EPA-R2-72-126
FINAL REPORT
DEVELOPMENT OF METHODS FOR
SAMPLING AND ANALYSIS OF
PARTICIPATE AND GASEOUS FLUORIDES
FROM STATIONARY SOURCES
prepared for the
ENVIRONMENTAL PROTECTION AGENCY
DURHAM, NORTH CAROLINA 277O1
CONTRACT NO. 68-O2-OO99
* ,
ARTHUR D. LITTLE, INC.
CAMBRIDGE, MASS. O214O
NOVEMBER 1972
Arthur D Little, Inc.
-------
DEVELOPMENT OF METHODS FOR THE SAMPLING AND ANALYSIS OF PARTICULATE
AND GASEOUS FLUORIDES FROM STATIONARY SOURCES
(Final Report)
April 1972
E.T. Peters (Project Manager)
J.E. Oberholtzer
J.R. Valentine
Prepared under Contract No. 68-02-0099
for the
Environmental Protection Agency
by
Arthur D. Little, Inc.
Cambridge, Massachusetts 02140
Case 73757
Arthur D Little, Inc
-------
ACKNOWLEDGEMENT
This study was performed under the technical direction of Dr. Roy L.
Bennett, Project Officer, Division of Chemistry and Physics, Research
and Monitoring. The authors are grateful for the contributions of
J. T. Funkhouser, P. L. Levins and K. N. Werner and would like to
acknowledge the participation of six companies who provided us with
their current fluoride sampling and analysis procedures.
Arthur D Little, Inc
-------
TABLE OF CONTENTS
Page
LIST OF TABLES iv
LIST OF FIGURES vi
I. SUMMARY 1
II. INTRODUCTION 2
III. STATIONARY SOURCE INDUSTRIES A
A. GENERAL 4
B. INDUSTRIAL PROCESSES 5
1. Primary Aluminum 5
2. Iron and Steel 8
3. Glass 13
4. Phosphate Rock Processing 15
C. SUMMARY OF EMISSION SPECIES 21
IV. SAMPLE COLLECTION 23
A. INTRODUCTION 23
B. REVIEW OF SAMPLE COLLECTION PROCEDURES AND EQUIPMENT 25
1. Glass Probe Sampling Apparatus 25
2. Inert Probe Sampling Apparatus 25
C. PRESENT INDUSTRIAL PRACTICE . 27
1. Primary Aluminum 27
2. Steelmaking 28
3. Glass Manufacturing 29
4. Phosphate Rock Processing 30
5. Summary of Methods Employed by Industry 31
D. LABORATORY EVALUATION OF SAMPLE COLLECTION AND STORAGE MATERIALS 33
1. Evaluation of Sample Probe Materials 33
2. Sample Storage 38
E. CONCLUSIONS AND RECOMMENDATIONS 42
V. SAMPLE ANALYSIS 45
A. INTRODUCTION 45
B. GENERAL CONSIDERATIONS 45
1. Selection of an Analytical Method 45
2. Fundamental Steps in Fluoride Analysis Procedures 48
ii
Arthur D Little, Inc.
-------
TABLE OF CONTENTS (cont'd.)
Page
C. REVIEW OF EXISTING METHODOLOGY 51
1. Measurement Procedures 51
2. Solubilization Techniques 56
3. Removal of Interfering Species 59
4. Current Industrial Practice 63
5. Comparison and Summary of Candidate Techniques 65
D. LABORATORY EVALUATION OF CANDIDATE TECHNIQUES 68
1. Evaluation of the Fluoride Specific Ion Electrode 68
as a Measurement Technique
2. Comparison of the Fluoride Electrode with the Zirconium- 75
SPADNS Method
E. SUMMARY AND RECOMMENDATIONS 109
VI. LITERATURE REFERENCES 112
APPENDIX A A-l
iii
Arthur D Little, Inc.
-------
LIST OF TABLES
Page
1. Summary of Gaseous and Particulate Emission Species from 22
Stationary Source Industries
2. Initial and Final Levels of Fluoride in an Inert Gas Stream 35
Passing Through Heated Glass and Stainless Steel Tubes
3. Stability of Dilute Sodium Fluoride Solution in Glass 40
and Polyethylene Containers
3a. Effect of Aluminum (III) on Fluoride Electrode Measurement 70
in 0.5 Molar Citrate Buffer
4. Effect of Iron (III) on Fluoride Measurement in 0.5 Molar 72
Citrate Buffer
5. Effect of Other Potential Interferences on Fluoride Electrode 73
Measurements Using Citrate Buffer
6. Comparison of Fluoride Electrode and Zr-SPADNS Measurement Methods 74
7. Total Soluble Fluorides in EPA Samples by Direct Electrode Measurement 76
8. .Description of Flux Compositions 79
9. Fluoride Recoveries from Test Fusions of Cryolite 80
10. Water-Addition of Fused Solid Samples 87
11. Recoveries of Fluoride Using Small-Volume Direct Distillation 90
From Sulfuric Acid
12. Effect of Potential Interferences on Direct Distillation— 91
75 ml Distillate
13. Recovery of Fluoride From Fused Cryolite in Presence of Aluminum 93
Via 180°C ASTM Distillation
14. Comparison of Direct Distillation Recoveries at 180 and 210°C 94
15. Recoveries of Fluoride from Portions of Particulate Field 95
Samples Via Direct Distillation and Electrode Measurement
16. Distillations from Sulfuric Acid 97
iv
Arthur D Little, Inc
-------
LIST OF TABLES (cont'd.)
Page
17. Comparison of Sulfuric and Perchloric Acid Water Addition 98
Distillations
18. Summary of Samples and Sample Collection Data - Primary 101
Aluminum Plant
19. Characterization of Impinger Catch from Primary Aluminum 102
Plant Samples
20. Emission Spectrographic Analysis of Filter Catch from a 104
Primary Aluminum Plant
21. X-ray Analysis of Insoluble Particulate from Series B 105
Probe Washings
22. Fluorine and Solids Distribution in Series B Samples 106
Collected From a Primary Aluminum Plant
23. Fluoride and Solids Distribution in Glass Industry Samples 108
24. Standard Additions of Fluoride to Primary Aluminum Plant 110
Sample Solutions
v
Arthur D Little, Inc
-------
LIST OF FIGURES
Page
1. Schematic Sketch of Unit Processes and Per Annum Materials 6
Flow and Fluoride Emissions in the Primary Aluminum Industry
2. Schematic Sketch of Unit Processes and Per Annum Materials 9
Flow and Fluoride Emissions in an Integrated Steel Mill
3. Schematic Sketch of Unit Processes and Per Annum Materials 14
Flow in Glass Manufacturing
4. Schematic Sketch of Unit Processes and Per Annum Materials 17
Flow in the Phosphate Rock Processing Industry
5. P.H.S. Particulate Sampling SystenT ' 26
6. Schematic Sketch of Wet-Stream Fluoride Stack Sampling Apparatus 32
7. Schematic Sketch of Apparatus for Evaluation of Materials in 34
Dilute HF Streams
8. A Comparison of Initial (Tap Off) and Final (Sample) Fluoride 37
Concentrations after Passage Through a Five Foot 316 Stainless
Steel Tube
9. A Comparison of Initial (Tap Off) and Final (Sample) Fluoride 39
Concentrations after Passage Through a Five Foot 316 Borosilicate
Glass Tube
10. Schematic Sketch of Fluoride Sampling Apparatus 41
11. Stepwise Procedures for Handling Fluoride Samples 50
12. Apparatus for Performing Water Addition Distillation 81
VI
Arthur D Little, Inc
-------
I. SUMMARY
This study, conducted by Arthur D. Little, Inc., for the Environmental
Protection Agency, has resulted in the development of tentative sampling and
analysis of fluorides emitted from various processes within the primary
aluminum, iron and steel, phosphate rock processing and glass manufacturing
industries. In most cases, distinction is made between gaseous, insoluble
particulate and soluble particulate fluoride species.
Stack emissions from most of the processes that were considered can
be sampled in a manner that separates the particulate from the gaseous species.
For the tentatively recommended procedure, the particulate is collected by
means of a filter or electrostatic precipitator heated above the water
dewpoint with gaseous fluorides subsequently collected in a series of water
impingers maintained at room temperature. Some very wet process streams,
such as for the production of diammonium phosphate (DAP), include
considerable water entrainment, precluding a particulate collector. In
this case, the particulate is collected in the impingers, and soluble partic-
ulate fluoride cannot be distinguished from gaseous fluorides.
Chemical analysis procedures have been developed which emphasize
procedural simplicity but which also provide reliable and reproducible
results. The fluoride specific ion electrode has been found to be the most
suitable approach for measurement of fluoride concentration. Direct
measurements on impinger solutions after appropriate buffering appears
feasible in many cases. Otherwise, fusion and distillation procedures
are required to assure solubilization and separation from interfering
species.
The tentative source method for fluorides utilizing the Standardization
Advisory Committee (SAC) format is presented in Appendix A. A program
providing for field evaluation of these tentative sampling and analysis
methods for stationary source fluoride emissions is recommended.
Arthur D Little, Inc.
-------
II. INTRODUCTION
In response to growing public demands for clean air, a considerable
effort is being expended by the Government in the identification of air
pollution sources and the specification of appropriate abatement procedures.
Fluoride emissions from several types of industries have been recognized as
being harmful to ground vegetation, including "citrus" fruits and gladiolus,
as well as cattle and other live stock. The medical effects on human
population exposed to fluoride emissions are not well known. The major
stationary sources of particulate and gaseous fluoride emissions have been
associated with the industries that employ high fluoride-bearing raw
materials or use fluorides as a part of their manufacturing process. These
industries include steelmaking, primary aluminum, phosphate rock processing,
and the manufacture of glass and ceramics.
To arrive at realistic estimates of the fluoride mass emissions
burden to the atmosphere and in order to specify appropriate control and
abatement methods for fluoride emissions, it is necessary to develop
reliable sampling and analysis procedures for gaseous and particulate
fluorides which are not encumbered by interferences from other species in
the stack gases. The current program is directed toward developing and
understanding the kinds, of species which are present in each source emission
and, in light of these species, to establish tentative methods for repre-
sentative sampling of these streams and analysis of gaseous and particulate
fluoride components. A follow-on study is anticipated in which the recommended
procedures will be field tested under the varying conditions that exist
within the several unit process operations in each of the four stationary
source industries of concern. The experience gathered during field testing
would provide an opportunity for making modifications to the sampling and
analysis procedures, if deemed appropriate. The collected fluoride emissions
data would be available as a resource to EPA in setting design and operating
specifications for flue gas pollution control equipment.
Arthur D Little, Inc
-------
The following sections cover:
• A review of the various unit processes within the primary
aluminum, iron and steel, glass and phosphate rock processing
industries; industry flow diagrams which include estimates
of fluoride throughout; and an inventory of emission species
encountered within each process.
• Descriptions of sampling apparatus and procedures for gaseous
and particulate fluorides, including present industrial prac-
tice, evaluations of the reactivity of sampling train components
and sample container materials in the presence of fluoride ion,
and a description of recommended apparatus for each industry.
• Development of analytical methods for measuring fluoride ion,
including procedures for separation and measurement in the
presence of interfering species.
• Analysis of field samples collected from each industry to
identify the chemical species present and to evaluate the
suitablility of developed analytical methods for measuring
fluoride.
• Recommendations for tentative sampling and analysis methods
for gaseous and particulate fluoride.
Arthur D Little, Inc.
-------
III. STATIONARY SOURCE INDUSTRIES
A. General
This study is restricted to consideration of four industries which,
because of the raw materials utilized in the chemistry of process relations,
are known to emit gaseous and particulate fluorides. These industries are
primarily aluminum, iron and steel, phosphate rock processing and glass
manufacturing. Some typical fluoride emission control is typically prac-
ticed for all unit operations in these industries. The glass industry
limits control to collection of particulate dispersoid and fume by haghouse
techniques while the other industries rely heavily on wet scrubbers which
can be very efficient for the collection of water-soluble gaseous fluorides,
including HF, F and SiF .
A recently completed comprehensive study of the engineering and cost
effectiveness of fluorides emission control for industrial sources was
carried out for the EPA by Resources Research, Inc., and TRW Systems
Group. The study tasks of their program included an inventory of
fluoride emitting processes, process modeling, assessment of the state-of-
the-art of measurement and control technology, determination of control
costs, projection of trends to the year 2000, and recommendations for.
research and development programs required for minimizing soluble fluoride
emissions in a cost effective manner. Descriptions of the various
stationary source industries, including production trends, unit process
operations, emissions control methods, and fluoride emission inventories,
were presented in detail. Since this extensive background is available,
this report will be limited to reviewing the processing operations that
contribute most heavily to fluoride emissions.
To supplement data presented in the RRI/TRW report, we have prepared
flow diagrams showing materials flow, unit processes and the fate of
fluorides (expressed as fluorine) for the major emission sources. These
diagrams are presented in Figures 1 and 4. Fluorine emission estimates
are based upon the RRI/TRW results as well as data from the Bureau of
Mines and our own files; the estimates for fluoride emissions from tha
Arthur D Little, Inc.
-------
scrubbers of the various phosphate rock processing industries are those
of RRI/TRW. Within the glass industry, the only major source of fluoride
emissions is in some of the special pressed"ware products including opal
glass, leaded glass and many of the technical-grade products.
B. Industrial Processes
1. Primary Aluminum. A schematic sketch of the unit processes and
materials flow of a primary aluminum facility is presented in Figure 1.
Aluminum metal is produced exclusively by the electrolytic refining of
purified alumina ore that is obtained from the mineral bauxite, a hydrated
aluminum oxide having the oxides of iron, silicon and titanium as impurities.
Electrolysis is carried out in a carbon black-lined reduction pot which is
fitted with carbon electrodes. The bath is a mixture of fused cryolite
(Na.AlF-), fluorspar (CaF0), and sodium fluoride (NaF). The primary
JO Z
electrolysis reaction is:
2 A1203 + 3 C + Al + 3 C02
The stoichiometric ratio yields about 20,000 cubic feet of C0_ per ton of
aluminum. Actual gas production is somewhat higher since appreciable
quantities of carbon monoxide are also present. The fluoride salts do not
take part directly in the reaction, but fluorine compounds, including
cryolite and chiolite (Na^Al^F ,), are carried off in the exhaust gases
along with the evolution of substantial quantities of C0«.
The top of the bath is covered with a frozen crust of alumina which
dissolves into the cryolite bath as aluminum production proceeds. The
carbon electrodes are renewed as they are consumed. There are two types of
electrodes in use. One is the prebaked electrode which is periodically lowered
into the bath until the stub end is reached. Another is the so-called
Soderburg electrode which is charged to the operation as a paste and is
baked into a hard electrode by the process heat as it is lowered into the
.pot. Makeup cryolite and fluorspar are also apread on top of the crust as
required along with the alumina charge. The molten aluminum sinks to the
bottom of the bath and is withdrawn through a tap-hole.
Arthur D Little, Inc
-------
c
ALUMINA
(7.2)
FLUORSPAR,
CRYOLITE
[21.1]
^
- -
h
1
L
(0.8)
[3.1]
(1.9)
[7.7]
VSS
HSS
L
(4.5)^
[10.3]
PreBake
[7.3] L^J
4 1
[10.4]
i i
FLUORINE
RECOVERY
|
[0.9]
[5.4]
[6 n
[13.8]
* —
[44.2]
[54.5]
&-
I I
FLUORINE
RECOVERY
T
[2.3]
[7.2]
A
"i T
FLUORINE
RECOVERY
i
[3.1]
(0.42)
(1.00)
(2.38)
ALUMINUM
(3.80)
(Millions of Tons)
[Thousands of Tons, Expressed as Fluorine)
| Stack and Vent Emissions
Liquid Sludge and Slag Losses
FIGURE 1 SCHEMATIC SKETCH OF UNIT PROCESSES AND PER ANNUM MATERIALS FLOW
AND FLUORIDE EMISSIONS IN THE PRIMARY ALUMINUM INDUSTRY
-------
An aluminum reduction potline is a very large-scale operation.
The individual pots are about 16 feet long, 8 feet wide and 6 feet deep.
They are arranged side by side in the pot building which is of open
high bay construction. It must be very well ventilated to dissipate
the substantial heat of reaction; ventilation is achieved by a roof
monitor that runs the length of the building.
In recent years, with increasing emphasis on control of fluorine
emissions, the pots are operated with hoods designed to collect and
control fluoride emissions. Because of the required pot mobilityj
however, it is extremely difficult to fabricate hoods which fit tightly
and operate satisfactorily.
There are several sources of serious fluoride contamination. Fluorides
are continually carried off by entrainment in the substantial quantities
by the CO evolved from the refining reaction. A second major source
of contamination occurs during "crust breaking," an operation conducted
from time to time to break up the frozen layer of alumina. In addition,
the pots must be periodically stirred to maintain uniform concentrations
in the bath and to keep the alumina ore in contact with the top of the
bath. During such an operation, it is not possible to efficiently
collect the evolved gases in the hood system, and the gases escape to
the atmosphere of the building. A similar problem exists whenever an
electrode must be changed and is even more serious than crust breaking
because more of the hood must be removed to get at the electrode.
A major portion of fluorine emissions can be recovered by passing
the gases collected in the hoods to a wet scrubbing system. Recovery
of fluorine from the building atmosphere is a much more difficult problem.
The quantity of gas to be scrubbed is of course very much larger than
that collected by the hoods, and a number of roof scrubbers must be
installed. Recently developed high efficiency control methods based on
adsorption of fluorides onto the incoming alumina feed, such as the
(2)
Alcoa 398 Process which incorporates a fluid bed reactor, may be
appropriate for cleaning roof monitor gases as well as hood gases.
Arthur D Little, Inc
-------
Fluoride emissions can be as high as ten pounds per ton of aluminum
produced. Particulate emissions include Al 0_, carbon soot, Na_AlF,,
A1F« and Na^Al^F..,. The major gaseous species are C0_, CO, organic
species (from the binders in the anode paste) and HF, which results from
the hydrolysis of A1F and, to a lesser extend, Na_AlF,. In addition,
NaAlF, is a volatile specie, but it readily decomposes to Na,-Al~F.., and
A1F3.
2. Iron and Steel A typical steel mill emits 0.2 to 1 pound
of fluorine (as fluorides) to the atmosphere per ton of steel produced.
About 5 to 10% of the emitted fluoride is solid particulate and the
remainder is gaseous.
The principal source of fluoride is fluorspar, CaF^, which is employed
in quantities of 2 to 15 pounds per ton of steel produced. The fluorspar
is added as a slag conditioner to improve the collection of oxide
impurities; the resulting increase in slag fluidity promotes desulfuriza-
tion and increases the rate of heat transfer through the slag, improving
process efficiency. Western ores, particularly those from southern
Utah, include another source of fluoride. These ores contain 2,000 to
4,000 ppm fluorine present as fluorapatite, fluorspar and fluorosilicates.
The remaining quantities of input fluorine are introduced as impurities
in the other raw materials.
An integrated steel mill is comprised of several combined facilities,
including a sintering plant, blast furnaces, coke and by-products plant
and several steel making shops which utilize open hearth, electric and
basic oxygen (EOF) furnaces. Support services are also required, including
boiler plants, furnace and annealing rooms, scarfing areas, water treat-
ment, etc. Each of these facilities contribute to pollution emissions to
some extent. The major fluoride emissions result from iron ore pelletizing
and sintering, and open hearth and EOF steelmaking processes. A flow
diagram of an integrated steel mill operation showing raw material require-
ments and evolved fluorides is presented in Figure 2.
Arthur D Little, Inc
-------
[4.0]
r
—*
D
n>
n
[0.1]
[Thousands of Tons]
(Millions of Tons)
t
Points of Fluoride
Emissions
FIGURE 2 SCHEMATIC SKETCH OF UNIT PROCESSES AND PER ANNUM MATERIALS
FLOW AND FLUORIDE EMISSIONS IN AN INTEGRATED STEEL MILL
-------
a. Pelletizing and Sintering
Palletizing and sintering are methods for recovering iron ore "fines"
that would otherwise be lost as dust and converting them into a product
suitable for charging into the blast furnace. Pelletization generally
occurs at the mine whereas sinter shops are integral with a blast furnace.
Both processes are similar and involve burning a mixture of ore-bearing
fines and an appropriate fuel, such as coke breeze or blast furnace off-
gases. Emissions consist almost entirely of particulates extrained in
the combustion gases, which are large in volume. The sintering process
is effective in removing sulfur from the charge so that sulfur oxides
also occur in the off-gases. Other species present in small quantities
include hydrocarbons, nitrogen oxides and fluorides. The fluorides are
largely confined to operations utilizing Western ores, particularly
those from southern Utah, which contain 0.02 to 0.04% mineral fluoride.
In general, the fluoride emissions from palletizing and sintering can be
kept very low through addition of limestone to 'the sinter mix, which ties
up much of the emitted fluoride in a slag.
b. Blast Furnace
Ore is reduced to iron in the blast furnace. The charge consists
of iron ore, coke and limestone which are added at the top of the furnace.
To promote combustion, a high volume of pre-heated air is forced up
through the furnace. A series of deoxidation reactions occur as the
charge descends down the furnace until finally, near the bottom of the
furnace, a molten pool of reduced iron is formed covered by a slag con-
taining many of the impurities. The process is continuous, with pig
iron and slag being tapped off periodically.
Any fluorine that enters the blast furnace as a constituent of the
ore, fuel or limestone will probably react with the limestone and leave
the furnace with the slag. Gases from the blast furnace are collected,
passed through stoves for heat recovery, and when reasonably cool, are
passed through dust control equipment, which removes most of the adsorbed
10
Arthur D Little, Inc
-------
fluoride gas. The gas is then burned to recover remaining latent
heat. Any gaseous fluorides remaining will be emitted to the atmosphere.
c. Steelmaking
Steelmaking involves a refinement of pig iron to reduce carbon
and silicon content to the desired level and to remove or control other
impurities. This refinement is carried out by heating a charge of pig
iron, scrap and fluxing agents (limestone and fluorspar) in the presence
of air. Carbon, silicon and manganese are oxidized and are either evolved
in the off-gases or are collected in the slag. Three basic Steelmaking
processes are used—open hearth, basic oxygen (EOF) and electric, the
latter of which is charged completely with steel scrap.
Open hearth operations are declining rapidly due to increased
capacity for basic oxygen and electric steel. The EOF method has be-
come increasingly important during the last decade as low cost supplies
of oxygen have become available through improvements in cryogenic
technology; steel refinement can be accomplished in 20 to 40 minutes
compared with 8 to 12 hours required for the open hearth process. In
addition, modern procedures for improving the quality and form of scrap
are resulting in an increased production of electric furnace steel.
In U.S. practice, fluorspar consumption in open hearth Steelmaking
may range from zero up to 10 pounds of fluorspar per ton of steel, with
an overall average consumption of about four pounds of spar per ton of
steel. More fluorspar is used in making high-carbon heats, because the
bath temperature and iron oxide content of the slag are generally lower
for these heats. Furnaces operating with oxygen lancing generally con-
sume less fluorspar, due to the higher bath temperature and stronger
boil which greatly accelerate the shaping of the slag.
The addition of fluorspar causes the slag to become more fluid,
thinning the "heavy slag" and facilitating the solution of lumps or
"floaters." The effect of fluorspar on the slag is only temporary,
lasting roughly one hour. About one percent of the calcium fluoride
11
Arthur D Little, Inc
-------
reacts with the silica according to the following reaction:
2 CaF2 + Si02 -> SiF^ + 2 CaO
This reaction increases the lime content and decreases the silica
content, thereby increasing the effective activity of the lime to remove
sulfur. While this reaction occurs only to a small extent, it does
create an emission problem due to the fact that the SiF, volatilizes.
There is also the possibility of direct reaction of fluorine on the
sulf ide to form SF^ . This would eliminate some of the sulfur in the
o
slag, throwing the equilibrium between sulfur in the slag and in the metal
out of balance. Further elimination of sulfur from the metal would
then occur to maintain equilibrium. The presence of moisture in the
combustion gases will tend to hydrolyze any SiF, to HF at the slag
surface. In addition, there is an opportunity for direct pyrohydrolysis
of any CaF^ floating on the surface of the bath or carried into the
gas stream as dust. There is also some evidence that magnesium forms
stable slag constituents that reduce the tendency of fluoride to leave
the slag system. Not only is the volatilization of SiF, suppressed
but the pyrohydrolysis of magnesium fluoride (MgF_) itself is insigni-
ficant up to 1300°C. Calcium does not appear to exert as strong an
influence in reducing the tendency of fluorine to volatilize.
