SEPA
United States Industrial Environmental Research EPA 600 7 79 192
Environmental Protection Laboratory August 1979
Agency Research Triangle Park NIC 2771 1
Calcium Sulfite Crystal
Sizing Studies
Interagency
Energy/Environment
R&D Program Report
-------
RESEARCH REPORTING SERIES
Research reports of the Office of Research and Development, U.S. Environmental
Protection Agency, have been grouped into nine series These nine broad cate-
gories were established to facilitate further development and application of en-
vironmental technology Elimination of traditional grouping was consciously
planned to foster technology transfer and a maximum interface in related fields.
The nine series are:
1. Environmental Health Effects Research
2. Environmental Protection Technology
3. Ecological Research
4. Environmental Monitoring
5. Socioeconomic Environmental Studies
6. Scientific and Technical Assessment Reports (STAR)
7. Interagency Energy-Environment Research and Development
8. "Special" Reports
9. Miscellaneous Reports
This report has been assigned to the INTERAGENCY ENERGY-ENVIRONMENT
RESEARCH AND DEVELOPMENT series. Reports in this series result from the
effort funded under the 17-agency Federal Energy/Environment Research and
Development Program These studies relate to EPA's mission to protect the public
health and welfare from adverse effects of pollutants associated with energy sys-
tems. The goal of the Program is to assure the rapid development of domestic
energy supplies in an environmentally-compatible manner by providing the nec-
essary environmental data and control technology. Investigations include analy-
ses of the transport of energy-related pollutants and their health and ecological
effects; assessments of, and development of, control technologies for energy
systems; and integrated assessments of a wide'range of energy-related environ-
mental issues.
EPA REVIEW NOTICE
This report has been reviewed by the participating Federal Agencies, and approved
for publication. Approval does not signify that the contents necessarily reflect
the views and policies of the Government, nor does mention of trade names or
commercial products constitute endorsement or recommendation for use.
This document is available to the public through the National Technical Informa-
tion Service, Springfield, Virginia 22161.
-------
EPA-600/7-79-192
August 1979
Calcium Sulfite Crystal Sizing Studies
by
Larry O. Edwards
Radian Corporation
8500 Shoal Creek Boulevard
Austin, Texas 78766
Contract No. 68-02-2608
Task No. 30
Program Element No. EHE624
EPA Project Officer: Robert H. Borgwardt
Industrial Environmental Research Laboratory
Office of Energy, Minerals, and Industry
Research Triangle Park, NC 27711
Prepared for
US. ENVIRONMENTAL PROTECTION AGENCY
Office of Research and Development
Washington, DC 20460
-------
CONTENTS
SECTION Page
1 INTRODUCTION 1
BACKGROUND 1
PURPOSE 2
2 SUMMARY 3
VISUAL METHODS 3
INSTRUMENTAL METHODS 4
NUCLEATION STUDIES 5
3 EXPERIMENTAL METHODS 6
INTRODUCTION 6
OPTICAL MICROSCOPY 6
SCANNING ELECTRON MICROSCOPE (SEM) 15
INSTRUMENTAL COUNTING 16
4 RESULTS OF VISUAL METHODS 18
INTRODUCTORY DISCUSSION OF CaS03 CRYSTALS 18
OPTICAL COUNTING 22
SEM COUNTING 27
5 RESULTS OF INSTRUMENTAL METHODS 37
COULTER COUNTER RESULTS 37
6 NUCLEATION STUDIES 52
INTRODUCTION 52
GRANULES 53
PLATELETS 55
7 DISCUSSION OF RESULTS 58
SUMMARY OF VISUAL SIZING METHODOLOGY 58
COMPARISON OF VISUAL AND INSTRUMENTAL RESULTS 59
NUCLEATION 61
8 REFERENCES 62
APPENDIX A - INSTRUMENTAL DATA 63
iii
-------
FIGURES
NUMBER Page
3-1 Optical photographs of CaSOa granules in 30% glycerol at
400X magnification 8
3-2 Optical photograph of CaSOs platelet sample that was treated
with 60-seconds of ultrasonic dispersion 10
3-3 SEM micrograph of CaSOa platelet sample ultrasonically dis-
persed during preparation 11
3-4 Optical photograph of CaSOs platlets (not sonicated) 12
3-5 Optical photographs of granules from a full-scale scrubber,
not sonicated 14
3-6 SEM photograph of scrubber platelets 17
4-1 SEM photograph at 207X of CaSOs platelets from a full-scale
scrubber 19
4-2 SEM photograph at 201X of CaS03 platelets from a full-scale
scrubber 19
4-3 SEM photograph of laboratory grown granular particles at
300X magnification 20
4-4
4-5
4-6
4-7
4-8
4-9
4-10
4-11
Optical photograph of some laboratory grown CaSOa platelets;
200X
SEM photograph of scrubber-grown CaSOs platelets, at 2000X. . .
Graphic display of number count of CaSOs granular particles..
Optical photograph of CaSOs platelets,
Graphic display of the number count of CaSOs platelet parti-
SEM photograph of sonically dispersed sample of CaS03 plate-
21
24
26
?8
30
let; 204X 32
4-12 Graphical display of the visual count of SEM photo (200X)
of platelets 33
iv
-------
FIGURES (continued)
NUMBER Page
4-13 Graphical display of volume count calculated from number
count given in Figure 4-12a ................................... 35
5-1 Sieved platelets in size range 8-13 ym ........................ 38
5-2 Logarithmic display of number count of platelets in the
8-13 ym size range ............................................ 39
5-3 Optical micrographs showing results of sieving CaSOa granules. 41
5-4 Optical micrograph showing results of sieving of CaSOa plate-
lets from a full-scale scrubber ............................... 44
5-5 Optical micrograph showing results of sieving of CaSOs plate-
lets from an RTF pilot scrubber ............................... 47
6-1 Optical micrograph of granules grown from seeded supersatura-
ted CaSOs solutions ........................................... 54
6-2 Product platelets from different seeding growth conditions.... 56
7-1 Graphical display of composite SEM counting (combined soni-
cated and non-sonicated counts) ............................... 60
A-l Granules in size range 8-13y; number count .................... 64
A-2 Logarithmic display of number count in Figure A-l ............. 65
A- 3 Granules in size range 13-20y ................................. 66
A-4 Logarithmic display of number count in Figure A-3 ............. 67
A-5 Granules in size range >20y; not sonicated .................... 68
A-6 Logarithmic display of number count in Figure A-5 ............. 69
A-7 Granules in size range >20y; sonicated ........................ 70
A-8 Logarithmic display of number count in Figure A-7 ............. 71
A-9 Platelets in size range 13-20y; not sonicated ................. 72
A-10 Logarithmic display of number count calculated from volume
count in Figure A-9 ........................................... 73
-------
FIGURES (continued)
NUMBER Page
A-ll Platelets in size range 13-20u; sonicated 74
A-12 Logarithmic display of number count in Figure A-ll 75
A-13 Platelets in size range 8-13y 76
A-14 Logarithmic display of number count in Figure A-13 77
A-15 Platelets in size range 5-8y 78
A-16 Logarithmic display of number count in Figure A-15 79
A-17 Platelets in size range >20y 80
A-18 Logarithmic display of number count in Figure A-17 81
A-19 Platelets in size range 13-20y; not sonicated 82
A-20 Logarithmic display of number count in Figure A-19 83
A-21 Platelets in size range 13-20u; sonicated 84
A-22 Logarithmic display of number count in Figure A-21 85
A-23 Platelets in size range 8-13y 86
A-24 Logarithmic display of number count in Figure A-23 82
vi
-------
TABLES
NUMBER Page
4-1 DETERMINATION OF FINE PARTICLES MISSED AT 200X MAGNIFICA-
TION; OPTICAL MICROSCOPE 27
4-2 DETERMINATION OF FINE PARTICLES "MISSED" AT 200X MAGNIFICA-
TION; SEM 31
5-1 ESTIMATES OF CENTRAL TENDANCY FOR VOLUME. COUNT ON GRANULES
OF CaSO 3 BY THE COULTER METHOD 42
5-2 LOG-NUMBER SLOPES OF SIEVED GRANULES 42
5-3 ESTIMATES FOR CENTRAL TENDANCY FOR VOLUME COUNT ON CaS03 PLATE-
LETS BY THE COULTER METHOD FOR FULL SCALE SCRUBBER 45
5-4 ESTIMATES FOR CENTRAL TENDANCY FOR VOLUME COUNT ON CaSO3
PLATELETS BY THE COULTER METHOD FOR PILOT SCALE SCRUBBER 46
5-5 LOG-NUMBER SLOPES OF SIEVED PLATELETS FROM FULL-SIZE
SCRUBBER 49
5-6 LOG-NUMBER SLOPES OF SIEVED PLATELETS FROM RTP PILOT SCRUBBER. 50
vii
-------
viii
-------
SECTION 1
INTRODUCTION
1.1 BACKGROUND
Sludge disposal represents a significant operating cost in most appli-
cations of lime/limestone or dual alkali scrubbing processes. In systems
where calcium sulfite is the major product, the waste sludge generally
settles slowly, has a low settled density and is difficult to dewater. An
earlier EPA report [1] developed a mathematical model of a scrubber to
improve the quality of calcium sulfite sludge. It was assumed that the size
distribution of calcium sulfite crystals can be used to predict sludge quali-
ty for most systems. That report related the crystal size distribution to
process conditions. By using a mass-balance approach in which the change
in the number of crystals in a size range is equal to the net flux via con-
vection and via growth, a mathematical statement of the problem was de-
rived:
1 V- ni
,. -~ + D(L) = 2- -i + B(L) (1-1)
J Yj i T±
where: L = one dimensional crystal characteristic dimension
... _ vessel volume
Y. • inlet time constant = rq r—
'i stream flow rate
, . . vessel volume
Y « outlet time constant • — jz —
'j stream flow rate
R(L) = crystal growth rate
B(L) » crystal birth function by other than nucleation
D(L) - crystal death function
n(L) « crystal size distribution function such that
n(L) £L is the number of crystals in the size
range (L) to (L + AL) per unit volume of
slurry.
This equation is essentially a conservation equation which simply accounts
for all sources and sinks and flows of crystals. The reader is referred to
the previous EPA report [1] for further details of derivation and application.
-------
The model depends upon nucleation and growth rate parameters. These
quantities are determined from experimental measurement of the crystal size
distribution n(L). The linear growth rate R(L) is determined from the slope
of the crystal size distribution curve and the solids residence time.
Since the nucleation birth rate B , which enters the differential (Equa-
tion 1-1) as-a boundary conditions, is defined as the limit of the popu-
lation times the growth rate for very small particles.
BQ = lira n(L)R(L), (1-2)
L+o
the nature of n(L) in the small particle range is very important. The suc-
cess of this type approach depends upon the function chosen to approximate
n(L). In the earlier work [1], n(L) was measured by several means for two
full-scale scrubbers. However, the variety of methods led to two rather
contrary results for n(L): (1) instrumental results gave a particle
number count exponentially increasing with decreasing particle size; (2)
optical microscope and sieving gave a number count which increased to a
maximum but then decreased toward zero for smaller particles. That report
recommended clarification of this discrepancy.
1.2 PURPOSE
The purpose of the current work was to resolve the discrepancy in the
measured values of n(L), if possible, to provide an analytical function for
n(L), and to recommend the best method for the routine sizing of particles.
In particular, a reliable method was sought which would at least approxi-
mately measure the slope in the small particle region (<2 ym) for n(L) versus
L. An understanding of the physical factors influencing the nucleation
process is valuable when trying to produce a higher quality sludge. Thus,
as time and resources permitted, laboratory, batch nucleation studies were
to be used to quantify some of the factors influencing calcium sulfite
nucleation. Specifically, the effects of temperature, concentration, pH,
energy input (stirring) and seeding were considered.
Two types of crystal forms are known to exist in full-scale scrubbers.
One shape is a flat platelet, previously judged to be about 1:20:30 in rela-
tive thickness:width:length. The other shape is granular and sphere-like,
although not always regular. The platelets are more dense and settle faster
and are the preferred morphology. The physical relation between the two
forms, if one exists, was not discerned in the previous study. It was hoped
that the nucleation studies could identify factors which influence the
crystal's form.
-------
SECTION 2
SUMMARY
2.1 VISUAL METHODS
The scanning electron microscope shows that CaS03 platelet crystals
are not flat, but tend to thicken in the middle. Visual methods, by both
light and electron microscopes, with manual counting do not produce a reli-
able particle size distribution for CaS03 crystals because they cannot
measure the thickness of the crystals. For the same reason, sieving, which
segregates platelets by area, also cannot produce a reliable particle size
distribution.
Visual methods can be used for a quick survey of the crystals. The
general size and geometry of the particles are readily apparent, as is
the presence of any significant amount of impurity or other particle type.
A series of samples from the same source may have a fairly consistent thick-
ness-to-width ratio, and visual inspection may be useful in qualitatively
determining relative sizes.
Visual sizing of CaS03 granules suffers from three drawbacks: they
are nearly transparent at their edges, they frequently clump together, and
there are numerous voids which make true particle volume uncertain.
