METHODS OF CHEMICAL ANALYSIS FOR OIL SHALE WASTES
' by
John R. Wallace, Linda Alden, Francis S. Bonomo,
John Nichols and Elizabeth Sexton
Charles H. Prien Center
for Synthetic Fuel Studies
Denver Research Institute
University of Denver
Denver, Colorado 80208
Contract No.
68-03-2791
October, 1982
EPA Project Monitor: Robert Thurnau
Industrial Environmental Research Laboratory
Cincinnati, Ohio
Prepared for
U.S. ENVIRONMENTAL PROTECTION AGENCY
Office of Research & Development
Washington, D.C. 20460
-------
DISCLAIMER
This report has been reviewed by the Industrial Environmental Research
Laboratory, U.S. Environmental Protection Agency, and approved for publication.
Approval does not signify that the contents necessarily reflect the views and
policies of the U.S. Environmental Protection Agency, nor does mention of trade
names or commercial products constitute endorsement or recommendation for use.
-------
ABSTRACT
This report describes methods of chemical analysis which are necessary in
order to adequately test the various pollution control technologies which are
presently being proposed for the treatment of oil shale retort gases and
wastewaters. It includes step-by-step protocols for determining important
species and it provides supporting evidence and discussion to allow the analyst
to adjust the procedure to the varied sample types which he may encounter.
Emphasis is placed on methods which are sufficiently rugged and rapid to be
useful on site during the field testing of pollution control systems.
A type of ion exchange chromatography, referred to in the literature as
suppressed ion chromatography, has been applied for the determination of major
and minor anions in retort wastewaters. Original experience with this method
for retort wastewaters indicated the' presence of very late-eluting compounds
which could not be removed from the column in a reasonable time and which would
interfere with subsequent analyses. This problem was solved with a column
switching arrangement which allows the late eluters to be separated on a
pre-column while the earlier-eluting compounds are separated on the main
analytical column. The same valve and column configuration can be used for
four different protocols: the first protocol, referred to as the majors
protocol, determines SOI, N°i> S20i, and SCN~ in a single run using 7 mM
Na2C03 as an eluent. The second protocol determines the late eluters,
S20f and SCN~ using an eluent of 7 mM NaC03 + 0.5 mg/1 of SCtt". Earlier
eluters are determined by the third protocol using a valve switching arrange-
ment which traps the late eluters on a pre-column, which is periodically
flushed to waste. Peaks which can be observed with this procedure include
Cl~, acetate, N02, SOf, NOi, P0|~ and SOJ, although not all have been success-
fully separated when present in the same solution. While the valving and
II
-------
column configuration remain unchanged, the analysts must be prepared to adjust
the eluent for the various types of samples and analytes encountered. The
fourth protocol is for the measurement of total sulfur and is carried out by
oxidizing the various sulfur species present to SOI, which is then determined
by the protocol for the early eluters.
The determination of sulfide by potentiometric titration with Pb(II) has
been investigated for retort wastewaters. Thipcyanate, thiosulfate, sulfate,
chloride, carbonate, ammonia, and hydroxide ions, which are components of
retort waters, have been tested as potential interferences for sulfide concen-
tration in the range 1-1,000 mg/1 and have been shown to be insignificant
within the normal range of interest. The titration has been evaluated under
field conditions with actual retort waters with regard to precision, recovery,
reliability, and ease of use. The titration procedure has also been compared
to the direct calibration method with the AgS ion selective electrode. The
titration procedure, which includes an AgS ion selective electrode to monitor
the titration and a Gran's ploferid point, is the preferred method of analysis.
Thermal evaporation, lyopholization, and the measurement of colligative
properties have been investigated as measurements of the total solute content
in retort wastewaters. Of these, both thermal evaporation and lyopholization
were shown to be inappropriate. Of the various colligative properties
considered, the freezing point depression was shown to be the best measurement
of total dissolved solutes. Using this method the total solute content can be
measured in units of moles/liter over the range 0.001 - 3.0 moles/liter (count-
ing each ion separately).
Three distinct methods have been developed for the analysis of ammoniacal
N in retort waters. The first method uses an ammonia selective electrode but
III
-------
minimizes many of the problems associated with this device by maintaining the
electrode in pure standard to which small amounts of sample are added. The
second method, ion chromatography, separates NH4+ on an ion exchange column
with detection by electrical conductivity. The third method involves
absorption of UV radiation by ammonia in the headspace over a basic sample
solution. All three methods are capable of quantitating ammoniacal. N in
turbid, briny, and organic-laden wastewaters but are distinguished by several
secondary characteristics. For example, the first method requires the least
investment in equipment but is the most labor intensive and least precise. The
ion chromatographic method is readily implemented with commercially available
equipment and also is capable of measuring Na and K simultaneously. The gas
absorption method is promising as the basis for an on-line, unattended monitor
and is capable of distinguishing between aqueous NH3 and NH4+. It is the most
precise measurement but requires spectral background correction in order to
achieve accurate results.
A method for measuring total sulfur in oil shale retort gas is described.
The method operates by converting the various sulfur species to S02 which is
then monitored by a commercially available monitor. Two devices, a heated tube
and a flame, have been evaluated for converting the various individual sulfur
species to S02. Of these, the conversion efficiency in the tube was shown to
depend both on the temperature and the species being oxidized, undesirable
performance for this purpose. The flame was shown to convert essentially 100%
of the various sulfur species to S02- Based on this performance, a device was
constructed for the measurement of total sulfur in retort gas and tested at an
oil shale retort. The total sulfur measured with the flame converter agreed
within experimental error with the sum of the individual sulfur species.
IV
-------
Gas chromatography with flame photometric detection was developed as a
means of measuring sulfur species in retort gas. Two potential problems were
considered in detail in order to establish the veracity of this technique:
First, hydrocarbons, one of the major components of retort gas, are known to
quench the fluorescence of the flame photometric detector (FPD), and it was
necessary to determine whether this would be a problem in realistic retort gas.
The second potential problem was the large number of sulfur species which could
occur in retort gas. Without prior knowledge one can easily number twenty
possible gases, which thus must be separated from each other if an unambiguous
assignment is to be made to each.
Fluorescent quenching effects were measured on two types of commercially
available FPD's, a single-flame detector and a dual-flame detector. The latter
exhibited no significant quenching effects over the concentration ranges of
interest in retort gas. However, for the single flame detector quenching
effects cannot be ruled out entirely. Although hydrogen sulfide in retort gas
is usually abundant enough to minimize quenching effects, the minor species
could be subject to quenching effects unless precautions are taken. These
precautions include operating the detector with the air and hydrogen flows
reversed and measuring peak height rather than peak area. In addition, columns
are selected that minimize co-elution with hydrocarbons! The single flame
detector exhibited both suppression and enhancement of the fluorescent signal.
Because of the large number of sulfur species which could conceivably
occur in retort gas, it became apparent that a single packed column could not
unambiguously separate all possible species. Efforts were therefore made to
find a column which would separate the compounds of primary interest—hydrogen
sulfide, carbonyl sulfide, sulfur dioxide, carbon disulfide, methyl mercaptan,
and ethyl mercaptan—from each other as well as from the later eluting sulfur
V
-------
compounds. Several columns were evaluated with regard to their ability to
achieve this separation as well as the required separation from hydrocarbons.
Columns were also tested for their ability to tolerate water vapor as well as
the other compounds in retort gas. The best general purpose column packing for
the determination of the sulfur compounds of primary interest was a Carbopack
B HT 100, although a Chromosil 310 packing would be useful for the occassional
determination of thiophenes. A protocol for the primary sulfur species is
described for the concentration range of 5-50,000 ppmv using a Carbopack
B HT 100 column arranged in a backflush-to-detector configuration.
Electrical conductivity, pH, alkalinity, and total inorganic carbon, while
not investigated explicitly in this study, are discussed briefly. It is
suggested that the measurement of pH and electrical conductivity with the
standard conductivity cell and pH electrode, respectively, has demonstrated no
obvious problems, although frequent cleaning and calibration should .be ex-
pected. It is recommended that the alkalinity test be discontinued as a
measurement of dissolved carbon dioxide because of interferences due to
ammonia and organic acids. Dissolved carbon dioxide should instead be deter-
mined by commercially available analyzers which are also suitable for total
s,
organic carbon measurements. Precautions for the latter two measurements are
discussed in the text.
VI
-------
TABLE OF CONTENTS
Disclaimer. . I
Abstract II
Table of Contents VII
List of Figures X
List of Tables. . ' XIII
Acknowledgement . XIV
1. Introduction 1-1
References 1-3
2. Minor Anions in Retort Wastewater by Ion Chromatography. . . . . 2-1
Experimental 2-5
Results and Discussion 2-10
Protocol for Major Anions Analysis 2-28
Principle. 2-28
Equipment 2-30
Reagents 2-31
Procedure 2-32
Interferences 2-33
Protocol for Thiosulfate and Thiocyanate .... 2-34
Principle 2-34
Equipment . . 2-34
Reagents . . . . . . . 2-35
Procedure 2-36
Protocol for Early Eluting Species 2-37
Principle 2-37
Equipment. . . ......... 2-37
Reagents 2-38
Procedure 2-39
Interferences, 2-40
Protocol for Total Sulfur 2-40
Principle 2-40
Equipment 2-41
Reagents 2-41
Sample Pre-treatment ... 2-41
Procedure . 2-43
References 2-43
3. The Determination of Sulfide in Oil Shale Wastewaters 3-1
Experimental Section 3-4
Results and Discussion 3-6
Conclusions 3-15
Recommended Method for Total Sulfide in Retort Wastewaters . . . 3-16
Principle 3-16
Reagents 3-16
Equipment 3-17
Procedure 3-18
Recording Data 3-20
Calculations 3-20
Instructions for Gran's Plot Program 3-21
References 3-25
Appendix 3A
Statistical Interpretation 3-26
VII
-------
4. Total Solutes in Retort Wastewaters 4-1
Experimental Section 4-2
Results and Conclusions 4-3
Acknowledgements . 4-7
Recommended Method for Total Solutes in Retort Water 4-7
Principle 4-7
Comment 4-7
Applicability 4-8
Reagents 4-8
Procedure . 4-9
Data Reporting . . . . 4-10
References 4-10
5. Total Ammoniacal Nitrogen in Retort Wastewaters. 5-1
Experimental Section 5-5
Sample Addition Technique 5-5
Ion Chromatography 5-6
Results and Discussion 5-9
Sample Addition Method 5-9
Ion Chromatography 5-12
Gas Absorption Method 5-15
Comparison of Methods . 5-25
Protocol for Total Ammoniacal N by Ion Chromatography 5-30
Principle. . 5-30
Comments 5-30
Reagents and Supplies. ..... 5-30
Apparatus 5-32
Procedures ' 5-32
Protocol for Total Ammoniacal N by Modified Electrode Method . . 5-33
Principle . . 5-33
Comments 5-33
Reagents and Supplies 5-33
Apparatus 5-34
Procedures 5-34
References 5-36
6. Total Sulfur in Retort Gas 6-1
Experimental ..... 6-4
Results and Discussions 6-5
Appendix 6A
Derivation of the Calibration Form for S Analyzer 6-23
Appendix 6B
Calibration of the Mass Flow Meter 6-28
7. Sulfur Species in Retort Gas . . . 7-1
Experimental 7-7
Reagents 7-7
Column Packing Process 7-8
Equipment 7-9
Preparation of Gas Standards 7-10
Procedure for Measuring Signal Quenching 7-11
Procedure for Measuring Retention Times. . 7-12
Procedure for Measuring Actual Retort Gas. . . 7-12
VIII
-------
7. Sulfur Species in Retort Gas (cont.)
Results and Discussion . 7-13
Flame Stability 7-13
Fluorescence Quenching 7-17
Uniformity of Response 7-32
Column Characterization 7-33
Application to Retort Gas 7-52
Protocol for Determination of Hydrogen Sulfide,
Carbonyl Sulfide, Sulfur Dioxide, Carbon Disulfide,
Methyl Mercaptan, and Ethyl Mercaptan in Retort Gas 7-63
Principle 7-63
Interferences 7-63
Equipment. . .' • • • 7-64
Standards 7-67
Procedure '. 7-68
References 7-68
8. Concluding Remarks 8-1
References 8-12
IX
-------
LIST OF FIGURES
Page
Figure 2-1. Ion chromatograph valve and column configuration. 2-8
Figure 2-2. Ion chromatograph of S203= and SCN~ in retort waters. . . . 2-13
Figure 2-3. Variation of adjusted retention time with eluent
strength. 2-15
Figure 2-4. Adjusted retention times for three eluents 2-16
Figure 2-5. Ion chromatogram of a retort wastewater using the
majors protocol ..... 2-19
Figure 2-6. Calibration curves obtained with the majors protocol
for N03~, S04 , S203 and SCN~ 2-21.
Figure 2-7. Calibration curves obtained with the protocol for late
eluting compounds 2-22
Figure 2-8. Ion chromatograms of standards obtained using the
early eluter protocol 2-24
Figure 2-9. Adjusted retention times for seven ions with
standard eluent (2.4 mM CO-" + 3.0 mM HC00~). . ... . . . 2-27
O O
Figure 2-10. Calibration curves for the early eluters. . . 2-42
Figure 5-1. Sample cell for measuring the uv absorption
of gaseous NH3 over a liquid sample 5-8
Figure 5-2. Ion chromatograms of Na+, NH4+ and K+ . 5-13
Figure 5-3. Effect of cell temperature on response.
(Standard concentration = 100 mg/1 NH,,
absorption wavelength = 197.4 nm) . . . .......... 5-17
Figure 5-4. Spectrum of NH2 in equilibrium with 100 mg/1
of NH4 at 60UC. (band pass = 0.6 nm). . 5-21
Figure 5-5. Calibration curves obtained from the gaseous
absorption method 5_22
Figure 5-6. Gas absorption spectrum over a retort wastewater 5-29
Figure 6-1. Equilibrium for the reaction S0? + % 09 = SO- 6-7
£ CO
Figure 6-2. Laboratory device for testing a flame as a
sulfur converter. 6-8
Figure 6-3. Total sulfur system operated during the fall,
1980 field test 6_14
Figure 6-4. Total S analyzer 6-21
X
-------
Figure 7-1. Column and valve configuration used during the
laboratory phase 7=14
Figure 7-2. I^S calibration curves obtained with the Dual
Flame Photometric Detector as a function of
peak area 7-18
Figure 7-3. HLS calibration curves obtained with the Dual
Flame Photometric Detector as a function of peak
height . 7-19
Figure 7-4. H2S calibration curves obtained with the Single Flame
Photometric Detector as a function of peak area 7-21
Figure 7-5. H£S calibration curves obtained with the Single Flame
Photometric Detector as a function of peak height 7-22
Figure 7-6. Quenching of the H£S Signal by co-eluting ethane for
the single flame Photometric Detector operated in the
normal mode 7-24
Figure 7-7. Quenching of the HLS signal by co-eluting ethane for
the Single Flame Pnotometric Detector operated in the
reverse mode 7-25
Figure 7-8. Quenching of the methyl mercaptan signal by co-eluting
butane for the Single Flame Photometric Detector operated
in the normal mode 7-27
Figure 7-9 Quenching of the methyl mercaptan signal by co-eluting
butane for the Single Flame Photometric Detector operated
in the reverse mode . . . . 7-28
Figure 7-10. Quenching of the H|S signal by co-eluting ethane for
the Dual Flame Photometric Detector. 7-30
Figure 7-11. Quenching of the methyl mercaptan signal by co-eluting
butane for the Dual Flame Photometric Detector 7-31
Figure 7-12. Retention times for the primary sulfur compounds with
acetone-washed Porapak QS 7-37
Figure 7-13. Retention times for the secondary sulfur compounds
with acetone-washed Porapak QS ..... 7-33
Figure 7-14. Retention times for hydrocarbons with the acetone-
washed Porapak QS : 7-39
Figure 7-15. Retention times for the primary sulfur compounds with 0V-
210 on Porapak QS ...... 7-40
Figure 7-16. Retention times for the secondary sulfur compounds
with OV-210 on acetone-washed Porapak QS 7-42
XI
-------
Figure 7-17. Retention times for hydrocarbons with OV-210
on acetone-washed Porapak QS ....... 7-43
Figure 7-18. Retention times for the primary sulfur compounds
with Chromosil 310 7-45
Figure 7-19. Retention times for the secondary sulfur compounds
__ with Chromosil 310 7-46
Figure 7-20. Retention times for hydrocarbons with Chromosil
310. 7-47
Figure 7-21. Retention times for the primary sulfur compounds
with Carbopack B HT 100 7-49
Figure 7-22. Retention times for the secondary sulfur compounds
with Carbopack B HT 100. 7-50
Figure 7-23. Retention times for hydrocarbons with Carbopack
B HT 100 7-51
Figure 7-24. Chromatogram of retort gas sample: 6' x 1/8"
Carbopack B HT 100 at 65 C isothermal 7-55
Figure 7-25. System for conditioning and supplying retort gas,
standards, and support gases to various analyzers 7-50
Figure 7-26. Column and valve configuration for monitoring
sulfur species in retort gases 7-65
Figure 8-1. Titration curve for a retort water sample. 8-4
XII
-------
LIST OF TABLES
Number Page
2-1 Oxidation by H202 of Various S Species to S04= . 2-29
3-1 Recovery of Sulfide at Various Concentrations
and in Various Matrices 3-10
3-2 Gran's Plot Outline 3-23
4-1 Total Solute by Freezing Point Depression 4-4
5-1 Comparison of Ammoniacal N Measurements
Obtained by Independent Methods 5-26
6-1 Conversion of Various S Compounds to S02
in Flame Converter 6-10
6A-1 Production or Consumption of Gases During Combustion . . . , . 6-25
7-1 Relative Response of the FPD to Four Sulfur Gases
with the Single-Flame FPD Operated in the Normal Mode 7-34
7-2 Integration and Valve Sequencing Programs for Figure 7-24. . . 7-56
8-1 Recommended Procedures for the Analysis
of Oil Shale Effluents . . 8-10
XIII
-------
ACKNOWLEDGEMENT
It is. indeed a pleasure to acknowledge those who made this effort possi-
ble, most notably the U.S. Environmental Protection Agency for financial
support and the project monitor, Robert Thurnau, for his continued confidence.
The authors are also indebted to the Laramie Energy Technology Center (Depart-
ment of Energy) for providing opportunities and financial support for .applying
many of the methods described in this report under realistic field conditions
in support of the test of various pollution control systems. This involvement
has helped provide direction to the effort reported here and has helped refine
and improve the methods discussed, especially those dealing with the measure-
ment of sulfur species and total sulfur in retort gas. Special thanks at this
institution go to Richard Poulson and Dave Sheesley for their advice and
encouragement.
XIV
-------
SECTION 1
INTRODUCTION
This report is written for chemists and engineers who are responsible for
developing pollution control technology for the oil shale and related synthetic
fuels industry. It describes several methods of chemical analysis which are
necessary in order to adequately test the various pollution control
technologies which are presently being proposed for the treatment of oil shale
retort gases and wastewaters. When possible, it includes step-by-step
protocols for determining important species, and in every case, it provides
supporting evidence and discussion to allow the analyst to adjust the procedure
to the varied sample types which he may encounter.
It has been recognized for some time that many methods of chemical
analysis which are appropriate for relatively simple stack gases and
wastewaters are not appropriate for retort gases and wastewaters (Fox et al. ,
1978; Felix et al., 1977; Farrier et al., 1977; Wildeman and Meglen, 1979;
Wildeman, 1979; Wallace, 1981). This situation arises largely because of the
large number of interfering materials found in many retort samples for which
the standard methods are not designed. For this reason, the U.S. Environmental
Protection Agency initiated an effort to develop improved methods of chemical
analysis for oil shale wastes. This effort is summarized in this and a
previous report (Wallace, 1981).
The objective of the work described in this report is to develop methods
of analysis which are helpful in selecting, monitoring and optimizing the
various pollution control systems which may be applied to oil shale processes.
In the case of wastewaters, present development plans call for "zero discharge"
of any contaminated water. The wastewater engineer, therefore, is concerned
1-1
-------
with treating the wastewaters for internal re-use and requires analytical
methods appropriate for this goal. This means that most methods must be
designed for the major and minor components (e.g. ammonia, thiosulfate, etc.)
and not necessarily for the same species which are normally of interest to the
regulator who is concerned with discharged waters. Retort gases, which are a
valuable by-product of the retorting process, represent a similar situation.
Such gases will either be burned for heat or otherwise consumed. Once the gas
is burned, the resulting flue gas would be similar to flue gases from any
number of other utility and industrial sources and should not present special
problems to the analyst. However, many development plans call for treating the
gas for sulfur and possibly ammonia removal prior to combustion. This requires
that the retort gas be analyzed, especially for the various sulfur species
which may be present. The measurement of sulfur species is thus one of the
procedures described in this report.
These objectives imply that the analytical methods which are developed
meet certain criteria. Methods for retort wastewaters must work well with
waters which are briney, turbid, colored and which contain high levels of
organic oils and solutes. Methods for retort gases must tolerate complex
mixtures of hydrocarbons, inert gases, water vapor, and entrained material. In
addition, the methods must be sufficiently rugged to operate in a field
laboratory where the environmental conditions are 1 ess-than-ideal. The
analyses described in this report are therefore designed to be used in a field
laboratory to provide rapid feedback to the plant operator to optimize system
performance. On the other hand, methods designed for these objectives are not
necessarily required to be as sensitive as methods designed for monitoring the
ambient environment or discharged streams.
1-2
-------
Most methods discussed in this report have been applied under field
conditions to a limited number of sample types with satisfactory results. For
example, application of the methods described to a series of retort wastewaters
herein resulted in satisfactory ion balances. At this stage it is prudent to
test these methods with a wider variety of waters and gases, an activity in
which the authors are presently engaged. The authors would also welcome
comments from other analysts regarding their experience with the methods
described in this report. It is hoped that these methods will be applicable
not only to oil shale retort process streams, but also to synfuel wastes
generally.
The recommended methods of analysis are summarized in the final chapter of
this report. For information-on additional methods, especially trace elements,
the reader is referred to the first report prepared in this project (Wallace,
1981).
REFERENCES
Farrier, D.S., R.E. Poulson, Q.D. Skinner, and J.C. Adams. 1977. Acqui-
sition, Processing, and Storage for Environmental Research of Aqueous
Effluents Derived from in Situ Oil Shale Processing. Proc. Second Pacific
Chemical Engineering Conference. Denver, CO.'
Felix, U.D., D.S. Farrier, and R.E. Poulson. 1977. High Performance Liquid
Chromatographic Characterization of Oil Shale Retort Waters. Proc. Second
Pacific Engineering Conference. Denver, CO.
Fox, J.P., D.S. Farrier, and R.E. Poulson. 1978. Chemical Characterization
and Analytical Considerations for an in Situ Oil Shale Process Water.
LETC/RI-78/7.
Wallace, J.R. 1981. The Analysis of Oil Shale Wastes: A Review. EPA-600/S7-
81-084. pb. 81-190.522..
Wildeman, T.R. March 1979. Sampling and Handling Of Oil Shale Solids And
Liquids. Oil Shale Symposium: Sampling, Analysis, and Quality Assurance.
Denver Research Institute. Denver, CO.
Wildeman, T.R., and R.R. Meglen. 1978. Analysis of Oil Shale Materials for
Element Balance Studies. In Analytical Chemistry of Liquid Fuel Sources.
P.C. Uden, and S. Siggia, eds. Adv. in Chem. Series 170. Washing-
ton D.C.
1-3
-------
SECTION 2
MINOR ANIONS IN RETORT WASTEWATER BY ION CHROMATOGRAPHY
Previous investigators have shown that, in addition to HC03~ and C03=,
several important minor anions also occur in retort wastewaters. For example,
Haas (1980) reported sulfide in the range of 500-1,000 mg/1 as the major anion
after HC03~ and CQ3= in condensate from the TOSCO II process. He also reported
total organic acids at levels of approximately 4,000 mg/1. On the other hand,
Fox et al. (1978) reported an absence of sulfide but the presence of the more
oxidized sulfur species S203=, S04=, and SCN~ in a retort wastewater from an
in-situ process. Fox's results for S03= were uncertain, ranging from <20 to
925 mg/1, due to the lack of adequate analytical techniques, while a single,
unconfirmed determination of S406= gave a level of 280 mg/1. On the other
hand, Stuber et al. (1978) reported S406= to be absent from most retort waters
(e.g., <5 mg/1). They also reported the major S species to be S203=, S04=, and
SCN~.
Previous work also has suggested the absence of several species. For
example, none of the investigators cited in the previous paragraph have
detected CN~, N02~, N03~, or P043~ in retort wastewaters, species whose
presence might be suspected in the condensate from a combustion source. It has
also been proposed that even if CN~ were initially present, it would be removed
rapidly by reaction with S203= to form SCN~ + S032~ (Luthy et al., 1977).