In comparison to the open hearth process, basic oxygen steelmaking
is quite recent in development, accounting for less than 1 percent of
the steel produced in 1960. The majority of new steelmaking installa-
tions, however, have been EOF, and basic oxygen steelmaking surpassed
the open hearth in annual tonnage in 1970. The EOF smelting process is
carried out in refractory lined cylindrical vessels that can accommodate
from 50 to 250 tons of hot metal and scrap per charge. To speed up
refinement, a stream of oxygen is impinged onto the liquid metal surface
which results in violent agitation and intimate mixing of the oxygen
with the molten iron. EOF steelmaking employs 12 to 15 pounds of CaF
per ton of steel, four to five times the amount used in open hearth.
12
Arthur D Little, Inc
-------
The majority of the fluoride is tied up in the slag; emissions to the
atmosphere are only a few percent of the fluoride input. It is postu-
lated, however, that a considerable portion of the HF generated by
slag reactions is adsorbed on the fine Fe20_ dust that is blown from
the furnace.
Finally, electric furnace steelmaking utilizes anywhere from 3
to 15 pounds of CaF_ per ton of steel, with an average of 8 pounds.
Since there is only a moderate gas input to the electric furnace, hoods
and vents can be effectively employed for collecting and controlling
particulate emissions.
3. Glass Glass processing produces emissions to the atmosphere
through physical entrainment in the air stream flowing through the glass-
making furnace and by volatilization from the molten glass. The major
effluent species are fluorides, silicates, sulfates and borates.
Fluoride emissions are generally restricted to the production of
specialty glasses, including opals, technical grade products and leaded
glasses. Major fluoride species include gaseous SiF, , HF and F« as well
as PbF«, NaF and CaF. particulate. A schematic sketch of the glassmaking
processes as well as a materials flow chart and fluoride emissions is
presented in Figure 3.
Glass manufacturing involves a high temperature conversion of raw
materials into a homogeneous melt. Principal raw materials include sand,
feldspar, soda ash and limestone; in addition, many other elements are
utilized in small amounts in the manufacture of specific glass composi-
tions.
With few exceptions, the major amounts of glass produced in the U.S.
are melted in continuous tanks with capacities up to 300 tons per day.
Furnaces are fired by natural gas or oil fuels and, to a lesser extent,
by electric melting. Glass melting furnaces and tanks are usually pro-
vided with regenerators to recover heat from the flue gas before it
passes out the stack. Reactions between the component raw materials and
13
Arthur D Little, Inc
-------
SAND (3.0)
m nnocnAD rin~l
rLUUKbrAK |_ IUJ
[r\TuirD 1 o r\ ^
U 1 HtK \ c. .1) )
k
w
k
p
h
STORAGE
D T MC
blNo
\/
^
v
MIXER
^
4 —
L*
^
p
T
mis?
LULLhLI 1UN
t r
[
L
1
HEAT
RcrnUFRY
FURNACE
2]
*
m
GLASS [8] PRODUCT
p P
FORMING ! STORAGE
i
(Millions of Tons)
[Thousands of Tons, Expressed as Fluorines]
t
Stack Emissions
i
D
FIGURE 3 SCHEMATIC SKETCH OF UNIT PROCESSES AND PER ANNUM MATERIALS FLOW IN GLASS MANUFACTURING
-------
the volatilization of the batch in the glass tank account for the emission
of sulfur dioxide, carbon dioxide, nitrous oxides, fluorides, etc., in
the stack emission. Fluoride compounds are often used as a source of
sodium, lithium and calcium. In addition, these compounds serve as
fluxes to enhance melting and as fining agents to improve homogeneity and
remove bubbles. Crystalline precipitates of CaF« provides the opacity
characteristic of the opal glasses. This latter process is the major
source of fluoride emissions in the glass making industry.
4. Phosphate Rock Processing Some 25 million tons of phosphate
rock is processed annually in the United States. The rock is mostly
fluorapatite [3Ca (PO^-CaF^ and hydroxyapatite [3Ca3(PO, )2«Ca(OH)2];
the fluoride content is generally 3 to 4% on a weight basis. Approximately
one third .of the rock is processed into phosphoric acid (by the wet
process method) and a quarter is used for electric furnace production
of elemental phosphorous and phosphoric acid. The remaining 40% is
used mainly in the production of fertilizers, including triple super-
phosphate (TSP) , normal superphosphate (NDP), diammonium phosphate (DAP),
and dicalcium phosphate, and defluorinated rock which is employed in
feed preparations for cattle and chickens.
A schematic sketch of the various process operations and corresponding
levels of fluoride emissions is presented in Figure 4. Most of the
phosphoric acid produced by the wet process is subsequently used in
other processes, including the preparation of TSP and DAP. The major
fluoride emissions from all of these processing operations are HF and
SiF,; the latter readily hydrolyzes to H^SiF,. The various operations
are briefly summarized below.
a. Wet-Process Phosphoric Acid
In the wet process, reaction of phosphate rock with an acid produces
free phosphoric acid and the salt of the acid. In most commercial pro-
cesses, sulfuric acid is used; the by-product salt is calcium sulfate
which precipitates as gypsum and is removed by filtration. The reaction
15
Arthur D Little, Inc
-------
is described by the following equation:
3Cair.(PO.),F0 + 30H-SO, + SiCL + 58H.O -> 30CaSO.-2H00 + 18HJPO. + H.SiF,
1U 4 D / Z 4 2. 2. q 2. J 4 2 o
There are a number of impurities in the phosphate rock which consume acid
and affect the reaction. One of the principal impurities is fluoride.
Generally, it is present in either of two forms — as part of the fluora-
patite mineral and as additional amounts of free calcium fluoride.
When the rock is acidified, fluoride is converted to hydrofluoric
acid. In most rocks, there is enough silica present to react with all
of this hydrofluoric acid. In most rocks, there is enough silica
present to react with all of this hydrofluoric acid to form fluosilicic
acid. At the temperatures and acidity conditions under which the diges-
tion reaction must be carried out, some of the fluosilicic acid is vapo-
rized as silicon tetrafluoride gas with traces of gaseous hydrofluoric
acid.
In most plants, the substantial heat of reaction is removed by sparging
the digester with air or by flash cooling of the hot liquor. In either
case, there is substantial emission of air, steam, the fluorine containing
gases, and P-jO,- mist to the stack. The effluent gas is processed in a
wet scrubber in which the fluorine compounds and the P~0,- are absorbed
in water to make dilute solutions . In some plants fluorine containing
by-products are recovered from the scrubber liquor. In most cases, however,
the liquor is discharged as an effluent by way of the gypsum disposal
pond and acid water cooling ponds.
b. Elemental Phosphorus
In the production of elemental phosphorus , phosphate rock is
reacted in an electric furnace with silica and coke, according to the
formula 2Ca3(P04)2 + 6Si02 + IOC ->• 6CaSi03 + P^ + 10C02
The volatized phosphorus is carried along by the gas stream, which is
mostly CO and C0» and is condensed. Iron present in the phosphate rock is
reduced to the elemental form and alloys with the phosphorus to form a
ferrosphosphorus by-product. The furnace effluent also contains nitrogen,
hydrogen, methane, and fluorides. Particulate matter entrained in the
16
Arthur D Little, Inc
-------
D
1. MET PHOSPHORIC_ACITJ
'.1.
FILTER
[4.1]
[64]
•c
STORAGE
(3.9)
[11]/SCRUBBER J
DUST
COLLECTION
LIQUID P
CONDENSOR
SLAG
4
SCRUBBER
[3.8]
~\ [1]
D
[4]
REACTOR
5. OTHER (i.e., Defluorinated Rock)
(0.2)
L2J
PA
[8]
MIXER/
GRANULATOR
W
KIL?
']
I
[4]
/
SCRUBBER
roi
[2]
STORAGE
(0.25)
EXPORTS AND OTHER
(Millions of Tons)
[Thousands of Tons, Expressed as Fluoride]
T Stack and Vent Emissions
^ Slaq and Gypsum Ponds
»PA
1
1 REACTOR
(Den)
GRANULATOR
CURING
[70]
STORAGE
CURING
[87]
STORAGE
FIGURE 4 SCHEMATIC SKETCH OF UNIT PROCESSES AND PER ANNUM MATERIALS
FLOW IN THE PHOSPHATE ROCK PROCESSING INDUSTRY
-------
furnace gases is collected by a preclpitator and returned to the furnace.
The phosphate rock generally undergoes treatment prior to being fed
to the electric furnace. The most common method of treatment is nodulizing
the phosphate rock in a rotary kiln at a temperature of 1200° or more in
which the rock is heated to the point of incipient fusion and agglomerated.
During this nodulizing operation, organic matter, moisture, carbon
dioxide and part of the fluoride are removed.
c. Triple Superphosphate (TSP)
A principal end product of the phosphate fertilizer industry is
triple superphosphate, which is a solid product formed by acidulating
phosphate rock with phosphoric acid. For run-of-pile triple, 50%
P^O- acid and ground rock are mixed and the product, which is a
pasty mass, is conveyed to storage, where reaction or "curing" continues
along with drying. The product is subsequently removed, crushed, screened
and shipped as a sized solid. This simple process, however, is difficult
to control because of the process conditions and the nature of materials.
Therefore, careful attention to fume control is required. The principal
contaminants are gaseous fluorides and phosphate rock dust. All of the
equipment must be closed or hooded and vented to a stack through suitable
fume control equipment, which is commonly some type of one- or two-
stage wet scrubber.
A special problem exists in the storage building where substantial
quantities of gaseous fluorides are evolved during the curing of the
triple superphosphate. To control this fume, the ventilating air must
be collected and passed through a scrubber.
d. Normal Superphosphate (NSP)
Normal superphosphate is formed by the reaction of phosphate rock
with sufficient sulfuric acid to convert the phosphate ore to monocalcium
phosphate. The sulfuric acid is tied up in the process as calcium sulfate.
Since the P-O,. content of NSP is only about 20% compared to 45-50% in
TSP, demands for this product are falling off.
18
Arthur D Little, Inc
-------
There is a great variety of equipment in use for manufacture of normal
superphosphate. The older plants were batch-type units; some of the
newer plants operate in a continuous manner and are equivalent to the
system used for triple superphosphate. Equipment required includes a
heavy duty mixer to contact the finely ground phosphate rock with sul-
furic acid, and then a means for discharging the partially reacted mass
to a storage area where completion of the reaction and drying can occur.
The evolution of steam, CO-, and the air required for ventilation
results in substantial emissions of gaseous fluorides containing fume
and rock dust. The appreciation of the need to control these emissions
came much later than the establishment of the plants and the adoption
of control devices varies widely throughout the industry.
e. Ammoniated Phosphate
In addition to phosphate, a complete fertilizer must also contain,
among other elements, nitrogen which can be in either the ammoniacal,
nitrate, or urea form. Such materials are manufactured by mixing and
reacting ammonia or nitrogen solutions with phosphoric acid or with
triple superphosphate. The degree of conversion can be to either mono-
ammonium phosphate or to diammonium phosphate. Diammonium phosphate
is preferred in today's market because of its higher total plant food
analysis. In many plants, particularly at midwest locations remote
from the primary phosphoric acid facilities, the preferred method is to
react ammonia and nitrogen solutions with triple superphosphate. In
this reaction, part of the monocalcium phosphate is converted to di-
calcium phosphate; the monoammonium phosphate so formed can be further
converted to diammonium phosphate. Ammonium nitrate or urea can also
be present in these ammoniating solutions and upon drying, remain with
the fertilizer material and fortify its plant food content. A substan-
tial portion of the fluorine present in the original phosphate rock is
present in the phosphoric acid or the triple superphosphate used for
the ammoniation reactions. In the reaction and drying step, appreciable
19
Arthur D Little, Inc
-------
quantities of this fluorine are driven off as silica tetrafluoride
together with ammonia gas and must be handled in a vent and scrubber
system. The drying and screening operations are very dusty and this
dust is recovered in cyclones which are part of the vent system.
Wet scrubbers commonly employed after the cyclones are of the vent
system. Wet scrubbers commonly employed after the cyclones are of the
venturi or the cyclonic contact type. There is wide variation in the
efficiency of these emission control devices in the industry.
f. Defluorinated Phosphate Rock
The animal feed industry requires supplements of phosphorus in
their feed preparations, primarily those for cattle and chickens. The
major source of this phosphorus today is from defluorinated phosphate
rock which is low in cost and readily available from phosphate processors.
Fluoride content of the rock used in animal feed preparations must be
reduced to about 0.2% fluoride for safe use.
The process used by several producers today consists of mixing pre-
weighted portions of phosphate rock, wet process phosphoric acid, and
sodium carbonate. The mixture is fed into a counter-current high tempera-
ture kiln or a high temperature fluid bed calciner. Gases from the kiln
or calciner contain essentially all of the fluorine in the phosphate
rock and phosphoric acid feed and dust. Multiple stage wet scrubbers
are used to remove the fluorides as a weak hydrofluoric acid solution
which is neutralized or disposed of by other means.
g. Dicalcium Phosphate
As an alternative to the process described above, some dicalcium
phosphate for animal feed supplements is produced by the reactions of
defluorinated phosphoric acid with lime. A relatively dilute acid
solution is neutralized to the dicalcium stage with a slurry of lime. The
operation is carried out batch-wise and the heat of reaction removes
essentially all of the water leaving a thick paste. This paste is then
dried, the limps broken up and the product packaged for use.
20
Arthur D Little, Inc
-------
Aside from particulate air pollution problems arising from the
handling of the lime and the dried product, the principal source of
pollution stems from the defluorination of wet process acid used in the
process. This defluorination is often accomplished by the addition of
a reactive form of silica such as diatomaceous earth to the crude
acid and then the injection of steam to help volatilize the contained
fluorine. The fluorine evolved from the acid is collected in wet
scrubbers.
C. Summary of Emission Species
A review of the various emission species associated with each of
the stationary source industries considered here are presented in summary
form in Table 1. The principal gaseous fluoride species are HF and SiF,,
both of which are highly soluble and are easily collected in water.
Other gaseous species that would be collected at the same time include
S0_ (and maybe some N0_ and S0_). A variety of particulate fluorides
can be entrained in process effluents. These depend largely on the
raw materials utilized in the unit process operations. In general,
fluorspar (CaF ) is the most prevalent fluoride particulate specie
in the glass and steel industries. Similarly, fluorapatite [Cain(PO,) /-F^
and cryolite (Na_AlF,) are the major fluoride particulate species in
J o
the phosphate rock processing and primary aluminum industries, respec-
tively.
21
Arthur D Little, Inc
-------
Table 1
Summary of Gaseous and Particulate Emission Species
from Stationary Source Industries
Industry
Gaseous
fluorides
Emission Species
Other
gases
Particulate
Fluorides
Other
Particulate
Primary Aluminum HF
C00, CO, Na.AlF
i 3 o
hydrocarbons A1F
Carbon soot
Iron and Steel HF,SiF,
CO ,CO,NO CaF
^ X £,
so2,so3,o2
MnO,Si02,
CaO,Na 0
Glass
so2,so3,co,
C1,NO
,NaF
Si02,PbO
CaO,Na2SO
Phosphate Rock
--Acidulation
so2,so3
CaF,
—Electric Furnace
P5,CO,CH4, Ca1()(P04)6F2, Si02
22
Arthur D Little, Inc
-------
IV. Sample Collection
A. Introduction
The greatest drawback to the reliable collection of fluoride emission
samples from stationary sources is the pronounced chemical reactivity of
most fluoride compounds, especially HF. This is especially important in
light of the fact that fluoride levels in reasonably controlled emissions
are typically in the range of 0.1 to 5 parts per million (ppm). Any
reaction of fluoride with the sample collection equipment or storage
containers could greatly influence the accuracy of the measurement.
Because of the substantial differences in the raw materials and manu-
facturing operations employed in the stationary source industries that are
being considered, and furthermore, within the various unit processes of
each industry, there are considerable variations in the makeup of the
stack emissions. It is necessary, therefore, to know as closely as
possible the types of species that are typically present in the stack
stream as well as their physical and chemical properties, such as particle
size, solubility and gas adsorptivity, so that appropriate stack sampling
procedures and equipment can be employed. Additionally, the analytical
requirements have a major impact on sample collection procedures. For
example, the apparatus and procedure to collect a sample for "total
soluble fluorides" would undoubtedly be much simpler than a requirement
for separate gaseous fluoride, soluble particulate fluoride and insoluble
particulate fluoride analyses. Consideration should also be given to
ancillary information that can be obtained from the collected samples. It
should be possible, in most cases, to measure "total particulate" in
addition to the required fluoride analyses. In addition to emitted
species, consideration must also be given to the physical condition of
the effluent stream, including stream velocity, homogeneity, temperature,
relative humidity and so forth. A very wet stream, for example, may
preclude the use of a filter for collecting particulate.
Two approaches have been suggested for dealing with the chemical
reactivity of fluoride compounds. The classical approach is to employ
23
Arthur D Little, Inc
-------
"inert" materials, such as stainless steel, various plastics and epoxies
and elastomers. However, a sampling system which appears to be "inert"
to one gas stream may, in fact, be highly reactive to another due to a
change in the chemistry of the emission species, temperature or humidity.
An alternate approach is to design the sampling apparatus so that all
gaseous fluorides are intentionally converted to a known, collectable,
relatively stable form. This approach has been applied by employing a
heated glass probe which presumably converts all HF and other gaseous
fluorides to H-SiF- which is highly soluble in water and is efficiently
collected in a liquid impinger. However, conversion efficiencies have
not been determined for the complete range of conditions which would be
encountered in the various industries and processes of concern.
Finally, after stack samples have been collected, it is necessary to
assure that there is no change in chemistry during the storage period
prior to analysis. As a result, it is necessary to establish suitable
procedures for sample transit and storage.
The remaining portion of this section presents a review of fluoride
stack sampling procedures given in the open literature; a discussion of
procedures and equipment employed by representative companies within the
primary aluminum, steel, phosphate rock processing and glass manufacturing
industries; a presentation of the results of some laboratory studies
carried out to evaluate the effect of fluoride ion on sampling train
materials of construction and on sample container materials; and finally,
our recommendations for tentative sampling methods to be evaluated in
an anticipated field study. As used herein, "dry" streams are those
having a relative humidity of less than 100% at stack conditions, whereas
"wet" streams imply the physical entrainment of water (or steam) in the
stream.
24
Arthur D Little, Inc
-------
B. Review of Sample Collection Procedures and Equipment
In terms of measuring the fluoride emissions from stationary sources,
the principal function of the sampling apparatus is to accurately and re-
producibly carry out the collection and separation (where possible) of
gaseous and particulate fluoride species in the presence of other pro-
cess emissions. Only a few studies addressing this subject have appeared
in the open literature. These studies have employed apparatus designed
for the collection of total particulate, such as the sorcalled PHS
(3)
(currently designated EPA) sampling train developed by W.S. Smith.
The apparatus, shown schematically in Figure 5, consists of a heated
glass probe (3), cyclone (4), and filter (5); water inpingers (8-11);
and a dry gas meter (19), vacuum pump (17) and flow meter (20).
(4)
1. Glass Probe Sampling Apparatus Dorsey and Kemnitz found
overall fluoride collection efficiencies of near 100%; however, in the
case of experiments utilizing the lowest range of HF concentration (0
to 0.4%), collection efficiencies were only about 90%. Because of the
limited amount of data presented in this paper, it is difficult to judge
the applicability of the "gaseous fluoride conversion" approach via a
heated glass probe without further laboratory and field-based evaluations.
The fragility of the glass probe in field work, the conversion effi-
ciency when the glass probe becomes coated by particulate dust or tarry
organic species, and the formation of geletinaceous silica hydrate during
the hydrolyzation of SiF, which may plug the impingers are additional
factors that must also be considered for this type of system.
2. Inert Probe Sampling Apparatus The only other stack sampling
procedures for fluorides reported in the literature ' rely on the
utilization of "inert" sampling equipment. Heated components, including
probes, cyclones and filter holders, are generally constructed from
316 stainless steel; other components, such as impinger bottles and
connectors, are made of polyethylene or polypropylene. No data has
been presented on the efficiency of these systems.
25
Arthur D Little, Inc
-------
NJ
—i
a
FIGURE 5 P.H.S. PARTICULATE SAMPLING SYSTEM
(3)
o
-------
Numerous procedures have been employed for ambient air sampling
of fluorides which, in some cases, may be applicable to stack sampling.
The various ambient air sampling methods have recently been reviewed
l'1'
(9)
by Robinson, et.al which updates previous reviews by Farrah
and Mandl, et.al.
C. Present Industrial Practices
During the course of our study, we carried out discussions with
representatives of one or more companies within each of the four indus-
tries to discuss present practice in fluoride emissions sampling and
analysis. Three of the companies had been analyzing stack emissions for
fluorides for over fifteen years, and the company with the least exper-
ience, for five years. As a result of this extensive experience, most
of these companies feel that their methods for the detection of fluorides
are reliable, accurate and well tested. There is a general reluctance
to employ other procedures.
Insofar as was possible, the points raised in these discussions
were directed to elicit the views of the industry in general and not
necessarily those of the company to which we were talking. The following
information was developed during these discussions.
1. Primary Aluminum Particulate emissions from the fused salt
electrolysis cells employed in primary aluminum production consist almost
entirely of alumina (A100_) and cryolite (Na^-AlF,). These species exhibit
£. J -3D
a bimodal size distribution, with a 30ym dust and 0.1-0.3ym fume average.
It is probable that both chemisorption and physical adsorption of HF
onto particulate species occurs in the stack fume. As a result, chemical
analysis of the fluoride distribution between gas and solid phases will
generally be biased in favor of fluoride particulate.
Stack gas streams have temperatures of 100-200°F. As a result,
it has been possible to use various plastics and elastomers that are
highly inert to fluorides for sampling train components. One aluminum
producer utilizes an epoxy paint-coated aluminum tube for a sampling
27
Arthur D Little, Inc
-------
probe; the epoxy provides chemical inertness and the aluminum imparts
mechanical strength and is light weight. Polypropylene, polyethylene
and viton were identified as being highly inert to humidified HF in
laboratory evaluations carried out by this company and were recommended
as construction materials for sampling apparatus components. Tygon
tubing was found to degrade rapidly in HF and should be avoided even for
short connections.
Typical sampling apparatus utilizes a stainless steel nozzle, epoxy-
lined metal probe, heated filter and impingers. Whatman 32 filter paper
has been found to be very efficient for collecting the fine particulate
in the fume. Impingers are held in an ice bath and contain distilled
water or dilute caustic scrubbing solutions to collect gaseous fluorides.
Sampling procedure includes isokinetic sampling at each of four to
six points located along two diameters 90° apart at 0.5 to 1.0 cfm. Each
point is sampled for 10 to 15 minutes, resulting in a total sampling
time of 2 hours.
2. Steelmaking Iron oxide dust is the major particulate emission
from most Steelmaking processes. Although there are some fluoride parti-
culates, most of the emitted fluoride is gaseous. Stack gases are usually
dry and, at the point of sampling, have temperatures of 200-300°F for
an open hearth.
Several unpublished studies carried out by Steelmaking companies
have shown that iron oxide dust is an efficient collector for HF. Since
particulate filters may therefore trap considerable quantities of HF in
addition to the particulate, distribution between solid and gaseous
fluorides could be greatly in error. One approach that has apparently
worked well has been to employ an electrostatic precipitator in place of
a cyclone and filter combination. In this way, the surface area of
the particulate is kept very small and gas-particulate reactions are
minimized. Gaseous fluorides are collected in distilled water or dilute
NaOH solutions.
28
Arthur D Little, Inc
-------
Stainless steel (Type 316) and polyethylene sampling train com-
ponents have been used over long periods of time (several years) with
no evidence of degradation due to exposure to fluorides. This lack of
reactivity may have resulted from the low water content in the gas
stream. Glass components have not been used to date due to their
potential adverse reactions, i.e., reaction of HF with glass to form
a stable phase such as CaF_. Furthermore, many operators consider
glassware to be too fragile for practical use in the field.