It was discovered in the course of this work that the granules are
actually hundreds of small platelets growing out of a common center. These
particles are nearly spherical and contain many internal void spaces. This
explains why they are less dense, settle more slowly and dewater more poorly
than the individual platelets. SEM work clearly showed the platelet-like
fine structure of the granules. Many intermediate particles also showed
multiple platelet growth from one platelet, but not enough platelets to
form a sphere. Nucleation studies, summarized in Section 2.3, showed that
the granules form when rapid growth occurs; individual platelets grow slowly
and only in slightly supersaturated solutions of CaSOa. Hence, even the
definition of a "single" particle (granule), which is composed of many small
platelets, may be uncertain.
In the particle sizing and counting work, a granule was considered to
be a single particle in granular sediments, and an individual plate was con-
sidered as a single particle in platelet sludges.
-------
2.2 INSTRUMENTAL METHODS
Instrumental counting is commonly done by either of two techniques,
Coulter Counter or sedimentation. The manufacturers' literature and the
results of work done in the report cited above indicate that the two tech-
niques are nearly equivalent. In this work, we report results only for the
Coulter method.
The Coulter method measures the particle's volume and reports it in
terms of an equivalent spherical diameter (BSD). When several platelet
samples from different sources were sieved into the same size range (by
area), very different ESD's were reported. This indicated that thickness-
to-width ratios of crystals from different sources vary appreciably. How-
ever, for a series of CaSOs platelets grown in the same source, Coulter
results paralleled expectations based on sieving. The Coulter method
is fast, yields repeatable results, is well suited for determining the
ESD's and is excellent for comparing size distributions.
If the number-size data are plotted on a logarithmic scale, they are
represented fairly well by several straight lines. This means that the popu-
lation is multimodel and each mode is exponential. However, most crystal
population theories predict one line (slope) for a particular sample; two
or three were nearly always found in this work. Speculation on crystal
growth mechanics that would produce such a multi-model population was
addressed in the EPA report cited above [1]. The exponential dependence
of particle number on size is in the form most compatible with the mathe-
matical model. Future use of that model will likely involve the determina-
tion, by instrumental methods, of a multi-termed, exponential, number-size
function.
The very large numbers of small particles as reported by the instrument-
al counting techniques in the earlier work implying high nucleation rates
were almost certainly fly ash particles. The sludges used were from on-line,
full-scale scrubbers. Work reported in this paper on pilot scrubbers (with-
out fly ash) indicates that extrapolation of the particle size distribution
curve to zero size to estimate the birth rate (Equation 1-2) is valid for
fly ash-free systems.
Preparation of the sample by using ultrasonic dispersion is not recommen-
ded. While it may break up some agglomerations, invariably they quickly
(re)form. And sonication does erode the edges and corners of the plate-
lets. In all cases but one, the size of the particle was reduced after soni-
cation. While some of the differences were significant on a volume-count
basis, there was less effect in the number-count presentation.
-------
Number-size trends for the granules appeared to be more regular than
those for platelets. Size comparisons between samples from the same source,
as determined by the Counter method, should be quite reliable. Since the
granules are really formed from platelets, sonication also erodes these
particles and is not recommended prior to instrumental counting.
2.3 NUCLEATION STUDIES
CaSOa nucleation mechanisms were also studied. The ultimate goal was
to understand processes which control the formation of calcium sulfite sludge.
In batch experiments, crystals were grown in a variety of environments that
simulated scrubber conditions. The effects of pH, impeller speed, concentra-
tion, temperature, kinetics and seeding on both crystal morphology (plate-
lets or granules) and size were monitored.
The following conclusions were reached:
• Solution temperature and stirring speed have little effect
upon crystal type or size.
• CaS03 crystals do not form at a pH of below 3.2, where bi-
sulfite is the dominant ionic form.
• In very supersaturated solutions (greater than three times
saturation), granules invariably form.
• Whenever spontaneous nucleation occurs the resultant crystal-
line form will be granular.
• Platelet CaS03 crystals can only be grown (in batches) at
low supersaturation levels with platelet seeding.
• The resultant size (area) of crystals grown by using seeds
appears independent of the size of the seed crystals.
This may not hold for the volume size.
• Granules dissolve more readily than do platelets in mild
acid, probably because the granules have a much larger
relative surface area. Hence, S02 scrubber contactors
may be preferentially destroying the granular forms.
-------
SECTION 3
EXPERIMENTAL METHODS
3.1 INTRODUCTION
Many particle sizing methods exist; a survey of the prominent tech-
niques was made in a previous Radian-EPA report [1, Appendix C]. In this
report the more applicable methods have been explored in greater depth and
compared to the others:
• optical microscopy,
• scanning electron microscopy (SEM), and
• Coulter counting.
The Coulter method is instrumental and is preferred for frequent use because
of its ease of use, the large size of the sample counted, and its unbiased
(by operator) results. Manual counting, whether on optical micrographs
or scanning electron micrographs, is very susceptible to error. Inaccuracies
arise because the number of particles counted is small and discretion must
be used by the observer; the counting is also very time consuming. Con-
sumption of time is not in itself a cause of inaccuracy. Instrumental
methods survey much larger, hence statistically better, samples. Each of
these techniques will be discussed and compared to the others in detail
below.
3.2 OPTICAL MICROSCOPY
A conventional optical microscope is certainly the simplest method
available for quickly surveying a sample of crystals. The general shapes,
the approximate size, and any major contaminants can be seen, but quantifi-
cation of these observations is difficult.
3.2.1 • Mounting Techniques
Mounting crystals for microscope viewing is very important. A brief
description of several techniques is given in this section along with the ad-
vantage and disadvantage of each.
Placing a drop of slurried CaSOs crystals on a slide leads to a variety
of problems. The larger crystals are found near the center of the drop;
-------
the smaller ones seem to be prominently near the edges. CaSO^, always present
to some extent in these CaSOa solutions due to oxidation, forms crystals
at the edge of the droplet. If the drop is allowed to dry, clusters of
crystals form and CaSOt appears throughout.
The microscope's depth of field is shallow and all of the crystals in
the drop cannot be simultaneously in focus. If a cover slip is placed on
top of the droplet, the depth of field is improved, but pressing the cover
slip down enough to get a thin layer of crystals causes crystal breakage.
Also, differential migration (by size) is observed as the smaller crystals
move to the edge of the coverslip; see Figure 3-1.
Dried sludge dusted onto a slide gives good contrast, but crystals may
breaK auaer the coverslip and there is much agglomeration. There is no
assurance that dusting maintains the original size distribution.
Immersion of the CaSOa crystals in oil or grease will protect the
crystals from breakage and help disperse agglomerates. The oils tested in
this study included glycerol, silicon oil, nujol, fluorolube and silicon
vacuum grease. The refractive index of the oil used must be substantially
different from the refraction index of the crystals to prevent loss of con-
trast.
Most oil-crystal mixtures showed aging changes: Some reaction, probably
dissolution, occurred slowly, and samples viewed several hours after immer-
sion were obviously altered. In the less viscous oils, differential migra-
tion was apparent; certain areas of view would consist predominantly of
large crystals while others contained only small crystals; again see Figure
3-1. The greases did not disperse agglomerates very well.
Filtering slurried crystals onto a Millipore^ filter was often ade-
quate for good viewing, but agglomerates frequently formed and smaller
particles could have been lost through the filter or obscured by larger
particles. The best preparative technique developed for CaS03 crystals
was aspiration of slurried crystals onto a Millipore filter (with no suction);
the filter was at no time saturated with water, and drying was allowed
between aspirations. This technique appeared to hold crystals where they
impacted, and agglomerations were fewer. Breakage was expected to some
degree, but little evidence (i.e., irregular breakage angles plus oddly
shaped bits and pieces) was observed. Also, the amount of CaS03 deposited
from the drying of the saturated CaSQ3 solution was calculated to be a
very small fraction of the total crystal weight.
After the crystals were distributed onto the dried Millipore filter sub-
strate, pieces were cut from the filter body, placed on a slide and immersed
breifly in acetone vapor. The acetone vapor changes the Millipore substrate
into a clear plastic film suitable for optical transmission microscopy.
The process is rapid (a few seconds), and no change in the CaSOs particles is
apparent.
-------
,
(a)
Figure 3-1.
Optical photographs of CaSOs granules in 30% glycerol at 400X
magnification. The two micrographs (a) and (b) were taken on
the same slide only a few millimeters apart.
-------
A special mention should be made of sonic dispersion. Recommended
procedure for use of a Coulter Counter includes immersion for 60 seconds
in a commercial ultrasonic vibrator bath. In anticipation of comparisons,
the effects of sonication were also investigated by microscopy. This treat-
ment of a few milliliters of crystal slurry does break-up agglomerates, but
it also appears to erode the crystals, as evidenced by the broken edges of
platelets shown in Figures 3-2 and 3-3. While only a few, new smaller broken
shards are observable, the larger platelet crystals look as though they
have been chipped away. The edges of the granules are too indistinct to be
resolved, but no change upon sonication is discernible. Finally, it should
be noted that the crystals rapidly reagglomerate (although probably not as
strongly); they definitely have an affinity for one another. A further
discussion on the effects of sonication is given in the section on instru-
mental counting; see Section 5.
3.2.2 Microscope Techniques
Although the use of a conventional microscope is relatively straight-
forward, several techniques were tried to increase contrast and distinguisha-
bility. However, none resulted in any significant improvement.
Various stains and dyes were tried. While they were picked up by the
crystals and color vision was improved, black and white contrast on Pola-
roid photos was not improved. Immersion oils were also tried, and while
they enhanced clarity at high magnifications (1000X and 2000X), nothing
additional was seen at these magnifications. Also, the oils tended to dis-
solve some of the crystals, especially the fines. Off-center lighting
slightly improved boundary definition. Polarization methods, useful for
single crystal work, were of no particular advantage in counting or sizing
fields of crystals.
Multiple photographs were taken of the same area at different magni-
fications. As the magnification was steadily increased, it was found that
few free particles were being missed in the less magnified photographs. If
smaller particles are there, this technique is not able to detect them.
However, many smaller CaSO3 platelets were revealed clinging to or growing
out of the larger platelets. Light polarization microscopy also confirmed
the presence of smaller crystals on the surfaces of large ones. These ad-
hering platelets were normally not counted as separate particles. Their
exact number and size were usually unresolvable, see Figure 3-4.
The CaSO3 granules clustered together and showed such a complex inner
structure that increased magnification revealed no new useable information,
see Figure 3-1. Within the range of conventional optical microscopes,
magnification greater than 200X proved to be of little additional use in
counting or sizing CaSO3 particles.
-------
<* ' .
•' .'- :.• .'V
. •», •: .•* • .
•f -
*
•
.
'v^
-
». tf y>
• • V
* * •
-. , ,
Figure 3-2. Optical photograph of CaS03 platelet sample that was treated
with 60-seconds of ultrasonic dispersion. Note the many broken
or eroded corners and edges as well as some broken, non-rectangu-
lar smaller particles; 200 X magnification.
10
-------
Figure 3-3. SEM micrograph of CaSOa platelet sample ultrasonically dis-
persed during preparation. Magnification at 501X. Note
irregularities around edges and on corners. Also note
notches in the ends of many crystals.
11
-------
/o/u
Figure 3-4. Optical photograph of CaSOs platelets (not sonicated).
a) typical field at 200 magnification
b) same area (near center) as above at 400 magnification.
Note that essentially no "new" particles are resolvable at
400X that were not identifiable at 200X. The multiple crystal
nature of a "single" platelet is more clearly revealed.
12
-------
3.2.3 Counting
Counting is very discretionary. First, one must decide which field to
select for counting. Then one must determine how many particles to count.
Additionally, it must be decided exactly what a particle is. For the reader
more interested in this phase of the work, each of these points is addressed
below.
For almost any preparative technique, areas with varying particle den-
sity will be seen in the microscope field. Some uncountable agglomerations
will be present. The more sparsely populated areas are easier to count,
but the number of particles counted is small, and one must decide if the
exclusion of agglomerations can be justified. Also, some areas will have a
greater abundance of large (or small) particles, due either to normal statis-
tical fluctuations or size differential migration of particles during prepara-
tion, see Figure 3-1. The observer must decide what a truly representative
area is.
Secondly, one must decide how many particles constitute a "good"
statistical sample. Certainly, the larger the count the better the sampling,
but one usually is limited by time and resources. The count totals by instru-
mental methods can be quite large, and one cannot hope to visually count a
comparable number of particles. Optically, perhaps the best one can do is
to terminate the count when it is observed that the distribution is not
"significantly" changing.
Thirdly, for platelets, it must be recognized that one cannot be sure
of the crystal thickness. The CaS03 platelets are too thin to measure on
an optical microscope. Realistically, to get a volume count, some ratio
of thickness-to-width must be assumed. While this may produce fairly con-
stant results for each sample, the thickness ratio may vary according to
the sample history. Comparison with instrumental counting (see Section 5)
indicates that the thickness-to-width ratio may vary significantly for
different samples of CaS03 platelets.