While Fox reported significant levels of Cl~, most other investigators report
Cl~ only at trace levels; it is thus likely that the Cl~ in Fox's samples was
due to the intrusion of ground water which was known to have occurred with
these samples. In the authors' experience, largely with retort wastewaters
from the 150-ton retort at the DOE facility in Laramie, the prevalent inorganic
2-1
-------
minor anions are S203-, S04-, and SCN . S03- is occasionally observed, for
example after waters have been exposed to H2S and air.
While the above-referenced studies indicate several minor species which
occur in retort water, they also indicate a degree of variability among waters
from different sources. These investigators, as well as others (Wallace, 1981)
also emphasized that many of the analytical techniques were not adequate to
determine the desired species unambiguously and quantitatively. This is
important because measurement of several of the species discussed above is
important to the design and operation of pollution control systems. In
addition, if unknown to the analyst, their presence can interfere with other
chemical analyses. Their measurement is also important for quality control,
since anion/cation balances cannot be completed until their concentrations are
known.
The objective of the work described herein was thus to develop reliable
methods of chemical analysis for the anions in retort wastewaters. The methods
were to be primarily for the purpose of evaluating and controlling the
operation of various pollution control systems under field conditions, and as
such were to be field-worthy and easily performed by a technician. Results
were to be available rapidly in order to aid in adjusting the operation of the
pollution control equipment. Another requirement was that the analyses require
a minimal amount of equipment and space, so that all necessary supplies and
equipment could be easily transported to a field site in a van or small
trailer.
The efficacy of both standard and less routine methods for analyzing
retort wastewaters has been reviewed by Wallace (1981), to which the reader is
referred for more detailed information. Suffice it to say that many methods of
analysis clearly needed to be improved or replaced. Others were not suitable
2-2
-------
for field use, requiring too much equipment or too much operator time. In many
cases, methods were simply not investigated or validated for retort waters in
terms of possible interferences, accuracy, or precision.
Because of the necessity of determining several of the minor species in
retort water, and because of the inadequacies of the available methods, efforts
were made to develop additional techniques for retort water. Ion
chromatography (Small et al. ; 1975) was selected as a likely technique because
of its ability to determine several species simultaneously: it is clearly
preferable to field one instrument rather than ten, and more feasible to
develop one method rather than several. In this regard, ion chromatography is
easily performed in field laboratories, since the instrument is easily
transported and minimal supporting apparatus is required.
The application of ion chromatography to complex samples has been
described by several authors. For example, Courtney (1976) applied ion
chromatography to the measurement of inorganic anions and cations in various
biological fluids. Hansen et al. (1979) reported the development of better
eluents for the determination of As043~ and S03= in leachates from flue dust
samples. They also found that S03= was difficult to determine by ion
chromatography because it readily oxidized during the chromatographic
separation. Holcombe et al. (1979) applied ion chromatography to the
measurement of S03=, S04=, and S203= in solutions from flue gas desulfurization
units. Two separate runs were required, the first with a weaker eluent for
S03= and S04=, and the second with a stronger eluent for S203=. McFadden and
Garland (1979), in attempting to develop an ion chromatographic method for oil
shale wastewaters, succeeded in separating the early eluters F~, Cl~, N02~,
P043~, S03=, and S04= on a single column; the later eluters, I", S203=, and
SCN~ were separated on a different column. Trujillo et al. (1981) measured
2-3
-------
S203= in leachate from spent oil shale with an ion chromatograph equipped with
a shortened column; however, all earlier peaks (F~ to S04=) eluted as a single
peak. For a more detailed review of the application of ion chrom'atography to
oil shale and other environmental samples, the reader is referred to Wallace
(1981), Mulik (1979), or Sawicki (1978).
In spite of these promising reports, early experience in the author's
laboratory indicated that further work remained before ion chromatography could
be applied routinely to the analysis of oil shale wastewaters. Although the
early eluters—F~, Cl~, N02~, P043~, S03~, and S04=—could indeed be separated
when injected as a pure standard, in real retort wastewaters the F~ and Cl~
peaks were often obscured by additional unknown species. The quantitation of
F~ and Cl~ was further complicated by the co-elution of the water dip and
carbonate dip (Small et a!., 1975). More importantly, after the first
injection of real retort waters, subsequent analyses were essentially
impossible because of a drifting baseline, presumably due to the elution of
highly retained retort water components. Another problem with the direct
application of ion chromatography to retort wastewaters was the tailing of
SCN~, which significantly decreased its detection limit and increased analysis
time.
The effort described in this report was thus directed towards developing
and improving ion chromatography for the routine and repetitive analysis of
retort wastewaters for non-carbonate anions.
Several different eluents and eluent additives were evaluated for the
purpose of improving the SCN~ peak. Solvent programming and non-standard
column configurations were also evaluated in an effort to elute S203= and SCN~
in reasonable time while still separating the early eluting compounds.
2-4
-------
Throughout this effort retort waters were run frequently in order to gain
experience with realistic samples.
In order to evaluate various eluents and eluent strengths as efficiently
as possible, it is helpful to use the expression (Breyer and Reiman, 1961)
(1) ta = t - tv = KAs/a
where t = adjusted retention time
u
t = absolute retention time
r
t = time for one void volume to elute
A = eluent concentration
s, a = valence of sample and eluent, respectively
K = constant for any given resin and eluent
This expression allows retention data to be plotted in a concise manner and
permits the estimate of retention times for a variety of conditions based on
relatively few measurements. The application of this approach is illustrated
in the discussion section of this report.
The following material describes the development of four separate ion
chromatographic assays for retort water. The first, termed the "majors
protocol", measures the major non-carbonate species normally encountered in
retort wastewater—S04=, S203=, and SCN~—as well as N03~. The second measures
the early eluters but misses S203= and SCN~. The third protocol detects SCN~
and S203= but misses the early eluters. The final protocol measures total S
after digestion with H202.
EXPERIMENTAL
All analyses were performed in a Dionex (Sunnyvale, CA) Model 10 ion
chromatograph equipped with an electrical conductivity detector. At the start
of this investigation the standard column sequence of injection valve -
2-5
-------
precolumn - analytical column - suppressor column - detector was employed. The
columns, also supplied by Dionex, were as follows: (1) precolumns, 6 mm x
50 mm, # 30986, fast run type; (2) analytical column, 6 mm x 250 mm, # 30985,
fast run type; (3) suppression columns, normal type, 12 mm x 100 mm, #30828.
All sample injections were made with a 100 pi sample loop. Peak area and
height measurements were accomplished on a Hewlett Packard (Avondale)
Model 3390 integrator.
Eluents and standard solutions were made from reagent grade chemicals-and
deionized, distilled water unless otherwise indicated. Stock standard
solutions, from which the working standards were prepared, were prepared at a
concentration of 10,000 mg/1, from their Na or K salts. Thiosulfate solutions
decomposed in a matter of hours unless prepared in freshly boiled, distilled,
deionized water, in which case they were stable for several months. Standards
for S03= were prepared from the formaldehyde-bisulfite addition compound
CH20:NaHS03 (Aldrich Chemicals). Stock standard solutions prepared from this
addition compound were stable for approximately 2 months although working
standards in the range 1-100 mg/1 had to be prepared fresh daily.
Samples of retort wastewater and steam-stripped retort wastewaters were
filtered after collection and stored at 5°C. Before injection into the sample
loop, the sample was diluted with d.i., distilled water if necessary and
injected through a 0.45 urn pore size membrane filter (Gelman Acrodisc #4184).
The conductivity detector was calibrated periodically with KC1 and
standards were run before and after each series of analyses and whenever
conditions were changed.
During the early stages of this investigation, a rudimentary form of
solvent programming was attempted in order to achieve an adequate separation of
the early eluters while still eluting S203= and SCN~ in reasonable time. Since
2-6
-------
the Diortex system is not equipped with solvent programming capabilities,
exponential solvent programming was achieved by feeding the high pressure pump
from a stirred beaker which in turn was fed from two reservoirs—one containing
a strong eluent and the other, the weak eluent. An exponential change in
solvent composition was achieved by switching from one reservoir to the other.
Various eluents, including a range of NaHC03 and Na2C03 concentrations,
were tested in order to find one capable of separating the early eluters while
still eluting SCN~ in a reasonable time. Eluent additives were also evaluated
for their ability to improve the SCN~ peak, which tended to tail badly with
most eluents. Eluents and additives included NaHC03 + Na2C03 over a range of
concentrations; 5 mM Na2C03/0.75 mM p-cyanophenol; 20 mM boric acid adjusted to
pH 11 with NaOH; 10 mM 4-amino-benzoic acid adjusted to pH 11 with NaOH; 7 mM
Na2C03 in 10% acetonitrile; 7 mM Na2C03/0.75 mM p-cyanophenol; 7 mM
Na2C03/0.5 mg/1 SCN~, added as NaSCN; 10 mM phthalamide adjusted to pH 11 with
NaOH; 10 mM n-butylboron adjusted to pH 11 with NaOH; 10 mM succinimide
adjusted to pH 11 with NaOH. Methylene-bis-thiocyanate and 2-thiohydantoin
were also evaluated as additives to NaC03.
Because of the long retention times and tailing of thiosulfate and
thiocyanate the valve arrangement in the Dionex Model 10 ion chromatograph was
modified to that shown in Figure 2-1. Addition of an 8 port slider valve
between the load/inject valve and the separator valve enables the sample to
progress through both pre-columns and main separator column or only through the
pre-columns. The air-driven slider valves are operated under program control
of the Hewlett Packard Model 3390 integrator, which operates through a Hewlett
Packard Model 19400A Event Control Module and home-made electronic/pneumatic
2-7
-------
Ul
«=r.
ce
o
o
o
CJ
Q.
ft
cc
1
o
or
i
CO
ce
CVJ
>
/\
00
*M
•5»
. «K
z o
UJ >
~~t llJ
w ul
cc
. a
3 0
UJ >
\ UJ
Ul *"
a:
1
/'-\
\
\
s
ro
^^
/\
ac
O
0 >
CM o:
X Ul
en
Ul
cc
II
cc
. 0
Ul CC
O UJ
UJ CO
CC Ul
2-5
-------
interface. With this design modification several different protocols were
designed and tested, as described below:
1. Major anions. This protocol allows the determination of the major
non-carbonate inorganic anions in retort wastewaters—S04=, S203=,
and SCN~—during a single run. N03~ is also determined.
For this protocol the sample is injected with both pre-columns
and separator columns in line. After four minutes the precolumn is
switched out of line (valve V6) leaving S203= and SON" on the pre-
column while the analysis of the earlier species continues on the
main analytical column. After the earlier peaks have eluted from the
main analytical column (approximately 15 minutes into the run), the
main analytical column is switched out of line and the pre-columns
are switched in line. S203= and SCN~ then elute from the pre-column
and proceed directly to the suppressor column. The entire analysis*
is complete in approximately 35 minutes. The eluent for this proce-
dure is 7.0 mM C03= at a flow rate of 2.3 ml/min.
2. Late eluters. S203= and SCN" are separated and analyzed with the
pre-columns switched in-line and the main analytical column switched
out of line using an eluent of 7 mM Na2C03/0.5 mg/1 NaSCN at a flow
rate of 2.3 ml/min. SzOs^ and SCN~ are well separated while early
components elute as a group. The analysis is complete in approxi-
mately 15 minutes.
3. Early eluters. The earlier eluting compounds (Cl~ through SOD are
separated using 3 mM Na2C03 at a flow rate of 2.3 ml/min. This
protocol begins with all columns in line. After the early eluters
have passed the pre-columns (Ca. 7 minutes), valve V6 is switched to
isolate the late eluters, S203= and SCN", on the column while the
2-9
-------
analysis of the early eluters proceeds. Immediately before the next
sample injection, the pre-columns are switched back in line to retain
S203= and SCN~ in the next sample.
If the pre-columns were never cleaned, the later eluters would
start to bleed onto the main analytical column. This is avoided by
rinsing the pre-column with 7 mM Na2C03 every third analysis. For
this purpose valves V7 and V8 are switched (Figure 2-1), thereby per-
mitting the late eluters to be rinsed to waste, a procedure requiring
approximately 10 minutes.
4. Total S. This procedure converts the various forms-of S to S04=
which is then determined. For this purpose 30% H202 is added to a
25 ml aliquot of sample in an open, stirred, beaker while measuring
the Eh with a Pt electrode. H202 is added until the Eh stabilizes at
approximately +0.4V vs. an Ag/AgCl electrode. (In practice, 5 ml of
H202 is normally more than sufficient.) After allowing the solution
to mix for several minutes, it is transferred to a hot plate and
boiled for 5 minutes to remove or decompose the H202. After cooling,
the solution is transferred to a 100 ml volumetric flask and brought
to volume. The resulting solution is analyzed for S04 using proce-
dure 1 above.
RESULTS AND DISCUSSION
In the Dionex system background conductivity due to the eluent is
minimized by passage through the suppressor column, which contains a
hydrogen-form cation exchange resin. Nevertheless, a residual conductivity
remains due to the partial dissociation of the weakly acidic eluents. In the
case of carbonate eluents, the background conductivity is typically 10-20 (j
Siemens while peak heights are typically 10-100 u Siemens.
2-10
-------
Original attempts to employ exponential solvent programming with
NaHC03/Na2C03 mixtures resulted in baseline shifts in the electrical
conductivity detector which were too large to permit quantitative measurements.
Attempts were made to minimize this problem by matching the conductivity of the
original and final eluents. In the case of a gradient from NaHC03 to Na2C03
such a matching can in theory be achieved simply by using equal-molar
concentrations of each, since the background conductivity is due only to the
amount of dissolved C02 eluting from the suppressor column. When this approach
was tried, the initial and final background conductivities were indeed the
same; however, an unacceptably large background peak occurred shortly after the
Na2C03 flow was started, obscuring many of the peaks of interest. This peak
can be explained by considering the action of C03= as it encounters the anion
exchange resin in the analytical column which has equilibrated with a dilute
HC03~ eluent. Since C03= is a stronger eluent than HC03~, it will displace
HC03~ from the resin according to the reaction
(2) 2(R-HC03-) + C03= -»• R2C03= + 2HC03~
As can be seen, this reaction causes a temporary increase in total carbonate
level and hence an increase in background conductivity in the detector. Under
the conditions tested, an exponential NaHC03/Na2C03 solvent gradient was
therefore unsatisfactory and was discontinued.
In order to minimize background shifts during gradient programming, a
search was made for other possible eluents. Ideally, such an eluent should
have a high pK in comparison to H2C03 (pK = 6.4) in order to minimize
3 cl
dissociation after the suppressor column. Also, it must be sufficiently
soluble in both the acid and ionized form, and be stable in the presence of
2-12
-------
air. Of course, a satisfactory eluent must also have a strong enough affinity
to the resin to elute the analytes rapidly enough.
Eluents tested for this purpose included hydroquinone (pK = 10.35), boric
a.
acid (pK = 9.1), 4-aminobenzoic acid, phthalimide, n-butylboron, succinimide,
a
and saccharin (2-sulfobenzoic acid imide). (Acids were converted to the basic
form by addition of NaOH.) Each of these proved to be unsatisfactory.
Hydroquinone promptly ruined the columns, presumably due to th'e irreversible
attachment of macromolecules which formed due to the exposure of the eluent to
air during the preparation process. The background conductivity of saccharin
was too high, most likely due to the presence of the sulfonic acid arising from
the decomposition of saccharin in basic solution. The other eluents either
were too weak or resulted in excessive background conductivity. While the
excessive conductivities may have been due to impurities, time did not permit a
more exhaustive study of alternate eluents.
Because of the apparent difficulty of finding an eluent system suitable
for solvent programming, efforts were directed tdWards optimizing the
NaHC03/Na2C03 eluent for retort waters. The first problem was the tailing of
the SCN~ peak with the standard eluent (2.4 mM NaHC03/3 mM Na2C03). To
minimize tailing two approaches were tried: (1) the use of stronger eluents,
and (2) the addition of trace species which resemble the SCN~. The latter
approach is based on the supposition that the SCN~ tailing is due to the
presence of a minority of sites which bind SCN~ more tightly and which may be
blocked through the action of similar species. Additives to Na2C03 included
p-cyanophenol, 2-thiohydantoin, acetonitrile, methylene-bis-thiocyanate, and
NaSCN. Of these, the first three additives showed no effect. The fourth
additive caused a slight improvement, but the most significant improvement was
for 0.5 mg/1 SCN~ in 7 mM Na2C03. Figure 2-2 compares the elution of SCN~
2-12
-------
START-}
A
TIME
START -
TIME
_ C2H302,C.f
SO=41N03
FIGURE 2-2
Ion chromatograph of S203= and SCN~ in retort waters
A. Eluent = 7mM Na2C03
B. Eluent = 7mM NaoCO, +0.5 mg/1 Na SCN
2-13
-------
using this eluent with two pre-columns. As can be seen, NaSCN both increases
the eluent strength and improves the SCN~ peak. For this reason the 7 mM
NaC03/0.5 mg/1 NaSCN eluent was chosen as optimum for SCN~.
In order to more systematically develop a separation with the
NaHC03/Na2C03 eluent system, the retention times for the ions of interest were
measured and plotted against eluent concentration, as shown in Figure 2-3. As
is illustrated for the case of Cl~, N02~, N03~, and S04=, the retention times
vary with eluent strength as predicted by equation 1. Thus, if the retention
times for the ions of interest are. known at one eluent strength and column
length, retention times can be predicted over a range of concentration and
column lengths. Next, this relationship was used to most efficiently select an
eluent to separate the desired species.
The utility of equation 1 is further illustrated by Figure 2-4, which
includes retention data for three different eluents: NaHC03, equimolar
NaHC03/Na2C03, and Na2C03. In this figure the abscissa for each of the eluents
is shifted in order to make the retention times overlap. One of the first
observations in Figure 2-4 is that the strength of the equimolar NaHC03/Na2C03
eluent is almost exactly one-half that of the Na2C03 eluent. This implies that
the NaHC03 component in the former solution is not active in displacing the
analyte ions from the resin. However, the NaHC03 component nevertheless
contributes to background conductivity and shortens the suppressor lifetime.
For these reasons NaHC03 was dropped from future eluents for most samples.
Figure 2-4 also suggests that much of the strength of the NaHC03 eluent is
due to the presence of carbonates. Were HCOg the only eluent ion binding to
the resin, the slopes for the monovalent ions would be -1. Instead, their
measured slopes are approximately -0.7, suggesting a mixed eluent system.
2-14
-------
10
Ul
•as.
LU
CK
O
t/5
O
A CHLORIDE
a NITRITE
• NITRATE
+ SULFATE
I I
I I
I I I I
I I i
1.0
ELUENT (No2C03) CONCENTRATION (mM)
10
FIGURE 2-3
Variation of adjusted retention time with eluent strength
2-15
-------
100
10
LU
2
ce
o
Q.
Na2C03
50^50 NagC03/NaHC03
NaHC0
ELUENT STRENGTH (mM//)
. . I
1.0
No2C03
I
5.0
10
50
1-0 50:50 Na2C03/NoHC03 5-(
J I
10
50
1.0
5.0
10.0
NoHCO.
50.0
FIGURE 2-4. ADJUSTED RETENTION TIMES FOR THREE ELUENTS
2-16
-------
A more thorough examination of the derivation of equation 1 (Reiman,
1961) would indicate that the constant K can be factored into two parts, K and
3
K , where K depends only on the sample ion and K , only on the eluent ion.
a.
Equation 1 can then be written as
(3)
log t = log K + log K - (f) log A
While the absolute values of Kg and Kg cannot be known, relative values of K
5
for a given column material can be assigned by injecting a series of ions while
keeping the eluent constant. Once relative values of K are assigned for one
eluent, other eluents can be evaluated, at least in theory, by measuring the
retention time of a single sample ion, at which point the retention times can
be calculated for the other ions.
In practice the approach described mathematically in the previous
paragraph can be most readily implemented graphically, as is illustrated in
Figure 2-4. Once retention times are measured at a few points, they can be
plotted linearly on log-log graph paper in order to establish the expected
retention times over a range of conditions. A new eluent can readily be
evaluated by injecting a few ions, at which point the corresponding retention
times for other ions can be graphically calculated by plotting the data as
shown in the figure. Retention times for other column lengths can also be
estimated since tr is proportional to column length, all else being equal.
While equation 3 and the associated graphical approach provide a
systematic basis for optimizing a separation, the practitioner must be aware of
additional factors. First, besides the eluent strength, other variables may be
manipulated in order to achieve the desired separation, including pH,
temperature, the presence of chelating agents, and column materials. Second,
2-17
-------
in the authors' experience, equation 3 is not necessarily obeyed exactly, and
therefore the chromatographic conditions must be tested close to the conditions
finally selected. Third, the previous paragraphs take no account of peak shape
or tailing, which must often be considered on an individual basis.
Using the approach described in the previous paragraph, four protocols
were developed for the ions encountered in retort wastewater. Since the late
eluters S203= and SCN~ could not be separated in a reasonable time under the
same conditions as the early eluters, the column switching procedure described
in the experimental section of this report was developed for the major species.
Figure 2-5 shows a chromatogram of a typical retort water sample analyzed
with the majors protocol. Although a peak is often observed near 4.9 minutes,
Cl~ and acetate are not adequately resolved in this region of the chromatogram
so that definitive assignment is not possible. (Occasionally retort waters are
encountered which give a much broader peak at 4/9 minutes, suggesting the
presence of mixed organic acids.) With the retort waters encountered in the
authors' laboratory, the region between the Cl~/Ac~ peak and the S04= peak is
normally empty, indicating the absence of S03~, N02~, and P043~. However, when
a peak occasionally does occur in this region, definitive assignment requires
the use of a weaker eluent. If enough information is known about the sample to
exclude two of the possible three ions in this region, analysis could proceed
for peaks occurring between Cl"/acetate peak and the S04= peak. As indicated
in the figure, S04=, S203=, and SCN~ are clearly resolved during a time period
of approximately 30 minutes. N03~, although normally not observed in retort
wastewaters in significant amounts, occurs immediately after S04=. The
separation of N03~ from S04~ is adequate for quantitation when NOg is present
in significant amounts.
2-18
-------
FIGURE 2-5.
Ion chromatogram of a retort wastewater using the majors protocol
2-19
-------
Figure 2-6 shows calibration curves for Cl , N03 , S04-, S203-, and SCN
obtained with the majors protocol. For these species the calibration curves
are linear over the range 3-100 mg/1. At higher concentrations the retention
times tend to decrease, especially for the late eluters, making peak
identification less certain. At lower levels baseline drift limits
detectability. Either peak height or peak area can be used for quantisation,
although peak height is normally less subject to interference due to nearby
peaks.
When S203= and SCN~ are the only species of interest, the most rapid
method is to use 7 mM NaC03 with 0.5 mM NaSCN as the eluent, which minimizes
tailing of the SCN~ peak, with only the two pre-columns in line. Under these
conditions acetate through nitrate elute unresolved in about 4 minutes, while
S203= and SCN" elute in approximately 8 and 15 minutes respectively
(Figure 2-2b). Figure 2-7 shows calibration curves over the concentration
range 1-300 mg/1. While linear over most of the range, slight curvature is
apparent between 100 and 300 mg/1. This curvature is most likely explained by
initial column saturation effects, especially since elution times start to
decrease at 300 mg/1.
The third protocol is for the early eluters, acetate - S04=, including
N03~, which elutes before S04= with the eluent used for this protocol. As
described in the Experimental section of this report, and as normally practiced
in the authors' laboratory, this procedure employs 3 mM Na2C03. However, as
will be seen in the following discussion, this eluent concentration is not
appropriate in every situation and must be adapted to sample requirements,
although the valving sequence can remain basically unchanged. It is therefore
proposed that the analyst use this basic procedure for measuring the early
eluters in retort water, but that he be prepared to alter the eluent to meet
2-20
-------
C/3
OS
as
o
Ul
UJ
Q.
IxlO
—A— CHLORIDE
—•— SULFATE
—o— NITRATE
—o~ THIOSULFATE
—*— THIOSYNATE
j—i—i
j i i i i
10
30
100
CONCENTRATION (mg/e)
FIGURE 2-6
Calibration curves obtained with the majors protocol for
Cl, NOg, SO^, SgOg, and SCN"
300
2-21
-------
1x10 V
1x10'
THIOSULFATE
THIOCYNATE
10 30 100
CONCENTRAflON (mq/1]
FIGURE 2-7
Calibration curves obtained with the protocol for late eluting
compounds
300
2-22
-------
sample requirements. The following discussion summarizes information of which
the analyst should be aware to obtain acceptable results.
