EPA-600/7-91-009
December 1981
MEASUREMENT AND PREDICTION OF THE RESISTIVITY OF
ASH/SORBENT MIXTURES PRODUCED BY SULFUR OXIDE CONTROL PROCESSES
by
Ronald P. Young
Environmental Sciences Department
Southern Research Institute
Birmingham, Alabama 35255
Cooperative Agreement No. CR812282
EPA Project Officer
Louis S. Hovis
Air and Energy Engineering Research Laboratory
U. S. Environmental Protection Agency
Research Triangle Park, North Carolina 27711
Prepared for:
U. S. Environmental Protection Agency
Office of Research and Development
Washington, DC 20460
-------
TECHNICAL REPORT DATA
(Please read Instructions on the reverse before compieti-
^"OFITNO. 2.
-600/7-91-009
3. 1
4. TITLE AND SUBTITLE
Measurement and Prediction of the Resistivity of Ash/
Sorbent mixtures Produced by Sulfur Oxide Control
Processes
5. REPORT DATE
December 1991
6. PERFORMING ORGANIZATION CODE
7. AUTHORiSl
Ronald P. Young
6. PERFORMING ORGANIZATION REPORT NO.
SRI-ENV-88-1076
9. PERFORMING ORGANIZATION NAME AND ADDRESS
Southern Research Institute
P. O. Box 55305
Birmingham, Alabama 35255-5305
10. PROGRAM ELEMENT NO.
11. CONTRACT/GRANT NO.
CR812282
12. SPONSORING AGENCY NAME AND ADDRESS
EPA, Office of Research and Development
Air and Energy Engineering Research Laboratory
Research Triangle Park, North Carolina 27711
13. TYPE OF REPORT AND FERIOO COVEREO
Final; 9/86-6/88
14. SPONSORING AGENCY CODE
EPA/600/13
is.supplementary ncipr0jec); officer Louis S. Hovis is no longer with the Agency
For details, contact Samuel L. Rakes, Mail Drop 4, 919/541-2828!
The report describes the development of (1) a modified procedure for obtai-
ning consistent and reproducible laboratory resistivity values for mixtures of coal
fly ash and partially spent sorbent, and (2) an approach for predicting resistivity
based on the chemical composition of the sample and the resistivities of thekey
compounds in the sample that are derived from the sorbent. Furnace and cold- side
sorbent injection technologies for reducing the emission of sulfur oxides from elec-
tric generating plants firing medium- to high-sulfur coal are under development for
retrofit applications. The particulate resulting from injecting this sorbent will be
a mixture of coal fly ash and partially spent sorbent. The presence of this sorbent
causes the resistivity of the mixture to be significantly higher than that of the fly ash
alone. Since higher resistivity dusts are more difficult to collect in an electrostatic
precipitator (ESP), accurate knowledge of the resistivity of the mixture is needed to
determine if the ESF will operate within an acceptable efficiency range. .1-
17. KEY WORDS AND DOCUMENT ANALYSIS
a. DESCRIPTORS
b.IDENTIFIERS/OPEN ENDED TERMS
c. COSATI Field/Group
Pollution Electric Power Plants
Electrical Resistivity
Coal Particles
Fly Ash Electrostatic Precipi-
Sorbents tators
Sulfur Oxides
Pollution Control
Stationary Sources
Particulate
13 B 10B
20C
21D 14G
21B
11G 131
07B
1S. DISTRIBUTION STATEMENT
Release to Public
19. SECURITY CLASS /Th^ Report)
Unclassified
21. NO. OF FAGES
54
20. SECURITY CLASS (Thispage)
Unclassified
22. PRICE
EPA Form 2220-1 (9-73)
-------
ABSTRACT
Furnace and cold-side sorbent injection technologies for reducing the
emission of sulfur oxides from electric generating plants firing medium- to
high-sulfur coal are under development for retrofit applications. The
particulate resulting from injecting this sorbent material will be a mixture
of coal fly ash and partially spent sorbent. The presence of this sorbent
material causes the resistivity of the mixture to be significantly higher than
that of the fly ash alone. And since higher resistivity dusts are more
difficult to collect in an electrostatic precipitator, accurate knowledge of
the resistivity of the mixture is needed to determine if the precipitator will
operate within an acceptable efficiency range. This report describes the
development of a modified procedure for obtaining consistent and reproducible
laboratory resistivity values for these mixtures, as well as development of an
approach for predicting the resistivity based .on the chemical composition of
the sample and the resistivities of the key compounds in the sample that are
derived from the sorbent.
ii
-------
CONTENTS
Abstract ii
Figures £v
Summary 1
Introduction 2
Development of Resistivity Measurement Procedure 2
Sample Characterization 3
Moisture Adsorption 3
Chemical Composition 3
Particle Morphology and Size Distribution 5
IEEE Resistivity Tests 10
Annealing Effects 17
Modified Resistivity Tests 23
Developing a Prediction Technique 31
Electrical Conduction Mechanise! 31
Correlation of Resistivity with Ionic Species 33
Ideas from Glass Research 34
Resistivity of Calcium Compounds 35
Simplified Model of Ash/Sorbent Mixtures 35
Determining Amounts of Calcium Compounds in Mixtures 39
Multiple Correlation Results 40
Conclusions 44
References 45
Appendix A. Quality Control Evaluation Report A-l
iii
-------
FIGURES
Number Page
1 BET area of the E-SOX sample as a function of time of exposure
to humidity. Circles are for exposure to ambient temperature
and humidity; squares are for exposure to flue gas conditions
(67°C and 11% water by volume) 4
2 Thermogravimetric analysis (TGA) of the E-SOX sample from 50 to
450°C in a nitrogen atmosphere 6
3 Thermogravimetric analysis (TGA) of the E-SOX sample from 50 to
450°C in air 7
4 Scanning electron microscope (SEM) photograph of particles in
the E-SOX sample at a magnification of 500X 8
5 Bahco particle size classification of the E-SOX sample.
Straight line is a linear, least-squares fit to the data ... 9
6 Resistivity of the E-SOX sample obtained according to the IEEE
Standard 548-1984 for resistivity measurements of fly ash . . 11
7 Resistivity for a single E-SOX sample obtained at successively
higher humidity levels 12
8 Comparison of resistivity data for a fresh E-SOX sample (E-SOX
2) and a sample tested once previously (E-SOX 1) 13
9 Comparison of resistivity data for a fresh E-SOX saaple (E-SOX
3) and samples tested once and twice previously (E-SOX 1 and
E-SOX 2, respectively) 14
10 Comparison of resistivity data for a fresh E-SOX sample (E-SOX
4) and samples tested once, twice, and three times previously
(E-SOX 1, E-SOX 2, and E-SOX 3, respectively) 15
11 Resistivity of E-SOX sample for various humidity levels. Data
for each humidity level were taken using a fresh sample ... 16
12 BET area of the E-SOX sample as a function of annealing
temperature in dry air . 18
13 BET area as a function of annealing temperature in an atmosphere
containing 10.4% humidity by volume 19
iv
-------
Numl
14
15
16
17
IS
19
20
21
22
23
24
25
FIGURES (Continued)
Page
Volume-vs-diameter data from mercury porosimetry measurements
of as-collected and annealed (450°C) E-SOX samples 21
Area-vs-diameter data from mercury porosimetry measurements of
as-collected and annealed (450CC) E-SOX samples 22
TGA of E-SGX sample: second run in nitrogen 24
TGA of E-SOX sample: second run in air 25
Resistivity of E-SOX sample taken in the ascending temperature
mode in dry air to a maximum temperature of 250°C 27
Resistivity of E-SOX sample taken in the ascending temperature
mode at various humidity levels to a maximum temperature of
250°C 28
Resistivity of E-SOX sample taken in the descending temperature
mode in dry air, beginning at 250°C 29
Resistivity of E-SOX sample taken in the descending temperature
mode at various humidity levels, beginning at 250°C 30
Comparison of measured resistivity values and those predicted
by the Bickelhaupt computer model (Model II, Reference 4) for
an E-SOX sample in an atmosphere containing 15% water vapor
by volume 32
Resistivities of calcium compounds expected to be present in
ash/sorbent mixtures 36
Comparison of the resistivities of an E-SOX sample, calcium
hydroxide, and calcium sulfite 37
Comparison of the resistivities of 4 LIMB samples (17 and 18
with hydrated lime injection, 101 and 119 with limestone
injection), calcium hydroxide, and calcium sulfate 38
Comparison of the measured resistivity of an E-SOX sample and
that predicted from a combination of the resistivities of
Ca(0H)2 and CaS03 in an atmosphere containing 10% water vapor
by volume 42
v
-------
SUMMARY
Injection of a sorbent material such as limestone or hydrated lime,
either into a furnace or the combustion flue gas downstream of the furnace,
will result in a particulate which is a mixture of coal fly ash and partially
spent sorbent material. It was found that the partially spent sorbent
material contained in these mixtures necessitated modifications to the
procedure normally used for measuring the resistivity of ordinary fly ash.