The granules presented other similar problems and were difficult or im-
possible to count. Several attempts were made, but results were not conclu-
sive. For example, refer to Figure 3-5. Figure 3-5a is a photograph at
400X magnification of a batch of CaSOs granules aspirated onto a substrate
without sonic dispersion. Counting the distinguishable dark spots as separate
particles still raises many questions. Some spots are rather more faint than
others; should they be counted equally? Are overlapping spots separate
particles? How large are particles partially obscured by other bodies?
Figures 3-5b is an enlargement of an area of Figure 3-5a. While some detail
is resolved, none of these discretionary questions is clarified. Circular
templates were used to estimate the area of each circular dark spot in the
photograph.
13
-------
«
v
iO JU.
(b)
Figure 3-5.
Optical photographs of granules from a full-scale scrubber,
not sonicated;
a) 400X
b) 800X of area in box in (a)
Note irregular shapes and difficulty in defining individual
particles. Also a vast number of small particles is not
apparent. A count of (a) is shown in Figure 2-12.
-------
Finally, for platelets, one must decide what to call a particle. This
can be most difficult, and those definitions may vary with one's objective.
For example, is an agglomerate to be counted as a single particle? Should
a single platelets with several smaller ones adhering to its surface be
counted as one or several particles? Should bits of foreign matter (like
fly ash or limestone) be counted? Should particles growing out of other
particles be counted separately? The granules present other similar pro-
blems, and they are more difficult or impossible to count.
When sizing particles for comparison with instrumental methods (which
don't distinguish between crystal types), all free particles, with agglomer-
ates counting as one, should be counted. When determining the number of
nucleation events, all independent CaSOa particles should be counted. When
counting to predict sedimentation rate, particles with different dry densi-
ties or shape factors should be counted separately.
In addition, one must decide whether a visual irregularity on a crystal
is a separte crystal, a new nucleation site, or just a structural irregulari-
ty. Individual particles within an agglomeration can only be counted approxi-
mately, if at all.
All these factors are at the discretion of the counter. The process is
tedious, time-consuming and could produce a wide spectrum of measurements
for the same sample. Thus, while it is a good method for a quick survey,
and it can distinguish between different types of particles, optical micro-
graphy can be a most difficult and potentially imprecise crystal sizing
method.
As explained in the preceding sections, optical counting has its
benefits and drawbacks. Some of the difficulties are overcome if the pur-
pose of the count is well defined. A discussion of the results will be post-
poned until after the section on the SEM work, at which point the two visual
counting methods will be compared.
3.3 SCANNING ELECTRON MICROSCOPE (SEM)
The SEM was used to look at the crystals in more detail and to look
into the size range below 5u where the optical microscope becomes unreliable.
The mounting technique that appeared to work best for optical micro-
scopy was also used for SEM. That is, a fairly dilute slurry of CaS03
crystals was aspirated onto a fine Millipore filter and allowed to dry.
This method yielded the most uniform distribution and minimized agglomera-
tions and breakage. The samples were then optically inspected, photographed
and counted. Subsequently, a small piece of filter, about 2 mm on a side,
was cut out and mounted on an SEM stage. It was sputtered with gold and
carbon to provide the conducting medium necessary for the SEM.
15
-------
Counting particles in an SEM micrograph is somewhat easier than in
an optical micrograph because of improved resolution and depth of field.
However, discretionary problems are heighthened. Multiple crystal formations
are more distinguishable, and a refined definition of particles is necessi-
tated; see Figure 3-6. In Figure 3-6b, how many particles are seen?
Other difficulties discussed in the counting of optical micrographs apply
equally to counting of SEM micrographs; see Section 3.2.3. Points made in
Section 3.2.4 are also applicable to SEM work.
3.4 INSTRUMENTAL COUNTING
In this work, two types of instrumental size counting techniques
were considered, Coulter Counter and sedimentation. Commercial sedi-
graphs (e.g., Micromeritics Instrument Corporation) require several grams
of materials and the sieving procedures described below yielded only milli-
gram quantities. Thus, only Coulter Counter data will be reported. In
other similar work ([1], Appendices C and D), including Micromeritics' own
literature, the sedimentation results are nearly identical to other instru-
mental results (Coulter Counter).
The Coulter Counter (Coulter Electronics, Inc.) instrument employs
an electrical current flowing across a small aperture in a conducting solu-
tion. As the solution flows, particles passing the aperture raise the
electrical resistivity approximately in proportion to their size. A signal
constitutes a count, and a measure of the volume of conducting solution
displaced is related to the volume of the particle. If the solution is too
concentrated with particles, coincidence counts will result in several smaller
particles counting as one larger particle. Coulter results, if in error, will
favor the larger particles. The ease of operation and reproducibility of
results make this method very desirable if many routine analyses are needed.
The samples, both granules and platelets, were analyzed at the Coulter
Electronics Lab. They were analyzed in a 4% NaCl/H20 solution saturated with
CaS03 and filtered. A Coulter Counter Model TAII with 140 ym and 50 ym
apertures was used. When sonicated, the samples were placed in an ultra-
sonic bath for 60 seconds. The data were processed on a Coulter M3 data
processor/calculator system. In most cases the particle size distribution
(PSD) was recorded as the number and volume percents finer than each size
range.
In certain experiments described later, crystals sieved into size ranges
were counted (Section 5) and used for seeding (Section 6). Micromesh sieves
were used with the aqueous CaSC>3 slurry. A repeated pull-push vacuum was
used to thoroughly wash the crystals. With this method about 500 mg could
be sieved at a time with about 2 liters of saturated CaS03 for washing.
The various size fractions were then stored in saturated CaSOs slurries.
16
-------
(a)
Figure 3-6.
SEM photograph blf Scrubber platelets.
a) 500X
b) 3000X of area in (a) at lower left tilted by 40°.
Note in (a) multiple incidents of crystals growing out of
crystals, (b) is a particularly graphic example of this.
Also note the srtall thickness of platelets.
17
-------
SECTION 4
RESULTS OF VISUAL METHODS
4.1 INTRODUCTORY DISCUSSION OF CaS03 CRYSTALS
Inspection of the SEM photos reveals several useful and interesting
facts about the CaSOs samples used. Since the platelet samples were from
a full-scale working scrubber, some limestone was observed; see Figure 4-1.
The limestone was usually in an ovoid shape with no angularities. There was
not enough limestone present to affect the total particle count. The parti-
cles that were seen were fairly large (>3y), and an elemental analysis
(by electron induced emission on the SEM) showed almost pure calcium. It
should be noted that only elements heavier than sodium are amenable to SEM
analysis.
Fly ash, previously unnoticed by optical microscopy, was detected by
SEM. In fact, large numbers of small particles were observed. Almost all
the fly r.sh was smaller than 2y and spherical; it was easily distinguishable
from the CaSOs crystals. An elemental (heavier than sodium) analysis of
the fly ash yielded a rather typical distribution. In order of approximate
abundance: Si, Al, Fe, P, Ca, Cl, Ti, K.
The morphological details are also more clearly seen with the SEM.
As observed elsewhere, crystal growth out of the face of parent crystals
is a very common crystal nucleation and growth mechanism. The enlarge-
ment frame, or cursor, in Figure 4-2 shows in detail a complex network of
crystals growing out of the surface of a 20 x 30y crystal. Inspection of
this and other photos confirms the fact that secondary growth is very common.
Figure 3-6b is even more dramatic; it is a close-up, tilted somewhat, of
the crystal complex in the lower left corner of Figure 3-6a.
The CaSOs granules are actually multiple growth centers on one parent
platelet. Figure 4-3 clearly shows platelets with "spine-like" projections
covering them. Here, the parent platelets are still resolvable, whereas in
Figure 4-4a, the parent rectangular shape has nearly been obscured. An
enlargement of the large granule very near the center of Figure 4-4a is
shown in Figure 4-4b. Clearly, the "spines" are rectangular platelets in
great numbers growing radially out of a center that had provided sites for
multiple secondary nucleation. This photo is also typical in that many of
the platelets grown in the same solution seem to be nearly the same size.
18
-------
Figure 4-1.
(a)
SEM photograph at 207X of CaS03 platelets from a full-scale
scrubber. Cursor at the right is 5X the 207X magnification.
Note fly ash (small, white spheres) scattered about; its ele-
mental analysis in decreasing order is: Si, Al, Fe, P, Ca,
^^^U, K (pn^eleme^s^jjia^jj^rjjjirjjgd^j^jrjjjlgigcied)!
shaped particle in
ground i| a millipore filter.
irsor is limestone. Back-
/o/u
Figure 3-4.
Optical photograph of CaS03 platelets (not sonicated).
a) typical field at 200 magnification
b) same area (near center) as above at 400 magnification.
Note that essentially no "new" particles are resolvable at
400X that were not identifiable at 200X. The multiple crystal
rf*PMogr^nS^01^
crubber. Cursor at right is 5X the 201X magnification. Dark,
ovoid particles are limestone. Note details of secondary crys-
tal formations on larger platelets.
&
-------
Figure 3-3. SEM micrograph of CaS03 platelet sample ultrasonically dis-
persed during preparation. Magnification at 501X. Note
irregularities around edges and on corners. Also note
notches in the ends of many crystals.
11
-------
Figure 4-4. SEM photographs of laboratory grown granular particles.
a) 198X
b) 2000X (cursor at right 10.000X); this particle is the large
granule near the center of (a).
Note carnation appearance of dense maze of plates. Also, note
extreme thinness of plates; at the bottom of (b), on the left,
the plates are so thin as to be transparent. In (b) on the
right, the thickness is no more than one-sixtieth of the width.
21
-------
Many distributions were observed in the course of this work. Nearly
always there seemed to be an abundance of crystals at a certain size, and
almost none larger. Figure 4-5 represents a batch of laboratory grown
platelets where one size seems predominant. Apparently, once the crystals
reach a certain size under these individual growth conditions, some other
type of growth dominates. It appears that many of these "large" crystals
began as secondary nucleations, and once a certain size was reached, they
broke off. Note the many crystals with notches in one end, where the
crystal was apparently nucleated and later broken off. Again, refer to
Figure 3-6b; it is believed that this is a crystal complex before these
"mature" crystals break off.
Figure 4-4b also provides a good opportunity to measure the thickness
of a platelet. It is impossible to determine whether any one platelet is
being viewed exactly edge on. But in the "carnation-like" granule, where
the length and width are fairly consistent, the thinnest edge-on view measures
about one-fiftieth to one-sixtieth of the width. One cannot be sure any
view is exactly edge-on, and some further rotation may reveal the crystal
edge to be even thinner. This ratio would then represent a maximum possible
thickness assuming that the length-to-width-to-thickness ratios are approxi-
mately constant for this platelet morphology.
Other results indicate that some particles may vary considerably in
thickness (i.e., thickness-to-width ratio). SEM photos taken of platelets
on edge show a wide variability, see Figure 4-6. Breakage may produce
particles excessively thick in relation to their length and width; the
large particle in the center of Figure 4-6 appears to be broken in this
manner. At the same time, the other end of this same platelet appears to
be much thinner. Also, the particle immediately beneath it has an air-foil
shaped cross-section and appears to be very thin toward its edges. In fact
the thickness-to-width ratio for crystals in this micrograph alone varied
from 1:20 to 1:60. Necessarily, the thickness measurement was very impre-
cise.
Thus, the assumption of a fairly regular thickness-to-width ratio may
not be a good one. Even in the same sample, a factor of 3 in the thickness-
to-width ratio exists. For different samples, the history of the sample could
markedly affect the thickness of the plates. Instrumental results (Section
5) also seem to indicate an inconsistent thickness-to-width ratio.
4.2 OPTICAL COUNTING
The difficulties inherent in visual counting have already been discussed
Discretion must be used by even an impartial observer and the results are
necessarily biased, but usually in an unknown direction.
22
-------
ft-
*"&
I b. WcH
/ *»
Figure 4-5. Optical photograph of some laboratory grown CaSOs platelets;
200X. Note predominance of crystals of about the same size.
Also note many crystals with notches in the end indicating
that the platelet was broken off from growth (nucleation)
center
23
-------
Figure 4-6. SEM photograph of scrubber-grown CaSOs platelets, at 2000X.
An edge view of crystals. Note great variability in thick-
ness between crystals or even for same crystals. The large
crystals in the center appear to be thicker toward their
center; the largest crystal is believed to be broken at
its left end.
24
-------
4.2.1 Granules
As mentioned in Section 3.2.3, the granules were very difficult to
count. Poorly resolvable boundaries and agglomerations were the most
troublesome problems. Circular templates were used to estimate the area of
each approximately circular dark spot on the photographs; see Figure 3-5.
A count of this figure is displayed in Figure 4-7a.
Figure 4-7b represents a count of a similar photograph (actually the
photograph in Figure 3-la) for the same initial sample but after sonic
dispersion. (Note: there is a difference in mounting technique between
the two figures, but it is not judged to be important in this comparison.)