Figure 2-8 contains chromatograms obtained with this protocol using two
different eluents. The first chromatogram shows an adequate separation of
acetate, Cl~, N02-, S03=, and S04= using 3 mM Na2C03. Comparison of the first
and second chromatograms illustrates one of the first problems with the early
eluters: the position of the acetate peak depends not only on the eluent
strength, but also on the condition of the suppressor column, moving to an
earlier position as the suppressor column is exhausted. This phenomenon is due
to the retention of acetate by the hydrogen-form but not the sodium-form of
resin in the suppressor column by ion exclusion effects. The analyst hoping to
determine both acetate and Cl~ clearly must consider this possible interfer-
ence. Another potential difficulty with Cl" and acetate is that their position
is near the water and carbonate dips, which appear as negative peaks near the
start of the chromatogram. Depending on the concentration range of acetate and
Cl~, these peaks must be separated through proper selection of eluent and
column in order to obtain quantitative results (Stevens et al., 1981). In
practice, interferences due to carbonate and acetate can be minimized by
minimizing ion exclusion effects in the suppressor columns, for example, by
using a sufficiently short suppressor column. (Unfortunately, shorter sup-
pressor columns require more frequent regeneration.) Another approach to
eliminating the migration of the carbonate and acetate peaks is the use of a
fiber suppressor column, for which retention by ion exclusion does not occur
(Stevens et al., 1981). Thus, while the analysis for Cl" and acetate should be
approached with caution, careful column selection should make these determina-
tions feasible.
2-23
-------
FIGURE 2-8
(Cont'd on next page)
Ion chromatograms of standards obtained using the early
eluter protocol
A. eluent = 3 mM Na2C039 suppressor column almost spent
B. same conditions as A, regenerated suppressed column
.C. standard eluent (2.4 mM Na2C03/3.0 mM NaHC03)
D. same as C, but sample also contains SO"
2-24
-------
Cl
S04=
STOP
— TIME
TIME
25.0 min.
.CJL~
-—TIME
22.0min.
TIME
22.0min.
2-25
-------
Figure 2-8d illustrates another potential interference. With a 3 mM Na2C03
eluent, S03= co-elutes with N03~ and P043~. However, for the retort waters
encountered in the authors' laboratory, this potential interference has not
posed a problem since the retort wastewaters encountered have contained neither
P043~ nor N03~ in significant amounts. Furthermore, this protocol is normally
carried out only after the "majors analysis" has been completed, so that the
absence of N03~ has already been established. However, should P043~ be
present, perhaps as an additive to a biological reactor, its peak can be
displaced by adjusting the pH, as is illustrated in Figure 2-8c. In this
figure acetate, Cl~, N02~, P043~, S03=, and S04= are separated using an eluent
of 2.4 mM Na2C03/3.0-mM NaHC03, although N03~ would still not be adequately
separated from S03=.
Figure 2-9 illustrates the adjusted retention times for seven early
eluters obtained with eluents of various strengths containing Na2C03 and NaHC03
in a ratio of 2.4/3.0 on a fast run column. In using this diagram it should be
kept in mind that the position of the P043~ peak can also be adjusted relative
to the other peaks by varying the pH of the eluent. This diagram suggests that
the best separation of the seven early eluters could be obtained with an eluent
of 1.4 mM Na2C03/1.8 mM NaHC03. (Such a separation is shown in Figure 2-8d and
indicates that the separation of NOi and Sol is not yet complete.) Figure 2-10
demonstrates calibration curves for the early eluters (with the exception of
N03~) obtained with this protocol.
The measurement of total sulfur, the fourth procedure reported, serves as
a check on the completeness of the analyses of the individual S species. This
procedure, as described in the Experimental section of this report, involves no
new chromatographic procedure, but only an oxidation with H202.
2-26
-------
100.0
10.0
e
"I
CO
UJ
as
O
LU
Ul
IK
tn
=3
"3
o—
o —
o—
'o—
o—
o—
o
•o •
-o-
• o-
• o-
-D o ACETATE
- + o CHLORIDE
o NITRITE
o PHOSPHATE
o SULFITE
.__.o o NITRATE
-o —o SULFATE
1.0
o.r
j_
I/3X IX 3X
( X TIMES ELUENT STRENGTH STD. ELUENT
FIGURE 2-9
Adjusted retention times for seven ions with
standard eluent (2.4 mM COg + 3.0 mM HCOj)
2-27
-------
The effect of this oxidation is shown in Table 2-1 for four retort water
samples. The original concentrations of S203=, S04=, and S03= were measured
using the "majors protocol"; S= and S03=, which were absent in the original
samples, were added using standard solutions. Closure is within experimental
error, which supports the validity of the individual determinations obtained
with the "majors protocol", as well as the oxidation procedure. Because of the
possibility of occurrence of other, unexpected S species in retort wastewaters,
it is recommended that the total S measurement be run periodically.
In the authors' laboratory, retort waters are analyzed routinely by ion
chromatography for S04=, N03~, $203=, and SOT and occasionally for Cl~, using
the procedures described above. Other than the occasional replacement of a
precolumn, no unusual problems have arisen. The precision for S203=, SCN~, and
S04= averages 3% relative (1 a) at naturally occurring concentrations. The
precision (1 a) of the N03~ measurement is poorer (9% relative) at naturally
occurring levels, as would be expected for the low levels normally encountered
(Figure 2-5). The precision of the Cl~ measurement averages 8% (1 a), which is
somewhat less than ideal due to the variable occurrence of the interfering
carbonate dip. Recovery of S04=, N03~, S203=, SCN~, and Cl~ spiked into retort
waters is normally 100% within experimental error. As of this writing,
insufficient data have been gathered to establish statistical parameters for
other species.
PROTOCOL FOR MAJOR ANIONS ANALYSIS
Principle
The major anions in retort wastewaters are separated and quantitated by
ion chromatography, a'form of high pressure liquid chromatography. In this
method the various anions are separated on a surface-agglomerated anion
exchange resin with an eluent of sodium carbonate. The eluent from the first
2-28
-------
Table 2-1
OXIDATION BY H202 OF VARIOUS S SPECIES TO S04=
Sample No. S-8 S-5 S-6 10/6 Feed
Original Species
S04= (mg/1)
S203= (mg/1 )
SCN- (mg/1)
S03= (mg/1)
S= (mg/1)
15.5
45.0
7.3
50.0
50.0
18.0
50.0
7.5
100.0
100.0
18.0
32.5
6.3
50.0
50.0
33.5
24.0
5
30
Theoretical yield of S04= (mg/1) '314 537 294 119
Measured yield of 504= (mg/1) 340 525 315 125
Measured yield of S04= (%) 108 98 107 105
2-29
-------
(separator) column proceeds to a second ion exchange column in the H+ form
where the eluent is converted to the weakly ionized hydrogen form. From the
second column the eluent proceeds to a low dead-volume electrical conductivity
cell where the analytes are detected as their strong acids.
For this protocol the major anions elute in the order sulfate, nitrate,
thiosulfate, and thiocyanate. (Carbonate/bicarbonate are not detected.)
Sulfite is also detected prior to the sulfate peak but has not yet been shown
to be entirely interference free. The earlier components (e.g. acetate,
chloride, phosphate, nitrite) elute as an unresolved group. Column switching
is necessary to achieve the separation described in this protocol.
Equipment
Dionex Ion Chromatograph with auxiliary valve installed as shown in
Figure 2-1
Integrator - such as HP 3390
2 - Dionex Fast Run Pre-Columns 3x150 mm .
1 - Dionex Fast Run Anion Column 3x500 mm
1 - Dionex Anion Suppressor Column 6x25 mm
3-4 liter carboys
1-1 liter volumetric flasks
10 - 100 milliliter volumetric flasks
1 - 500, 100 and 10 milliliter graduated cylinder
10 cubic centimeter disposable syringes
0.45 urn syringe filters
1 - variable pipettor 100-1,000 pi
pipettors - 10 ul and 30 pi
3 cycle logarithmic paper
2-30
-------
Reagents
1. 0.01 N KC1 - Potassium Chloride Stock Standard
Dry KC1 in oven at 105° to constant weight at 20°C. Weigh out
0.7456 grams of KCL and add to 500 ml of distilled water in a 1 liter
volumetric flask. Shake to dissolve KC1 and bring up to volume with
distilled water.
2. 0.001 N KC1 - Potassium Chloride Calibration Standard
Measure 400 ml of 0.01 N KC1 stock standard and add.to 3,500 ml
distilled water in 4 liter carboy. Mix solution and dilute to volume
with distilled water.
3. 0.007 M Na2C03
Measure 2.3744 grams of ACS grade Na2C03 on an analytical balance.
Add to distilled water in 4 liter carboy and dilute to volume with
distilled water.
4. 1 M H2S04 - Regeneration Solution
Measure 111 ml of concentrated sulfuric acid with graduated cylinders.
Dilute to 4 liters in carboy using distilled water.
5. Anion Stock Standards - 10,000 mg/1
a. Chloride - measure 1.6484 grams NaCl on an analytical balance.
Dissolve in 100 ml volumetric flask using distilled water.
Dilute to volume.
b. Sulfite - measure 1.7089 grams 98% formaldehyde sodium bisulfite
addition complex and dissolve in 100 ml volumetric flask with
distilled water. Bring up to volume with distilled water.
c. Sulfate - measure 1.8142 grams K2S04 on an analytical balance.
Dissolve in 100 ml volumetric and bring up to volume with
distilled water.
2-32
-------
d. Nitrate - measure 1.3708 grams NaN03 on an analytical balance.
Dissolve in a 100 ml volumetric flask and bring up to volume with
distilled water.
e. Thiosulfate - measure 1.4101 grams Na2S203 on an analytical
balance. Dissolve in a 100 ml volumetric flask and bring up to
volume with distilled water.
f. Thiocyanate - measure 1.3958 grams sodium thiocyanate on an
analytical balance. Dissolve in a 100 ml volumetric flask and'
bring up to volume with distilled water.
6. Working Standards
1 ppm - using pipettor add 10 |jl of each anion stock standard to a
100 ml volumetric and dilute to volume with distilled water.
3 ppm - as above using 30 pi pipettor
10 ppm - 100 pi each
»
30 ppm - 300 pi each
100 ppm - 1,000 (jl each
Procedure
1. Pretreatment of Sample. Dilute sample if a concentration of greater
than 100 mg/1 of any component of interest is expected.
2. Preliminary Procedures
a. Calibrate conductivity meter using 0.001 N KC1 solution with
pre-columns, anion column and anion suppressor column out of
line. Reading should be 147 ± 2 pmhos/cm.. If not within
acceptable range, follow conductivity meter instructions for
correcting the reading.
b. Prepare calibration plot using the following protocol. The
system should be set up so that the pre-columns, the main
2-32
-------
separator column and the suppressor column are separately
controlled (Figure 2-1). Before injection of a standard, all
three columns should be in the sample stream. Four minutes after
injection the pre-column should be taken out of line to trap the
thiosulfate and thiocyanate on the pre-column. Chloride,
sulfide, sulfate and nitrate will elute in 11 minutes. At that
time the main separator should be switched out of line and the
pre-columns back in line. The thiosulfate and thiocyanate will
now elute. The complete analysis takes approximately 35 minutes.
The anion standards should be run twice for each concentration
and the resulting peak height plotted against the concentration
on log-log paper. The calibration graph should be linear with a
slope close to one.
3. Sample analysis. Inject diluted sample onto sample loop through a
0.45 ul syringe filter to remove particulate matter. Follow the same
procedure as for the anion standards. The concentrations are deter-
mined by comparing the peak heights to the calibration curve.
Interferences
Acetate and other weakly acidic organic acids can coelute under some
conditions with chloride. Also, a carbonate dip can occur if the carbonate
concentration in the sample is much smaller than in the eluent. This dip
occurs just before the chloride peak and can interfere with chloride peak
height measurement. It can be minimized by adjusting the sample carbonate
concentration to equal the eluent. Acetate will sometimes co-elute with
chloride depending on the state of the suppressor column. (See text).
2-33
-------
Species eluting prior to SOj are not entirely resolved, so that in the
presence of more than one of these anions the sample must be run using a weaker
eluent such as 1.8 mM NaHC03 and 1.4 mM Na2C03.
PROTOCOL FOR THIOSULFATE AND THIOCYANATE
Principle
Thiosulfate and thiocyanate are separated and quantitated by ion
chromatography, a form of high pressure liquid chromatography. In this method
the various anions are separated on a surface agglomerated anion exchange resin
with an eluent of sodium carbonate/Na SCN. The eluent from the separator
column proceeds to a second ion exchange column in the H+ form where the eluent
is converted to the weakly ionized carbonic acid and the sample ions are
converted to the hydrogen form. From the second column the eluent proceeds to
a low dead-volume electrical conductivity cell where the analytes are detected
as their ions.
Equipment
Dionex Ion Chromatograph
Integrator - such as HP 3390
2 - Dionex Fast Run Pre-Columns 3x150 mm
1 - Dionex Anion Suppressor Column 6x25 mm
3-4 liter carboys
1-1 liter volumetric flask
4 - 100 milliliter volumetric flasks
1 - 500, 100, and 10 milliliter graduated cylinder
10 cubic centimeter disposable syringes
0.45 urn syringe filters
2-34
-------
1 - variable pipettor 100-1,000 ul
1 - pipettor - 10 pi and 30 pi
3 cycle logarithmic paper
Reagents
1. 0.01 N KC1 - Potassium Chloride Stock Standard
Dry KC1 in oven at 105° to constant weight at 20°C. Weigh out
0.7456 grams of KC1 and add to 500 ml of distilled water in a 1 liter
volumetric flask. Shake to dissolve KC1 and bring up to volume with
distilled water.
2. 0.001 N KC1 - Potassium Chloride Calibration Standard
Measure 400 ml of 0.01 N KC1 stock standard and add to 3,500 ml
distilled water in 4 liter carboy. Mix solution and dilute to volume
with distilled water.
3. 0.007 M Na2C03
Measure 2.3744 grams of ACS grade Na2C03 on an analytical balance.
Add to distilled water in 4 liter carboy and dilute to volume with
distilled water.
4. 1 M H2S04 - Regeneration Solution
Measure 111 ml of concentrated sulfuric acid with graduated cylinders.
Dilute to 4 liters in carboy using distilled water.
5. Anion Stock Standards - 10,000 ppm
a. Thiosulfate - measure 1.4101 grams Na2S203 on an analytical
balance. Dissolve in 100 ml volumetric and bring up to volume
with distilled water.
b. Thiocyanate - measure 1.3958 grams NaSCN on an analytical
balance. Dissolve in 100 ml volumetric and bring up to volume
with distilled water.
2-35
-------
6. Working Standards
1 ppm - using pipettor add 10 (jl of each anion stock standard to a
100 ml volumetric and dilute to volume with distilled water.
3 ppm - as above using 30 pi pipettor
10 ppm - 100 |jl each
30 ppm - 300 jjl each
100 ppm - 1,000 nl each
Procedure
1. Pretreatment of Sample: dilute sample if a concentration greater than
100 mg/1 of any component of interest is expected.
2. Preliminary Procedures
a. Calibrate conductivity meter using 0.001 N KC1 solution with
pre-columns, anion column and anion suppressor column out of
line. Reading should be 147 ± 2 umhos/cm. If not within accept-
able range, follow conductivity meter instructions for correcting
the reading.
b. Prepare calibration graph by running the anion standard series
twice and plotting the resultant peak heights versus the anion
concentrations on logarithmic paper. Note that only the two
pre-columns are used as separator columns. Each analysis should
take approximately 20 minutes.
3. Sample Analysis: inject diluted sample onto sample loop through a
0.45 urn syringe filter to remove particulate matter. Follow the same
procedure as for the anion standards. The actual concentrations are
determined by comparing the peak heights to the calibration curve.
2-36
-------
PROTOCOL FOR EARLY ELUTING SPECIES
Principle
The major anions in retort wastewaters are separated and quantitated by
ion chromatography, a form of high pressure liquid chromatography. In this
method the various anions are separated on a surface-agglomerated am"on
exchange resin with an eluent of sodium carbonate. The eluent from the first
(separator) column proceeds to a second ion exchange column in the H+ form
where the eluent is converted to the weakly ionized hyrogen form. From the
second column the eluent proceeds to a low dead volume electrical conductivity
cell where the analytes are detected as their strong acids.
For this protocol the late eluters, which would otherwise bleed from the .
column during subsequent analyses, are trapped on a precolumn while the early
eluters proceed to analysis. The precolumn is periodically flushed to waste
without upsetting the baseline from the analytical column. This procedure
requires that"a timed valve switching sequence be performed during the
separation. While this valve sequence can remain unchanged, the analyst should
be prepared to adjust the eluent strength slightly (based on sample character-
istics) in order to achieve the optimum separation.
Equipment
Dionex Ion Chromatograph with valves and columns arranged as shown in
Figure 2-1
Integrator - such as HP 3390
2 - Dionex Fast Run Pre-Columns 3x150 mm
1 - Dionex Fast Run Am"on Column 3x500 mm
1 - Dionex Anion Suppressor Column 6x25 mm
3-4 liter carboys
2-37
-------
1-1 liter volumetric flask
6 - 100 milliliter volumetric flasks
1 - 500, 100 and 10 milliliter graduated cylinder
10 cubic centimeter disposal syringes
0.45 urn syringe filters
1 - variable pipettor 100-1,000 pi
1 - pipettor - 10 ul and 30 pi
3 cycle logarithmic paper
Reagents
1. 0.01 N KC1 - Potassium Chloride Stock Standard
Dry KC1 in oven at 105° to constant weight at 20°C. Weigh
0.7456 grams of KC1 and add to 500 ml of distilled water in a 1 liter
volumetric flask. Shake to dissolve KC1 and-bring to volume with
distilled water.
2. 0.001 N KC1 - Potassium Chloride Calibration Standard
Measure 400 ml of 0.01 N KC1 stock standard and add to 3,500 ml
distilled water in 4 liter carboy. Mix solution and dilute to volume
with distilled water.
3. 0.003 M Na2C03 - Weak Eluent •
Measure 1.0176 grams of ACS grade Na2C03 on an analytical balance.
Add to distilled water in 4 liter carboy and dilute to volume with
distilled water.
4. 1 M H2S04 - Regeneration Solution
Measure 111 ml of concentrated sulfuric acid with graduated cylinders.
Dilute to 4 liters in carboy using distilled water.
2-38
-------
5. Anion Stock Standards - 10,000 mg/1
a. Chloride - measure 1.6484 grams NaCl on an analytical balance.
Dissolve in 100 ml volumetric using distilled water. Dilute to
volume.
b. Sulfite - measure 1.7089 grams 98% formaldehyde sodium bisulfite
addition complex and dissolve in 100 ml volumetric flask with
distilled water. Bring to volume with distilled water.
c. Sulfate - measure 1.8142 grams K2S04 on an analytical balance.
Dissolve in 100 ml volumetric and bring to volume with distilled
water.
6. Working Standards
1 mg/1 - using pipettor add 10 ul of each anion stock standard to a
100 ml volumetric and dilute to volume with distilled water.
3 mg/1 - as above using 30 ul pipettor
10 mg/1 - 100 pi each
30 mg/1 - 300 ul each
100 mg/1 - 1,000 ul each
Procedure
1. Pretreatment of Sample: Dilute sample if a concentration greater than
100 mg/1 of any component of interest is expected.
2. Preliminary Procedures
a. Calibrate conductivity meter using 0.001 N KC1 solution with
pre-columns, anion column and anion suppressor column out of
line. Reading should be 147 ± 2 umhos/cm. If not within
acceptable range, follow conductivity meter instructions for
correcting the reading.
2-39
-------
b. Prepare a calibration curve using the working am"on standards.
With the columns in line, pump eluent through system until
baseline stabilizes. Zero the meter. Flush sample loop and
inject 1 ppm standard. The standard run should take approxi-
mately 15 minutes. Complete the am"on standard series, running
each twice. Plot the resulting peak height vs. concentration on
log-log paper. The calibration curve should be linear and close
to a slope of one.
3. Sample Analysis
Inject diluted sample onto sample loop through a 0.45 }jm syringe
filter to remove particulate matter. Determine concentrations by
comparing the peak heights- to the calibration curve.
Interferences
Acetate and other weakly acidic organic acids can coelute under some
conditions with chloride. Also, a carbonate dip can occur if the carbonate
concentration in the sample is much smaller than in the eluent. This dip
occurs just before the chloride peak and can interfere with chloride peak
height measurement. It can be minimized by adjusting the sample carbonate
concentration to equal the eluent.
Phosphate, sulfite and nitrate coelute with 3 mM Na2Co3 eluent, so in the
presence of more than one of these anions the use of another eluent such as
1.8 mM NaHC03, 2.4 mM Na2C03 is recommended.
PROTOCOL FOR TOTAL SULFUR
Principle ,
Total sulfur in water is measured by oxidation of the sulfur species to
sulfate and quantisation of the sulfate by ion chromatography.
2-40
-------
Equipment
Electrometer - reading to 0.1 volts
Platinum wire electrode
Reference electrode
Magnetic stirrer
stir bars
buret - 10 ml
25 ml pipets
100 ml beakers
hot plate, watch glass
100 ml volumetric
Reagents
30% H202 (fresh)
Sample Pre-treatment
Put 25 ml of sample into a 100 ml beaker. Place platinum wire electrode
and reference electrode in the sample. Connect electrodes to electrometer.
Stir the sample with a magnetic stirrer. Note the initial electrical
potential, then begin adding 30% H202 dropwise from the buret. Continue adding
the 30% H202 until the oxidation potential is +0.4V. At this potential all
sulfur species will have oxidized to S04.
Heat the oxidized sample to a slow boil, covering it with a watch glass to
prevent loss of sample. Boil for 10 minutes to remove excess H202, then cool
to room temperature. Pour into 100 milliliter volumetric flask. Wash the
beaker with distilled water, pouring washes into 100 ml volumetric also. Add
distilled water to bring volume up to mark.
2-41
-------
uioV
UIO
=>
>-
-------
Procedure
Analyze sample for sulfate using weak eluent. (See procedure for early
eluters.) Correct calculated concentration for the dilution.
REFERENCES
Anderson, C. 1976. Ion Chromatography: a new technique for clinical
chemistry. Clin. Chem., 22(9):1424.
Fox, J.P., D.S. Farrier, R.E. Poulson. 1978. Chemical Characterization and
Analytical Considerations for.an In-Situ Oil Shale Process Water. Laramie
Energy Technology Center Report No. LETC/RI-78-7.
Hansen, L.D., B.E. Richter, O.K. Rollins, J.D. Lamb, D.J. Eutough. 1979.
Determination of As and Species in Environmental Samples by Ion
Chromatography. Anal. Chem., 51(6):633.
Haas, F.C. 1980. Analysis of Tosco II Oil Shale Retort Water in Analysis of
Water Associated with Alternate Fuel Production. L.P. Jackso'n,
C.C. Wright, eds. ASTM Philadelphia, PA.
Holcombe, L.J., B.F. Jones, E.E. Ellsworth, F.B. Meserole. 1979. The
Quantitative Determination of Aqueous Solutions of Sulfite, Sulfate and
Thiosulfate using the Ion Chromatograph, in Ion Chromatographic Analysis
of Environmental Pollutants. Vol. II. J.D. Mulik, F. Sawicki, eds. Ann
Arbor Science, Ann Arbor.
Luthy, R.G., S.G. Bruce, Jr., R.W. Walters, D.U. Nakles. 1977. Identification
and Reactions of Cyanide and Thiocynate in Coal Gasification Wastewaters.
Proceedings of the 50th Annual Conference of the Water Pollution Control
Federation. Philadelphia, PA. October 1977.
McFadden, K.M., T.R. Garland. 1979. Determination of Species in Oil Shale
Waste Waters by Ion Chromatography. 34th Regional ACS meeting, Reckhard,
Washington. June 13-15, 1979.
Miner, W. ed. 1953. Ion Exchange Resins in Medicine and Biological Research.
Annals, of the New York Academy of Sciences, 57:61.
Mulik, J.D., E. Sawicki. ed. 1979. Ion Chromatographic Analysis of
Environmental Pollutants, Vol. 2. Ann Arbor Science, Ann Arbor.
Peterson, S. 1954. Anion Exchange Processes. Annals of the New York Academy
of Sciences, 57:144.
Reimar, III, W. 1961. Chromatography: Columnar Liquid-Solid Ion-Exchange
Process, in Treatise on Analytical Chemistry. Part I, Vol. 3.
J.M. Kolthoff, P.J. Elving, eds. Interscience, NY.
Sawicki, E. , J.D. Mulik, E. Willgenstein, eds. 1978. Ion Chromatographic
Analysis of Environmental Pollutants. Ann Arbor Science, Ann Arbor.
2-43
-------
Small, H. , T.S. Stevens, W.C. Bauman. 1975. Novel Ion ' Exchange
Chromatographic Method Using Conduct!metric Detection. Anal. Chem.,
47:1801.
Steinle, K. 1962. Uber die Bestandteile der Wackenroderschen Flussigkeit und
ihren Biedungsmechanismus. Ph.D. dissertation.