Sampling is typically carried out at a single point of average velo-
city at a rate of about 1 cfm. The sampling rate is adjusted for cyclic
operations to maintain gas velocity within 10% of average velocity.
Sampling is usually carried out for one hour. The probe and particulate
collector (precipitator or filter) is maintained at a temperature
greater than the water dewpoint and the impinger solutions are held in
an ice bath.
3. Glass Manufacturing Glass furnaces usually do not employ
any anti-pollution control (APC) devices; control is effected by good
operating practice. Stack temperatures can be as high as 1000°F,
although 600°F is typical. Emissions can vary considerably, depending
on the product being manufactured. There are greater fluoride emissions,
for example, in the manufacturing of fiber glass as well as opal, boro-
silicate and leaded glasses. Except for PbF~, the particulate fluorides
are water soluble. The major gaseous fluoride specie is SiF,, although
HF, F2 and BF, may also be present.
Stack sampling in the glass industry is generally carried out to
monitor visible particulate, SCL, SO, and fluorides. The company we
visited utilized a sampling system consisting of a stainless steel
probe and glass fiber filter (both heated to about 500°F to avoid S0_
and water condensation) followed by an SO- condensor and several water
impingers in an ice bath. The filter is often removed when samples are
being collected for fluoride measurements alone, since all particulate
29
Arthur D Little, Inc
-------
fluorides (except for PbF.) are soluble. Of course, in using this
latter approach, one gets a total fluoride analysis, for it is impossible
to distinguish between soluble particulate and gaseous fluoride fractions.
Sampling is generally carried out isokinetically at a single point
of average velocity at a rate of 0.75-1.25 cfm. Tests usually cover two
reversals in cycle of the heat regenerator system, requiring 50 to 80
minutes. Experimentation is being conducted by one company on the
use of an in-stack particulate collector. Alundum thimbles were found to
pass the fine particulate or, for smaller pore thimbles, to quickly
become plugged and were therefore ruled out as a particulate collector.
An in-stack glass fiber filter in a stainless holder is presently being
evaluated. The advantage of in-stack collectors is to reduce the opportu-
nity for gas-solid phase interactions and transformations within the cooler
portions of the sampling apparatus, which would lead to a false analysis.
4. Phosphate Rock Processing A common characteristic of almost
all of the processes that utilize phosphate rock is a very wet exhaust
gas stream; most of the process streams carry entrained water droplets
or contain large quantities of steam. As a result, particulate filters
have not been used in stack sampling for even though they are heated,
the amount of water entrainment is sufficient to cause plugging. There-
fore, all of the particulate, as well as the soluble gaseous fluorides,
are collected in the impingers, and analyses are limited to providing
the total fluoride content.
Most phosphate rock processing companies employ the procedure adopted
by the State of Florida for the sampling and analysis of water soluble
fluorides in stack emissions. A typical sampling train for use with
this method consists of a stainless steel probe, two impingers in series
containing 0.1N sodium hydroxide for the collection of fluorides, a dry
impinger, a dry gas meter, and finally a vacuum pump. In some cases,
a filter is included between the second impinger and the mist trap to
collect any particulate or aerosol that passes through the impingers.
30
Arthur D Little, Inc
-------
These filters consist of glass fiber filters, alundum thimbles or
funnels packed with glass wool. A schematic representation of the
equipment, which is also used for particulate analysis with the filter
included, is shown in Figure 6.
Sampling is carried out isokinetically at the point of average
velocity as long as there is a low concentration of material entrained
in the stream. In cases where either wet or dry cyclones are employed,
significant tengential stratification occurs in the stream and the samples
are collected incrementally at a series of traverse points.
5. Summary of Methods Employed by Industry The primary aluminum,
glass and steel industries all have "dry" emission streams and can there-
fore carry out separation of gaseous and particulate fluorides by straight-
forward methods. Several types of particulate collectors have been em-
ployed, apparently with good reliability, including electrostatic precipi-
tators and high efficiency filters. The former tends to minimize particulate
surface area which would be expected to reduce the potential for gaseous
interactions with the collected particulate. Pre-treated filters have
been used efficiently for some applications where only light particulate
loadings are encountered. In either case, some of the particulate is
generally collected along the walls of the sampling probe. . Attempts
have been made to reduce the amount of particulate in the probe by em-
ploying an in-stack filter; although there is insufficient data reported
to demonstrate the effectiveness of this approach, further consideration
appears to be warranted. All sampling trains employ impingers to collect
gaseous fluorides. There does not appear to be any difference in the
collection efficiency by water or dilute caustic
Phosphate rock processing operations are characterized by "wet"
streams which generally preclude the use of a particulate collector
before the impingers. Consequently, there is no provision for carrying
out a separation of gaseous and water soluble particulate fluorides for
this case. The apparatus generally used for sampling stack emission
31
Arthur D Little, Inc
-------
v^
Stack H^
(7^ —
.Port
\ Probe &
1 *" n
n
Gas
FT nw
^ — -v
— 1 Tl^.
Thermometer -^
Ice .^
Bath s^
n
\
^
\
1
JJ
f
/ » "N "• "N
X 30-50
^v°F^
«?^
Impingers Fi
200 ml . Disti
Water
Fill
^
^ /^~l>:
•\
7
^
i_
I^—~
&
J)
^
-1
II ^
lied Trap
lied
Filter (Optional)
Meter
Pressure Gauge
Temperature Gauge
Control Valve
Vacuum
Pump
-!
D
cr
FIGURE 6.
SCHEMATIC SKETCH OF WET-STREAM FLUORIDE STACK SAMPLING APPARATUS
o
-------
is therefore the same as for the "dry" stream industries except for
elimination of the particulate collector.
Sampling practice within all industries is reasonably uniform, in-
volving single point isokinetic sampling for a sufficient length of time
(30 minutes to 2 hours) to average the effect of process variations.
In cases where there is good evidence for large variations in stream
velocity from point to point, or for stratification or "cycloning,"
sampling is carried out along a point traverse. Collected samples are
generally brought directly to the analytical laboratory or, when necessary,
are stored and sealed in plastic containers.
D. Laboratory Evaluation of Sample Collection and Storage Materials
1. Evaluation of Sample Probe Materials There is uncertainty in
the efficiency of hot glass probes for converting gaseous fluorides to
(4)
SiF, at low fluoride concentrations. A study by Dorsey and Kenmitz
yielded an efficiency of about 90% for a range of concentrations from 0
to 0.4% HF in air at temperatures of 200 to 360°F. No measurements were
made at the 0.1 to 20 ppm levels which are likely to be found in
stationary source emissions. Furthermore, the reported "inertness" of
316 stainless steel to gaseous fluorides has never been documented.
To aid in the selection of an appropriate material for sampling
apparatus, a series of laboratory evaluations were carried out to
evaluate the suitability of glass and 316 stainless steel as probe
materials. A schematic sketch of the experimental apparatus is pre-
sented in Figure 7. Procedurally, nitrogen containing 1000 ppm HF was
mixed with humidified air to provide concentrations within the range of
0.1 to 100 ppm HF. This stream was then preheated in a furnace coil
to a temperature within the range 100 to 350°F in simulation of stack
conditions. The stream was then split, with equal gas volumes passing
through both legs of a Tee, as determined by calibrated flow meters.
One branch of the Tee passed directly to a series of two impingers while
the other branch passed through a 5-foot length of the sample material
33
Arthur D Little, Inc
-------
=r
Exhaust
t itcli
1 1 i Ai I
, " 1
i ' 1
1 ' fc 1 ' — ' J ' 1
' X^x'^x'^X'^X i ™ ' 1 T/-a Ra-l-hi '
. . ' rro^ Tee M*-*®1-'
"T | (250-350°F) 1 | 1
***• ••' 1
III \ 1
i (200-350°F) !
O
Dieter
hr
_
. .-
i
Pump
£L .. . (I Sample • • rumij
Water p K 1 | — | | — | 1 Meter
Bubbler |(J_ce_Bath]_|
Impingers
HF in N-
Air
C
FIGURE 7 SCHEMATIC SKETCH OF APPARATUS FOR EVALUATION OF MATERIALS IN DILUTE HF STREAMS
-------
TABLE 2
INITIAL AND FINAL LEVELS OF FLUORIDE IN AN INERT GAS STREAM PASSING THROUGH
HEATED GLASS AND STAINLESS STEEL TUBES
Run
Time
(min)
Tap
F~/ft3
(moles
X10~5)
Test
F~/ft3
(moles
X10~5)
F Test/F Tap
(X100)
Run
Time
(min)
316 STAINLESS STEEL
320°F
A-l
2
3
4
5
6
7
8
9
10
11
12
200°F
C-13
14
15
D-16
17
18
19
20
21
22
23
24
30
30
30
30
40
30
30
30
30
30
180
160
30
30
30
30
30
30
30
30
30
30
30
27
0.9
0.8
1.25
1.3
0.03
1.15
1.45
1.15
1.3
0.2
0.01
0.00
Average
Standard
1.2
1.3
0.1
1.25
1.2
1.25
1.65
1.5
1.4
0.06
0.03
0.02
1.0
1.05
1.65
1.6
0.2
1.15
2.0
1.2
1.35
0.2
0.01
0.01
(8 runs) 1
Deviation
1.2
1.2
0.1
1.0
1.55
1.15
1.4
1.25
1.5
0.13
0.02
0.01
111
131
132
123
100
138
104
104
118
13.7
100
92
80
129
92
85
83
107
320°F
H-39
40
41
42
43
44
45
46
47
200°F
F-29
30
31
32
G-33
34
35
36
37
38
30
30
30
30
30
30
30
30
30
30
30
30
30
30
30
30
30
30
30
Tap
F'/ft3
(moles
X10'5)
Test
F~/ft3 I
(moles
X10~5)
r Test/F Tap
(XLOO)
BOROSILICATE GLASS
1.45
1.45
1.35
1.35
1.35
1.30
1.35
0.05
0.01
Average
Standard
1.55
1.35
0.15
0.02
1.45
1.40
1.35
1.45
0.09
0.01
1.30
1.40
1.35
1.30
1.30
1.25
1.30
0.13
0.04
(7 runs)
Deviation
1.20
1.30
0.30
0.06
1.25
1.25
1.35
1.35
0.30
0.09
90
97
100
96
96
96
96
—
—
96
2.6
77
96
—
—
86
89
100
93
—
—
Average (8 runs) 96
Standard Deviation 15.0
Average (6 runs) 90
Standard Deviation 7.3
35
Arthur D Little, Inc.
-------
to be evaluated and then into impingers. The impingers were held in an
ice bath and contained 0.1 N NaOH. During the course of a run, the sample
tube was maintained at a preselected temperature in the range 200-350°F
by means of heating tape to approximate stack sampling conditions. The
impinger solutions were subsequently analyzed for fluoride by adjusting
pH and carrying out a direct measurement with a specified ion fluoride
electrode previously calibrated by standard additions of sodium fluoride.
A series of experiments was carried out with this apparatus at an
HF concentration of 10 ppm and sampling rate of 0.1 to 0.2 cfm, utilizing
5-foot lengths of 316 stainless steel and borosilicate glass as samples.
Measurement of the volume increase in the impingers showed a 4% water
content in the gas stream. Experimental runs generally covered 1 to 4
hours, with collection and analysis of impinger solutions after each
30-minute interval. At the conclusion of a run, the HF stream was
turned off and air was allowed to continue to flow through the system.
Several additional 15-minute impinger samples were collected to determine
the length of time required to purge fluoride from the system.
The experimental data, which is normalized to constant volume, is
presented in Table 2. A comparison between the initial (tap off) and
final (sample) fluoride concentrations are presented for the stainless
steel and glass tubing samples in Figures 8 and 9, respectively. In
the case of stainless steel, Figure 8, the observed increase in fluoride
after passing through the sample (Run A at 320°F) is attributed to some
defluorination of a teflon plug which was used for the gas seal at the
front of the stainless tube. With the exception of one bad data point,
indicated by an arrow, Run B shows excellent collection efficiency. The
200°F data for stainless steel exhibits considerable scatter, especially
in Run D, which is due mostly to experimental adjustments during the
course of the run to achieve equal flow rates in the tap off and sample
streams. If the six data points are averaged, initial and final stream
concentrations agree quite closely; the average of all 200°F data yields
a collection ratio of fluoride out to fluoride in of 96% with a stan-
36
Arthur D Little, Inc
-------
2.0
t/1
O
iH
X
w
O
M
H
W
U
z
O
O
w
Q
l-l
Pi
O
1.0 -
O SAMPLE
TAP OFF
FIGURE 0
1 2
TIME - HOURS
A COMPARISON OF INITIAL (TAP OFF) AND FINAL (SAMPLE)
FLUORIDE CONCENTRATIONS AFTER PASSAGE THROUGH A FIVE
FOOT 316 STAINLESS STEEL TUBE
Arthur D Little, Inc.
-------
dard deviation of 15.0. After sampling experiments with the stainless
steel were completed, the tube was washed with water. These washings
were found to contain 3.1 x 10 moles of fluoride, indicating very little
adsorption of fluoride onto the tube walls.
With the benefit of having determined proper flow rates and other
experimental settings from the stainless steel experiments, the data for
the glass sample, Table 2 and Figure 9, are considerably more uniform.
The outlet concentration was found to be lower than the initial concen-
tration in all samples measured. By averaging the data, the collection
ratio for the glass sample is 90% at 200°F and 96% at 320°F, with
standard deviation of 7.3 and 2.6, respectively. Probe washings from
the glass tube at the conclusion of experiments yielded 1.56 x 10
moles of fluoride, which is about five times more than what was found in
the stainless sample.
The experiments with a 316 stainless steel sample should probably
be repeated to reflect the improvements in the experimental procedure
that were incorporated in the glass sample experiments. Also, variations
in HF concentration and flow rate should be explored. We feel that it
is imperative to carry out similar evaluations of collection ratios at
levels of HF that more nearly represent well controlled stack (i.e., 0.1
to 1 ppm HF) and flow rates up to 1.5 cfm before a specific probe material
for sampling fluorides can be recommended.
2. Sample Storage Since a very significant part of the sample
collection procedure involves containment of the collected samples prior
to analysis, a laboratory evaluation was carried out to determine the
degree to which fluoride ion is lost to the container. For these
-4
evaluations, acidic, neutral and basic solutions containing 1 x 10
molar solutions of fluoride added as NaF and as Na0SiFc were stored in
2. o
glass and polyethylene bottles for a period of 27 days. During this
time, aliquot samples were periodically removed from the containers and
their fluoride content measured to ascertian whether or not losses were
occuring.
38
Arthur D Little, Inc.
-------
2.0
u-i
o
,-H
X
V)
w
o
M
H
w
o
a
o
w
Q
h-l
Pi
s
i.o
i.o
320°F
0 SAMPLE
A TAP OFF
Run 1
Run 1
Run 2
TIME - HOURS
FIGURE 9 A COMPARISON OF INITIAL (TAP OFF) AND FINAL (SA?tPLE)
FLUORIDE CONCENTRATIONS AFTER PASSAGE THROUGH A FIVE
FOOT BOROSILICATE GLASS TUBE
39
Arthur D Little, Inc.
-------
TABLE 3
STABILITY OF DILUTE SODIUM FLUORIDE SOLUTION
Solution
NaF in 0.1N NaOH
NaF in H20
NaF in 0.001N
H,,SO,
Na0SiF, in 0.1N
2. O
Na0SiF, in H00
f. O 2.
Na_SiF,. in
z o
0.001N H,
IN GLASS AND POLYETHYLENE CONTAINERS
Container
Material
*
Fluoride (moles/liter)
Start
iOH glass
polyethylene
glass
polyethylene
glass
polyethylene
N glass
iTJ
polyethylene
i glass
polyethylene
glass
polyethylene
9.
0.
10.
10.
9.
9.
9.
9.
9.
9.
9.
9.
IxlO-5
1
1
1
5
5
9x10-5
9
8
8
3
3
9
9
9
9
9
9
9
9
9
9
9
9
2 Days
.0x10-5
.4
.5
.6
.0
.1
.4x10-5
.5
.4
.6
.3
.4
7
9.
9.
9.
9.
10.
10.
9.
9.
9.
9.
9.
9.
Days
9x10-5
9
8
9
1
1
6x10-5
8
8
9
8
9
27
10.
10.
9.
10.
10.
10.
9.
9.
9.
9.
9.
9.
Days
3x10-5
1
9
3
3
3
8xlO~5
9
6
9
9
9
Forty (40) ml of sample (pipet) plus 5 ml 1 M sodium citrate
were neutralized (citric acid) to bromthymol blue and diluted
to 50 ml for measurement by the direct fluoride electrode.
40
Arthur D Little, Inc
-------
(a)
Fi
(b)
M
••
•^•^M
Iter
\
\
\
\
^
1^^
\
\
\
i
«
^M
Heated Probe
1
\
\
\
M^M
\
\
N
Heated Probe
and Fi
1_J 0 f^\
1
" 1-
1
1
i
M^H
-
4,
-1
i
1
i
\ V J i
Gas Pump
— ' Meter
l(Ice Bath) | Mist
Impingers Trap
I
i o^\
1 ,
Her j |
i
*
>*<
4
1
I
I
1
1 ^
(c)
f
\
\
HM^
\
Heated Probe l\\ '
^nd 1 l__\s- —
1 ' r \
Preci pita tor ,
1 1 1
1
i
\
i. f
1
o
FIGURE 10
SCHEMATIC SKETCH OF FLUORIDE SAMPLING APPARATUS
41
Arthur D Little, Inc
-------
The glass containers were new flint glass bottles which had been
washed with distilled water prior to use. The polyethylene bottles
had been used previously, although not for the storage of fluoride
solutions and were also washed prior to use. The results of this
evaluation are shown in Table 3. In no case is any significant loss
of fluoride readily apparent. It is concluded that for these levels
of fluoride either flint glass or polyethylene bottles are suitable
sample containers. Because of their fragility, however, glass containers
are less acceptable for use in the field.
E. Conclusions and Recommendations
Within the four industries considered in this study, primary aluminum,
steel and glass manufacturing plants are characterized by "dry" stack
gas streams. We would expect that a single sampling train and sampling
procedure would be appropriate for all process streams within these
industries. A schematic representation of the required sampling train
is presented in Figure 10. Particulate and gaseous fluorides would be
collected separately. Furthermore, suitable analytical methodology
would permit the separation of total fluoride into soluble particulate,
insoluble particulate and gaseous fractions. In most cases, the total
collected particulate could also be determined with no change in sampling
or analysis procedure.
With reference to Figure 10 there are three optional procedures for
collecting particulate. In-stack collectors (a) have been used infre-
quently and there is insufficient background data to support their use at
this time. Consequently, the tentatively recommended procedure for
particulate collection is either by a filter utilizing either a glass
fiber or organic membrane collection media, (b), or an electrostatic
precipitator, (c). Any of these would be maintained at temperatures
above the water dew point but, in the case of a membrane filter, not in
excess of 250°F. Gaseous fluorides would be collected in a series of
two Greenburg-Smith impingers containing either distilled water or dilute
NaOH.
42
Arthur D Little, Inc
-------
In difference to these industries, the phosphate rock processing
operations are characterized by "wet" emission streams. As a result
of copious amounts of water and steam, it is not sensible to attempt
a separation of particulate and gaseous fluorides. The same sampling
apparatus as shown schematically in Figure lOa could be employed by
removing the particulate collection component from the train.
Based upon an examination of the open literature, discussions with
industry representatives and our own laboratory evaluations, both 316
stainless steel and borosilicate glass materials are suitable for heated
sampling train components, including the probe, precipitator body and
filter holder. At lower temperatures (250°F and below), epoxy or plastic-
coated metal tubes can be used for the probe. Impingers, mist traps and
connectors can be made of glass or polyethylene. The fragility of glass
makes it less suitable than the other materials for field sampling.
Sampling should be carried out in a careful manner to provide a
high level of reliability. Pre-sampling checkouts should include a
velocity traverse, measurement of typical stream water content, etc.
In general, sampling should be isokinetic and carried out at one or
more points to reflect the "average" stream conditions at a rate of
0.75 to 1.5 cfm. Sampling times should reflect cyclic variations in
the process being monitored, but should be at least one hour. All
required data should be recorded in an appropriate format to allow
subsequent calculations of emission levels and rates.
After sampling, cleanup should be performed carefully to avoid any
contamination or loss of sample. Separate containers should be employed
for filter (or precipitator) particulate, impinger catch and probe washings.
These may be combined during analysis, depending upon analytical require-
ments. Either glass or polyethylene containers may be used; the latter
are recommended due to their ruggedness.
A procedural outline of the tentatively recommended stack sampling
procedure is presented in Appendix A. It must be recognized that the al-
ternative components, materials and procedures that are specified must
43
Arthur D Little, Inc
-------
be subjected to field evaluation before final recommendations can be made.
Finally, there is presently no basis for determining or estimating the
precision, accuracy, sensitivity or detectability of the tentatively recom-
mended sampling methods.
44
Arthur D Little, Inc
-------
V. SAMPLE ANALYSIS
A. Introduction
Measurement of the fluoride content of solid, liquid and slurry sam-
ples collected from various industrial stationary sources requires a
combination of analytical techniques including solubilization of the
sample, removal of interfering species and fluoride measurement. Numerous
methods for accomplishing these procedures have been employed over the
years. The most promising approaches, as judged by a critical review of
the literature and interviews with several industrial laboratories, have
been incorporated into a comprehensive laboratory evaluation of methods,
utilizing samples representative of typical emissions from primary alumi-
num, steelmaking, phosphate rock processing and glass manufacturing indus-
tries. From these studies, tentative fluoride analysis procedures have
been developed and are recommended as primary candidate methods to be
employed in subsequent field test evaluation studies.
This section presents an overview of the requirements of an analy-
tical method; a comprehensive review of existing methodology, including
current industrial practice; the results of a laboratory evaluation of
candidate methods for sample fusion, distillation and measurement in
the presence of various potential interfering chemical species; analysis
of several stationary source samples collected in the field to provide
insight into the nature of field samples as well as to evaluate the candi-
date methods; and finally, our conclusions and recommendations.
B. General Considerations
1. Selection of an Analytical Method Before describing the various
methods that are potentially useful for measuring fluoride, the major
criteria by which the applicability of a method can be judged are presented.
Factors which must be considered include the physical and chemical nature
of the samples to be analyzed, the manner in which they were obtained, and
the conditions under which the analyses are performed. The chemical
45
Arthur D Little, Inc
-------
nature of the fluoride-containing compounds in conjunction with other
compounds which may constitute potential interferences and also the range
of fluoride concentrations to be encountered are clearly important.
Other factors that should be considered include the simplicity, adapta-
bility, accuracy and precision that is required.
A wide range of fluoride concentrations can be expected for the
process effluents encountered in the aluminum, iron and steel, phosphate
rock, and glass and ceramic industries. Depending on the particular pro-
cess and the efficiency of any APC devices employed, fluoride concentra-
tions in the emissions stream can range from less than 0.1 ppm (v/v) to
upwards of 50 ppm or more. The amount of fluoride collected during
sampling obviously depends on both the sampling flow rate and time;
typical practice involves a range on the order of one cfm for a minimum
of one hour. For these conditions, the total amount of fluoride collected
can range from 0.15 to 75 mg—a dynamic range of 500. Fluoride emissions
consist of both particulate and gaseous species; the particulate/gaseous
ratio covers a range of 0.1 to 10. In cases where particulate and gaseous
fluorides are collected separately and analyzed individually, it is possi-
ble that the amount of fluoride in one or the other of the two fractions
could be as low as 15 yg. The dynamic range requirement for an analytical
method is therefore increased another order of magnitude to about 5,000.
The necessity to measure fluorides over such a wide dynamic range with
little if any ji priori knowledge of the fluoride concentration is clearly
a significant constraint on the measurement method.