However, the uncountable background appeared much more cluttered in the
sonicated sample. Compare Figure 3-5a at 400X to Figure 3-la at 400X.
In the former, the background looks "clean", and no small particles are
apparent in the enlargement (Figure 3-5b). But Figure 3-la shows a myriad
of small, unclear spots. Such increases in background "noise" upon soni-
cation were not always this dramatic, but a general trend toward more,
smaller particles in sonicated samples was noticed. Finally, to once again
emphasize the discretionary nature of the technique, Figure 3-lb pictures
an area less than 2mm away from the area shown in Figure 3-la on the same
slide. Because even matched samples often showed considerable variation
in appearances, all microscopic and SEM results must be considered to be
only qualitative.
4.2.2 Platelets
Particle definition was easier with the CaS03 platelets than with the
granules. Frequent clumps (agglomerates) were uncountable, but these were
assumed to be random conglomerations of all types of particles with an
approximate average distribution. Thus, to be tractable, the areas chosen
to be counted simply avoided such clumps of platelets.
Confusion can arise in the definition of platelet size. Instrumental
counts measure a particle's volume and report sizes as equivalent spherical
diameter (ESD). Visually one sees flat rectangles with an average ratio
of 2:3 for width:length. One can therefore describe the "area" of the
particle on the basis of one linear dimension; i.e., width, length or the
average of the two. To convert from area to ESD for platelets (to facili-
tate later comparison):
4/37TR3 - t x w x H (4-1)
where t is the thickness and 2R - ESD. The typical crystal morphology deter-
mined in this work on the SEM indicated an average geometry of 1:30:45,
resulting in the crystal width being 2.18 times the ESD. Since the plate-
let shape factor can vary considerably, this ratio should be used only as
an approximate conversion factor.
25
-------
Num
GRANULES (Nor SONICATED)
Cumulative % Greater Than
Number %
10 15 20 25 30 35 40 45 50
DIAMETER (u)
(a)
25-
20-
Num
% 15-
10-
5-
n
\ GRANULES (SONICATED)
^x. /-Cumulative %
\
~
7
\.
S^~
^
^*+
Greater Than
•Number 1,
^* — \ — • i
•100
• 80
Cum
• 60 5
1 40
• 20
n
0 5 10 15 20 25 30 35 40 45 50
DIAMETER (u)
(b)
C1r;ure 4-7. Graphic display of number count of CaSOi granular particles.
a) non-sonically dispersed preparation; Total count, 78;
Count of Figure 3-Sa.
b) sonlcally disaersed pi-eoaration; Total count, i3.
Count of Figure 3-la.
26
-------
A typical example of platelets is shown in Figure 4-8a. The field
is rather sparsely populated to avoid agglomerates. The sample, from an
actual scrubber, was aspirated onto a Millipore filter substrate; it was
not sonically dispersed and magnification is 200X. Figures 4-8b and A-8c
show sections of the same field at 400X and 800X magnification. Very few
new particles were detected at the increased viewing power; certainly no
vast number of small particles dominates the sample. Figure 4-9a represents
the count of this crystal field. Very few extremely large (>40y in width)
particles were seen, and if the samples were handled with care, relatively
few small broken pieces were observed.
A typical photograph of a sonicated sample is seen in Figure 4-10a.
That sample, from the same batch as before (i.e., in Figure 4-8), was mounted
as before and is shown at 200X magnification. Comparison photographs
at 400X and 800X of part of the same field are in Figures 4-10b, and c,
respectively. Typical of several other such microscopic enlargements of
areas of photographs, very few particles were "missed" in the 200X photograph
compared to the 800X photograph; see Table 4-1. Size measurements at 800X
TABLE 4-1. DETERMINATION OF FINE PARTICLES MISSED AT 200X MAGNIFICATION;
OPTICAL MICROSCOPE
Sample
Fig 4-8
Fig 4-10
Total
Particles
Counted
at 200X
47
155
Total <5y
Particles
Counted
at 200X
14
73
Number of
<5y Particles
"Missed" in
Box at 200X
From 800X
Photograph
1
2
By Extrapolation,
Total Number
of <5P Particles
in 200X
Photograph
4
8
Percent
Missed
at 200X
8X
5Z
compared to 200X were only slightly (M.0%) more precise. Again, some of the
smallest particles (specks) were later determined to be fly ash. Indeed,
30% of the particles are below 5y (counting only small ones which show
angularity - presumably therefore not fly ash). Again, the background
appears to contain more "trash". Many of the smaller particles are angular,
probably being "broken-off" chips. These particles, more common in sonicated
samples, are somewhat thicker than the 1:30:45 relative dimensions; their
calculated ESD will be perhaps a factor two too small.
4.3 SEM COUNTING
The scanning electron microscope produces much higher quality micro-
graphy, but of course it is much more expensive and time consuming. It
was used primarily to check on the optical method. Significant differences
were found. The optical microscope cannot be used to identify particles
smaller than 2y in width, and is somewhat questionable below for widths less
than 4 or 5y. Above that size, with proper technique, optical microscopy
should be as accurate as SEM.
27
-------
Figure 4-8. Optical photograph of CaSOs platelets.
The field chosen is sparsely populated to avoid agglomerates
Not sonicated.
a) 200X
b) 400X, area of (a) in box
c) 800X, area of (b) in box
The count of (a) appears in Figure 4-9a.
28
-------
Num
%
25 -
20 •
15
10
PLATELETS (Noi SONICATED)
Cumulative % Greater Than
Number %
10 15
20 25 30
WIDTH (y)
(a)
100
80
Cum
60
-40
20
Num
PLATELETS (SONICATED)
Cumulative % Greater Than
•Number %
10 15 20 25 30 35 40 45 50
100
•80
• 60
40
20
Cum
Figure 4-9. Graphic display of the number count of CaSOs platelet particles;
a) non-sonically dispersed sample; Total count, 47
Count of Figure 4-8a.
b) sonically dispersed sample; Total count, 166.
Count of Figure 4-1 Oa.
29
-------
#>,
(a)
• }
. f / A * «
s
. . '
-
*«-
^7
g
;••>;•*
i . -^
* ^^\
•*
«
4
• t ^ i^
.*
, -
• «'• ' i
,/
•
' »
* *
;
•
^^i
v^?^
J^ '
<• . ' ^
^ ^f-
,
,
,
.
;
:••
v
(b)
(c)
Figure 4-10. Optical phocograph of CaSO; platelets, the sample has been
ultrasonically dispersed.
a) 200X
b) 400X, area of (a) in box
c) 80QX, area of (b) in box
The count of (a) appears in Figure 4-9b.
30
-------
When the SEM is operated at 200X, the pictures are more clearly defined
than with the optical microscope at a comparable magnification; see Figure
4-11. This is another photograph of CaSOs platelets from a full-size
scrubber (also see Figure 4-2). In the preparation, the sample in Figure
4-2 was not ultrasonically dispersed; the sample in Figure 4-11 was. A
comparison of the photos for the effects of sonication will be given after
the counting is discussed; see Section 4.3.3.
4.3.1 Number Count
In either photograph, Figure 4-2 or 4-11, the left side is at about
200X magnification, and the cursor, enlarged on the right, is at about
1000X. A careful number count of the left sides was made and is graphically
displayed in Figure 4-12. The number of smaller particles increases general-
ly in an exponential fashion. However, when the 1000X cursor is carefully
inspected, many particles (<1.5y) are found that were not identified in the
same area at 200X magnification. Some are particles deeply imbedded in the
Millipore filter support; some are particles adhering to or growing out of
the surface of larger crystals. Others are just too small to be defined as
particles at 200X magnification. Table 4-2 gives the actual counting data
on these photographs; see Table 4-1 for a comparison with optical micro-
scopy.
TABLE 4-2. DETERMINATION OF FINE PARTICLES "MISSED" AT 200X MAGNIFICATION ; SEM
Sample
Total
Particles
Counted
at 200X
Total <1.5y
Particles
Counted
at 200X
Number
of <1.5y
Particles
Missed
at 200X,
From 1000X
Cursor
By
Extrapolation,
Number
of <1.5p
Particles
Missed in
Whole Photo
Fly Ash
Particles
Missed
at 200X,
From 1000X
Cursor
By Extrapolation
Number
of Fly Ash
Particles
Missed in
Whole
Photo
Fig 4-2
(non-sonicated)
125
15
10
250
200
Fig 4-11
(sonicated)
152
27
20
500
200
The cursor covers only 1/25 of the area of the left side. As Table 4-2
shows, if the cursor area is considered to be typical, when the number of
particles newly identified in it is multiplied by 25, an estimate of the small
particles (<1.5u) missed at 200X is obtained. This estimate shows that many
more particles are being missed (small particles) than are being identified
(all particles). Recall that the slurry was aspirated onto the filter with-
out vacuum, so few particles are presumed to have been lost into the Milli-
pore filter matrix. If these assumptions are approximately correct, then
approximately 70% of the particles are smaller than 2y (80% if fly ash is
31
-------
Figure 4-11.
SEM photograph of sonically dispersed sample of CaS03 plate-
lets; 204X. The cursor on the right is 5X the 204X magnifi-
cation. Note that many of the "rectangular" platelets are
erroded at the edges and corners.
32
-------
25'
20.
15.
Mum
% 10]
5.
0
PLATELETS (NOT SONICATED)
Cumulative % Greater Than
Number %
20 25
WIDTH
(a)
Num
TOO
80
60
Cum
40 *
120
30 35 40 45 50
25'
20.
15
10
5 .
0
•m
\
PLATELETS
\
(SONICATED)
^ — Cumulative %
WH
\
Greater Than
/ Number %
\|
f
^
^_
r~^_
—.^•^
1 1
^J
I
L
:as=
— •
100
,80
.60
Cum
• 40
.20
0
0 5 10 15 20 25 30 35 40 45 50
WIDTH (v)
(b)
Figure 4-12. Graphical display of the visual count of SEM photo (200X) of
platelets.
a) not sonlcally dispersed; Count of Figure 4-2; Total Count
125.
b) sonically dispersed; Count of Figure 4-11; Total Count 152.
Note general shift of probability toward the smaller size
range in sonicated sample.
33
-------
included) in the non-sonicated sample and 80% (90% with fly ash) in the
sonicated sample. This conclusion will be shown to be in reasonable
agreement with the results of instrumental methods.
It is therefore clear that optical number counting is inaccurate below
about 4y for optical microscopes and below about 2p for SEM (at 200X). If
higher magnifications are used, optical methods lose resolution and few
new, small particles are identified. For SEM the magnification can be in-
creased with good resolution, but then the sample becomes so small that
several fields of view are required for a representative sample. Such tech-
niques should be fairly good for particles in the 4-8y size range and quite
accurate for larger sizes (assuming the other difficulties discussed
earlier in this section can be overcome).
4.3.2 Volume Counting
If one casually glances at a 200X SEM photograph, either Figure 4-2
or 4-11, one does not see a preponderance of small particles. Rather the
eye tends to see a volume count; that is, most of the volume (or mass)
appears to be intermediate in size. Figure 4-13 gives a volume count of
Figure 4-2 (also see Figure 4-12a for the number count), and the presenta-
tion satisfies the intuitive impression that one gets from viewing these
photographs. The three large particles at the lower left in Figure 4-2
represent fully 30% of the volume of all CaS03 platelets in the picture.
The vast numbers of fine particles occupy only a miniscule percentage of
the volume.
Figure 4-13 shows that the particles smaller than 2y account for
only 0.013% of the volume. If the population of fines extrapolated from the
cursor results is included (70% of the numbers), the small particles still
occupy only 0.2% of the CaSOa volume. Inclusion of the estimated 200 fly
ash particles raises the fines volume to 0.3%. Hence, the volume count,
reflective of what the eye sees, is dominated by large particles. The number
count, the data required by most mathematical models, is dominated by small
particles. Numerical results, derived from volume count results via shape
factors, may be inaccurate since the small particle region of the volume
count is poorly determined. It is just this size region that will be most
important in the number count.
4.3.3 Effect of Ultrasonic Dispersion
Ultrasonic dispersion, or sonication, is routinely used to break up
agglomerates in samples for instrumental counting. It has been assumed that
sonication does not damage the crystals. It may or may not dislodge crystals
adhering to or growing out of larger crystals. These phenomena were investi-
gated with the SEM. The effects of sonication on instrumental counts is
discussed in Section 5.
34
-------
PLATELETS (NOT SONICATED)
100-
Cum
Cumulative % Greater Than
Volume %
Vol
Figure 4-13. Graphical display of volume count calculated from number count
given in Figure 4-12a.
Note the dramatic shift in the shape of the histogram from
Figure 4-12a.
35
-------
According to the limited number of SEM photographs used, it does appear
that sonication reduces the size of CaSOa crystals. Figure 4-12 compares
number counts of the same sample, one (a) prepared without ultrasonics and
the other (b) prepared with the treatment. The sonicated sample does show
a shift in the number percent count toward smaller sizes.
Two features also noted in the optical microscopic comparison are simi-
larly found in these samples. First, very large particles are seen after
sonication; see Figure 4-12a where a few particles were identified as large
as 45y wide. Secondly, many more smaller particles are seen; again see
Figure 4-12 and compare the regions below lOy.