Stevens, T.S., J.C. Davis, H. Small. 1981. Hollow Fiber Ion-Exchange
Suppressor for Ion Chromatography. Anal. Chem., 53:1488.
Stuber, H.A., J.A. Leenheer, D.S. Farrier, 1978. Inorganic Sulfur Species in
Wastewaters from in Situ Oil Shale Processing. J. Environ. Sci. Health.
Part A, A13:663-675.
Trujillo, F.I., M.M. Miller, R.K. Shogerboe, H.E. Taylor, C.H. Grant. 1981.
Ion Chromatographic Determination of Thiosulfate in Oil Shale Leachates.
Anal. Chem., 53:1944.
Wallace, J.R. 1981. The Analysis of Oil Shale Wastes: A Review. EPA Report
No. EPA-600/57-81-084.
J& —*T
-------
SECTION 3
THE DETERMINATION OF SULFIDE IN OIL SHALE WASTEWATERS
As with most synthetic fuel projects, oil shale retorting produces a
series of wastewaters which contain principally dissolved C02 and NH3, but also
numerous inorganic and organic ions. Total dissolved material may be in the
range of several tens of grams per liter, presenting numerous potential
problems to the analyst in terms of interferences. Present development plans
for oil shale plants call for "zero discharge" of contaminated retort
wastewaters, so that the retort wastewaters must be treated for internal
re-use, a goal which requires the application of reliable and rapid chemical
analyses of the process waters before, during, and after the various treatment
and application steps.
Although sulfide is one of the species of environmental concern, the
available evidence suggests that the standard methods of chemical analysis for
sulfide are not appropriate fof- retort wastewaters, or for that matter,
wastewaters from other synthetic fossil fuel processes (Wallace, 1981). For
example, the iodometric method responds to any reducing agent which reacts with
iodine, including the various sulfur oxides which are found in retort waters.
Other possible interferences in retort water would include ammonia and
unsaturated organic compounds. The methylene blue method is also subject to
interferences from reducing compounds; in addition, since retort waters are
generally highly colored and turbid, such colorimetric methods are generally
contraindicated. Because of these interferences, methods have been developed
for separating sulfide from the rest of the sample. For example, Kolthoff and
Elving (1961) recommend acidifying the sample and purging the resulting H2S
into a bubbler containing zinc acetate. The resulting ZnS precipitate can then
3-1
-------
be analyzed by either standard method. Similarly, the APHA (1976) recommends
separating sulfide by adding zinc acetate and removing the ZnS precipitate.
Either separation method is most likely to succeed when sulfide is the only
form of S present. However, for retort waters, which contain other reduced S
compounds, the removal of sulfide may upset the equilibrium and produce
additional sulfide. In addition, it is not clear whether these separation
techniques cleanly remove potentially interfering organic compounds which may
either co-distill or co-precipitate.
These difficulties may explain why Fox et al. (1978) obtained erratic
results in their efforts to measure sulfide in retort water with the
distillation/iodometric titration. In any case, even if distillation and
precipitation procedures may be helpful in the laboratory, they are too
cumbersome and time consuming for field use.
In order to avoid these problems with the standard methods, several
investigators have turned to a potentiometric titration for the determination
of sulfide in complex, colored, and turbid wastewaters from other sources.
These methods generally involve titrating with Pb, Ag, or Cd while monitoring
the precipitation with the AgS membrane electrode, although several variations
have been reported in the literature.
While any titrant forming a metal sulfide precipitate could be used, the
selectivity of the titration with retort wastewaters is improved by selecting a
titrant which reacts only with sulfide. For a Ag+ titrant, Schmidt and
Punger (1971) considered a number of possible interferences, including SCN~ and
S203=, both of which occur in retort wastewaters. The latter two species were
shown to be capable of deforming the end point of the Ag+ titration. Gruen and
Harrap (1971) also observed that the presence of thiosulfate deformed the shape
of the sulfide end point. Schwartz and Light (1970) explored the utility of
3-2
-------
Ag+ as a titrant for the determination of sulfide in wastewaters from a pulp
and paper mill. Although this titrant was acceptable for pure solutions of
sulfide, it did not consistently produce a sharp end point with wastewaters,
presumably because of the high level of complexing and reducing organic
compounds. In addition, a silver mirror was observed with some wastewaters,
indicating that the Ag+ had been reduced. Schwartz and Light also investigated
Hg(II) as a titrant and found non-stoichiometric results, especially in the
presence of dissolved NH3. Florence and Farrar (1980) investigated Ag(I),
Hg(II), Zn(II), Cd(II), and Pb(II) as titrants using artificial standards.
Pb(II) was seen to give the most accurate result based on a stoichiometric
reaction.
For these reasons, Pb(II) was selected as the titrant of choice for
further evaluation. Previous authors have employed either Pb(N03)2 or
Pb(C104)2 which they standardized with a Na2H2EDTA titration (Baumann, 1974;
Florence and Farrar, 1980). However, Florence and Farrar also claimed that the
Pb(II) titration was not quite stoichiometric, yielding a result 6% lower than
would be expected based on the formation of PbS. This finding led them to
standardize the Pb solution against a previously standardized sulfide solution
in order to obtain the most accurate results.
Haas (1979) and Prien et al. (1977) have reported the use of the AgS
electrode for the measurement of sulfide in retort wastewaters. Neither
reported operational problems such as drifting or erratic results. In
addition, Haas1 results for sulfide agreed well with his measurements for total
sulfur, a result which is reasonable for the type of retort waters^which he was
analyzing. On the other hand, neither author reported to have examined the
possibility of interferences. Neither did they include data on the range,
precision, or accuracy of their method.
3-3
-------
The objective of the work described herein is, therefore, to rigorously
determine the efficacy of the Pb(II) titration method for determining sulfide
in oil shale wastewaters. Of specific concern is the effect of potential
interferences from compounds which are known or suspected to occur in retort
wastewaters, as well as the figures of merit such as the precision, accuracy
and range of the technique. Because of many possibilities for interferences
due to known and unknown species in retort wastewater, a Gran's plot method was
selected for the end point determination (Gran, 1962). This method determines
the equivalence point by considering several potential/volume points throughout
the titration and is thus less dependent on occurrences near the equivalence
point where spurious reactions are most likely to occur.
Since the proposed technique is to be used for the evaluation of pollution
control systems under field conditions, it is important that the titration be
sufficiently rugged to give reliable results under conditions which are less
ideal than those found in the average laboratory. The method must be reliable
and insensitive to environmental conditions. It also must be rapid enough to
provide the plant operator with analytical data soon enough to evaluate and
optimize plant operation. The ruggedness of the method under discussion has
*
therefore been evaluated through extensive application under field conditions
with actual retort water samples. Equally as important as the analytical
method is the sampling and preservation technique, which also has been
investigated with fresh retort waters.
EXPERIMENTAL SECTION
Pb(II) titrant was prepared from Analytical Reagent Pb(N03)2
(Mallinckrodt, Inc.), at concentrations of 1,000 and 10,000 mg/1 of Pb.
Titrant was protected from C02 in the air with guard tubes containing ascarite.
NaI03 was primary standard grade from Anachemia. Na2S203 and starch solutions
3-4
-------
were prepared from "Acculute" concentrates from Anachemia. All other chemicals
were reagent grade or better unless otherwise indicated. Sulfide anti-oxidant
buffer (SAOB) was prepared by dissolving 80 g of NaOH and 35 g of ascorbic acid
in 1 liter of distilled water. Sulfide stock solution was prepared by
dissolving 7.0 g of Na2S«9H20 in 100 ml of 50% SAOB in distilled water.
Sulfide standards were prepared by diluting the sulfide stock solution with 50%
SAOB in water.
Electrodes were an Orion AgS membrane electrode (Model 941600) and an
Orion double junction reference electrode (Model 900200) with 10% w/v KN03 in
the outer chamber.' During all titrations cell voltages were measured with a
Keithley Model 616 digital electrometer, which for the potential range
encountered could be read to ± one mV. Attempts were made to determine sulfide
directly from the electrode calibration curve, for which a more precise
potential measurement was required. For this purpose the potential was
measured with a Data Technology Corporation Model 370 digital voltmeter
connected to the analog output of the Keithley electrometer.
Titrations were carried out by placing 50 ml of sample in a stirred beaker
containing 50 ml of SAOB. The electrodes were introduced into the beaker and
the titrant [either 10,000 mg/1 or 1,000 mg/1 of Pb(II)] was selected bas.ed on
the original cell potentials in order to keep the titrant volume in the range
0.5 to 10 ml.
For routine application, the cell potential and titrant volume are
recorded typically three to four times during the course of the titration, and
the equivalence point is determined by the Gran's plot technique (Gran, 1962),
either graphically or on a desk top calculator.
3-5
-------
Analysis under field conditions were performed at the North Site of the
Laramie Energy Technology Center in support of the tests of various pilot scale
pollution control systems.
RESULTS AND DISCUSSION
Original examination of the sulfide electrode indicated that it gave a
linear Nernstian response over the range 1-1,000 mg/1 of sulfide. The cell
potential normally stabilized to within ± one mV (the limit of readability of
the electrometer) within a few seconds. Response time gradually degraded when
the electrode was used with retort wastewaters, but could be restored by
burnishing the electrode surface according to the manufacturers instructions.
First efforts with the electrode were directed to selecting a method for
determining the end point for a manual titration. For this effort, artificial
sulfide standards were used to establish the shape of the titration curve,
which was a typical "S"-shaped curve with a break point of approximately
200 mV. With such a large break point, it was evident that the end point could
be determined by the deflection point by plotting cell potential, E, vs added
volume. However, this approach proved to be tedious, requiring the careful
addition of titrant near the end point and the plotting of several data points.
Rather than plotting the data, an attempt was made to determine the end
point by monitoring the cell potential with an analog meter and observing the
sudden change in cell potential. This approach proved to be less reproducible
than plotting the data.
Finally, a Gran's plot method was tested and proved to be the most
convenient for use with a manual titration. (The Gran's plot method permits
the analyst to collect several data points, each consisting of a cell potential
and volume of added titrant, throughout the midrange of the titration. The
volumes added are otherwise arbitrary, and special care is not required near
3-6
-------
the end point.) For this reason, and for the reasons discussed in the
introduction, the Gran's plot technique was used for the rest of this
investigation. However, since either the deflection point technique or the
Gran's plot technique gave equivalent results, it might be possible to use the
former technique for an automated titrator, many of which use derivative
techniques for end point determination.
Three different variations of the Gran's plot method, each of which give
equivalent results but differ in ease of use, were used throughout this
investigation. The first variation used the traditional Gran's plot paper
(Orion). This approach required a supply of specialized paper which was not
always readily available, and also lacked the resolution which was sometimes
helpful in achieving the most precise results.
For the second approach the fraction, f, of the sulfide which had been
titrated was calculated at several points throughout the titration from the
expression:
(1) f =
where V0 = total volume in titration vessel at start of titration
V. = volume of titrant added
E0 = starting cell potential
E = cell potential
m = Nernstian slope
The fraction f was then plotted against the volume of titrant added, resulting
in a straight line with an intercept equal to the equivalent volume. The third
variation was completed entirely with a Hewlett Packard model 9825A calculator,
which calculated the fraction f at each data point, fit a least squares line to
the data pairs, and then calculated the end point and sample concentration.
(This program is listed in the Protocol section.) Data corresponding to
3-7
-------
0.05
-------
are shown in Table 3-1, where the "simple" matrix refers to standards prepared
in SAOB and the "complex" matrix is discussed below. For the data in this
table the recovery of the 967 mg/1 sample was set equal to 100%, based on the
results of the last paragraph, and the remaining recoveries were calculated
therefrom. The recoveries for one and five mg/1 samples differ slightly from
100%. While the statistical interpretation of this data is discussed in more
detail in Appendix 3A, suffice it to say at this point that the significance of
the difference is marginal for the 5.8 mg/1 sample but is substantial for the
0.967 mg/1 sample. These data suggest that recovery of the titration is
approximately 10% low at 1 mg/1, and possibly 3% low at 5 mg/1. Similar
results have been observed by Florence and Farrar (1980) for low level sulfide
titrations. However, from the data available it is not clear whether the
titration actually yields low results at the mg/1 level or the sulfide standard
is slightly lower than expected, perhaps due to partial oxidation. Thus, the
data in Table 3-1 imply that the accuracy of the titration is at least 97% at
the five mg/1 level and at least 87% at the one mg/1 level. As discussed in
the previous paragraph, the accuracy is essentially 100% at higher levels.
For the purposes of wastewater treatment in the oil shale industry, there
is little importance in determining sulfide below one mg/1, since the water is
destined for internal re-use rather than discharge. In fact, most samples
encountered by the authors during the operation of pilot-scale wastewater
treatment plants at oil shale facilities contain over 100 mg/1 sulfide.
Referring to Table 3-1, it is apparent that the results for the 5.8 and
0.967 mg/1 samples are both within 0.2 mg/1 of the expected value, which is
sufficiently close to satisfy this purpose. However, titrations below one mg/1
should be approached with caution.
3-9
-------
rH
1
OO
0>
o
ro
r-
^
to
LU
O
1— 1
OH
r-
P_
«^
LU
o
i— i
u_
i
^3
to
LL.
O
^~
rv
LU
O
O
LU
oi
^5
a)
i.
3
01
ro
O)
s:
ai
CO
ro
at
-P
o
01
a.
X
LU
•X
•K
^
Ol >-£>
O *•*>
O
Ol
rv
C •
O
•r"
•P
ro
S-
•P
c
Ol /""^
O i —
O O)
O £
ai
•o
4!
•—
to
01
c
o
+J /""s
ro i—
j- \
•P D)
C E-
Ol s^-/
O
c
o
c
o
•p >->
s- \,
•P CD
C E
Ol ^*>
u
c
o
o
X
•r™*
s-
-p
s:
ai J-
r— Ol
^ ^
ro 3
to z,
rH rH rH
+1 +1 +1
O CM 1^-
O
cn co
CO rH
* O O
cn
+1 +1 +1
r^« *^~ co
to to «3-
cn . co
to
o
in
CM
1^. QQ
oo to
in . o
cn in
»\
•» *• CO
^* cn in
co m oo
cn
^
in *• "
to to r"»
cn to to
00
m
0
f^.
r~. o to
to oo cn
cn
in o
' Ol Ol Ol
"o. "o. "a.
E E E
•r— »r— *r—
to to to
rH CM CO
rH
+1
O
o
rH
f^»
rH
+1
to
to
cn
cn
^*
cn
*t
o
CD
O
rH
*
O
in
cn
i^
to
cn
X
a>
a.
E
O
CJ
«^-
rH
+1
CO
00
cn
CO
o
.
+1
CO
1*^
•
in
cn
to
•
in
*•
CO
.
in
VI
CO
r---
in
0
CO
in
X
ai
Q.
E
O
O
in
rH
+1
in
.
rH
cn
rH
O
.
+1
in
00
00
d
cn
cn
oo
d
M
cn
CO
C)
*\
CO
00
^3
f*^
to
C7)
.
0
X
ai
Q.
E
O
to
t
c
^£
o
.c
01
01
ai
r~
a.
E
ro
Ul
o>
O)
s-
r*
^j
ai
f
+j
c
o
•o
Ol
ro
.a
O)
O)
ro
&»
ai
>
ro
0)
jCI
•P
q-
o
r"
O
•i—
-P
ro
•^
>
ai
•a
-a
i.
ro
T3
c
ro
-p
01
Ol
o
Ol
i-
ro
c
2^
o
r"
01
o>
•r-
E
•i—
<—
i-
O
^
i.
01
ai
f*
t—
•x
Ol
<~
4^
S-
o
H—
Ol
S-
ro
01
-P
•^
pi
• |^
r^
s-
o
S-
s-
ai
Ol
f~
I—
,
S~\
£^
O
•^*
4->
ro
S-
^^
c
Ol
u
c
0
u
T3
Ol
-P
0
O)
Q.
X
Ol
SWH^
*v^
/^*N «
c — ' 3
r—
II ro
~^
s- at
01 cn
> ro
0 i-
U O)
Ol >
o; ro
^
*
3-10
-------
Before the titration procedure can be applied to retort waters
consideration must be given to the potential interfering effects of the many
compounds in retort water. While it is not possible to prepare artificial
standards containing all compounds found in retort water, major components can
be added to solutions containing known amounts of sulfide. For this purpose,
sulfide standards were prepared containing 1-1,000 mg/1 sulfide in addition to
the following levels of possible interfering compounds: SCN~, S20f, SOI, C-l~,
and C0§ at a concentration of 1,000 mg/1 were added as their sodium salts;
68.0 g/1 of (NH4)2C03; and 40 g/1 of NaOH to neutralize NH| during titration.
The results of the addition of the possible interfering compounds is seen
in Table 3-1 by comparing the results obtained for the "complex" samples with
those obtained for the "simple" samples with the same sulfide concentration.
Any interference occurs only in the low level sulfide and at levels of only a
few percent. All errors due to interferences are less than 0.1 mg/1 for the
low level samples and are not measurable at the higher levels of sulfide.
Concentrations of possible interfering compounds in the samples under
consideration are larger than normally encountered in the authors' laboratory,
and this implies that errors due to these added compounds are not significant
in regard to the treatment of retort waters. However, for application of this
technique to other wastewaters where low level measurements are more important,
it should be noted that the results for samples 3 and 6 are statistically
different at the 95% confidence level. It is therefore recommended that
application of this technique to samples containing less than one mg/1 be
preceeded by additional investigation of possible interferences.
At this time the titration method has been applied to retort wastewaters
in support of tests of various pollution control systems, including venturi
scrubbers, steam strippers, and reverse osmosis units; results have been very
-------
encouraging. Precision of the method when applied to the routine analysis of
retort waters containing sulfide in the range of 1-1,000 mg/1 is typically 2-3%
relative or ± 0.1 mg/1. The contribution of the electrometer to this level of
precision was examined as follows: a Gran's plot calculation was first carried
out on a set of titrant volumes and cell potentials which were artifically
selected so as to be exactly consistent, that is, to yield an exactly straight
line when plotted on Gran's plot graph paper. Next, the same calculation was
repeated after altering a potential by small increments while leaving all other
parameters unchanged. This procedure resulted in a variation of approximately
2-3% in the calculated sulfide concentration for errors of one mV in the cell
potential. Since the electrometer used for the field measurements was readable
to only ± one mV, it appears likely that the readibility of the electrometer
was a major contributor to the imprecision of the -method. These data also
suggest that using an electrometer capable of reporting to ±0.1 mV could
improve the precision if that should be needed. The recovery as measured by
adding known amounts of sulfide to the sample is 100% within the precision of
the method.
The titration can be carried out routinely with a minimum of equipment,
and has caused few problems. The only difficulty encountered has been with
samples from a steam stripper which contained less than a few mg/1 of sulfide,
for which a drifting electrode potential was sometimes observed. This made the
titration more difficult. In order to compensate for such drift, the titration
was performed more rapidly than usual with apparently no ill effects and
results were reported only to the closest one mg/1.
Although in theory the titration can be completed without ever calibrating
the AgS electrode, it is helpful to carry out a calibration occasionally,
perhaps every few weeks. When this is done, the calibration is stable enough
3-22
-------
to permit selection of a titrant of proper strength by observing the cell
potential at the start of the titration, and to thereby avoid titrating the
same sample twice. Qrder-of-magnitude agreement between the concentrations
indicated by the titration and cell potential assures against mistakenly
titrating a non-sulfide species, such as COf, when sulfide is absent. An
occasional calibration is also helpful in checking for proper electrode
response.
Preparation of standards from solid Na2S'H20 in a field laboratory is
usually inconvenient and standards are not sufficiently stable to allow for
advance preparation and storage in polyethylene containers. Concentrated
standards (11,000 mg/1) were therefore prepared and sealed in 20 ml, pre-scored
glass vials which are sealed by melting an elongated glass neck. Before
sealing, the head space but not the liquid in the vial, was flushed with N2.
It was hoped that by hermetically sealing the container and by using
concentrated solutions, standards would be stable indefinitely. In the field
laboratory standards could be prepared simply by diluting the contents of the
vial. Measurements of solutions stored in vials showed no measurable decay
over a 9 month period. The use of standard concentrates in hermetically-sealed
glass containers is thus a promising method for storing sulfide standards.
Performing the titration technique on site during various field tests has
made it possible to examine the stability of sulfide in fresh retort
wastewaters. During normal operation, samples designated for sulfide analysis
are buffered with 50% SAOB immediately upon collection and are analyzed within
one hour. Select samples were set aside for sulfide time decay measurements.
Samples which were filtered through a 0.45 micrometer filter and stored at 4°C
in tightly sealed polyethylene bottles were observed to lose approximately 20%
of their sulfide content per day and to be essentially free of sulfide in two
3-13
-------
weeks. In a separate test, fresh retort wastewaters stored at room
*
temperatures in similar containers lost 99% of the sulfide within 24 hours.
Samples stored in SAOB were more stable, losing approximately 20% in 10 days,
mostly in the first two days. While an insufficient number of samples have
been examined to draw any conclusions regarding the typical decay rate of
sulfide, it is clear that samples must be analyzed immediately upon collection
to assure a realistic sulfide measurement.
If retort waters are analyzed promptly upon collection, it is observed
that sulfide is the major form of sulfur present. In comparison, several other
investigators, who did not have access to fresh retort waters, have reported
that the major species in retort water are S20|, S0|, and SCN~ (Stuber et al.,
1978). However, based on the data observed in this study it appears likely
that the latter species arise, at least in part, from the oxidation of sulfide
during storage rather than being an original component in retort water.
Finally, the titration method was compared to the direct calibration
technique as a method for measuring sulfitie in retort wastewater. That is,
does the potential from the AgS electrode give an adequate measurement of
sulfide, assuming the electrode to be calibrated with sulfide standards? If
so, the direct calibration technique would be an attractive basis for a
* continuous-flow, automated analyzer.
Under normal laboratory conditions it was observed that the electrode
potential would vary approximately 3 mV.when immersed in a series of solutions,
each containing the same sulfide concentration. For the sulfide electrode 3 mV
corresponds to an error of approximately 23%, and an effort was therefore made
to define and minimize the sources of this drift. Since temperature variations
are one possible source of error, a series of identical samples was analyzed
3-14
-------
over the range of 20-30°C. No effort was made to control the temperature of
the un-immersed part of the electrode, so that temperature gradients were
likely. Under these conditions, a temperature change of 1°C caused a 2% error
in the concentration measured. Changing the ionic strength of the solution by
approximately one mole/1 also caused the measured sulfide concentration to vary
by a few percent. Although all sources of error were not identified in this
limited effort, it appeared that best results could be obtained by controlling
both the temperature and ionic strength of samples and standards.
The principal difficulty with using the direct calibration technique was
the need for frequent calibration with sulfide standards. This difficulty was
compounded because the sulfide standards are unstable and require frequent
preparation. In the final analysis, the direct calibration technique proved
too time consuming and difficult for routine application, although it has been
used in applications requiring a continuous monitor. In comparison, the Pb
titration is especially attractive because the titrant is stable Indefinitely,
thereby contributing to its reliability.
CONCLUSIONS
The titration with Pb(II) using an AgS electrode as an indicator and a
Gran's plot procedure to determine the end point is an acceptable method for
measuring sulfide in retort wastewaters over the range 1-1,000 mg/1.
Significant interferences have not been found. The precision of the technique
with retort wastewaters, as applied during routine monitoring, is typically
2-3% relative ,or ±0.1 rag/1, whichever is greater. The data suggest that
samples containing one mg/1 give results approximately 0.1 mg/1 lower than
expected. No bias is observed at higher concentrations, and the accuracy of
the method is therefore limited only by the precision for samples containing
more than 5 mg/1 of sulfide. The titration has proven reliable, simple, and
3-25
-------
rapid when applied under field conditions. The titration method as described
in this text is thus attractive as a reference method for the measurement of
sulfide in retort wastewaters.
In comparison to the titration method, the direct calibration curve method
is less accurate and precise for routine use with a variety of samples, but may
be appropriate for monitoring a series of similar samples under closely
controlled conditions of temperature and ionic strength. The main advantages
of the direct calibration curve technique are that it requires only a single
measurement for each sample and it can be applied as a monitor in a continuous
flow system. However, these advantages are more than offset by the lesser
accuracy and reliability of the direct calibration method. In addition, the
*
direct calibration 'method requires frequent calibration with unstable
standards, while the titrant for the titration method is stable indefinitely.
Concentrated sulfide standards have been shown to be stable when stored in
SAOB in hermetically sealed vials for periods up to 9 months. This approach is
suggested as a possible means of providing sulfide standards to field
laboratories.
RECOMMENDED METHOD FOR TOTAL SULFIDE IN RETORT WASTEWATERS
Principle
Sulfide is titrated with Pb(II) using the AgS ion selective electrode as
an indicator. The end point is determined by a Gran's plot technique.