For the four industries of concern in this study, the chemical nature
of the collected sample is widely variable. Chemical compounds that can
be expected to be present in the process effluent streams have been given
previously in Table 1. Both gaseous and particulate fluoride-containing
compounds are emitted from most processes. Hydrogen fluoride and SiF,
are the most commonly encountered gases. Fortunately, these gases are water
soluble and solutions containing them are not particularly difficult to
handle analytically. In contrast, particulate fluorides can range from
the very insoluble and refractory dryolites and chiolites, to slightly
46
Arthur D Little, Inc
-------
soluble phosphate rock and fluorspar, to very soluble sodium fluoride
and sodium fluosilicate. Other species present in the effluent which do
not contain fluoride but which can cause potential analytical interferen-
ces include gases, such as the sulfur and nitrogen oxides, and solids,
having varying solubility in water, containing metals such as aluminum,
iron, calcium and silicon.
Another consideration concerns the simplicity of the candidate
analytical method. Potential users of the method include personnel asso-
ciated with federal, state, and municipal regulatory agencies. Since
fluorides constitute only one of many pollutants for which laboratory
analyses will be required, the candidate method should be sufficiently
simple so that a technician who has never previously made fluoride mea-
surements can obtain reliable results in a short time and with a minimum
of training and practice.
It is desirable that the method be sufficiently flexible to be easily
adapted to future requirements for fluoride measurement. If particular
emphasis should eventually be placed on fewer industrial processes than at
present, it is possible that the nature of the sample obtained from these
processes would be such as to permit a simpler analytical scheme to be
used. The present analytical method should therefore be easily modified,
preferably by deletion of one or more steps, to meet new requirements.
The accuracy and precision of analytical method is obviously an
important factor which must be considered; however, undue emphasis should
not be placed on accuracy and precision at the expense of the other con-
siderations mentioned previously. A given analytical method in the hands
of a highly experienced analyst may produce results with an accuracy and
precision of better than + 1%. An alternate method may be good to only
jh 5-10%, even in the hands of a relatively inexperienced operator having
little information on the nature of the sample. This second method may,
in fact, be preferable to the more precise method, which in the hands of
the inexperienced technician, may produce errors far larger than + 10%.
Additionally, and of more importance, the analytical precision and accuracy
47
Arthur D Little, Inc
-------
must be considered in terms of the magnitude of errors introduced in other
steps of the overall measurement. Errors associated with sampling, while
not well defined, could be 10-25%, in which case an analytical method
precise and accurate to + 1% is not at all warranted.
The variance (square of the standard deviation) of the overall mea-
surement equals the sum of the variances of the individual measurement
steps. Thus, if random sampling errors amount to + 25%, the use of an
analytical procedure good to +^ 10% would increase the overall measurement
error by only 2%. If random errors in sampling and analysis were both
+ 10%, the error in the overall measurement would be 14%. As a base line,
it is assumed that an analytical procedure having a precision and accuracy
of + 5-10% is probably sufficient for the measurement of fluorides.
2. Fundamental Steps in Fluoride Analysis Procedures With the
exception of certain instrumental techniques, such as X-ray fluorescence
and neutron activation analysis, virtually all fluoride analytical pro-
cedures involve the measurement of free fluoride ion in solution. In
many cases, the actual measurement is the quickest and easiest part of
the overall procedure. If the sample to be analyzed contains particulate
material, the solids must be put into solution. In cases where there are
also other ions in solution which can bind fluoride into complexes, the
fluoride must be freed from the complex. Other species present in solution
may interfere directly with the chemical reaction or the indicator used
in the measurement; such interferences must also be dealt with. Thus,
the analysis of fluoride typically involves three steps—dissolution of
solids, separation of fluoride from interferences, and finally, measure-
ment of fluoride ion.
Samples obtained from process effluent streams and subsequently re-
turned to the laboratory for analysis can range from clear impinger solu-
tions, containing only dissolved gaseous species, to filter papers contain-
ing dry particulate to aqueous slurries containing varying amounts of
dissolved and undissolved chemical species. Impingers are normally
charged with a total of 200-400 ml of collecting solution. Some condensa-
48
Arthur D Little, Inc
-------
tion occurs during sampling, which increases the final amount of impinger
solution. Thus, the gaseous fluorides that are collected are usually pre-
sented for analysis in a total volume of 300-500 ml. When the relative
humidity of the process stream is very high, as is the case in many streams
found in the phosphate rock processing industry, significantly more
condensation can occur, and a final impinger catch in excess of one liter
is sometimes obtained.
Filter papers containing particulate are usually returned to the
laboratory intact; total particulate catch can range from 10-1,000 mg.
If a cyclone collector is used ahead of the filter, the filter catch may
be only a fraction of the total particulate. If a cyclone is not used,
most of the particulate material will be found on the filter paper. In
either case, some portion of the particulate usually adheres to the walls
of the sampling probe and is collected in the probe washings.
Regardless of the sampling train employed, the sampling probe, cyclone
(if used), and filter holder or electrostatic precipitator must be washed
to remove all particulate material. The final volume of these washings
is again usually on the order of 300-500 ml; in some cases, these washings
may contain most of the particulate material collected.
Typical procedures for handling the three types of samples which can
be encountered are shown schematically in Figure 11. Because of their
widespread use and general applicability, fusion and distillation have been
substituted for the more general terms,solubilization and separation,
respectively. The normal procedural routes are shown by the solid arrows.
In addition, a number of abbreviated procedures are shown by the dashed
lines; the condition which must be met for the shortcut to be valid is also
given. One common simplification is the elimination of the fusion step if
the solids can be brought into solution by other means and if the resulting
solution is compatible with the measurement technique being used. This
abbreviated procedure is often employed in the phosphate rock industry
where the particulate material is usually soluble in acid. A second
simplification that is sometimes possible is the elimination of the separa-
tion step; this is permissible if the level of interfering species is such
49
Arthur D Little, Inc
-------
CO
l-i
0)
tn
0)
o
g
C
•H
T3
•H
U
nj
O
W
tn
T3
•H
O
cn
Add CaO and
evaporate
Filter Paper
Add CaO Solution
Dry and Ash
. FIGURE 11
Stepwise Procedures for Handling Fluoride Samples
(Slurry can also be filtered and the two fractions treated
following procedures for filter paper and
clear impinger solutions)
50
Arthur D Little, Inc
-------
that they do not adversely affect the measurement method being used.
This simplification may be widely applicable when the fluoride ion
electrode is used in conjunction with a complexing buffer to measure
fluoride concentrations impinger solutions which contain only dissolved
gaseous species.
C. Review of Existing Methodology
Based on a critical review of the literature and on discussions with
laboratory personnel in the industries of concern to this study, a number
of techniques for sample dissolution, separation of interferences, and
measurement of fluoride have been compared to select candidate techniques
for further testing in the laboratory. Means for effecting each of the
three principle analytical steps are discussed in the sub-sections which
follow. In addition, a discussion of techniques which are being used
successfully in the laboratories of the industries of concern is also
included.
1. Measurement Procedures Candidate procedures for measuring fluo-
ride which can be considered candidates primarily involve volumetric or
colorimetric determinations or the use of the fluoride specific ion
electrode. A variety of other techniques have been used for measuring
fluoride in specific cases. Many of these are of limited general
applicability and/or require complex, expensive instrumentation. For
these reasons, these other techniques are not considered to be candi-
dates for use in a widely applicable standard method, and they will not
be discussed further. For a rather complete discussion of such techniques
along with references to the original literature, the reader is referred
to the recent RRI/TRW report. '
a. Volumetric Techniques
Fluoride ion can be determined volumetrically by titrating with salts
of one of a number of metals such as thorium, zirconium, iron, aluminum,
or lanthanum, all of which form stable fluoride complexes in solution.
The most commonly used titrants are the thorium and lanthanum nitrates.
51
Arthur D Little, Inc
-------
The titration endpoint is usually determined from the color change of an
indicator dye such as Alizarin Red-S, Purpurin Sulfonate, or Eriochrome
Cyanine R. The indicator color change at the endpoint is either
observed visually or measured photometrically; the latter procedure is
more time consuming but eliminates the human judgmental errors that are
often associated with a visual observation of the faint color change at
(12 13)
the endpoint. ' Non-colorimetric means of endpoint detection have
(14)
been used. They include the fluoride ion electrode, fluorescing
indicators, oscillometry, and conductometry.
Of the many reported schemes for volumetric fluoride analysis, the
most widely accepted and employed is still the titration with thorium
nitrate using Alizarin Red-S indicator. This method is described in
detail in the ASTM Method for Inorganic Fluoride in the Atmosphere
(D 1606-60). Its advantages and limitations are generally similar to
most volumetric fluoride methods. It can be used for a wide range of
fluoride concentrations—0.005-10 mg—a 2,000-fold range. However,
variations in the procedure are recommended for making measurements of fluo-
ride concentration in the range of 0.005-0.01 mg, 0.01-0.05 mg, and
0.05-10 mg. Thus, to use the method over its widest possible dynamic
range, some knowledge of the approximate fluoride concentration must be
available—either from prior experience or from trial and error determina-
tions on separate aliquots of the same sample.
If the solution to be titrated is carefully freed of interferences
and the method is put in the hands of an experienced technician, the
thorium nitrate titration is one of the most accurate and precise fluo-
ride measurement techniques, particularly in the upper part of its appli-
cable concentration range. Precision and accuracy of better than +1%
can be obtained.
The method suffers from a variety of interferences; a careful distilla-
tion from perchloric acid is almost always required to provide an inter-
ference-free solution for titration. Ions such as phosphate, sulfate,
and oxalates interfere in the titration by forming precipitates or
52
Arthur D Little, Inc
-------
complexes. Nitrates, peroxides, sulfides, and sulfites interfere with
the formation of the indicator color. Aluminum and silicon interfere by
inhibiting the release of fluoride during the distillation. In the pre-
sence of significant amounts of these metals, a double distillation,
first from sulfuric acid and then from perchloric acid, is often required.
One of the primary reasons for choosing a volumetric determination,
the fact that there is usually a known constant stoichiometry of reaction
between the sample and titrant, is often not true for titrations of fluo-
ride, particularly at low fluoride concentrations. Consequently, the
use of a calibration curve generated by titrating known amounts of sodium
fluoride is generally required. Finally, the method requires a high de-
gree of manipulative technique. The titration is usually performed with
a microburet to minimize sample dilution by the titrant; reproducible detec-
tion of the indicator color at the endpoint requires a good deal of
practice on the part of the analyst.
b. Colorimetric Techniques
Most of the colorimetric techniques utilized for the determination
of fluoride involve the bleaching by fluoride ion of a colored metal-
indicator complex. A few of the better known methods of this type include
(18) (19)
aluminum-Eriochrome cyanine R, thorium-thoron, zirconium-Alizarin
(2fH (21 22)
Red-S,v ' and zirconium-SPADNS. ' ' Methods of this type which involve
the bleaching of a color have the disadvantage that the initial solution
is highly colored. Low concentrations of fluoride reduce the color inten-
sity only slightly; thus the analysis requires that a small difference
between two large absorbence readings be measured accurately.
Recently, a number of direct color reactions, in which the color
intensity is directly proportional to fluoride concentration, have been
(23)
reported. These include use of the chloranilate salts of lanthanum
(24)
and thorium; fluoride forms a complex with the metal ion releasing
highly colored free chloranilic acid. Fluoride reacts with Lanthanum-
Alizarin complexone reagent to form a single complex species which is
(25)
intensely blue in color.
53
Arthur D Little, Inc
-------
Colorimetric methods in general are more sensitive than volumetric
methods and can be used with a smaller sample as well. Precision and
accuracy are generally about + 5%. Many of the methods can detect as little
as 0.01 yg/ml of fluoride. Virtually all methods require that a calibra-
tion curve be generated from known standard fluoride solutions. Using a
given reagent system, the calibration curve is usually linear over a
ten- or twenty-fold concentration range. To extend the range of concentra-
tion to which a method can be applied, the sample can be diluted or a
different reagent system must be chosen. It is possible, by selecting
appropriate reagents, to measure fluoride concentrations ranging from
0.2 mg/ml (Iron-Ferron reagent, range 0.01-0.2 mg/ml) to 0.01 yg/ml
(8)
(Lanthanum-Alizarin Complexone reagent, range 0.01-0.4 yg/ml.) In
any event, the use of colorimetric techniques over a wide dynamic range
requires some prior knowledge of the approximate fluoride concentration.
Similar to the volumetric methods, colorimetric fluoride methods are
subject to a variety of interferences; many of them the same species
interfere for the same reasons. Typical interferences include aluminum,
iron, phosphate, sulfate, chloride, oxalate, and nitrate. Reliable
colorimetric methods, therefore, usually require a prior separation of
fluoride from interferences by distillation or diffusion.
c. The Fluoride Specific Ion Electrode
(26)
In 1966, Frant and Ross reported the development of an electrode
containing a membrane fabricated from a single crystal of lanthanum fluo-
ride which had been doped with divalent europium. The electrode was
capable of selectively measuring fluoride activity in solution over a
wide range of fluoride concentrations. Since that time the electrode
has received a great deal of attention and has been evaluated in many
laboratories on a wide variety of fluoride-containing samples.
Of all the measurement techniques for fluoride, the fluoride electrode
is the least sensitive to interfering species and has the widest dynamic
range. It is useable (and a linear calibration curve can be obtained)
for fluoride concentrations ranging from 10~ to 10 M (2,000-0.02 yg/ml) .~!'
54
Arthur D Little, Inc
-------
Measurements can be made routinely in as little as one milliliter of solu-
and
(29)
( 28}
tion, and on even smaller volumes if the electrode is modified
slightly.
The only known interference in the electrode reaction is due to
hydroxyl ion; the electrode response to hydroxyl ion is the same as its
response to fluoride. Thus at high pH, small variations in pH can intro-
duce sizable errors into the measurement of low concentrations of fluo-
ride. However, if pH is held in the region of 5-7, pH-variations have
no effect even at the limit of detection of the electrode.
Like all analytical techniques employing an electrode reaction, the
fluoride electrode response is proportional to activity rather than to
concentration. However, valid concentration measurements can be made
if care is taken to be sure that the fluoride ion activity coefficient is
the same in calibration and unknown solutions. Changes in activity
coefficient are usually caused by variations in the ionic strength of the
unknown solution. These effects can be eliminated by adding to the un-
known solution a volume of high ionic strength buffer (often 2 M) to
swamp out small differences in ionic strength of dilute solutions. Frant
and Ross have coined the acronym TISAB (total ionic strength adjust-
ment buffer) to describe a family of buffers of this type. Sodium chloride
is usually used to achieve the described high ionic strength. Sodium
acetate/acetic acid is sometimes included to provide pH buffering.
Like the volumetric and colorimetric techniques discussed previously,
interference in the fluoride electrode measurement can be .caused by the
presence of species in the sample which complex fluoride ions; the elec-
trode only measures the activity of free fluoride ion. If cations such as
aluminum, iron, or calcium are present at a very high level, a separation
step may be necessary. However, at moderate levels, the effect of these
interferences can be removed by adding a complexant to tie up the metal
(31) (32)
and release the fluoride. Edmond and Tusl both report good
success by adding 0.5 M citrate to complex calcium ion in measurements
(33)
of phosphate rock samples. Oliver and Clayton used IM citrate to
55
Arthur D Little, Inc.
-------
minimize the interference due to aluminum that was present in a variety
of samples. Unfortunately, citrate also tends to react with the lanthanum
in the electrode membrane and causes the electrode membrane and causes
the electrode response to become increasingly slow at low fluoride levels
-4 (34)
(less than 10 M). Ingram has found that the use of nitrate along
with the citrate is beneficial in minimizing the deleterious effects of
citrate on the electrode.
With the exception of hydroxide, ionic strength, and certain cations,
the fluoride electrode is virtually interference free. Notably absent
are interferences due to anions such as sulfate, nitrate, phosphate, and
chloride which plague many other methods. The precision and accuracy
of fluoride electrode measurements has been reported by many workers
and generally is in the range of from 1-5%. This range of precision and
accuracy probably reflects in large measure the varying amount of care
with which the measurements were made. The temperature of the sample
solution will change the electrode response; a change of 1°C will produce
a 1.5% relative error in the measurement.^ Lack of stability and/or
resolution in the pH meter or electrometer used to measure EMF can intro-
duce significant error. An error of 1 millivolt in the EMF measurement
produces a relative error of 4% regardless of the absolute concentration
being measured. Failure to adequately deal with the problem of slow
electrode equilibration particularly at low fluoride concentrations can
(34)
also lead to significant errors.
2. Solubilization Techniques
a. Fusion
The most generally applicable technique for putting unknown insoluble
materials into solution is to perform a fusion with an alkaline flux.
In the case of fluoride, the insoluble material is converted to the
alkali metal fluoride, which, upon subsequent dissolution of the melt,
is water soluble. General procedures for handling particulates on filter
papers or in slurries are described in the ASTM Standard Method for
56
Arthur D Little, Inc
-------
Inorganic Fluoride in the Atmosphere (D 1606-60). Filters must be ashed
prior to fusion to destroy the paper and any other organic materials
that are present. Ashing at a temperature less than 600°C until no
carbon remains is of vital importance. If carbon is present during the
fusion, fluoride losses can occur. The filter paper is first wetted with
low-fluoride lime water to prevent loss of volatile fluorides, and the
ashing is then performed at 550-600°C. The solids remaining after
ashing are mixed with a quantity of the flux and fused over a burner or
in an electric furnace. Because of the corrosive nature of the fusion,
it is usually carried out in nickel, platinum, or Inconel crucibles.
Slurries can be treated in one of two ways. The slurry can be
filtered and the filter paper treated as described above. After fusion,
the solids can be redissolved in the filtrate fraction. Alternatively,
separate analyses of the fused solids and filtrate can be performed.
A second procedure is to evaporate the total slurry to dryness, add
flux, and perform the fusion. Low-fluoride calcium oxide must be added
prior to evaporation to minimize evolution of volatile fluorides as the
evaporation proceeds. The choice between filtering or evaporating a
slurry is not well defined. The filtration procedure is probably more
time consuming in that an ashing and perhaps two fluoride measurements
are required, but filtration does eliminate the possible fluoride losses
to the atmosphere and to the sides of the vessel that can occur during
the evaporations step.
The choice of a fusion flux is somewhat arbitrary; sodium hydroxide
and sodium carbonate are the most widely used. The ASTM Method D 1606-60
[
(8)
(35)
recommends a fusion with sodium hydroxide; VanLoon has reported
successful fusions of cryolite and chiolite with this flux. Farrah
(33)
and Oliver and Clayton have used sodium carbonate to fuse a variety
of samples encountered in the aluminum industry. The latter workers
also used a mixture of sodium carbonate and sodium borate to more com-
pletely solubilize particularly intractable materials. Evans and
Sergeant and Guth and Wey both have reported the use of sodium
57
Arthur D Little, Inc
-------
carbonate fusions in conjunction with the analysis of fluorides in rocks
and minerals.
b. Acid Dissolution
For materials known to be acid soluble, the fusion can be replaced
with a simple dissolution in acid. This approach is widely used in the
(31 32 38 39}
analyses of phosphate rock and related materials ' ' ' Care
must be taken to minimize losses of volatilized fluorides from the
acidic solutions. Acid dissolution followed by a direct electrode
measurement of fluoride in the resulting solution is often feasible.
If an acid distillation must be performed prior to measurement, the
solid material can be introduced directly into the distillation flask.
c. Pyrohydrolysis
A pyrohydrolytic technique which, in essence, combines the steps
of solubilization and separation into one has been reported by Clement,
et al. They report that the procedure is particularly effective in
removing fluoride from intractable rock and mineral samples. The sample
is mixed with a flux composed of bismuth trioxide, sodium tungstate,
and vanadium pentoxide. The mixture is then heated at 700-750°C in the
presence of a flowing stream of water vapor. Fluoride is evolved from
the sample as HF and is collected in alkaline solution for subsequent
measurement.
d. Ion Exchange
Cation exchange resins have been used to effect the dissolution of
(39)
phosphate rock samples. About 5 grams of resin in the acid form is
agitated with a 50-ml water slurry of the sample to be dissolved. The
cation exchanger binds calcium, and the fluoride anions go into solution.
A period of shaking of up to four hours may be required for complete
dissolution, after which the mixture is slurried into buret tube and
rinsed with a few column volumes of water to flush out all fluoride. The
resulting solution is then diluted to volume for subsequent analysis.
58
Arthur D Little, Inc
-------
3. Removal of Interfering Species The need for and choice of a
technique for removal of potentially interfering species depends both on
the nature of the sample to be analyzed and the characteristics of the
measurement method which is to be employed. The better known techniques
include distillation, diffusion, and ion exchange; these three techniques
are discussed separately in subsequent sections. The removal of the
effects of interfering cations by masking with ligands that complex the
cation has been discussed previously in connection with the fluoride ion
electrode measurement technique.
a. Distillation
The steam distillation of fluoride as fluosilicic acid from strong
acid solution is one of the most widely employed separation techniques.
Its use for purifying fluoride samples prior to volumetric measurement
(41)
was reported by Willard and Winter in 1933. Their procedure, which
is still widely employed, involves a distillation from perchloric acid
at 135°C. Steam is introduced into the heated distilling flask to help
strip out fluoride and to keep the pot volume and temperature constant
throughout the distillation. Later variations of this basic approach
(42) (43)
include the use of other acids such as sulfuric and phosphoric
as well as the drop-wise addition of water in place of the introduction
(44)
of steam.^ Despite the variations, all of the distillations are
carried out at a constant, relatively low temperature so that the carry-
over of volatile materials which could interfere in subsequent volumetric
or colorimetric measurement is minimized.
In the presence of large amounts of materials such as aluminum and
silicon which form soluble fluoride complexes in solution, the vapor
pressure of fluoride is markedly reduced and a long distillation time is
required for complete fluoride evolution. Since the vapor pressure in-
creases with temperature, a double distillation is usually employed when
large amounts of these interferences are present. The sample is first
distilled from sulfuric acid at 165°C to isolate the fluoride from the
aluminum. The sulfuric distillate is then redistilled from perchloric
59
Arthur D Little, Inc
-------
acid at 135°C to separate the fluoride from the traces of sulfuric acid
and other volatile materials that had been carried over during the sul-
furic acid distillation.
Despite the tedium of the constant temperature distillation, the
basic Willard-Winter approach is still widely used and has become a
standard of comparison by which other separation techniques are judged.
It is specified in the ASTM Method for Inorganic Fluoride in the
Atmosphere (D 1606-60). The distillation can be applied to widely
ranging amounts of fluoride in a sample. The distillation as described
in the ASTM method can handle samples containing 0.005-10 mg fluoride in
a solution volume of 50 ml or less. Cross contamination from high
fluoride samples to low ones is a distinct problem, and separate stills
are usually reserved for samples containing very small amounts of
fluoride. A second reason for the continued use of the Willard-Winter
distillation is due to the fact that titration with thorium nitrate
continues to be a "reference" measurement method. That titration demands
an interference-free solution that generally requires a distillation from
perchloric acid.
(45)
In an attempt to develop a simpler approach to distillation, Bellack
studied the distillation of fluoride from sulfuric acid and concluded that
a direct distillation without addition of either water or steam could
work effectively and required a great deal less operator attention. The
procedure involves the use of a 1-liter distillation pot which is first
charged with a mixture of 200 ml concentrated sulfuric acid and 400 ml
of water. The acid solution is then distilled until the pot temperature
reaches exactly 180°C. The collected distillate is discarded. This
"pre-distillation" serves to adjust the acid/water ratio in the pot and
also to strip out traces of fluoride that are normally found in the
sulfuric acid. After the distillation pot has cooled, the sample in
300 ml of solution is introduced into the pot and distillation of the
sample is carried out until the pot temperature again reaches 180°C.
60
Arthur D Little, Inc.
-------
The acid in the distillation pot can be reused for subsequent samples
until the metal ion concentration becomes so high that the evolution
of fluoride is retarded. The ASTM procedure for Determining Fluoride
Ion in Industrial Water and Industrial Waste Water (D 11179-68) incor-
porates what is essentially Bellack's method. The maximum and minimum
amounts of fluoride that can be distilled using this procedure have
not, to our knowledge, been fully explored. In his original work,
(45)
Bellack obtained good recoveries using 300 ml samples containing
fluoride concentrations ranging from 0.09-10.0 mg/&. At fluoride
concentrations in excess of about 3 mg/£, a small amount of fluoride has
been found to remain in the condenser at the end of the distillation.