Application of the IEEE Standard 548-1984 procedure for measuring resistivity
of fly ash to measuring resistivity of an ash/sorbent mixture resulted in
Inconsistent values of resistivity and an overall lack of repeatability, as
well as large discrepancies between values measured using the ascending and
descending temperature methods. This behavior was traced to large changes in
the nature of the as-collected sample caused by heating to 450°C, as specified
by the Standard. In particular, it was found that calcium hydroxide in the
as-collected sample decomposed at about 300°C, accompanied by large changes in
the surface area of the sample. Then in order to preserve the nature of the
as-collected sample, it should not be exposed to temperatures above 250cC
during resistivity measurements. Even with this stipulation, resistivity
measurements made using ascending temperatures did not show good repeatabili-
ty: the first test on a sample gave resistivity values which were lower than
those obtained on subsequent tests at identical conditions. In contrast,
resistivity measurements made using descending temperatures showed excellent
repeatability and consistency. Therefore, resistivity measurements of
ash/sorbent mixtures should be made in the descending temperature mode
beginning at 250°C.
Multiple regression analyses of the resistivity data showed very strong
correlations between the resistivities of the ash/sorbent mixtures and the
resistivities of the calcium compounds found in the mixtures. In fact, the
resistivity of the mixture appeared to be primarily determined by the resis-
tivity of one or more of these calcium compounds. Based on these correla-
tions, considerable work was done to develop a method for predicting the
resistivity of an ash/sorbent mixture from its chemical composition. However,
these efforts were not generally successful. One evident source of difficulty
was the insufficient chemical analysis of the several calcium compounds in the
sample (consisting of both the residue of the original sorbent and the
products of reaction in the flue gas). Another possible source of difficulty
was the omission of any resistivity effect due to components of the fly ash or
products of reaction between the ash and the sorbent. Poor correlation was
obtained between the chemical compositions and resistivities of the mixtures.
But, for samples of sorbent only, exposed to SO« in the absence of fly ash,
more satisfactory agreement between chemical composition and resistivity was
obtained in following the same approach. It thus seems that this approach
should be successful for predicting the resistivities of ash/sorbent mixtures
provided that an accurate method of determining the chemical composition of
the mixtures can be developed.
1
-------
INTRODUCTION
Furnace and cold-side sorbent injection technologies are under develop-
ment for retrofit applications at electric generating plants firing medium- to
high-sulfur coal. Both coal fly ash and sorbent material are expected to be
collected in an existing electrostatic precipitator (ESP). In most candidate
sites for retrofit installations, the existing ESP is quite small (SCA
generally less than 300 ft2/1000 acfm or 60 m2/[m3/s]) since it was designed
to collect low-resistivity fly ash. The presence of the sorbent causes the
resistivity of the mixture to be significantly higher than that of the fly ash
alone. And since the resistivity of the particulate influences the collection
efficiency of the dust (higher resistivity dusts being more difficult to
collect), it is important to be able to accurately measure the resistivity of
the mixture to determine if the precipitator will operate within an acceptable
efficiency range.
DEVELOPMENT OF RESISTIVITY MEASUREMENT PROCEDURE
A total of 40 ash/sorbent samples produced from burning pulverized coal
and injecting sorbent material at various locations were used in this study.
Eighteen of these samples involved injection of the sorbent into the flue gas
downstream of the regenerative air heater. The other 22 samples were obtained
from the LIMB program where the sorbent material is injected into the burner.
Most of the work was performed with E-SOX dust, which is the residue from a
process in which a slurry of Ca(0K)2 is injected downstream from the air
heater. The remainder of the work, with LIMB dust from furnace injection of
limestone or Ca(0H)2 and with HALT dust from low-temperature injection of dry
Ca(0K)2, was performed to determine whether the results with E-SOX samples
might extend to the other types of materials.
Two additional samples that were studied were residues of slurries of
Ca(0H)2 that had been injected into a gas stream containing S02 but not
containing fly ash. These samples came from a reactor operated by Babcock &
Wilcox. The S02 concentration ranged from 1500 to 2500 ppm. The slurry
injected produced Ca/S mole ratios of 0.6 to 1.3 and lowered the gas tempera-
ture to the range of 22 to 33°C from saturation.
Many of the samples mentioned above were subjected to a battery of tests
designed to characterize them as completely as possible. These tests included
chemical analysis, particle size analysis, density, thermogravimetric analysis
(TGA), scanning electron microscope (SEM) photography, mercury porosimetry,
and surface area measurements. Because the important qualitative features of
the data were the same for all of the samples, the results reported here will
focus on the sample generated by burning Pittsburgh coal and spraying a slurry
of Longview hydrated lime into the a humidifying chamber in the Southern
Research pilot-scale combustor, with an entering flue gas temperature of about
150eC. This sample is designated the "E-SOX sample" because one type of
cool-sice lime slurry injection process is the EPA E-SOX process.
2
-------
SAMPLE CHARACTERIZATION
Moisture Adsorption
Because some calcium compounds, especially the hydroxide, have a
distinct affinity for water vapor, it was necessary to test the E-SOX sample
for adsorption of moisture to see if the material remained in its as- collect-
ed state long enough to permit routine laboratory tests on a sample. Any
interaction of the E-SOX sample with moisture from the flue gas during its
collection cannot be easily controlled in the field and was not investigated
for this work. As a simple test for the adsorption of water, the specific
surface area was measured according to the theory of Brunauer, Emmett, and
Teller (1) (referred to here as the "BET area") for (a) a sample exposed to
room air (relative humidity of about 60%), and (b) a sample continuously
exposed to the flue gas conditions at which it was collected (67°C and 11%
humidity by volume). Water wetting the surface of the particles would
decrease the roughness of the surface and cause a corresponding decrease in
the specific surface area. The results are shown in Figure 1. Although the
changes are quite small, a slight decrease in the BET area of the sample
exposed to room air can be seen, indicating adsorption of moisture. This was
confirmed by noting that the sample gained weight during the 8 days of
exposure. However, the weight gain was only 0.03 g for a 100 g sample,
corresponding to about 1.7 x 10~3 moles of water. The sample maintained at
flue gas conditions showed a very slight increase in BET area, indicating a
possible small loss of adsorbed water. Due to the need for keeping its
temperature elevated, it was not possible to confirm this by measuring the
weight of the sample. Therefore, the data of Figure 1 indicate that, for
reasonably short time delays (a few days) between acquiring the sample and
transporting it to the laboratory, the change in BET area of the sample due to
exposure to atmospheric humidity will be small.
Chemical Corn-position
Results of chemical analysis of the E-SOX type sample were as follows:
Chemical
Weight
Compound
Percent
Li20
0.02
Na20
0.19
K20
0.87
MgO
1.0
CaO
32.4
Fe203
13.1
AI2O3
11.6
Si02
24.9
Ti02
0.5
P2O5
0.12
S03
13.5
LOI
11.1
where LOI is the loss-on-ignition when the sample was heated to 750°C. The
presence of the sorbent material in this sample is reflected in the relatively
3
-------
10.00
9.00 —
8.00
-------
large amount of calcium present, conventionally reported as CaO. Analysis of
the amount of carbon, hydrogen, and nitrogen (CHN) in this sample yielded the
following (also in weight percent):
C 4.35
H 0.89
N 0.02
Results of thermogravimetric analysis (TGA) , which measures the change
in weight of a sample as it is heated, are shown in Figure 2. The initial
decrease in the weight of the sample is due to driving off adsorbed moisture.
The weight loss which occurs between about 300 and 450CC has been identified
as decomposition of calcium hydroxide in the sample.
TGA is normally performed on samples in a nitrogen atmosphere in order
to eliminate any effects caused by chemical reactions between the atmospheric
gas and the surface of the sample. But since heating of the samples in our
experiments takes place in an atmosphere of air, it was necessary to see if
any differences were obtained when the weight loss occurred in air. The
result for the E-S0X sample in air are shown in Figure 3. Comparison of
Figures 2 and 3 shows that only small differences in weight loss were seen
when TGA was run in the two different atmospheres.
Particle Morphology and Size Distribution
Figure 4 shows a typical area of the E-S0X sample, at a magnification of
500X, using a scanning electron microscope (SEM). Several spherical, amor-
phous particles typical of coal fly ash are present. The clusters of many
small spheres sticking together, which can be seen in some areas of the
photograph, are also seen in SEM photos of fly ash which contain relatively
large (e.g., > 20% by weight) amounts of CaO. In addition, several clusters
of crystalline material, which are probably composed mostly of sorbent matter,
are evident. The spherical and crystalline particles appear to be dispersed
uniformly in the photograph.
The results of Bahco particle size classification of the E-SOX sample
are shown in Figure 5. The data in this figure have been corrected for the
density of the sample, 2.35 g/cm5. The straight line indicates the result of
linear, least-squares analysis of the data. The mass-median-diameter (mmd)
obtained from this analysis was 6.3 pm. However, the data appear to show a
slight bimodal behavior, with a transition point somewhere around 5 pm. Then
characterization of the sample by a single mode (6.3 pm) may not be rigorously
correct, but the departure from unimodal behavior is so small that the
computed mmd is probably sufficient for our purposes. The measured porosity
(ratio of the volume of voids within the sample to the total volume occupied
by the sample) was 77%, which is at the high end of the range of porosities
typically observed for fly ashes (40 to 80%).
5
-------
TEMPERATURE, °C 6868-67
Figure 2. Thermogravimetric analysis (TGA) of the E-SOX sample from 50 to
450°C in a nitrogen atmosphere.