Other samples that were sonically dispersed and photographed on the SEM
showed a definite increase in the finer particles; see Figure 3-3 at 500X.
For instance, 75% of the sonicated particles are at 9.0y or smaller; the com-
parable figure was 54% for the uonsonicated sample. Again, no large particles
(>30y in width) are seen in this sonicated sample. The "gnawing" around
the edges of the crystals due to sonication is well depicted in the 2500X
cursor of Figure 3-3. It appears that sonic dispersion does alter the
number count by breaking large particles and generating more smaller ones.
As we shall see later, instrumental counting results generally show a
smaller average size for the whole distribution after sonication.
No attempt was made to count granular crystals in SEM micrographs
because it is impossible, for the purpose of identifying nucleation sites,
to define what a particle is. Inspection of Figures 4-3 or 4-4 clearly
shows what is present, but also dramatically illustrates the impossiblity
of counting, let alone sizing.
36
-------
SECTION 5
RESULTS OF INSTRUMENTAL METHODS
Instrumental particle sizing techniques have several distinct ad-
vantages over visual or manual counting methods. They are objective, fast,
size or count large samples, and give reproducible results. However,
they do not distinguish between different particle types in mixtures of
particles. This includes the inability to differentiate large particles
from agglomerates. They measure an effective spherical diameter and
shape factors must be determined by other means if the true particle
dimensions are sought. They measure a size distribution based on
mass (or volume); this can be converted to a number distrubution.
Finally, the particles must be immersed in a fluid.
5.1 COULTER COUNTER RESULTS v
One of the primary objectives of this project was to determine the
most suitable, reliable means to count particles by size. This includes
an understanding of the reasons for discrepancies between optical and
instrumental methods [1]. To this end, samples of CaSOa particles, both
granules and platelets, were sieved, photographed and counted by the Coulter
method. Identical samples were run with and without sonic dispersion. A
complete compilation of results will be found in Appendix A.
Two presentations of the data are frequently used. The Coulter
Counter actually measures a volume count, which can be converted to mass
count if the particle density is known. The relative or percentage number
size count may be obtained by dividing the number count in each size in-
terval by the average particle volume for that size range. In most cases
the number count will be given the Coulter Counter data processor. Thus
a small volume of tiny particles can convert to a very large number of
tiny particles. Nearly all the counts in this report show such a
characteristic. For example, the data shown in Figure 5-1 (and included in
Appendix A) show a volume maximum at 5.5y with very little volume below 2y.
Still, the number count increases nearly monotonically as the particles
diminish in size. Figure 5-2 is a logarithmic plot (where a true exponen-
tial curve plots as & straight line) of the relative number count. These
plots are linear over most regions and indicate that the population is be-
having exponentially. The only exceptions appear to be in ranges that were
enhanced by sieving or transition regions between true exponential regions.
37
-------
PLATELETS (FULL-SIZE SCRUBBER)
SIZE RANGE, 8-13u
VOLUME PERCENT COUNT
-I—I I I I II11 I "1 1 I I I I 111—I—I 'I I I I I I, I I I
.5 1.0 5 10 50 100
MICRON DIAMETER Log Scale
PLATELETS (FULL-SIZE SCRUBBER)
SIZE RANGE, 8-13u
NUMBER PERCENT COUNT'
n I i >
10
MICRON DIAMETER Log Scale
I '' " I
50 TOO
Figure 5-1. Platelets in size range 8-13U-
Top: Volume Count
Bottom: Number Count
38
-------
X
a
u
ESD XI0
PLATELETS
NUMBER VS ESD
8-13y
FULL-SIZE SCRUBBER
O ©
8 10 12 14 16 18 20 22 24 26 28
Figure 5-2. Logarithmic display of number count of platelets is the
3-13 urn size range. See Figure 5-lb.
39
-------
Most crystal growth theories (1, 2) predict that the number of particles
in a size range is (negatively) proportional to the size. That is:
dN X7 dN
-TT a ~ N or -rr = -ON
dL dL (5_1)
where a is the constant of proportionality and L is a one dimensional
crystal size parameter (like equivalent spherical diameter). The solution
of this differential equation is:
N = Ae~aL or In N - -aL + Constant (5-2)
Thus, a plot of In N versus L will be a straight line with a slope of -a.
Speculation on the relation of a to the crystal growth mechanism may be
found in either of the two references at the beginning of this paragraph
and will not be specifically addressed here. However, the occurrence of
several straight line regions in the present data indicates that different
mechanisms are in effect in different size ranges. It is proposed that
the steep slopes (about -10) below 2y found in particles from production-
line scrubbers are due to fly ash.
The occurrence of several straight lines on the log-population plot
indicates that multimodal population control mechanisms are operative.
For the purposes of mathematically modeling a scrubber, it may be neces-
sary to deal with a compound exponential population equation of the form:
N =» Ae'aL + Be"3L + Ce~6L + . . . (5-3)
whsre different ranges are dominated by different terms in this type of
equation.
5.1.1 Granules
The principal goal of this project was to determine a method that
could satisfactorily size a population of sludge crystals. Instrumental
counting of the granular CaSOs crystals appears to be an acceptable solu-
tion. To this end, laboratory grown CaSOs crystals were seived into size
ranges, photographed and counted visually and instrumentally. The micro-
graphs of the three size ranges, >20y, 13-20y, and 8-13y appear in Figure
5-3. The volume-per-size-range count, an analysis that reflects what the
eye sees, agrees fairly well with the trend expected from sieves. The
maximum population location, the mode, appears to be in best agreement
with the expectations from the sieving (see Table 5-1). Appendix A is
a complete exhibit of the data.
40
-------
'*§
>'
(c)
Figure 5-3. Optical micrographs showing results of sieving CaSOa granules,
Magnification at 400X. General size differences are readily
apparent. Instrumental counts of these batches are included
in Appendix A.
a) >20y batch
b) 13-20y batch
c) 8-13y batch
41
-------
TABLE 5-1. ESTIMATES OF CENTRAL TENDANCY FOR VOLUME COUNT
ON GRANULES OF CaSOs BY THE COULTER METHOD
Particle
Size (y)
>20
>20
13-20
8-13
Sonic
Dispersion
No
Yes
No
No
Geometrical
Mean (y)
13.58
11.40
7.31
8.13
Median
(U)
16.77
13.82
7.77
8.54
Mode
(y)
20.34
15.34
12.06
7.55
'Standard
Deviation
2.10
2.28
2.47
2.50
Since these samples were sieved from successive batches of laboratory
grown CaSOs crystals, they contained no fly ash. However, regardless of
the amount of washing, a large number of small particles were always pre-
sent, see Figure 5-3. Although the volume count paralleled expectations
from sieving, in the number count the fine particles still dominated. The
sieves effectively segregated larger particles but appear to have had little
effect on the smaller particles.
For the batches of larger sized particles, one would expect the number
count to begin at a greater BSD. For the batches of smaller sized particles
the percentage count should begin at smaller diameters, but then increase
more quickly (steeper slope). These expectations are met for these data,
see Figures A-2, A-4, A-6, A-8 and Table 5-2. The 8-13y batch shows some
irregularities from the others in that it appears to be bi-modal over this
range. The average of the two slopes is probably a more representative
figure (see Table 5-2).
TABLE 5-2. LOG-NUMBER SLOPES OF SIEVED GRANULES
Particle
Size (y)
>20
>20
13-20
8-13
Sonic
Dispersion
No
Yes
No
No
Slope
>5y
- 0.82
- 1.00
- 1.38
-1.83 (-1.42)*
Slope
<5y
-3.62
-3.98
-3.30
-10.90
*The average of the two slopes for particles >5y.
The 8-13y fraction also has an inordinately steep slope in the below 5U reei
where all other samples were quite similar. But otherwise, the Coulter °n>
results agree with the expectations from the sieving preparation.
42
-------
The Coulter data are reported in terms of equivalent spherical dia-
meter. The granules appear to be approximately spherical; see Figure 4-4.
The ESD's observed should give good relative size correlations. However,
granules are not solid. Since the Counter method actually measures the
true volume displaced, it effectively reports a solid sphere. Thus, the
apparent optical size concentration maximum in the micrographs may be
larger than the volume concentration maximum determined by the Coulter
method. A direct comparison is not possible since the void volume of the
granules is difficult to estimate; again refer to Figure 4-4.
Finally, the behavior of the log-number curve for the >20y granule
fraction should be discussed: refer to Figure A-6. The plateau region
around 8y is inconsistent with any exponential population mechanism.
Similar behavior is seen in some of the platelet fractions, and mechanisms
may be operable that result in such populations (e.g., new particles
formed exclusively from attrition). However, recall that each batch was
sieved into size fractions, and one might therefore expect a maximum in the
number count curve. In practice this was seldom found. Generally, the
sieving effectively sized larger particles only, and the smaller particles
were always present in each size fraction. The 20y sieve was the easiest to
use (less clogging), and the plateau may be a reflection of that efficien-
cy.
5.1.2 Platelets - Volume Count
In general the Coulter results on CaS03 platelets were less reliable
from one sample to another than were the granular particle counts. Two
sets of samples were run, one from a full-scale, operative scrubber (fly
ash included) and another from a model or pilot scrubber (no fly ash).
Both sets were sieved consecutively into several fractions, photographed
and Coulter counted.
The first set run was CaSOa scrubber sludge from a full-scale scrubber.
Figure 5-4 shows three photos of the size ranges after sieving; 13-20y,
8-13y, and 5-8y. Again, the seiving appears to be effective on the larger
particles. The 13-20y fraction shows no particles whose width is greater
than 20y, and only a few crystals whose length is slightly greater than 20y.
The 8-13y fraction micrograph similarly shows no particles wider than
lOy and none longer than 16y; however, the background shows an unusually
large number of smaller spherical particles, probably fly ash. The 5-8y
fraction shows many fine particles but none whose width exceeds 8y. These
surveys of size maxima cover at least three micrographs similar to those in
Figure 5-4 for each size fraction. In summary, the pictures indicate
that the sieving has restricted the size of the large particles to the
expected range.
Again, the volume count reflects the trends expected from sieving;
see Table 5-3.
43
-------
(a)
(b)
(c)
Figure 5-4. Optical micrograph showing results of sieving of CaS03 plate-
lets from a full-scale scrubber. Magnification at 400X. General
size differences are readily apparent. Instrumental counts of
these batches are included in Appendix A.
a) 13-20y batch
b) 8-13u batch
c) 5-8|j batch
44
-------
TABLE 5-3. ESTIMATES FOR CENTRAL TENDANCY FOR VOLUME COUNT
ON CaSO$ PLATELETS BY THE COULTER METHOD FOR
FULL SCALE SCRUBBER
Particle
Size (u)
13-20
13-20
8-13
5-8
Sonic
Dispersion
No
Yes
No
No
Geometrical
Mean (y)
8.05
7.73
5.48
5.18
Median
(y)
8.38
8.18
5.52
4.45
Mode
(y)
10.08
7.64
5.66
3.56
Standard
Deviation
2.28
2.37
1.66
2.27
The complete population curves are in Appendix A, Figures A-9, A-ll, A-13,
A-15. These data appear quite regular and show no exceptions to the ex-
pected trends. The small standard deviation of the 8-13y batch means that
the volume count is bunched rather tightly around the average value
(or most common value [mode]), about 5.5y; see Figure A-13a. The micro-
graph shows most of the crystals to have one of their dimensions (length
or width) within this range; however, the thickness is impossible to deter-
mine with the optical microscope. Other studies have judged the thickness
to be about one-twentieth of the width (1); SEM work has shown that some
are much thinner. An estimate can be made from the Coulter analysis of
the sieving product if one assumes 10.5y as an average rectangular dimension
for the 8-13y batch determined by microscopy. Recall that the Coulter data
are reported in terms of equivalent spherical diameters. For this case
BSD =* 2R - 5.5y (from Table 5-3):
t x 10.5 x 10.5 - j IT (R - 2.75)3 (5-2a)
t x (110.25) - 4.189 (20.797) (5-2b)
t = 0.790y (5-2c)
for a width-to-thickness (t) ratio of something like:
w/t - 10.5/0.790 - 13.3. (5-3)
The earlier studies [1] assumed an average crystal geometry of 1:20:30,
so the agreement for thickness estimates is only approximate. With w/t •
13.3, the predicted ESD based on the sieving would be: 13-20y batch, 8.65p;
8-13y batch (as calculated above), 5.5y; and, the 5-8y batch, 3.4ly. The
agreement to the geometrical mean size for each batch (Table 5-3) is only
fair; about the same can be said for agreement with the median and mode.
45
-------
The second set of data resulted from a similar procedure run with CaSOs
platelets from a pilot scrubber at RTF; there was no fly ash in the sample.
Again the samples were sieved, photographed and Coulter counted. The results
are given in Table 5-4.
TABLE 5-4. ESTIMATES FOR CENTRAL TENDANCY FOR VOLUME COUNT
ON CaS03 PLATELETS BY THE COULTER METHOD FOR
PILOT SCALE SCRUBBER.