When this is the case, it is recommended that the condenser be discon-
nected and flushed with 300-400 ml of water, and the washings added to
the distillate prior to measurement. This distillation procedure can
be used only if appropriate measurement techniques which can tolerate
a small amount of sulfate carry-over near the end of the distillation are
(45)
employed. Bellack successfully used the Zirconium-SPADNS colorimetric
method on distillates obtained by the direct distillation procedure.
b. Diffusion
Diffusion of fluoride from an acid solution to an alkaline solution or
solid alkaline absorbent at temperatures well below the boiling point of
the acid solution is a well established procedure. It has been utilized
primarily as a technique for separating microgram and sub-microgram
amounts of fluoride from very small samples in biochemically oriented
research. ' ' The technique is based on the early developments of
Conway; the acid solution containing the fluoride sample is placed
in one compartment of the diffusion cell and the alkaline receiving medium
is placed in a second compartment. The cell is then sealed and diffusion
is allowed to proceed. After diffusion is complete, the cell is opened
and the alkaline receiving medium is subjected to one of a variety of
fluoride measurement techniques.
61
Arthur D Little, Inc
-------
The diffusion is usually performed at temperatures ranging from
25-60°C. Over this temperature range, the fluoride vapor pressure is
very low, but this effect is counteracted by placing the alkaline and aci-
dic solutions in close proximity and employing diffusion times of up
to 24 hours or more. It has been reported that the addition of
siloxane derivatives can markedly speed the diffusion process; diffusion
times at 25°C of under 6 hours have been reported. Tusl found
that silicone grease had a similar accelerating effect.
The technique of diffusion can produce good fluoride separations,
but it has not been used extensively for the types of samples expected
from process effluent streams. Of major concern is the effect of
sizable amounts of aluminum or iron on the vapor pressure and consequently
the diffusion rate of fluoride.
c. Ion Exchange
A variety of procedures for separating fluoride from cationic and/or
anionic interferences by the use of ion exchange have been reported. Anion
exchange resins in the dydroxide or acetate form have been most widely
employed. Fluoride can be trapped on Amberlite IRA-400 and subsequently
eluted with 10% sodium chloride solution. ' Ziphin, et al^ ' have
performed a gradient elution with sodium hydroxide to separate fluoride
from phosphate. The separation of fluoride from iron, aluminum, phos-
phate, and sulfate by stepwise elution from an ion exchanger with sodium
acetate has also been reported.
While ion exchange has been used successfully and is an attractively
simple separation approach, it should be used cautiously on ill-charac-
terized field samples. The separation itself depends upon the formation
of a resin-fluoride complex. The presence of unexpected interferences
in solution which could also form fluoride complexes could change the
fluoride distribution coefficient between the stationary and mobile
phases and lead to unexpected erroneous results.
62
Arthur D Little, Inc.
-------
4. Current Industrial Practice Analytical methods for fluorides
that are currently being used in the aluminum, iron and steel, glass and
ceramic, and phosphate rock industries were surveyed by means of personal
communication with analytical chemists and environmental control engineers
at selected companies within each industry. With very few exceptions,
the same general methodology and practice is found in all four industries,
so a single comprehensive discussion of our findings will be presented.
Analytical methods presently in use vary to some extent from company to
company and even between the various plant locations of a single company.
In general, the Willard-Winter perchloric acid distillation followed
by titration with thorium nitrate is the "accepted" procedure and alternate
methods are evaluated by comparison to it. With regard to the acceptabil-
ity of newer techniques such as the fluoride specific ion electrode and
various colorimetric methods, one encounters opinions ranging from out-
right rejection to acceptance as valid. The tendency to reject the newer,
simpler methods seems to be related somewhat more to the sense of achieve-
ment that a given analyst experiences when he does a good thorium nitrate
titration than it is to the results of comparative collaborative tests.
For many industrial laboratories, fluoride analysis for emission control
purposes is relatively new and represents only a small portion of the overall
analytical load. These laboratories were previously performing fluoride
analyses for the purposes of process control or product specification many
years ago when the thorium titration was the only available reliable
technique. Consequently the titrimetric method was written into a
great many process control and customer acceptance analysis procedures,
and once so written, there is a great deal of inertia which prevents
change.
For industries and processes where particulates are collected as a
separate fraction, the normal procedure involves ashing the filter paper,
fusing the solids, distilling from perchloric acid, and finally titrating
with thorium nitrate. One exception is the phosphate rock industry,
where solids are acid soluble and the fusion step is unnecessary.
Sodium hydroxide seems to be the preferred flux for fusing samples.
63
Arthur D Little, Inc.
-------
Perchloric acid in the distillation is retained because of its low
contribution to the reagent blank and also because a final distillation
from perchloric acid at 135°C is necessary for a successful thorium
titration. Although the constant-temperature Willard-Winter distilla-
tion can be tedious when performed only occasionally be an inexperienced
technician, it becomes more or less routine when the same technician
performs a sizable number every day. Operating conditions become
optimized to the point where the distillation flask is' charged, set
on the heater, the steam line is connected, and the distillation pro-
ceeds virtually unattended at the correct temperature. Similar consi-
derations apply to the thorium nitrate/Alizarin Red-S titration with its
"hard to see" endpoint. After much experience with the technique,
the faint color change is easily perceived. In one laboratory, for exam-
ple, a certain single analyst does all of the "important" thorium nitrate
titrations; he has been doing so for nearly 20 years. In one laboratory
visited, the Zirconium-SPADNS colorimetric method has been used to analyze
modified Willard-Winter distillates for some time and has been found to
be completely satisfactory. We encountered no one who was using the
fluoride specific electrode to measure distillates.
(54)
One major company has found the Technicon Auto-Analyzer procedure
to be an ideal compromise between speed, simplicity, and accuracy. In
the Technicon procedure, a continuous micro-distillation from sulfuric
acid is performed and the distillate is measured colorimetrically with
the Lanthanum-Alizarin complex one reagent. A second laboratory tried
the technique and discontinued it; because their sample load was rather
low, the ratio of analyses performed to required maintenance time was
poor. Since the technique has a limited dynamic range, they found that
it was frequently necessary to rerun samples using a new aliquot size
when analyses were being performed on samples having unknown fluoride
concentrations. It was suggested that a fluoride ion electrode measure-
ment in conjunction with the Auto-Analyzer might result in an automated
method with better dynamic range.
64
Arthur D Little, Inc
-------
The direct fluoride ion electrode measurement after buffering seems
to be generally accepted and is employed for measuring fluoride in impin-
ger solutions collected after a particulate filter. In the phosphate
rock industry, where wet streams demand that gases and particulates
be collected together in an impinger, direct electrode measurements are
made in the slurry after buffering. Initially, all process effluent
samples collected by one company were run by both this technique and
the conventional Willard-Winter distillation/thorium titration. Reasonable
agreement between results was found, and now only an occasional sample is
distilled as a check and in the event that elevated levels of fluoride
are found in the effluent. In that case, the sample is submitted for
a confirmatory analysis by distillation and titration.
Attention to "good analytical practice" is necessary for reliable
fluoride measurements, especially at low concentrations. Collaborative
testing using real samples is done continually within given industries.
This procedure serves as a quality control check on the ability of a
given laboratory to run "the accepted method," and also provides a means
for evaluating and comparing potentially useful new methods.
It is very important to exercise caution to avoid external and cross
contamination of samples. Some laboratories distill their own water
and filter the laboratory air through charcoal and lime. Seemingly
innocuous occurrences such as the use of a holocarbon-charged aerosol
can in the laboratory can result in sample contamination. The use of
completely separate distillation flasks and muffle furnaces for samples
of high and low concentrations is strongly suggested. Finally, it is
absolutely necessary that alkaline conditions be maintained when flouride-
containing samples are ashed or evaporated so that there is no fluoride
loss through volatilization.
5. Comparison and Summary of Candidate Techniques
a. Measurement Procedures
Of the available fluoride measurement techniques, the fluoride
65
Arthur D Little, Inc
-------
electrode appears to be the clear choice by virtually all criteria. Its
wide dynamic range of about 10 , as compared to a maximum of about 100
for colorimetric techniques and 200 for volumetric (titrimetric) tech-
niques , makes it well suited for measuring the wide range of fluoride
concentrations which can be encountered. Titrimetric procedures can
produce the best precision and accuracy; at the higher fluoride levels
within the range to which titrimetry can be applied, precision and accuracy
of better than + 1% is possible, but can become + 5% or more when per-
formed by an unskilled analyst. In comparison, colorimetric and specific
ion electrode measurements are typically good to about + 2-5% when per-
formed by a relatively unskilled technician. In any event, if one consi-
ders the magnitude of other errors which could be encountered in the over-
all sampling and analysis procedure, a precision and accuracy of better
than +_ 2-5% in the measurement step is probably not warranted.
Because of its freedom from anionic interferences, the fluoride ion
electrode is particularly well suited to the direct measurement of gaseous
fluorides collected in impingers. Interferences in colorimetric and
volumetric procedures resulting from oxides of sulfur and nitrogen as well
as other gases which are frequently encountered in emission samples
generally require a separation step prior to measurement. Interferences
due to cations usually manifest themselves by complexing fluoride in the
sample solution being measured, and consequently, affect all measurement
techniques in more or less the same way.
There is very little difference in the actual time required to perform
a measurement by either of the three types of techniques. Both the colori-
metric and electrode methods are equally simple to perform and require
little operator skill. On the other hand, the titration definitely
requires a dedicated and skilled technician. While the visual titrimetric
procedure requires less expensive laboratory apparatus, this factor alone
cannot make up for its limitations.
66
Arthur D Little, Inc.
-------
b. Solubilization
Fusion in sodium hydroxide or sodium carbonate seems to be the only
reliable way to handle a variety of unknown particulate materials. Pyro-
hydrolysis would probably work as well, but the technique seems needlessly
complex. However, if pyrohydrolysis were to be run more or less continu-
ously on a high volume of samples, it would probably be faster than the
equivalent fusion/distillation combination. Other means of dissolving
samples exist, but they are not as universally applicable to ill-charac-
terized materials as is the alkaline fusion technique.
c. Separation of Fluoride
Removal of interfering species by distillation appears to be the
method of choice. While ion exchange techniques are simple, the possibil-
ity exists that unexpected species in solution might disturb the
sorbtion behavior of fluoride ion; incomplete separation and/or non-
quantitative recovery could result. The choice between distillation
and diffusion is less clear. The fundamental principles involved
are so similar that both should work. However, there has been very
little experience with diffusion separations of samples of the type
expected in this application—relatively large volumes of solution
containing significant amounts of aluminum and iron.
There is no evidence that sulfuric acid is any less capable of pro-
ducing quantitative fluoride recoveries than any of the other acids
commonly used. In fact, with perchloric acid at 135°C, the reverse is
sometimes true. When significant amounts of aluminum are present in the
sample, a pre-distillation from sulfuric acid must be performed prior
to the perchloric acid distillation. On the assumption that the fluoride
specific ion electrode, with its freedom from sulfate interference, is
used for the measurement step, a single distillation .from sulfuric acid
should suffice. The fact that the fluoride electrode is generally free
from anionic interferences (most of the volatile interference that
distill over hydrolyze to anions) should also relax the restriction on
67
Arthur D Little, Inc
-------
precise control of distillation temperatures; this, in turn, should result
in less operator effort if not in overall distillation time.
We believe that the base separation method is a distillation from
sulfuric acid. This could take the form of a direct distillation simi-
lar to the one described in ASTM Method D1179-68, or it could be a
distillation with steam or drop-wise water addition with only minimal
temperature control. The relative merits of the two approaches have been
compared and will be discussed in the section on laboratory studies which
follows.
D. Laboratory Evaluation of Candidate Techniques
1. Evaluations of the Fluoride Specific Ion Electrode as a
Measurement Technique
Studies were preformed to evaluate the effect of interfering
species such as iron and aluminum on the fluoride electrode measurement.
Its performance was compared with the zirconium-SPADNS colorimetric method
using a set of simulated field samples. In addition, a set of actual
phosphate rock industry samples were analyzed in two different laboratories
using the fluoride electrode to assess its performance in a practical
situation.
a. Apparatus and Procedure Fluoride measurements were made with an
Orion fluoride electrode (Model 94-09) used in conjunction with silver/
silver chloride, 4M potassium chloride reference electrode. In the initial
phase of our study, the EMF between the fluoride electrode and the reference
electrode was measured with a Beckman Expandomatic pH meter. Later in the
program we switched to an Orion Model 800 digital pH meter for all suc-
ceeding measurements.
Fluoride standards were prepared by diluting Orion standard fluroide
(0.1M) solution. A calibration curve was prepared by plotting EMF as a
function of logarithm of concentration using seven known solutions span-
—2 —6
ning the concentration range from 10 - 10 moles/liter fluoride.
68
Arthur D Little, Inc
-------
Measurements of unknowns were performed by reading the electrode EMF and
converting to concentration by referring to the calibration curve.
-4
When measuring fluoride concentrations less than- 10 M, a slight
drift on the order of 1-2 millivolts was observed during the first few
minutes after immersing the electrode. When working at these levels,
measurements were made at a fixed time interval of five minutes after
immersion to minimize the effect of drift.
The electrode measurement was performed on a final solution volume of
50 ml. A 25 ml sample of the solution to be measured was taken, bromthymol
blue indicator was added, and the pH adjusted, if necessary, to be within
the range of 6.6-7.1, the range where the indicator is green in color.
Sodium citrate solution was then added (10 ml of 2.5 M or 5 ml of 1.0 M
for final citrate concentrations of 0.5 M and 0.1 M, respectively),
and the resulting solution diluted to 50 ml.
Measurements using the zirconium-SPADNS method followed the proce-
dures described in ASTM Method D1179-68. Details of the distillation
procedures are found in Part 3 of this Section.
b. Results and Discussion
1) Effect of Interfering Species To evaluate the potential use of
the fluoride specific ion electrode the direct measurement of fluoride-
containing solutions without prior separation of interferences, a rela-
tively high level of sodium citrate (0.5 M) was utilized. In addition
to complexing interfering cations, this high concentration of sodium
citrate tends to produce a solution of essentially constant ionic strength
by swamping out the effects of small differences in salt content of the
unknown sample.
Two of the most important interfering cations for the industries of
concern are aluminum and iron. Measurement of known concentrations of
fluoride in solutions containing varying excesses of aluminum are shown in
Table 3A. Accurate measurement is clearly related to both the concentra-
tion of aluminum (more precisely the aluminum/citrate ratio) and the
69
Arthur D Little, Inc
-------
Table 3A
Fluoride
Added(a'b)
1.0 x 10
1.0 x 10
-2
-2
1.0 x 10
1.0 x 10
-3
-3
1.0 x 10
1.0 x 10
1.0 x 10
1.0 x 10
1.0 x 10
1.0 x 10
1.0 x 10
1.0 x 10
1.0 x 10
-4
r4
-4
-4
r4
-4
-4
-4
Effect of Aluminum (III) on Fluoride Electrode Measurement
Aluminum
Added(a»c)
5.0 x 10~2
1.0 x 10"1
5.0 x 10~3
1.0 x 10~2
1.0 x 10~4
1.0 x 10~4
2.0 x 10~4
2.0 x 10~4
5.0 x 10~4
1.0 x 10~3
2.2 x 10~3
7.2 x 10~3
1.2 x 10~2
in 0.5 Molar Citrate Buffer
Mole Ratio
Al+V
5
10
5
10
1
1
2
2
5
10
22
72
120
Fluoride
Found (a)
5.5 x 10~4
1.2 x 10~4
9.6 x 10~4
7.2 x 10~4
9.8 x 10~5
1.0 x 10~4
1.0 x 10~4
9.9 x 10~5
1.0 x 10~4
1.0 x 10~4
9.4 x 10~5
7.0 x 10~5
6.6 x 10~5
Percent
Found
6
1
100
72
98
100
100
99
100
100
94
70
66
(a) Expressed as moles/liter.
(b) Added as sodium fluoride.
(c) Added as aluminum sulfate.
-------
aluminum/fluoride ratio. For fluoride at the 10 molar level, accurate
measurements were possible in the presence of a ten-fold excess of
aluminum, while the measurement was only 5% low with a twenty-fold
_3
excess. At the 10 M fluoride level, a ten-fold excess of aluminum
produced a fluoride measurement that was low by over 25%; but with a
five-fold excess, an accurate measurement possible.
The absolute amount of aluminum that can be tolerated decreases as
the amount of fluoride to be measured decreases. However, it does not
_3
appear to decrease as rapidly. Thus at the 10 M fluoride level,
-3 -4 -3
5 x 10 M aluminum can be tolerated; at 10 , at least 1 x 10 M
aluminum does not interfere. This effect can also be seen by comparing
-2 -3
the effects of 1.0x 10 M and 7.2 x 10 aluminum at fluoride concentra-
-3 -4
tions of 10 and 10 M, respectively. In each case the amount of
fluoride measured was about 30% low. Consequently, it appears that for
solutions containing large amounts of fluoride, the tolerance for
aluminum can be improved by performing a dilution of the sample.
Iron also interferes in the direct electrode measurement, but to a
significantly less extent than does aluminum. The results of a number of
measurements to evaluate the interference due to iron are shown in Table 4.
-4
At the 10 M fluoride level, a 100-fold excess of iron does not introduce
significant error into the measurement.
Anions such as silicate, sulfate, and phosphate, which might be
expected in field samples, were studied; their effects on the direct
electrode measurement are shown in Table 5. Silicate at the 1 M level
-4
has no measurable effect on the measurement of 10 M fluoride. Large
excesses of phosphate and sulfate could also be tolerated. In these mea-
surements the citrate was not acting as a complexant; it only served to
adjust ionic strength. As a result, the citrate level could be reduced
to 0.1 M without significantly affecting the insensitivity of the elec-
trode to these anionic interferences.
71
Arthur D Little, Inc
-------
Table 4
Effect of Iron (III) on Fluoride Measurement
in 0.5 Molar Citrate Buffer
Fluoride
Added(a'b)
Added
Iron
(a,c)
Mole Ratio
Fe+3/F-
Fluoride
Found(a)
Percent
Found
1.0 x 10
-4
1.0 x 10"
1.0 x 10
1.0 x 10
1.0 x 10
-4
I
-4
-4
1.0 x 10
1.0 x 10
-4
-4
3.0 x 10
5.0 x 10~4
1.0 x 10~3
1.0 x 10~2
1.0 x 10"1
3
5
10
100
1000
0.99 x 10
1.02 x 10
1.03 x 10
1.03 x 10
1.02 x 10
0.99 x 10
0.97 x 10
0.76 x 10
-4
-4
I
-4
I
r4
-4
-4
-4
-4
99
102
103
103
102
99
97
76
(a) Expressed in moles/liter.
(b) Added as sodium fluoride.
(c) Fe added as
12'H20
72
Arthur D Little, Inc.
-------
Table 5
Effect of Other Potential Interferences on
Fluoride Electrode Measurements Using Citrate Buffer
Citrate
Level(a)
Interference
(c)
Fluoride
Added(a'b)
Mole Ratio
Interference/F
Fluoride
Found(a)
Percent
Found
0.5
Silicate
1.0 x 10
-4
1
10
1.0 x 10
1.0 x 10
-4
I
-4
100
100
0.1
Silicate
1.0 x 10
-4
100
1000
10,000
9.8 x 10
-5
1.03 x 10
1.03 x 10
-4
i
-4
98
103
103
0.1
0.1
Sulfate
Phosphate
1.0 x 10
1.0 x 10
-4
1000
1000
9.3 x 10
1.0 x 10
-5
I
-4
93
100
(a) Expressed as moles/liter
(b) Added as sodium fluoride
(c) Interferences added as sodium silicate, sodium sulfate, and
disodium hydrogen phosphate.
73
Arthur D Little, Inc
-------
Table 6
Comparison of Fluoride Electrode and Zr-SPADNS Measurement Methods
(a)
Percent Fluoride Recovery
Distillation
Sulfuric
(to 210°C)
Perchloric
(at 135°C)
Species Moles F Electrode
NaF 1.0 x 10~5 101
1.0 x 10~4 98
1.0 x 10~3 97
CaF (b) 1.1 x 10"4 88
2
CaF (c) 1.1 x 10"4 86
Cryolite 2.1 x 10"4 86
-4
Phosphate Rock 1.1 x 10 100
Zr-SPADNS
138
106
97
86
83
81
100
(a) Comparative analyses performed on aliquots of the same
sample of distillate.
(b) Sample also contained 1.0 g PbO.
(c) Sample also contained 1.0 g
74
Arthur D Little, Inc
-------
2. Comparison of the Fluoride Electrode with the Zirconium-SPADNS
Method
Distillations of a variety of samples were performed, and the collec-
ted distillates were split and measured by the fluoride electrode
as well as by the zirconium-SPADNS method. A comparison of the results
obtained is shown in Table 6. For sulfuric acid distillates, agreement
_3
is excellent at 10 M fluoride. However, at lower fluoride levels,
the Zr-SPADNS results become progressively higher probably due to the
positive interference of small amounts of sulfuric acid carried over
in the distillate at high temperatures. The sulfuric acid carried over
has no effect on the electrode measurement even at the 10 M fluoride
level (see Table 5).
With the perchloric acid distillates, there is good general agreement.
Recoveries, as determined by both methods, are low for the calcium
fluoride and cryolite samples. It is generally known that a single dis-
tillation from perchloric acid at 135°C will not yield quantitative
fluoride recovery from cryolite; a pre-distillation from H-SO, at 165°C
is required to effect the separation of fluoride from aluminum. To stimu-
+2 -t-T
late field samples, Pb and Fe J were added to the calcium fluoride
samples. The presence of large excesses of these metals probably caused
a low fluoride recovery for calcium fluoride during the distillation step.
Therefore, the low recovery appears to be due to incomplete fluoride
evolution from distillations and is probably not related to the measure-
ment procedures per se.
c. Inter-Laboratory Comparison of the Fluoride Electrode
To provide a realistic evaluation of the possibility of making fluo-
ride measurements via the fluoride electrode directly on a field sample,
a series of samples taken at diammonium phosphate and phosphoric acid
plants were split and were independently analyzed in this laboratory
and in the laboratories of EPA. With only two exceptions the results,
shown in Table 7, are in excellent agreement. Samples C-l and C-2 were
75
Arthur D Little, Inc
-------
TABLE 7
TOTAL SOLUBLE FLUORIDES IN EPA SAMPLES BY DIRECT ELECTRODE MEASUREMENT
Sample Designation
165PF + 166PF + 16
168PF + 230PF, D-l
172PF + 232PF, F-l
187PF + 238PF, C-2
189PF + 234PF, D-
191PF + 192PF + 1
270PF + 271PF + 2
273PF + 339PF, K-
275PF + 342PF, M-2
277PF + 341PF, P-2
F Std., 10 yg/ml, #11
.on
167PF, C-l*
)-l
'-1
:-2*
)-2
193PF, E-2
272PF, R-2
:-2
1-2
'-2
L, #11**
ADL
moles /I
S.OxlO"2
1.62xlO~3
3. 3x10" 3
4. 5xlO~ 2
7.4xlO~4
1.44xlO~3
2.7xlO~3
5.2xlO~3
3.9xlO~4
1.40xlO~4
U
-4
5.3x10
Results
yg/ml
950
31
63
860
14
27
51
99
7.4
2.8
10.1
EPA Results
ye /ml
960
29.5
62
944
13.9
27
52.4
98
7.2
2.7
10.0
Remarks
strongly acidic
strongly acidic
very alkaline
moderately acid
very alkaline
c
Sample stability in doubt, see text.
** Supplied as known standard by EPA, concentration derived from ADL
calibration curve.
-------
strongly acidic and it is not unlikely that fluoride was lost from these
two samples during transit, storage, and handling.
d. Conclusions
The results of these studies support the earlier conclusions from our
literature survey that the fluoride electrode is capable of producing
precise and accurate results quickly and easily without the need for
highly trained or experienced personnel. If a complexing buffer is used,
significant excesses of aluminum and iron in solution can be tolerated.