6
-------
100.00
g
LL
SO.OD
80.00
I I I I
1 1 1 1
98.16%
1 1
—
93.13%
1 ^
E-SOX. FIRST RUN
-------
Reprcducsd Irom
best gvallabla copy.
6S8S-46
Figure 4. Scanning electron microscope (SEM) photograph of particles in
the E-SOX sample at a magnification of 500X.
8
-------
101
PARTiCLE SIZE, fim
Figure 5. Bahco particle size classification of the E-SOX sample.
Straight line is a linear, least-squares fit to the data.
9
-------
IEEE RESISTIVITY TESTS
The E-SOX sample was subjected to a resistivity test according to IEEE
Standard 548-1984 (designed for testing fly ash only) at 9.6% water vapor by.
volume. This test calls for measuring the resistivity of a sample, in
increments of about 30CC, beginning at 95°C and increasing to a temperature of
about 215°C, and then at temperatures near 315 and 425°C (the "ascending
temperature phase"). Next, the material is maintained at 460°C overnight in
dry air, followed by resistivity measurements at about 30°C increments as the
temperature decreases from 450 to 85°C (the "descending temperature phase").
The results of this test are shown in Figure 6. For most of the temperature
range, the resistivity measured in the ascending phase Is higher than that
measured in the descending phase for a given temperature. However, this
relationship was reversed for other samples. In the low temperature limit,
the resistivity measured in the descending mode is three orders of magnitude
less than that measured in the ascending mode. This discrepancy is Intolera-
ble since various sorbent injection processes will likely need to operate in
this low temperature region.
Often, it is necessary to perform multiple resistivity measurements on a
single sample at different atmospheric conditions. So, the next step in this
work was to examine the effect of "cycling" the E-SOX sample in the descending
temperature mode for measuring resistivity — i.e., using a single sample to
measure resistivity at increasing humidity levels. The results are shown in
Figure 7. In general, the data are similar to that typically seen with fly
ash and no sorbent addition: coincident data in the volume conduction region
(above about 250°C) and decreasing resistivity with increasing humidity levels
in the surface conduction region (below about 180°C). For a truly dry
atmosphere with ideal electrical conditions, the resistivity would be a
straight line, since only the volume conduction mode would be active.
However, the curvature in the dry data at low temperatures is typical of
"real" conditions and is attributed to the presence of a trace amount of
moisture (measured to be 0.21 volume percent) and leakage currents within the
electrical connections and through the ambient air at these very high resis-
tivities. The lack of coincidence between the dry and humid data in the
volume conduction region is not uncommon with fly ashes and is not considered
important here.
Between each test shown in Figure 7, a fresh sample was put in the
environmental chamber to compare with the results for the cycled samples.
Figures 8, 9, and 10 show the differences in resistivity measured for the
fresh sample (circles) and the samples which have been cycled for each
humidity level. For example, the sample labeled E-SOX 1 in Figure 10 was the
fresh sample used in the dry atmosphere test and remained in the environmental
chamber during the humidified atmosphere tests. Thus, it had been cycled a
total of four times at the end of the 15% moisture (by volume) test. In
general, the resistivity of the fresh sample was always higher than that of
the cycled samples. Sometimes this difference was as much as an order of
magnitude for a given temperature. Figure 11 shows the resistivity data for
all fresh samples at the indicated humidity levels. Comparison with Figure 7
shows that, except for the dry air data (which was obtained only on a fresh
sample), the resistivities of the fresh samples were slightly higher than
those of the cycled samples.
10
-------
10
14
T
O E-SGX, ASCENDING
O E-SGX, DESCENDING
9.6% H20
10
13
10
12
E
.c
c
£ 1011
>
55
in
cc
10™
109
108
X
2.8
84
183
2.6
112
233
2.4
144
291
2.2
182
359
2.0
227
441
1.8
283
641
1.6
352
666
1.4
441
826
TEMPERATURE
1Q00/K
°C
°F
6863*9
Figure 6. Resistivity of the E-SOX sasple obtained according to the IEEE
Standard 548-1984 for resistivity measurements of fly ash.
11
-------
84 112 144 182 227 283 352 441 °C
183 233 291 359 441 541 666 826 °J=
TEMPERATURE
Figure 7. Resistivity for a single E-SOX sample obtained at successively-
higher humidity levels.
12
-------
10
15
T
T
10
14
O E-SOX2
~ E-SOX 1
4.9% H20
10
13
1012
E
u
E
o
t 10"
10
10
10s
10'
10'
X
2.8
84
183
2.6
112
233
2.4
144
2S1
2.2 2.0
182 227
359 441
1.8
283
541
TEMPERATURE
1.6 1.4
352 441
666 826
1000/K
°c
°F
6868-16
Figure 8. Comparison of resistivity data for a fresh E-SOX sample
(E-SOX 2) and a sample tested once previously (E-SOX 1).
13
-------
1015
10
14
I I
o E-SOX 3
O E-SOX 2
O E-SOX 1
9.6% H20
1013
1012
E
U
i1011
>
H
W
05
K 1010
10s
10®
10?
2.S
84
183
2.6
112
233
2.4
144
291
2.2
182
359
2.0
227
441
1.8
2E3
541
1.6
352
666
1.4
441
826
TEMPERATURE
1000/K
°C
°F
6868-17
Figure 9. Comparison of resistivity data for a fresh E-SOX sample
(E-SOX 3) and samples tested once and twice previously
(E-SOX 1) and E-SOX 2, respectively).
14
-------
1G15
1014
13
10
£ 1012
o
£
o
t
> 1011
i-
u
w
UJ
cc
10
10
109
10s
107
o E-SDX 4
& E-SOX 3
O E-SOX 2
o E-SOX 1
15.0% H20
2.8 2.6 2.4 2.2 2.0 1.8 1.6 1.4
84 112 144 182 227 283 352 441
183 233 291 35S 441 541 666 826
TEMPERATURE
1000/K
°C
°F
(263-13
Figure 10. Comparison of resistivity data for a fresh E-SOX sample
(E-SOX 4) and samples tested once, twice, and three times
previously (E-SOX 1, E-SOX 2, and E-SOX 3, respectively).
15
-------
101E
10'4
10
13
12
10
| 10"
0
£
1 10«
10s
10*
107
10*
E-SOX 1 DRY A!H
E-SOX 2 4.S%H20
h2o
0
2.8
84
183
2.6
112
233
2.4
144
2S1
2.2
182
35S
2.0
227
441
1.8
283
541
1.6
352
666
1.4
441
826
1000/K
°C
°F
TEMPERATURE
6868-19
Figure 11. Resistivity of E-SOX sample for various humidity levels. Data
for each humidity level were taken using a fresh sample.
16
-------
ANNEALING EFFECTS
The effect of temperature on the surface area of the E-SOX sample is
shown in Figure 12. For this test, the sample was maintained in an atmosphere
of dry air (measured water concentration of 0.21% by volume) at a temperature
of 50eC. After 24 hours, a portion of the sample was removed for BET analy-
sis. The temperature was then raised in increments of 50CC (to a maximum of
450CC) and the procedure repeated. A steady increase in the surface area with
temperature up to 400 °C was observed. Then when the sample was heated to
450eC for 24 hours, the surface area dropped sharply to a value which was a
factor of 3 lower than the initial value.
Using the same procedure described in the preceding paragraph, the test
was repeated in a humidified atmosphere (10.4 volume percent water). Also
included in this test was a LIMB type sample (EPA Demo coal plus Tennessee
Luttrel limestone injection) for comparison. The results are shown in
Figure 13. Again, the E-SOX sample increased in surface area with increasing
temperature, but only up to 350°C instead of 400°C like its counterpart in dry
air. Then at temperatures of 400 and 450CC the surface area again dropped
sharply. The reason for the 50"C discrepancy in the temperature at which the
BET ares decreases between the dry and humid data is not clear, but this
temperature range corresponds to the area where the decomposition of Ca(0H)2
begins in the TGA results of Figure 8. Thus, this discrepancy is probably due
to interaction between the sample and atmospheric humidity. The surface area
of the LIMB type sample showed the same general behavior, but in a much less
dramatic fashion.
Additional investigations were made of the effect of annealing the E-SOX
sample at 450°C. Because of the large change in the BET area, it was hoped
that differences between the annealed and as-collected sample could be seen in
the SEM photographs. However, the annealed sample was indistinguishable from
the as-collected sample shown in Figure 10, even up to a magnification of
lOkX. Bahco particle size classification of the annealed sample showed that
the mmd increased from 6.3 /im for the as-collected sample to 7.7 pm for the
annealed sample. Pursuing this further, the BET area of the size fractions
collected in the Bahco machine were measured:
BAHCO
BET AREA
(m2/g)
MODE
ORIGINAL
ANNEALED
%
(fim)
SAMPLE
SAMPLE
CHANGE
1.6
7.0
4.7
33
2.5
6.1
2.9
52
4.7
5.6
2.0
64
8.3
5.5
1.8
67
11.6
5.2
1.4
73
19.0
5.4
1.2
78
24.1
6.2
1.2
81
26.9
6.1
0.9
85
remainder
9.4
1.7
82
where "original" refers to the as-collected sample. Although the surface area
changed across the entire range of sizes, the percent change increased as the
17
-------
17.00
IB.00
100 150
200 250 300 350 400 45C 500
TEMPERATURE, °C sttt-i
Figure 12. BET area of the E-SOX sample as a function of annealing
temperature in dry air.
