta tistica l analysis for the m easu rem en ts  w hen  the datase t w as 
divided accord ing  to  the c lassif ica tions above. The num ber o f  filte rs  con tribu ting  to the 
data set is g iven as num ber o f  days toge ther w ith  the m ean, standard  dev ia tion  (S td  Dev.), 
median, m inim um  and  m ax im um  values.
TABLE  5-2: CONCENTRAT ION  OF  FINE  PM  - E LEM ENTS , BC  AND  MASS  
(28™  DECEM BER  2005  -  31 M ARCH  2006)
Element/ng/m3
# o f
Days Mean StDev Median Min Max
Al 20 381.51 318.54 301.80 49.16 1352.40
Si 65 448.36 521.82 327.65 117.55 3078.72
S 69 145.95 99.20 116.38 32.62 567.04
K 69 144.04 99.48 116.07 17.44 462.73
Ca 68 46.22 60.68 32.80 8.48 415.62
Ti 56 14.59 15.61 10.90 3.17 98.74
Mn 48 3.54 2.79 2.68 1.14 16.65
Fe 69 112.59 121.84 88.28 31.20 847.32
Ni 24 2.22 0.62 2.13 1.48 4.32
Cu 26 2.42 0.67 2.23 0.25 4.02
Zn 60 2.17 1.32 2.00 0.52 7.46
Br 68 2.00 1.01 1.69 0.79 5.42
Rb 16 0.73 0.37 0.61 0.18 1.72
Sr 15 1.17 0.98 0.97 0.33 3.45
Pb 29 1.29 0.57 1.18 0.51 2.87
BC (ug/m3) 69 0.70 0.66 0.52 0.01 2.96
Mass (|jg/m3) 69 10.59 7.72 7.91 2.03 37.84
106
TABLE 5-3: CONCENTRATION OF COARSE PM  - ELEMENTS, BC AND
MASS (4™ APRIL -  31 OCTOBER 2006)
Element/ng/m3
# of 
Days Mean StDev Median Min Max
Al 90 1128.45 456.63 1052.18 400.98 3185.81
Si 97 3423.89 1530.08 3350.16 118.63 8581.07
S 96 360.40 140.21 339.55 71.09 662.47
Cl 97 1790.63 988.34 1753.71 45.58 4900.17
K 97 418.90 168.14 413.50 12.08 1432.84
Ca 97 482.38 186.97 471.09 23.62 1048.34
Ti 97 94.12 44.18 88.14 4.21 247.06
V 23 5.14 1.13 4.80 3.64 7.43
Cr 79 4.27 2.00 3.73 1.65 13.98
Mn 96 17.34 6.61 16.82 2.63 37.84
Fe 100 1023.49 460.57 1034.55 2.52 2212.74
Ni 100 3.17 0.80 3.16 1.55 6.44
Cu 100 4.48 1.00 4.32 2.52 7.93
Zn 97 8.14 4.03 7.57 1.10 36.47
Se 18 0.50 0.07 0.53 0.27 0.56
Br 100 6.23 1.76 6.12 2.01 12.01
Rb 93 1.44 0.57 1.55 0.49 3.20
Sr 96 4.14 1.41 4.13 0.99 8.90
Pb 93 2.77 1.53 2.52 0.66 8.46
BC (pg/m3) 100 0.67 0.24 0.68 0.04 1.46
Mass (Mg/m3) 100 42.81 16.56 42.58 0.16 87.84
TABLE  5-4: CONCENTRAT ION  OF  F INE  PM  ELEM ENTS , BC  AND  MASS  
(4™ A PR IL  -  31 OCTOBER  2006)
Element/ ng/m3
# of 
Days Mean StDev Median Min Max
Al 27 294.73 184.51 243.82 136.44 1128.73
Si 65 333.84 445.18 222.42 70.27 3578.00
s 98 513.26 275.76 493.77 50.85 1252.09
K 98 293.45 181.06 249.16 21.14 744.58
Ca 90 33.62 51.54 24.11 7.02 494.39
Ti 78 10.07 10.69 7.70 2.67 91.63
Mn 91 3.20 1.97 2.76 0.99 18.12
Fe 100 91.21 109.38 75.58 2.07 1087.95
Ni 100 2.30 0.58 2.14 1.09 4.98
Cu 100 2.94 0.85 2.68 1.71 7.90
r  Zn 97 5.78 3.12 5.52 0.99 25.90
Br 100 6.47 3.51 5.76 1.04 17.13
Rb 69 0.96 0.49 1.01 0.37 2.14
Sr 23 0.67 0.63 0.53 0.37 3.52
Pb 95 2.15 1.15 2.04 0.49 6.87
BC (uq/m3) 100 1.92 0.86 1.90 0.10 4.63
Mass (|jq /m3) 100 17.69 7.71 17.76 0.50 41.14
107
TABLE 5-5: CONCENTRATION OF COARSE PM - ELEMENTS, BC AND
MASS (2nd NOVEMBER 2006 -  15™ FEBRUARY 2007)
Element /ng /m3
# of 
Days Mean StDev Median Min Max
Al 43 12889.96 15874.20 5396.71 937.75 52309.42
Si 47 47291.35 61286.19 19919.45 120.01 233707.50
S 47 1224.69 826.17 991.58 188.41 3425.45
Cl 47 2070.95 1659.78 1438.02 158.36 6981.94
K 47 4839.69 5861.45 2094.16 219.75 24135.81
Ca 47 8C39.03 10977.46 3336.12 342.28 49899.12
Ti 47 1391.15 1743.45 507.47 54.37 7413.49
V 27 56.20 60.51 20.65 3.67 233.23
Cr 37 15.98 14.91 9.47 3.18 70.18
Mn 46 271.54 355.30 101.90 11.34 1644.17
Fe 47 12364.79 15424.33 4490.79 498.41 67143.01
Co 10 42.64 38.58 43.41 4.23 120.46
Ni 47 16.64 17.28 8.49 2.65 64.08
Cu 47 20.77 21.27 10.68 3.78 89.71
Zn 47 55.49 58.17 26.51 4.86 210.55
Br 47 22.63 20.71 12.54 3.78 84.58
Rb 47 24.09 30.42 10.04 1.08 137.12
Sr 46 75.10 104.19 29.89 2.70 495.94
Pb 46 20.23 19.73 10.44 1.62 78.94
BC(|jg/m3) 47 5.30 5.58 2.63 0.63 20.03
Mass (ng/m3) 47 473.37 549.55 192.85 24.35 1794.01
TABLE 5-6: C ONCENTRAT ION  OF  FINE  PM  - ELEM ENTS , BC  AND  MASS  
(2ND NOVEM BER  2006 -  15™ FEBRUARY  2007)
Element/ng/m3
# of 
Days Mean StDev Median Min Max
Al 36 1366.06 1336.78 869.96 320.84 6601.87
Si 47 4080.85 4478.26 2402.82 483.59 22310.09
S 47 518.24 285.93 458.89 153.83 1158.51
K 47 678.39 413.09 609.34 182.37 2283.64
Ca 47 391.58 509.03 220.02 46.18 2702.69
Ti 47 91.54 92.67 53.61 13.27 465.19
Mn 46 17.39 17.15 9.99 3.20 84.58
Fe 47 772.43 751.27 477.03 133.77 3909.44
Ni 47 5.75 4.28 3.65 2.12 18.80
Cu 47 7.22 6.33 5.13 2.29 37.22
Zn 47 10.02 8.45 6.80 2.98 45.61
Br 47 6.53 3.26 5.71 2.11 15.38
Rb 45 2.25 1.93 1.81 0.52 9.40
Sr 46 4.42 5.10 2.33 0.52 23.49
Pb 45 3.57 2.28 2.70 1.19 10.02
BC (uq/m3) 47 2.44 1.27 2.11 0.46 5.97
Mass (Mq/m3) 47 85.91 83.09 51.84 22.09 430.23
108
The PCA analysis value of the Coarse data for 2nd November 2006  — 15th February 2007  
is presented as an example in Tab le  5-7. The PCA score less the “tracer” scores for the 
same set of samples is also presented in Table 5-8. The mean values obtained for the 
mass balance computation were as follows: 400.91 , 199.19, 283 .31 , 81 .07  and 0.36 for 
the factor scores 1 to 5 respectively.
The computed source, S, values greater than zero (S>0) are presented in Tab le  5-9. The 
source profile values are presented in Tab le  5-10.
Fig. 5-1, 5-2, 5-3 and 5-4 show the correlation between the model values and 
experimental data for some selected species of the aerosol (mass, BC, Fe and Cl 
respectively). Fig. 5-5, 5-6, 5-7, and 5-8 show a time series plot of the model and 
experimental data for the selected species above.
The same procedures were repeated for the fine and coarse data for the dataset with 
statistical values given in Tab les 5-1 to 5-6. The gradients, squared correlation 
coefficient -  R2, intercept are given in Tables 5-11 to 5-16.
109
TABLE 5-7: PCA FACTOR SCORES FOR COARSE DATA
(2nd N OVEM BER  2006 -  15th FEBRUARY  2007)
F ilte r# F Scorel F Score2 F Score3 F Score4 F Score5
C248 -0.399 -0.503 -0.373 -0.796 0.001
C249 -0.390 -0.358 -0.452 -0.813 0.103
C250 -0.362 -0.296 -0.403 -0.856 0.288
C251 -0.322 -0.440 -0.301 -0.845 0.485
C252 -0.400 -0.304 -0.427 -0.427 0.250
C254 -0.321 -0.594 -0.002 -0.235 -0.327
C255 -0.269 -0.153 -0.349 -0.291 0.226
C256 -0.337 -0.177 -0.198 0.231 -0.383
C257 -0.204 -0.657 -0.193 0.393 -0.111
C258 -0.458 -0.555 -0.038 0.308 0.261
C260 -0.280 -0.218 0.113 -0.509 0.470
C261 -0.338 -0.341 0.077 -0.344 0.276
C262 -0.207 -0.127 -0.048 -0.264 1.222
C263 -0.465 -0.208 -0.374 -0.163 -0.736
C264 -0.362 -0.273 -0.634 -0.472 0.974
C266 -0.460 -0.447 -0.316 -0.047 -0.469
C267 -0.458 -0.220 -0.396 -0.266 -0.421
C268 -0.523 0.351 -0.585 0.547 0.120
C269 -0.478 0.444 -0.633 1.247 -0.996
C270 -0.525 -0.052 -0.224 0.665 -1.105
C271 -0.586 -0.268 -0.139 0.522 -1.071
C272 -0.453 -0.075 -0.252 0.012 -0.532
C274 -0.462 -0.086 -0.182 -0.200 -0.135
C275 -0.473 -0.126 -0.302 -0.076 -0.205
C276 -0.568 -0.262 -0.271 0.294 -0.775
C277 -0.551 -0.161 -0.405 0.490 -0.702
C278 -0.551 -0.178 -0.366 0.322 -0.262
C279 -0.580 -0.024 -0.489 0.513 -0.233
C281 1.334 -0.107 -1.410 1.173 -0.998
C282 1.643 4.124 -0.388 -3.573 -0.464
C283 5.300 -1.215 -2.956 1.103 0.720
C284 2.238 -2.761 3.266 -0.814 -0.691
C285 1.497 1.276 2.094 -0.733 -0.492
C286 0.550 1.770 1.291 1.891 -1.038
C287 0.277 0.410 1.863 0.439 -1.798
C288 -0.268 1.791 0.314 3.323 2.262
C289 0.111 -0.958 1.562 1.303 -0.798
C290 0.011 1.420 0.893 0.774 3.650
C291 0.135 1.584 0.383 1.363 -0.359
C292 0.505 -1.081 2.721 -0.490 2.407
C293 0.659 1.567 0.884 -0.080 -1.431
C294 -0.256 -0.099 -0.196 -0.750 1.170
C296 -0.269 0.133 -0.457 -1.158 0.590
C297 -0.336 -0.182 -0.289 -0.595 0.617
C298 -0.401 -0.153 -0.385 -0.410 -0.119
C299 -0.229 -0.541 -0.334 -0.580 1.213
C300 -0.308 -0.138 -0.258 -0.118 -0.358
COOO -0.410 -0.533 -0.436 -1.008 -0.299
StDev 1 1 1 1 1
Mean -6.6x10 '17 -4.7x10"16 -3.7x10"16 -8.6x10~16 -3x1 O'16
TABLE 5-8: PCA FACTOR SCORES LESS TRACER VALUE (C000) FOR
COARSE DATA  (2nd NOVEMBER 2006 -  15™ FEBRUARY 2007)
Filter # FS1-Zero FS2-Zero FS3-Zero FS4-Zero FS5-Zero
C248 0.011 0.030 0.063 0.213 0.300
C249 0.021 0.175 -0.017 0.196 0.402
C250 0.049 0.237 0.033 0.152 0.587
C251 0.088 L 0.093 0.134 0.164 0.784
C252 0.010 0.229 0.009 0.581 0.548
C254 0.089 -0.061 0.434 0.773 -0.028
C255 0.142 0.380 0.087 0.717 0.525
C256 0.074 0.356 0.238 1.239 -0.085
C257 0.207 -0.124 0.242 1.401 0.187
C258 -0.048 -0.022 0.398 1.316 0.560
C260 0.130 0.316 0.549 0.500 0.769
C261 0.072 0.193 0.513 0.664 0.574
C262 0.203 0.407 0.387 0.744 1.520
C263 -0.054 0.325 0.061 0.845 -0.437
C264 0.048 0.260 -0.199 0.536 1.273
C266 -0.049 0.086 0.120 0.961 -0.170
C267 -0.048 0.313 0.040 0.742 -0.123
C268 -0.112 0.884 -0.149 1.555 0.419
C269 -0.068 0.977 -0.197 2.256 -0.698
C270 -0.114 0.481 0.211 1.673 -0.806
C271 -0.176 0.265 0.296 1.530 -0.772
C272 -0.043 0.458 0.183 1.021 -0.233
C274 -0.052 0.447 0.253 0.808 0.164
C275 -0.063 0.408 0.133 0.932 0.093
C276 -0.157 0.271 0.165 1.302 -0.477
C277 -0.141 0.372 0.031 1.498 -0.403
C278 -0.141 0.355 0.069 1.330 0.037
C279 -0.170 0.510 -0.054 1.522 0.066
C281 1.745 0.426 -0.974 2.182 -0.699
C282 2.053 4.657 0.047 -2.564 -0.165
C283 5.711 -0.682 -2.521 2.111 1.019
0284 2.649 -2.228 3.702 0.195 -0.393
C285 1.907 1.809 2.529 0.275 -0.193
C286 0.960 2.303 1.727 2.899 -0.739
C287 0.688 0.943 2.299 1.447 -1.500
C288 0.143 2.324 0.749 4.331 2.561
C289 0.522 -0.425 1.998 2.311 -0.499
C290 0.421 1.953 1.329 1.782 3.948
C291 0.545 2.117 0.818 2.371 -0.061
C292 0.915 -0.548 3.157 0.518 2.706
C293 1.069 2.101 1.320 0.929 -1.132
C294 0.155 0.434 0.240 0.258 1.469
C296 0.142 0.666 -0.022 -0.150 0.888
C297 0.075 0.351 0.146 0.413 0.916
C298 0.010 0.381 0.050 0.598 0.179
C299 0.182 -0.008 0.101 0.429 1.512
0300 0.102 0.396 0.178 0.890 -0.059
cooo 0.000 0.000 0.000 0.000 0.000
111
TABLE 5-9: COMPUTED SOURCE CONTRIBUTION TO THE VARIOUS
COARSE FILTER  MASS (2nd NOVEMBER 2006 -  15th FEBRUARY 2007)
Filter # Source 1 Source 2 Source 3 Source 4 Source 5
C248 4.57 5.99 17.71 17.23 0.11
C249 8.27 34.94 0.00 15.87 0.15
C250 19.62 47.19 9.34 12.34 0.21
C251 35.42 18.60 38.03 13.27 0.29
C252 4.17 45.66 2.44 47.11 0.20
C254 35.68 0.00 122.87 62.70 0.00
C255 56.82 75.71 24.63 58.12 0.19
C256 29.55 70.89 67.31 100.44 0.00
C257 82.83 0.00 u 68.59 113.56 0.07
C258 0.00 0.00 112.72 106.71 0.20
C260 52.26 62.84 155.54 40.51 0.28
C261 28.91 38.35 145.26 53.84 0.21
C262 81.48 80.95 109.74 60.34 0.55
C263 0.00 64.69 17.36 68.50 0.00
C264 19.29 51.83 0.00 43.47 0.46
C266 0.00 17.21 33.98 77.93 0.00
C267 0.00 62.38 11.25 60.16 0.00
C268 0.00 175.98 0.00 126.09 0.15
C269 0.00 194.62 0.00 182.85 0.00
C270 0.00 95.793 59.83 135.63 0.00
C271 0.00 52.76 83.92 124.04 0.00
C272 0.00 91.29 51.91 82.74 0.00
C274 0.00 89.10 71.82 65.53 0.06
C275 0.00 81.15 37.80 75.55 0.03
C276 0.00 53.93 46.74 105.54 0.00
C277 0.00 74.02 8.69 121.45 0.00
C278 0.00 70.65 19.65 107.83 0.01
C279 0.00 101.48 0.00 123.35 0.02
C281 699.56 84.78 0.00 176.86 0.00
C282 823.23 927.26 13.42 0.00 0.00
C283 2289.41 0.00 0.00 171.13 0.37
C284 1061.92 0.00 1048.81 15.79 0.00
C285 764.72 360.29 716.63 22.33 0.00
C286 384.87 458.60 489.24 235.01 0.00
C287 275.66 187.76 651.27 117.34 0.00
C288 57.16 462.72 212.31 351.12 0.93
C289 209.15 0.00 566.03 187.36 0.00
C290 168.85 388.89 376.47 144.50 1.43
C291 218.52 421.63 231.87 192.24 0.00
C292 366.86 0.00 894.42 42.01 0.98
C293 428.63 418.27 374.00 75.29 0.00
C294 61.97 86.35 67.95 20.94 0.53
C296 56.76 132.65 0.00 0.00 0.32
C297 29.94 69.92 41.44 33.47 0.33
C298 3.97 75.77 14.26 48.52 0.07
C299 72.86 0.00 28.71 34.74 0.55
C300 40.89 78.77 50.32 72.18 0.00
COOO 0.00 0.00 0.00 0.00 0.00
Mean 180.99 123.16 147.80 86.32 0.18
112
TABLE 5-10: COMPUTED SOURCE PROFILE VALUES FOR COARSE DATA
(2nd NOVEMBER 2006 -  15™ FEBRUARY 2007)
Element
(ng/m3) Profile 1 Profile 2 Profile 3 Profile 4 Profile 5
Al 59.51 31.30 9.13 2046.65
Si 4.16 177.91 186.71
s 1.22 1.95 6.90 16.62
Cl 2.23 3.95 2.77 4.13 989.840
K 9.84 9.67 8.01 9.55
Ca 20.90 10.95 13.53 10.94 722.27
Ti 2.93 2.65 2.18 3.32
V 0.10 0.08 0.04 0.08
Cr 0.08 0.02
Mn 0.67 0.49 0.34 0.47 13.02
Fe 26.63 23.43 18.21 28.95
Ni 0.02 0.02 0.03 0.04 8.56
Cu 0.03 0.03 0.02 0.07 14.44
Zn 0.07 0.18 0.04 0.17 9.95
Br 0.03 0.04 0.03 0.06 9.95
Rb 0.06 0.05 0.03 0.05
Sr 0.21 0.09 0.109 0.120 6.35
Pb 0.02 0.05 0.02 0.10
BC (tag/m3) 0.01 0.01 0.02
Mass((jg/m3) 0.64 1.21 1.09 0.69 0.66
113
C
oa
rs
e 
M
od
el
 M
as
s 
(M
g/
rf
i)
FIGURE 5-1: COARSE PM MASS -  MODEL COMPARED WITH EXPERIMENTAL
Coarse Mass - Model vrs Expt
Coarse Expt Mass (|jg/m3)
114
Co
ar
se
 
M
od
el
 
BC
 
(p
g/m
 
)
Coarse BC - Model vrs Expt
FIGURE 5-2: COARSE BC -  MODEL COMPARED WITH EXPERIMENTAL
Coarse Expt BC (|jg/m3)
115
C
oa
rs
e 
M
od
el
 F
e 
(n
g
/m
3)
FIGURE 5-3: COARSE Fe -  MODEL COMPARED WITH EXPERIMENTAL
Coarse Fe - Model vrs Expt
Coarse Expt Fe (ng/m3)
116
Co
ar
se
 
M
od
el
 C
l 
(n
g/
m
3)
Coarse Cl - Model vrs Expt
FIGURE 5-4: COARSE Cl -  MODEL COMPARED WITH EXPERIMENTAL
Coarse Expt Cl (ng/m3)
117
M
as
s 
C
o
n
ce
n
tr
at
io
n
 
(|
jg
m
FIGURE 5-5: COARSE PM MASS -  COMPARISON OF MODEL AND EXPERIMENTAL
2000
1800
1600
1400
1200
1000
800
600
400
200
0
■  Expt Mass
■  Model Mass
-
r -  III A n ... u  _ . ■ |
„i n m m  ini l i L U I ____ ,___ 1 1  1 l l
&
cT  a>n N° V  o ' kV  k\Q ^  ^$ o r sS' K^x rfy cs* c&
f§> &
o N' c?