Reagents
1. Sulfide Anti-Oxidant Buffer (SAOB). (50 ml/sample)
' In a 1 liter volumetric flask place 600 ml distilled water, 80 g NaOH
(or 175 ml of 33% w/w NaOH), and 35 g ascorbic acid. Dissolve and
bring to volume.
3-16
-------
2. 10 M NaOH. (0-10 ml/sample)
Dissolve 40 g NaOH in 100 ml volumetric flask. Dilute to volume.
This is needed only to check electrode performance and not for
routine work.
3. Pb stock solution. (10,000 mg/1 as Pb)
Add 15.985 g Analytical Reagent grade Pb(N03)2 to a 1 liter flask.
Dissolve and bring to volume.
(Pb) = (0,6256)x[mass Pb(N03)2]
4. Pb titrant. (1-10 ml/sample)
Dilute the Pb stock solution 10:1 volumetrically, or as needed to
arrive at a titrant volume of 1-10 ml.
5. S= Stock Solution (10,000 mg/1.)
Add 7.49 g of AR grade Na2S-9H20 to a 100 ml volumetric flask. Add
50 ml SAOB and bring to volume with de-oxygenated, d.i. H20.
(S=) = 0.1335 X Na2S-9H20
This is needed only to check electrode performance and not for
routine work.
6. Standard S= Solutions, (needed only to check electrode)
Dilute the S= stock solution with 50% SAOB in distilled H20.
Equipment
1. 2 burets, 10 ml each
2. 250 ml beakers, 1 per sample
3. pH electrode and meter
4. AgS ion selective electrode (Orion)
5. Double junction reference electrode with an inner filling solution
from manufacturer and an outer filling solution of 10% w/v KN03
3-17
-------
6. electrometer, capable of reading 800 mV to the nearest mV
7. 10 ml graduated pipet (optional)
8. 50 ml dispenser for SAOB
Procedure
The procedure as described below is applicable to waters containing
sulfide in the range 1-310 mg/1, and minor adjustments must be made for more
concentrated samples. For samples containing more than 310 mg/1 it is
convenient to reduce the sample volume in order to keep the titrant volume
within 10 ml. Similarly, for samples containing unusually large amounts of NHJ
and HCOf (approaching one m/1) it is recommended that additional base be added
in the form of 10 M NaOH. The function of the additional base is to convert
the H2S and HS~ to S=, which requires a pH of approximately 14. However,
excessive amounts of NH| and HCO| buffer the solution and prevent this pH from
being attained.
The following procedure is divided into protocols which should be'
performed occasionally, perhaps every few weeks and during the initial set up,
and protocols which should be performed during each analysis. Performing a
calibration of the electrode every few weeks is normally sufficient to permit
the analyst to estimate the concentration based on the original electrode
reading, and thereby to select the titrant of proper strength.
Procedures to be Performed Occasionally—
Standardize the electrode. Standardize the stock sulfide solution by
titrating with Pb(II) as described below. Prepare sulfide standards by
volumetric dilution in 50% SAOB covering the range 1-1,000 mg/1. Into a 250 ml
beaker, place 50 ml of SAOB and 50 ml of standard solution. Place the beaker
on a magnetic stirrer, immerse the electrodes, and record the potential. Plot
potential vs concentration on semi-log paper. The resulting calibration curve
3-18
-------
should be straight with a slope close to 30 mV per decade change in
concentration.
NaOH requirement for concentrated samples. Add 50 ml of SAOB and 50 ml of
sample to a 250 ml beaker. Add a magnetic stirrer and insert the AgS
electrode, the reference electrode, and a pH electrode. The pH should be close
to 14. If it is not, adjust with 10 M NaOH, until the pH is 14 or until the
potential of the sulfide electrode stabilizes within a few mV. Note the volume
of NaOH required and add the equivalent amount to samples containing similar
concentration of NH| or COf.
Drifting electrode response. During the analysis of retort waters, the
AgS electrode often develops a slow and drifting response after a few days use.
A more rapid response can normally be restored by burnishing the electrode as
described by the manufacturer.
Procedures to be Performed With Each Sample--
Immediately upon collection add 50 ml of sample to 50 ml of SAOB in a
250 ml beaker. Place on a magnetic stirrer. This solution may remain in a
covered beaker for up to an hour before titration although immediate analysis
is preferred. Do not allow the sample to stand after collection without the
addition of SAOB.
Place the beaker on a magnetic stirrer, begin stirring, and insert the AgS
and reference electrode. Based on the original cell potential, select a
titrant which will result in a titrant volume of 1-10 ml. Gradually add
titrant with stirring. (A black precipitate is evident immediately after the
addition of titrant unless the sample is opaque.) Stop adding the titrant at
several locations and record the buret reading and cell potential. Four
readings spaced in the range 2<(E-E0)<35 mV are normally an adequate number of
data points.
3-19
-------
Recording Data
The following data must be recorded in order to calculate the sulfide
concentration:
Once per Sample
E0 = original potential in mV .
V0 = total volume in beaker (SAOB + sample)
Vstart = Startin9 buret reading
(Pb) = titrant concentration
Vsample = volume of Samp1e
At each stopping point
E = cell potential in mV
VB = buret reading
Calculations
Determine the equivalence point, V using Gran's Plot Paper or by
plotting the function:
(E-E0)/m
Where V. = titrant volume
m = Nernstian slope from the calibration curve = 29.5 mV at 25°C
Calculate the sulfide concentration from the equation:
•CS-) = V. (Pb)
eg.
H sample
Where (Pb) is the lead concentration in mg/T and V is the equivalence
point.
Alternately, the calculations can be performed automatically by the
following program designed to operate on a Hewlett Packard Model 9825
calculator. This program performs a least squares fit of the function f to a
straight line and thereby determines the V and the sulfide concentration.
3-20
-------
Values of f>0.95 and <0.05 are rejected, so that at least two measurements are
required which give values of f in the range 0.05 to 0.95. This normally
corresponds to changes in the cell potential in the range 1 to 35 mV.
Instructions for Gran's Plot Program
1. If the 9825 is off, insert the tape cartridge and turn it on.
If the 9825 is on, but the program is not loaded, insert the tape
cartridge and type 'Idp 0 EXECUTE1, where everything inside the
quotes is typed and EXECUTE means press the key marked EXECUTE.
If the 9825 is on and the program is loaded, just type 'RUN'.
2. If the calculator beeps and displays "error nn", where nn are numbers
or letters, start again. If it happens again, consult the manual.
3. The program should be running and displaying the first prompt.
4. The following data are entered once for similar sample conditions:
Date (mm/dd/yy) - .maximum of 8 characters
Log coefficient (29.5 or other)
V0 - displayed as "V(0)"
Vsample " d1sPlayed as "V(sample)"
Pb concentration
5. For each sample the following items are requested:
Sample ID - maximum of 16 characters
E0 - displayed as E(0)
Starting buret reading
6. Enter the data points for the sample:
E and V. (buret reading), these will be printed. Continue until
t
there are no more, then just press 'CONTINUE' when asked for the
next value of E.
3-21
-------
7. Make any needed corrections. When asked "Remove data (y/n)?", type
'y1 or 'CONTINUE1 for yes, 'n' for no. If no, go to Step 8.
«t
If you wish to remove data, the prompts are in the same order as for
entering data. When you have removed all the points you wish, just
press 'CONTINUE'.
After removing data, you will be asked "Enter more data (y/n)?". If
you wish to enter corrected data points, type 'y' °r 'CONTINUE' for
yes, 'n' for no. If yes, go to Step 6.
8. The results will be printed. V is the equivalent volume, Cs is the
calculated sample concentration and R is the correlation coefficient.
9. If for the next sample, the data entered in Step 4 is the same, go to
Step 5. •
If the data in Step 4 should be different, press 'STOP' and then
press 'RUN1 and go to Step 4.
3-22
-------
Table 3-2
GRAN'S PLOT OUTLINE
0: "Sulfide concentration":gto "start"
1: "f":ret (V+T)tn"((E-C)/G)/V
2: "Cs":ret HP(.1548/8)
3: "yes":ret flglS or cap(Y$[l,l])="Y"
4: "sum":
5: S[l]+p3pl-*S[l]
6: S[2]+p3p2^S[2]
7: S[3]+p3plpl-»S[3]
8: S[4]+p3p2p2+S[4]
9: S[5]+p3plp2^S[5]
10: N+p3^N;it N>=0;ret
11: beep;dsp "Too many points removed.";end
12: "regress":
13: S[1]/N->X
14: S[2]/N+Y
15: (S[3]-NXX)/(N-1>L
16: (S[4]-NYY)/(N-1>*M
17: (S[5]-NXY)/L(N-1>B
18: Y-BX-^A
19: -A/B-»H
21: ret
22: "ent data":
23: prt "Entered data:","E","V(t)"
24: ent E; if flg!3;ret
25: ENT "Burette reading?",Z
26: if flg!3;gto -1
27: Z-D-»T;prt E,Z;spc
28: Jf'(E,T>F
29: if F>.05;cll 'sum'(T,F,l)
30: if F>.05 and F<.95;cll -'sum1 (T,F,1-)
31: gto -7
32: "rmv_data":
33: prt "Removed data:","E","V(t)"
34: ent E; if flg!3;ret
35: ent "Burette reading?",Z
36: if flg!3;gto -1
37: Z-D-»T;prt E,Z;spc
38: 'f (E,T)-»F
39: if F>.05;cll 'sum'(T,F,-1)
40: if F>.05 and F<.95;cll 'sum1(T,F,-1)
41: gto -7
42: "get_data":
43: ell 'ent_data'
44: ent "Remove data(y/n)?",Y$
45: if not 'yes1;ret
46: ell 'rmv_data'
47: ent "Enter more data(y/n)?",Y$
48: if 'yes1;gto -5
49: ret
3-23
-------
Table 3-2 (cont.)
50: "prt_hdr":
51: spc 2;wrt 16.1
52: prt "Sample:",I$
53: prt "Date:",D$
54: spc
55: wrt 16.2,G;wrt 16.3,C;wrt 16.4.V
56: wrt 16.5,J;wrt 16.6,P;wrt 16.7,D
57: spc ;ret
58: "prt results": .
59:
60:
61: fxd 5;prt "R:",R
62: wrt 16.1
63: spc 3;fxd 2;ret
64: "get_info":
65: ent "Date(mm/dd/yy)?",D$
66: ent "Log coeff?",G;if flg!3;gto +0
67: ent "V(0)?",V;if flg!3;gto +0 • •
68: ent "V(sample)?",U;if flg!3;gto +0
69: ent "Pb cone?",P;if flg!3;gto +0
70: ret
71: "get_smpinfo":
72: ent "Sample ID?",I$
73: ent "E(0)?",C;if flg!3;gto +0
74: ent "Starting burette reading?",D;if flg!3;gto +0
75: ret
76: "mem_init":
77: dim D$[8],I$[16],S[5],Y$[6]
78: fmt 1,16"-"
79: fmt 2,"Log coef:",f7.2
80: fmt 3,"E(0):",fll.2
81: fmt 4,"V(0):",fll.2
82: fmt 5,"V(smpl):",f8.2
83: fmt 6,"Pb conc:",f8.2
84: fmt 7,"Bur start:",f6.2
85: fmt 8,"Veg:",fl2.2
86: fmt 9,"Cs:",f!3.2
87: 0-»N;fxd 2
88: ret
89: "0->S[*]":O^S[l
90: "start":
91: ell 'mem_init'
92: ell 'get_info'
93: ell 'get_smp.info'
94: ell 'O^S[*]'
95: ell 'prtjidr'
96: ell 'get_data'
97: ell 'regress1
98: ell 'prt_results'
99: gto -6
*30484
3-24
-------
REFERENCES
APHA. 1975. Standard Methods for the Examination of Water and Wastewater.
14th Edition. Washington, DC.
Bauman, E.B. 1974. Determination of Parts per Billion Sulfide in Water with
the Sulfide-Selective Electrode. Anal. Chem., 46(9):1345.
Florence, T.M. and Y.J. Farrar. 1980. Titration of Microgram Amounts of
Sulfide with a Sulfide-Selective Electrode. Anal. Chim. Acta, 116:175.
Fox, J.P., D.S. Farrier and R.E. Poulson. 1978. Chemical Characterization and
Analytical Considerations for an In Situ Oil Shale Process Water.
LETC/RI-78/7. Laramie Energy Technology Center, Laramie, Wyoming. 47 pp.
Gran, G. 1952. Determination of the Equivalence Point in Potentiometric
Titrations, Part II. Analyst, 77:662.
Gruen, L.C.- and B.S. Harrar. 1971. A Titrimetric Method for Sulfide Analyses
with a Specific-Ion Electrode. J. Soc. of Leather Trades Chemists,
55:131.
Haas, F.C. 1979. Analysis of TOSCO II Oil Shale Retort Water, in Analysis of
Waters Associated with Alternative Fuel Production. L.P. Jackson and
C. C. Wright, eds. American Society for Testing and Materials,
Philadelphia, PA. p. 18.
Kolthoff, I.M. and P.J. Elving. 1961. Treatise on Analytical Chemistry,
Part II. Analytical Chemistry of the Elements, Volume 7. Interscience
Publishers, New York, NY.
Prien, C.H., J.J. Schmidt-Collerus, R.E. Pressey, C.H. Habenicht, K. Gala,
J.E. Cotter, D.J. Powell, R. Sung. 1977. Research of Sampling and
Analysis Procedures: Paraho Demonstration Plant. Denver Research
Institute Report on Contract 68-02-1881.
Stuber, H.A., J.A. Leenheer, D.S. Farrier. 1978. Inorganic Sulfur Species in
Wastewaters from in Situ Oil Shale Processing. J. Enveron Sci. Health.
Part A, A13:663-675.
Schwartz, J.L. and T.S. Light. 1970. Analysis of Alkaline Pulping Liquor with
Sulfide Ion-Selective Electrode, Tappi, 53(1):90.'
Volk, W. 1969. Applied Statistics for Engineers. 2nd Edition. McGraw-Hill.
New York, NY.
Wallace, J.R. 1981. The Analysis of Oil Shale Wastes: A Review.
EPA-600/S7-81-084. U.S. Environmental Protection Agency, Cincinnati,
Ohio.
3-25
-------
APPENDIX 3A
STATISTICAL INTERPRETATION
This section describes the application of the "t" test to the data in
Table 3-1. First, consider the standard deviation which should be associated
with each recovery. In theory the uncertainty in the recovery, as a quotient
of two numbers, should depend on the uncertainty in both the "measured" and
"expected" concentrations (Table 3-1). However, the uncertainty in the latter
can be neglected since volumetric dilution errors are insignificant in
comparison to the errors associated with the titration, and the standard
deviation of the recovery can therefore be obtained simply by dividing the
standard deviations in Column 4 by the expected concentrations.
While this calculation yields the standard deviation based on a set of
three analyses, a more precise estimate of the standard deviation can be
obtained by averaging all standard deviations in Column 4 of the table, as
follows:
~" lilxil + s2(x2) s2(xj
—A I —1} ... -—J}
(Al) - i fi x^
where s2. = variance of the recovery for the j sample
J
s2(x.) = variance of the average for the i sample set
x. = average concentration for the i sample
R. = recovery for the .th sample
J J
Carrying out this calculation yields an estimated standard deviation of 0.011
for the recovery for all samples. (This value is reasonable in view of the
typical precision obtained with this analysis during routine application.)
Determining whether the recoveries differ from unity, or from each other,
thus requires application of the t test to the recoveries in Column 5
(Volk, 1969).
3-26
-------
To determine whether the recoveries, R, differ from unity, t is calculated
from the expression:
fA2) t = R " 1-°Q
l/^; L 0.011 VZ
Referring to tables of the t distribution indicates that the recovery of the
5.80 mg/1 pure sample differs from unity with a 90% probability (t = 1.8).
That is, if the recovery is actually 1.00, the probability is 10% that t > 1.8.
For the 0.848 mg/1 pure sample (Sample 3), t = 7.9, and the recovery is not
equal to unity with a 99.9% probability.
The effect of interferences can be measured by calculating the t value for
sample pairs i and j:
R. - R.
(A3) t =
J J
0.011 V2
Obviously, no measureable interference occurs at a sulfide concentration of
1,000 mg/1 (Samples 1 and 4). For the 5.80 mg/1 samples, t = 1.0, also not
large enough to, conclude that an interference exists,. At 0.967 mg/1 (sample
pair 3 and 6), t.= 2.4, a value that would be.exceeded in 5% of the cases if no
interference existed. It thus appears that the results for Samples 3 and 6 are
thus (barely) significantly different.
3-27
-------
SECTION 4
TOTAL SOLUTES IN RETORT WASTEWATERS
The measurement of total residue, which is defined as the mass remaining
when' a sample is dried at 103-105°C, is performed frequently on ground and
surface waters to determine total solutes in the sample (EPA, 1979).
Unfortunately, many major components of retort waters, most notably carbonate,
bicarbonate, ammonia, and volatile organic compounds, are lost during the
drying process. The traditional residue measurement, therefore, is not
normally meaningful when applied to retort wastewaters. Nevertheless, a
measure of the total solute content is still important to both the wastewater
engineer—who needs a simple indication of the overall effect of a treatment
process—and to the analyst—who needs a quality control'check for individual
determinations. An alternate technique is therefore clearly desirable.
One alternative method considered in this study, as well as by other
investigators, is lyophilization (freeze drying), whereby the sample remains
cold throughout the drying process, only reaching room temperature (at maximum)
at the moment of total desiccation. The residue is then weighed as an
indication of dissolved residue.
It is widely known that for ideal solutions the colligative properties,
such as boiling point elevation, freezing point depression, and osmotic
pressure, are an indication of total solute content. Unlike the drying
procedures commonly used for residue measurements, the colligative properties
are related to total molality rather than to dissolved mass. Nevertheless,
total molality provides the wastewater engineer with a gross indicator of
treatment progress, and therefore can serve much the same purpose as the
traditional "TDS" measurement. Similarly, total molality provides a quality
4-1
-------
control check for the analyst, since the individual components should sum to
the total molal strength, providing that the major components have been
measured correctly. Thus, if the colligative properties of retort wastewaters
could be measured and related to total solute strength, they would fill much
the same role as the total residue measurement.
This chapter describes efforts to investigate both lyophilization and the
measurement of colligative properties as an indication of total solute content.
The objective of the work described below is to develop a method for measuring
total dissolved material in retort wastewaters which is simple and rugged
enough to be performed in a field laboratory in support of pollution control
tests. The analysis should also be rapid enough to provide timely and
pertinent data to the pollution control plant operator. To be of most value,
the technique developed also should be applicable to other synfuel wastewaters,
most of which contain similar major components as oil shale retort waters.
EXPERIMENTAL SECTION
Standards of known osmolality (osmotic strength) were prepared from "Baker
Analyzed" reagent-grade NaCl dissolved in de-ionized, distilled water.
Solutions of NH4HC03/NH3 were prepared from reagent grade ammonium hydroxide
(Fisher Scientific) and analytical reagent NH4HC03 (Mallinckrodt). The
ammonium hydroxide was standardized by titration with 0.1 N H2S04 prepared from
Acculute concentrated standards using a pH meter to monitor the progress of the
titration. For this titration the pH at the equivalence point varies with the
concentration, a factor which was taken into account.
Retort waters of various types were obtained from the Laramie Energy
Technology Center (DOE), mainly in conjunction with operations at their North
Site Facility.
4-2
-------
Selected retort water samples were analyzed on a Digimatlc Model 3DII
(Advanced Instruments, Inc.) to establish repeatability with retort waters.
This instrument operates by automatically determining the freezing point
depression of a sample. Additional freezing point depression measurements,
including those listed in Table 4-1, were obtained on a Knauer manually
operated cryoscope operated according to the manufacturer's instructions.
Samples were stored at 4°C from preparation until analysis.
RESULTS AND CONCLUSIONS
Various retort water samples were analyzed for total dissolved residue
using the standard drying technique at 103°C. As expected, residue measured by
this procedure was only a small fraction of that expected based on analysis of
the individual components. Lyophilization of retort waters gave similar
results, implying that the ammonium carbonates also evaporate under cooled,
low temperature conditions. To confirm this implication, solutions of ammonium
bicarbonate and ammonium carbonate were evaporated using both lyophilization
and thermal drying. With either procedure over 98% of the dissolved material
was lost. It is thus clear that neither lyophilization nor thermal drying, as
carried out in this experiment, are appropriate for retort wastewaters.
It is natural to ask at this point whether lyophilization carried out at a
different temperature would more selectively evaporate water in comparison to
the dissolved material. The answer to this question can be found by examining
the vapor pressures of water and the major constituents in retort water as a
function of temperature. Application of simple thermodynamic formulae
indicates that the vapor pressure of ammonium bicarbonate, one of the major
components of retort waters, exceeds the vapor pressure of water at all
temperatures above approximately -10°C, corresponding to 0.003 atm. In theory,
then, lyophilization carried out well below -10°C could selectively evaporate
4-3
-------
Table 4-1
TOTAL SOLUTE BY FREEZING POINT DEPRESSION
Sample
Ideal
Osmolality
(moles/1)
Measured
Osmolality
NaCl
Equivalent
(moles/1)
1.0 M NH4HC03 +
0.1 M NH4OH
2.1
1.90
2.0
0.1 M NH4HC03 +
0.01 M NH4OH
0.21
0.216
0.21
0.01 M NH4HC03 +
0.001 M NH4OH
0.021
0.021
0.021
1.0 M NH4HC03 +
0.1 M NH4OH +
0.921 M. Acetone
3.02
2.64
2.8
Omega-9 Retort Water.
0.58
4-4
-------
water from retort water samples. However, because the vapors pressure of most
organic compounds are less affected by temperature than is water, as the
temperature is lowered the organic compounds in retort water would be more
selectively evaporated than they are at 103°C. It thus appears that
evaporation techniques, regardless of the temperature under which they are
carried out, are inappropriate for retort wastewater.
The most common colligative properties for measuring total solute content
include -osmotic pressure, boiling point elevation, and freezing point
depression. The first of these requires a semi-permeable membrane capable of
passing water but not the dissolved materials. In the case of retort
wastewaters, the membrane must be impermeable to both ammonia and carbon
dioxide, as well as additional organic compounds. Most membranes are permeable
to both these gases, and for this reason osmotic pressure was not examined.
The depression of the dew point, which is equivalent to the elevation of-
the boiling point, can be measured by commercially-avail able instruments, one
of which was evaluated using NaCl standards, NH3/NH4HC03 standards, and retort
waters. While the NaCl standards yielded reasonable results, no meaningful
numbers could be obtained with the other samples. In retrospect, this result
is reasonable since the proportionality between dew point depression and solute
content can only be expected for non-volatile solutes (Lewis and Randall,
1961). Hence, dew point depression (and boiling point elevation) are
inappropriate methods for measuring solute content of retort wastewaters.
Freezing point depression also has been evaluated using both artificial
standards and actual retort waters and has proven to be the best method of
those tested. Table 4-1 illustrates the application of this technique to real
and simulated retort waters. In this table ideal osmolality refers to the
number of moles of solute added, counting each ion separately. Measured
4-5
-------
osmolality is the freezing point depression in °C divided by 1.86°C, the
freezing point depression per mole at infinite dilution. The measured and
ideal osmolality will agree to the extent that the sample behaves like an ideal
(i.e., infinitely dilute) solution. The measured osmolality is sufficiently
close to the actual molal strength for most field applications. The final
column in the table shows the concentration of NaCl which gives the same
freezing point depression as the solution under test. Comparing the data in
columns 2 and 4 in this table suggests that the "NaCl equivalent molarity" is a
slightly better indication of total solute content than is the measured
osmolality. The measured osmolality of the retort water listed in Table 4-1 is
also reasonable in view of published data on this sample (Fox et al., 1978). In
conclusion,, data in Table 4-1 indicate that total solute content in retort
waters can be measured to an accuracy of 7% relative at concentrations up to
3 moles/1. At higher concentrations, solution non-ideality may become
increasingly important, and the freezing point method is therefore not
recommended for such samples. The accuracy of most devices for measuring
freezing point depression is typically ±0.001 moles/1, which therefore
constitutes the lower limit of this technique.
For the purpose of checking the total of the individual analyses, the data
in Table 4-1 suggest that the "NaCl equivalent molarity" is the better measure.
Since most species are determined in units of mg/T, they can be easily
converted to moles/1 and compared directly to the "NaCl equivalent". This
approach may prove to be difficult with waters containing high levels of .total
organic carbon (TOC) which are not analyzed for individual organic species. In
this case the TOC measurement cannot be related directly to molarity. However,
in the author's limited experience in applying the freezing point method, the
organic fraction did not appear to constitute a large molar fraction. In any
4-6
-------
case, a high molar organic content suggests that at least the major organic
species should be determined.
Repeatability of the freezing point method was determined by analyzing a
variety of retort waters with the commercially available instrument listed in
the experimental section of this report. Most results could be repeated to
within ±0.001 osmols, indicating that adequate precision is available from
commercially available instruments. Using such instruments, the freezing point
measurement can be made in a matter of minutes with a minimum of equipment.