18
-------
17.00
15.00
200 250 300 350
TEMPERATURE. °C
Figure 13. BET area as a function of annealing temperature in an
atmosphere containing 10.4% humidity by volume.
19
-------
particle size increased. The density also increases from a value of
2.35 g/cci3 for the as-collected material to 2.51 g/cm3 for an annealed sample,
while the porosity stayed relatively constant (77 and 74%, respectively).
Examination of the pore structure of the E-SOX sample before and after
annealing, using a mercury porosimeter, is also of interest. Since mercury
has a high surface tension, it does not readily wet most surfaces and must be
forced, by pressure, to enter a small pore. The relationship between the
applied pressure and the corresponding diameter of the pore which mercury will
intrude was first derived by Washburn (2) and can be written as
Pd - -47cosfi . (1)
where P is the applied pressure; d is the pore diameter; 7 is the surface
tension of mercury (taken to be 480 dynes/cm); and 6 is the contact angle
between mercury and the wall of the pore (taken to be 140c). Kote that
equation (1) is derived for a capillary process, so cylindrical geometry is
assumed. If P is measured in psia, then solving equation (1) for d gives
d = 213.4/P (micrometers) (2)
The results of the porosimeter measurements on the E-SOX sample are shown in
Figure 14. In general, the distribution of pores was shifted toward larger
diameters in the annealed sample compared to the distribution for an as-
collected sample.
The surface area of the sample, S, can be obtained from the mercury
intrusion data of Figure 14 by relating the total surface area to the pres-
sure-volume work required to force mercury into the sample:
dW = -P dV = 7cos0 dS (3)
where W is the work and V the volume of mercury. The total surface area is
obtained by integration of equation (3) over the range of pressures used in
the porosimeter:
S f FdV (4)
7cos0
The value of the integral can be obtained directly from the pressure-volume
data of Figure 14. The resulting surface areas of the as-collected and
annealed samples are shown in Figure 15. These areas are in favorable
agreement with the BET surface areas.
20
-------
PORE DIAMETER, fm es62-4s
Figure 14. Volume - vs - diameter data from mercury porosimetry measureissnts
of as-collected and annealed (450eC) E-SOX samples.
21
-------
Figure 15. Area-vs-diameter data from mercury porosimetry measurements of
as-collected and annealed (450°C) E-SOX samples.
22
-------
Based on the TGA and BET area results, the differences in the resistivi-
ty obtained with the ascending and descending temperature techniques must be
due to the decomposition of Ca(0H)2 and the accompanying change in the surface
area above 350°C. Therefore, the IEEE procedure for measuring resistivity
produces fundamental changes in the nature of the ash/sorbent sample.
One explanation for the difference in resistivities of fresh and cycled
samples may be that the process of decomposing the Ca(0H)2, is incomplete
after a single annealing of the sample at 450EC, but is completed during the
second annealing. To test this hypothesis, TGA was run twice for several
samples to see if any additional weight loss occurred during the second
heating to 450°C. Examples of the results of these second runs for the E-SOX
sample are shown in Figures 16 and 17. The small increase in weight at the
beginning of the second runs has been attributed by the operators of the
instrument to a slight buoyancy of the weighing pan in the TGA due to thermal
convection currents. This effect would be obscured by the weight loss due to
adsorbed moisture during the initial runs. For every sample tested, the
weight change for these second runs was less than 1% so the above hypothesis
does not seem very credible. However, it cannot be positively ruled out.
A second possibility is that the change in surface area is incomplete
after a single annealing. This was confirmed during chemical analysis and BET
measurements of the samples used for the data of Figure 7:
E-SOX 1
E-SOX 2
E-SOX 3
Chemical
(0,5,10,
(5,10,15
(10,15
E-SOX 4
Compound
15% H20)
% H20)
% H20)
(15% H20)
Liz0
0.02
0.02
0.02
0.02
Naz0
0.18
0.19
0.19
0.18
k2o
0.79
0.80
0.80
0.80
MgO
1.00
1.00
1.00
1.00
CaO
32.40
32.30
32.30
32.30
Fe203
13.40
13.40
13.20
13.20
11.90
11.90
11.90
11.80
Sx02
25.00
25.10
24.60
25.00
Ti02
0.58
0.50
0.58
0.58
P20s
0.12
0.13
0.13
0.13
so3
13.20
12.80
13.20
12.90
LOI
13.50
13.50
13.80
13.70
BET, mz/g
1.80
1.80
1.90
3.00
Although no significant difference in the chemical composition of the samples
could be seen, there was a 40% change in the BET area between the sample
annealed only once (E-SOX 4) and the cycled samples.
MODIFIED RESISTIVITY TESTS
The preceding results indicate that, in order to avoid the large changes
in BST area and the decomposition of Ca(0H)2 induced by annealing the E-SOX
sample at 450°C, it will be necessary to limit the upper temperature at which
resistivity is measured. From the data of Figures 8, 12, and 13, restricting
23
-------
100.00
I 90.00
U3
80.00
1 1 I 1 1 . 1 ! 1 1 1
i
100.03%
1
99.84%
E-SOX, SECOND RUN {NITROGEN)
WT: 6.7805 mg
RATE: 10.00 °C/min
—
FROM: 300 °C
TO: 450 °C
WT % CHANGE: 0.39
I I I I I
I I I
1 1
50.00 S0.00 130.00 170.00 210.00 250.00 2SO.OO 330.00 370.00 410.00 450.00
TEMPERATURE. °C sm-ss
Figure 16. TGA of E-SOX sample: second run in nitrogen.
24
-------
100.00
I-
x
g
S 90.00
80.00
1 1 1 I 1 1 1 1 1 1
I —
100.02%
r
99.75%
E-SOX, SECOND RUN (AIR)
WT: 8.4832 mg
RATE: 10.00 °C/m:n
FROM: 300 °C
TO: 450 °C
WT % CHANGE: 0.27
I I I I
I I I I
I I
60.00 90.00 130.00 170.00 210.00 250.00 290.00 330.00 370.00 410.00 460.00
TEMPERATURE, °C 6868-66
Figure 17. TGA of E-SOX saEple: second run in air.
25
-------
the maximum temperature to 250cC should avoid such changes in the nature of
the E-SOX sample. Thus, a series of resistivity measurements using a maximum
temperature of 250°C were made to investigate the reliability and reproduc-
ibility of such a procedure.
The resistivity of the E-SOX sample measured in dry air in the ascending
temperature mode is shown in Figure 18. The test was performed on the same
sample three times. Note that the resistivity of the fresh sample (circles)
was lower than that measured in the two subsequent tests, but the second and
third test data showed excellent reproducibility. The reason for this
discrepancy is probably due to the presence of adsorbed moisture which is
gradually driven from the surface of the sample as the temperature is in-
creased and would not be indicative of the resistivity of the sample in the
flue gas of a precipitator. Figure 19 shows the resistivity of the E-SOX
sample, again using the ascending temperature mode, as a function of atmo-
spheric humidity. The dry data were acquired first, followed in order by the
5, 10, and 15% water data. Because the first (dry) data set gives resistivity
values lower than subsequent tests, the dry and 5% water data are unusually
close together. However, the data for the humidified samples should be
representative of the sample. Heating is always done with the sample in dry
air, although water vapor may be added when the resistivity values are
measured at descending temperatures.
The tests described in the preceding paragraph were repeated for the
descending temperature mode, starting at 250°C. The results for an E-SOX
sample in dry air are shown in Figure 20. Except for some scatter at the
lowest temperatures, the reproducibility is excellent, with no discrepancies
seen between fresh and cycled samples. The data for humidified samples are
shown in Figure 21. Differences in resistivity between the dry and 5%
moisture data are much larger here than in the case for ascending temperature
data. Comparison of Figures 18 and 20 shows that the resistivity of the E-SOX
sample in dry air is greater when measured in the descending temperature mode
than that measured with ascending temperatures, due to adsorbed moisture
already being driven from the sample before the dry data were obtained in the
descending mode. Figures 19 and 21 show that, up to about 150CC, the resis-
tivities measured with ascending and descending temperatures agree well.
Above 150°C, however, the resistivity measured with descending temperatures
tends to be slightly higher than that measured with ascending temperatures.
Therefore, resistivity measured in the descending temperature mode (beginning
at 250°C) agrees well with the corresponding results from the ascending
temperature mode, but avoids the problem of low values associated with the
first test using the ascending temperature method. Resistivity measurements
made in the descending temperature mode (beginning at 250°C) are then to be
preferred for ash/sorbent mixtures because this method produces the most
reliable and repeatable measurements.
The same heating procedure is deemed necessary for samples of LIMB dust
as well as for samples of E-SOX dust. This conclusion was reached because the
LIMB samples also showed evidence by TGA that the thermal decomposition of
Ga(0H)2 occurred in them at temperatures above 250°C, even though the evidence
of changes in BET surface area was weak (Figure 13).