<v
Date  Ended
118
M
as
s 
C
on
ce
nt
ra
ti
on
 
(n
g
m
FIGURE 5-6: COARSE BC -  COMPARISON OF MODEL AND EXPERIMENTAL
25
20
15
10
I Expt BC 
I Model BC
II h i
^  ^  / •  ^  / ■  / •  ^  ^  ^  ^  / •  / •
D a t e  E n d e d
119
M a s s  C o n c e n t r a t i o n  ( n g /m 3)
ro CO cn CD 00o o o o o o o oo o o o o o o oo o o o o o o oo o o o o o o o
02/11/2006
09/11/2006
16/11/2006
23/11/2006
30/11/2006
07/12/2006
14/12/2006
o
“  21/12/2006 
<0
m
3
O. 28/12/2006
<D
Q.
04/01/2007
11/01/2007
18/01/2007
25/01/2007
01/02/2007
08/02/2007
15/02/2007
M
as
s 
C
on
ce
nt
ra
ti
on
 
(n
gm
FIGURE 5-8: COARSE Cl -  COMPARISON OF MODEL AND EXPERIMENTAL
D a te  E n d e d
121
TABLE 5-11: COARSE PM  CONCENTRATION MODEL AND
EXPERIMENTAL DATA FIT -  ELEMENTS, BC AND MASS
(28™ DECEMBER 2005 -  31 MARCH 2006)
Zero Fit Non Zero Fit
Element Gradient R2 Gradient Intercept R2
Al 0.99 0.95 0.94 89.96 0.95
Si 0.99 0.96 0.92 380.29 0.96
S 0.99 0.94 0.94 15.84 0.94
Cl 0.98 0.88 0.85 101.01 0.91
K 0.99 0.97 0.97 13.27 0.97
Ca 0.96 0.83 0.79 154.37 0.89
Ti 0.99 0.95 0.97 3.54 0.95
V 0.98 0.86 0.92 0.75 0.86
Cr 0.93 0.84 0.96 -0.18 0.84
Mn 0.99 0.96 0.94 1.64 0.97
Fe 0.99 0.97 0.96 52.37 0.97
Zn 0.98 0.95 0.95 0.43 0.95
Br 1.00 0.89 1.10 -0.36 0.90
Rb 0.99 0.94 0.93 0.15 0.94
Sr 0.97 0.90 0.85 0.97 0.92
Pb 0.96 0.79 0.83 0.45 0.82
BC 0.97 0.88 0.91 0.05 0.88
Mass 0.97 0.96 1.04 -5.53 0.96
TABLE 5-12: F INE  PM  CONCENTRAT ION  M ODEL  AND  EXPER IM ENTAL  
DATA  F IT  ELEM ENTS , BC  AND  M ASS  
(28th D ECEM BER  2005 -  31 M ARCH  2006)
Element
Zero Fit N on Zero Fit
Gradient R2 Gradient Intercept R2
Si 1.00 0.94 0.98 27.83 0.94
s 0.99 0.94 0.98 2.66 0.94
K 0.95 0.95 0.92 5.88 0.95
Ca 0.98 0.93 0.96 2.49 0.93
Ti 0.98 0.89 0.96 0.75 0.89
Mn 0.97 0.84 0.96 0.07 0.84
Fe 0.99 0.90 0.97 5.16 0.90
Zn 1.01 0.88 1.06 -0.15 0.88
Br 0.97 0.77 1.11 -0.33 0.78
BC 0.95 0.89 0.81 0.19 0.95
Mass 0.99 0.98 1.00 -0.11 0.98
122
TABLE 5-13: COARSE PM  CONCENTRATION MODEL AND
EXPERIMENTAL DATA FIT -  ELEMENTS, BC AND MASS
(4th APRIL -  31 OCTOBER 2006)
Zero Fit Non Zero Fit
Element Gradient R2 Gradient Intercept R2
Al 0.97 0.83 0.92 65.05 0.84
Si 0.99 0.94 0.95 150.47 0.94
S 1.02 0.92 0.92 37.60 0.93
Cl 1.10 0.78 0.86 549.65 0.87
K 1.01 0.85 0.93 35.20 0.86
Ca 0.99 0.90 0.90 45.52 0.91
Ti 0.99 0.94 0.96 3.54 0.94
V 0.94 0.47 0.76 0.96 0.50
Cr 0.95 0.34 0.60 1.85 0.61
Mn 0.98 0.88 0.92 1.20 0.89
Fe 0.98 0.89 0.87 120.73 0.91
Ni 0.99 0.79 0.85 0.46 0.81
Cu 0.99 0.82 0.92 0.36 0.83
Zn 0.97 0.78 0.78 1.87 0.83
Br 0.99 0.83 0.86 0.84 0.85
Rb 0.97 0.74 0.80 0.30 0.79
Sr 0.99 0.90 0.94 0.24 0.90
Pb 1.03 0.80 0.83 0.73 0.87
BC 0.98 0.80 0.86 0.09 0.82
Mass 0.99 0.92 0.94 2.45 0.93
TABLE 5-14: F INE  PM  CONCENTRAT ION  M ODEL  AND  EXPER IM ENTAL  
DATA  FIT  -  ELEM ENTS , BC  AND  MASS (4th A PR IL  -  31 O CTOBER  2006)
Element
Zero Fit Non Zero Fit
Gradient R2 Gradient Intercept R2
Al 0.85 0.90 1.05 -82.50 0.95
Si 0.98 0.98 0.99 -11.55 0.98
S 1.00 0.91 0.92 50.12 0.92
K 1.02 0.95 0.94 34.28 0.96
Ca 1.04 0.94 0.95 9.13 0.96
Ti 0.97 0.96 0.99 -0.58 0.96
Mn 0.96 0.91 0.95 0.04 0.91
Fe 0.99 0.98 0.97 3.04 0.98
Ni 1.00 0.97 0.97 0.07 0.97
Cu 1.00 0.99 1.00 -0.01 0.99
Zn 1.01 0.81 0.84 1.32 0.86
Br 0.98 0.88 0.89 0.81 0.89
Rb 1.07 0.74 0.83 0.29 0.82
Sr 0.89 0.96 0.97 -0.10 0.97
Pb 0.97 0.87 0.90 0.19 0.87
BC 0.98 0.88 0.92 0.13 0.89
Mass 0.96 0.76 0.79 3.48 0.81
123
TABLE 5-15: COARSE PM CONCENTRATION MODEL AND
EXPERIMENTAL DATA FIT -  ELEMENT, BC AND MASS
(2nd NOVEMBER 2006 -  15™ FEBRUARY 2007)
Element
Zero Fit Non Zero Fit
Gradient R2 Gradient Intercept R2
Al 0.95 0.94 0.88 2196.10 0.96
Si 1.02 0.97 0.97 5237.70 0.98
S 0.9b 0.95 1.06 -130.75 0.96
Cl 0.92 0.83 0.96 -121.27 0.83
K 1.00 0.99 0.97 328.59 0.99
Ca 0.99 0.99 0.97 458.77 0.99
Ti 0.99 0.98 0.95 161.52 0.99
V 0.99 0.96 0.99 -0.88 0.96
Cr 0.99 0.97 1.04 -1.43 0.98
Mn 0.99 0.99 0.97 18.83 0.99
Fe 1.00 0.98 0.96 1230.70 0.99
Ni 0.99 0.97 0.95 1.22 0.98
Cu 0.97 0.94 0.90 2.88 0.95
Zn 0.98 0.95 0.91 7.65 0.96
Br 0.97 0.94 0.98 -0.30 0.94
Rb 1.00 0.99 0.97 1.75 0.99
Sr 0.99 0.99 0.97 5.03 0.99
Pb 0.97 0.92 0.87 3.97 0.94
BC 0.98 0.92 0.91 0.82 0.93
Mass 1.00 0.99 0.96 37.33 0.99
TABLE 5-16: F INE  PM  CONCENTRAT ION  M ODEL  AND  EXPER IM ENTAL  
DATA  FIT  -  ELEM ENT , BC  AND  MASS (2nd N OVEM BER  2006  -  15™
FEBRUARY  2007)
Element
Zero Fit Non Zero Fit
G radient R2 Gradient Intercept R2
Al 1.01 0.97 1.04 -80.77 0.97
Si _j 0.99 0.99 0.98 113.25 0.99
S 0.98 0.93 0.95 22.85 0.93
K 0.98 0.94 0.95 34.33 0.94
Ca 1.02 0.94 0.97 49.50 0.95
Ti 0.99 0.97 0.97 3.46 0.98
Mn 1.00 0.98 0.99 0.14 0.98
Fe 0.99 0.98 0.97 26.17 0.98
Ni 0.98 0.93 0.91 0.56 0.93
Cu 0.97 0.98 0.98 -0.04 0.98
Zn 1.08 0.94 0.99 1.40 0.95
Br 1.00 0.97 1.01 0.14 0.97
Rb 0.97 0.97 0.96 0.05 0.97
Sr 1.02 0.98 1.01 0.13 0.98
Pb 1.04 0.73 0.86 0.90 0.78
BC 0.99 0.96 0.95 0.12 0.96
Mass 1.00 1.00 1.00 0.31 1.00
124
CHAPTER  6
QUALITY  CONTROL  AND  D ISCUSSION  OF  RESULTS
6.1 QUALITY  A SSURANCE  AND  QUALITY  CONTROL  AS  A PPL IED  TO  
AEROSOL ANALYSIS
Modern societies rely on measurements to function, hence the need for sound, accurate 
and reliable measurements cannot be overstated.201 These measurements inform decisions 
on important matters such as air pollution, forensic and criminal investigations, 
pharmaceutical and narcotic issues, public surveys, etc. It is very essential that the user of 
the results is confident and assured that the data are truly representative of the sample and 
that the results are defendable, traceable and can be trusted by other laboratories or 
analysts. The social and economic impact of “wrong” result is too enormous to be left to 
chance. For example, in environmental monitoring, under reporting of values could lead 
to serious health hazards while over reporting could also lead to identification of unreal 
hazards.
Analysis of aerosol involves the use of different analytical techniques and different 
measuring instruments. Much effort has been put into the development of analytical 
techniques and instruments aimed at improving the sensitivity, selectivity and reliability 
of operation. Hence these techniques and instruments over the years have demonstrated 
their ability to yield reproducible results of repeated measurements. The detection limits 
of most of the analytical techniques have improved. This is mainly due to improved 
electronics, which are usually the cause of high signal to noise-ratio. But all these are still 
not free from the bias (systematic errors) of the analyst undertaking the analysis. In 
addition, sampling, sample preparation (where applicable) and analysis of limited 
quantity materials usually affect the quality of trace element analysis. It is therefore
125
necessary that the entire analytical procedure has to be taken into account to be able to 
estimate the reliability of the method. Hence there is the need to develop guidelines to 
ensure that the results are within control. These guidelines should ensure that:
• Measurements are accurate, precise and credible
• Data are representative of ambient conditions
• Results are comparable and traceable
• Measurements are consistent over time
• High and evenly distributed data capture are ensured
Quality assurance and control (QA/QC) are very important in any analytical or 
monitoring program such as aerosol analysis. These activities will ensure that the 
measurements meet the defined objective of the project. It must however be noted that the 
function of QA/QC is not to achieve the highest possible data quality.202'203 This is not 
attainable because of resource constraints but it is a set of activities that will ensure that 
the data generated meet the specific objectives of the project.
Quality assurance activities cover all activities undertaken before sampling including the 
thesis objectives and data quality objectives, site selection, equipment evaluation and 
operation. Quality assurance also addresses the process-related functions such as project 
design, project management and process audit.
Quality control activities cover the direct measurement-related activities such as 
operation, calibration, data management and field audits. This includes ensuring the 
integrity of output-related measurement activities among others, sample site operations,
126
maintenance o f  sam plers and  o ther equ ipm ents, ca lib ra tion , sam ple  analysis, data 
collation and reporting .
Fig. 6-1 show s the stages in the aeroso l analysis used in th is thes is. T hese  stages can  be 
grouped into three m ain  c lasses -  sam pling , analy tical p rocess  and  data  evalua tion . Each 
main class had at least a quality  con tro l activ ity  to check  the va lue  being  generated .
W eigh ing  o f  F ilters
Sampling
W eigh ing  o f  B lanks 
and  Loaded  F ilters
 i r_______
BC  M easu rem en t
 A___
EDXRF Measurement
Data
Evaluation
FIGURE 6-1: STAGES IN THE AEROSOL ANALYSIS USED IN THIS THESIS WORK
It is the successfu l im plem en ta tion  o f  each com ponen t o f  the m easu rem en t chain  that 
ensures the success o f  th is w ork .
D ata  A nalysis
R ecep to r Model 
A nalysis
Analytical
Process
Sam pling  w ith  (JEN 1 
S tacked  F ilte r C assette
127
Sampling:
•  Site se lec tion  -  In th is w ork  the ob jective w as no t to study  any  m a jo r source but 
the genera l am bien t air. H ence the sam pling  site w as se lec ted  such th a t it is not 
near any m ajo r source  and  the aerodynam ic c lea rance  w as good  and no sheltering  
o f  the sam pler by any  ob ject.
•  The consistency  o f  the m icrobalance and the hum id ity  o f  the w eigh ing  
env ironm ent w ere  checked  by w eigh ing  a b lank  filte r a fte r every  10th filter 
w eigh ing  th roughou t the w eigh ing  period.