The freezing point depression thus appears to be a method which could easily be
performed in field laboratories in support of water pollution control tests.
However, this author has not yet applied this test under field conditions, nor
to a large number of samples and therefore cannot yet attest to its ruggedness
and reliability under such conditions.
ACKNOWLEDGEMENTS
Freezing point depression measurements from the Knauer cryoscope were
obtained by Huffman Laboratories (Wheat Ridge, Colorado).
RECOMMENDED METHOD FOR TOTAL SOLUTES IN RETORT WATER
Principle
The freezing point of an ideal aqueous solution is lowered 1.86°C for
every mole of solute dissolved in one kg of water. The following procedure
estimates the total solute content by measuring the freezing point depression.
Comment
Freezing point depression is proportional to the molality of the solution
at infinite dilution (One molal = one gram-mole/kg of solvent). However,
retort waters containing less than three moles (counting each ion and counter
ion separately) are sufficiently dilute to give results accurate to within a
few percent. In addition, at such concentrations the difference between
4-7
-------
molarity and molality can be neglected for the purpose of this analysis.
Results considered as molarity can thus be compared directly with the sum of
the individual determinations carried out on the sample.
Applicability
Retort waters containing 0.01 to 3.0 moles/1 of dissolved material,
counting each ion separately.
Reagents
1. Osmolality* standard. Prepare standards of known osmolality covering
the range of concentrations expected in the samples using reagent
grade NaCl and de-ionized water. The following table is condensed
from more extensive data in Handbook of Chemistry and Physics:
wt %
NaCl
0.100
0.60
2.90
6.20
8.60
Total**
Molarity
(g-moles/1)
0.034
0.206
1.01
2.21
3.12
Freezing
Point
Depression
(°C)
0.062
0.358
1.729
3.837
5.512
Osmolality
0.033
0.192
0.930
." 2.063
2.964
100 g of each standard is normally sufficient.
* Osmolality is defined as the freezing point depression in °C divided
by 1.86°C.
**Total molarity = moles Na+ + moles CT~.
2. Standard sulfuric acid solution, 1 N (for set-up only). Prepare 11
of standard 1.00 N H2S04 solution from pre-standardized concentrates
available from most supply houses. Alternately, H2S04 standards can
be prepared and standardized according to the directions given in
4-8
-------
"Standard Methods" (APHA) or "Methods for the Chemical Analysis of
Water and Wastes" (EPA).
3. Aqueous ammonia, 1 M (needed for set-up only). Dilute 67 ml of
concentrated (28%) reagent, grade ammonium hydroxide to 1 1 with
d. i. H20. Standardize by titration with I H H2S04: To a 100 ml
beaker add 25.0 ml of aqueous ammonia and titrate to an inflection
point (near pH = 4.6) with H2S04 using a pH meter and electrode as
indicators. Calculate the ammonia concentration from the expression
,MU , _ (Volume of H2S04)(Normality of H2S04)
(NH3) 25.0
4. Instrument test solution (needed for initial set-up only). Fill a
1 1 volumetric flask 1/2 full with d.i. H20. Add 100.0 ml of aqueous
ammonia. Add 79.06 g of reagent grade NH4HC03 and dissolve. Add
29.0 g of acetone and bring to 1 1 total volume with d.i. H20. This
solution has a total molarity of 2.6 assuming that the aqueous
ammonia is exactly 1.0 M and that complete disassociation occurs:
Total molarity = 1 M NH4 + 1 M HC03 + 0.1 M NH3 + 0.5 M acetone
Prepare additional instrument test solutions by volumetric dilution
with d.i. H2p to cover the range of interest. Keep all test
solutions refrigerated when not in use. Note: The aqueous NH3 must
be added before NH4HC03 in order to avoid the loss of C02 from the
solution.
Procedure
Most freezing point depression measurements will be made using
commercially-available cryoscopes, and the manufacturer's directions should be
4-9
-------
followed for this purpose. However, as a minimum the cryoscope should be
calibrated with the NaCl standards before and during each series of samples.
It is also recommended that the cryoscope be checked at least once with a
series of "instrument test solutions" described above. The measured "NaCl
equivalent molarity" should be within 7% of the ideal osmolality if the
instrument is to be used for retort waters.
Data Reporting
Plot the measured freezing point depression (or osmolality) vs. the
molarity of the NaCl standards. Use this figure to determine the NaCl
equivalent total molarity, the total molarity of an NaCl solution giving the
same freezing point depression as the sample. Report both the freezing point
depression and the NaCl equivalent concentration.
REFERENCES
EPA. 1979. Methods for Chemical Analysis of Water and Wastes. EPA-600/
4-79-020.
Lewis, G.N., M. Randall. 1961. Thermodynamics. Revised by K.S. Pitzer and
Leo Brewer. McGraw-Hill Book Co., NY.
Fox, J.P., D.S. Farrier, R.E. Poulson. 1978. Chemical Characterization and
Analytical Considerations for an In Situ Oil Shale Process Water.
LETC/RI-78/7. Laramie Energy Technology Center, Laramie, WY.
APHA. 1976. Standard Methods. American Public Health Association, Wash-
ington, D.C.
4-10
-------
SECTION 5
TOTAL AMMONIACAL NITROGEN IN RETORT WASTEWATERS
Total ammoniacal nitrogen, referred to here as t-NH3, is typically the
predominant cationic species in retort water and is frequently of interest to
pollution control engineers. The levels of t-NH3 found in retort wastewaters,
for example, are believed to inhibit biological oxidation, and such treatment
would require the removal and monitoring of t-NH3. The evaluation of other
treatment processes, such as reverse osmosis and stream stripping, also
requires monitoring of t-NH3. In addition, ammonia in retort wastewaters may
be reclaimed and sold to offset the costs of treatment (Probstein and Gold,
1978), thereby providing added incentive for accurate and precise measurements
of total ammoniacal nitrogen.
For these reasons t-NH3 has been measured frequently and with a variety of
methods. The common methods include (1) direct measurement by ion selective
electrode, (2) distillation from a basic medium followed by titration with
H2S04, and (3) automated or manual colorimetric procedures (Skougstad, 1979;
EPA, 1979; APHA, 1976). For example, Wildeman and Hdeffner (1979) and Prien
et al. (1977) both determined t-NH3 using method (2). Haas employed method (1)
for the analysis of TOSCO II retort water. Fox et al. (1978) compared three
methods for the analysis of a wastewater from an in situ process:
(1) distillation from a basic solution into H3B03 followed by titration with
H2S04, (2) distillation from a basic solution into H3B03 followed by the
automated phenolate finish, and (3) direct measurement by ion selective
electrode with no distillation. Comparable values for t-NH3 were obtained for
all three methods, suggesting that each one would be appropriate for monitoring
retort wastewaters.
5-1
-------
However, when attempting to apply these procedures in the environment of a
field laboratory in support of the simultaneous operation of various pollution
control systems, it became apparent that existing tests were unable to provide
accurate and reliable results in a timely manner. In such an environment
distillation procedures require excessive time, space, and equipment and
therefore are not practical. The colorimetric procedures are generally prone
to interferences from organic material, alkalinity, and turbidity, all of which
are plentiful in retort waters, and therefore were not attempted (APHA, 1976).
The ammonia electrodes manufactured by both Orion and Hnu have been applied to
various retort wastewaters with unsatisfactory results: the principal
difficulty with the electrode technique was the occurrence of erratic shifts in
the calibration curve. These shifts were presumably caused by high ionic
strength or high organic content of retort waters. Due to these effects,
results obtained in the field with the electrode were sometimes inaccurate by a
factor of two. Results could be obtained in the laboratory within a precision
of ± 20% (la) only by recalibrating after every few samples and repeating the
analyses whenever the calibration shifted. Thus, even to obtain results
reliable to within 20% required an unacceptable expenditure of labor and
excessive time periods. Improved techniques for the measurement of t-NH3 in
retort wastewaters were clearly required.
This chapter describes three parallel efforts to develop a practical
method for measuring t-NH3 in retort wastewater. Although not anticipated at
the onset, three different techniques were developed, each with its own
advantages and disadvantages. The first is a modification of the electrode
technique which minimizes exposure of the electrode to briny, organic-laden
retort waters. The second is an ion chromatographic technique, which also
5-2
-------
measures Na and K simultaneously. The third method involves absorption of UV
radiation by gaseous NH3 which is evolved from a basic solution.
Early observation of the performance of the ammonia electrode indicated
that while the intercept of the calibration curve would change erratically
during a series of samples, the slope and intercept would remain relatively
constant. The most dramatic changes would often occur when the electrode was
exposed to the most concentrated samples; however, such samples could not be
anticipated, resulting in frequent recalibration and repetition. The modified
electrode technique consists of keeping the electrode immersed in a low level
(e.g. 1-10 mg/1) standard adjusted to the ionic strength recommended by the
manufacturer. Microliter quantities of the sample are then added to the
standard without removing the electrode, and the concentration of the sample is
determined by the resulting change in electrode potential rather than by the
absolute potential. This procedure will be referred to in this chapter as the
sample addition technique (Orion, 1978).
In the ion chromatographic technique the cations in the "sample are
separated on a low capacity, pellicular, cation exchange resin using a weak
acid eluent. The eluent from the cation exchange column proceeds to a column
of anion exchange resin in the hydroxide from where the acid eluent is
converted to water. The eluent from the second column proceeds to an
electrical conductivity detector, where the cations are detected by the
resulting change in electrical conductivity. This type of ion exchange
chromatography is referred to in the literature as suppressed ion exchange
chromatography to indicate the presence of the second column, which suppresses
the background conductivity of the eluent (Mulik and Sawicki, 1979). Because
of the chromatographic nature of this method, the other monovalent cations in
5-3
-------
retort water are also separated, and It is common to observe Na and K during
the determination of t-NH3.
It has long been recognized that ammonia vapor exhibits a series of strong
absorption peaks in the UV region 180-210 nm (Tannenbaum et al. , 1953; Thompson
et al., 1963). While those peaks below 195 nm lie in the vacuum ultraviolet,
the remaining peaks are easily accessible by instruments available in most
laboratories and would seem to be an ideal means for measuring ammonia.
Preliminary investigation of the same references indicated that few serious
interferences would be found in retort water, assuming that the ammonia in the
water sample is evolved into the gas phase by the addition of base. The
absorbtivity of ammonia at 197 nm is approximately five orders of magnitude
larger than that of either 02 or H20. The simple amines exhibit broad,
relatively unstructured absorption spectra in the same region as ammonia. At
the outset of this study it was believed that such amines would not be present
in retort waters in sufficient quantities to cause a serious problem.
The possibility of using the absorption spectrum of ammonia as a probe for
measuring ammonia in various waters has been investigated by a number of
authors. These investigators typically added strong base to an aliquot of
sample in a gas bubbler, which was then purged with a continuous gas stream
into an absorption cell mounted in an atomic absorption or UV spectrometer. As
the ammonia evolved from the basic solution, it was swept into the absorption
cell, resulting in a transient absorption peak. Gunther et al. (1956) were one
of the first to use this absorption to measure ammonia. Their original
observations indicated that the ammonia peaks were sufficiently sharp to
require a spectrometer with a band pass of approximately one nm or better in
order to achieve a linear calibration curve. They also observed that the
5-4
-------
ammonia would gradually decay if left in the light path, a result they attrib-
uted to photodissociation. Factors affecting the response of the system to
ammonia were studied by several investigators, and these were seen to be the
concentration of base added to the sample, the shape and design of the bubbler,
the flow rate of the purging gas, and the temperature (Muroski and Syty, 1980;
Cresser, 1977; Cresser, 1976; Takahaski et a!., 1980). Thus Cresser recom-
mended that the temperature of the sample reaction vessel be controlled to
avoid excessive errors. Muroski and Syty also examined this method for inter-
ferences due to a number of ions, and observed no effects due to S03=, I~,
N02~, N03-, SCN~, S04=, C03=, S=, and CIT.
In spite of these investigations, the gaseous absorption method has not
gained widespread popularity, presumably because of the ease of applying either
the electrode method or the automated colorimetric methods to surface and
ground waters. However, the results support its application to retort waters,
which are analyzed only with difficulty by the standard methods. For this
reason the gaseous absorption technique was pursued simultaneously with other
approaches.
EXPERIMENTAL SECTION
Ammonium standards were prepared from analytical reagent NH4C1 in
deionized (d.i.), distilled water. All other chemicals were reagent grade
unless mentioned otherwise.
Sample Addition Technique
Measurements were made with an Orion Research (Cambridge) membrane ammonia
electrode, Model 95-10, in conjunction with a Keith!ey Model 616 Digital
electrometer. Potentials were readable to ± 0.1 mV.
Before analyzing samples the electrode was first standardized. After the
electrode had been equilibrated as specified by the manufacturer, it was
5-5
-------
immersed in 50 ml of 0.1 M NaOH in a magnetically stirred beaker. 10 (jl of
10,000 mg/1 NH4+ standard was added to give a solution concentration of 2 mg/1.
Successively larger volumes of standard were added to calibrate the electrodes
over the range of 2-800 mg/1. These data were then plotted on semi-log paper,
producing a straight line at concentrations over 10 mg/1 but frequently exhib-
iting a slight curvature below that level.
Before analyzing the samples, the electrode was rinsed and again immersed
in 50 ml of 0.1 M NaOH. 1.0 ml of 250 mg/1 NH4+ was added to give a concen-
tration of 4.90 mg/1, and the cell potential was recorded in two minutes. The
sample was added from disposable pipettes, starting with 0.01 ml and proceeding
to larger aliquots, until a potential change of 20-30 mV occurred. A reading
was recorded after one minute. Additional aliquots of sample were added until
a total change of 40-50 mV was recorded.
The concentration in the sample was then calculated based on the change in
cell potential. Referring to the calibration figure, it was determined what
concentration would have to be present in the beaker to cause the potential
change observed. Based on this and the volume of sample injected, the
concentration of NH4+ in the original sample was calculated. This procedure
was carried out for the two data points recorded (at 20-30 mV and 40-50 mV),
and the results were averaged.
Ion Chromatography
Ion chromatography was performed on a Dionex (Sunnyvale) Model 10 ion
chromatograph equipped with a cation separator column (#030831), cation
pre-column (# 030830), and cation suppressor column (# 030834). Eluent was
either 5 mM or 7.5 mM HN03 in distilled water; detection was with the elec-
trical conductivity detector which comes as standard equipment on the Model 10.
5-6
-------
Peak signals were processed with a Hewlett Packard Model 3390 digital
integrator. Quantisation was by peak height.
Gaseous Absorption Methods—
Original investigations of the gaseous absorption method were carried out
with a separate absorption cell and gas evolution chamber, similar in principle
to the devices described previously by others. The gas evolution chamber
consisted of a 155 ml erlenmeyer flask containing a magnetic stirrer and
thermometer. The flask was immersed in a water bath to maintain constant
temperature. The absorption cell was a 10 cm heated quartz cell mounted in a
Beckman Model DK2-A ratio recording spectrophotometer. The headspace over the
sample in the erlenmeyer flask was circulated through the absorption cell and
returned to the flask with a variable speed diaphragm pump (Bendix Model BDX44
Super Sampler). In practice, 50 ml of the sample or standard was placed in the
flask, 5 ml of 50%, w/w NaOH was added, and the flask was immediately sealed.
Stirring was initiated and the headspace gas was recirculated through the
absorption cell until a constant value was obtained.
Although this device was used for several preliminary experiments,
limitations soon became apparent. The temperature equilibration was too slow,
the apparatus needlessly complex, and the measurement cycle too slow. In
addition, a somewhat better precision and accuracy was judged to be desirable.
For these reasons the majority of work was carried out with the device shown in
Figure 5-1, which combined gas evolution, temperature equilibration, and
absorption measurement into a single cell. This device was mounted in the
sample chamber of a Gary Model 219 UV/visible spectrometer so that the light
path passed through the fused silica (Suprasil, Heraeus Amersil) windows shown
in the figure (70 mm path length). Rapid temperature equilibration was
achieved by circulating water from a temperature bath through the coils
5-7
-------
THERMISTER
SAMPLE INTRODUCTION
HEATED TOP HAT
SAMPLE CONTAINER
HEATING FLUID
STIRRING MOTOR
BASE PLATE
CLAMPS
FIGURE 5-1
Sample cell for measuring the uv absorption of gaseous NH
over a liquid sample '-
5-8
-------
soldered to the chamber body, which was constructed from a stainless steel,
125 ml beaker. Total chamber volume is approximately 150 ml. The top hat in
the figure serves to prevent condensation on the windows or window mounting
block, a feature which is necessary for work with heated samples. The window
mounting block was constructed of acrylic, although it is recommended that
future models be made from light metal in order to improve the rate of temper-
ature equilibration of the cell. Stirring of the sample is achieved with the
water driven stirring motor shown coupled to a magnetic stirring bar.
Temperature of either the solution or headspace is made with the thermister
shown. All connections for heating fluid, sample injection, etc. are made with
tubing through a demountable wall on the spectrometer, so that the spectrometer
sample area need not be opened once the sample chamber is installed.
The sample chamber is mounted in the spectrometer and heating fluid is
circulated until temperature equilibration is achieved. Approximately 30 ml of
sample or standard is injected followed with 3 ml of 10 M NaOH, both with
plastic disposable syringes. After the ammonia has equilibrated between the
_vapor and aqueous phase, the absorption of the gaseous NH3 at 197 nm is
measured, either by scanning across the 197 nm peak or by remaining fixed at
the peak maxima. All measurements were made with a fixed bandpass of 0.6 nm
with the spectrometer operated in the double beam mode.
RESULTS AND DISCUSSION
Sample Addition Method
At the onset of this investigation various approaches were attempted in
order to minimize the effect of the drifting intercept of the calibration
curve. The first attempt avoided the use of a calibration curve entirely. As
with the other procedures, this procedure began by immersing the electrode in a
5 mg/1 NHJ/0.1 M NaOH standard. Aliquots of the sample were then added to
5-9
-------
effect potential changes of approximately 20 and 40 mV. Potential and volumes
were recorded at each of the three data points, including the 5 mg/1 standard.
Based on these three data pairs, it is possible in theory to calculate the
concentration in the original sample by using the Nernst equation in the form
(1) E - E = m log £-
° Lo
r Co Vo - Vs Cs
Vo - Vs
m = slope (constant)
V = volume of sample added
VQ = original volume of standard
C - concentration in the sample
o
CQ = concentration of the original standard (i.e., 5 mg/1)
EQ = cell potential for the original standard
E = cell potential after sample addition
and solving for C . Here the known quantities are E , C , V , and two pairs of
E and V , and the objective is to solve for m and C . This calculation was
s s
carried out by numerical iteration on a Hewlett Packard 9825 desk top computer
to a precision of 0.1% relative.
Although computational errors were reduced to negligible amounts, results
of this procedure were erratic, the concentration often varying by a factor of
two for sequential measurements of the same sample. For this reason the
sensitivity of the calculated result to errors in the measured potential was
determined by artificially varying the potential which was entered into the
calculation. It was observed that 1.0 mV error in a 10 mV potential change
caused an error in the calculated concentration of 94% relative; an error of
1.0 mV in a 20 mV change, an error of 30% relative. Since the electrode
5-10
-------
potential cannot be measured to better than ± one or two mV due to drifting
responses, this approach was abandoned.
The first attempt to incorporate a standard curve into the sample addition
technique used the same procedure as described in the Experimental section of
this chapter, except that the calibration curve was assumed to be linear (on a
semi-log plot). With this assumption the sample concentration is still calcu-
lated based on equation 1, except that m is fixed and is obtained from a
previously obtained calibration curve. Fixing m results in a greatly simpli-
fied calculation which can be carried out with a pocket calculator. However,
it soon became apparent that the calibration curve obtained in the laboratory
with pure standards exhibited sufficient curvature at concentrations below
10 mg/1 to cause a major source of error. This approach was therefore
abandoned in favor of the method described in the Experimental section using a
plot of the calibration curve to calculate the relative concentrations.
However, the approach described above would result in no loss in accuracy
should the calibration curve prove linear down to 5 mg/1, as is sometimes the
case. The simplified calculation possible with this procedure results in
significant time savings to the analyst.
The remaining measurements with the sample addition technique were made
using the calibration curve as described in the Experimental section. The
precision of this method was estimated by analyzing 12 oil shale retort and
synthetic fuel wastewaters in duplicate. Each individual determination itself
consisted of the average of two measurements made after two separate sample
additions. The average difference between each (averaged) determination was 6%
relative. However, this estimate of precision is probably overly optimistic
since duplicates were measured sequentially, and perhaps more importantly, the
same quantities of sample were added during each analysis. Another estimate of
5-11
-------
the precision can be obtained by examining the two concentrations which were
averaged to obtain the individual results. Examining 24 such data pairs yields
an average difference of 17%, with an average precision of 17/V2 = 12% for the
average of two determinations. Thus, the precision for an individual determi-
nation, consisting of the average resulting from two sequential sample
additions is 6-12% (la).
The lower range of the sample addition technique is limited by the neces-
sity of adding sufficient sample to cause a shift in potential which is large
enough to measure accurately, while at the same time not adding so much sample
that the matrix of the resulting solution is changed. Referring to equation 1
and setting CQ = 5 mg/1, AE = 40 mV, and the maximum value for V = 5.0 ml, one
calculates a minimum value for the sample concentration, C = 210 mg/1.
Decreasing CQ or increasing V both improve the lower limit, albeit at the cost
of poorer performance since the electrode is less stable at lower concen-
trations and is also affected by changes in the matrix of the sample.
Ion Chromatography
Figure 5-2 is included to illustrate several features of the ion chroma-
tographic determination of ammonia and other ions in retort waters.
Figure 5-2a shows the elution of Na+, NHJ, and K+ using a 7.5 mM HN03 eluent.
NHJ and K+ elute close enough to interfere if either are present in large
excess. For this reason, among others, quantitation was accomplished with peak
height measurements, which are less sensitive than peak area to the occurrence
of nearby peaks. Fortunately, K+ is not normally observed in retort waste-
waters and the chromatogram shown in Figure 5-2b is typical of those encoun-
tered. Both NH| and K+N+ are clearly resolved and can be quantitated if
desired. Figure 5-2c shows a chromatogram for a 10 mg/1 standard of Na+, NHJ,
and K+ obtained with a 5 mM HN03 eluent. The improved resolution between NHJ
5-12
-------
Start
I4min
.<
Start 22min
Start
I4min
i—ft
Start
No"1
28min
FIGURE 5-2
Ion Chronidtoqrams of Na+, NH.+ and K+
5-13
-------
and K+ resulting from the weaker eluent is readily apparent, and the
determination of these three species in the same sample is feasible, assuming
the absence of unknown interferences. Methyl amine elutes immediately after
the NH| peak, and interferences due to amines should be considered in samples
such as retort waters containing high levels of organic material. However, the
retort waters analyzed in the authors' laboratory have shown no peak following
the NH| peak, suggesting that amines are not a problem in these waters.
Figure 5-2d shows a chromatogram of leachate from burned, spent shale. Na+,
K+, and a trace of NH| are clearly separated and were reported.
The ion chromatographic method has been used routinely in support of
various field tests for a year, and during this time has performed with few
problems. However, a gradual deterioration of column performance was noted
over this time, especially during the analysis of leachate samples. The latter
effect is likely due to the occurrence of large concentrations of alkali earth
elements in the leachate which are difficult to remove from the analytical
column. Periodic treatment with sodium tartate eluent temporarily improves but
does not restore entirely column performance. It is recommended, therefore,
that additional precautions be included for such samples. For example, more
frequent changing of the pre-column may eliminate degradation of the analytical
column due to divalent cations. Most experience with the ion chromatographic
system has been with condensates from pilot-scale oil shale retorts and such
waters have shown little effect on the column. Column manufacturers now
recommend the use of HC1 instead of the HN03 eluent as a means of prolonging
column life.
Statistical performance criteria, of course, depend on the exact
conditions of the analyses and vary between location, environmental conditions
and operator. However, the following has been typical of the authors'
5-14
-------
experience. During a two week field test the ion chromatograph was operated in
an unheated laboratory where temperatures fluctuated widely. During this
period 65 samples were analyzed in duplicate for NH^ with an average difference
of 2.3%. This difference is somewhat optimistic for such environmental
conditions, since sample pairs were analyzed in immediate sequence rather than
at different times. For samples analyzed in the laboratory the recovery has
averaged 104% (n = 4), not significantly different than unity. The main
limitation to reproductibility has been a drifting response, perhaps due to
temperature fluctuations, and a slight shift in calibration upon switching
suppressor columns, problems which may be minimized in the future by thermally
controlling the detector and installing a fiber suppressor column, which can be
used continuously. For determination of Na+ and K+ in leachate, the average
difference for 6 pairs of samples selected at random was 5.2% and 5.8%
respectively. Again, the principle limitation for these species has been a
drifting response which occurs as the analytical column is exposed to the high
Ga and Mg levels expected in leachate samples.