Particle
Size (y)
>20
13-20
13-20
8-13
Sonic
Dispersion
No
No
Yes
No
Geometrical
Mean (y)
15.56
10.23
11.57
9.31
Median
(y)
16.55
10.22
11.23
10.01
Mode
(y)
17.38
9.97
10.16
13.25
Standard
Deviation
1.69
1.59
1.70
2.26
Once again, the sieving appears to have successfully restricted the
sizes of the larger particles; see Figure 5-5. In the greater-than-20y
batch, many smaller particles are seen, but particles whose width is 18 to
20y are common (Figure 5-5a). The 13-20y batch is clearly made up of smaller
particles. From several micrographs, no particle was seen larger than 15y
in length. Nor was any particle seen greater than 13y in width; see
Figure 5-5b. The 8-13y batch was not as well size resolved; it did contain
particles as large as 15y in length as well as many smaller particles. The
trends in Table 5-4 reflect the population trends in the micrographs, except
for the mode in the 8-13y sample.
The oddity in these data is the instrumentally determined increase in
size after sonic dispersion. No other example showed this behavior. The
operator insisted that the probability of a "switch" in the data between
the 13-20y sonicated and nonsonicated samples is very small. The increase
in mean size after sonication, especially when a decrease is expected, is
outside the expected range of technique variability. The result stands
without explanation.
However, another problem is introduced by these data. They do not
compare well with the other sieved platelet Coulter counts. A comparison
of the micrographs of similarly sieved samples (Figure 5-4a, b with Figures
5-5b, c, respectively) suggest that the pilot scale scrubber samples appear
under the microscope to be slightly larger. Yet the Coulter data statis-
tical figures are from 25 to over 200 percent larger for the second set
of batches. Again, this discrepancy is not believed to be instrumental
46
-------
u :
,
£'•
(a)
(b)
i/ r
- • *
^^^•p *
m
Figure 5-5. Optical micrograph showing results of sieving of CaSOs platelets
from an RTP pilot scrubber. Magnification at 200X for (a),
400X for (b) and (c). General size differences are readily
apparent. Instrumental counts of these batches are included in
Appendix A.
a) >20y batch
b) 13-20y batch
c) 8-13y batch
47
-------
error. It is not possible to account for these consistently higher re-
sults for the second set of samples (i.e., those from RTF) by comparison
of length and width measurements of the two samples. This observation led
to a more careful analysis of the third dimension, thickness.
Measuring the thickness of thin platelets, even on the SEM, is diffi-
cult. The best example obtained was shown in Figure 4-6. Here the edges
appear at oblique angles. A great range of thicknesses is seen. The largest
crystal in the center appears to be rather thick on the left end (as viewed
in this micrograph). This end is the result of breakage, and the resulting
cross-section is clearly exposed. However, the other end appears to taper
off and is comparatively quite thin. The back length edge may also be thin
suggesting an "airfoil" shape. The crystal immediately below it clearly
is thicker in the center than towards its edges. Thus, the crystals appear
to fatten in the middle where the crystal-molecular alignment forces are
greatest. If breakage occurs, then all width-to-thickness ratios are lost.
In this picture alone, the width-to-thickness ratio range from 12:1 to 54:1.
Hence, the thickness of a platelet, necessary for comparing the optically
determined length and width size with the instrumentally reported equiva-
lent spherical diameter, are optically indeterminable. The width to thick-
ness ratio can vary over a range of at least four, and that ratio is pro-
bably quite dependent upon the crystal history. That is, growth time and
conditions, in addition to breakage, very likely significantly influence
crystal geometry.
The constant size differential between the two sets of samples that
were Coulter counted is probably real. The thickness variation may account
for the reported differences in the equivalent spherical diameters.
Optical inspection for volume determination is consequently nearly
impossible. The Coulter method is probably the best available to volume-
size crystals of this type. Still, results like the sonicated sample in
Table 5-4 caution against overconfidence even with this method.
5.1.3 Platelets - Number Count
The Coulter method measures the volume of particles in a size range.
For the purposes of mathematical scrubber models (1), the number-per-size-
range is needed. The volume percent in a size range may be converted to a
number percent by dividing by the volume. For large particles (>20y),
the differences are not great. But for very small particles, the change in
distribution is enormous. A tiny volume percent may account for 40% of the
number count. The volume and number percent histograms are given for eight
samples in Appendix A.
48
-------
The number counts are generally exponential-type population curves.
They are best analyzed as log plots where true exponential behavior results
in straight lines.
At larger particle sizes (BSD >10u, which corresponds to a width at
least >20vO, all four of the sieved platelet batches show nearly identical
behavior; see Table 5-5.
TABLE 3-5. LOG-NUMBER SLOPES OF SIEVED PLATELETS FROM FULL-SIZE
SCRUBBER
Particle
Size (y)
13-20
13-20
8-13
5-8
Sonic
Dispersion
No
Yes
No
No
>10y
-1.38
-1.31
-1.17
-1.29
Slope
Near 5y
-2.00
-1.91
-3.77
-4.20
Slope
<2y
-4.92
-5.33
-9.80
-1.47
Again, these four platelet samples were sieved consecutively from a full-
size scrubber's sludge, and they therefore contain fly ash.
Since no particles were optically observed in the large size range (>10y),
these counts probably represent agglomerates or coincidence counts. The curve
in this region is a very good exponential, and either of the above explana-
tions should produce such behavior. That is, the occurrence of agglomerates
or coincidence counts is proportional to the total number of particles.
If the sieves do "knock-out" larger particles but discriminate little
among the smaller particles, percent number count curves should be steeper
for the smaller size range batches. This trend was indeed observed for these
platelet samples; see Table 5-5. The intermediate size region, near 5y
BSD or 10-15y platelet width, is the region of sieving size discrimination.
While the volume count showed some differences between sonicated and non-
sonicated samples, the number count gives almost identical results for both.
The number count for the smaller size batches would also be expected to
begin at smaller sizes. This behavior is observed for the 5-8y and 8-l3y
ranges; for the 13-20y range, the slope differs so little from the larger
particle slope (coincidence event background) that it is difficult to tell
where the straight line for the intermediate sized particles begins, and
the behavior may well be within expected bounds. Refer to Figures A-10
and A-12 in Appendix A.
At smaller size ranges (<5u ESD), nonexponential behavior is encountered.
Both the 5-8y and 8-13y range platelet batches show local maxima. Pre-
sumably the number count would sharply decline if not for the very steep
49
-------
slopes found below 2u. Similar transition regions (between good straight
line-exponential regions) are also found in the number count displays of
the pilot scrubber platelets.
Regions of no increase or an actual decrease in the number count indicate
that the exponential-type growth mechanism is no longer operative. These
regions exist in most number-count curves, including the optical count curves.
Apparently, a preferred size is generated by certain mechanisms. In fact,
the 2-10y region of Figure A-14 appears Gaussian shaped; i.e., of the general
form,
e-a(x-x0)
(5-4)
Breakage of certain size particles or clustering or agglomeration of particles
governed by some size preferential mechanism could result in such a curve,
but verification of such speculation is beyond the scope of this project.
The very steep curve below 1.5y in Figure A-14 is probably due to fly
ash. Indeed, the SEM reveals large numbers of spherical fly ash particles
below 2u. Thus, in full-scale scrubbers, the fly ash will probably mask
the population statistics of CaSCh platelets below 2y. In both platelet
and granular samples grown in laboratory and pilot plant scrubbers, where no
fly ash is present, such steep slopes are not seen.
The CaSOa platelets grown in the RTF pilot plant scrubber exhibit
sizing statistics similar to those of the platelets from a full-scale
scrubber. Two or three regions show rather straight lines on log paper,
usually with a plateau region in the intermediate size range (5-lOy ESP).
These transition regions are not believed to be due to sieving because:
(1) the sieving did not effect smaller size ranges, and (2) the largest
size batch of the full-size scrubber sample which physically sieved most
efficiently, showed no plateau. Again, certain of the curves display Gaussian
characteristics (see Figures A-18, A-20). The slopes of the rather well
defined straight lines are given in Table 5-6.
TABLE 5-6. LOG-NUMBER SLOPES OF SIEVED PLATELETS FROM RTP PILOT SCRUBBER
Particle
Size (y)
>20
13-20
13-20
8-13
Sonic
Dispersion
No
No
Yes
No
Slope
>10y
-0.90
-1.87
-1.46
-1.28
Slope
-1.86
^0
M)
-3.60
Slope
-4.49
-6.12
-10.30
-6.19
!"• 1 -
50
-------
The slopes of the 13-20y samples near 3y were nearly flat, but no straight
line could be defined.
In conclusion, it seems that the Coulter method gives internally con-
sistent results when analyzing samples from a common source. Comparison
between samples with different histories may not be valid. Volume-counts
tend to reflect what the eye "sees," but number-counts are more useful for
mathemetical models. The number-count curves tend to give straight lines
when plotted on log paper, an indication of exponential population behavior.
The plateau regions unpredictably encountered in sieved samples may or may
not reflect the efficiency of the sieving. Certainly, they are nonexponential
regions; in fact, several curves appear rather Gaussian. Population regu-
lating processes appear to be multimodal and no easily identifiable mecha-
nisms suggest themselves. Possibly, the steep slopes (<-10) below 2y are
due to fly ash which can be seen in SEM photos adhering to larger particles.
51
-------
SECTION 6
NUCLEATION STUDIES
6.1 INTRODUCTION
One of the secondary goals of this project was to study CaSOs crystal
nucleation and growth processes in order to produce a higher quality sludge-
quality here, is defined by settling rate and dewaterability. CaSOs
platelets certainly settle faster than granules [1] and they generally
dewater better. The goal then was to study certain process variables
to see which ones had an effect on crystal growth and nucleation.
Ottmers, et al. [3] have reported that CaS03 supersaturation levels
of up to 3 resulted in no spontaneous precipitation. They did note that
precipitation below this level could be induced by seeding.
In laboratory batch crystallization experiments without seeding it was
not possible to produce platelets. Granules invariably formed independent of
solution temperature, solution pH, stirring rate, and changes in CaSOa concen-
tration. Occasionally, a few platelets formed initially, but after more
crystal growth, granules were found to be in the vast majority.
Growth of platelets was possible only with platelet seeding. Slightly
supersaturated CaSOs solutions, when seeded with platelets, did yield pre-
dominately platelets. Granule seeds produced only granules. If supersatura-
tion ever passed a level where nucleation became spontaneous (about 3 times
saturation) or the CaSOs concentration changed rapidly enough for one to
notice momentary changes in opacity, granules were the result. Hence, only
platelet seeding followed by slow growth (over periods of hours) would pro-
duce platelets.
It was found that solution pH was critical in crystal nucleation. At
concentrations of Ca and SO 3 around .01 to .05M*, no precipitate spon-
taneously formed at pH lower than 4.6. Solid CaSOa dissolved in aqueous
solutions below about 3.2 pH. Equilibrium calculations indicate that
bisulfite, HSOa , will predominate below about pH 3.5 for this concentration
range. It was much simpler to initiate crystallization in a controlled
manner by pH variation rather than by direct concentration variation.
*M represents molar concentration. Technically the concentration should
be a formality, but molarity is far more commonly used.
52
-------
is always a contaminant in SO3 studies. Extreme measures, like
double gaseous barrier insulation, are necessary to prevent oxidation of the
sulfite. For example, purified Na still contains enough Oa so that if labora-
tory solutions are purged with Nz, some of the SOs will be oxidized in the
process. At temperatures above 40°C oxidation is quite rapid (seconds)
and complete to the extent possible, limited only by available Oa . Also,
CaSOn is relatively water insoluble and pH levels below 2 are required to
dissolve it (to HSOiT).
6.2 GRANULES
Little need be said about nucleation mechanics and kinetics for granules.
Any rapidly precipitating solution will produce granules. Without seeding
they form more readily than do platelets. As the concentration (or pH) of
a non-precipitating saturated solution is increased, the driving force toward
crystallization increases. At some point the first crystals will form as
tiny platelets. But by that time the supersaturation levels are high enough
to cause incipient nucleation and growth, and many platelets grow out of the
original crystal in a rapid, haphazard fashion. An irreversible change
takes place. The product is a nearly spherical ball consisting of hundreds
of platelets cocked at all angles. Figure 4-4 shows an excellent sample
of this type of granular growth.
Thus, granules are really spheres of multiple platelets jammed awk-
wardly together. This morphology encloses many dead spaces and results
in voids between the platelets. This accounts for the lower dry density
of the granules (1.25), compared to the platelets (1.39), and explains
why they are difficult to dewater. The entropy of the granules is apparen-
tly greater than that of the platelet crystals, and granules are the mor-
phology of choice for rapid crystallization and growth.
Granules grown at higher supersaturation levels (hence, more rapid
growth) appeared under the microscope to be slightly smaller than more
slowly grown granules. In any case, the spheres formed were 2-10y in
diameter, with a much tighter range in any one sample; see Figure 6-1.