•  The m icrobalance  used undergoes period ic  rev iew  and ca lib ra tion  by SW EDAC  -  
The Sw edish  B oard  o f  A ccred ita tion  and C onfo rm ity  A ssessm en t.
•  M ore than  2 stacked  filte r casse ttes w ith  covers are used  in the sam pling , hence 
filters are loaded and rem oved  in a clean laboratory  env ironm en t instead o f  at the 
sam pling site.
• G ent sam pler flow  rate  is checked  at least 3 tim es during  the 24 hours sam pling  
period fo r any  p ressu re  drop.
•  The G en t flow  rate  is again  checked  after sam pling  by  p lo tting  o f  vo lum e o f  air 
sam pled aga in s t sam pling  tim e as shown in Fig. 6-2. The corre la tion  coeffic ien t 
(R2=  0 .91) and the g rad ien t o f  the slope (1 .03) show  tha t co rre la tion  betw een the 
volum e o f  a ir sam pled  and sam pling  tim e is very  good.
•  B lank filte r m ass, w h ich  fo llow ed  the sam e analy tical p rocedu re  as the loaded 
filters, are a lso  used  as a check  on the m icrobalance , the filte r m edia  and the 
sam pling  p rocess  as shown in Fig. 6-3. The very  good  ag reem en t betw een  the
128
Vo
lu
m
e 
of 
Ai
r 
Sa
m
pl
ed
 
(n
3)
m asses im plies tha t the m icrobalance  perfo rm ance w as excellen t. In  add ition , it 
show s th a t the sam pling  p rocess w as free from  con tam ina tion .
FIGURE 6-2: VAR IAT ION  OF  VOLUME  OF  A IR  SAM PLED  W ITH  SAM PL ING  
TIM E  OF THE  GENT  SAM PLER
Sampling Time (hrs)
Analytical Process:
•  In the BC analysis, m easurem en ts are repeated  4 tim es at d iffe ren t parts o f  the 
filter to  check  fo r th e  hom ogeneity  o f  the filters and the average  used in the 
analysis (T ab le  6-1). The very  sm all d ifferences in the va lues show ed tha t the 
aerosol deposited  on the filte rs are hom ogeneous.
• S ince EDXRF  is a  non-destructive  techn ique, som e selected  filters have their 
m easurem en ts repea ted  to verify  the repeatab ility  o f  the e lem en ta l analysis and 
corre la tion  be tw een  re la ted  m easurem en ts are checked.
•  EDXRF analysis o f  b lank  filters to check  for concen tra tions in fluenced by either 
the filte r m edia  and /o r signals from  the EDXRF spectrom etry .
129
FI
NA
L 
BL
AN
K 
C
O
AR
SE
 
M
AS
S 
(p
g
• Analysis of Standard Reference Material (SRM), which is closely matched in 
major and trace elements, to check analytical procedure as shown in Tab le  4-1.
FIGURE 6-3: COM PAR ISON  OF  IN IT IAL  BLANK  COARSE  F ILTER  MASS  
W ITH  F INAL  BLANK  COARSE  M ASS
INITIAL BLANK COARSE FILTER MASS (Mg)
Data Evaluation:
• The data are routinely checked for internal consistency. For example, the BC and 
mass are independently measured in both fractions. The BC should always be less 
than the loaded mass before the BC value could be meaningful. The BC values are 
checked against the respective mass.
• Inter-element relationships of various elements which originate from a common 
source are checked as shown in Fig. 4-10 (where there is a very good correlation 
between elements like Al, Si, Mn. Fe, etc.)
130
•  The physica l m odel constra in ts are checked w ith  the data  genera ted  i.e. the factor 
load ing  scores from  the PCA  should  have m ean  =0 and standard  d ev ia tio n= l (an 
exam ple is show n in Tab le  5-7).
• The m odel de te rm ined  - m ass, e lem ental concen tra tion s, BC  and the 
experim en ta lly  d eterm ined  - m ass, e lem ental concen tra tions and  BC are com pared  
as shown in Fig. 5-1 to 5 -8  and  Tables 5-11 to 5-15.
TABLE  6-1: SOM E  OUTPUT  VOLTAGE  M EASUREM ENT  FOR  BLACK  
CARBON  DETERM INATION
Filters Voltages Aver. Voltage
White 8.0 8.0 8.0 8.0 8.00
Black 0.4 0.4 0.4 0.4 0.40
C101 7.0 6.8 7.0 7.0 6.95
C102 7.0 6.8 6.9 6.9 6.90
C103 6.6 6.4 6.5 6.5 6.50
C104 6.8 6.6 6.7 6.8 6.73
C105 6.9 6.7 6.8 6.8 6.80
C106 7.3 6.9 7.1 7.1 7.10
C107 7.5 7.3 7.4 7.4 7.40
C108 7.5 7.4 7.5 7.5 7.48
C109 7.3 7.0 7.1 7.1 7.13
C110 7.2 7.0 7.0 7.1 7.08
c m 7.3 7.0 7.1 7.1 7.13
C112 6.6 6.4 6.5 6.6 6.53
C113 8.0 7.9 8.0 8.0 7.98
C114 6.4 6.2 6.3 6.4 6.33
C115 6.3 6.1 6.2 6.2 6.20
C116 7.8 7.5 7.7 7.7 7.68
C117 6.2 6.1 6.2 6.2 6.18
C118 6.4 6.3 6.3 6.4 6.35
C119 7.1 6.9 7.0 7.0 7.00
C120 6.0 5.9 6.0 6.0 5.98
6.2.1 MASS CONCENTRAT ION
The sampling period (28th December 2005 -  12th February 2007) included two harmattan 
seasons and a rainy season. The 2006/07 harmattan was very severe and the 
Meteorological Services Department of Ghana reported it to be the most severe harmattan 
in about 40 years. This was evident in the mass concentration of the air particulate matter 
(APM) deposited on the filters. The total mass of particulate matter collected on the 
filters varied from day to day as shown in Fig. 4-4 , 4-5  and 4-6. The PM]0 data is mainly 
influenced by the coarse mass concentration as this is more than three times the fine mass 
concentration in all cases.
6.2.1.1 COARSE  M ASS  CONCENTRAT ION
The data of the coarse mass concentration is given in Tab le  6-2. The mean mass 
concentrations of APM for the 2005/06 and 2006/07 Harmattan seasons were 88.66 and 
473.37 |igm‘3 respectively. These were higher than the 2006 rainy season of 42.81 ngm'3, 
which was expected since Harmattan dust is absent during the rainy season and the 
washout effect of the rains are also pronounced. There were still days which had very low 
mass concentration in the Harmattan seasons than the rainy season as shown in the 
Minimum (Min) mass concentration values. Some days in the rainy season also had very 
high values. This could be due to a long period of stable weather conditions.
The median values for the 2005/06 Harmattan and the 2006 Rainy seasons showed that 
the data was evenly spread around the mean. This is further confirmed by the relatively 
small standard deviation of the 2005/06 Harmattan and 2006 Rainy seasons. The data
6.2 DISCUSSION OF RESULTS
132
spread for the 2006/07 Harmattan shows that there were some few days with extreme 
high mass concentration which influenced the mean as shown by the standard deviation 
and median.
TABLE  6-2: COARSE  M ASS CONCENTRAT ION  (ngm  3)
Period No. o f  Days M ean StDev M edian M in M ax
2005/06 Harmattan 69 88.66 48.25 89.59 8.64 228.17
2006 Rainy 100 42.81 16.56 42.58 0.16 87.84
2006/07 Harmattan 47 473.37 549.55 192.85 24.35 1794.01
All Data 216 151.14 308.04 57.04 0.16 1794.01
6.2.1.2 F INE  MASS CONCENTRAT ION
The fine mass concentration data is given in Tab le  6-3. The mean fine mass 
concentration for the 2005/06 Harmattan, 2006 Rainy, and 2006/07 Harmattan were 
10.59, 17.69 and 85.91 Hgm'3 respectively. The 2005/06 Harmattan fine mass 
concentration was less than the 2006 Rainy season which was expected since the 
harmattan season is usually characterized by high coarse mass concentration. The 
2006/07 Harmattan values were very high (about 800% and 500% of the 2005/06 
Harmattan and 2006 Rainy seasons respectively). As stated above this was a very 
abnormal Harmattan season.
TABLE 6-3: F INE  M ASS  CONCENTRAT ION  (ngm  3)
Period No. o f  Days M ean StDev M edian M in M ax
2005/06 Harmattan 69 10.59 7.72 7.91 2.03 37.84
2006 Rainy 100 17.69 7.71 17.76 0.50 41.14
2006/07 Harmattan 47 85.91 83.09 51.84 22.09 430.23
All Data 216 30.27 48.97 18.03 0.50 430.23
The median and standard deviation values for the 2006 Rainy seasons showed that the 
data was evenly spread around the mean. The influence of high extreme values for the
133
Harmattan seasons were more pronounced in the 2006/07 Harmattan than the 2005/06 
Harmattan.
6.2.1.3 COM PAR ISON  OF  TH IS  W ORK  W ITH  SOME  SELECTED  W ORKS  
AND INTERNAT IONAL  STANDARDS
Comparing the results from this work with work done by other researchers is presented in 
Table 6-4. The results show that the coarse fraction is very high but the fine fraction 
which contains most of the respirable particulate matter is low except for the 2006/07 
Harmattan season where there were few days with extreme high values affecting the 
average for the period. The fine fraction values from most of the developed countries 
were higher than the coarse fraction showing that their air particulate matter (APM) is 
mainly from anthropogenic sources.
TABLE 6-4: COM PAR ISON  OF  PM  W ITH  RESULTS  FROM  L ITERATURE
Place
Concentration (Mg/m3) Reference
Fine Coarse
Skukuza (South Africa) 9.4 16.2 Maenhaut2M
Prestoriaskop (South Africa) 12.3 19.4
Palmer (South Africa) 18.0 15.2
Watertown, Boston (USA) 17.4 8.6 Chan2"5
Long Beach, Califonia (USA) 48.6 22.5
Tapada du Quterio (Portugal) 18.5 13.1 Alves2"6
Calgary (Canada) 11.1 26.3 Cheng2"7
Sofia (Bulgaria) 44.0 38.0 Houthuijs2”*
Ostrava (Czech) 60.0 29.0
Dorog (Hungary) 42.0 25.0
Swietochlowice (Poland) 58.0 22.0
Brastislava (Solvak) 36.0 13.0
Serowe (Botswana) 10.1 18.4 Moloi131’
Goteborg (Sweden) 7.0 7.2
Vienna (Austria) 18.6 7.9 Hauck19
Streithofen (Austria) 15.0 6.1
Linz (Austria) 18.8 11.1
Graz (Austria) 21.1 9.9
Kwabenya (Ghana) -  All Data 30.3 151.1 This work
Kwabenya (Ghana) -  2005/06 Harmattan 10.6 88.7
Kwabenya (Ghana) -  2006 Rainy Season 17.7 42.8
Kwabenya (Ghana) -  2006/07 Harmattan 85.9 473.4
134
The world-wide guidelines values given by the World Health Organization209 are:
PM2.5 -  10 |igm'3 (annual mean) and 25 Hgm'3 (24-hours mean) and 
PMio-20 |igm'3 (annual mean) and 50 ngm'3 (24-hour mean) respectively.
The Ghana Environmental Protection Agency (Ghana EPA) has a guideline limit only for 
PM10 -  50 Hgm'3 (annual mean) and 70 ngm"3 (24-hour mean) respectively.
It should be noted that in this work the PM , 0 is not measured directly. The Gent size- 
fractionate into Coarse fraction (PM 10-2.5) and Fine fraction (PM2.5). Both fractions have to 
be added to get the PM 10 values. Averagely, the PM 10 value have been exceeded and in 
the extreme over eight times the guideline limits, while the PM2 5 values are far below the 
guideline values except during the 2006/07 Harmattan season where some extremely high 
values were measured. In the said WHO guideline it has been stated that there is no 
threshold below which damage to health is not observed so effort needs to be made to 
limit these PM10 emission.
6.2.2 BLACK  CARBON  (BC ) CONCENTRAT ION
Generally, black carbon (BC) is not measured in the coarse fraction, but it was 
determined in this work. This was done taking into account the peculiar situation in the 
country (Ghana) where most people and City Authorities indiscriminately burn waste. 
The high incidence of bush fires during the Harmattan period which gives rise to the 
coarse fraction BC was also considered. The Coarse was about 1.09% (1.65 |igm'3) as 
shown in Table 6-5.
135
For fine particulates the mean contribution of black carbon (BC) for the whole sampling 
period was about 5.45% (5.45 (igm-3) of the fine mass concentration (T ab le  6-6). The 
daily BC concentration is shown in Fig. 4-8 and that in terms of percentage is shown in 
Fig. 4-9. This shows that the finer fraction is dominated by BC, which is mostly 
generated by anthropogenic activities.
The BC of the coarse fraction is low but that of the fine fraction (about 5.5%) is on the 
higher side, especially that for the Rainy season which is about 10.85%. This value is 
comparable to industrialized countries. Some of the effects of BC include lowering of 
intelligent quotient (IQ)210 in children and global warming.211 BC is mainly from 
anthropogenic sources hence serious effort need to be made to reduce its generation.
6.2.2.1 COARSE  BLACK  CARBON  CONCENTRAT ION
The data of the coarse BC concentration is given in Tab le  6-5 . The coarse black carbon 
(BC) concentration for the 2005/06 Harmattan, 2006 Rainy and 2006/07 Harmattan were
0.57, 0.67 and 5.30 |igm'3 respectively. In terms of percentage of mass of particulates it 
was 0.64%, 1.57% and 1.12% for the 2005/06 Harmattan, 2006 Rainy and 2006/07 
Harmattan respectively. The percentage contribution to particulate mass was expected; 
during the Harmattan period dust from the Sahel region dominate the particulate matter. 
Though bush fires are common occurrences, the amount of coarse BC generated is not 
proportionally high enough to significantly affect the BC-to-mass ratio.
The median and standard deviation value for the BC for the 2005/06 Harmattan and the 
2006 Rainy seasons showed that the data was evenly spread around the mean. The BC
136
data for the 2006/07 Harmattan shows that there were some few days with extreme high 
concentration which influenced the mean.
TABLE 6-5: COARSE  BLACK  CARBON  CONCENTRAT ION  (jigm  3)
Period
N um ber
M ean StDev M edian M in M ax
% o f  | 
Masso f  D ays
2005/06 Harmattan 69 0.57 0.33 0.50 0.00 1.69 0.64
2006 Rainy 100 0.67 0.24 0.68 0.04 1.46 1.57
2006/07 Harmattan 47 5.30 5.58 2.63 0.63 20.03 1.12
All Data 216 1.65 3.24 0.71 0.00 20.03 1.09
6 2 .2.2 FINE  BLACK  CARBON  CONCENTRAT ION
The fine fraction BC concentration is shown in Tab le  6-6 . The mean concentrations 
(percentage of mass) of fine black carbon were 0.70 (6.61%), 1.92 (10.85%) and 2.44 
|igm'3 (2.84%) for the 2005/06 Harmattan, Rainy and 2006/07 Harmattan season 
respectively. The fine BC was evenly distributed around the mean as shown by the 
median and standard deviation. Though the Harmattan data were influenced by few
extremely high values it did not significantly affect the mean.
TABLE 6-6: F INE  BLACK  CARBON  CONCENTRAT ION  ( f igm 3)
Period
N um ber
M ean StD ev M edian M in M ax
% o f  
Masso f  Days
2005/06 Harmattan 69 0.70 0.66 0.52 0.01 2.96 6.61
2006 Rainy 100 1.92 0.86 1.90 0.10 4.63 10.85
2006/07 Harmattan 47 2.44 1.27 2.11 0.46 5.97 2.84
All Data 216 1.65 1.13 1.68 0.10 5.97 5.45
6.2.2.3 COM PAR ISON  OF  BC  VALUES W ITH  SOME  SELECTED  W ORKS
Comparing the results from this work with those of other researchers is presented in 
Table 6-7. From the table the BC from Africa (Tanzania, Sudan, Botswana and Ghana) 
are very high, considering the fact that these countries are not very industralised. In terms
137
of BC-to-mass ratio the values from Ghana and Botswana is slightly lower than those 
from the industrialized countries. This would be further discussed under sources of the 
particulate matter.
TABLE 6-7: COM PAR ISON  OF  BC  W ITH  RESULTS  FROM  L ITERATURE
Place
Concentration
(fig/ra3)
Ratio 
BC-Mass (%)
Reference
Fine Coarse Fine Coarse
Vaxjo (Sweden) 0.96 BDL Selin Lindgren11’"
Gothenbuirg (Sweden) 0.90 0.01 13.0 0.1 Moloi13'’
Serowe (Botswana) 0.96 0.03 9.5 0.2 , ,
Gaborone (Botswana) 0.90 0.01 Bennet212
Dar es Salam (Tanzania) 4.90 0.54 , ,
Khartoum (Sudan) 3.00 0.30 E l-T ah ir13
Riga (Latvia) 2.40 0.11 Viskna214
Vienna (Austria) 3.60 0.59 19.0 0.5 Gromiscek215
Streithofen (Austria) 2.30 0.10 12.7 1.2 „
Ghent (Belgium) 3.80 0.59 19.6 0.5 Maenhaut216
Waasmunster (Belgium) 3.40 0.49 14.2 0.4 , ,
Kwabenya (Ghana) -  All Data 1.65 1.65 5.45 1.09 This work
Kwabenya (Ghana) -  2005/06 Harmattan 0.70 0.57 6.61 0.64
Kwabenya (Ghana) -  2006 Rainy Season 1.92 0.67 10.85 1.57 , ,
Kwabenya (Ghana) -  2006/07 Harmattan 2.44 5.30 2.84 1.12 , ,
BDL  =  below  de tec tio n  lim it
From Table 6-7, it is very evident that the fine fraction contains most part of the BC and 
this needs to be taken into account in mass comparison from different location. Apart 
from the health and radiation balance consideration, the fine fraction with its high BC 
concentration presents a large surface to volume ratio than the coarse particles. Other 
gases could therefore attach themselves to the BC surface and are deposited directly into 
the lungs.