If intolerably high amounts of interfering cations are present, a separa-
tion is necessary, but after a separation has been performed, the fluoride
electrode appears to be as good as or better than other techniques for
fluoride ion measurement.
2) Studies of Techniques for Fusing Particulates Experiments to
compare a variety of fusion fluxes were performed using cryolite (3NaF«AlF )
as a model compound because it is one of the more intractable fluoride-
containing compounds which one might expect to encounter in field samples.
Fluxes evaluated included the following:
• mixed carbonates (sodium/potassium eutectic)
• sodium hydroxide
• sodium hydroxide/borax
• sodium carbonate/borax
• mixed carbonate/borax
The admixtures of borax (sodium borate) were tested since borax has been
reported by some industry sources to aid in the dissolution of refractory
fluoroaluminates.
Mixed carbonates and sodium hydroxide were equally effective for fusing
cryolite. The mixed carbonates were found to be easier to use than sodium
hydroxide.
a. Apparatus and Procedure All fusions were carried out in covered,
50-ml nickel crucibles over a Fisher burner operating on natural gas/air.
77
Arthur D Little, Inc
-------
The crucibles were cleaned between each test by fusing fresh portions of
mixed carbonate flux to ensure that no residual fluoride remained in the
crucible. After cooling, the melt was removed by rapping the inverted
crucible on a hard surface, and the crucible was washed carefully with
distilled water.
After the crucible had been cleaned, a portion of the requisite amount
of flux to be tested was first placed in the cleaned crucible and the
sample was added, followed by the remainder of the flux. The solids were
then mixed, the crucible covered, and the fusion carried out at medium red
heat, approximately 700-750°C. The usual fusion time was 10 minutes,
but some 30-minute tests were performed to determine if a longer fusion
was necessary; fusions were timed from the point when all of the flux
had melted. After fusion, the crucibles were cooled to room temperature
before treating the fusate further.
About half of the fluoride measurements for the fusion tests were
performed with the specific ion electrode directly on solutions of the
fusate containing 0.5 M citrate. The amounts of each flux were chosen so
that when the cooled melt was dissolved in water, neutralized with citric
acid to a pH of ca.7 as indicated by bromthymol blue, and diluted to 50 ml,
the final citrate concentration would be 0.5 M.
The remaining fusate samples were distilled from sulfuric acid and
the distillate measured with the fluoride electrode after pH adjustment
and addition of sodium citrate. Complete details of this procedure is
included in Part 3 of this Section.
b. Results and Discussion The compositions of the five fluxes used
in these experiments are shown in Table 8. Fluoride recoveries are shown
in Table 9. Initially, fusions were performed using a cryolite sample
(cryolite A) which, in retrospect, appears to have been impure. The
recoveries in Runs 3, 4, and 8-12, using three different fluxes all range
from 84-90% with an average recovery of 87%. Subsequent to these measure-
ments, a new sample of cryolite (cryolite B) was obtained. It was fused
78
Arthur D Little, Inc.
-------
TABLE 8
Description of Flux Compositions
Code
Type
Chemical
Weight(g)
Sodium Carbonate/Borax
3.4
1.1
B Mixed Carbonate/Borax Na-CO.,
1.5
1.9
1.1
Sodium Hydroxide/Borax NaOH
2.8
0.9
D Mixed Carbonate
2.0
2.5
Sodium Hydroxide
NaOH
3.0
79
Arthur D Little, Inc
-------
TABLE 9
Fluoride Recoveries from Test Fusions of Cryolite
Fusion ,, . . . Fluoride , . Percent
Run No. Time (min) Flux ' Cryolite ' Added (moles)^C) Recovery Average
1
2
3
4
5
6
7
8*
9*
10
11*
12*
13*
14*
15*
10
10
10
30
10
30
30
10
10
30
10
10
10
10
10
A
A
B
B
C
C
D
D
D
E
E
E
D
D
E
A
A
A
A
A
A
A
A
A
A
A
A
B
B
B
_5
2.4 x 10 ,
2.0 x 10
-4
2.2 x 10 ,
2.2 x 10
-4
2.0 x 10 ,
2.1 x 10
-4
2.1 x 10 7
— a
1.9 x 10 7.
— X
2.1 x 10
-4
2.1 x 10 .
1.8 x 10",
—3
2.0 x 10
-4
2.0 x 10 ,
2.0 x 10
-4
2.1 x 10
87
74
88
88
77
70
100
87
89
84
86
90
99
100
97
81
88
74
92
86
100
97
(a) - Direct electrode measurement on dissolved fusate expect for runs
indicated (*) where the fusate was distilled from l^SO^. prior to
electrode measurement.
(b) - Description of fluxes in Table 8.
(c) - Assuming Cryolite was 100% 3 NaF • A1F».
(d) - See text
80
Arthur D Little, Inc.
-------
FIGURE 12 APPARATUS FOR PERFORMING WATER ADDITION DISTILLATIONS
81
Arthur D Little, Inc
-------
with both sodium hydroxide and mixed caroonates (Runs 13-15) and very
nearly 100% recovery was obtained. We are, therefore, quite certain that
the first cryolite was probably about 87% pure; the single 100% recovery
for this material in Run 7 is most likely in error.
An examination of the results in Table 9 shows that sodium hydroxide,
mixed carbonates, and mixed carbonate/borax yield equivalent recoveries.
The inclusion of borax in the flux did not increase fluoride recovery
from cryolite. In fact, with the sodium hydroxide/borax flux (Runs 5 and
6), the opposite may be true. The reason for lower recovery in those
two runs is unclear. However, it was observed that this flux corroded
the nickel crucibles very.rapidly, and loss of fluoride could have con-
ceivably accompanied corrosion. In the three cases examined, there seems
to be no significant increase in fluoride recovery when the time of fusion
is extended from 10 min. to 30 min.
One can conclude that the mixed carbonate flux and the sodium hydrox-
ide flux are equally capable of effecting complete fusion. We have
found in the course of these experiments, however, that the mixed carbonate
flux is more convenient to use since the fusion proceeds at a lower temp-
erature. The mixed carbonates have less of a tendency to creep up the
sides of the crucible and they generally contain less water. The latter
point is significant, because water trapped in the flux is evolved with
sputtering when the fusion is initiated and may lead to mechanical loss of
the sample.
3) Evaluation of Distillation Procedures
a. Introduction Two distillation procedures using sulfuric acid
were studied as potential alternates to the tedious Willard-Winter
perchloric acid distillation. The first approach involved the dropwise
addition of water during the H^SO, distillation to hold the pot tempera-
ture more or less constant. This "water-addition distillation" is similar
to the constant temperature distillation for sulfuric acid at 165°C
that is widely used prior to a perchloric acid distillation when one is
dealing with "hard to distill" samples.
82
Arthur D Little, Inc
-------
The second approach, termed a "direct distillation," involved the
addition of the sample to a sulfuric acid/water solution in the still pot,
applying heat, and allowing the distillation to proceed with no further
additions to the pot. This latter approach is essentially the one recom-
mended in the ASTM Method for Fluoride Ion in Industrial Water (D1179-68).
Tests on a variety of samples representative of those expected to be
found in process effluent samples showed that both of the sulfuric distil-
lations could be operated to produce essentially quantitative recovery.
In general, recoveries from the sulfuric acid distillations were as good
as, and sometimes better than, the Willard-Winter perchloric acid distil-
lation. Because it is the simpler procedure and produces a lower reagent
blank, the direct distillation is recommended over the approach involving
continuous water addition.
b. Apparatus and Procedure The apparatus used for the constant
temperature water addition distillations (from both sulfuric acid and
perchloric acid) is shown in Figure 12. It consists of a 300-ml distil-
lation flask equipped with an addition funnel, a thermometer, and a
Kjeldahl-type spray trap leading to a straight condenser mounted nearly
vertical. The distillation flask^was heated by a heating mantle and the
spray trap and connections to the condenser were wrapped with electric
heating tape and insulation to prevent condensation. In practice, the
sample and some 4 mm diameter soft glass beads were added to the
flask, the system was closed, and the sulfuric acid added from the
addition funnel. After heat had been applied and the temperature had
reached the desired value, water was added from the addition funnel at
a rate which held the temperature at the desired value. For collection
of the distillate, the condenser outlet was kept just beneath the surface
of dilute (approximately 0.05 m) sodium hydroxide solution to which
phenolphtalein had been added. Thus, the fluorides were trapped in an
alkaline medium, and the indicator showed if excess acid had distilled
over and lowered the pH.
83
Arthur D Little, Inc.
-------
It was decided to measure the distilled fluorides via the specific ion
electrode procedure using 0.1 M citrate buffer to compensate for any pH
and ionic strength variations which could have resulted from acid
carry-over. Thus, after the distillation was complete, the requisite
volume of 1.0 molar sodium citrate was added along with a few drops of
bromthymol blue indicator. The solution was neutralized to the green
of the indicator with 6 M hydrochloric acid and was then diluted to a
known volume with water. The fluoride content was measured using the
fluoride electrode and concentrations were determined from a calibration
curve prepared from sodium fluoride in the same buffer.
The apparatus used for the initial direct distillation (without water
addition) experiments was the same apparatus that had been used for the
preceding distillations with water addition. The samples, whether fused or
just mixed with the carbonate flux material, were transferred to the
flask, together with a total of 60 ml of distilled water and a few soft
glass beads. The flask was then coupled to the distillation train,
and a total of 70 ml concentrated sulfuric acid was added carefully via
the addition funnel. This water/acid ratio was selected because its
initial boiling point was around 155°C. The solution temperature generally
rose to approximately 140°C during acid addition. The solution was
then heated with the heating mantle at constant power, and boiling general-
ly commenced around 155-160°C.
The distillate was collected in 0.05 molar sodium hydroxide contain-
ing phenolphthalein indicator, and extra hydroxide was added as required
to keep the solution alkaline. The distillate was collected in two 25 ml
fractions. At the end of the second fraction, the pot temperature had
reached 240°C, and excessive amounts of acid began to appear in the dis-
tillate as indicated by the need to add increasing amounts of sodium
hydroxide to keep the distillate alkaline.
To study further the effect of distillate volume collected on
fluoride recovery, a series of distillations were performed using the same
84
Arthur D Little, Inc
-------
apparatus, but the volumes of water and acid were increased by 50% to 90
and 105 ml, respectively. The distillate volume collected was increased
proportionately; a total of 75 ml of distillate was collected in three
equal fractions. Pot temperatures at the end of each 25 ml fraction were
approximately 175°C, 210°C and 240°C. Relatively high heating mantle
voltages (approximately 100 volts) were used; the fractions came off in
about 20 minutes each.
The distillation incorporated in ASTM Method D1179-68 is also a direct
distillation but involves a further scale-up of pot volume and amount of
distillate collected. Because it could have a potentially lower fluoride
blank and could be somewhat simpler to use than the previous direct distil-
lation procedure, the ASTM method was evaluated as written and also with
modification to go to a higher final pot temperature.
The apparatus used for this direct distillation was essentially the
same as described in ASTM D1179-68. It was quite similar to the apparatus
shown in Figure 12 except that a larger, one-liter still pot was used. A
Graham condenser was substituted for the straight condenser, and the addi-
tion funnel was omitted.
The procedure for the ASTM-type direct distillation involved first
charging the pot with a mixture of 200 ml concentrated sulfuric acid and
400 ml of water. The acid solution was then distilled until the pot tem-
perature reached a predetermined temperature (180°C in the ASTM method;
210°C in our final method). Due to the higher efficiency of the Graham
condenser, an alkaline receiving solution was not used, and the tip of
the condenser was not immersed beneath the surface of the liquid in the
receiver.
After the initial distillation of the acid solution, the pot was
cooled to under 40°C and the sample, together with 300 ml of distilled
water, was added slowly with careful mixing. Heat was then applied, and
the sample distilled until the pot reached the same maximum temperature
as used during the preceding cleanup distillation.
used during the preceding cleanup distillation.
85
Arthur D Little, Inc.
-------
Preparation of the collected distillate for measurement using the fluoride
electrode was done in the same way as described above for the water addition
distillations. However, as an alkaline collecting solution was not used in
this procedure, the collected distillate was weakly acidic, and the citrate
buffer alone was usually capable of bringing the solution to the proper pH
for measurement.
In the ASTM-type direct distillation procedure, the acid in the distil-
ling pot can be reused for subsequent samples until the metal content becomes
high enough to significantly retard the evolution of fluoride. In this work
the pot was recharged after every third sample.
c. Results and Discussion
1) Water Addition Distillation The results of a series of water
addition distillations from sulfuric acid performed at several temperatures
on a variety of sample types are summarized in Table 10. In order to better
understand the evolution of fluoride as the distillation proceeded, the dis-
tillate was collected in separate, successive, 50-ml fractions and the fluo-
ride in each was determined.
When aluminum is present in the sample to be distilled, it tends to
form soluble complexes with fluoride and thus reduces the fluoride vapor
pressure at a given temperature. The slower evolution of fluoride in the
presence of aluminum can be seen by comparing Runs 3 and 4 with the distil-
lation of a phosphate rock sample in Run 9. Evolution of fluoride from the
phosphate rock was virtually complete at a temperature somewhat less than
165°C after the first 50-ml fraction of distillate had been collected, but
4-6% of'the fluoride from the cryolite sample was found in the second 50-ml
fraction.
Increasing the distillation temperature markedly increased the rate
of fluoride evolution from a giveiv sample type. This effect can be observed
by comparing Runs 1 and 2 with 'Runs 3 and 4. At the lower temperature, 8
to 10% of the fluoride was found in the third fraction, while at 165°C
evolution was virtually complete in the first 100 ml. Although there is a
large difference in the amount of added fluoride in Runs 1 and 2 as compared
to 3 and 4, we believe it is temperature rather than sample size that affects
distillation recoveries. Runs 5 and 6, which started with only lOOy moles
of fluoride yields results more like those of 1 and 2 than like 3 and 4.
86
Arthur D Little, Inc
-------
TABLE 10
Run
1
2
3
4
5
00
6
7
8
9
>
rt-
c
a
r-
o
Water-Addition Distillation of Fused Solid Samples
— — — Fluoride Found in Distillate Fractions (v moles)
Fluoride Added
Material Temp (°C) (u moles) 1st 50 ml 2nd 50 ml 3rd 50 ml 4th 50 ml Total
Cryolite 130-135 1750 1260 (72%) 310 (18%) 140 (8%) NC[dJ 1710
130-135 1720 1050 (61%) 420 (24%) 180 (10%) NC 1650
165 201 190 (95%) 13 (6%) NC NC 201[c]
7nfi[c]
165 210 195 (93%) 9 (4%) NC NC 206
NaF + Al(III)[b] 140 100 73 (73%) 23 (23%) 13 (13%) NC 107 °
140 100 79 (79%) 18 (18%) 5 (5%) NC 10° °
19[c]
140 20 3.5 (18%) 15 (75%) 2.2 (11%) < 0.1
Blank 140 0 1.4 0.3 0.3 < 0.1 2-°
Phos. Rock 150-165 108 108 (100%) 2 (2%) < 1 < 1 108 [c]
[a] - Percentages in parentheses are the percentages of the amount added that was found in each distillate fraction.
[b] - Ten-fold mole excess of Al(III) added as aluminum salt.
[c] - Total recovery corrected for reagent blank of 2 \i moles.
[d] - NC - fraction not collected.
Percent
Recovery
98
96
100
96
107
100
95
—
100
-------
The effect of aluminum in depressing vapor pressure and increasing dis-
tillation time appears to be more or less independent of the aluminum/fluoride
ratio. If one compares Runs 1 and 2 with Runs 5 and 6, the profile of percent
fluoride distilled as a function of distillate volume collected appears quite
similar, even though in the cryolite samples the aluminum/fluoride ratio was
about 1:6, while in Runs 5 and 6 the aluminum/fluoride ratio was 10:1.
In the water distillation, the distillation time and consequently the
amount of operator attention required increases in proportion to the amount
of total distillate to be collected. From this standpoint, it would be de-
sirable to operate at the highest temperature possible to minimize distilla-
tion time. However, it was observed that at temperatures very much in excess
of 170°C, the entering water droplets tended to explode violently upon hit-
ting the surface of the hot acid, and to maintain a relatively constant tem-
perature, required the undivided attention of the operator.
Since the measurement of fluoride with the ion specific electrode is not
affected by varying small amounts of sulfuric acid that might occasionally
distill over, it was apparent that the requirements for close control of
temperature could be relaxed. It was found that if the temperature were
allowed to vary between 140-170°C, the explosive water/acid reaction could
be minimized, and only a periodic check of temperature and minor adjustment
of water addition rate was required. Under these conditions, the only con-
cern then was that sufficient distillate be collected to insure complete
fluoride recovery.
To determine the amount of distillate collection required under worst
case conditions, three experiments were performed at a carefully controlled
temperature of 140°C. As essentially quantitative recovery was obtained
after collecting 150 ml of distillate (Runs 5-7), it. seemed reasonable to
assume that one should be able to obtain quantitative recovery at pot tem-
peratures anywhere within the range of 140-170°C for a total of 200 ml of
distillate collected. In Run 7, the fourth fraction was found to contain
an insignificant amount of fluoride, lending support to this assumption.
2) Small-Volume Direct Distillations from Sulfuric Acid The water-
addition distillation from sulfuric acid with only minimal temperature con-
trol is a less tedious procedure than the normal Willard-Winter distillation
which requires very close control of temperature. The evaluation of an even
88
Arthur D Little, Inc
-------
simpler procedure, a direct distillation with no additions to the pot while
the distillation procedures, is described in this section. The distillation
apparatus used for these initial experiments was the same as had been used
in the preceding water-addition distillations.
Results of these initial direct distillations are shown in Table 11.
For the first five runs, the pot was charged with a total of 130 ml of acid
solution, and a total of 50 ml of distillate was collected. Recoveries of
fluoride from NaF were excellent. The first run with cryolite, however,
yielded only a 92% recovery. In all cases, 11-15% of the total fluoride
added was recovered in the second 25 ml fraction.
In an attempt to improve the cryolite recovery, the volumes of water
and acid charged as well as distillate collected were increased by 50%.
Two cryolite samples and one sample of NaBF^ were distilled under these
conditions, with an average recovery of 86% for the former and 100% for
the NaBF4 sample.
The third fraction contained 1-3% of the total fluoride collected,
which is hardly significant to the overall analysis. The low level of
fluoride in this fraction provides assurance that significant amounts of
fluoride will not be lost by failing to collect enough distillate.
A variety of species, including aluminum and iron, are known to retard
the distillation of fluoride. Therefore, several distillations were carried
out to evaluate potential interferences from phosphate, silicate, iron and
aluminum. The results are presented in Table 12.
In order to see whether phosphoric acid might influence the rate of
distillation of fluoride from solution, an addition and recovery of cryolite
was carried out in the presence of a 1000-fold excess of phosphoric acid.
The data suggest that the presence of phosphoric acid in the distillation
pot may have resulted in a slight increase in the percentage of fluoride
found in the first fraction, but this effect was not marked. Overall
recovery is not significantly different from that found previously for
cryolite alone.
The effect of large excesses of both aluminum and iron on the recovery
of fluoride via direction distillation were studied using the appropriate metal
89
Arthur D Little, Inc.
-------
TABLE 11
RECOVERIES OF FLUORIDE USING SMALL-VOLUME DIRECT DISTILLATION
vo
o
FROM SULFURIC ACID
FLUORIDE ADDED
SAMPLE
NaF(a)
NaF(b>
Cryolite (b)
(M moles)
10
10
10
10
278
197
322
FLUORIDE FOUND (jA MOLES)
1st 25 ml
10.4
10.2
10.0
9.9
215
162
244
2nd 25 ml
1.3
1.1
1.3
1.5
41
30
51
5rd 25 ml Total
NC(c) 11.7
NC 11.3
NC(c) 11.4
NC(c> 256
3 195
2 297
PERCENT 'd*
RECOVERY
103
99
100
101
92
99
92
NaBF
(b)
358
250
100
10
360
100
c
-t
D
ET
(a) Mixed with 4.5 g mixed carbonates, not fused
(b) Fused with 4.5 g mixed carbonates
(c) Initial pot solution 60 ml HO plus 70 ml H SO^, a total of 50 ml distillate
collected. In other runs, initial pot solution 90 ml HO plus 105 ml H2SO ,
a total of 75 ml distillate collected.
(d) Percent recoveries computed from total recovery after correcting for a re-
agent blank of 1.4 M moles fluoride.
-------
TABLE 12
EFFECT OF POTENTIAL INTERFERENCES ON DIRECT DISTILLATION - 75 ML DISTILLATE
FLUORIDE OTHER SPECIES
(a)
SPECIES
Cryolite B
NaF
NaF
NaF
NaF
NaF
NaF
NaF(d)
(a)
(b)
(c)
(d)
(e)
c
D
AMOUNT /b) AMOUNT
P* MOLES) TYPE (M MOLES)
214 H,PO,
3 4
10 Al (III)
10 Al (III)
10 Fe (III)
10 Fe (III)
10 Silicate
10 Silicate
10 Silicate
All samples mixed with 4.5 g
Interfering species added as
Corrected for reagent blank
200
10
10
10
1
10
1
1
MOLE RATIO FLUORIDE FOUND
(^MDLES)
OTHER/FLUORIDE 1st 25ml 2nd 25ml 3rd 25ml TOTAL
1000 182 11 NA(e) 193
1000 6.2 4.5 1.1 11.8
1000 7.6 3.1 0.2 10.9
1000 8.6 0.7 0.3 9.6
100 9.4 0.8 0.2 10.4
1000 2.3 2.1 1.6 6
100 6.5 2.6 0.5 9.6
100 5.5 2.8 0.4 8.7
PERCENT
RECOVERY
90
104
95
82
90
46
82
73
mixed carbonates, only cryolite was fused
aluminum
(acid and
-3
1.0x10 moles MgSO added in attempt
4
Fraction not analyzed
sulfate, ferric ammonium sulfate, and sodium metasilicate
carbonates only) of 0.14x10 moles F~
to precipitate silicate
—
n
-------
salt and sodium fluoride. In these experiments, aluminum was found not to
retard fluoride evolution, whereas iron seemed to have an effect, i.e., an
increase in the amount of iron resulted in a poorer recovery. It must be
noted that variations in the 1.4y mole reagent blank at the lOy mole sample
level can reduce the statistical significance of these findings. However,
we do believe the effect of lOy moles of iron on fluoride recovery is real,
and the possible influence of ly mole iron is reason enough to keep iron
concentrations below ly mole.
In comparison to the above, the deleterious effects of large amounts
of silicate are quite significant. The addition of acid to the water soluble
silicate results in rapid formation of a gel of silicic acid. Fluoride may
be absorbed into this gel and trapped. Also, portions of this gel are
splashed up onto the upper area of the flask during boiling away from further
contact with the acid. An attempt to precipitate the silicate as magnesium
silicate was of no avail and indeed may have resulted in still poorer
recovery.
3) ASTM-Type Direct Distillation from Sulfuric Acid The ASTM-type
direct distillation procedure was first evaluated by distilling five fused
cryolite samples according to ASTM Method D1179-68. The results of these
experiments are shown in Table 13; an average recovery of about 92% was
observed for the five runs. The aluminum/fluoride ratio varied from 0.17
to 5.0 and appeared to have no significant effect on the recovery.
With the water-addition distillation discussed previously, recoveries
closely approaching 100% could be obtained for the same cryolite sample.
In an attempt to achieve better recovery with the direct distillation, it
was decided to let the distillation proceed to a higher final pot temperature.
The distillate was collected in two fractions—that which came over at pot
temperatures up to 180°C and that which came over between 180 and 210°C.
The results of this experiment are presented in Table 14; for a variety of
fluoride levels and aluminum/fluoride ratios, approximately 95% of the added
fluoride was recovered in the first fraction. By allowing the distillation
to proceed to 210°C, an overall average recovery of 99% was achieved.
Similar experiments were performed on actual field samples from several
industries of concern in this work. The results presented in Table 15 show
92
Arthur D Little, Inc
-------
TABLE 13
RECOVERY OF FLUORIDE FROM FUSED CRYOLITE IN PRESENCE OF ALUMINUM
VIA 180°C ASTM DISTILLATION
Run
Al Present
(ymoles)
F added
(ymoles)
(a)
F Found
(ymoles)
Percent
Recovery
40
232
205
88
60
124
118
95
100
1000
1000
(b)
(b)
251
138
203
235
118
194
94
86
96
(a) Added as Cryolite B (see text) and assumed to be 100%
3NaF-AlF3. Each sample was fused with 4.5g mixed
carbonates (Na2C03~ COo) .