The dynamic range for NH| is approximately from 1-300 mg/1, although the
calibration curve is non-linear throughout parts of this region. Below ~5 mg/1
the peak height is proportional to concentration, while above vLO mg/1 peak
height is proportional to ~(NH3)-5~-6. This is reasonable for a weak base
which is only partially ionized in the detector.
Gas Absorption Method
Experiments were performed with the first device described in the Experi-
mental Section (separate absorption and gas evolution chambers) to assess in a
preliminary manner the feasibility of the gaseous absorption technique.
Characteristics of the spectra obtained with standard solutions and real retort
wastewaters were initially observed. As expected, artificial standards yielded
5-15
-------
spectra with a series of sharp peaks from approximately 210-197 nm (the vacuum
uv, cutoff for the instrument in use). More importantly, with retort waters a
clean ammonia spectrum was observed with no apparent interfering peaks. This
observation was especially important because organic compounds absorb in this
spectral region, and it was of obvious concern that organic compounds in retort
water would evolve and interfere significantly. Closer examination for other
absorbing species involved suppression of the ammonia spectrum by acidifying
the samples with HC1 to a pH of approximately 3. Under these conditions no
absorption was observed in the region 210-197 nm, except for one sample which
exhibited an intense absorption spectrum consisting of a series of sharp peaks.
Based on the appearance of the spectrum (Thompson et al., 1963), as well as the
fact that it disappeared when the sample was made basic, this absorption was
assigned to S02. This assignment is also supported by the common occurrence in
retort waters of thiosulfate, which decomposes to S02 upon the addition of
acid. Preliminary investigation based on a limited number of samples thus
indicated that retort waters when made basic by the addition of NaOH, yielded
an easily observed ammonia spectrum free of obvious interfering peaks.
Next, a calibration curve was established using standards over the range
3-300 mg/1, and was observed to be linear over this range but curved at higher
concentrations. Using this curve, three retort water samples were analyzed for
ammonia. Results compared reasonably with previous analyses by electrode.
Using this same device the importance of temperature control was also
established by comparing the absorbance of a 100 mg/1 standard over a range of
temperatures (Figure 5-3). At 50°C a change of 1°C was seen to cause a 3%
change in absorbance, a major source of error, especially since the addition of
NaOH to the sample often caused a temperature increase of 5-10°C, which was
only slowly dissipated.
5-16
-------
0.8
0.6
8
CO
QZ
O
CO
aa
0.4
0.2
Effect of Temperature of Cell on the
Absorbance at 197.35 nm of a lOOppm
NH3 Standard
0
10
20
30
40
50
60
70
CELL TEMPERATURE
FIGURE 5-3
Effect of cell temperature on response. (Standard concentration =
100 mg/1 NH3, absorption wavelength = 197.4 nm)
5-1?
-------
This implied that better temperature control was required. In addition, a
more rapid method of changing samples was required to make this approach
practical. For these reasons attempts were made to construct a continuous flow
system which would automatically enter the sample, meter the reagents,
equilibrate the temperature, and transfer the gaseous headspace sample to the
absorption cell. Such a device, modeled after the automated Hg method
described in the EPA (1979) handbook was assembled, including a constant
temperature bath to maintain temperature control to better than ± 1°C. While
this device worked well for standards and even some retort waters, it failed as
a general technique for retort waters due to a persistent foaming problem with
some retort waters which prevented the complete separation of gas and water
phases. This caused liquid sample and reagent to be carried into the
absorption cell, resulting in a worthless spectrum and a lengthy cleaning
process. Several defoaming agents, such as anti-foam B (Alpkern Corp.,
Portland), and mechanical separators of various size and design including
hydrophobic filters were evaluated, but none proved adequate for all samples.
Because the foaming of one sample invalidated the results for all subsequent
samples, the continuous flow approach was abandoned in favor of apparatus shown
in Figure 5-1.
The cell shown in Figure 5-1 was designed to'(1) minimize the effect of
foaming, and (2) improve the rate of temperature equilibration. The sample
container in this cell is made from 0,025" thick stainless steel as opposed to
the standard laboratory glassware used in the original device. The new sample
container achieved temperature equilibration for a 30 ml sample in a few
minutes.
Tolerance to foamy materials was tested with several samples known to
readily produce stable foams, which were introduced into the chamber and
5-18
-------
treated normally. Samples for this purpose included 5% Joy dish detergent in
water, 5% Joy + 5% vacuum oil in water, 0.07% sodium dodecylsulfate in water,
0.07% sodium dodecylsulfate + 1% 3-in-l household oil, and several retort
waters from a variety of sources. No problems with foaming were encountered
with any of these samples, although water could be splashed onto the cell
windows if the magnetic stirrer were operated too fast. This cell also over-
comes certain disadvantages which have been reported by previous authors for
similar flow through devices designed for measuring t-NH3 by gas phase absorp-
tion (Muroski and Syty, 1980; Cresser, 1977; Takahaski et al., 1980). While
the response of flow-through systems depends on the gas flow rate, no similar
phenomenon occurs with the static cell. More importantly, the response is
relatively insensitive to the exact volumes of NaOH and sample which are added.
Although the samples were introduced only with the precision achievable with a
plastic, disposable syringe, the precision of the final analysis was still the
the best of the methods tested. (See below.) All results discussed below were
thus obtained using this sample cell.
Certain apparent problems should be corrected in future designs. When
first placed in the spectrometer, the cell required two hours to stabilize at
the desired temperature for analysis. Sample introduced before this period
would condense on the windows, where it would remain for hours. This long
initial warm up was ascribed to the thermally large window mount, which was
made from 16 mm thick acrylic. A window mount constructed from more thermally
conductive material would be preferable in the future. In addition, the flat
bottomed cell did not drain thoroughly between samples, requiring both water
and sample rinses to obtain accurate results. This could be improved by
changing the shape of the cell bottom. The stirring mechanism was not suffi-
ciently reliable for routine use and would occasionally stop. Thus, a
5-29
-------
mechanically-driven stirrer or a more strongly coupled magnetic stirrer, would
appear to be preferable to the design used in this model.
Figure 5-4 shows the spectrum of a 100 mg/1 NH£ standard over the range
210-188 nm. As can be seen, the peaks become increasingly more sensitive with
decreasing wavelength. Although not obvious from the figure, the noise level
also increases below 197 nm, making this region less desirable. Possible
advantages of using shorter wavelengths were investigated by first flushing the
Cary 219 spectrometer with dry nitrogen evolved from a liquid nitrogen tank.
However, the lines below 197 nm still tended to be noisier than those of longer
wavelength. Background absorption also becomes increasingly important below
approximately 200 nm. For example, when deionized, distilled water is
introduced into the cell at 60°C, a spectrum is observed in which absorption
increases with decreasing wavelength and results in an absorption of 0.004 at
197 nm. Qualitatively, this matches the reported spectrum of water
(Thompson e't. al., 1963). Since the absorbtivity of water increases approxi-
mately ten fold for every 5 nm decrease in wavelength in tfciis spectral region,
the use of wavelengths below 197 nm does not appear promising. The remaining
experiments discussed in this report were thus performed using the 197 nm peak
(spectrometer setting = 197.35 nm) and a band pass of 0.6 nm which represented
a compromise between maximum sensitivity and minimum noise.
Three slightly different calibration curves are shown in Figure 5-5 for
the concentration range 1-1,000 mg/1. The first curve was obtained with the
cell at 50°C using the total height of the 197 nm peak. The second curve was
obtained at a cell temperature of 60°C using the same procedure. The third
curve was determined by measuring the height of the 197 nm peak above a base-
line drawn between the two adjacent valleys to better reject any absorption due
to broad-band interfering peaks. The figure illustrates that the calibration
5-20
-------
NH, SPECTRUM
i ill i t i i i t i
WAVELENGTH (nm)
FIGURE 5-4
Spectrum of NHo in equilibrium with 100 mg/1 of
NH4+..at 60 C. (band pass = .0.6 nm)'
5-21
-------
o
o
o
o
o
o
o
4-3
Q.
S-
o
in
=3
O
O)
tfl
(O
01
o
o
o
p
o
-M
J3
O
in
OJ
>
3
O
c
o
us
I I I I I I I
I I I
o
o
-o
O
o
(Wrfd) NOIlVyiNHONOO CHN
5-22
-------
curve is linear from 1-100 mg/1, but slight curvature is apparent at 300 mg/1.
Nevertheless, the curve is usable over the entire range except that concen-
trations above 330 mg/1 are not accessible at 60°C. The importance of temper-
ature in this analysis suggests that the range could be adjusted to both lower
and higher concentrations by adjusting the temperature appropriately. Consid-
erably higher concentration could also be measured using a cell with a shorter
absorption pathlength.
In developing this procedure, the importance of technical procedures
become apparent. These included the necessity of choosing a stirring bar of
the proper size (ca 15 mm) to provide reliable stirring, and selecting a
stirring rate which was fast enough yet did not splash liquid onto the windows.
The rinsing procedure between samples could also affect the results, and a
rinse with deionized water followed by a rinse with the next sample was found
to be necessary for reliable results.
Many retort wastewaters contain a high content of dissolved ions and the
gas absorption technique was, therefore, tested for its ability to tolerate
samples of high ionic strength. Standards of 33 and 100 mg/1 were prepared
with and without the addition of strong electrolyte (NaCl at 47 g/1 and Na2S04
at 8 g/1). When the absorption of these samples was compared, no difference
was observed. Thus, the gas absorption technique appears to be insensitive to
the occurrence of strong electrolytes at ionic strengths far exceeding that
expected in retort waters.
Another major interest in testing the gas absorption method was evaluating
its ability to analyze solutions containing high levels of surfactants and
oils. The tolerance of the other two methods discussed in this report for
surfactants and oils is limited due to fouling of the electrode membrane and to
degradation of the chromatograph column. Thus, attempts were made to analyze
5-23
-------
solutions containing 100 mg/1 of NH3 and the same levels of surfactants and
oils used to test the cell with regard to its ability to tolerate foamy samples
(see above). These samples contained more oil and surfactant than would
normally be expected in a retort water sample and would form a separated layer
if allowed to stand for a minute. Success with these samples would therefore
imply more than sufficient tolerance to analyze retort water samples. Unfortu-
nately, the analysis of such samples gave rise to certain problems. The
equilibration time was excessively long, often requiring 30 minutes. Oils
caused a broadband absorption which was 30-50% of the NH3 absorption. (This
was subtracted by acidifying the sample to pH 3 and measuring the absorption
before adding the base.) The oil often left a residue which required flushing
out by pulling room air through the cell until the absorption decreased. Most
importantly, when the organic background was subtracted, the absorbance was too
low for the concentration of
-------
(one standard deviation). Precision of the second method, identically deter-
mined, was 0.6% relative. The recovery for the two methods was 103.6% and
99.3% respectively. The detection limit of'the gas absorption method, defined
as three times the background noise, was determined by measuring the absorbance
of a pure water sample six times. The detection limit so calculated was
0.24 mg/1 of NH3. Prudent laboratory practice would thus permit reporting
concentrations over approximately 3 mg/1.
Comparison of-Methods
In order to establish the veracity of a technique, it is clearly
advantageous to compare the results obtained on identical samples using
independent techniques. This was done for a set of 12 samples consisting of
oil shale retort wastewaters as well as a tar sands wastewater and coal
gasification condensate using the methods discussed in this chapter. These
results are tabulated in Table 5-1. Waters listed in this table are from
various field tests and interlaboratory comparison studies ("round robins").
However the waters are not necessarily representative of the process listed but
should cover the spectrum of sample types expected during the treatment of oil
shale retort and synfuel wastewaters. Results of the two variations of the gas
absorption method are listed, the "single" heading and the "scan" heading
referring to the total absorption and peak-above-valley variations discussed
previously.
For the purposes of interpretating the data in this table, the ion
chromatographic method was used as a reference to which the other results are
compared. The average difference listed on the bottom row was then calculated
according to the expression
2 n |xx - x2.l1
"1=1
5-25
-------
Table 5-1
COMPARISON OF AMMONIACAL N MEASUREMENTS OBTAINED BY INDEPENDENT METHODS
Concentration (mg/1 as NH3)
Gas Absorption
Sample
Number
1
2
3
4
5
6
7
8
9
10
11
12
Average
Description
Mixed feedwater to steam
stripper, 10/6/81
Mixed effluent from steam
stripper, 10/6/81
Geokinetics #17
Dilute Paraho
Oxy-6
S-7, reflux from steam stripper
Concentrate from reverse
osmosis treatment
S-ll, reflux from steam
stripper
S-3, reflux from steam stripper
Omega-9 (3,125)
Tar sands wastewater
Condensate from coal
gasification
di f f erence
Sample
Addition
895
441
2,128
6,736
8,197
11,290
4,702
12,934
18,282
3,595
73
3,903
12%
Ion
Chromatography
725
370
2,200
6,000
8,500
10,200
3,700
12,000
17,500
3,300
55
4,000
0.0%
Single
780
405
2,300
6,400
8,600
11,000
4,400
, 14,800
19,800
3,600
102
4,200
13%
Scan
720
325
2,150
6,250
8,400
9,800
4,200
12,000
17,000
3,450
50
4,200
5%
•,
5-26
-------
The average difference for the sample addition method is reasonable considering
the precision of the two methods involved in the calculation. Furthermore, no
systematic bias is apparent, and the accuracy of the standard addition method
is therefore limited by its precision. Similar statements can be made for the
scanning variation of the gas absorption method, except that agreement of this
method to the ion chromatographic method is better. The five percent agreement
between these two methods suggests that both are accurate to that amount or
better. However, the single wavelength variation of the gas absorption method
clearly gives results which are biased high, since every result obtained with
this method is higher than the corresponding result obtained with the scanning
variation and ion chromatographic method. This bias is most evident for
sample 12 (tar sands wastewater), but is less pronounced for the other samples.
In spite of the obvious bias with the single wavelength variation, the results
obtained with this method agree on the average with the ion chromatographic
results as well as do the results from the sample addition method. This
reflects the better precision of the absorption technique.
In completing the analyses using the single wavelength variation, attempts
were made to correct for background absorption by first neutralizing or acid-
ifying the sample and measuring the resulting absorption. The sample was then
made alkaline and the resulting absorption measured. Using this procedure,
corrections for background absorption were negligible. Apparently, the
background absorption resulting in the high bias for the single wavelength
variation results from compounds which are volatile in alkaline but not in
neutral or acidic solution. Thus, while the identity of the interfering
material in this technique has not been established, it must be an organic
base. Organic amines are possibilities although their presence has not been
5-27
-------
reported in oil shale retort wastewaters in significant quantities. Exami-
nation of spectra obtained with selected samples confirms the presence of
broadband absorbing species. Thus, the background spectrum can be calculated
by measuring the peak height above the valley and calculating the background by
comparison to the spectrum of pure NH3. The results of such a calculation are
shown in Figure 5-6. While the background is clearly a significant part of the
total absorption, not enough detail is present to suggest a particular compound
or class of compounds.
Performing the methods described in this chapter confirms that each has
definite advantages and disadvantages. The sample addition method requires the
minimum investment in equipment, the major items being an electrometer, an
ammonia electrode, and a set of microliter pipettes. It is also highly labor
intensive and automation of this procedure would require a substantial effort.
The ion chromatographic method can be performed with commercially available
equipment, the principal items being a suppressed ion chromatograph and peak
integrator or strip chart recorder. Labor requirements are less than the other
methods and consist primarily of periodically checking on the "proper operation
of the instrument and reducing the data. The major difficulty with the ion
chromatograph has been degradation of the columns when exposed to Ca and Mg and
the drifting response which occurs over time or when the suppressor columns are
exchanged for regeneration. The gas absorption method with background correc-
tion appears promising as a continuous, on-line monitor because of the sim-
plicity of the measurement and the minimum sample handling requirements.
Optical absorption instruments are often used for process monitors because of
their inherent simplicity and ruggedness, yet no suitable gas absorption
instrument is presently available. If such an instrument is developed, it
should include a method for correcting broadband background absorption. This
5-28
-------
0.?
0.6
0.5
0.4
o
CQ
cc
o
to
CO
<£
0.3
0.2
0.1 -
190
200
Measured
Spectrum
1 L II 1 I 1 I
Calculated
Background
210
WAVELENGTH (nm)
FIGURE 5-6
Gas absorption spectrum over a retort wastewater
5-29
-------
could complicate the device somewhat. Another advantage of the gas absorption
method is that it responds only to ammoniacal N in the form of aqueous NH3.
This device could thus be used to distinguish between aqueous NH3 and NH| and
also to measure the volatility of ammoniacal N as a function of temperature and
pH, a measurement of interest to those involved in removing ammoniacal N from
wastewaters.
PROTOCOL FOR TOTAL AMMONIACAL N BY ION CHROMATOGRAPHY
Principle
NH| is separated from the concomitant cations on a cation exchange resin
using weak acid eluent. The NH| is detected by electrical conductivity after
the conductivity of the eluent is suppressed by passage through an anion
exchange resin in the OH~ form.
Comments
This method also measures Na and l< simultaneously with NH|. However, K
co-elutes with methyl amine and therefore cannot be quantitated when both are
present.
Samples with major amounts of alkaline earth metals cause the separator
column to degrade. Such samples (e.g., leachates) must therefore be approached
cautiously.
The experimental work supporting this method was performed on a suppressed
ion chromatograph. No comment is made or implied regarding the efficacy of
non-suppressed ion chromatographs.
Reagents and Supplies
1. Ammonia-free water. Waters stored in the laboratory are frequently
contaminated due to the ubiquitous presence of NH3. It is therefore
important to check the purity of the water used to make up reagent
solutions by injecting an aliquot into the ion chromatograph. No NH£
peak should be visible on the recording.
5-30
-------
2. Ammonia stock solution 10,000 mg/1 as NH3. Dissolve 31.30 g of
analytical reagent grade NH4C1 in water and bring to 1.00 liter volume
with ammonia free water.
3. Stock Na solution, 10,000 mg/1 as Na. Dissolve 25.42 g of analytical
reagent grade NaCl and bring to 1.00 liter volume with ammonia free
water.
4. Stock K solution, 10,000 mg/1 as K. Dissolve 19.07 g of analytical
reagent grade KC1 and bring to 1.00 liter total volume.
5. Mixed working standards. Prepare mixed standards over the range of
1-330 mg/1. Because the calibration relationship is often curved,
especially near the end of the working curve, the standard concentra-
tions should bracket the sample concentration.
6. Disposable syringe, 3 or 10 ml capacity. Exposure of the samples and
standards to the atmosphere can be minimized by outfitting the syringe
with barbed adaptors to permit the connection of small tubing.
Transfer of the sample or standard into the syringe -can then be
effected conveniently with minimum exposure to air.
7. Syringe-type filters, 0.45 urn pore diameter. The samples should be
filtered before injection into the chromatograph. This is accom-
plished most easily with a filter holder which fits into the luer-lock
syringe tip during injection of the sample.
8. Eluents. Prepare a stock solution of 1 M HC1 or HN03 using concen-
trated, pre-standardized reagent available from most supply houses.
a. 7.5 mM eluent is used for Na+ and NHJ but does not adequately
separate K+. Add 30 ml of the stock solution to a 4 liter
container and bring to volume.
b. 5.0 mM eluent is used to separate Na+, NH| and K+. Add 20 ml of
stock solution to a 4 liter container and bring to volume.
5-31
-------
Apparatus
1. Suppressed ion chromatograph equipped with a 100 |jl sample loop,
electrical conductivity detector, and a standard cation separator
column and pre-column.
2. Dual channel strip chart recorder with separate ranges or a digital
peak integrator capable of reporting peak height.
3. Apparatus or supplies for diluting samples up to 1,000 fold.
Procedure
Refer to the operator's manual of the ion chromatograph for general
operating procedures.
The operating range of the instrument equipped with a 100 pi injection
loop is approximately 1-300 mg/1. However, this can be readily shifted to a
higher range by decreasing the volume of the sample loop, should the operator
wish to avoid diluting sampVes and has no need for the lower concentration'
measurements.
While the analysis can be performed with a single suppressor column, it is
recommended for routine work to include two suppressor columns. The second
suppressor column can be regenerated while the first column is in use, so that
a suppressor column is always available. Otherwise, all analyses must stop
while the suppressor column is regenerated. An alternative to the use of two
suppressor columns may be the use of a single cation fiber suppressor which
needs no regeneration. This device has recently been made commercially avail-
able but has not yet been tested in the authors' laboratory.
To begin the analysis, inject a series of standards covering the range
1-300 mg/1. Plot the resulting peak heights on log-log paper and select a
region on the calibration curve which is linear or nearly so,. Because the
response of the ion chromatograph often drifts during the course of analyses,
5-32
-------
it is important to recalibrate periodically, including before and after each
series of samples and upon changing suppressor columns. It is frequently more
convenient to maintain the calibration of only part of the possible dynamic
range, and to dilute samples to fall within that range.
Samples are injected identically to the standards and are quantitated by
use of the calibration curve prepared as described in the last paragraph.
The precision of this analysis should be determined, not by repeating a
series of samples in sequence, but by operating a set of samples spaced over a
several day period. The latter approach will include contributions due to
instrument drift, while the former approach may result in an overly optimistic
estimate.
PROTOCOL FOR TOTAL AMMONIACAL N BY MODIFIED ELECTRODE METHOD
Principle
An ammonia ion selective electrode is maintained in a solution of pure
NH4C1 standard to which small amounts of sample are added. The concentration
in the sample is determined by the change in potential upon addition of the
sample.
Comments
This method is applicable to samples containing 200 mg/1 or more of
ammoniacal N. It requires less equipment than the ion chromatograph method but
is also more labor intensive and less precise.
Reagents and Supplies
1. 0.1 M NaOH. Add 4.0 g NaOH to a 1 liter flask. Dissolve and bring to
volume with deionized water. Prepare 50 ml per sample. Note: water
contaminated with NH3 will make this analysis more difficult by
causing increased curvature of the working curve. See APHA (1976)
5-33
-------
method 418A for methods to minimize NH3 contamination of reagent
water.
2. NH3 stock solution, 10,000 mg/1 of NH3. Dissolve 31.30 g of
analytical reagent grade NH4C1 in d.i. water and bring to 1 liter
total volume.
3. Working NH3 standard, 1,000 mg/1. Dilute 10.0 ml of the 10,000 mg/1
stock solution to 100.0 ml total volume.
Apparatus
1. Ammonia ion selective electrode, membrane type.
2. Electrometer, suitable for use with ammonia electrode and readable to
± 0.1 mV on 100 mV scale.
3. Beakers, 100 ml, one per sample.
4. Magnetic stirrer.
5. Set of micropipets covering the range 10-1,000 jjT- A set including
volumes of 10, 25, 50, 100, 250, and 1,000 (jl is suitable.
Procedure
1. Preliminary. Place 50.0 ml of 0.1 M NaOH in a 100 ml beaker, add a
magnetic stirring bar, and immerse the electrode at a slight angle to
discourage the formation of bubbles on the surface. Begin stirring
and observe the potential. This potential should be at least 60 mV
less than the potential of the 2 mg/1 standard prepared below. A
smaller difference suggests contamination of the 0.1 M NaOH with NH3.
2. Standardization. To the beaker prepared in the previous step, add a
series of aliquots of NH3 standard stock solution to prepare a series
of standards. Record the cell potential after each addition, allowing
5-34
-------
a few minutes for the electrode to stabilize. A suitable series of
standards follows:
Aliquot
Volume
(Ml)
100
250
50
100
250
1,000
3,000
Concentration of
Solution Added
(ma/1 )
1,000
1,000
10,000
10,000
10,000
10,000
10,000
Total
Vol ume
(ml)
50.10
50.35
50.40
50.50
50.75
51.75
53.75
Final
Concentration in
Beaker (mg/1)
2.00
6.95
16.9
36.6
85.7
277
825
Plot the log of the concentration vs. the cell potential on semi-log
paper. Repeat this calibration procedure to assure that the slope
and shape of the curve is reproducible. Remove and rinse the
electrode.
3. Analysis. Add 50.0 ml of 0.1 M NaOH to a 100 ml beaker as above.
Immerse the electrode and begin stirring. Add 0.250 ml of 1,000 mg/1
standard, resulting in a concentration of 4.98 mg/1, and record the
cell potential, E , after allowing a few minutes for equilibration.
Begin adding the sample, starting with 10 ul aliquots and progressing
to larger volumes until a total change in cell potential of
approximately 20 mV occurs. Record the cell potential, EI and total
volume'added to this point. Proceed to add larger, volumes, up to
5.0 ml, until a total change of approximately 40 mV is observed.
Record the cell potential, E2, and the total volume added to this
point.