138
The following elements were identified and quantified in both coarse and fine filters 
collected during the sampling period: Al, Si, S, K, Ca, Ti, Mn, Fe, Cu, Zn, Br, Rb, Sr and 
Pb. In addition, the following elements were identified and quantified in only the coarse 
filters: Cl, V, Cr and Ni. The following elements were also identified but not used in 
modeling because they are present in very few samples or below detection limit: P, Co 
and Se in the coarse fraction and Cl, V, Cr and Co for the fine fraction.
The elemental concentrations in the two fractions vary from season to season; this 
additional information is an advantage that is not possible with only gravimetric analysis. 
Using simple correlation analysis some elements correlate well and could serve as 
possible source indicators as specified below. For example Al, Si, Mn and Fe are typical 
elements from dust or soil. These correlations also vary from season to season. Below are 
some examples:
Fine 2005/06 Harmattan:
1. Si, Ca, Ti, Mn, Fe, Zn
2. S, K,Zn, Br 
Fine 2006 Rainy:
1. Al, Si, Ca, Ti, Mn, Fe, Sr
2. S, K, Br, Rb,
3. K,Zn, Br, Pb,
4. Zn, Pb
6.2.3 ELEMENTAL CONCENTRATIONS
139
Fine 2006/07 Harmattan:
1. Al, Si, Ca, Ti, Mn, Fe, Rb, Sr
2. S, K, Br, Pb
3. Al, Si, Ca, Ti, Mn, Fe, Ni, Cu, Rb, Sr
4. Cu, Zn, Pb
5. S, Cu, Zn, Pb 
Coarse 2005/06 Harmattan:
1. Al, Si, S, K, Ca, Mn, Fe, Br, Rb, Sr
2. S, Cl, V, Br, Sr
3. Cu, Zn 
Coarse 2006 Rainy:
1. Al, Si, Ca, Ti, Mn, Fe, Rb, Sr
2. S, Cl, Br
3. Zn, Pb
Coarse 2006/07 Harmattan:
All elements seem to have some amount of correlation with each other except Cr 
which has no correlation with Ca, V and Sr, In addition, V does not correlate with 
Al.
The high correlation between elements in the 2006/07 Harmattan could be attributed to 
the high air particulate matter concentration dominated by dust from the Sahel region. 
This dust particulate acts as nucleation point for all the other elements emitted by the 
other prevailing sources to be attached to.
140
In the fine fraction it was observed that S during the Harmattan correlates with K, Br and 
Zn but during the Rainy season S does not correlate with Zn. It was also observed that 
some elements correlate but not all these elements collate with other elements. For 
example during the Rainy season in the fine fraction, S correlated with K and Br likewise 
Pb correlated with K and Br but not with S. This suggests that there are more than one 
source emitting K and Br.
6.2.4 SOURCE PROFILES
To identify the major aerosol sources and to apportion the various elements to these 
sources, the data generated over the measuring period were subjected to principal 
component analysis217 (PCA) with varimax rotation. Statistical methods have been 
commonly used for identification of the relative importance of different
129,133,190,200,218,219 t * ' * -C • j. u • Isources. Input aata tor the source assignments were chemical species
namely elemental, black carbon and mass.
In the PCA analysis, several runs were made. The number of factors was varied and 
varimax as well as promax rotations were performed. The pollution sources were 
assumed to be independent of each other and because varimax gave the most consistent 
results it was the rotation that was used in the factor analysis. The same factors appeared 
in the analysis even if some variables were omitted, although the factor loadings for the 
different elements varied slightly.
141
Table 6-8  shows the five source profiles generated by the receptor model used in this 
work (as per Eqn. 5 -6 ) for the coarse particulate for the 2005/06 Harmattan which 
resulted in an average contribution of:
• Profile 1: 4.84 |ig /nvVday (5.5%)
• Profile 2: 68.46 |ig /mVday (78.2%)
• Profile 3: 1.30 pg /m3/day (1.5%)
• Profile 4: 8.73 pg /m3/day (10.0%)
• Profile 5: 4.12 pg /m3/day (4.8%)
The elements and BC values in the profile are per unit mass of aerosol deposited on the filter. 
Hence the Mass values should be 1.0 and the values obtained also shows the accuracy of 
the modeled profile fit to the data. Source profile values shown in bold are the suggested 
main species contributing to the respective sources. The criteria used includes: the species 
high values compared with their respective standard deviations and the contribution of 
the element distributed over the total number of sources.
Using source information from other researchers in both Africa and Europe136,190,218 the 
most likely sources which fit the profiles in the coarse fraction of the 2005/06 Harmattan
• Profile 1- Soil/Dust
• Profile 2 — Biomass burning intermixed with long distance transport (LDT) and 
dust
• Profile 3 — Metal Industrial -  Cu and Zn
6.2.4.1 COARSE FRACTION PARTICULATE MATTER
142
• Profile 4 — Sea spray intermixed with long distance transport (LDT)
• Profile 5 -  Industrial emission -  Pb
Profile 1 in the 2005/06 Harmattan contains over 56% Al, 58% Si, 54% Ca, 53% Ti, 47% 
Mn, 47% Fe, 46% Rb and 56% of Rb which are all strong signatures of soil/dust. This is 
not surprising since it was during the harmattan season but its contributing to the mass is 
only about 5% and hence attributed to activities in the vicinity of the sampler.
Profile 2 which contains 82% of the mass looks like a mixture of three sources with 31% 
of BC, 27% Al, 24% Si, 41% S, 32% K, 21% Ca, 35% Mn, 28% Fe, 35% Zn, 32% Rb 
and 22% Sr. The very high S is major signature of long distance transport (LDT). Al, Si, 
Ca, Mn, Fe, Rb, and Sr are all signatures of soil/dust. BC, K and Zn are also signature of 
biomass. The coarse biomass is due to the localised bush fires and domestic waste 
burning.
Profile 3 contains 76% Cu and 34 % Zn with less than 5% BC which shows that it is not 
associated with combustion process. This is appearing in the coarse fraction so can only 
be attributed to metal polishing/grinding industry.
Profile 4 contains 31% S, 63% Cl, 52% Br and 35% BC. Kwabenya is less than 10 km
from the sea hence the presence of Cl and Br, which are strong signature of sea spray,
was expected. The high percentage of S and BC in the source shows that it is intermixed 
with LDT.
143
TABLE 6-8: COARSE PARTICLE SOURCE PROFILE
(28™ DECEMBER 2005 -  31st MARCH 2006)
Element (ng/fig) Profile 1 Profile 2 Profile 3 Profile 4 Profile 5
Al 128.76 4.35 17.89 15.86 3.99
2.51 0.27 7.45 2.36 2.34
Si 392 .16 11.42 53.75 47.17 22.74
7.36 0.8 21.83 6.92 6.86
S 6.42 0.97 3.21 5.75 1.66
0.41 0.04 1.23 0.39 0.39
Cl 22.26 0.37 1.8 27.74 1.39
2.02 0.22 6.01 1.91 1.89
K 26.69 1.48 5.78 6.43 3.33
0.59 0.06 1.75 0.55 0.55
Ca 50 .89 1.44 2.75 10.4 3.31
2.58 0.28 7.66 2.43 2.41
Ti 10.80 0.40 2.06 1.26 1.12
0.23 0.02 0.67 0.21 0.21
V 0.16 0.06 0.05 0.03
0.01 0.002 0.04 0.01
Cr 0.07 0.06 0.13 0.06
0.01 0.001 0.04 0.01
Mn 1.72 0.09 0.18 0.24 0.18
0.03 0.003 0.09 0.03 0.03
Fe 88.39 3.74 19.11 16.58 10.33
1.31 0.14 3.9 1.24 1.22
Ni 0.2 0.11 0 .10
0.01 0.03 0.01
Cu 0 . 1! 1.75 0.02
0.01 __ 0.04 0.01
Zn 0.14 0.03 1.35 0.07 0.07
0.02 0.002 0.05 0.01 0.01
Br 0.20 0.13 0.14
0.01 0.02 0.01
Rb 0.13 0.01 0.02 0.03 0.01
0.003 0.0003 0.01 0.003 0.003
Sr 0.45 0.01 0.09 0.01
0.01 0.001 0.01 0.01
Pb 0.11 0.03 0.03 0 .10
0.005 0.02 0.005 0.005
BC (ng/fig) 0.02 0.01 0.01 0.01
0.002 0.002 0.005 0.002
Mass (ng/ng) 0.84 1.02 1.36 0.61 1.28
0.19 0.02 0.58 0.18 0.18
NB: 1. Values in i t a l i c s  are the standard deviation of the various species. 
2. Bold values are suggested main species contributing to Source.
144
Profile 5 contains over 27% Pb and account for about 5% of the loaded mass, which is 
significant. This could be attributed to an industrial source emitting Pb.
The coarse fraction source profiles for the 2006 Rainy Season is given in Table 6-9 with 
an average contribution of:
• Profile 1: 14.94 (ig /m3/day (34.5%)
• Profile 2: 13.46 \ig /mVday (31.1%)
• Profile 3: 3.98 (ig /m3/day (9.2%)
• Profile 4: 8.46 ng /m3/day (19.6%)
• Profile 5: 2.82 ng /mVday (5.6%)
The most likely sources that fit the profiles are:
• Profile 1 — Soil/Dust
• Profile 2 -  Sea spray/LDT
• Profile 3 — Biomass
• Profile 4 -  Industrial -  Cr, Ni, Cu
• Profile 5 -  Industrial -  V, Cr
The 2006/07 Harmattan coarse fraction source profiles are given in Table 6-10 with an 
average contribution of:
• Profile 1: 180.99 jag /m3/day (33.6%)
• Profile 2: 123.16 (ig /m3/day (22.9%)
• Profile 3: 147.80 pg /m3/day (27.4%)
• Profile 4: 86.32 pg /nrVday (16.0%)
• Profile 5: 0.18 fig /m3/day (0.03%)
145
TABLE 6-9: COARSE PARTICULATE SOURCE PROFILE
(4th APRIL 2006 -  31st OCTOBER 2006)
Element (ng/Mg) Profile 1 Profile 2 Profile 3 Profile 4 Profile 5
Al 36 .38 11.62 28.46 18.57 18.51
1 .55 2 .5 6 7 .35 5 .6 5 9.02
Si 114.12 32.98 107.45 68.57 59.31
2 .79 4 .61 13 .21 10.16 16.22
S 2.23 18.88 11.39 8.9
0 .2 8 0 .4 6 1.33 1 .63
Cl 13.92 125.42 50.14
2 .5 4 .1 4 14 .5 8
K 7.43 14.89 20 .58 1.33
0 .6 9 1 .1 4 3 .2 8 2 .5 2
Ca 13.25 7.95 12.15 10.64 9.47
0 .4 4 0 .72 2 .0 7 1 .59 2 .5 4
Ti 3.29 0.35 3.29 2.41 1.33
0 .0 8 0 .1 4 0 .3 9 0.3 0 .4 8
V 0.03 0.02 0.17 0.57
0 .0 0 6 0 .01 0 .03 0 .0 4
Cr 0.01 0.04 0.21 0.63
0.01 0 .0 2 0 .0 6 0.66
Mn 0.49 0.17 0.75 0.39 0.28
0 .02 0 .03 0 .0 8 0 .06 0.1
Fe 29.81 11.29 57.58 17.33 18.5
1 .04 1.72 4 .9 4 3 .8 6.07
Ni 0.03 0.04 0.29 0.02
0 .0 0 3 0 .01 0 .0 1 0.02
Cu 0.02 0.02 0.23 0.32 0.03
0 .0 0 3 0 .005 0 .01 0 .01 0 .02
Zn 0.07 0.06 1.26 0.09 0.10
0.01 0 .02 0 .0 6 0 .0 5 0 .0 7
Br 0.03 0.18 0.31 0.22 0.08
0 .0 0 5 0 .01 0 .0 2 0 .02 0 .03
Rb 0.04 0.02 0.07 0.02 0.01
0 .0 0 2 0 .003 0 .01 0 .01 0 .01
Sr 0.10 0.88 0.07 0.09 0.08
0 .0 0 3 0 .0 0 6 0.02 0 .01 0 .0 2
Pb 0.06 0.43 0.08
0 .0 0 4 0 .02 0.02
BC (Mg/fig) 0.01 0.01 0.07 0.01 0.01
0.001 0.001 0 .0 0 3 0 .0 0 2 0 .0 0 4
Mass (fig/ng) 1.08 1.06 0.89 0.71 0.96
0 .03 0 .0 5 0 .1 6 0 .1 2 0 .2 0
146
TABLE 6-10: COARSE PARTICULATE SOURCE PROFILE
(2nd NOVEMBER 2006 -  15™ FEBRUARY 2007)
Element (ng/ng) Profile 1 Profile 2 Profile 3 Profile 4 Profile 5
Al 59.51 31.3 9.13 2046.65
5 .2 3 .7 5 10 .5 2 8 8 4 .1 6
Si 4.16 177.91 186.71
3 .6 2 7 .92 5 .71
S 0.57 1.22 1.95 6.89 16.61
0 .0 8 0 .18 0 .13 0 .3 7 100 .91
Cl 2 .23 3.95 2.77 4.13 989.84
0 .31 0 .67 0 .4 8 1 .35 3 7 1 .1 3
K 9 .84 9.67 8.01 9.55
0 .2 8 0 .61 0 .43 1 .22
Ca 20.9 10.95 13.53 10.94 722.27
0 .4 5 0 .9 8 0 .71 1 .9 8 3 3 6 .1 3
Ti 2.93 2.65 2.18 3.31
0 .01 0 .21 0 .1 5 0 .4 2
V 0.10 0.08 0.04 0.08
0 .0 0 5 0 .01 0.01 0 .0 2
Cr 0.08 0.02
0 .0 0 2 0 .0 0 4
Mn 0 .67 0.49 0.35 0.47 13.01
0 .01 0 .03 0 .0 2 0 .0 6 16 .91
Fe 26.63 23.43 18.21 28.95
0 .7 8 1.72 1 .24 3 .4 6
Ni 0.02 0.02 0.03 0.04 8.56
0.001 0 .003 0 .0 0 2 0 .0 0 5 1 .39
Cu 0.03 0.03 0.02 0.07 14.44
0 .0 0 2 0 .0 0 4 0 .0 0 3 0 .01 2 .4 6
Zn 0.07 0.18 0.04 0.17 9.95
0 .0 0 5 0 .01 0 .01 0 .0 2 6 .2 7
Br 0.03 0.04 0.03 0.06 9.95
0 .0 0 2 0 .0 0 5 0 .0 0 3 0 .01 2 .5 9
Rb 0.06 0.05 0.03 0.05
0.001 0 .003 0 .0 0 2 0 .0 0 5
Sr 0.20 0.09 0.11 0.12 6.35
0 .0 0 4 0 .01 0.1 0 .0 2 5.06
Pb 0.02 0.05 0.02 0.10
0 .0 0 2 0 .0 0 5 0 .003 0 .01
BC <Hg/(Jg) 0.01 0.01 0.02
0.001 0.001 0 .0 0 2
Mass (ng/ng) 0 .64 1.21 1.09 0.69 0.66
0 .0 2 0 .0 4 0 .03 0 .0 9 0 .2 9
147
The m ost likely  sources th a t fit the  p ro files are:
•  P ro file  1 -  S ea sp ray  +  Industria l Zn
•  P ro file  2 -  So il/D ust +  Zn
• Profile 3 -  Soil/Dust + Ni
• Profile 4 -  Biomass intermixed with LDT
• Profile 5 -  Industrial -  Ni, Cu
6.2.4.2 F INE  FRACT ION  PART ICULATE  M ATTER
Three source profiles were generated by the receptor model for the fine particulates in
2005/06 Harmattan as shown in Tab le  6-11 with an average contribution of:
• Profile 1: 0.69 |ig /m3/day (6.3%)
•  P rofile 2: 3 .89  ^g  /m 3/day  (35 .2% )
• Profile 3: 6.46 |_ig /m3/day (58.5%)
The fine fraction of the 2005/06 Harmattan Profile 1 contains over 35% Si, 40% Ca, 34% 
Ti, 27% Mn and 32% Fe. These are all strong signatures of soil/dust, though the Rb and 
Sr levels in the samples for the period were too few to be used for the model, this is still a 
soil/dust source. This profile contributes only about 6.3% to the loaded mass on the 
filter.
Profile 2 contains over 78% S, 66% K. 46% Mn, 42% Fe, 78% Zn, 75% Br and 85% BC. 
The high S levels indicate long distant transport (LDT) and the high K, Br and BC 
indicate biomass burning. Hence this profile is a mixture of LDT and Biomass burning.
148
Profile 3 is similar to Profile 1 but without Ti, it contains over 36% Si, 27% Ca, 26% Mn, 
24% Fe. This profile account for 62.7% of the loaded mass, and contains all the signature 
of dust/soil. It appears that there were two sources of fine particulate soil/dust during the 
2005/06 harmattan season.