26
-------
10
17
I I I i I I
1G18
O FIRST TEST
~ SECOND TEST
O THIRD TEST
10«
f ,0«
I 1°13
Ui
cc
~
o
C
R>
o
o
O
o
O
o
O
o
10
12
1011
10
101 I I I I I I I I I I I
2.S 2.6 2.4 2.2 2.0 1.8 1.6 1.4 10CO/K
84 112 144 182 227 283 352 441 °C
183 233 291 35S 441 541 '688 828 °F
TEMPERATURE
ES6S-4S
Figure 18. Resistivity of E-S0X sample taken in the ascending temperature
mode in dry air to a maxima temperature of 250eC.
27
-------
10
17
I I I I I I
10
16
O DRY AIR
~ 5.0% H20
O 9.6% H20
A 14.7% HzO
10
IB
£ 10™
X
c
£
>
H
5 io1j
11]
c
10
12
10
11
10
10
I I I I 1 I 1 1 I I I
2.8
84
183
2.6
112
233
2.4
144
291
2.2
182
359
2.0
227
441
1.8
283
641
1.6
352
66S
1.4 100G/K
441 °C
828 °F
TEMPERATURE
B86S-44
Figure 19. Resistivity of E-SOX sample taken in the ascending tenperature
mode at various humidity levels to a maximum temperature of
250°C.
28
-------
1017 ^
ITT
I I I I I I
I I 1
10
16
Oo
O TEST 1
~ TEST 2
OTEST 3
p-\ _u o ^
i|0 LCD
10™
| 1014
,Qe
o
d>
2 101S
05
1G12
1011 fc-
10
10
2.8 2.6
64 112
183 233
2.4
144
2S1
2.2
182
359
2.0
227
441
1.8
2S3
541
1.6
332
68B
1.4 1000/K
441 °C
E2S °F
TEMPERATURE
5868-42
Fig-ore 20. Resistivity of E-SOX sample taken in the descending temperature
mode in dry air, beginning at 250°C.
29
-------
10
17
1016
1015
f 10™
8 1013
uj
cc
1C12
1011
10
10
5 I I
l 1 i 1 ! 1 1 1
I t 1 1 I
I =
-
O DRY AIR
-
=~ c
z O
o
~ 4.9% H20
O 9.9% H20
A 14.9% K20
—
D
o
-
r—
o
—
o
-
o
o
-
llllll 1 1 1 llll
^u—G-n-r ° _
-
= /
/
-
7/
/
-
it
-
' i i
1 1 1 1 1 i 1 1
I I I I I
I
2.8
84
183
2.6
112
233
2.4
144
291
2.2
182
35S
2.0
227
441
1.8
283
541
1.6
352
6SS
1.4 1G00/K
441 °C
826 °F
TEMPERATURE
5S6J-41
Figure 21. Resistivity of E-SOX sample taken in the descending temperature
mode at various humidity levels, beginning at 250eC.
30
-------
DEVELOPING A PREDICTION TECHNIQUE
Knowledge of the resistivity of the ash/sorbent mixture resulting from
use of the E-SOX or other sorbent injection process is important in predicting
the collection efficiency and, consequently, the economic and technical
feasibility of applying the process on a given precipitator. To facilitate
planning and to hold down costs, it would be helpful to be able to predict the
resistivity of an ash/sorbent mixture in a manner similar to the computer
model developed by Bickelhaupt for fly ash (3,4). However, applying the
Bickelhaupt computer model (Model II, reference 4) to the samples used in this
work yields predicted values of resistivity which are consistently lower than
the measured values. Figure 22 shows an example of the failure of the
computer model to predict the resistivity of the E-SOX sample. Thus, it is
obvious that either the Bickelhaupt model must be modified, or a new computer
model must be written for predicting the resistivity of ash/sorbent mixtures.
It must be understood that the expected value of a satisfactory model
for predicting resistivity will be in estimating trends or relative values in
laboratory data, not in predicting the values with absolute accuracy.
Moreover, it will be apparent that the model may be substantially less
successful in predicting the resistivity values that prevail in a flue gas
duct than in predicting the resistivity values that occur under laboratory
conditions. No laboratory procedure can be expected to duplicate all of the
dynamic physical and chemical processes involved in the injection of a sorbent
and the subsequent reaction of the sorbent with S02.
ELECTRICAL CONDUCTION MECHANISM
When unloading the test cells used for making the measurements shown in
Figures 7 through 10, it was noted that some of the E-SOX sample was sticking
to the negative electrode in each cell. Further, the amount of sample
sticking to the electrode increased with the number of times the sample had
been cycled. This stuck material was carefully removed from each cell and
combined to obtain enough material for chemical analysis. The results were as
follows:
Chemical
Compound
As-Collected
Sample
"Stuck" to Percent
Neg. Elect. Difference
Fe203
A1203
Si02
Ti02
p2o5
SO,
LOI
Li20
Na20
K20
MgO
CaO
0.02
0.19
0.87
1.0
32.4
13.1
11.6
24.9
0.50
0.12
13.5
11.1
0.01
0.26
0.87
1.0
34.4
13.0
11.7
25.0
0.52
0.02
12.5
10.6
50
37
0
0
6
1
1
0
4
83
7
5
31
-------
10
15
i—i—i—i—i—r
10™
j
10'
E 1012
u
X
E
£
O
£ 1011
>
t-
K
S 1010
10s
10B
m E-SOX, MEASURED
A E-SOX, MODEL II
It?
3.0
60
140
2.8
84
183
2.6
112
233
2.4
144
291
2.2
182
359
J ! L
2.0
227
441
1.8
283
541
1.6
352
66S
1000/K
Cc
°F
TEMPERATURE
6868-64
Figure 22. Comparison of measured resistivity values and those predicted
by the Bickelhaupt computer model (Model II, Reference 4) for
an E-SOX sample in an atmosphere containing 15% water vapor by
volume.
32
-------
The results of the analysis are in weight percent and were performed on the
samples after ignition at 750CC. Differences of less than 10% should be
viewed with skepticism since they are within experimental error. In addition,
the 50% difference for Li20 nay not be meaningful since it results from
calculations on very small numbers. The 37% increase in the amount of Na20
would appear to indicate a migration of sodium ions to the negative electrode.
Although this evidence is limited, it is consistent with the results of
Bickelhaupt (5,6), which show that sodium ions are the dominant charge
carriers in coal fly ash. Migration of phosphorous away from the negative
electrode has no ready explanation.
A more thorough search for correlations between the resistivity of an
ash/sorbent mixture and its chemical composition was made. The first step in
this search was to identify the charge carrying species through a "charge
transference" experiment: applying voltage across a sample for a long period
of time so that the mobile charge carriers can move through the sample away
from the electrode of like charge and toward the electrode of opposite charge.
To accomplish this, 0.5 cm thick ash/sorbent samples were loaded between
parallel plate electrodes, one of which had a 0.102 mm recess. This recess
allowed us to obtain a thin layer of the material adjacent to the electrode
for chemical analysis. Two samples of each species were used, one with
positive polarity on the recessed electrode and the other with negative
polarity. In this way, we could look for an ion which showed an increase in
concentration at one electrode and a corresponding decrease at the other
electrode, as compared to its concentration in the bulk. The samples were
then placed in an environmental chamber at a temperature of 60CG in an
atmosphere containing 11% water vapor by volume. Then an electric field of
2 kV/cm was continuously applied across the samples for a period of 1 month.
At the end of that time, the portions of the samples within the recesses were
removed for chemical analysis.
In order to increase the chances of getting useful information from such
a test, five LIMB samples were used in this transference experiment, in
addition to the single E-SOX sample. Unfortunately, the results of chemical
analysis of the samples from the recesses did not show any consistent pattern
of ion migration. The three LIMB samples produced by injecting limestone did
show a migration of the alkali metal ions (Li, Na, K), similar to what would
be expected for fly ash only. However, the E-SOX sample and the two LIMB
samples obtained by injecting hydrated lime did not show any evidence of ion
migration. This was disappointing because the evidence from the cycling
experiments discussed above indicated a possible sodium ion migration.
However, the careful analysis of the charge transference results did not
confirm this, even with the use of duplicate samples. Since the amount of
charge transferred in each test was on the order of 40 coulombs, it cannot be
argued that the migration of an ionic species was too small to be detected.
Instead, it must be assumed that the charge carriers were not included in the
compounds tested for in the usual chemical analysis.
Correlation of Resistivity with Ionic Species
Because the charge transference experiments did not immediately identify
the charge carriers in an ash/sorbent mixture, an exhaustive search was made
33
-------
for a correlation between the measured resistivity of the samples and all
reasonable combinations of the amount of alkali and alkaline earth ions in the
sample in order to uncover more subtle relationships. In general, no correla-
tions were found. Since the two LIMB samples with limestone injection showed
a migration of sodium in the charge transference experiment, it would be
expected that they would show a correlation with the atomic concentration of
sodium (or sodium plus lithium). The two data points available appeared to
show this to be the case. However, the small number of data points are
insufficient to draw such a conclusion with any confidence.
Ideas from Glass Research
Since fly ash is composed of predominantly amorphous material, some
researchers have found that many of the concepts developed to explain electri-
cal conduction in glass can also be applied successfully to fly ash. Scanning
electron microscope photographs of ash/sorbent mixtures also show a large
amount of amorphous material, so there is reason to believe that some of the
results from research on glasses may apply to these mixtures also.