The spheres immediately formed into agglomerations.
Due to their much larger surface area, the granules dissolved more
rapidly in acid than did platelets. When granules and platelets formed a
mixed population and mild acid (pH 3) was added, the granules quickly broke
up into hundreds of tiny platelets which subsequently quickly dissolved.
Meanwhile, nearby platelets were more resistant to dissolution. In a mixture
of granules and platelets, it is possible to dissolve nearly all the granules
before dissolving very many platelets. This process is probably what happens
in a S02 scrubber tower where the return tower make-up liquors still contain
many calcium sulfite crystals. In the tower they are briefly exposed to the
acidic environment created by the dissolved S02. The S02 contactor may be
working as a fines destruction device, operating principally on the granules,
and contributing greatly to the production of platelets as the final dominant
product form.
53
-------
Figure 6-1. Optical micrograph of granules grown from seeded supersaturated
CaSOa solutions. Note the size uniformity and agglomeration.
54
-------
6.3 PLATELETS
In batch preparations, platelets could be induced only by seeding with
other platelets. Stirring speed and temperature had no apparent effect
(i.e., under microscopic examination) upon the size or type of crystals
formed. The concentration of Ca4"1" and SOs* were the principal determining
factors. As was mentioned earlier, it was found to be easier to control
the S03= concentration by varying the pH (IT1" + S03=+ HSOs") than by adding
As the pH of solutions at about . 04M Ca"1""1" and S03~ was slowly increased
by the addition of aqueous NaOH, crystals formed spontaneously at a pH of
about 4.8. If NaOH addition was stopped, more crystals continued to form
(invariably granules) , and the pH fell back (in several hours) to a value of
about 3.5. In more concentrated solutions, spontaneous crystallization began
at lower pH; for example, at .2M in Ca and 80$ =, cloudiness appeared at
pH 4.2. Temperatures differences between 0 C and 60°C had no noticeable
effect on the pH at which spontaneous crystallization occurred.
A solution of .04M CaSOs at a pH of 4.5 without seeds would set for
days with no crystallization, indicating that the threshold for spontaneous
nucleation had not been crossed. Thus, nucleation experiments were carried
out by adjusting, from below, the pH of a .04M solution to pH 4.5. Seed
crystals were added and the container was sealed and allowed to stand at
constant temperature for 16 hours. Crystals formed in the first few hours
and further standing past 16 hours did not appear to have any effect. Tem-
perature ranges of 0 C to 60 C also appeared to make little difference. All
examinations were made with the optical microscope.
Seed crystals were sieved into different size batches. The amount of
seed crystals added varied between 1 and 10 percent of the final product,
again resulting in little apparent difference. No matter what size range
was used for seed or how much seeding was used, the products looked very
much alike in size; see Figure 6-2. The crystals seemed to grow to a certain
surface area in size and few larger platelets were observed. That is, at
some point the preferential growth coordinate becomes the thickness. This
maximum area was typically 200-400y2 (e.g., 15 x 20y or 18 x 18y) for these
growth experiments; again see Figure 6-2. Apparently, in spite of ample
opportunity, under these conditions larger crystals rarely form. There was
little chance for mechanical breakage; the few smaller, irregular crystal
fragments seen confirmed this.
The size distribution of the laboratory grown samples showed an inordin-
ately large percentage of certain large crystals; in fact, they looked more
like the sieved sample (where one size range predominates) than a random
sample- Recall that the seed crystals were only about 5% of the final
crystal mass. A similar frequent size was seen in scrubber samples as well.
The maximum predomonant size there was somewhat larger, up to 800y2, although
55
-------
(b)
Product platelets from different seeding growth conditions.
a) sieved seed crystals, 13-20y
b) product grown from 13-20y seed at 0 C, pH 4.5 ->• 3.42;
5% seed
c) product grown from 13-20y seed at 25 C, pH 4.5 -»• 3.42;
5% seed
d) product grown from 13-20y seed at 25 C, pH 4.65 ->• 4.20-
1% seed
56
-------
a few crystals as large as 1500y2 have been seen. Although time and condi-
tions were available for "extremely" large crystals to form, none were seen.
The constancy in size of the largest crystals in a batch, and more approxi-
mately from one batch to another, suggests that some growth limiting process
(at least in area) appears to be operative.
There are many other variables that will affect crystal growth. The
total ionic concentration or a host of surfactants are two known factors
that can drastically alter crystal growth patterns. At one point in this
work, buffers were used to maintain a constant pH during growth. However,
the total ionic concentration was considerably different with the buffer,
and a completely new crystal geometry resulted. This points to another po-
tentially fertile area of research for improving the quality of sludge.
Time and resources did not permit more extensive studies in this area.
57
-------
SECTION 7
DISCUSSION OF RESULTS
7.1 SUMMARY OF VISUAL SIZING METHODOLOGY
The specifics of counting from photographs have been dealt with in
earlier sections. This section will compare alternate visual methods of
counting and arrive at more generalized conclusions.
Conventional optical microscopes are adequate for identifying and count-
ing particles as small as a few microns. There may be problems of prepara-
tion, mounting or defining the particles or their geometry, but they are
inherent to any type of microscopy. Depending somewhat on shape, particles
below 4 or 5 microns are less well resolved in shape and size and a few
are hidden or simply missed. For particles below 2 microns in size, opti-
cal microscopy is simply inadequate for size counting. All other methods
indciate that large numbers of micron and submicron particles are missed
under the best of optical conditions.
The counting of particles seen in optical micrographs is slow, tedious
and discretionary. The problems of preparation, mounting and definition of
particles mentioned in the previous paragraph are not to be casually dis-
regarded. They may be, in fact, insurmountable as is the case for CaSOs
granules. Increased magnification by immersion techniques allows only a
slightly improved resolution of the particles. But the higher the magni-
fication, the smaller the field of view and the smaller the sample
per field of view. Indeed, if all other difficulties can be tolerably over-
come, the size of the sample that can be reasonably counted may still be
far smaller than is desirable. In the end, the trade-off between magni-
fication and the size of the sample counted must be suited to the objective
The problem of sample size is even more acute for SEM studies.
The magnification is readily available and the resolution is high, but
to generate a reliable result would require many micrographs and much
counting. The technique is also costly and more time consuming than opti-
cal microscopy. And the mounting and preparation difficulties may be even
more complex for SEM.
58
-------
In conclusion, microscopy appears to be a poor technique for size
counting. The optical microscope is relatively quick and easy. It is a
good way to survey the sample for general characteristics such as general
size, different morphologies and varieties of particles present. It is even
an acceptable method for obtaining an approximate count for particles larger
than 5 microns.
The SEM, although slower and more costly, provides more detail. It
is best used to survey the small particle range (below 5 microns) and to
more carefully determine crystal shape, especially for complex particles.
The combined optical count method could never be used for routine exami-
nation. Its chief use would be to complement instrumental methods being
used or considered. Also, the instrumental methods do not directly distin-
guish between different types of particles, and for that reason a quick,
optical inspection of any sample is advisable.
The count of CaS03 platelets (both with and without sonic dispersion)
for several fields is shown in Figure 7-1. In this figure, the large num-
bers of submicron particles believed to be missed on 200X to 400X magnifi-
cation micrographs have not been included. If they were, the number of
particles below 2y would vastly increase; however, the total volume of fine
particles would remain below 1%. This total sample represents 766 particles
counted, a number large enough to have some statistical credibility. The
shape of the curve is in general agreement with instrumental counts dis-
cussed in Section 3.
7.2 COMPARISON OF VISUAL AND INSTRUMENTAL RESULTS
Some comparisons between visual and instrumental methods were included
in Sections 4 and 5. There, a discussion of sizing involved reference to
all possible techniques. The result of that comparison was that very little
about the particle volume for platelets can be discerned using optical
techniques. The platelet thickness varies widely, and the statistical
averages resulting from instrumental counting are deceptively simplistic.
That is, the clean, numerical results belie the complexities involved in
size-counting a crystal distribution.
Because the thickness of platelets appears to have a broad range and
cannot be measured optically, visual sizing (to volume) is nearly fruitless.
For granules, even the definition of a particle is nearly impossible, and
visual counting is impractical. Optical inspections can tell if all the
particles are of the same general morphology and substance—information not
readily available from instrumental methods. Visual methods could be useful
for gross size approximations of similar samples. For example, the differ-
ences between the sieved samples were readily apparent at 200X using an
optical microscope in transmission.
59
-------
Num
25
20-
15-
10-
0
\
NUMBER
V
\
COUNT
^~- Number %
'
.x
\f
\
\
V
,x-
H
Cu
*v
mul
|«^«
^<»
ati
>*-,
ve % Greater Than
i 1
.— .j
i 1
^s'Xl 0
^ r — ,
4
u
i
a««-_L [ i !
rlOO
• 80
Cum
• 60 %
• 40
' 20
0
0 5 10 15 20 25 30 35 40
WIDTH (u)
(a)
45 50
Num
%
25-
20-
15
10
5 -
0
VOLUME COUNT
Cumulative % Greater Than
Number %
Figure 7-1.
10 15
20 25
WIDTH
flOO
80
C
60
Cum
20
30 35 40 45 50
(b)
Graphical display of composite SEM counting (combined soni-
cated and non-sonicated counts).
a) number-count
b) calculated corresponding volume count
60
-------
The SEM confirms that the thickness will vary even in the same batch
of crystals. This may be due to different growth patterns, different histories
or breakage. But the SEM is not a practical way to size a group of crystals.
Optical microscopy is easier and less expensive than SEM and adequate for
cursory inspection, but it is tedious and inherently subject to error.
Neither method would be practical for a large amount of counting.
The Coulter method (and presumably sedimentation techniques) indiscrim-
inately counts large numbers of particles. It is less subject to the vagar-
ies of preparation. However, it is not free of inconsistencies, particularly
in the volume-count statistics. A method that seems to give some order to
the data is to plot the log of the number-count versus size; straight lines
usually result. What an exponential growth population means is not within
the scope of this work, but the result is in the form required by the mathe-
matical model.
The granular CaSOs particle sizing results from the Coulter method
appear fairly regular and straight-forward. However, for the purposes of the
mathematical model, this approach may be misleading. The SEM shows the
"particles" to actually be clusters of thin platelets growing radially out
of nucleation centers. The number of individual crystals or nucleation
events is undefinable. And, as with platelets, agglomerations are always
present but to an undeterminable extent. Thus, although the granular sizing
results seems to be internally consistent, their true meaning is uncertain.
One of the purposes of this project was to determine the best way to
size-count particles and to make that information compatible with the mathe-
matical modeling presented elsewhere [1]. The Coulter method is clearly
the best choice to that end. It should be very good for particles with
similar histories, such as repeated samples from a pilot scrubber. Optical
or SEM inspection from time to time is strongly advised to determine general-
ized geometry and the presence of foreign particles. The visual inspection
would be particularly necessary on a full-scale scrubber to check for lime-
stone build-up or fly ash, either of which can significantly distort the
Coulter results.
7.3 NUCLEATION
The nucleation studies showed that the pH is a very critical parameter
in regulating the crystal growth. Platelets can grow only slowly and with
seeding; granules are the fast-growth morphology. Among the other factors
affecting growth patterns which may be operative in scrubbers, the total
ionic strength of the solution appears to be most influential.
The nucleation studies support the conclusion reached elsewhere that
at first growth occurs preferentially along the length and width coordinates;
later, after a size is reached which depends upon solution conditions to
some extent, the crystals thicken in the middle. Further growth appears to
occur primarily in thickness. Thus, nucleation studies also support the
conclusion reached elsewhere that judging crystal size by looking at the
platelet area is ill-advised.
61
-------
SECTION 8
REFERENCES
1. Phillips, J. L., et al., "Development of a Mathematical Basis for Re-
lating Sludge Properties to FGD-Scrubber Operating Variables."
EPA-600/7-78-072 (NTIS PB281582), April 1978, (Radian Corporation).
EPA Project Officer: Robert H. Borgwardt.
2. Randolph, Alan D. and Maurice A. Larson, "Theory of Particulate
Processes. Analysis and Techniques of Continuous Crystallization."
New York, Academic, 1971.
3. Ottmers, D. Jr., et al., "A Theoretical and Experimental Study of the
Lime/Limestone Wet Scrubbing Process." EPA-650/2-75-006, (NTIS PB
243399), December 1974, (Radian Corporation). EPA Project Officer:
Julian W. Jones.
62
-------
APPENDIX A
INSTRUMENTAL DATA
Some of the raw data, along with an alternate presentation, are collec-
ted in this appendix to avoid discontinuity in the body of the report. The
odd-numbered figures display some raw data as received from Coulter Elec-
tronics which did the analysis. The subsequent even-numbered figure is a
logarithmic plot of the number count for that sample. On these plots
are given the slopes for each straight line segment that is reasonably well
defined. All sizes in this appendix are in equivalent spherical diameters
(BSD) in microns (urn).
63
-------
GRANULES (FULL-SIZE SCRUBBER)
SIZE RANGE, 8-13„
NUMBER PERCENT COUNT
5 10
MICRON DIAMETER Log Scale
50
100
Figure A-l. Granules in size range 8-13|i; number count.