TABLE  6-11: F INE  PARTICLE  SOURCE  PROFILE  
(28™  DECEM BER  2005 -  31ST M ARCH  2006)
Element (ng/ng) Profilel Profile2 Profile3
Si 165.22 22.37 18.17
3 .9 3 2 .6 9 1 .5 4
s 7.66 19.37 2.38
0 .7 3 0 .5 0 .2 9
K 13.61 15.65 3.33
0 .6 5 0 .4 5 0 .2 6
Ca 19.70 2.77 1.46
0 .4 4 0 .3 0 .1 7
Ti 4.65 1.02 0.34
0 .1 4 0.1 0 .0 5
Mn 0.78 0.23 0.08
0 .03 0 .02 0 .01
Fe 38.93 9.05 3.08
0 .9 9 0 .6 8 0 .3 9
Zn 0.24 0.29 0.02
0 .01 0 .01 0 .01
Br 0.11 0.24 0.03
0 .02 0 .0 1 0 .0 0 5
BC (Mg/Mg) 0.04 0.11
0 .01 0 .0 0 3
Mass (ng/ng) 1.01 0.81 1.01
0 .0 5 0 .03 0 .0 2
The most likely sources that fit the profiles are:
• Profile 1 -  Soil/Dust with Ti
• Profile 2 -  LDT/Biomass
• Profile 3 -  Soil/Dust without Ti
149
TABLE 6-12: FINE PARTICULATE SOURCE PROFILE
(4™ APRIL 2006 -  31st OCTOBER 2006)
Element (ng/ne) P rofile  1 Profile 2 Profile 3 Profile  4 Profile  5
Al 52.69 2.03 11.62 21.17
2.31 0.75 2.31 9.58
Si 175.36 5.54 1.37 40.69
3.1 <5.65 3.1 12.84
S 44.38 38.09 10.52 1.82
1.14 7.45 3.48 14.43
K 6.37 28.97 19.98
1.81 0.59 3.87
Ca 22.39 0.47 5.04 5.24
0.48 0.16 1.04 2.01
Ti 4.60 0.61 0.33 1.1
0.11 0.24 0.11 0.47
Mn 0.75 0.06 1.03 0.02 0.46
0.03 0.01 0.07 0.03 0.13
Fe 49.82 1.41 13.02 0.17 11.21
0.81 0.26 1.73 0.81 3.35
Ni 0.02 0.01 0.02 0.39 0.23
0.01 0.002 0.01 0.005 0.02
Cu 0.1 __  0.01 0.23 0.12 1.89
0.004 0.001 0.01 0.004 0.02
Zn 0.16 0.22 2.62 1.19
0.06 0.02 0.12 0.23
Br 0.54 0.57 0.14 0.17
0.02 0.12 0.05 0.22
Rb 0.03 0.08 0.05 0.05
0.01 0.003 0.02 0.04
Sr 0.15 0.01 0.01 0.10
0.01 __  0.0/ 0.01 0.02
Pb 0.14 0.05 1.06 0.01 0.33
0.02 0.01 0.04 0.02 0.08
BC (fig/(ig) 0.02 0.12 0.46 0.05 0.01
0.01 0.004 0.03 0.01 0.06
Mass (|ig/fig) 0.99 1.01 0.94 0.99 1.01
0.15 0.05 0.31 0.15 0.61
The fine fraction source profiles for the 2006 Rainy Season is given in Table 6-12 with 
an average contribution of:
• Profile 1: 1.06 \xg /m3/day (6.1%)
• Profile 2: 9.49 ^g /m3/day (54.8%)
• Profile 3: 0.94 \ig /mVday (5.4%)
150
• Profile 4: 4.79 jig /mVday (27.7%)
• Profile 5: 1.04 jig /mVday (6.0%)
The most likely sources that fit the profiles are:
• Profile 1 -  Soil/Dust
• Profile 2 -  LDT/Biomass
• Profile 3 -  Industrial -  Mn, Zn and Pb
• Profile 4 -  Industrial -  Al, Ti and Ni
• Profile 5 -  Industrial -  Cu and Zn
The 2006/07 Harmattan fine fraction source profiles are given in Table 6-13 with an 
average contribution of:
• Profile 1: 50.60 |ag /m’/day (58.0%)
• Profile 2: 23.93 fxg /m3/day (27.4%)
• Profile 3: 7.97 ng/m3/day (9.1%)
• Profile 4: 4.70 ng /m3/day (5.4%)
• Profile 5: 0.11 jag /nrVday (0.1%)
The most likely sources that fit the profiles are:
• Profile 1 -  Soil/Dust
• Profile 2 -  LDT/Biomass
• Profile 3 -  Industrial -  Cu, Zn and Pb
• Profile 4 -  Industrial -  Ca, Ni and Br
• Profile 5 -  Industrial-Ni and Cu
151
TABLE 6-13: FINE PARTICULATE SOURCE PROFILE
(2nd NOVEMBER 2006 -  15™ FEBRUARY 2007)
Element (ng/jig) P rofile  1 Profile  2 Profile  3 P rofile  4 P rofile  5
Al 16.22 12.51 0.76
0.6 2.25 3.41
Si 55.47 37.32 11.38 49.34
1.04 3.90 5.91 15.37
S 0.2 19.48 3.06 0.5 11.98
0.14 0.56 0.84 2.18 52.01
K 3.94 16.39 0.43 11.2 136.18
0.19 0.72 1.09 3.17 67.83
Ca 6.08 23.70
0.22 3.17
Ti 1.12 1.00 0.30 1.13 17.55
0.03 0.11 0.16 0.42 10.09
Mn 0.21 0.17 0.06 0.37
0.01 0.02 0.03 1.97
Fe 9.08 8.36 3.25 16.04
0.21 0.77 1.18 3.06
Ni 0.04 0.07 0.03 0.29 5.27
0.002 0.01 0.01 0.03 0.77
Cu 0.02 0.03 0.39 0.31 4.54
0.004 0.01 0.01 0.04 0.83
Zn 0.02 0.23 0.54
0.004 0.02 0.02
Br 0.02 0.13 0.04 0.40
0.001 0.004 0.01 0.02
Rb 0.02 0.04 0.01 1.12
0.001 0.006 0.01 0.33
Sr 0.06 0.02 0.01 0.16
0.001 0.004 0.01 0.02
Pb 0.12 0.10
0.01 0.01
BC (ng/fig) 0.08 0.01 0.03 0.28
0.002 0.003 0.01 0.18
Mass (fig/ng) 1.03 1.04 1.26 1.06 0.67
0.01 0.04 0.66 0.16 0.28
152
6.2.5 C OM PAR ISO N  BETW EEN  MODEL  AND  EXPER IM ENTAL  DATA
Generally, the model was able to reproduce the experimental data as shown in Tables 5- 
11 to 5-16  with the very high R2 values. The gradient of the fits was also very close to 1. 
The fit for Cl in the coarse fraction consistently gave R2 value (0.88, 0.78 and 0.83 for the 
zero fit and 0.91, 0.87 and 0.83, 2005/06 Harmattan, 2006 Rainy Season and 2006/07 
Harmattan respectively) which were among the lowest.
The fine particulate BC and the Coarse BC for the 2005/06 Harmattan and 2006 Rainy 
seasons model fit the data very well. The 2006/07 Harmattan Coarse BC (as shown in 
Fig. 5-2), the model fits the experimental data better for BC concentrations less than 10 
Hgrn3. In the theory for the calculation of the BC (Eqn  3 -13 ) it was assumed that a thin 
layer of aerosol would be deposited on the filter giving rise to a single dust layer 
thickness. For the extreme Harmattan of 2006/07, it appears that the particulate matter 
deposited on some of the filters was more than a single layer thick.
153
SUM M ARY , CONCLUSIONS  AND  R ECOM M ENDAT IONS  
7.1 SUM M ARY
Airborne particulate matter (APM) in the semi-rural town of Kwabenya, near Accra- 
Ghana, was collected using Gent sampler onto nuclepore filters. The Gent sampler size- 
segregated the APM into coarse (2.5<aerodynamic diameter<10 urn) and fine 
(aerodynamic diameter < 2.5 pm). A total of 216 samples were collected from 28th 
December 2005 to 12th February 2007. The sampling period spanned two Harmattan 
seasons and one rainy season. The parameters determined were mass, elemental and 
black carbon concentrations. This was achieved using gravimetric analysis, EDXRF and 
reflectometric techniques respectively.
The filters used for air pollution studies are very important, in terms of the material they 
are made of, since all the parameters measured come from the loaded filter. It was 
therefore very important that the filter type used had very low blank values since this will 
give very low detection limit. The nuclepore filters used in this work satisfy this 
condition. The mass of the loaded filter is also very critical in atmospheric aerosol 
analysis, because all the other measurement will be fractions of this. To minimize any 
discrepancy in the weighing procedure before and after sampling, measurements were 
done with microgramme-sensitive (|ig) balance in a temperature- and relative humidity- 
controlled environment.
The mass concentration of the particulate matter (PM,0) at Kwabenya was dominated by 
the coarse particles (PM 10-2.5). The average mass concentration for the whole measuring
CHAPTER 7
154
period was 151.14 |xgm'3 for the coarse fraction and 30.26 ngm'3 for the fine fraction. The 
coarse mass concentration for the Harmattan seasons generally was higher than the Rainy 
season. The coarse-to-fine ratios were also higher for the Harmattan seasons than the 
Rainy season. A total of 185 days out of the 216 days of sampling had PMio values above 
the WHO limit of 50 |igm'3 for 24-hour mean. Using the Ghana EPA guideline limit of 70 
jigm~3 for 24 hours, 130 days had values above the guideline limit. A total of 60 days had 
PM2.5 values that exceeded the WHO limit of 25 ngm"3 for 24-hour mean. Though mass 
concentration alone does not provide sufficient information when the limit values are 
exceeded, it serves as basis for the recommendation of abatement strategies and high 
PM 10  concentrations are not favourable for people with respiratory health problems 
(asthma, cough and catarrh), allergic eye diseases, and sickle cell anaemia.
Black carbon (BC), also known as elemental carbon (EC) is essentially a primary 
pollutant emitted during incomplete combustion of fossil and biomass carbonaceous 
fuels. The “Black smoke” or Gagel150 method was used to measure the BC concentration 
on the loaded filters. On the average the BC concentration in the fine fraction was higher 
than the coarse fraction except for the 2006/07 Harmattan period. In terms of percentages 
of loaded mass on filter, the fine fraction BC was much higher than the coarse fraction 
BC.
Elemental concentration was determined on the loaded filters using energy dispersive x- 
ray fluorescence analysis. The spectrometer used was optimised for air filter analysis 
through the design of the sample holder. This facilitated the reduction of scattering of x- 
rays into the detector, enhancement of signal-to-background ratio, and consequently
155
lower detection limits. The x-ray spectra were fitted using the AXIL188'189 software and the 
fundamental parameters’ quantitative method. The analytical procedure was validated 
using a NIST air filter standard reference material SRM2783. The elements identified in 
the coarse fraction and quantified were Al, Si, S, Cl, K, Ca, Ti, V, Cr, Mn, Fe, Ni, Cu, Zn, 
Br, Rb, Sr and Pb. In addition, the following elements were identified; P, Co and Se but 
not quantified in the coarse fraction because they were mostly below detection limit or 
were present in very few samples. In the fine fraction the following elements were 
identified Al, Si, S, K, Ca, Ti, Mn, Fe, Cu, Zn, Br, Rb, Sr and Pb and quantified. Cl, V, 
Cr and Co were identified but were not quantified in the fine fraction. The elemental 
concentrations varied from season to season. As expected those of the crustal origin -  Al, 
Si, Ti, Mn, Fe. Rb and Sr had higher values during the Harmattan seasons. The 
correlation between the various elements served as pointers to the sources emitting the 
particulate matter.
A receptor model using principal component and regression analysis was used for the 
pollution source apportionment of the loaded aerosol. The species used for the receptor 
model were -  mass, BC and elemental concentrations. The coarse aerosol was influenced 
by some specific sources over the study period, including Soil/Dust, Biomass/LDT and 
Sea aerosol/spray in addition to other sources depending on the season. The fine fraction 
is dominated by Soil/Dust, Biomass/LDT and some metal industrial sources. The model 
data generated fitted the experimental data very well.
156
This work is the first known sustained air particulate matter measurement in Ghana 
spanning fourteen months. Other works sited in Chapter 1 of this thesis (Section 1.2.3.) 
were for brief periods sometimes lasting less than three weeks. The sampling period 
covered two Harmattan seasons and a Rainy season. The Harmattan season with its 
characteristic dry desert wind from the Sahara is very peculiar to the West African region 
and the study of the Harmattan dust is currently among one of the focused area of air 
pollution experts53'68'220'224 because of the aerosol load that it generates and its effect on 
health and climate. Hence studies that can multi-characterise the Harmattan aerosol 
(mass, BC and elemental concentrations) are very important, as in this work.
The project design and the type of filters used in this work allowed for the measurement 
and analyses of different parameters such as mass, BC and elemental concentration. The 
mass was further segregated into fine (PM2.s), coarse (PM 10-2.5) fractions. Consistently, the 
mass concentration of the particulates showed low values during the rainy season and 
very high levels during the harmattan period, especially during the 2006/07 Harmattan. 
The coarse PM ranged from 0.16 (ig/m3 to 1794.01 ng/m3; PM2.5 (fine) fraction from 0.5 
Hg/m3 to 430.23 |ig/m3; PM 10 from 0.87 (ig/m3 to 2064.89 |ig/m3.
Though fine particles can originate from natural sources in addition to anthropogenic 
sources, it is a well known fact that the coarse-to-fine (C/F) ratio can be used as possible 
indicator for natural and anthropogenic emissions. Usually elements of natural origin 
have high C/F ratios (>1.5) and elements of anthropogenic origin have low C/F ratio.136'225 
In this work, for all the data the coarse-to-fine ratio average is 5.0 (8.4 for the 2005/06
7.2 CONCLUSIONS
157
Harmattan; 2.4 for the Rainy Season and 5.5 for the 2006/07 Harmattan). From this result 
it can be said that most of the APM source at Kwabenya are from natural origin.
The fine fraction (PMz.s) aerosols are more mobile in the environment than the coarse 
fraction. The fine fraction impact on the environment is therefore greater since it can 
incorporate in raindrops or be easily transported in the biosphere, etc. It also has 
implication for human health as they are easily absorbed in the human body. All these 
facts are important in epidemiological and hazard evaluation when aerosol particles are 
being investigated.
Determination of BC by reflectometer on the nuclepore filters shows that the coarse 
fraction contains very little BC contribution. The fine fraction mass however is 
dominated by BC accounting for over 10% during the 2006 Rainy Season. This shows 
that the BC is coming from anthropogenic sources.
Conventional EDXRF is severely limited in air filter analysis because of the low mass 
loading; this limitation arises because the results generated are very close to the detection 
limit. One of the factors contributing to this high detection limit is scattering of excitation 
radiation into the detector thereby producing a high background. The EDXRF 
spectrometer used at the Royal Veterinary and Agricultural University of Denmark, 
Copenhagen - Denmark was optimised for air filter analyses.187 This was achieved by the 
modification of the source-sample-detector arrangement (F ig . 4 -10) such that the 
scattered radiation are mainly transmitted through the filter giving rise to a high signal-to- 
background ratio and lower detection limits. A total of 21 elements ranging from Al -  Pb
158
were identified but 18 quantified in the coarse fraction (with concentrations ranging from
0.5 ngm"3 to 47295 ngm'3). In the fine fraction 18 elements were identified and 14 
quantified (with concentration ranging from 0.67 ngm'3 to 4080 ngm'3). The elemental 
analysis of the loaded aerosol show high levels of air pollution originating from natural 
and anthropogenic sources. Ghana is a developing country and the level of 
industrialisation is quite low but the levels of BC and elements-such as S, Ni and Pb, are 
comparable to that of Sweden212 which is a developed country. However S is a strong 
indicator of long distance transport (LDT), this could be resolved in future work when air 
mass trajectories are added to determine if the S from local sources is more dominating 
than that from LDT.
This work did not set out to do direct source measurements, measurements sited at the 
pollutant emission points (as shown in Fig. 2-4), because it is difficult and very 
expensive. In addition, the modification and transformation of the species in the 
atmosphere is not fully understood because the atmosphere is a very complex system. It is 
for this reason that the receptor model approach was used. It is satisfying that some 
characteristic elements appeared which were used to explain the source profiles from the 
receptor model used in this work. The metal industries' signature in the fine are very 
striking since during the study no known metal processing industries/smelters were in 
operation in the Accra Metropolitan Area.
Biomass burning was identified as a major source contributing to the APM at Kwabenya, 
especially in the fine fraction with the high BC, K and Br. In the coarse fraction sea spray
159
(aerosol) is a major source of Cl. The high source of Cl in the air could account for the 
high corrosion of appliances at Kwabenya and along the coast of Ghana.
The lack of vital meteorological data from the Ghana Atomic Energy Commission mini­
weather station, such as wind speed and direction, has hampered the ability of this work 
to identify the source directions. The nearest station that could be used is from the Kotoka 
International Airport which is about fifteen kilometers away but given the topology and 
rainfall patterns at the two stations, the airport results could not be applied to this work. 
This situation is currently being addressed with the installation of a new weather station.