(b) Excess Al(III) added as aluminum sulfate.
93
Arthur D Little, Inc
-------
TABLE 14
COMPARISON OF DIRECT DISTILLATION RECOVERIES AT 180 and 210°C
Fluoride Added
(y moles)
(a)
Ratio
Al/F
(b)
Fluoride Found (y moles)
< 40 - 180°C 180 - 210°C Total
Percent
Recovery
100
10
95
99
99
10
100
100
10
9.9
93
0.2
10.1
98
101
98
1000
950
23
973
97
(a) Added as sodium fluoride
(b) A constant amount of Al(III), 1000 y moles, was added as aluminum
sulfate.
94
Arthur D Little, Inc
-------
TABLE 15
RECOVERIES OF FLUORIDE FROM PORTIONS OF PARTICULATE FIELD SAMPLES
VIA DIRECT DISTILLATION AND ELECTRODE MEASUREMENT
Sample Description
Aluminum Facility
Filter catch, ^ 19% of total sample
Water probe wash, ^ 21% of total
sample
Acetone probe wash, 50% of total
sample
Fluoride Found (y moles)
180°
Fraction
178
98
64
180-210°
Fraction
3.0
1.9
1.0
%F in 180°
Fraction
98
98
98
Glass Facility
Filter catch, 13% of total sample
Probe wash, 25% of total sample
205
8
11
0.1
95
99
Phosphate Rock
Standard sample
69
73
2.1
2.5
97
97
95
Arthur D Little, Inc
-------
that a 2-5% improvement in recovery was obtained by extending the maximum
distillation temperature from 180 to 210°C; furthermore, the results suggest
that the direct distillation can provide good quantitative recovery under
these conditions.
In carrying out the direct distillation to 210°C, a small amount of
sulfuric acid is carried over at the high temperature range. While this
has no effect on the fluoride electrode measurement, it can represent a
significant positive interference at low fluoride levels when other measure-
ment techniques such as Zr-SPADNS are used.
4) Comparison of the Two Sulfuric Acid Distillations and Comparison
with the Willard-Winter Perchloric Acid Procedure To further
compare the water addition and ASTM-type direct sulfuric acid distillations,
four samples of standard phosphate rock were distilled using each procedure.
These samples were first fused with mixed carbonates, and the final measure-
ment was made with the fluoride electrode. The fluoride content of the
phosphate rock had been determined by collaborative testing to be 3.80+0.018%
(dry basis). The results of the distillation comparisons are shown in
Table 16. For the water addition method, an average of 3.83% fluoride was
obtained in very good agreement with the accepted value. Four determina-
tions by the direct distillation procedure yielded an average fluoride
content of 3.74%, within 2% relative of the accepted value. Although we
have not been able to pinpoint the reason, we believe that the two low values
determined by direct distillation result from a systematic error, perhaps
the result of a leak in the apparatus. From our overall experience with
them, we believe that the two techniques for distillation from sulfuric acid
are capable of producing equally precise and accurate results.
In another series of experiments, distillation from sulfuric acid was
compared with the "accepted" Willard-Winter perchloric acid distillation.
Samples were chosen to be representative of those that might be encountered
in the industries of primary concern to this work—calcium fluoride/lead
(glass and ceramic industry), calcium fluoride/iron oxide (iron and steel),
cryolite (primary aluminum), and standard phosphate rock (phosphate industry).
The samples were fused with mixed carbonates prior to distillation and the
final measurement was performed with the fluoride electrode. The results of
this comparison are shown in Table 17. For three of the four samples the
96
Arthur D Little, Inc
-------
TABLE 16
DISTILLATIONS FROM SULFURIC ACID
Comparison of 150 - 165°C Water Addition with 210°C Direct
Method
Water Addition
Sample (mg)
54.1
. 57.4
53.9
54.7
[a]
Fluoride Found (\i moles)
108
115
109
111
[b]
Weight % F
3.80
3.81
3.84
3.86
Direct
51.2
46.0
35.0
38.6
103
89.0
70.0
74.4
3.82
3.68
3.80
3.67
[a] - Standard phosphate rock, 3.80% F (dry basis).
[b] - Electrode measurement. Corrected for reagent blank—water
addition, 2 p moles; direct 0.6 y moles.
97
Arthur D Little, Inc
-------
c
—t
o
TABLE 17
COMPARISON OF SULFURIC AND PERCHLORIC ACID WATER ADDITION DISTILLATIONS
H2SOit Distillation (150 - 165°C)
Distillation (135°C)
vo
oo
Sample
CaF2 + 1.0 g PbO
CaF2 + 1.0 g Fe203
Cryolite
Phosphate Rock
F Added
(y moles)
99.5
99.4
210
108
Percent
Recovery
82
97
97
100
F Added
(jj moles)
106
105
208
111
Percent
Recovery
92
90
87
98
-------
sulfuric acid distillation yielded recoveries equal to or greater than those
for the Willard-Winter distillation. The low recoveries with perchloric
acid at 135°C for the samples containing iron and aluminum are not unexpected.
For "hard to distill" samples such as these, a preliminary distillation from
sulfuric acid is then subsequently distilled from perchloric acid at 135°C.
Whether the 82% recovery for the simulated lead glass furnace sample by the
sulfuric acid distillation is real or results from gross experimental error
is not known. The fact that the distillation profile for the filter catch
particulates obtained from a lead glass facility is not significantly different
from the profile of other samples, as shown in Table 15, suggests that the
82% recovery may be due, at least, in part to experimental error.
d. Conclusions From the preceding experiments it is concluded that
either the water addition or direct distillation from sulfuric acid are
acceptable alternatives to the perchloric acid distillation if final fluoride
measurement is performed with the fluoride ion specific electrode. The
water addition distillation can be made less tedious by requiring that the
temperature be controlled within the relatively wide range of 140-170°C;
200 ml of distillate must be collected to ensure quantitative recovery.
To achieve quantitative recovery by the direct distillation, the pot tem-
perature should be taken to 210°C. For direct distillation of small amounts
of fluoride, measurement with the ion specific electrode is probably a
definite requirement; there is sufficient sulfuric acid carryover at tem-
peratures in excess of 180°C to interfere with most other measurement
techniques.
Of the two sulfuric acid distillation techniques, the direct distilla-
tion is both easier and faster. It can be made almost completely automatic
by incorporating a high-limit temperature sensor in the pot to turn off
the heat "once the 210°C temperature limit has been reached. The direct
distillation does require a nonproductive pre-distillation of the initial
acid charged into the pot. This limitation is not of major importance
because once it has been pre-distilled, the pot can be used for a number of
sample distillations. The pre-distillation has a distinct advantage in
reducing the reagent blank to a lower level than is normally found in the
water addition distillation.
99
Arthur D Little, Inc
-------
e. Analysis of Field-Collected Samples
1) Primary Aluminum Samples To gain Insight into the chemical nature
of samples that one would encounter when performing fluoride analyses on
process emissions, a variety of field samples were collected and studied in
the laboratory. As there was a greater opportunity to obtain samples from
a primary aluminum facility than from the other stationary sources, a larger
number of aluminum industry samples were studied. Limited studies were
also performed on samples provided by cooperating companies within the glass
and phosphate rock processing industries.
Seven sets of samples were obtained from primary aluminum reduction
plants. Sampling was carried out in a PHS-type sampling apparatus (the
model supplied commercially by Research Appliance Company was used), con-
sisting of a heated glass-lined probe, cyclone and filter assembly followed
by a series of water impingers. A summary of the collected samples and
sampling conditions is presented in Table 18. The six Series "A" samples
were taken at the exit from an APC device while the set denoted "B" were
collected ahead of the APC device.
The chemical nature of the impinger catch was studied by a variety of
methods including the following:
• Dissolved Solids—drying at 105°C and ignition at 850°C.
• Fluoride—direct electrode measurement after buffering with
0.5 M citrate.
• Aluminum—atomic absorption spectrometry.
• Silicon—colorimetric analysis using molybdenum blue reagent.
• Total Acidity—potentiometric (pH) titration using 0.01 N NaOH.
• SC>2—iodometric titration.
• Chloride—mercurimetric titration.
The results of these studies are presented in Table 19. The ratio of
aluminum to fluoride is less than unity in all cases. As has been shown
previously for direct electrode measurement, a ten- to twenty-fold excess
of aluminum over fluoride can be tolerated. The very high fluoride/silicon
ratio observed for these samples strongly suggests that most of the gaseous
fluoride enters the impinger at HF. A glass-line probe was used for obtain-
ing these samples, and if it were converting all HF to SiF^ as has been
100
Arthur D Little Inc
-------
TABLE 18
Summary of Samples and Sample Collection Data -
Primary Aluminum Plant
Sample
No
A-l
-2
-3
-4+
-5+
-6+
B-l
Dry Gas
Volume
(SCF)
24.8
23.9
23.7
23.6
23.7
23.5
35.3
Sampling
Time
(min)
60
60
60
60
60
60
60
Ave Stack
Temp
(°F)
105
95
95
100
100
105
200
Outlet
Temp
/ OT^N
75
75
75
75
80
75
80
Water
Impinge r
(ml)
19
22
—
—
—
—
__
Pickup
Silica Gel
(8)
4
8
9
-
-
-
_
Filter
Catch
(mg)771"
3
1
1
1
**
***
60
Filter
A
C
C
D
C
B
E
+ These runs were for filter catch only.
* Filter Types
A - 4" diameter MSA 1106B glass fiber
B - 2" diameter MSA 1106B glass fiber
C - 4" diameter Gelman "Acropore" membrane
D - 4" diameter Whatman No. 41 paper
E - 2" diameter Millipore HAW (0.45y pore)
** Not tared.
*** Tare weight not available.
101
Arthur D Little, Inc
-------
Table 19
Characterization of Impinger Catch from Primary Aluminum Plant Samples
Solids (mg) A -j / i \
6 Acid (mmoles)
Sample Volume (ml) 1Q5°C 850°C Fluoride (mg) Al (mg) Si (mg) 1st EP 2nd EP S02 (mmoles) Cl (mg)
M A-l 325 33 9 0.2 < 0.1 0.04 0.51 0.60 < 0.1 2
o
S3
A-2 390 17 3.5 0.2 < 0.1 NA 0.78 1.4 0.2 NA
A-3 370 7 1 0.35 < 0.1 0.02 0.75 1.5 0.2 9
B-l 341 69 NA 126 NA 0.74 NA NA NA 41
c
-t
D
cr
NA - Not analyzed
-------
postulated, significantly more silicon should have been found. It is also
possible that hydrolysis of silicon tetrafluoride occurred after absorption
into water yielding insoluble silicic acid which precipitated and was thereby
missed in the silicon analysis. As there were no solids visible in the im-
pinger solutions, this latter hypothesis should be valid for this group of
samples.
The amount of total acidity exceeds that which can be accounted for on
the basis of the measured fluoride, chloride and SC>2 contents. The titration
curves indicated that the major acidic species had pH -'s of about 3 and 7.
3.
The pH of HF is about 3.5 so it is not likely to be the stronger acid; in
Si
any event, the fluoride level is only about 0.01 millimoles or about 2% of
the amount of the stronger acid found. Since in subsequent studies it was
found that the acidic material did not interfere in the fluoride measurement,
no further attempt was made to provide a positive identification.
Several of the particulate filters were analyzed by emission spectro-
graphy; the results are presented in Table 20. In addition, the presence
of chlorine and sulfur was detected by X-ray fluorescence. X-ray diffraction
analysis yielded amorphous patterns; there was no evidence for the presence
of any crystalline phases.
Separate containers of water and acetone probe washings were obtained
as part of the Series B sample. These washings contained a significant
amount of undissolved particulate which was filtered off and analyzed by
X-ray methods. The results, shown in Table 21, indicate the presence of
the same general inorganic elements that have been observed previously in
the filter catch. However, crystalline alumina and cryolite were found in
the probe washings. Apparently these materials are attracted to and adhere
tenaciously to the probe walls during sample collection.
A summary of the distribution of fluoride and solids in the Series B
sample is presented in Table 22. The following observations should be noted:
1. The insoluble fluoride particulate collected in the probe
represents 7% of the total fluoride. X-ray diffraction
data identifies the insoluble fluoride as cryolite.
2. The filter catch yields an amorphous diffraction pattern,
yet accounts for 10% of the fluoride. This change in
103
Arthur D Little Inc.
-------
Table 20
Emission Spectrographic Analysis of Filter Catch
from a Primary Aluminum Plant
Level in A-4 ^-5
Sample (Paper}_ (Membrane)
> 10%
0.3-3% Al
o 0.03 - 0.3% — Si
0.01 - 0.1% Na, Si Al, Ti
30 - 300 ppm Fe, Ga
10 - 100 ppm Ca, Ti Ca, Fe, Ga, Na
3-30 ppm Cu, Mo, Ni, V
1-10 ppm — Cu, Mo, Ni, V
Paper Membrane B-l
Blank Blank (Membrane)
Al
Na
Si Fe
—
Ca,Pb, Si, V
Si Ca As, Cu, Ga, Sn, Ti
Ca, Fe Fe, Na Mg
Al, B, Cu, Na Al, Cu Ag, B
"> (a) This membrane blank was run using a filter of the type used for Sample A-5.
£3 The B-l membrane was from a second vendor, and no blank was run.
D
cr
-------
Table 21
X-Ray Analysis of Insoluble Particulate from Series B Probe Washings
Sample
Water Wash
Acetone Wash
Fluorescence
Ni, Ga, Fe
Al, Ca, S, Ti
Ni, Fe
Al, Ca, S, Ti
Diffraction
a-Al203, Na3AlF6
a-Al203
105
Arthur D Little Inc
-------
TABLE 22
FLUORINE AND SOLIDS DISTRIBUTION IN SERIES B SAMPLES
COLLECTED FROM A PRIMARY ALUMINUM PLANT
Filter"*"
Probe Washings Catch
Water Acetone
Initial Volume (ml) 100 144
Insoluble
Solids (mg) 212 16
Fluoride (mg) 10 2.5
Soluble
Solids (mg) 41 NA
Fluoride (mg) 15 0.5
Total
Solids (mg) 253 16 60
Fluoride (mg) 25 3 18
Distribution
Solids (%) 68 15
Fluoride (%) 15 10
Impinger
Catch
341
0
0
69
126*
69
126*
17
74
Total
228
12
110
142*
398
172
Soluble and insoluble fractions were not determined
NA Not analyzed
* Includes HF which was driven off during the evaporation step
106
Arthur D Little, Inc
-------
ratio of fluoride to solids from probe to filter suggests
a preferential sorption of gaseous fluorides on the filter.
3. Altogether, the probe collects 16% of the total fluoride
and 70% of the solids. Good probe cleaning techniques
are therefore absolutely necessary to achieve quantitative
results.
4. Since little silicon was found in the irapingers, it is
determined that HF is the only major gaseous fluoride
species, accounting for at least 74% of the total fluoride.
2) Glass Furnace Samples Portions of a sample collected in the stack
of a lead glass furnace were received from a cooperating company in the
glass industry. They employed a heated stainless probe, glass fiber filter
and water impingers. The analytical results obtained on the submitted samples
are presented in Table 23. The majority (98%) of the fluoride was collected
in the probe or on the filter. It is surmised that most of the fluoride was
present as PbF-, although X-ray analysis of the filter catch identified
nothing other than PbO. The fact that 14% fluoride was collected in the
probe implies that appropriate probe cleaning procedures are required.
3) Phosphate Rock Processing Samples To aid us in our understanding
of potential problems associated with_the analysis of actual field samples,
a phosphate fertilizer company provided us with aliquots of samples collected
from seven gas streams in the emission control system of a DAP plant.
Sampling was carried out with a stainless steel probe followed immediately
by a set of three impingers in series; the first impinger contained sul-
furic acid for the collection of ammonia and the others contained dilute
sodium hydroxide for collecting fluorides. At the conclusion of the sampling
period, the contents of all three impingers were transferred into a single
container.
The received samples were obviously acidic (pH < 1) and contained only
a small amount (<_ 1 mg) of particulate matter. This particulate was fil-
tered off; no fluoride was detected in these solids. An aliquot of the
filtrate was measured directly with the fluoride ion-specific electrode.
Good recoveries of standard additions of fluoride were obtained, indicating
107
Arthur D Little; Inc.
-------
TABLE 23
FLUORIDE AND SOLIDS DISTRIBUTION IN GLASS INDUSTRY SAMPLES
Initial Volume (ml)
Insoluble
Solids (mg)
Fluoride (mg)
Soluble
Solids (mg)
Fluoride (mg)
Total
Solids (mg)
Fluoride (mg)
Probe
Washings
140
Filter
Catch
1082
5.4 32
Impinger
Catch
280
Total
41
0.6
NA
4.8
0
0
NA
0.7
41
0
__
5
.6
.5
0.7
38.1
Soluble and insoluble fractions were not determined.
108
Arthur D Little, Inc
-------
that no significant amounts of interfering species were present in these
solutions. The fact that all fluoride was water soluble and that levels
of interfering species were tolerable, suggests that direct electrode
measurements of DAP samples without pretreatment is feasible and should be
carefully evaluated during field testing.
4) Direct Electrode Measurements on Field Samples The possibility
of using the fluoride electrode for direct fluoride measurments of collected
solution samples with no pre-treatment other than buffering is attractive.
Direct measurement should be applicable in a variety of cases, particularly
in the case of clear impinger solutions. To test this possibility, a number
of fluoride standard addition and direct measurement experiments were per-
formed on aliquots of the .Series A and Series B sample solutions collected
from a primary aluminum company; the results are shown in Table 24. Good
recoveries were obtained for the impinger solutions as well as for the
water probe washings. The results suggest that direct measurement of im-
pinger solutions is not significantly affected by interfering species. That
fact that fluosilicic acid, which was used for the Series "A" additions,
could be recovered quite well indicates that the direct measurement should
be feasible regardless of whether the fluoride enters the impinger as HF
or as SiF^. The good fluoride recovery from the Series "B" probe washings
suggests that water soluble fluoride could be measured in an unfiltered
sample without interference from dissolved cations. It is important to re-
member that particulate cryolite was present in the washings, and conse-
quently, a total fluoride measurement would certainly require fusion and
distillation.
E. Summary and Recommendations
Analytical methods development for the measurement of gaseous and
particulate fluorides has been approached with a view of providing a pro-
cedure which is easily and reliably performed and has an accuracy and pre-
cision consistent with sample collection and storage procedures, but no
worse than + 10%. Based upon the review of the literature, discussions with
workers in various industrial laboratories and our comprehensive laboratory
studies that have been presented in this section, several methods have been
identified as being consistently reliable. As the major laboratory effort
109
Arthur D Little Inc
-------
Sample
(a)
c
•n
A-l-I
A-2-I
A-3-I
B-l-I
Initial
0.60
0.60
0.60
0.79
0.79
0.79
1.25
1.25
1.25
155
Amount of Fluoride (ymoles)
(b)
Added
0.33
1.97
3.61
0.33
1.97
3.61
0.33
1.97
3.61
200
o Primary Aluminum
c\ _ _ _
;S^ —
Total Expected
0.93
2.57
4.21
1.12 •
2.76
4.40
1.58
3.22
4.86
355
525
725
Plant Sample
Found
0.89
2.5
4.5
1.0
2.7
4.6
1.4
3.2
4.7
340
490
690
Solutions
Found Minus P
Expected
- 0.04
- 0.07
+ 0.3
- 0.02
- 0.06
+ 0.2
- 0.2
+ 0.01
- 0.2
- 15
- 15
- 35
ercent
Error
- 4%
- 3%
+ 7%
- 2%
- 2%
+ 4%
- 13%
0%
- 4%
- 4%
- 7%
- 5%
B-l-W
325
325
200
400
525
725
490
690
- 15
- 35
- 7%
- 5%
(a) Suffix I indicates impinger solution, W indicates water probe wash.
(b) Fluosilicic acid added to Series A Solutions, NaF to Series B solutions.
Fluosilicic acid analyzed by acid/base titration.
-------
was directed toward evaluating a variety of techniques for measuring soluble
and insoluble fluorides in samples containing other potentially interfering
compounds, there has been insufficient data obtained for any specific method
to statistically specify method accuracy and precision, although probably
reliable estimates have been given in the previous discussions of individual
methods.
In striving for procedural simplicity in the hands of the inexperienced
analyst, yet demanding a reliable and reproducible result, the use of the
fluoride specific ion electrode for measuring fluorides in process effluent
samples seems to be the optimum approach. Direct measurements on impinger
solutions after buffering and addition of a metal complexant (0.5 M citrate
has been shown effective) seems feasible in some cases, and should be com-
pared critically to more rigorous sample fusion and distillation preparation
procedures as part of a continuing field evaluation study.
The direct measurement approach seems most promising when either a
particulate filter or electrostatic precipitator precedes the filter.
Direct measurement may also be feasible for the phosphate rock industry;
although both particulates and gases are collected in the impinger, the
particulate fluorides are generally quite water soluble and thus would be
included in an analysis for total fluorides.
In cases where insoluble particulate is present and a solubilization
step is required, distillation from sulfuric acid, either with continuous
water addition at temperatures in the range 140-170°C or directly according
to ASTM Method D1179-68, but with a final pot temperature of 210°C should
produce quantitative separation of fluoride from interferences prior to
measurement with the fluoride electrode. If the water addition method is
used, a minimum of 200 ml of distillate should be collected. The direct
distillation is the preferred procedure because it is simpler and has a
lower fluoride blank. The final fluoride measurement on the distillate
using the fluoride electrode can be made easily; adjustment of pH and
addition of 0.1 M citrate precedes the final measurement.
Ill
Arthur D Little; Inc
-------
VI. LITERATURE REFERENCES
1. F. M. Robinson, G. I. Gruber, W. 0. Lusk and M. J. Santy, "Engineering
and Cost Effectiveness Study of Fluoride Emissions Control" EPA Report
No. SN 16893.000 (January 1972).
2. C. C. Cook, G. R. Swany and J. W. Colpitts, "Operating Experience with
the Alcoa 398 Process for Fluoride Recovery," JAPCA 21, 479-483 (1971).
3. W. S. Smith, Paper presented at the 60th Annual Meeting, APCA,
Cleveland, Ohio (1967).
4. J. A. Dorsey and D. A. Kemnitz, "A Source Sampling Technique for
Particulate and Gaseous Fluorides," JAPCA 18^, 12-14 (1968).
5. L. A. Elfers and C. E. Decker, "Determination of Fluoride in Air and
Stack Gas Samples by Use of an Ion Specific Electrode," Anal Chem 40,
1658-1660 (1968).
6. "Standard Sampling Techniques and Methods of Analysis for the Deter-
mination of Air Pollutants from Point Sources," Florida Department
of Air and Water Pollution Control (January 1971).
7. G. L. Rounds and H. J. Matoi, "Electrostatic Sampler for Dust-Laden
Gases," Anal Chem 27_, 829-830 (1955).
8. G. H. Farrah, "Manual Procedures for the Estimation of Atmospheric
Fluorides," JAPCA _17, 738-741 (1967).
9. R. H. Mandl, L. H. Weinstein, G. J. Weiskopf and J. L. Major, "Separa-
tion and Collection of Gaseous and Particulate Fluorides," Second
International Clean Air Congress, Washington, D.C. (December 1970).
10. J. A. Dorsey, Environmental Protection Agency—Private Communication.
11. H. H. Willard and C. A. Horton, Anal Chem J2.2, 1190 (1950).
12. J. A. Dean, M. H. Buehler and L. J. Hardin, Assoc. Office Agr.
Chemists 40, 949 (1957).
13. R. S. Allison, "The Determination of Small Amounts of Fluorine in
Glass," J. Soc. Glass Technol., 3T_, 213-18 (1953).