-------
X
-------
100-
90-
80-
70-
60-
. SO-
W-
SO-
20-
10-
0-
GRANULES (FULL-SIZE SCRUBBER)
SIZE RANGE, 13-20u
VOLUME PERCENT COUNT
MICRON DIAMETER Log Scale
100-
90-
80-
70-
60-
« 50-
40-
30-
20-
10-
0-
GRANULES (FULL-SIZE SCRUBBER)
SIZE, RANGE 13-20U
NUMBER PERCENT COUNT
-i—i f i i M| r—T 1 i] I i li | T-i 1—I | ! i II | r-T-
.5 1.0 5 10 50 100
MICRON DIAMETER Log Scale
Figure A-3. Granules in size range 13-20U.
Top: Volume Count
Bottom: Number Count
66
-------
X
a
c
s_
GRANULES
NUMBER VS ESD
13-20u
FULL-SIZE SCRUBBER
1
0 2 4 6 8 10 12 14 16 18 20 22 , 24 26 28
ESO (u)
Figure A-4. Logarithmic display of number count in Figure A-3.
67
-------
100-
90
80
70'
60-
i 50
40
30
20
10
0
GRANULES (FULL-SIZE SCRUBBED
SIZE RANGE, >20y
VOLUME PERCENT COUNT
-i 1—r | i i np^t-i 1 i | i i m "—i 1—i | • 11 q r-r
.5 1.0 5 10 50 TOO
MICRON DIAMETER Log Scale
GRANULES (FULL-SIZE SCRUBBER)
SIZE RANGE ,>20y
NUMBER PERCENT COUNT
50 TOO
MICRON DIAMETER Log Scale
Figure A-5. Granules in size range >20u; act sonicated.
Top: Volume Counc
Bottom: Number Count
68
-------
X
a>
5-
1 •>
5 -
0
O
GRANULES
NUMBER VS ESD
>20U, NOT SONICATED
FULL-SIZE SCRUBBER
0 I 4 6 3 10 12 14 16 18 20 22 24 26 28
ESO (u)
Figure A-6. Logarithmic display of number count in Figure A-5.
69
-------
(FULL-SIZE SCRUBBER)
SIZE RANGE, >20U; SONICATED
VOLUME PERCENT COUNT
1.0 5 10
MICRON DIAMETER Log Scale
50 100
GRANULES (FULL-SIZE SCRUBBER)
SIZE RANGE, >20U; SONICATED
NUMBER PERCENT COUNT
100
MICRON DIAMETER Log Scale
Figure A-7. Granules in size range >20u; sonicated.
Top: Volume Count
Bottom: Number Count
70
-------
X
t)
•g
5-
1 -
5 •
5 -
1
ESD X5
GRANULES
NUMBER VS ESO
>20u; SONICATED
FULL-SIZE SCRUBBER
10 12 14 16 13 20 22 24 25 28
• 0 2 4 6
ESO (u)
Figure A-3. Logarithmic display of number count in Figure A-7.
71
-------
100
90 -
80-
70-
60-
50 -
40 -
30 -
20
10
0
PLATELETS (FULL-SIZE SCRUBBER)
SIZE RANGE. 13-20y
VOLUME PERCENT COUNT
1.0
10
MICRON DIAMETER Log Scale
50
100
Figure A-9. Platelets in size range 13-20p; not sonicated.
Volume Count.
72
-------
PLATELETS
NUMBER VS ESO
13-20U,NOT SONICATED
FULL-SIZE SCRUBBER
0 2 4 6 8 10 12 14 16 \13 20 22 24 26 28
ESO (y)
Figure A-10. Logarithmic display of number count calculated from
volum* count in Figure A-9.
73
-------
PLATELETS (FULL-SIZE SCRUBBER)
SIZE RANGE, 13-20u; SONICATED
VOLUME PERCENT COUNT
-i—I-T-I—i" | I II11 T
.5 1.0 5 10
MICRON DIAMETER Log Scale
PLATELETS (FULL-SIZE SCRUBBER)
SIZE RANGE, 13-20u; SONICATED
NUMBER PERCENT COUNT
MICRON DIAMETER Log Scale
Figure A-11. Platelets ia ai2e range 13-20U; sonicated.
Top: Volume Count
Bottom: Number Count
74
-------
X
I
01
•9
5-3
5 •
5 -
1 -
5 •
•ESD XI0
vrS.33
PLATELETS
NUMBER VS ESO
13-20u; SONICATED
FULL-SIZE SCRUBBER
8 10 12 14 16 \18 20 22 24 26 23
0246
ESD (U)
Figure A-12. Logarithmic display of number count in Figure A-ll.
75
-------
100-
90-
80-
70-
60-
504
40-
30-
20-
10-
0-
PLATELETS
(FULL-SIZE SCRUBBER)
SIZE RANGE, 3-13u
VOLUME PERCENT COUNT
1 I " " I ^ ,,,,,,, . .
.5 1.0 5 10
MICRON DIAMETER Log Scale
SO 100
(FULL-SIZE SCRUBBER)
SIZE RANGE, 8-13u
NUMBER PERCENT COUNT
1.0 5 10
MICRON DIAMETER Log Scale
Figure A-13. Platelets in size range 3-1Ju.
Top: Volume Count
Bottom: Number Count
100
76
-------
X
01
I
s.
PLATELETS
NUMBER VS ESD
8-13
FULL-SIZE SCRUBBER
O
0 2 4 6 8 10 12 14 16 18 20 22 24 25 28
ESO (U)
Figure A-14. Logarithmic display of number count in Figure A-13.
77
-------
100-
90-
80-
70-
60-
50-
40-
30-
20-
10-
0-
PLATELETS
(FULL-SIZE SCRUBBFR)
SIZE RANGE, 5-8u
VOLUME PERCENT COUNT
.5 1.0
5 10
MICRON DIAMETER Log Scale
50
(FULL-SIZE SCRUBBER)
SIZE RANGE. 5-8u
NUMBER PERCENT COUNT
.5 1.0 5 10 50 100
MICRON DIAMETER Log Scale
Figure A-15. Platelets in size range 5-8u.
Top: Volume Count
Bottom: Number Count
78
-------
X
•o
c
L.
NUMBER VS ESD
5-3u
FULL-SIZE SCRUBBER
0 2 4 6 3 10 12 14 16\ 18 20 22 24 26 28
ESD (li)
Figure A-16. Logarithmic display of number count in Figure A-15.
79
-------
100-
90-
80-
70-
60-
i 50-
40-
30-
20-
10-
0-
J»UTELETS
(PILOT SCRUBBER)
SIZE RANGE, >20u
VOLUME PERCENT COUNT
1.0 5 TO
MICRON DIAMETER Log Scale
(PILOT SCU88ER)
SIZE RANGE, >20U
NUMBER PERCENT COUNT
50 100
MICRON DIAMETER Log Scale
Figure A-17. Platelets in size range >20u.
Top: Volume Count
Bottom: Number Count
80
-------
X
(U
I
5 •
1 •
5-
1
O
O
•ESD X5
v-4.49
PUTELETS
NUMBER VS ESO
>20U
PILOT SCRUBBER
0 2 4 5 8 10 12 14 16 18 20 22 24 26 28
ESO (w)
Figure A-13. Logarithmic display of number count in Figure A-17.
81
-------
(PILOT SCRUBBER)
SIZE RANGE, 13-20U
VOLUME PERCENT COUNT
MICRON DIAMETER Log Scale
(PILOT SCRUBBER)
SIZE RANGE, 13-20U
NUMBER PERCENT COUNT
1.0 5 10
MICRON DIAMETER Log Scale
Figure A-19. Platelets in size range 13-20U; not sonicated.
Top: Volume Count
Bottom: Number Count
82
-------
I
5-
1 •
5-
1 -
5-J
5-
1
•ESO X5
PLATELETS
NUMBER VS ESO
13-20U;NOT SONICATED
PILOT SCRUBBER
02 4 6 8 10 12 14 16 18 20 22 24 26 28
ESO (u)
Figure A-20. Logarithmic display of number count in Figure A-19.
83
-------
100-
90-
80-
70-
60-
50-
40-
30-
20-
10-
0-
. PLATELETS
(PILOT SCRUBBER)
SIZE RANGE, 13-20u; SONICATED
^VOLUME PERCENT COUNT
i I i 11 11 i i i I i i i 11 [ 1—r—
.5 1.0 5 10
MICRON DIAMETER Log Scale
50 100
PLATELETS (PILOT SCRUBBER)
SIZE RANGE, 13-20p; SONICATED
NUMBER PERCENT COUNT
50 100
MICRON DIAMETER Log Scale
Figure A-21. Platelets in size range 13-20u; sonicated.
Top: Volume Count
Bottom: Number Count
84
-------
5-
1 -
1 -
5 •
5 -
PLATELETS
NUMBER VS ESD
13-20U; SONICATED
PILOT SCRUBBER
—I 1 1 I 1 I I 1 1 1 ' I
6 2 4 6 8 10 12 14 16 18 20 22 24 26 28
ESD (u)
Figure A-22. Logarithmic display of number count in Figure A-21.
85
-------
(PILOT SCRUBBER)
SIZE RANGE, 8-13u
VOLUME PERCENT COUNT
1.0 5 10
MICRON DIAMETER Log Scale
SO 100
100-
90-
80-
70-
60-
»« 50-
40-
30-
20-
10-
0-
PLATELETS
(PILOT SCRUBBER)
SIZE RANGE, 8-13U
NUMBER PERCENT COUNT
MICRON DIAMETER Log Scale
50
100
Figure A-23. Platelecs in size rang* 3-13u.
Top: Volume Count
Bottom: Number Count
86
-------
PLATELETS
NUMBER VS ESO
8-13u
PILOT SCRUBBER
ESD X5
i 1 1 1 1 1 1
0 2 4 6 8 10 12 14 16 18 20 22 24 26 28
ESD (u)
Figure A-24. Logarithmic display of number count in Figure A-23.
87
-------
TECHNICAL REPORT DATA
(Please read Instructions on the reverie before completing)
1. REPORT NO.
EPA-600/7-79-192
2.
3. RECIPIENT'S ACCESSION1 NO.
4. TITLE AND SUBTITLE
Calcium Sulf ite Crystal Sizing Studies
5. REPORT DATE
August 1979
6. PERFORMING ORGANIZATION CODE
7. AUTHOR(S)
8. PERFORMING ORGANIZATION REPORT NO.I
Larry O. Edwards
9. PERFORMING ORGANIZATION NAME AND ADDRESS
Radian Corporation
8500 Shoal Creek Boulevard
Austin, Texas 78766
10. PROGRAM ELEMENT NO.
EHE624
11. CONTRACT/GRANT NO.
68-02-2608, Task 30
12. SPONSORING AGENCY NAME AND ADDRESS
EPA, Office of Research and Development
Industrial Environmental Research Laboratory
Research Triangle Park, NC 27711
13. TYPE OF REPORT AND PERIOD COVERED
Task Final; Through 12/78
14. SPONSORING AGENCY CODE
EPA/600/13
13. SUPPLEMENTARY NOTES TjgRL-RTP project officer is Robert H. Borgwardt, Mail Drop 65,
is. ABSTRACT
repOrt describes a reliable experimental method that can be used rou-
tinely to determine the crystal size distribution function, a measure that is required
for a mathematical representation of the nucleation and growth processes involved in
the settling, dewatering, and disposal of calcium sulfite sludge from lime/limestone
or dual alkali scrubbers, a major problem associated with coal burning. (A recent
EPA report presented a mathematical description of the SO2 scrubbing process , but
found discrepancies in the particle size distribution function when measured by dif-
ferent techniques.) Optical and instrumental crystal sizing methods were compared
and the merits of each discussed; Coulter counting is recommended. The two primary
crystal forms, platelets and granules, were shown to be related: granules are clus-
ters of platelets. Platelets, the preferred shape, form only in slow growth condit-
ions and require seeding. Particle clusters complicate the crystal definition and
counting processes. The crystal size distribution was shown to be the sum of seve-
ral exponential population curves. Fly ash, if present, will dominate counts for par-
ticles smaller than i micron.
17.
KEY WORDS AND DOCUMENT ANALYSIS
DESCRIPTORS
b.IDENTIFIERS/OPEN ENDED TERMS
COSATl Field/Group
Pollution
Calcium Oxides
Sulfites
Sludge
Crystals
Measurement
Alkalies
Mathematical Mo-
dels
Nucleation
Crystal Growth
Scrubbers
Calcium Carbonates
Sulfur Oxides
Pollution Control
Stationary Sources
Calcium Sulfite
Dual Alkali System
Instrumental Measure-
ment
13 B
07B
07A
20B
14 B
07D
12A
13. DISTRIBUTION STATEMENT
Release to Public
19. SECURITY CLASS (This Report)
Unclassified
21. NO. OF PAGES
96
20. SECURITY CLASS (This page)
Unclassified
22. PRICE
EPA Form 2220-1 (9-73)
88
------- |