Developed countries have a well coordinated air quality management system, including 
routine air particulate matter monitoring. Most of these developed countries are in the 
temperate regions. To fully understand the behaviour of aerosol particles under different 
climatic conditions there is the need to characterise aerosol particles from different 
places including developing countries like Ghana. This work is therefore contributing to 
the bridging of this information gap on the nature and characteristics of aerosol in Ghana 
in terms of mass. BC, elemental concentrations and sources emitting them.
The high aerosols load during the Harmattan dominated by sources of crustal origin is in 
agreement with the findings of other researchers.53,68'221
The ability of this work to use source signatures from outside the region to identify 
sources contributing to the APM at Kwabenya is very interesting. This will curtail the 
need for direct source measurement and characterisation of all identifiable sources since 
this is very expensive and tedious.
160
For any nation to develop appropriate air policy there is the need for sound scientific 
evidence backed by reliable monitoring for public acceptance. This would assist to make 
regulatory and mitigation actions more acceptable to those who have to pay in the short­
term to achieve the long-term benefits. Other pollutants such as gaseous (SOx, CO, 0 3 
and NOx) and chemical carcinogens (PAFls, Benzene and formaldehydes) are all very 
important to assess air quality but the equipment for their measurements and monitoring 
are not available in the country. Without reliable and well-trusted monitoring, carried out 
with techniques tailor-made to the specific situation and equipment available, policy 
development may be severely hampered. This is very crucial in a developing country like 
Ghana, where resources are scarce given a host of competing priorities. Flence studies 
that can generate multiple parameters such as mass, elemental concentrations, and BC 
concentrations are preferred, as in this work. In addition, air pollution monitoring is very 
expensive (in terms of equipment, logistics and consumables) hence monitoring in 
combination with modeling is the best and preferred option. These were the criteria that 
influenced this thesis.
7.3. R ECOM M ENDATIONS
From the results, discussions and conclusions of this work it is recommended that:
• The measurement currently continuing at Kwabenya should be supported to gen­
erate a larger data set for the improvement of the model and developed into a 
monitoring site with data spanning decades (like the super sites in the USA) for 
long term trend studies.
161
• Source profiles of some of the local sources identified are to be determined (at 
source) and compared with those generated in this work (at the receptor).
• There is also the need to study the amount of loaded aerosol that will make the 
BC concentration exceed the single layer of particulate matter on the filter.
• According to Reichhardt32 there exists credible evidence that PMio (24 hour mean) 
in the range 30 — 200 |igm‘3 has effects on human health. There is also the need to 
combine the mass concentration measurements with epidemiological studies to es­
tablish the possible effects in Ghana. This study is very important for developing 
countries because most of the studies have come from countries where there is a 
high fine-to-coarse ratio. This could be helpful in showing whether fine concentra­
tions are more important with regard to health effects than Coarse in the PMio.
• The high anthropogenic (especially fine) S, Ni, Pb and BC is worrying and there 
is the need to take some mitigation measures to reduce the level.
• For a more complete air quality situation in the country there is the need to set 
monitoring stations in the Northern sector (for example in Navorongo or Tamale), 
the Central sector (preferably Kumasi) and the one at Kwabenya in the South.
• Finally, with the discovery of oil on the Western coast of the country, it would be 
very appropriate to setup a monitoring station in the area to determine the baseline 
data upon which the production of oil would impact.
162
References
1. Hinds, W. C., Aerosol Technology: Properties, Behavior, and Measurement o f 
Airborne Particles, John Wiley, New York, 1999, pp l-6 .
2. Colbeck I., Particle Emission from Outdoor and Indoor Sources. , in The Hand­
book o f Environmental Chemistry: Airborne Particulate Matter, Kouimtzis T. and 
Samara C. (eds.), Springer, Berlin, 1995, pp 2 -  34.
3. Harrison R.M., Air Pollution: Sources, Concentration and Measurement, in Pollu­
tion: Causes, Effects and Control, Harrison R.M. (ed), The Royal Society of 
Chemistry, Cambridge, 1996, pp 144-168.
4. Jaenicke R., Atmospheric Aerosol Size Distribution, in Atmospheric Particles, 
Harrison and van Grieken R. E. (ed), John Wiley, Chichester 1998, pp 1-25.
5. Colls J., Air Pollution — An Introduction, E & FN Spon, London, 1997, p p l.
6 . Buseck P. R. and Posfai M., Aerosol minerals and related aerosol particles: Ef­
fects on climate and the environment. Proceedings o f  the National Academy o f 
Sciences o f the United States o f America, 96, pp3372-3379, (1999)
7. Graedel T. E. and Crutzen P. J., Atmospheric Change: An Earth System Perspect­
ive, W. H. Freeman and company, New York, pp77, (1993).
8 . Raes F., Dingenen R. V., Vignati E., Wilson J., Putaud J., Sienfeld J. H. and 
Adams P., Formation and cycling of aerosols in the global troposphere, Atmo­
spheric Environment, 34(25), pp4215-4240, (2000).
9. World Health Organisation (WHO), WHO guidelines fo r  air quality. Fact sheet 
no. 187, 2000, [http://www.who.int/inf-fs/factl87.html]
163
10. Salby M. L.. Fundamentals o f Atmospheric Physics, Academic Press, San Diego, 
1996, pp258-262.
11. Brimblecombe P., Air Pollution and Health History, in Air Pollution and Health, 
Holgate S. T., Samet J. M., Koren H. S. and Maynard R. L. (eds.), Academic 
Press, London, 1999, pp5-18.
12. Ramazzini B., De morbis artificium (1713), Reprinted by Arbetsmiljofolaget, 
1989.
13. Linne C. von, Dalaresan (Iter Dalekarlicum), 1734, Reprinted in Bertil Gullander 
(ed.), 1980, Linne i Dalarna. Forum, Stockholm.
14. Loomis D., Sizing up air pollution research, Epidemiology 11, pp2-4, (2000).
15. Mark D., Atmospheric Aerosol Sampling, In Atmospheric Particles. Harrison 
RM, van Grieken R (ed). Wiley: Chichester; 1998, pp29-94.
16. Schwartz J and Neas LM., Fine particles are more strongly associated than course 
particles with acute respiratory health effects in school children, Epidemiology 11:
pp6 - 1 0 , (2 0 0 0 ).
17. Donaldson K, Li XY, MacNee W., Ultrafine (nanometre) particles mediated lung 
injury. Aerosol Science; 29: pp553-560, (1998).
18. Harrison, RM., Pollution Causes, Effects and Control, Royal Society of Chem­
istry, Cambridge, 2001.
19. Hauck H., Berner A., Frischer T., Gomiscek B., Kundi M., Neuberger M., 
Puxbaum H., Preining O. and AUPHEP- Team , AUPHEP -  Austrian Project on 
Health Effects of Particulates -  general overview, Atmospheric Environment. 38, 
pp3905-3915, (2004).
164
20. Anderson K. R., Avol E. L., Edwards S. A., Shamoo D. A., Peng R. C„ Linn W.
S., and Hackney J. D., Controlled exposures of volunteers to respirable carbon 
and sulphuric acid aerosol, Journal o f Air and Waste Management Association, 
42, pp770-776, (1992).
21.CSIRO Atmospheric Research, Air Pollution: eight thousand deaths per day, 
[http://www.dar.csiro.au/news/1999/mr997.html]
22. Statement on behalf of the American Academy of Pediatrics before the Clean Air 
Scientific Advisory Committee on the USEPA regarding National Ambient Air 
Quality Standards for Particulate Matter on April 7, 2005.
23. American Academy of Pediatrics, Committee on Environmental Health, Ambient 
Air Pollution: Health Hazards to Children, Pediatrics, 114, ppl699-1707, (2004)
24. Gauderman W. J., Avol E., Gilliland F., Vora H., Thomas D., Berhane K., Mc­
Connell R., Kuenzli N., Lurmann F., Rappaport E., Margolis H., Bates D., and 
Peters J., The effect of Air pollution on Lung development from 10 to 18 years of 
age, New Engl. J. Med., 351(11), ppl057-1067, (2000).
25. Yang C., Chang C., Chuang H., Ho C. and Tsai S., Evidence for Increased Risks 
of Preterm Delivery in a Population Residing near a Freeway in Taiwan, Archives 
o f Environmental Health, 58 (10), pp649-654, (2003).
26. Walters S. and Ayres J., Health Effects of Particulates, in Pollution: Causes, Ef­
fects and Control, Harrison R.M. (ed), The Royal Society of Chemistry, Cam­
bridge, 1996, pp 248-250,
165
27. Dockery D. and Pope A., Epidemiology of Acute Health Effects: Summary of 
Time-Series Studies, in Particles in Our Air: Concentration and Effects, Wilson R 
and Spengler J. D. (eds), Harvard University, Cambridge, 1996, pp 123-148.
28. Vedal S., Ambient Particles and Health: Lines that Divide, J. Air & Waste Man­
agement Assoc., 47, pp551-581 (1997)
29. Hornberg C., Maciuleviciute L., Seemayer N. H., and Kainka E., Induction of sis­
ter chromatid exchanges (SCE) in human tracheal epithelial cells by the fractions 
PM-10 and PM-2.5 of airborne particulates, Toxicol. Lett. 96, 97, pp215-220,
(1998)
30. Nriagu J., Jinabhai C. C., Naidoo R. and Coutsoudis A., Lead Poisoning of Chil­
dren in Africa, II. Kwazulu/Natal, South Africa, Sci. Total Environ., 197, ppl-11,
(1997).
31. Cohen B.S., Xiong J. Q., Fang C. and Li W., Deposition of Charged particles on 
Lung Airways, Health Phys., 74, pp554-560, (1998).
32. Reichhardt T„ Weighing the Health Risks of Airborne Particulates, Environ. Sci. 
Technol. 29 No.8 . pp360-364, (1995).
33. Pope A. and Dockery D., Epidemiology of Chronic Health Effects: Cross-Section­
al Studies, in Particles in Our Air: Concentrations and Effects, Wilson R and 
Spengler J. D. (eds), Harvard University, Cambridge, 1996, pp 149-167.
34. Zhang J., Song H., Tong S., Li L., Liu B., and Wang L., Ambient Sulfate Concen­
tration and Chronic Disease Mortality in Beijing, Sci. Total Environ., 262, pp63- 
71,(2000).
166
35. Kiehl J. T. and Briegleb B. P., The Relative Roles of Sulfate Aerosols and green­
house Gases in Climate Forcing, Science, 260, pp311-314, (1993).
36. Coakley Jr. J. A., Cess R. D. and Yurevich F. B., The effect of tropospheric aero­
sols on the earth’s radiation budget: a parameterization for climate models, J. At­
mospheric Sciences, 40, ppl 16-138, (1983)
37. Hansen J. E., Sato M., Lacis A., Ruedy R., Tegen I. and Matthews E., Climate 
forcing in the industrial era, Proceedings o f  the National Academy o f Science o f 
the United State o f  America, 95, ppl2753-l 2758, (1998).
38. Charlson R. J. and Heintzenberg J., Aerosol Forcing o f  Climate, John Wiley, 
Chichester, 1995, pp403.
39. Floughton J. T„ Meira Filho L. G., Bruce J., Lee h., Callander B. A., Haites E., 
Harris N., Maskell K. (eds), Climate Change 1994: Radiative Forcing o f Climate 
Change and an Evaluation o f the IPCCIS92 Emission Scenarios, Cambridge Uni­
versity, 1995, ppl5.
40. Charlson R. J., Schwartz S. E., Hales J. M., Cess R. D., Coakley J. A., Hansen J. 
E., and Hofmann D. J., Climate forcing by Anthropogenic Aerosol, Science, 255, 
pp423-430, (1992).
41. Andreae M. O., Raising dust in the greenhouse. Nature, 380, pp 389-390, (1996).
42. Sokolik I. N. and Toon O. B., Direct Radiative Forcing by Anthropogenic Air­
borne Mineral Aerosols, Nature, 381, pp681-683, (1996).
43. Langmann B., Herzog M. and Graf H., Radiative Forcing of Climate by Sulfate 
Aerosols as Determined by a Regional circulation Chemistry Transport Model, 
Atmospheric Environment, 32 No.16, pp2757-2768, (1998).
167
44. Jacobson M. Z., Strong radiative heating due to the mixing state of black carbon 
in atmospheric aerosols, Nature 409: pp695-697, (2001).
45. National Aeronautics and Space Administration (NASA), NASA ties El Nino in­
duced drought to air pollution from fires,
[http://www.nasa.gov/home/hqnews/2003/apr/HP_news_03128.html]
46. Nash T. H., The Effect of Air Pollution on Other Plants, Particularly Vascular 
Plants, in Air Pollution and Lichens, Ferry B. W., Baddeley M. S. and Hawks- 
worth D. L. (eds), The Athlone Press of the University of London, London, 1973, 
pp208-209.
47. Gytarsky M. L., Karaban R. T., Nazarov I. M., Sysygina T. I. and Chemeris M. 
V„ Monitoring of Forest Ecosystems in the Russian Subarctic: Effects of Industri­
al Pollution, Sci. Total Environ., 164, pp57-64, (1995).
48. Mankovska B. and Steinnes E., Effects of Pollutants from an Aluminium Reduc­
tion Plant on Forest Ecosystems, Sci. Total Environ., 163, ppl 1-23, (1995).
49. Vike E., Air-Pollutant Dispersal Patterns and Vegetation Damage in the Vicinity 
of Three Aluminium Smelters in Norway, Sci. Total Environ., 236, pp75-90,
(1999).
50. Clarke A.G., Azadi-Boogar G.A and Andrews G.E., Particle Size and Chemical 
Composition of Urban Aerosols, Sci. Total Environ., 235, ppl 5-24, (1999).
51. Landsberger S. and Biegalski S., Analysis of Inorganic Particulate Pollutants by 
Nuclear Methods, in The Handbook o f Environmental Chemistry: Airborne Par­
ticulate Matter, Kouimtzis T. and Samara C. (eds.), Springer, Berlin, 1995, 
pp 175—198.
168
52. Kent G. S., Trepte C. R., Skeens K. M., Winker D. M., LITE and SAGE II Meas­
urement of Aerosol in the Southern Hemisphere Upper Troposhere, J. Geophys. 
Res., 103, D15, 19 111-19 127, (1998)
53. Baumbach G„ Vogt U., Hein K. R. G., Oluwole A. F., Ogunsola O. J., Olaniyi H. 
B. and Akeredolu F. A., Air pollution in a large tropical city with high traffic 
density — results of measurements in Lagos, Nigeria, Sci. Total Environ., 169, 
pp25-31, (1995).
54. World Bank, Rural Energy and Development: Improving energy supplies for two 
billion people, The World Bank, 1996, pp26.
55. Jayarantne E. R. and Verma T. S., The impact of biomass burning on the environ­
mental aerosol concentration on Gaborone-Botswana, Atmospheric Environment, 
35, ppl 821-1828, (2001).
56. Mbuligwe S. E. and Kassenga G. R., Automobile air pollution in Dare es Salaam 
City -Tanzania, Sci. Total Environ., 199, pp227-235, (1997).
57. Cooper W. C., The health implications of increased manganese in the environ­
ment resulting from the combustion of fuel additives: a review of the literature, J. 
Toxicol. Environ. Health, 14, pp23-46, (1984).
58. Loranger S. and Zayed J., Environmental and occupational exposure to man­
ganese: a multimedia assessment, Int. Arch. Occup. Environ. Health, 67, pp 101- 
110,(1995).
59. Sierra P., Loranger S., Kennedy G. and Zayed J., Occupational and environmental 
exposure of automobile mechanics and nonautomotive workers to airborne man­
169
ganese arising from combustion of methylcyclopentadienyl manganese tricar­
bonyl (MMT), ,4m. In. Hyg. Assoc. J., 56, pp713-716, (1995).
60. Ferraz H. B., Bertolucci P. H. F., Pereira J. S., Lima J. G. C., Andrade L. A. F.. 
Chronic Exposure to the fungicide maneb may produce symptoms and signs of 
CNS manganese intoxication, Neurology, 38, pp550-553, (1988).
61. Roels FI., Lauwerys R., Buchet J. P., Genet P., Sarhan M. J., Hanotiau I., deFays 
M., Bernard A., Stanescu D., Epidemiological survey among workers exposed to 
manganese: effects on lung, central nervous system, and some biological indices, 
Am. J. Ind. Med., 11, pp307-327. (1987).
62. Wang D. X., Zhou W. M., Wang S. Z., Zheng W., Occupational exposure to man­
ganese in welders and associated neurodegenerative diseases in China, Toxicol. 
Sci.. 42, pp24, (1998).
63. Zheng W., Ren S. and Graziano J. H., Manganese inhibits mitochondrial acon- 
itase: a mechanism of manganese neuroxicity, Brain Research, 799, pp334-342,
(1998).
64. http://:www.epa.gov/ttnatw01/hlthef/manganese.htlm
65. Barbeau A.. Inoue N. and Cloutier T., Role of Manganese in dystonia, Adv. Neur­
ol., 14, pp339-352, (1976).
6 6 . Mena I., Court J., Fuenzalida S., Papavasiliou P. S., and Cotzias G. C., modifica­
tion of chronic manganese poisoning treatment with L-dopa or 5-OH tryptophane, 
New Engl. J. Med., 282, pp5-10, (1970).
67. Tepper L. B., Hazards to Health: Manganese, New Engl. J. Med., 264, pp347-348, 
(1961).