The ionic conductivity of virtually all oxide glasses is due to the
motion of monovalent cations (7). For the majority of these glasses, the
conducting ion is sodium, although lithium and potassium ions can also
contribute to the conductivity. However, there are various "alkali-free"
silicate glasses whose conduction mechanism is uncertain. These glasses
contain small amounts of alkali ions, but these ions do not carry appreciable
current in the glass.
Since the ash/sorbent samples have such large amounts of calcium in them
and no consistent motion of alkali ions was detected in the transference
experiments, the results for alkali-free glasses with high calcium content are
of special interest here. Within this group of glasses, the amount of alkali
metal ions appears to have no influence on their electrical conductivity. For
example, the conductivity of calcium silicate glass was found to be insensi-
tive to sodium concentration up to 1.26 mole % Na20 (8). Two possibilities
for the charge carrying species of these glasses have been suggested in the
literature: (a) oxygen ions, and (b) hydrogen ions. In calcium aluminum
silicate glasses, oxygen ion diffusion is apparently too low to account for
their conductivity (9), although there is some dispute on this Issue (10).
Other studies have shGwn that the conductivity of lead silicate (11), calcium
silicate (12,13), and barium aluminum borate (14) glasses was affected by the
presence of OH groups, indicating that hydrogen ions are the charge carriers
in these glasses. It is known that the mobility of hydrogen Ions In most
silicate glasses Is several orders of magnitude lower than the mobility of
sodium ions (15). But the presence of alumina groups tends to make the
hydrogen ions less tightly bound In silicate glasses (15), so alumlnosillcate
glasses may well have relatively high hydrogen ion mobilities. These studies
suggest that hydrogen ions are the charge carriers in these glasses, although
electronic conduction cannot be completely ruled out. If these results can be
extended to ash/sorbent mixtures, then it may be that hydrogen ions are the
charge carriers in these mixtures as well. However, this work did not address
that issue.
34
-------
Resistivity of Calcium Compounds
Since the primary difference between an ash/sorbent mixture and the fly-
ash alone is the large amount of calcium in the mixture, an investigation of
the resistivity of calcium-based compounds expected to be found in the
mixtures was conducted to see if they were correlated with the resistivity of
the mixtures. The resistivities of calcium hydroxide, calcium sulfite,
calcium sulfate, and calcium carbonate are shown in Figure 23. The data were
taken in an atmosphere containing 15% water vapor by volume. Calcium sulfite
has the highest resistivity of the four compounds. The resistivity of calcium
sulfate is about an order of magnitude lower for most of the temperature
range. Magnitudes of the calcium hydroxide and carbonate curves are much
lower than those of the sulfite or sulfate. Note that the peak resistivity
values for the ash/sorbent mixtures ranged from IxlO12 to 1x1014 ohm-cm, so the
data of Figure 23 seem to bracket the data for the mixtures.
Figure 24 compares the resistivity of the E-SOX sample with that of the
two calcium compounds shown to be dominant in the sample. The shape of the
E-SOX data is very much like that of the calcium hydroxide, but its magnitude
is nearer that of the calcium sulfite. Similar data for four LIMB samples are
shown in Figure 25. The resistivities of the two samples with limestone
injection (101 and 11S) are comparable to the calcium hydroxide resistivity;
the resistivities of the two samples with hydrated lime injection show very
good agreement with the calcium sulfate resistivity. The data in these two
figures suggest that the E-SOX sample resistivity may be a combination of the
resistivity of calcium hydroxide and sulfite, while the LIMB samples with
hydrated lime injection have resistivities nearly identical to that of calcium
sulfite. The resistivities of the other two LIMB samples are not as clear.
They could be combinations of calcium carbonate, hydroxide, and sulfate.
The data in Figures 24 and 25 indicate the possibility that the resis-
tivity of an ash/sorbent mixture is primarily determined by the resistivity of
one or more calcium compounds in the mixture. This possibility was also
indirectly indicated by other pieces of information from previous work:
measurements made under Cooperative Agreement CR810284-01 indicated that the
effective dielectric constant of high calcium content fly ash was independent
of temperature, in contrast to results for more typical fly ash; inability of
the Bickelhaupt computer model for fly ash resistivity to correctly predict
the resistivity of ash/sorbent mixtures; and field data under this project
which show the in-situ resistivity to be insensitive to the sorbent feed rate
under many conditions. Therefore, an attempt was made to correlate the
resistivity of an ash/sorbent sample with the resistivity of the sorbent
compounds contained in the mixture.
SIMPLIFIED MODEL OF ASH/SORBENT MIXTURES
Scanning electron microscope photographs of ash/sorbent mixtures show
that they consist of predominantly spherical particles and clusters of
spheres. Although the precise dispersion of fly ash and sorbent particles
throughout the mixture Is not known, assume, for simplicity, that the layer
consists of discrete particles of either fly ash or one of the sorbent
compounds. In addition, assume that the resistivities of all fly ash parti-
cles are the same and those of all sorbent particles of a given compound are
35
-------
101£
101S
10™
1013
E
X
£ 1°12
5
f—
CO
23
111
c 1011
10
10
o CALCIUM HYDROXIDE
O CALCIUM SULFITE
O CALCIUM SULFATE
& CALCIUM CARBONATE
(16% H20]
10s
108
3.0
60
140
2.8
84
183
2.6
112
233
2.4
144
291
2.2
182
259
2.0
227
441
1.8
283
541
TEMPERATURE
1.6 1C00/K
352 °C
668 °F
SSS8-S1
Figure 23. Resistivities of calcium compounds expected to be present in
ash/sorbent mixtures.
36
-------
io1«
10
15
O E-SOX
O CALCIUM HYDROXIDE
O CALCIUM SULFITE
(1B%H20)
10
14
g 1013
C
E
x
o
£ 1012
h-
M
W
10
11
10
10
10s
10*
3.0 2.8 2.6 2.4 2.2 2.0 1.8
60 84 112 144 182 227 283
140 183 233 2S1 369 441 541
TEMPERATURE
1.6 1000/K
352 °C
66S °F
6168-90
Figure 24. Comparison <5f the resistivities of an E-SOX sample, calcium
hydroxide, and calcium sulfite.
37
-------
1016
i r
15
10
10^
1013
E
U
£
£
C
t 1012
>
H
M
W
HI
O LIME {119}
~ LIMB (18)
O LIMB (17)
A LIMB (101)
^7 CALCiUM HYDROXIDE
k CALCIUM SULFATE
(15%H20)
10
11
10™
10-
10*
3.0 2.E 2.6 2.4 2.2 2.0 1.8
60 84 112 144 182 227 283
140 183 233 291 359 441 541
TEMPERATURE
1.6 1000/K
352 °C
68E °F
6868-92
Figure 25. Comparison of the resistivities of 4 LIMB samples (17 and 18
with hydrated lime injection, 101 and 119 with limestone
injection), calcium hydroxide, and calcium sulfate.
38
-------
the same. Then a charge traversing the bulk of the sample through or along
these resistive particles can be represented by an electrical circuit composed
of resistors representing the fly ash and sorbent particles. Because the
resistivities of the calcium compounds are so high (Figure 23) and no correla-
tion between resistivity and chemical composition was found, the resistivity
of the fly ash is expected to be negligible in comparison to that of the
sorbent. Also, the results of multiple regression (discussed below) suggest
that only two compounds are important for a given ash/sorbent mixture. Then
by the usual method of adding resistors in series, we could write:
R - (niRx) + (n2R2) (5)
where R is the equivalent resistance of the entire circuit, and nx and n2 are
the number of calcium compounds of type 1 and type 2, respectively, in the
path of the charge.
Although the sample will be highly inhomogeneous on the local level, we
would expect that, on the average, a charge would encounter the same number of
resistors of type 1 and the same number of resistors of type 2 during any
complete traverse of the sample. Therefore, the relative number of each type
of resistor would be the important parameters in determining the contribution
of each species to R, and we could rewrite equation (5) as:
R' — (ajRx) + (a2R2) (6)
where R' — R/(n1+n2)
ai = ^/(ni+nj.)
a2 ~ n2/(nx+n2)
1 = ax + a2
Note that the resistance calculated from equation (6) will always be less than
that of the larger resistor and greater than that of the smaller resistor.
Moreover, if a.l and a2 are comparable and one resistance is much greater than
the other, then the overall resistance will be essentially equal to this
higher resistance. This is the type of relationship suggested by Figures 24
and 25.
Of course, the true nature of an ash/sorbent mixture will be much more
complicated than that assumed here. But the development of equation (6) can
be easily extended to the case where more than two calcium compounds contrib-
ute to the overall resistivity of the mixture. This simple picture is
intended only as a starting point for trying to understand the resistivity of
the ash/sorbent samples.
DETERMINING AMOUNTS OF CALCIUM COMPOUNDS IN MIXTURES
Determining the amounts of calcium sulfite, sulfate, hydroxide, and
carbonate within an ash/sorbent mixture using available methods proved to be a
39
-------
difficult task. A serious handicap in this endeavor is the difficulty of
physically separating the ash and sorbent fractions of a sample and thus
avoiding the influence of the ash constituents on the calcium constituents to
be determined in the sorbent fraction. The steps followed to achieve a
reasonably satisfactory solution to the analytical problem are described in
the following paragraphs.