14. J. J. Lingane, Anal Chem 39_, 881 (1967).
15. H. H. Willard and C. A. Horton, Anal Chem 24_, 862-865 (1952).
16. C. I. Grant and H. M. Haendler, Anal Chem ^8_, 415 (1956).
17. F. W. Dykes, G. L. Bouman, M. C. Elliot and J. E. Rein, U. S. Atomic
Energy Comm. Rept. IDO-14405 (1959).
112
Arthur D Little Inc.
-------
18. B. J. MacNulty, G. J. Hunter and D. G. Barrett, Anal Chim Acta 14, 368
(1956).
19. F. S. Grimaldi, Ingram and F. Cuttica, Anal Chem 27_, 918 (1955).
20. APHA, "Standard Methods for the Examination of Water, Sewage and
Industrial Waste," llth ed. (1965).
21. E. Bellack and P. J. Schouboe, Anal Chem _30_, 2032-2034 (1958).
22. W. H. Wharton, Anal Chem _34, 1296-1298 (1962).
23. Fisher Scientific Co., Application Data Sheet L-174 Lanthanum
Chloranilate.
24. Fisher Scientific Co., Application Data Sheet T-401 Thorium
Chloranilate.
25. M. A. Leonard and T. S. West, J. Chem. Soc., 4477 (1960).
26. M. S. Frant and J. W. Ross, Jr., "Electrode for Sensing Fluoride
Ion Activity in Solution," Science, 154. 1553 (1966).
27. E. W. Baumann, "Trace Fluoride Determination with Specific Ion
Electrode," Anal. Chim. Acta, 42_, 127 (1968).
28. J. L. Stuart, "A Simple Diffusion Method for the Determination of
Fluoride," Analyst, j)5_, 1032 (1970).
29. R. A. Durst, "Fluoride Microanalysis by Linear Null-Point Potentiom-
etry," Anal Chem, _40, 931 (1968).
30. M. S. Frant and J. W. Ross, Jr., "Use of a Total Ionic Strength Ad-
justment Buffer for Electrode Determination of Fluoride in Water
Supplies," Anal Chem, 40_, 1169 (1968).
31. C. R. Edmond, "Direct Determination of Fluoride in Phosphate Rock
Samples Using the Specific Ion Electrode," Anal Chem, 41, 1327 (1969).
32. J. Tusl, "Direct Determination of Fluoride in Phosphate Materials in
Mineral Feeds with the Fluoride Ion Activity Electride," JAOAC,53,
267 (1970).
33. R. T. Oliver and A. G. Clayton, "Direct Determination of Fluoride in
Miscellaneous Fluoride Materials with the Orion Fluoride Electrode,"
Anal. Chim. Acta, 5l_, 409 (1970).
34. B. L. Ingram, "Determination of Fluoride in Silicate Rocks Without
Separation of Aluminum Using a Specific Ion Electrode," Anal Chem,
4-2, 1825 (1970).
113
Arthur D Little Inc
-------
35. J. C. VanLoon, "The Rapid Determination of Fluoride in Mineral Fluorides
Using a Specific Ion Electrode," Anal Lett, 1(6), 393 (1968).
36. W. H. Evans and G. A. Sargent, "The Determination of Small Amounts of
Fluorine in Rocks and Minerals," Analyst, 92, 690 (1967).
37. J. L. Guth and R. Wey, "Sur un dosage du fluorure dans les
mineral," Crystallogr., 92, 105 (1969).
38. E. J. Duff and J. L. Stuart, "Determination of Fluoride in Calcium
Phosphates with a Fluoride-Selective Electrode," Anal. Chim. Acta,
52, 155 (1970).
39. L. Evans, R. D. Hoyle and J. B. McHaskill, "An Accurate and Rapid
Method of Analysis for Fluorine in Phosphate Rocks," N. Z. Jl. Sci.,
13, 143 (1970); See also H. N. S. Schaefer, Anal Chem, 15, 53 (1963).
40. R. L. Clements, G. A. Sergeant and P. J. Webb,' The Determination of
Fluorine in Rocks and Minerals by a Pyrohydrolytic Method," Analyst,
96, 51 (1971).
41. H. H. Willard and 0. B. Winter, "Volumetric Method for Determination
of Fluorine," Ind. Eng. Chem. (Anal. Ed.), .5, 7 (1933).
42. H. H. Willard and C. A. Norton, "Indicators for Titration of Fluoride
with Thorium," Anal. Chem. 22^, 1190 (1950).
43. D. S. Reynolds, "Further Studies on the Willard and Winter Method for
the Determination of Fluorine," J. Assoc. Official Agr. Chemists, 18,
108 (1935).
44. P. J. Kie, L. W. Rieger and H. E. Power, "Determination of Fluoride in
Biological Samples by a Nonfusion Distillation and Ion-Selective
Membrane Electrode Method," Anal. Chem., 41, 1081 (1969).
45. E. Bellack, "Simplified Fluoride Distillation Method," Am. Water Works
Assoc. J., 50_, 530 (1958).
46. D. R. Taves, Talanta, 15, 1015-1023 (1968).
47. B. Marshall and R. Wood, Analyst, 94, 493-499 (1969).
48. E. J. Conway, "Microdiffusion Analysis and Volumetric Error,"
MacMillan Pub. (1958).
49. D. R. Taves, "Separation of Fluoride by Rapid Diffusion Using
Hexamethyldisiloxane," Talanta, 15, 969 (1968).
114
Arthur D Little, Inc.
-------
50. J. Tusl, "A Study of the Influence of Sillcone Grease on Spectro-
photometric Methods of the Determination of Fluoride," Anal. Chem.,
41, 352 (1969).
51. W. Funasaki, et al, "Removal of Fluoride Ion with Anion-Exchange
Resins," Mem. Fac. Eng., Kyoto Univ., 18, 44-50 (1956).
52. I. Ziphin, W. D. Armstrong and L. Singie, "Chromatographic
Separation of Fluoride and Phosphate," Anal. Chem., 29, 310 (1950).
53. H. M. Nielsen, "Determination of Microgram Quantities of Fluoride,"
Anal. Chem. ,'_30, 1009 (1958).
54. Anon., Health Lab. Sci., 6(2). (1969).
115
Arthur D Little Inc
-------
APPENDIX A - TENTATIVE REFERENCE METHOD FOR SAMPLING AND ANALYZING
PARTICULATE AND GASEOUS FLUORIDES FROM STATIONARY SOURCES*
1. PRINCIPLE AND APPLICABILITY
1.1 Sampling
1.1.1 Principle - Samples are collected isokinetically by apparatus
which provides for the separate collection of particulate and gaseous
fluorides.
1.1.2 Applicability - The method is directly applicable for analyzing
"dry" streams [where "dry" refers to a relative humidity of less than 100%
at stack conditions] from various industrial stationary sources, includ-
ing the various processes within the primary aluminum, iron and steel, and
glass and ceramic industries. With only slight modification (described
in Section 11), the method can be applied to analyzing "wet" streams
[where "wet" refers to water or steam entrainment in the stream] such as
occur for unit operations within the phosphate rock processing industry.
In the latter case, particulate and gaseous fluorides cannot be separated.
1.2 Analysis
1.2.1 Principle - Fluoride ion in the collected sample is measured
with the fluoride specific ion electrode. If the sample to be analyzed
contains particulate, a caustic fusion must be performed to ensure that
all fluoride will be soluble. A distillation of the sample from sulfuric
acid prior to the electrode measurement is required for removal of in-
terfering ions.
1.2.2 Applicability - The method as written is applicable to any
type of sample collected from process effluents which are discharged
into the air from unit processes within the primary aluminum, iron and
steel, glass and ceramic, and phosphate rock industries. The nature of
the sample collected has little effect on the success of the analysis,
so that the procedure should be generally applicable to stationary source
fluoride emissions from other industries.
*This method is based on laboratory evaluations, discussions with
industrial representatives and reference to the open literature.
The method cannot be accepted as final until suitable field tests
have been conducted and modifications to the method are made, as
appropriate.
A-l
Arthur D Little, Inc
-------
2. RANGE AND SENSITIVITY
2.1 Sampling - Not Available
2.2 Analysis - The fluoride electrode can measure fluoride concen-
trations in the range of 0.02-2,000 yg/ml; however, measurements of less
than 0.1 yg/ml require a great deal of care and should be avoided. Since
the collected distillate is diluted to 500 ml and a 25-ml aliquot of that
solution is then diluted to 50 ml for measurement, the practical lower
limit is 100 yg of fluoride; the reagent blank for fusion and distillation
averages 10 yg fluoride. The upper limit of fluoride which can be measured
successfully.
3. INTERFERENCES
3.1 Sampling - Preliminary measurements indicate that as much as 10%
of the total fluoride may be lost in the sampling train, principally through
interaction with the probe. Also, physical and chemical adsorption of
gaseous fluorides onto particulate can effect the apparent gaseous to
solid fluoride distribution.
3.2 Analysis - Metals such as aluminum, iron and lead form soluble
fluoride complexes reducing the fluoride vapor pressure which can result
in incomplete evolution of fluoride during distillation. Silica forms a
precipitate which entraps fluoride, sticks to the walls of the still pot
and prevents complete evolution of fluoride.
Maximum amounts of these interferences which can be tolerated have not
been determined as a function of fluoride concentration. However, initial
results indicate that for 10 y moles of fluoride, the presence of more
than 1 mmole of silicon or iron, more than 5 mmoles of lead, or more than
10 mmoles of aluminum can adversely affect distillation recovery.
4. PRECISION, ACCURACY AND STABILITY
4.1 Sampling
4.1.1 Precision - Not Available
4.1.2 Accuracy - Not Available
4.1.3 Stability - The particulate catch, probe washings, and impinger
solutions remain stable and unchanged for at least a month as long as they
are stored in sealed polyethylene or flint glass containers.
4.2 Analysis - Reliable estimates of precision, accuracy and stability
when analyzing field samples are not yet available. Although very tenta-
tive, some estimates drawn from laboratory studies are present below.
A-2
Arthur D Little, Inc
-------
4.2.1 Precision - At the 2 mg fluoride level, a relative standard
deviation of + 2% or better was achieved on sets of quadruplicate samples.
4.2.2 Accuracy - Accuracy depends first upon the accuracy with which
the fluoride content of the standardization samples is known. The more
important limitation on accuracy is the effect of metals (see 3.2) on
distillation recovery. Within the constraints of 3.2, 95% or better
recovery can be achieved.
4.2.3 Stability - The one source of instability in the method is
potential drift in the electrode response. A calibration curve, if
used, should be generated daily and a known, mid-range standard should
be measured as a check at least once per hour. If direct readout is
employed, the span should be checked hourly.
Electrode response is temperature sensitive; concentration indi-
cated will change by about 1.5%/°C. If ambient lab temperature fluc-
tuates more than a few degrees, it is advisable to plan samples in a
water bath prior to measurement. If an electrically-driven magnetic
stirrer is used, precaution should be taken to see that it does not
heat the solution.
5. APPARATUS
5.1 Sampling
5.1.1 Sample Probe - A 5-foot (or longer) by 1/2 to 5/8 inch
diameter 316 stainless steel with stainless steel nozzle and orifice;
or
A 5-foot by 1/2 to 5/8 inch diameter borosilicate glass tubing encased
in a metal support tube fitted with a stainless steel nozzle and ori-
fice, both with provision for heating the probe to maintain a gas tem-
perature of 250°F.
5.1.2 Particulate Collector - A 316 stainless steel or Pyrex filter
holder with suitable filter media (glass fiber, organic membrane, or
paper);
or
A stainless steel enclosed electrostatic precipitator as described in
Reference A-l; both systems including a heating system capable of
maintaining any temperature up to 250°F.
5-1.3 Impingers - Two, 500 ml Greenburg-Smith impingers, made from
polyethylene or glass and containing 200 ml of distilled water, to be
contained in an ice bath.
A-3
Arthur D Little, Inc.
-------
5.1.4 Mist Trap - A 500 ml bottle containing 175g of dried silica
gel for protection of down stream components from moisture.
5.1.5 Vacuum Pump - Rated at 4 cfm at 0 in.Hg and 9 cfm at 26 in.Hg.
5.1.6 Dry Gas Meter - Rated at a maximum of 175 cu. ft. per hour.
5.1.7 Air Flowmeter - A calibrated rotameter or critical orifice
capable of measuring airflow within 2% over the range of 1 to 50 ft/sec.
5.2. Analysis
5.2.1 Fusion Crucible - Crucible, nickel or Inconel, 60 ml capacity.
5.2.2 Distillation Assembly - Glassware (shown in Figure A-l) con-
sisting of a 1-liter, round bottonu borosilicate boiling flask' ', and
adapter with a thermometer opening-c^, a 300 mm Graham condenser(f) ,
and a thermometer reading to 250°C^ '. Standard taper or spherical
ground-glass joints shall be used throughout.
(a)
Heat is provided by a hemispherical heating mantle connected to
a laboratory variable transformer. Distillate is collected in a 500-ml
volumetric flaskAg).
5.2.3 Fluoride Specific Ion Electrode System
5.2.3.1 Fluoride Ion Specific Electrode (Orion Model 94-09 or
equivalent solid-state fluoride ion activity sensing electrode using
a rare-earth doped lanthanum fluoride crystal)
5.2.3.2 Reference Electrode - Silver/silver chloride or saturated
Calomel suitable for use with the fluoride electrode and electrometer
used. A large area liquid junction is preferable to a fiber-type liquid
junction.
5.2.3.3 Electrometer (more commonly a pH meter with millivolt
scale, or a "Specific Ion Meter" made specifically for ion-specific
electrode use) capable of + 0.5 mv. resolution.
6. REAGENTS
6.1 Sampling
6.1.1 Distilled Water
6.2 Analysis
6.2.1 Fusion Flux - Either NaOH or Na2C02/K.2C03 (equimolar mixture)
may be used. Flux components shall be ACS reagent grade.
A-4
Arthur D Little, Inc
-------
6.2.2 Calcium Oxide - Fluoride content less than 50 ppm.
6.2.3 Concentrated Sulfuric Acid - ACS Reagent grade.
6.2.4 Citrate Buffer - 1.0 M stock solution. Dissolve 294.1 g of
reagent grade Na,.C,.H,-0 .2H?0 (Fisher CERTIFIED, or equivalent) in 1
liter of distilled water.
6.2.5 Bromthymol Blue Indicator Solution - A 0.4% aqueous solution
is commercially available.
6.2.6 Fluoride Standard Solution - A O.lM fluoride standard solu-
tion is commercially available, or it can be prepared by dissolving
ACS Reagent grade sodium fluoride in water.
7. PROCEDURE
7.1 Sampling
7.1.1 Prior to sampling, measure stack gas velocity at a series of
test points to determine the velocity profile of the stack, using pro-
cedure given in Reference A-2. If variation is 10% or less, sampling
may be carried out at point of average velocity. If variation exceeds
10%, sampling must be carried out incrementally at a series of test
points to reflect an average velocity per the procedure given in
Reference A-2.
7.1.2 Sampling will be isokinetic at a rate between 0.75 and 1.25
cfm.
7.1.3 Sampling time will reflect all cyclic variations in the pro-
cess being monitored, but will not be less than 60 minutes.
7.1.4 After completion of sampling, the train shall be disassembled
and the collected materials placed in polyethylene storage containers.
The final volume of the impinger catch together with the weight increase
of the silica gel will be noted to calculate moisture content of the
gas. Separate containers will be used for the particulate catch, im-
pinger catch and washings, and probe washings.
7.2 Analysis
7.2.1 Preparation of Sample for Distillation - Whether the fluo-
ride samples have been collected by an electrostatic precipitator in
NaOH solutions, or on filters, they generally require some treatment
before a distillation can be made. Organic matter must be destroyed
without incurring any fluoride losses, and large volumes of solution
may need to be evaporated to volumes which will permit their placement
in the distilling flasks. Several precautions are mandatory when pre-
paring samples for distillation.
A-5
Arthur D Little, Inc
-------
7.2.1.1 Filter Paper Sample - Fold the filter paper with the clean
side out and place in a nickel or Inconel crucible. Saturate the paper
with saturated lime water solution prepared from fluoride-free CaO.
Complete saturation of the paper is essential to prevent loss of fluo-
ride. Heat the crucible and contents on a hot plate to remove excess
water. When dry, ash the sample in a muffle furnace at a temperature
not in excess of 600°C until no carbon remains. Ordinarily this is
accomplished in less than 1 h. Remove the crucible from the furnace,
cool, and add 3 g of NaOH (or 4.5 g mixed carbonates). Fuse the con-
tents over a burner for ten minutes, and then allow to cool to a tem-
perature less than 100°C.
7.2.1.2 Glass Fiber Filter Sample - Proceed in accordance with
7.2.1.1 for filter papers except to omit the ashing step.
7.2.1.3 Organic Membrane Filter Sample - Proceed in accordance
with 7.2.1.1 for filter papers, except to omit folding, and thoroughly
char the filter on the hot plate at low heat, prior to ignition. Ex-
cessive heat causes an instantaneous ignition of the filter with an
attendant loss of sample.
7.2.1.4 Impinger Sample - Depending on whether or not a filter
or electrostatic precipitator was used ahead of the impingers in the
sampling train, the impinger sample can be either a clear solution or
can contain suspended or settled solids.
If the impinger sample contains no undissolved solids, it is ready
for distillation. If the sample volume is greater than 300 ml, a
300 ml aliquot can be taken for distillation; if the fluoride level is
low, the volume can be reduced to 300 ml or less by first adding 0.1 g
CaO and then evaporating on a steam bath. During evaporation, care
must be taken to avoid mechanical losses due to splattering.
If the impinger sample contains insoluble particulate, it should
be filtered and the filtrate set aside while the filter paper and
collected solids are treated as in 7.2.1.1. (If the impinger sample
containing insoluble particulate is greater than 300 ml in volume,
0.1 g of CaO should be added and the solution volume reduced by evap-
orating over a steam bath prior to performing the filtration.) After
the solids which were filtered out of the impinger sample have been
ashed and fused, and the fusion allowed to cool, the fusion solids
along with the impinger sample filtrate are ready to be transferred
together into the distillation flasks.
7.2.1.5 Electrostatic Precipitator Sample - Wash the contents of
the precipitator into a 250 ml nickel or Inconel beaker with distilled
water. Add 0.1 g CaO and 3 g of NaOH, evaporate to dryness, and fuse
over a burner for 10 minutes. Allow the fusion to cool to less than
100°C.
A-6
Arthur D Little, Inc.
-------
7.2.2 Distillation
7.2.2.1 Charging the Flask with Acid and Adjusting the Acid/Water
Ratio - Before the actual samples can be distilled, the distilling
flask must be charged with a mixture of sulfuric acid and water and
the acid/water ratio adjusted by distilling the contents of the flask
until the pot temperature reaches exactly 210°C. This pre-distillation
also serves to strip out traces of fluoride in the H-SO,. Place 400 ml
of water in the distilling flask and add 200 ml of concentrated H SO,.
Observe the usual precautions while mixing the H«SO,; the acid should
be added slowly with constant swirling of the flask. Add about a half
dozen glass beads and assemble the apparatus as shown in Figure A-l.
The electric heating mantle is then turned on and the distillation of
the initial acid/water charge is allowed to proceed until the temper-
ature of the still pot reaches exactly 210°C. The heat is then im-
mediately removed and the temperature of the still pot allowed to cool
to less than 60°C before proceeding to add the first fluoride sample.
The distillate that was collected during this initial adjustment of
the acid/water ratio is of no use and is discarded.
7.2.2.2 Distillation - The sample in a total liquid volume of
300 ml is then slowly transferred into the distilling flask with good
mixing by swirling. Solid samples from fusions should be broken up
with a spatula, the fine chunks introduced directly into the flask
with the aid of a water wash, and additional water added so that the
total volume added to the distilling flask is about 300 ml. In this
case, the exact volume is not critical. However, in the case of
fluoride-containing solutions with a volume greater than 300 ml, if
an aliquot is taken for introduction into the still, the volume of
the aliquot must obviously be known accurately.
7.2.3 Analysis
7.2.3.1 Calibration - Fluoride standards are prepared by taking
aliquots of the 0.1 M stock solution, adding 5.0 ml of 1 M citrate
buffer, and diluting to a final volume of 50 ml. For wide dynamic
range measurements, seven standards spanning the range 10 - 10~" M
are adequate. The standards are measured and a calibration curve
(millivolts vs. log concentration) is constructed.
For measurements over a narrower range, certain meters may be cali-
brated to read directly in ppm fluoride by calibration with two known
solutions.
7.2.3.2 Measurement - The collected distillate (<210°C) is diluted
to 500 ml and a 25 ml aliquot is pipetted into a 50 ml volumetric flask.
Four drops of bromthymol blue are added and, if necessary, pH adjusted
to be in the range of 6.6 - 7.1 (indicator is green in color). Five
milliliters of 1.0 M citrate buffer is added, and the resulting solu-
tion diluted to 50 ml.
A-7
Arthur D Little, Inc.
-------
The solution is transferred into a beaker, the fluoride electrode
immersed and the EMF in millivolts (or ppm F if instrument is direct
reading) is read. Concentration of fluoride is read from the calibra-
tion curve.
8. CALIBRATION. STANDARDS. AND EFFICIENCIES
8.1 Sampling - Not available.
8.2 Analysis - The electrode measurement is calibrated daily and
checked hourly. Fusion and distillation are not normally calibrated
but assumed to be quantitative. Tests to date indicate that this
assumption is valid, but field testing is required to substantiate it.
The entire method should be run occasionally at appropriate times using
a primary standard such as ACS Reagent grade NaF or a collaboratively
analyzed standard sample of a material such as cryolite or phosphate
rock to ensure that it is indeed working properly in a particular
laboratory for a particular analyst.
9. CALCULATIONS
9.1 Sampling - From measurements of gas temperature, barometric
pressure and collected volume of water, calculate dry gas volume and
total gas volume at standard conditions. These volumes are required
for subsequent calculations of fluoride concentration.
9.2 Analysis - Total mg F in the submitted sample is computed from
the mg/ml value derived from the calibration curve as follows:
.,,. „ / / -, x -innn total sample volume received
milligrams F = (mg/ml) x 1000 x —: r* : — —
volume of sample aliquot taken
10. REFERENCES
10.1 Sampling
A-l G. L. Rounds and H. J. Matoi. "Electrostatic Sampler for
Dust-Laden Gases," Anal. Chem. 27. 1955, 829-830.
A-2 "Standards of Performance for New Stationary Sources,"
Federal Register 36, August 17, 1971, 15704-15722.
10-2 Analysis
A-3 ASTM Method D1606-60 "Standard Method of Test for Inorganic
Fluoride in the Atmosphere."
A-4 ASTM Method D1179-68 "Standard Method of Test for Fluoride
Ion in Industrial Water and Industrial Waste Water."
A-8
Arthur D Little, Inc.
-------
11. MODIFICATION FOR MEASUREMENT OF WET GAS STREAMS
The procedure for measuring fluorides in wet gas stream is the
same as above, except for the following modifications:
5.1.2 Delete
7.1.4 Change third sentence to: The impinger catch and washings
and the probe washings will be combined and placed in one container.
A-9
Arthur D Little, Inc
-------
Figure A-l Distillation Assembly for Fluoride Insolation
A. Electric Heat Mantle
B. Round-bottom flask, 1000 ml
C. Adapter with thermometer opening
D. Thermometer 250°C
E. Connecting tube
F. Graham condenser, 300 mm
G. Volumetric flask, 500 ml
A-10
Arthur D Little, Inc.
1
-------
DOT 1977
DA07B24150792J "
Development of methods for. sampling and analysis of
Darticulate and gaseous fluorides from stationary sources
Author: Peters, t. T., Oberho1tzer, j. t., Valentine, j. R.
Location: little (Arthur' D.). Inc., Cambridge. Moss.
Section: CA05&C02, CA070000 Pubi Class: T
Journal: U. G.-!!nt. Tocn. Inform, Sorv.. PS Rep. Coden:
XPOF^CA Publ : 72 . lor.uc-: No. 213313/0, Pages'. 133 pp.
Citr.tion: Govt. f?ep. Announce. (U.3.) 1973. 73(2), 68
Ava i 1 : NT IS
Identifier's: fluoride, cic-tn air-, fluorine detn air, air
analysis fluorine fluorice, pollution air fluorine fluoride
------- |