170
6 8 . Schwanghart W. and Schiitt B., meteorological causes of Harmattan dust in West 
Africa, Geomorphology, 95, pp412-428, (2008)
69. Levin Z, Ganor E. and Gladstein V., The effects of desert particles coated with 
sulfate on rain formation, J. o f  Appl. MeteoroL, 35, pp 1511-1523, (1996)
70. Yin Y., Levin Z., Reisin T. G. and Tzivion S., The effects of giant cloud condens­
ation nuclei on the development of a precipitation in convective clouds -  a numer­
ical study, Atmos. Res., 53, pp91-116, (2000)
71. Rosenfeld D.. Rudich Y. and Lahav R., Desert Dust suppressing precipitation: A 
possible desertification feedback loop. Proceedings o f the National Academy o f 
Science o f the United State o f America, 98 (11), pp5975-5980, (2001)
72. Etyemezian V., Kuhns H., Gillies J., Green M., Pitchford M. and Watson J., 
Vehicle-based road dust emission measurement 1: Methods and Calibration, At­
mospheric Environment., 37, pp4559-4571, (2003).
73. Martens C. S., Wesolowski J. J., Harriss R. C. and Kaifer R, Chlorine loss from 
Puerto Rican and San Francisco Bay area marine aerosol, J. Geophys. Res., 78, 
pp8778-8792, (1973)
74. Mouri H., Nagao I., Okada K., Koga S. and Tanaka H., Elemental compositions 
of individual aerosol particles collected over the Southern Ocean: A case study, 
Atmos. Res., 43, ppl83-195, (1997).
75. United Nations, World urbanization prospects: the 2003 revision, United Nations, 
New York, 2004.
76. Mukoko S., On sustainable urban development in sub-Saharan Africa, Cities, 13, 
pp265-271, (1996).
171
77. Stren R and Halfani M., The cities of sub-Saharan Africa: from dependency to 
marginality, in Handbook o f Urban Studies, Paddison R. (ed.), Sage Publications, 
2001, pp466-485.
78. Cohen B., Urban growth in developing countries: a review of current trends and a 
caution regarding existing forecasts, World Development, 32, pp23-51, (2004).
79. Miller R. B. and Small C., Cities from space: potential applications of remote 
sensing in urban environmental research policy, Environmental Science and 
Policy, 6 , p p l29-137, (2003).
80. Tirabassi T., Listening out for urban air pollution, Atmospheric Environment, 33, 
pp4219- 4220, (1999).
81. Fenger J., Urban air quality, Atmospheric Environment, 33, pp4877- 4900, (1999).
82. Mage D., Ozolins G., Peterson P., Webster A., Orthofer R., Vandeweerd V. and 
Gwynne M., Urban air pollution in megacities of the world, Atmospheric Envir­
onment, 30, pp681- 6 8 6 , (1996).
83. Cuhadaroglu B. and Demirci E., Influence of some meteorological factors on air 
pollution in Trabzon city, Energy and Buildings, 25, ppl 79-184, (1997).
84. Romero H., Ihl M., Rivera A., Zalazar P. and Azocar P., Rapid urban growth -  
land use changes and air pollution in Santiago-Chile, Atmospheric Environment, 
33, pp4039-4047, (1999).
85. Carinanos P., Galan C., Alcazar P. and Dominguez E., Meteorological phenomena 
affecting the presence of solid particles suspended in the air during winter, Int. J. 
o f Biometeorology, 44, pp6-10, (2000).
172
8 6 . Panter B. C., Hooper M. A. and Tapper N. J., A comparison of air particulate mat­
ter and associated polycyclic aromatic hydrocarbons in some tropical and temper­
ate urban environments, Atmospheric Environment, 33, pp4087-4099, (1999).
87. Public Agenda Newspaper of 20th June 2005, under the heading Bloody Gold, as 
corporate interest clashes with people’s right.
8 8 . Nyarko B.J.B.,Serfor-Armah Y., Akaho E. H. K. and Kyere A. W. K., Biomonit­
oring of trace-element air pollution in a gold mining area in Ghana using the gen­
eralized kO-standardization NAA method, International Conference on Isotopic 
and Nuclear Analytical Techniques for Health and Environment, IAEA- 
CSP22/CD,2004.
89. Nyarko B.J.B., Serfor-Armah Y., Akaho E.H.K., Adomako D. and Osae S., De­
termination of heavy metal pollution levels in lichens at Obuasi gold mining area 
in Ghana, J. o f Applied Scie. Tech. (JAST), 9 No. 1&2, pp215-228, (2004).
90. Pilinis C. and Pandis S. N., Physical, Chemical and Optical Properties of Aero­
sols, in The Handbook o f  Environmental Chemistry: Airborne Particulate Matter, 
Kouimtzis T. and Samara C. (eds.), Springer, Berlin, 1995, pp 102.
91. Chow J. C., Measurement Methods to Determine Compliance with Ambient Air 
Quality Standards for Suspended Particles, J. Air and waste Management Assoc., 
45, pp320-382, 1995.
92. Finlayson-Pitts B. J. and Pitts Jr. J. N., Chemistry o f  the Upper and Lower Atmo­
sphere: Theory, Experiments, and Applications, Academic Press, San Diego, 
2000, pp. 297, 349, 354, 383.
173
93. O’Dowd C. D., Lowe J. A. and Smith M. H., Observations and modeling of aero­
sol growth in marine stratocumulus -  case study, Atmospheric Environment, 33, 
pp3053-3062, 1999.
94. Colls J., Air Pollution — An Introduction, E & FN Spon, London, 1997, ppl48.
95. Zhuang H., Chan C. K., Fang M. and Wexler A. S., Size distributions of particu­
lates sulfate, nitrate, and ammonium at a costal site in Hong Kong, Atmospheric 
Environment, 33, pp843-853, 1999.
96. Haaf W. and Jaenicke R., Results of improved size distribution measurements in 
the Aitken range of atmospheric aerosols, J. Aerosol Sci., 11, pp321-330, (1980).
97. Foltescu V. L., Fine Atmospheric Particles: Formation, Transport and Depos­
ition, PhD thesis, Chalmers University of Technology -  Sweden, 1995.
98. Kerminen V-M. and Wexler A. S., Post-fog nucleation of H2S0 4-H20  particles in 
smog, Atmospheric Environment, 28 (15), pp2399-2406, (1994).
99. Blando J. D. and Turpin B. J., Secondary organic aerosol formation in cloud and 
fog droplets -  A literature review of plausibility, Atmospheric Environment, 34, 
pp 1623-1632, (2000).
100. Lavanchy V. H. M., Gaggeler H. W., Nyeki S. and Baltensperger U., Ele­
mental Carbon (EC) and Black Carbon (BC) measurements with thermal methods 
and an aethalometer at the high-alpine research station Jungfraujoch, Atmospheric 
Environment, 33, pp2759-2769, (1999).
101. Mark D., Atmospheric Aerosol Sampling, In Atmospheric Particles. Har­
rison RM, van Grieken R (ed). Wiley: Chichester; 1998, pp31-42.
174
102. Zufall M. J. and Davidson C. I., Dry Deposition of Sulphur at High-Alti- 
tude Background Station in South Africa, In Atmospheric Particles. Harrison RM, 
van Grieken R(ed). Wiley: Chichester; 1998, pp426-463.
103. Seinffeld J. H. and Pandis S. N., Particulates Matter (Aerosols), in Atmo­
spheric Chemistry and Physics: From Air Pollution to Climate Change, John 
Wiley, New York, 1998, pp97-108.
104. McMurry P. H., A review of atmospheric aerosol measurements, Atmo­
spheric Environment, 34, 1959-1999, 2000.
105. International Atomic Energy Agency, Sampling and Analytical Methodo­
logies for Instrumental Neutron Activation Analysis of Airborne Particulate Mat­
ter, Training Course Series No. 4, Vienna, 1992
106. Wedding J. B.. Weigand M., John W. and Wall S., Sampling effectiveness 
of the inlet to the dichotomous sampler, Environ. Sci. Technol., 14, ppl367 -  
1370, (1980).
107. Liu B. Y. H. and Pui D. Y. H., Aerosol sampling inlets and inhalable 
particles, Atmospheric Environment, 15, pp589-600, (1981).
108. Esmen N. A. and Lee T. C., Distortion of cascade impactor measured size 
distribution due to bounce and blow-off, Am. In. Hyg. Assoc. ./., 41, pp410-419,
(1980).
109. Lawson D. R., Impaction surface coatings intercomparison and measure­
ments with cascade impactors, Atmospheric Environment, 14, pp 195-199, (1980).
175
110. Wang L., Yan Q. and Liu L., Effect of a stick on the flow field in a cyc­
lone and the pressure drop reduction mechanism, Aerosol Sci. Technol., 35, 
pp909-913, (2001).
111. Enliang L. and Yingmin W., A new collection theory of cyclone separat­
ors, AIChE Journal, 35 (4), pp6 6 6  -  669, (1989).
112. Wang L. and Ye L., Reducing pressure drop in cyclones by a stick, Aero­
sol Sci. Technol., 31, ppl 87 -  193, (1999).
113. Mothes H. and Loffler F., Prediction of particle removal in cyclones separ­
ators, International Chemical Engineering, 28(2), 1988
114. Coker A. K., Understanding Cyclone design, Chemical Engineering Pro­
gress, D ecem ber , pp51 -55, (1993).
115. Air & Waste Management Association, Air Pollution Engineering Manual, 
Van Nostrand Reinhold, New York, NY, 1992
116. U. S. EPA Office of Air Quality Planning and Standards, Stationary 
Source Control Techniques Document for Fine Particulate Matter, EPA-452/R- 
97-001, Research Triangle, NC, October, 1998
117. Perry’s Chemical Engineer’s handbook, Robert Perry and Don Green
(eds.), 6 th Edition, McGraw-Hill, New York, NY, 1984.
118. U. S. EPA, Air Pollution Control Technology Fact Sheet EPA-452/F-03- 
005, 2003.
119. British Columbia Ministry of Water, Land and Air Protection, Particulate 
Matter in British Colombia - A report on PM10 and PM2.5 mass concentration up 
to 2000, May 2003, pp21.
176
120. Spurny K. R., Lodge Jr. J. P., Frank E. R. and Sheesley D. C., Aerosol fil­
tration by means of nuclepore filters structural and filtration properties. Environ. 
Sci. Techno!., 3, pp453-468, (1969).
121. Liu B.Y. H. and Lee K. W., Efficiency of membrane and Nuclepore filters 
for submicron aerosols, Environ. Sci. Technoi, 10, pp345-350, (1976).
122. John W. and Reischl G., Measurement of the filtration efficiencies of se­
lected filter types, Atmospheric Environment, 12, pp2015-2019, (1978).
123. Dzubay T. G. and Barbour R. K., A method to improve adhesion of aero­
sol particles on Teflon filters, J. Air Pollution Contr. Assoc., 33, pp692-695, 
(1983).
124. Watson J. G. and Chow J. C., Clear Sky Visibility as a Challenge for So­
ciety, Annual Rev. Energy Environ., 19, pp241-266, (1994).
125. Brimblecombe P., The Big Smoke: A Flistory of Air Pollution in London 
since Medieval Times, Methuen, London (1987).
126. Engelbrecht D. R., Cahill T. A. and Feeney P. J., Electrostatic effects on 
Gravimetric Analysis of Membrane Filters, Journal o f  Air Pollution Control As­
sociation, 30, pp391-392, (1980).
127. Witz S., Eden R. W., Wadley M. W., Dunwoody C., Papa R. P. and Torres 
K. J., Rapid Loss of Particulate Nitrate, Chloride and Ammonium on Quartz Fiber 
Filters during storage, J. Air Waste Manage. Assoc., 40, pp53-61, (1990).
128. Witz S., Eden R. W., Liu C. S. and Wadley M. W., Water Content of Col­
lected Aerosols in the South Coast and Southeast Desert Air Basins, Journal o f  
Air Pollution Control Association, 38, pp418-419, (1988).
177
129. Begum B. A., Kim E„ Jeong C-H., Lee D-W. and Hopke P. K , Evaluation 
of the potential source contribution function using the 2002 Quebec forest fire 
episode, Atmospheric Environment, 39, pp3719-3724, 2005
130. Wierzbicka A., Lillieblad L., Pagels J., Strand ML, Gudmundsson A., 
Swietlicki E., Sanati M. and Boghard M., Aerosol optical, chemical and physical 
properties at Gosan, Korea during the Asian dust and pollution episode in 2001, 
Atmospheric Environment, 39, pp39-50, 2005
131. Duan F., Liu X., Yu T. and Cachier H, Identification and estimate of bio­
mass burning contribution to the urban aerosol organic concentrations in Beijing, 
Atmospheric Environment, 38, pp l275-1282, 2004
132. Wang Y., Zhung G., Tang A., Yuan H., Sun Y, Chen S. and Zheng A., 
The ion chemistry and the source of PM2 5 aerosol in Beijing, Atmospheric Envir­
onment, 39, pp3771-3784, 2005
133. Yli-Toumi T., Hopke P., Paatero P., Basumia M. S., Landsberger S., 
Viisanen Y., Paatero J., Atmospheric aerosol over Finnish Arctic: Source analysis 
by the multilinear engine and the potential source contribution function, Atmo­
spheric Environment, 37, pp4381-4392, 2003
134. Graham B., Falkovich A. H., Rudich Y., Maenhaut W., Guyon P. and An- 
dreae M. O., Local and regional contribution to the atmospheric aerosol over Tel 
Aviv, Israel: a case study using elemental, ionic and organic tracers, Atmospheric 
Environment, 38, p p l593-1604, 2004
178
135. Venkataraman C., Reddy C. K., Josson S. and Reddy M. S., Aerosol size 
and chemical characteristics at Mumbai, India during the INDOEX-IFP (1999), 
Atmospheric Environment, 36, pp 1979-1991 , 2002
136. Moloi K., Chimidza S., Selin Lindgren E., Viksna A. and Standzenieks P., 
Black carbon, mass and elemental measurements of airborne particles in the vil­
lage of Serowe, Botswana, Atmospheric Environment, 36, pp2447-2457, 2002
137. Jenkins Ron, Gould R.W. and Gedcke Dale, Quantitative X-ray Spectro­
metry, Mercel Dekker Inc, New York and Basel, 1981, ppl.
138. Jaklevic J. M. and Giauque R. D., Energy-Dispersive X-ray Fluorescence 
Analysis using X-ray tubes Excitation, in Handbook of X-ray Spectrometry: 
Methods and Techniques, van Grieken R. E. and Markowicz A. A. (eds), Marcel 
Dekker, New York, 1993, pp 151-180
139. Jenkins Ron, Gould R.W. and Gedcke Dale, Quantitative X-ray Spectro­
metry, Mercel Dekker Inc, New York and Basel, 1981, pp35.
140. Aboh I. J. K., Developing a Model for Multi-element Analysis of Prestea 
Gold Ore using Energy Dispersive X-Ray Fluorescence Technique- M. Phil Thes­
is submitted to the University of Ghana, Legon, 1990
141. Dzubay T. G. and Nelson R. O., Self absorption corrections for x-ray ana­
lysis of aerosol, Advances in X-ray Analysis, 18, pp619-631, (1975)
142. Pierson W. R., Brachaczek W. W., Gorse R. A. Jr., Japar S. M., Norbeck 
J. M. and Keeler J. G., Atmospheric acidity measurements on Allegheny Moun­
tain and the origin of ambient acidity in northeastern United States, Atmospheric 
Environment, 23, pp431-459, (1989).
179
143. Shah J. J., Johnson R. L., Heyerdahl E. K., and Huntzicker J. J., Aerosol 
chemical composition and light scattering in Portland, Oregon: the role of carbon, 
Atmospheric Environment, 18, pp235-240, (1984).
144. Japar S. M. and Szkarlat A. C., Measurement of diesel vehicle exhaust 
particulates using photoacoustic spectroscopy, Combustion Science and Techno­
logy, 24, pp215-219,(1981).
145. Japar S. M., Szkarlat A. C. and Gorse R. A. Jr., Optical properties of par­
ticulate emissions from on-road vehicles, Atmospheric Environment, 15, pp2063- 
2070,(1981)
146. Huntzicker J. J., Heyerdahl E. K., McDow S. R., Rau J. A., Griest W. H. 
and MacDougall C. S., Combustion as principal source of carbonaceous aerosol in 
the Ohio River Valley, Journal o f  Air Pollution Control Association, 36, pp705- 
709,(1986).
147. Gray H. A., Cass G. R., Huntzicker J. J., Heyerdahl E. K. and Rau J. A., 
Characteristics of atmospheric organic and elemental carbon particle concentra­
tions in Los Angeles, Environmental Science and Technology, 20, pp580-589, 
(1986).
148. Chow J. C., Waston J. G., Prichett L. C., Pierson W. R., Frazier C. A., and 
Purcell R. G., The Dri Thermal/Optical reflectance carbon analysis system: de­
scription, evaluation and applications in U. S. air quality studies, Atmospheric En­
vironment, 27A(8), ppl 185-1201, (1993).
149. Waston J. G., Chow J. C., Lowenthal D. H., Pritchett L. C., Frazier C. A., 
Neuroth G. R. and Robbins R., Differences in carbon compositionof source pro­
180
files for diesel- and gasoline-powered vehicles, Atmospheric Environment, 28, 
pp2493-2505, (1994).
