Jim Anderson, Arizona State University -  5/6/99
Contact me at janderson@asu.edu for additional details.


Individual particle data from Cape Grim samples collected during ACE-1.

SAMPLING:  These are from dual in-line 47 mm filters collected in most cases over 24-hour periods with sector control (the extended Sector 2).  Top filter was Nuclepore with 8.0 micron pores ("coarse" samples) and the bottom filter was Nuclepore with 0.2 micron pores ("fine" samples).  Flow rate was typically 17 l/s.

The sample names (column 12)day the sample was started; 23 through 28 were in November, 4 through 14 were in December.

ANALYTICAL METHODS:  A 1 cm by 1 cm section of each fine and coarse sample was mounted on a 12 mm aluminum SEM stub using conductive-carbon double-stick tape.  Samples were coated with 200 Angstron thick carbon to provide electrical conductivity.  Samples were analyzed using an automated JEOL 5800 scanning electron microscope interfaced to a NORAN Voyager IV X-ray/image analysis system.  X-ray spectra were collected using a NORAN Pioneer EDS detector with a Norvar thin window.  

The compositional data are not listed here except that each particle is assigned to a particle type based on composition.  The compositional type of the particles are indicated by the cluster number (column 11).

SIZE DATA:
The dimension of all sizes (Length, Width, Perimeter) is microns.  Column 3 has the 2-dimensional area of each particle in square microns.  The aspect ratio (column 5) and circularity (column 9) are dimensionless and calculated form other parameters.  Column 10 has the squareroot of particle area, a better measure of particle size than average diameter for complex shapes.

COMPOSITIONAL TYPES (Cluster assignments in column 11):

Compositions are in atomic fractions using only those elements analyzed.

Below are the clusters that fell out a group of 23,000+
particles. K-means cluster conditions were that at least 0.1% of
the particles had to have compositions that fell within a 20-
degree hypercone in compositional multispace (24 elements).  The
clustering was iterated until stability was achieved. 

Problems with minor elements in cluster compositions: I use a
fairly conservative set of detection limits for all of the
elements.  Detection limits are applied to corrected weight
percents and any value below a detection limit is set to 0.0. 
This creates a bit of a problem for the smallest particles (0.1
to maybe 0.4 microns diameter) because the minor elements in
something like sea salt tend to get set to 0.0.  (particles like
sea salt with diameters less than about 2 microns are semi-
transparent with regard to primary electron penetration and
therefore produce compositional totals less than 100% - the
smaller the particle, the bigger the effect) A good example is in
the first set of clusters, C1 through C3.  These should be lumped
into a single particle type.

Another thing to be aware of in regard to minor elements in a
cluster composition is the effect of aggregation.  A few
particles in a cluster may be aggregated with something else and
the rest of the particles not aggregated.  The cluster
composition is an average, so whatever is in the aggregations
gets averaged through the population.  Of course the individual
compositions in the cluster have to be within the defined limits.

I have grouped clusters together below based on shared
compositional features.  In many cases, similar particle types
will be lumped together for use in any papers, so there really
are less particle types than the 35 clusters might suggest.
Percents listed are % to total particles in any cluster.

A. (C1-C3 should be combined into 1 type)
C1. Na.44 Cl.48 Mg.023 S.025 K.009 Ca.011 (13.5%) S/Cl=.05
C2. Na.50 Cl.47 Mg.003 S.010 K.004 Ca.005 (40.1%) S/Cl=.02
C3. Na.56 Cl.40 Mg.006 S.014 K.003 Ca.005 (12.7%) S/Cl=.03

B.
C5. Na.47 Cl.41 S.055 Mg.022 Ca.024 K.009 (6.6%) S/Cl=.13

C.
C14 Na.64 S.28 Cl.02 Mg.01 Ca.02 (0.3%)

D.
C18 Na.45 S.50 Mg.02 Cl.01 (0.2%)

E.
C30. Na.08 Mg.01 S.87 Ca.02 (0.4%)

F.
C7. Na.39 Cl.35 S.12 Mg.02 Ca.08 (1.6%) S/Cl=.35

G.
C11. Na.47 S.25 Cl.11 Ca.14 (0.3%)

H.
C33. Na.01 S.52 Cl.02 Ca.43 (0.7%)

I. 
C19. Na.18 S.39 Cl.04 Ca.33 Mg.02 K.02 (0.6%)

J.(aggregates - should lump together)
C10. Na.13 S.34 Cl.10 Ca.29 Mg.08 K.03 (0.1%)
C12. Na.33 S.20 Cl.27 Ca.14 Mg.02 (1.2%)
C17. Na.24 S.29 Cl.17 Ca.22 Mg.03 K.02 (0.6%)

K.(maybe a continuous series) (these look like points spread
throughout a volume in compositional multispace.  I should
probably call this a single type, but define by the range of
composition rather than a single composition)

C20. Na.11 Mg.17 S.22 Cl.28 K.03 Ca.16 (0.4%)
C21. Na.10 Mg.18 S.28 Cl.18 K.02 Ca.21 (0.7%)
C22. Na.11 Mg.09 S.34 Cl.12 K.02 Ca.29 (1.0%)
C24. Na.04 Mg.31 S.18 Cl.30 K.03 Ca.11 (0.2%)
C26. Na.03 Mg.30 S.26 Cl.21 K.02 Ca.16 (0.5%)
C28. Na.02 Mg.31 S.19 Cl.32 K.04 Ca.10 (0.6%)
C31. Na.01 Mg.13 S.34 Cl.24 K.01 Ca.25 (0.7%)

L.(at least three types here)
C13. Na.30 Mg.20 S.32 Cl.11 Ca.04 (0.1%)
C23. Na.08 Mg.28 S.09 Cl.46 K.04 Ca.02 (0.4%)
C25. Na.02 Mg.28 S.21 Cl.27 K.18 Ca.02 (0.1%)
C27. Na.02 Mg.26 S.21 Cl.28 K.19 Ca.01 (0.4%)
C32. Mg.47 S.09 Cl.37 K.02 Ca.02 (0.5%)
C34. Mg.45 S.15 Cl.20 K.17 Ca.01 (0.1%)

M.
C4. Na.34 Mg.10 Cl.42 S.06 Ca.02 (1.3%)

N. (a type like this and also the excess Na type always pop up in
marine aerosols - they are NOT poorly analyzed sea salt)

C6. Na.28 Cl.66 S.02 (2.7%)

O.
C16. Na.32 Cl.43 S.17 Mg.01 K.01 Ca.03 (0.8%)

P.
C8. Na.59 Cl.28 S.08 Ca.02 (2.2%)
C9. Na.72 Cl.13 S.10 Ca.04 (0.5%)

Q.
C29. Na.02 Si.90 S.01 Cl.02 Cr.01 (0.3%)

R.
C35. Mg.04 Si.33 S.04 Cl.01 Cr.55 (0.1%)

___________________________________________________

Misc. minor clusters, all less than 0.1%