The total calcium content of the sample was obtained by igniting the
sample in air at 750cC, dissolving the residue in a combination of mineral
acids, and then determining calcium by atomic absorption spectroscopy. The
total amount of calcium thus found was assumed to represent the contribution
of the sorbent alone. This was done in the absence of information of the
amount of calcium in the ash component of the mixture. In a literal sense,
the assumption that the ash made no contribution to the calcium content of the
sample cannot be right; however, in a practical sense, the assumption is
acceptable if the ash has a low calcium content or if the weight fraction of
sorbent in the sample is high.
The sulfite, sulfate, and hydroxide constituents were determined by
extracting a portion of the original, uxiignited sample with water and then
following a set of analytical steps for the several anions associated with
calcium. Sulfite was determined both by titration (with aqueous iodine) end
by ion chromatography. Sulfate was determined by treating a portion of the
sample extract with peroxide and then determining the total amount of sulfate
then present by ion chromatography; the original quantity of sulfate was then
calculated by subtracting from the total amount of sulfate the amount of
sulfite determined previously (which was converted to sulfate by peroxide).
Hydroxide was determined by titrating a portion of the extract with aqueous
acid.
Any part of the calcium not associated with sulfite, sulfate, and
hydroxide was assumed to be present as the carbonate. A possible error in
this assumption, of course, is that calcium in the original fly ash was likely
to be present as a silicate, rather than as the carbonate. Attributing all of
the sulfate found to the sorbent is also a possible source of error since part
of the sulfate may have occurred in the ash. In special circumstances, as
mentioned later in this report, the amount of carbonate present was determined
more directly by performing an elemental analysis of the sample and calculat-
ing the amount of carbonate that was equivalent to the amount of elemental
carbon found.
MULTIPLE CORRELATION RESULTS
In accordance with the simplified model represented by equation (6), a
search was made for correlations between the resistivity of the calcium
compounds of Figure 23 and the various ash/sorbent mixtures using step-wise
multiple regression. The regression coefficients (r2) for this purely
mathematical approach generally exceeded 0.97. For example, the resistivity
of the E-SOX mixture could be expressed as the sum of the resistivities of
calcium hydroxide and calcium sulfite according to the equation
P(E-SOX) - 1.80xl012 + 24.62p[Ca(0H)z] + O.Ol^CaSOa)
(7)
-------
where p is the resistivity (in ohm-cm) and rz — 0.98. The comparison between
the measured resistivity and that predicted by this equation is shown in
Figure 26. Note that the constant term in the equation is an artifact of the
regression process. Later modifications to the computer program which forced
the additive constant to zero in the regression calculations showed that the
correlation coefficient actually increased in general. For instance, the
correlation of equation (7) becomes
p(E-SOX) - 26.56p[Ca(OH)2] (8)
with an r2 of 0.99 when the additive term is set to zero.
It was unfortunate that the coefficients in equations (7) and (8) could
not be directly related to the amounts of calcium hydroxide and calcium
sulfite in the sample determined by the chemical analysis described above.
The problem is almost certainly due to inaccuracies in the analytical proce-
dure .
The correlation procedure gave somewhat better results when it was
applied to two samples (referred to here as 821 and 901) that were derived by
injection of a Ca(0H)2 slurry into a gas stream containing S02 but not fly
ash. The correlation equations for these two samples were
p(821) - 48.10p(CaC03) + 2.00p(CaSO*) (9)
and
p(901) = 11.35p(CaC03) + 0.51p(CaSO*) (10)
with correlation coefficients (r2) of 0.97 and 0.99, respectively. Adjustment
of .these equations to the form of equation (6) produced the following:
p' (821) - 0.96p' (CaC03) 0.04p'(CaSOJ (11)
p'(901) - 0.96p'(CaC03) + 0.04p'(CaS04) (12)
The coefficients 0.96 and 0.04 in equations (11) and (12) are to be interpret-
ed as the apparent weight fractions of CaC03 and CaSO*, respectively, in both
samples.
For comparison, the actual results of determinations of the four calciun
compounds in these two samples are as follows:
41
-------
101S
T
I
O E-SOX. MEASURED
~ E-SOX, PREDICTED
10% H20
10™
Q
Q
£
c
>
«
LU
cr
~
~
1013
10
12
10
11
I I I I I
3.0
60
140
2.8
84
1SS
2 6
112
233
2.4
144
291
2.2
182
359
2.0
227
441
1.8
283
541
1.6 1000/ K
352 °C
668 °F
TEMPERATURE
6868-94
Figure 26. Comparison of the measured resistivity of an E-SOX sample and
that predicted from a combination of the resistivities of
Ca(0H)2 and CaS03 in an atmosphere containing 10% water vapor
by volume.
42
-------
Weight percent in
Compound
Sample 821
Sample 901
Ca(0H)z
9.9
4.2
CaS03
CaS04
CaC03
47
35
5.4
5.8
30
48
The above concentrations of CaC03 were calculated by difference, as already
explained. Since such high concentrations of CaC03 in samples of this type
seemed anomalous, CaC03 was redetermined on the basis of elemental carbon;
essentially the same results were obtained — 30% for sample 821 and 42% for
sample 901. If the calcium compounds are reported on the basis of just CaC03
and CaSOi,, the equivalent concentrations in sample 821 are as follows:
Similarly, the values in sample 901 are %CaC03 = 89 and %CaS0* - 11. The
comparison of the interpretations of sample compositions derived from the
resistivity correlations and the experimental analyses gives the basis of an
evaluation of the predictive model. This comparison is more favorable than
that referred to earlier for samples of the E-SOX type but obviously is still
not very satisfactory. Improved analytical information, it may be hoped,
would provide a more satisfactory comparison in an absolute sense.
The fact that the method for determining the relative amounts of the
four calcium compounds was relatively unsuccessful for those samples contain-
ing both ash and sorbent is not surprising. Interactions of the fly ash and
sorbent materials would be expected to form a very complex mixture that would
be difficult to analyze. In addition, the unavailability of the chemical
composition of the fly ash made estimation of the amounts of calcium compounds
within the mixtures very uncertain. However, the overall results of this work
indicate that the approach of correlating the resistivity of these mixtures
with the resistivities of the four calcium compounds will be successful,
providing a more adequate analytical method for estimating the amount of these
compounds within the mixture can be obtained.
Results of multiple correlation analysis of the resistivities of the
three general classes of mixtures studied and the four sorbent compounds of
interest can be summarized as follows:
1) The resistivity of samples with sorbent injection in the
downstream flue gas correlates directly with the resistivity
of Ca(0H)2, as indicated by equation (8).
2) The resistivities of the LIMB samples are a linear combination
of the resistivities of CaC03 (largest coefficient) and CaS04.
The data underlying this observation are not included in this
report.
%CaC03 - 100(30/[30 + 5.4]) « 85
%CaS0A = 100(5.4/[30 + 5.4]) = 15
43
-------
3) The resistivities of the samples derived from sorbent only are
linear combinations of the resistivities of CaC03 and CaS04,
as shown by equations (9) and (10). The fact that the larger
of the two compositional coefficients is associated with CaC03
Is anomalous, in that in samples of the type CaC03 is not
usually an important constituent. The conclusions given here,
therefore, may not apply to samples of more usual compositions
that result from low temperature reactions of Ca(0H)2 slurry
with S0Z, which consist mainly of CaS03 and Ca(0H)2.
Although the results summarized above are true in general, It should be
kept In mind that slight variations can be encountered with samples from
individual sorbent injection processes. The relative importance of each of
the four calcium compounds to the overall resistivity of the mixture will
depend on the chemical interaction history of the sample.
CONCLUSIONS
Application of IEEE Standard 548-1984 (intended for measuring the
resistivity of fly ash only) is not appropriate for use with fly ash/sorbent
mixtures. This Is due to decomposition of calcium hydroxide and large changes
in the surface area of these samples which occur when they are heated above
3O0CC. In order to avoid the large changes in the sample caused by heating,
the IEEE procedure should be modified to limit the upper temperature at which
resistivity is measured to 250°C for these mixtures. And to obtain best
repeatability of the resistivity measurements, the test should be conducted in
the descending temperature mode. Measurements using this modified procedure
produce consistent and repeatable values of resistivity for ash/sorbent
mixtures.
Results of multiple regression analysis show that the resistivity of an
ash/sorbent mixture is primarily determined by the resistivities of one or
more calcium compounds within the sample. Indeed, the correlation coeffi-
cients (r2) obtained for all samples generally exceeded 0.97. For a sample
consisting of only sorbent exposed to SOx, the agreement between the amounts
of calcium compounds found within these samples by analysis and the amounts
deduced from the resistivity coefficients is plausible if not quantitatively
satisfying. For a sample containing both ash and sorbent, however, the
agreement is unacceptable, presumably because of inadequate procedures for
estimating the amounts of the calcium compounds in an ash/sorbent mixture.