150. Gagel A., Simultaneous black smoke and airborne particulate emission 
measurement by means of an automated combined instrument, VDI-Report, 1257, 
pp631-645, 1996
151. Bern H, Gallorini M, Rizzio E. and Krzeminska M., Comparative studies 
of the concentrations of some elements in the urban air particulate matter in Lodz 
City of Poland and in Milan, Italy, Environment International, 29, pp423-428, 
(2003).
152. Landsberger S., Improved Methodology for the Determination of the Sev­
en Elemental Tracer Long-Distance Pollution signatures Using Thermal and epi- 
thermal Neutron Activation Analysis, Analytical Chemistry, 60, ppl842-1845, 
(1988).
153. Var F., Narita Y. and Tanaka S., The concentration, trend and seasonal 
variation of metals in the atmosphere in 16 Japanese cities shown by the results of 
National Air Surveillance Network (NASN) from 1974 to 1996, Atmospheric En­
vironment, 34, pp2755-2770, (2000).
154. Heller-Zeiller S. F., Borgoul P. V., Moore R. R., Smoliar M., Suarez A. E. 
And Ondov J. M., Comparison of INAA and RNAA methods with thermal-ioniz- 
ation mass spectrometry for iridium determinations in atmospheric tracer studies, 
Journal o f Radioanalytical and Nuclear Chemistry, 244(1), pp93-96, (2000).
155. Hashimoto Y., Sekine Y., Kim H. K., Chen Z. L. and Yang Z. M., Monit­
oring of atmospheric aerosol components by multielement neutron activation ana­
181
lysis in Seoul, Korea, April 1987 -  March 1989, Journal o f Japan Society o f  Air 
Pollution, 25, pp330-323, (1990).
156. Guinn V. P., Nuclear Analytical Methods in the Life Sciences, Zeisler, R., 
and Guinn, V. P. (eds), Humana Press, Clifton, NJ, 1990, pp 1-7, 1990.
157. Ehmann W. D. and Vance D. E., Radiochemistry and Nuclear Methods of 
Analysis, Wiley, Toronto, 1991, pp271-272.
158. DeSoete, D., Gijbels, Hoste, J., Neutron Activation Analysis, Wiley-Inter- 
science, Toronto, 1972, pp445-447.
159. El Nimr T., De Corte F., Moens L., Simonts A. and Hoste J., Epicadmium 
neutron activation analysis (ENAA) based on the kO comparator method, Journal 
o f Radio analytical Chemistry, 67, pp421-435, (1981).
160. Allen C. C., Desert varnish of Sonoran Desert -  optical and electron mi­
croprobe analysis, Journal o f  Geology, 86, pp743-752, (1978)
161. Casuccio G. S., Janocko P. B., Lee R. J., Kelly J. F., Dattner S. I. and 
Mgebroff J. S., The use of computer controlled scanning electron microscopy in 
environmental studies, Journal o f Air Pollution Control Association, 33, pp973- 
943, (1983).
162. Wernisch J., Quantitative electron microprobe analysis without standards, 
X-ray Spectrometry, 14, pp 109-119, 1985
163. Davis B. L., Additional suggestions for x-ray quantitative analysis of high- 
volume filters, Atmospheric Environment, 12, pp2403-2406, (1978)
164. Davis B. L., Standardless x-ray diffraction quantitative analysis, Atmo­
spheric Environment, 14, pp217-220, (1980)
182
165. Davis B. L., A study of the errors in x-ray quantitative analysis procedures 
for aerosol collected on filter media, Atmospheric Environment, 15, pp291-296,
(1981)
166. Davis B. L., Johnson L. R., Stevens R. K., Courtney W. J. and Safriet D. 
W., The quartz content and elemental composition of aerosols from selected sites 
of the EPA inhalable particulate network, Atmospheric Environment, 18(4), 
pp771-782, (1984)
167. Elsma R., Vermeulen A. T. and Kieskamp W. M., Determination of 
European emission using concentration and isotopic measurements, Environment­
al Mon. Assess., 31(1-2), ppl97-202, (1995)
168. Goldstein S. L., O’Nions R. K. and Hamilton P. J., A Sm-Nd isotopic 
study of atmospheric dusts and particulates from major river systems, Earth Plan­
et Sci. Lett., 70, pp221-236, (1984)
169. Johnson B. J and Dawson G. A., A preliminary study of carbon isotopic 
content of ambient formic acid and two selected sources: automobile exhaust and 
formicine ants, J. Atmos. Chem., 17(2), ppl23-140, (1993)
170. Kelly W. R., Chen L-T., Gramlich J. W. and Hehn K. E., Determination of 
sulphur in low sulphur steels isotope dilution thermal ionization mass spectro­
metry, Analyst, 115, pp 1019-1024, (1990)
171. Kelly W. R., Paulsen P. J., Murphy K. E., Paulsen P. J., Vocke R. D. Jr. 
and Chen L-T., Determination of sulphur in fossil fuels by isotope dilu­
tion-thermal ionization mass spectrometry, Anal Chem, 66, pp2505-2513, (1994)
183
172. Lin Z-C., Ondov J. M., Kelly W. R., Paulsen P. J. and Stevens R. K , Tag­
ging diesel and residential oil furnance emissions in Roanoke, VA, with enriched 
isotopes of Samarium, J. Air Waste Mange. Assoc., 42, ppl 057-1062, (1992)
173. Lin Z-C., Ondov J. M.and Kelly W. R., Tracing emissions from coal-fired 
power plants with enriched rare-earth isotopes, FUEL, 72, pp697 (1993)
174. Ondov J. M., Kelly W. R., Holland J. Z., Lin Z-C., and Wight S. A., Tra­
cing fly ash emitted from a coal-fired power plant with enriched rare-earth iso­
topes, An urban scale test, Atmospheric Environment, 26B, pp453-462, (1992).
175. Sohns E. Gerling P. and Faber E., Improved stable nitrogen isotope ratio 
measurements of natural gases., Analytical Chemistry, 66(17), pp2614-2620, 
(1994)
176. Currie L. A., Klouda G. A. and Voorhees K. J., Atmospheric carbon: the
importance of accelerator mass spectrometry, Nucl. Instr. Methods, 233, pp371- 
379, (1984)
177. Klouda G. A., Currie L. A., Donahue D. J., Jull A. J. T. and Naylor M. H., 
Urban atmosphere 14CO and 14CH4 measurements by accelerator mass spectro­
metry, Radiocarbon, 28(2A), pp625-633, (1986)
178. Linick T. W., Jull A. J. T., Toolin I. J. and Donahue D. J., Operation of the 
NSF-Arizona accelerator facility for radioisotope analysis and results from selec­
ted collaborative research projects, Radiocarbon, 28, pp522, (1986)
179. Verkouteren R. M., Klouda G. A., Currie L. A., Donahue D. J., Jull A. J. 
T. and Linick T. W., Preparation of microgram samples on iron wool for radiocar­
184
bon analysis via accelerator mass spectrometry: a closed-system approach, Nucl. 
Instr. Methods, B294, ppl-4 (1987)
180. Venegas L. E. and Mazzeo N. A., Design methodology for background air 
pollution monitoring site selection in an urban area, International Journal Envir­
onment and Pollution, 20(1-6), pp 185-195 (2003).
181. Wu. H. W. Y. and Chan L. Y., Comparative study of air quality surveil­
lance networks in Hong Kong, Atmospheric Environment, 31, pp935-945, (1997)
182. Seinfeld J. H., Optimal location of pollutant monitoring stations in an air­
shed, Atmospheric Environment, 6, pp847-858, (1972)
183. Noll K.. E., Miller T. L., Norco J. E. and Raufer R. K., An objective air
monitoring site selection methodology for large point sources, Atmospheric En­
vironment, 11, pp 1051 -1059, (1977)
184. Noll K. E. and Mitsutomi S., Design methodology for optimum dosage air
monitoring site selection, Atmospheric Environment, 17, pp2583-2590 (1983)
185. Mazzeo N. A. and Venegas L. E., Practical use of the ISCST3 model to se­
lect monitoring site locations for air pollution control. International Journal En­
vironment and Pollution, 14(1-6), pp246-259, (2000)
186. Haas T. C., Redesigning continental-scale monitoring networks, Atmo­
spheric Environment, 26, pp3323-3333, (1992)
187. Laursen J, Stikans M, Karlsen K, Pind N. A versatile and easy to handle 
EDXRF instrumentation. In. Proceedings o f European Conference on Energy 
Dispersive X-ray Spectrometry 1998. Fernandez JE, Tartari. A. (ed). Editrice 
Compositori, Bologna, Italy, 1999, pp 139-144.
185
188. Van Espen P., Janssens K. and Nobels J., AXIL-PC, Software for the 
analysis of complex x-ray spectra, Chemometrics and Intelligent Laboratory Sys­
tem, 1, ppl09-l 14, (1986)
189. Vekemans B., Janssens K., Vincze L., Adams F., Van Espen P., Analysis 
of X-ray Spectra by Iterative Least Squares (AXIL): New Developments, X-Ray 
Spectrom., 23, 278-285,( 1994)
190. Selin Lindgren E, Henriksson D, Lundin M. Therning P, Laursen J. and 
Pind N.,. Possible indicators for biomass burning in a small Swedish City as stud­
ied by energy dispersive x-ray (EDXRF) spectrometry, X-Ray Spectrom, 35, 
ppl9-26, (2006)
191. IAEA, QXAS 3.6 -  Quantitative X-ray Analysis System -  Developed Un­
der the Auspices of the International Atomic Energy Agency, Vienna, 2005 [Cited 
March 27, 2009].
http://www.iaea.org/OurWork/ST/NA/NAAL/pci/ins/xrf/pciXRFdown.php>
192. Zannetti P., Numerical simulation modeling air pollution: an overview, in 
Air Pollution, Zannetti et al (eds), Computational Mechanics Publications, 
Southampton, 1993, pp3-I4
193. Kukkonen J., Air quality modeling -  Dispersion models, in Monitoring 
Ambient Air Quality fo r  Health Impact Assessment, WHO Regional Publications, 
European Series No. 85, 1999, pp 155-160
194. Noorduk H., The national smog warning system in the Netherland: a com­
bination of measuring and modeling, in Air Pollution 2, Pollution control and
186
monitoring, Baldasano J. M. et al (ed), Southampton, Computational Mechanics 
Publications, 1994, pp35 -  42
195. Hopke P. K., Receptor Modeling for Air Quality Management, in Air 
Quality Management, Hester R. E. and Harrison R. M. (eds.), Royal Society of 
Chemistry, Cambridge, 1997, pp95 -  117
196. Henry R. C., Hopke P. K. and Williamson H. J., Review of Receptor Mod­
el fundamentals, Atmospheric Environment, 18, p p l507 - 1515, 1984
197. Henry R. C., Current Factor Analysis Receptor models are ill-posed, At­
mospheric Environment, 21, pp 1815 -  1820, 1987
198. Antilla P., Paatero P., Tapper U. and Jarvinen O., Source Identification of 
bulk wet deposition in Finland by Positive Matrix Factoration, Atmospheric En­
vironment, 29, ppl705 -  1718, 1995
199. Huang S., Rahn K. A. and Arimoto R., Testing and optimizing two factor- 
analysis techniques on aerosol at Narragansett, Rhode Island, Atmospheric Envir­
onment, 3 3 ,pp2169 -  2185, 1999
200. Thurston G. D. , Spengler J. S., A quantitative assessment of source con­
tribution to inhalable particulate matter pollution in Metropolitan Boston, Atmo­
spheric Environment, 19, pp9-25, (1985)
201. Prichard E. and Barwick V., Quality Assurance in Analytical Chemistry, 
Analytical Techniques in the Sciences, Wiley, 2007, ppl
202. Bower J. and Mucke H-G., Design, operation and quality assurance and 
control in a monitoring system, in Monitoring Ambient Air Quality tor Health Im­
187
pact Assessment, World Health Organisation Regional Publications, European 
Series No. 85, 1999, pp41
203. Bower J., Ambient Air Quality Monitoring, in Issues in Environmental 
Science and technology 8 - Air Quality Management, Hester R. E. and Harrison 
R. M. (eds.), Great Britain Royal Society of Chemistry, 1997, pp43-44
204. Maenhaut W. Salma I., and Cafmeyer J., Regional atmospheric aerosol 
composition and sources in the Eastern Transval, South Africa, and impact of bio­
mass burning, J. Geophys. Res., 101 D19, pp23631-23650, 1996.
205. Chan Y. C., Vowles P. D., McTainsh G. H., Simpson R. W., Cohen D.D., 
Bailey G. M., and McOrist G. D., Characterisation and source identification of 
PM10 aerosol collected with a high volume cascade impactor in Brisbane (Aus­
tralia), Scie. Tot Environ., 262, pp5-19, 2000.
206. Alves L. C., Reis M. A. and Freitas M. C., Air particulates matter charac­
terization of rural area in Portugal, Nucl. Instrum. Methods Phys. Res., B136-138, 
pp941-947, 1998
207. Cheng L., Sandhu H. S., Angle R. P., McDonald K. M., and Myrick R. H., 
Rural particulate matter in Alberta, Canada, Atmospheric Environment, 34(20), 
pp3365-3372, 2000
208. Houthuijs D., Breugelmans O., Hoek G., Vaskovic E., Mihalikova E., 
Pastuszka J. S., Jirik V., Sachelarescu S., Lolova D., Meliefste K., Uzunova E., 
Marinescu C., Volf J., de Leeuw F., van de Wiel H., Fletcher T., Lebret E. and 
Brunekreef B., PM10 and PM2.5 concentrations in Central and Eastern Europe: 
results from the Cesar study, Atmospheric Environment, 35, 2757-2771, 2001
209 . W o r ld  H e a lth  O rg a n is a t io n ,  w w w .w h o . in t /m e d ia c e n t r e / fa c t s h e e t / fs313/ e n , 
2006
210. Suglia S. F., Gryparis A., Wright R. O., Schwartz J. and Wright R. J., As­
sociation of black carbon with cognition among children in a prospective birth co­
hort study, American Journal Epidemilogy, 167(3), 280-286, 2008
211. Ramanathan V., Transported black carbon a significant player in pacific 
ocean climate, ScienceDaily, March 27, 2008
212. Bennet C., Jonasson P., and Selin Lindgren E., Concentrations and sources 
of trace elements in particulate air pollution, Dar es Salam, Tanzania, studied by 
EDXRF, X-ray Spectrometry, 34, 1-6, 2005
213. El-Tahir H. M., Selin Lindgren E., and Habbani F. I., Elemental character­
ization of airborne particles in Khartoum, Sudan, X-ray Spectrometry, 34, 144- 
152,2005
214. Viksna A., Selin Lindgren E., Standzenieks P. and Jacobsson J., EDXRF 
and TXRF analysis of elemental size distributions and environmental mobility of 
airborne particles in the city of Riga, Latvia, X-ray Spectrometry, 33, 414-420, 
2004
215. Gomiscek B., Frank A., Puxbaum H., Stopper S., Preining O. and Hauck 
H., Case study analysis of PM burden at an urban and rural site during the 
AUPHEP, Atmospheric Environment, 38, 3935-3948, 2004
216. Maenhaut W. and Cafmeyer J., Long-term atmospheric aerosol study at 
urban and rural sites in Belgium using multi-elemental analysis by particle-in­
189
duced x-ray emission spectrometry and short-irradiation instrumental neutron ac­
tivation analysis, X-ray Spectrometry, 27, 236-246, 1998
217. SPSS, Factor analysis, in SPSS for Windows version 16,2007
218. Aboh I. J. K., Flenriksson D., Laursen J., Lundin M., Pind N., Selin Lind­
gren E., Wahnstrom T., EDXRF characterisation of elemental contents in PM2.5 
in a medium-sized Swedish city dominated by a modern waste incineration plant, 
X-Ray Spectrometry, 36, 104-110, (2007)
219. Morishita M, Keeler G.J, Wagner, J. G, Harkema, J. R., Source identifica­
tion of ambient PM2.5 during summer inhalation exposure studies in Detroit, MI, 
Atmospheric Environment', 40: 3823-3834, (2006)
220. Wong S., Colareo P. R. and Dessler A. E., Principal component analysis of 
the evolution of the Saharan air layer and dust transport: Comparisons between a 
model stimulation and MODIS and AIRS retrievals, J. Geophys. Res., I l l ,  
D20109 (1-12), (2006).
221. Bou Karam D., Flamant C., Knippertz P., Reitebuch O., Pelon J., Chong 
M. and Dabas A., Dust emissions over the Sahel associated with the West African 
monsoon intertropical discontinuity region: A representative case, Q. J. R. Met- 
eorol. Soc., 134, pp621 -  634, (2008)
222. Antoine D. and Nobileau D., Recent increase of Sahara dust transport over 
the Mediterranean Sea, as revealed from ocean color satellite (SeaWIFS) observa­
tions, J. Geophys. Res., I l l ,  D12214 (1-19), (2006).
190
223. Caminade C., Terray L. and Maisonnave E., West African monsoon re­
sponse to greenhouse gas and sulphate aerosol forcing under two emission scen­
arios, Climate Dynamics, 26, pp531 -  547, (2006).
224. Ogunjobi K. O., He Z. and Simmer C., Spectral aerosol properties from 
AERONET Sun-photometric measurement over West Africa, Atmospheric Re­
search, 88, pp89 -  107, (2008).
225. Oblad M. and Selin E., Measurement of elemental composition in back­
ground aerosol on the West Coast of Sweden, Atmospheric Environment, 20, 
pp 1419-1432, (1986).
191