If the obstacles to relating sample resistivity to both sample composi-
tion and resistivities of the individual calcium compounds could be overcome,
the prediction of the resistivity of an ash/sorbent sample would likely
proceed as follows: 1) the sample would be identifled as to type based on
origin (LIMB or E-SOX process, for example), 2) it would be analyzed for the
different calcium compounds contained, and 3) the normalized concentration
factors, as illustrated earlier in this report, would be used, along with
resistivity values of the appropriate calcium compounds, in an equation that
predicts resistivity for the given type of sample.
Efforts to improve the resistivity model are now continuing as part of
the contract work for the Electric Power Research Institute under RP 1833-17.
44
-------
REFERENCES
1. Brunauer, S., Ernmett, P.H., and Teller, E. Adsorption of Gases in
Multimolecular Layers. J. Amer. Chem. Soc., 60:309-19, 1938.
2. Washburn, E. The Dynamics of Capillary Flow. Fhys. Rev., 17: 273-83,
1921.
3. Bickelhaupt, R. A Technique for Predicting Fly Ash Resistivity.
EPA-600/7-79-204 (NTIS PB80-102379), U. S. Environmental Protection
Agency, Research Triangle Park, NC. 1979.
4. Bickelhaupt, R. Fly Ash Resistivity Improvement with Emphasis on
Sulfur Trioxide. EPA-600/7-86-010 (NTIS PB86-178126), U. S.
Environmental Protection Agency, Research Triangle Park, NC. 1986.
5. Bickelhaupt, R. Electrical Volume Conduction in Fly Ash. J. Air
Pollut. Control. Assoc., 24:251-5, 1974.
6. Bickelhaupt, R. Surface Resistivity and the Chemical Composition of Fly
Ash. J. Air Pollut. Control. Assoc., 25:148-52, 1975.
7. Doremus, R. H. Glass Science. John Wiley & Sons, New York, 1973.
8. Schwartz, M. and Mackenzie, J. D. Ionic Conductivity in Calcium
Silicate Glasses. J. Am. Ceram. Soc., 49:582-5, 1966.
9. Hagel, W. C. and Mackenzie, J. D. Electrical Conduction and Oxygen
Diffusion in Calcium Alxominoborate and Aluminosilicate Glasses. Fhys.
Chem. Glasses, 5:113-9, 1964.
10. Owen, A. E. Properties of Glasses in the System Ca0-B203-Al203. Part 1.
The D. C. Conductivity and Structure of Calcium Boroaluminate Glasses.
Phys. Chem. Glasses, 2:87-98, 1961.
11. Hughes, K., Isard, J. 0., and Milnes, G. C. Measurement of Ionic
Transport in Glass. Part 2. Soda-Lead-Silica Glass. Phys. Chem.
Glasses, 9:43-6, 1968.
12. Schwartz, M. Mass Transport in Calcium Silicate Glass. Rensselaer
Polytechnic Institute, Troy, New York, 1969.
13. Terai, R. and Okawa, E. Calcium Self Diffusion end Electrical Conduc-
tivity of Calcium Aluminosilicate Glasses. Bull. Govt. Ind. Res. Inst.
Osaka, 26:136-40, 1975.
14. Gough, E., Isard, J. 0., and Topping, J. A. Electrical Properties of
Alkali-Free Borate Glasses. Phys. Chem. Glasses, 10:89-100, 1969.
15. Doremus, R. H. Ion Exchange in Glasses. In: Ion Exchange, Vol. 2,
Marinsky, J. A., Ed., M. Dekker, New York, 1969, p. 1.
45
-------
APPENDIX A.
QUALITY CONTROL EVALUATION REPORT
Quality Assurance Project Plan
The work described in this topical report was supported through Coopera-
tive Agreement CR 812282. A Quality Assurance Project Plan (QAPP) was not
originally required for this project. However, the measurement efforts for
this work were performed in accordance with the QAPP for CR 811683, Task 1.
This task concerned the development of a method for predicting the resistivity
of an ash/sorbent mixture resulting from a dry, post furnace flue gas desul-
furization process. The QAPP was submitted on June 1, 1985 end approved by
EPA in January of 19S6. All tasks of CR 811683 were designated as EPA QA
Category III. Since CR 812282 involves measurement and sampling efforts with
procedures identical to those addressed in the CR 811683 QAPP, it has been
assumed that the same QA classification and QA procedures are applicable.
Management and Technical Systems Audits
On July 15, 1985, EPA conducted a Management Systems Audit (MSA) and a
Technical Systems Audit (TSA) on project CR 811683. Based on the results of
these audits, the program was rated as "Acceptable, with Qualifications". The
major qualifications were that separate QAPPs be prepared for Tasks 3 and 4
and that Standard Operating Procedures (SOPs) be more thoroughly documented
for certain measurement systems. SRI responded by upgrading existing SOPs,
writing new SOPs, and by preparing and submitting separate QAPPs for Tasks 3
and 4. The resistivity, bulk chemical analysis, particle size distribution,
and scanning electron microscopy measurements are covered under the approved
QA plan for Task 1. Auxiliary measurements of specific surface area, particle
density, and thermogravimetrlc analysis are covered under the separate QA plan
for Task 4.
Evaluation of Measurement and Analytical Systems
All measurement and analytical systems were operated in accordance with
the procedures given in the QAPP unless stated otherwise in the following
paragraphs. The critical measurements which relate to this topical report are
discussed below.
Resistivity. The electrical resistivity of ash/sorbent mixtures is the
most important parameter addressed in this topical report. The different
nature of these mixtures compared with the properties of fly ash alone
required alterations of the standard laboratory measurement procedure (IEEE
Standard 548-1984, descending temperature technique). This procedure was
originally developed for fly ash from pulverized coal combustion. But by the
middle of 1986, it had been determined that this method caused changes In the
physical and chemical properties of ash/sorbent mixtures during the measure-
ment, resulting in erratic and nonreproducible data. By the middle of 1987,
It had been determined that consistent and reproducible data could be obtained
by limiting the highest test temperature to 250°C instead of the value of
455°C which is normally used for fly ash.
A-l
-------
Chemical Analysis. The mineral content of an ash sample is normally
determined after igniting it at 750°C. However, it was found that the
physical and chemical nature of ash/sorbent mixtures was altered by exposing
them to temperatures higher than 300CC. In particular, calcium hydroxide
contained in the sample decomposed above 300°C. Then in order to determine
the chemical composition and have it be representative of the sample, the
standard chemical analysis was performed on an unignited sample.
In addition to the standard chemical analysis normally run on fly ash,
auxiliary tests were performed to determine the amounts of calcium sulfate,
calcium sulfite, calcium hydroxide, and calcium carbonate in the mixture.
These four compounds occur in varying amounts in ash/sorbent mixtures and
largely determine the resistivity of the sample. Free and hydrated water was
determined from the weight loss of the sample at 110 and 300°C, respectively.
The carbon and hydrogen in the sample were determined with an elemental
analyzer in which a weighed quantity of sample is burned in an oxygen-rich
system and the products of combustion are fixed in an absorption train after
complete oxidation and purification for interfering substances (standard CHN
measurement). S03 was determined by both titration and ion chromatography.
The sulfate content was determined by measuring the soluble S0A (which
determines both S03 and S04 as S0A) and taking the value of S05 determined by
titration, converting it to its equivalent SOj, value, and subtracting it from
the soluble S04 value.
The amount of Ca(0H)2 in the mixture was determined directly from
titration. However, in order to estimate the remaining calcium compounds of
the sample and the fly ash to sorbent ratio, several assumptions were made:
(a) the sulfur and calcium concentration of the fly ash is insignificant
compared to the total sulfur and calcium present in the samples, (b) SC^""2,
S04"2, and OH"1 are all associated with calcium and any excess calcium is in
the form of CaC03, and (c) the sorbent material was relatively pure so that
the fly ash concentration can be approximated by a summation of those oxides
other than calcium, magnesium, and sulfur. These assumptions were necessary
because samples of the corresponding fly ashes were unavailable.
Sample Custody and Tracking
Sample custody and tracking requirements were met as specified in the
QAPP. All samples were assigned unique SRI notebook numbers referring to the
page and section in which they were described in the laboratory notebook.
Sample containers were marked with the corresponding numbers and other
appropriate identifying information.
Internal Audits. PC Checks. and Corrective Measures
An internal audit was performed by the SRI QA manager on CR 811683. The
audit revealed no serious deficiencies in the internal QC checks and data
reduction. However, recommendations were made to standardize the use of
laboratory notebooks for recording data during field measurement activities.
This replaced the use of loose-leaf binders and loose forms. In cases where
the forms were desirable, they were attached in the notebook. This facilitat-
ed access to the data for subsequent QC checks.
A-2
-------
Limitations of Use of Data
The most significant limitation on the use of these data is that they
should not be applied outside the range of conditions studied. The unique
electrical characteristics of ash/sorbent mixtures indicate resistivity data
require careful interpretation. In particular, the resistivity characteris-
tics obtained in this work should not be compared with that of samples of
significantly different origin or with data obtained when the ash/sorbent
sample was exposed to temperatures in excess of 250"C.
The Quality Assurance Review evaluation requires this Statement: "QA/QC
requirements apply to this project. Data are NOT supported by QA/QG
documentation."
It is not deemed feasible to incorporate the evaluation comments into
the report.
A-3
------- |