4. Calculation. The sample concentration is calculated based on the
potential difference EI~EO and E2-E0 and not on the absolute cell
potential. Consider first EI. Referring to the calibration plot
5-35
-------
prepared in Step 2, set EQ to the potential corresponding to
4.98 mg/1. Determine the concentration resulting in a potential
increase of Et-Eo and call this C. The sample concentration
calculated at this point is then given by:
r - C (Vi + 50) - 250
cs "—71-
Where volumes are in ml and concentrations in mg/1. Repeat this
calculation for the second data point (E2, V2) to obtain a second
estimate of C . Average these two results.
REFERENCES
APHA, 1976. Standard Methods for the Examination of Water and Wastewaters,
14th ed. American Public Health Assoc., Washington D.C.
Cressor, M.S. 1976. Determination of Nitrogen in Solution by Gas-Phase
Molecular Absorption Spectrometry. Anal. Chimica. Acta., 85:253.
Cressor, M.S. 1977. Factors governing the sensitivity of Ammonia-Nitrogen
Determination by Gas-Phase Molecular Absorption Spectrometry. Laboratory
Practice. Jan., .1977:19,
EPA. 1979. Methods for Chemical Analysis of Water and Wastes. EPA-600/
4-79-020. United States Environmental Protection Agency, Cincinnati,
Ohio.
Gunther, F.A., J.H. Barkley, M.S. Kolbezen, R.K. Blenn, E.A. Staggs. 1956.
Quantitative Microdetermination of Gaseous Ammonia by its Absorption at
204.3 nm. Anal. Chem., 28:1985.
Haas, F.C. 1979. Analysis of Tosco II Oil Shale Retort Water, Analysis of
Waters Associated with Alternative Fuel Production. L.P. Jackson and
C.C. Wright, eds. American Society for Testing and Materials publ.
Philadelphia, PA.
Mulik, J.D., E. Sawicki, ed. 1979. Ion Chromatographic Analysis of
Environmental Pollutants. Vol. 2. Ann Arbor Science, Ann Arbor.
Muroski, C.C., A. Syty. 1980. Determination of the Ammonium Ion by Evolution
of Ammonia and Ultraviolet Absorption Spectrometry in the Gas Phase.
Anal. Chem., 52:143.
Orion. 1978. Instruction Manual for the Ammonia Electrode. Orion Research,
Cambridge, Mass.
5-36
-------
Prien, C.H. et al. 1977. Sampling and Analysis Program at the Paraho Oil
Shale Demonstration Facility. DRI Report 5624.
Probstein, R.F. and H. Gold. 1978. Water in Synthetic Fuel Production. MIT
Press. Cambridge, Mass.
Skougstad, M.W., M.J. Fishman, L.C. Friedman, D.E. Endrmann, S.S. Duncan, ed.
1979. Techniques for Determination of Inorganic Substances in Water and
Fluvial Sediments. U.S. Government Printing Office, Washington, D.C.
Takahashi, M. et al. 1980. Improvement of the Sensitivity in the
Determination of Trace Ammonium-Nitrogen by Gas-Phase Ammonia Molecular
Absorption Spectrometry. Can. J. Spectroscopy 25(1):25.
Tannenbaum E., E.M. Coffin, A.J. Harrison. 1953. The Far Ultra-Violet
Absorption Spectra of Simple Alky! Amines. J. Chem. Physics, 21:311.
Thompson, B.A., P. Hortech and R.R. Reeves. 1963. Ultra-violet Absorption
Coefficients of C02, 02, H20, N20, NH3, NO, S02 and CH4. J. Geophysical
Research, 68(24):6431.
Wildeman, T.R. and S.R. Hoeffner. 1979. Paraho Waters: Characteristics and
Analysis of Major Constituents, ASTM Symposium. Pittsburgh, PA.
June 1979.
5-37
-------
SECTION 6
TOTAL SULFUR' IN RETORT GAS
In addition to liquid fuel, oil shale retorting produces a synthetic fuel
gas, known as retort gas, as a major energy-containing product. The compo-
sition and volume of retort gas vary widely depending on the retorting process,
and analytical methods developed for such gas must be insensitive to matrix
composition. For example, retorting processes which heat the oil shale with a
heat exchange media ("indirect processes") produce a gas containing approxi-
mately 80-100% combustible components (H2, CO and hydrocarbons) with a heat
content of approximately 1,000 Btu/SCF (^9 kcal/1). The direct retorting
processes, which combust a portion of the oil shale in the retort, yield a gas
containing approximately 80-90% inert material (N2, C02) with a heat content
approximately ten times less (Nevens et a!., 1980).
A major air pollution concern associated with proposed oil shale develop-
ment is the emission of S contained in the retort gas. Untreated retort gas
may contain S gases at percent levels in the form of COS, H2S, CS2, or various
mercaptans. However, since present development plans call for the combustion
of retort gas as a source of steam and electricity, the various S compounds
would be emitted as S02.
Numerous S-control technologies, such as the Stretford process and various
amine scrubbing techniques, are presently being considered for the removal of S
species from retort gas prior to combustion, and such systems may be field
tested during the coming years. It is thus important that a means be available
for measuring total S in retort gas so that the control efficiency of such
devices can be determined.
6-1
-------
A measure of total S is also needed for quality control purposes as a
check on the determination of the individual S species. As discussed in
Chapter 7, methods are not available at the present time for the determination
of all possible S species in retort gas. Even if such methods were available,
the difficulty of calibrating for each species would normally be too time
consuming to perform routinely. A measurement of total S thus provides some
assurance that major species have not been overlooked or measured incorrectly.
The method should be sufficiently rugged for use in a field laboratory.
In addition, automatic and rapid operation is essential. Reagent requirements
should be minimal. The analytical technique must also be compatible with a
sampling system which avoids cooling the sample or otherwise exposing the
sample to liquid water, since the S species found in retort gas may be removed
by exposure to condensed water (Wallace, 1980).
Of the instruments available for measuring total S in related types of
samples, the flame photometric detector (FPD) is probably the most common
(APHA, 1977). However, several characteristics of this device suggest that it
is particularly unsuited for retort gas. These characteristics include inter-
ferences due to hydrocarbons and perhaps H2 and C02, instability of the flame
upon exposure to combustible gases, and a dynamic range which is too low
(~10~10 - 10~7 g/sec) for retort gas. The latter feature requires that retort
gases be diluted approximately 1,000 fold prior to introduction into the flame,
an operation which is difficult to perform well even under laboratory
conditions. In addition, the response of the FPD is compound-specific, that
is, the response per mass of S depends on the type of S compound
(Wallace, 1980; Lucero and Paljug, 1974; Gangwal et a!., 1979; Pearson and
Nines, 1977; Doehler et al. , 1977). The latter feature would be particularly
6-2
-------
trouble.some when trying to determine Sin retort gas without a prior knowledge
of the forms of S present.
ASTM method D 1072 provides a manual method for measuring total S in a
combustible gas. In this method the fuel gas is burned, the combusted S gases
are scrubbed in a solution of Na2C03, oxidized to H2S04, and titrated with
BaCl2. This method is obviously time consuming, although the basic chemistry
appears sound.
A second ASTM method, D 3031, provides for the determination of trace
levels of S in natural gas by hydrogenating the sample in a hydrogen stream at
900°C and determining the H2S produced with the methylene blue method. This
method also appears time consuming, and it is questionable whether the hydro-
genation step would work as efficiently with retort gas as with natural gas.
Aside from the FPD, the literature in general describes two main cate-
gories of instruments for measuring S in hydrocarbon matrices: those which
reduce all S compounds to H2S prior to analysis, and those which oxidize all S
compounds to C502 or S03. Both methods are discussed extensively in the
literature (Kolthoff and Elving, 1961; Kuck, 1978; Wallace, 1980). In summary,
either operation can be performed quantitatively, although problems with coking
have been reported in reductive systems. For this study an oxidative approach
was selected because of the ready availability of continuous and
interference-free S02 analyzers. The objective was then to develop a system
for producing S02 in constant proportion for each of the various S compounds.
This goal was first attempted by passing various S compounds through a
heated tube in the presence of air, an approach which is used in several
laboratory instruments (Martin and Grant, 1965; Wallace et al. , 1970; Culmo,
1972). When this approach failed, the next approach was to pass the S gases
6-3
-------
through a flame, an approach which achieved stoichiometric conversion of all
the S compounds tested.
The final system was challenged with a variety of S compounds to test its
ability to measure total S independently of molecular form. Finally, the
integrated sampling and analysis system was field tested during operation of
the 150-ton simulated in situ retort located at the Laramie Energy Technology
Center. The data presented in this paper suggest that the system developed
should be applicable to a wide variety of synthetic fuel gases as well as oil
shale retort gas.
EXPERIMENTAL ' .
Gas flow rates were measured with Matheson Model 7642T rotameters cali-
brated with the gas in use. Selected flow measurements were also made with a
Tylan mass flow meter with sensing heads covering the range 1-1,500 seem. The
calibration of gas flow equipment was carried out with either dry test meters
or bubble meters, both of which were calibrated in reference to NBS standards.
Liquid flows were controlled with a Fisher Labortechnik syringe pump which had
been calibrated gravimetrically. "
Gas standards of the volatile gases—S02, H2S, and COS—were prepared by
dilution with a flowing stream of carrier gas—either H2 or N2. Gas standards
for the less volatile species, CS2, dimethyl disulfide and thiophene, were
prepared by metering a constant flow of liquid into a flow of diluent.
Evaporation occurred in a glass tube under a heat lamp. Gas standards of
methyl mercaptan were prepared by metering a flow of 0.40% methyl mercaptan in
N2 into a flowing diluent stream.
• S02 was detected with a Thermo Electron Corp. Model 40, pulsed
fluorescence stack gas monitor. The range of this device is approximately
5-5,000 ppmv. All gases were supplied by Matheson. Reagents were as follows:
6-4
-------
Thiophene, analytical reagent grade (MaTHnckrodt); 0.4% methyl mercaptan in
N2, analyzed mixture (Matheson); Thiophene liquid (Eastman); Methyl mercaptan
liquid (Eastman); CS2, analytical reagent (Mallinckrodt).
RESULTS AND DISCUSSIONS
A large number of S species can conceivably occur in retort gas and it was
not feasible to test the conversion of every possible gas. Therefore, a group
of six compounds representative of different compound classes were selected for
detailed testing. These included S02s H2S, COS, methyl mercaptan, thiophene,
and carbon disulfide.
Measurement of total S requires that all forms of S in the retort gas must
be converted to a species which can be readily measured. Because of the ready
availability of reliable and accurate S02 monitors, attempts were made to
develop a method which would convert all S compounds to S02 with the same
conversion efficiency. This was first attempted by passing the gases of
interest through a heated tube in the presence of air. The heated tube used in
these experiments was approximately 12" long by 1" diameter and packed with
Raschig rings. A residence time of approximately 4 seconds was required to
effect optimal conversion. Effluent S02 was measured by a pulsed fluorescence
S02 monitor which had been calibrated with pure S02. Results were compared to
values expected for 100% conversion.
Two main problems were evident with this approach. When S02 was intro-
duced into the heated tube, 100% recovery was achieved with temperatures up to
720°C. Above this temperature loss of S02 occurred, presumably due to
formation of S03. When introducing H2S into the heated tube, a temperature of
at least 650°C was required to effect 100% conversion. Unfortunately, in the
temperature range 650-720°C the conversion of .carbon disulfide, methyl
mercaptan, and thiophene were incomplete and variable, approximately 84*, 83 and
6-5
-------
93%, respectively. Thus, the problems with this approach were (1) the depen-
dence of the conversion efficiency on temperature and molecular form, and
(2) the production of significant amounts of S03 which would be expected to
damage eventually the optics of the S02 monitor.
These results can be explained by consideration of Figure 6-1, which shows
that the equilibrium for the reaction S02 + %02 = S03 is highly dependent on
both temperature and oxygen content at temperatures below 1,500°C. Significant
amounts of S03 can be formed in this range. Dependence upon oxygen content is
also an important consideration since retort gases contain variable amounts of
fuel which would consume variable amounts of the added combustion air.
One approach to this problem would be to find a catalyst which would
selectively convert all species to S02 even when operated well below 1,500°C.
However, a more straight forward approach, and the one used in this effort, is
to carry out the conversion at temperatures well above 1,500°C where 100%
conversion is most likely. In order to achieve a sufficiently high temper-
ature, a flame was used as a converter in all subsequent experiments.
Laboratory experiments were conducted to determine the conversion effi-
ciency of a methane/air flame using the apparatus illustrated in Figure 6-2.
Artificial standards were introduced through the inlet labeled "sample," and
the response of the S02 monitor was compared for the various species. The S02
monitor was calibrated with a known concentration of S02 introduced without
passage through the flame, so that the absolute as well as the relative conver-
sion efficiencies could be determined. The condenser shown was included
because the combustion gases from the flame contained more water than could be
tolerated by the S02 monitor.
The device shown in Figure 6-2 was entirely enclosed to avoid contamina-
tion of the combustion gases by air prior to analysis. This was achieved by
6-6
-------
100 r
I500°C
tn
-------
UJ
1
Lu,
*o
r >
tr,
Ul
tn
UJ
Q
0
o
X
>"
or
— — -^ <
USK -n
S-.
a>
-t->
CD
>-
o
u
nr
tu-
c
1
10
LU
oc
X<
I
UJ
a: <
a.
£_
o
(U
p
f
fff
O
J3
Of
tt. o
22
(O o
o
6-8
-------
lighting the flame in open air, and then introducing it gradually into the
burner chamber. Although a flame could easily be maintained in open air, it
proved difficult at first to introduce the flame into the closed system without
extinguishing it. A stable enclosed flame was finally achieved by balancing
the flow rates of methane, sample, and air such that the gas velocity at the
burner jet was slightly larger than the flame velocity. This could be achieved
at the nominal flow rates listed in the figure, although it was often helpful
to adjust the methane and air flow rates slightly to achieve a conically-
shaped, blue flame. When operated properly, the flame was almost invisible and
best seen in subdued light. Once the flame was enclosed it 'could still be
extinguished by sudden changes in downstream pressure, resulting from attaching
the receiving flask to the exhaust line. However, models of this device which
have been assembled subsequently for field monitoring exhibited more stable and
more easily ignited flames. This improvement was achieved by using small
diameter (1/8") tubing for the gas supply lines, for which slight pressure
fluctuations are less likely to cause flow disruptions, and by including a
valve downstream from the condenser which could be closed gradually once the
flame was lit and enclosed. With such modifications the flame can normally be
lit externally to the system and enclosed without extinguishing it. It was
observed that small leaks in the feedthrough to the combustion chamber resulted
in less stable flames and more difficulty in enclosing the flame. It is thus
important that the feedthrough be an air-tight seal.
The conversion efficiency for seven different S gases was determined for
concentrations ranging from 283-34,000 ppmv. Both H2 and N2 were used as
diluent gases in preparing the standards to simulate both lean (direct mode)
and rich (indirect mode) retort gases. Table 6-1 shows that the conversion
efficiency is essentially 100% within the precision of the method. The single
6-9
-------
Table 6-1
CONVERSION OF VARIOUS S COMPOUNDS TO S02 IN FLAME CONVERTER
S Species
H2S
H2S
H2S
H2S
H2S
H2S
H2S
H2S
H2S
H2S
H2S
H2S
H2S
H2S
H2S
H2S •
H2S
CH3SH
CH3SSCH3
CH3SSCH3
Thiophene
Thiophene
COS
COS
COS
CS2
CS2
CS2
S02
Carrier Gas
N2
N2
N2
N2
N2
N2
N2
H2
H2
H2
H2
H2
H2
H2
H2
H2
H2
N2
N2
N2
N2
N2
N2
N2
N2
N2
N2
N2
N2
Concentration
of S Species
(ppmv/v)
34,000
15,800
7,000
2,000
1,184
536
283
26,570
12,435
4,174
1,530
1,180
2,640
9,130
26,600
1,820
18,800
4,000
1,960
980
2,220
1,110
3,200
26,000
16,000
7,250
2,900
1,450
1,430
% Conversion
to S02
102.5
96
98
106
98.4
105.6
23
93
98
101
104
. 108
96
101
97
104
95
98-100
101
102
100
101
97
92
100
96
98
95
99
6-20
-------
low conversion efficiency for H2S at 283 ppmv is due to baseline drift in the
S02 monitor. For this run the sample was diluted 70 fold, resulting in an S02
concentration of approximately 4 ppmv, which is near the detection limit of the
S02 monitor used in this experiment. Since the total S content of retort gas
is typically within the range tested in this experiment, these data indicate
that the methane/air flame is an efficient device for non-selectively convert-
ing the various S species in retort gas to S02. While this device could also
be used at concentrations lower than those shown in the table, further testing
should be performed before applying this method to gases containing less than
500 ppmv of total S.
Although data in Table 6-1 do not indicate an effect due to the sample
matrix, a minor effect can be expected, as illustrated by the following calcu-
lation: assuming that essentially 100% of the S species are converted to S02,
the resulting S02 concentration depends on both the flow rate of the sample
into the flame and the total flow rate of gas to the S02 monitor. However, to
a small extent the total flow rate of combusted gas to the monitor, Q , depends
C*
on the components "of the sample gas. For example, one mole of H2 in the sample
reacts with \ mole of 02 to form one mole of H20, which is removed by the
condenser. Thus, introducing 100 ml/min of H2 in the sample causes the loss of
50 ml/min in total flow. On the other hand, CO in the sample causes a net
increase in Q
C ..
CO + hQ2 — > 2 C02
Thus, 4,000 ppmv of H2S in H2 would give a slightly higher response at the S02
monitor than would 4,000 ppmv of H2S in CO. Such differences are minimized
primarily by maintaining a total flow rate approximately 40 times higher than
the sample flow rate. Errors of this type are considered in Appendix 6A for a
range of retort gases. Under worst case conditions (e.g., using a standard of
6-21
-------
H2S in N2 for monitoring a direct mode retort gas) an error of 5% could be
incurred. Thus, for best results a standard should be selected which resembles
somewhat the gas to be analyzed. For example, choosing a N2-based standard for
the analysis of direct-mode retort gases would produce errors of approximately
0.5%. Thus, by using a reasonable standard, and by keeping the auxiliary air,
premix air and methane flow rates constant between sample and standard, the
concentration of S in the sample is given by the expression (Appendix 6A):
(S02)
S-.T.
RG ~ STD 2 RG
Where:
(S)D~ = total concentration of S in the retort gas on a dry basis
Kb
(S)STD = concentration of S in the standard on a dry basis (ppmv)
(S02)RG = concentration of S02 produced by the retort gas during
combustion (ppmv)
(S02)STD = concentration of S02 produced by the standard during
combustion (ppmv)
Since S02 is measured by the pulsed fluorescent analyzer and (S)STD is known,
(S)D~ is readily calculated. The absolute value of (S02) need not be known,
Ku
but only some value in proportion. Thus, the S02 analyzer need not be cali-
brated against an S02 standard, but rather against any standard containing an S
species. This feature of equation I greatly simplifies the calibration proce-
dures. Although equation 1 is generally applicable, the analyst should be
aware of the assumptions and limitations inherent in its derivation, as de-
tailed in Appendix 6A.
Because of the encouraging results obtained in the laboratory, a total S
system was assembled and tested under field conditions courtesy of an oppor-
tunity provided by the Laramie Energy Technology Center. This test was carried
6-12
-------
out in conjunction with the September 8-15, 1980 operation of the 150 ton
retort at the' North Site Test Facility of the LETC. The apparatus assembled
for this purpose was similar to the device shown in Figure 6-2. It was
designed to operate for a short period in the mobile laboratory which was on
site at the time, but was not sufficiently rugged for long term, routine
application. Changes in the device included an improved, positive seal feed
through and the incorporation of the combustion chamber into the end of the
condenser. The primary purpose of this test was to determine the mechanical
and pneumatic integrity of the analyzer, not to provide monitoring data for the
retort.
However, to deal with actual retort gases several more substantive changes
were required in peripheral equipment. The retort gas being tested contained
suspended matter and was saturated with water and possibly oil at approximately
60°C and could not be introduced directly into the total S analyzer without
certain precautions. Similar gases from other industrial sources are cooled to
remove water and oils. However, in the case of retort gas such treatment could
result in the loss of sulfur compounds, and efforts were therefore made to
avoid this treatment. Instead of cooling the gas, all sampling parts which
were exposed to retort gas were heated to approximately 70°G. Suspended matter
was removed by an in-line filter. Since the response of the total S analyzer
is proportional to the sample flow rate, the sample rotameter was replaced with
a mass flow meter to obtain more precise and accurate flow measurements. Since
response of the mass flow meter to retort gas would likely be different than
its response to the standard gas, provisions were included to calibrate the
mass flow meter for each type of gas. .
The entire flow scheme, including facilities for sampling, sample con-
ditioning, combustion, calibration and analysis is shown in Figure 6-3.
6-13
-------
CO
I
UJ
o:
4J
to
CU
-a
'cu
O
CO
01
01
C71
C
-a
o -a
I-H QJ
OJ
Q.
O
QJ
s
13
•i?
6-14
-------
In order of occurrence, the gas exited the process gas line through a port
which was shared with LETC, passed through a glass fiber filter cartridge
(Balston Corp.), and moved to the sampling trailer through 150 feet of Teflon-
lined hose which was heated to 70°C. The flow of retort gas through the sample
line was controlled by valve V2 at approximately two liter minute"1. At this
rate, transit time in the sample line was approximately one minute. The
pressure downstream from the sample pump was maintained at approximately 4 psi
with a back pressure regulator to provide a constant pressure for metering •
valve V4, which regulated the flow of gas to the combustor.
The dry test meter in Figure 6-3, with the drying tube and valve V5, was
included to calibrate the mass flow monitor. Although the mass flow monitor is
relatively insensitive to temperature and pressure, its response is propor-
tional to the specific heat of the gas. In theory, then, the mass flow monitor
could be calibrated either volumetrically or by measuring the specific heat of
the sample gas.
To operate this device the air and fuel flows were first started (with V2
off), the flame was lit and introduced into the chamber and the S02 analyzer
was zeroed. The span gas was introduced next with the flame burning burning,
V4 was adjusted to admit approximately 100 ml/min to the burner, and the span
of the S02 detector adjusted. Leaving the air and fuel rates constant, the
span gas was stopped, V2 was opened enough to admit approximately 2 liter min"1
of retort gas, and V4 was adjusted again, if necessary, to admit approximately
100 ml of retort gas to the flame. The concentration of total S was then
calculated according to equation 1.
During this test the total S analyzer (TSA) was operated from 16:43 on
9/13/80 until the end of the retorting operation at 00:13 on 9/18/80. This
allowed for observation of changes in S concentrations on a real-time basis
6-15
-------
during much of this period and also provided valuable information on the
operation of the analyzer.
Operating parameters were initially set as follows and were left constant
except for minor changes in sample flow to the combustor.
Flow rate in sample line
Premix air flow
Auxiliary air flow
Methane flow
Pressure at back pressure regulator
Pressure at burner (gauge P3)
Mass flow meter reading for standard
Mass flow meter reading for sample
Temperature of sample line
Temperature of mass flow meter,
sensing head, and other components
exposed to retort gas
Span gas, CH3SH in N2
Burner head
2 al min-"1
0.8 al min"1
3.1 al min"1
90 ami min"1
4 psi
0 (atmospheric)
083 std ml/min
125 std ml/min
165°F
150°F
4000 ppm v
quartz w 7 mm ID at tip
All parameters were identical for both sample and standard except sample flow
rates through the mass flow meter, which varied as indicated.
The TSA first operated with span gas and was then switched to retort gas.
Stable operation was observed in both cases. The flame burned well in the
enclosed combustion chamber and produced readings which were stable, reasonable
and well within the working range (1-5.,000 ppmv) of the S02 analyzer. However,
no independent measurements of S species were available for comparison.
The next step was to have been calibration of the mass flow meter with the
dry test meter. However, this operation was not performed during this test
6-16
-------
because of the unexpectly high level of oily materials in the retort gas which
could have resulted in fouling and poor operation of the dry test meter.
Instead, calibration factors were calculated based on the concentrations of the
major gas components (Appendix 6B). In the future a bubble meter could provide
a simpler and more rugged method of calibrating the mass flow meter.
Once measurement of the retort gas began (16:55 on 9/13/80), the analyzer
ran continuously until the end of the burn although the recorder was operated
only intermittently. It was apparent that S levels were essentially constant
during this period. The response time for the complete sampling and analysis
system, including 150' of sampling line, could also be estimated during this
period. From the time the retort gas was first allowed into the sampling line
until the S02 analyzer achieved full response required 4 minutes.
During the evening of 9/13 to 9/14 the most serious malfunction of the
test occurred. The temperature controller for those parts of the system
exposed to retort gas malfunctioned, allowing the temperature to increase to
100°C. This temperature is well above the rating for the mass flow meter, and
readings from this device were erratic thereafter. For this reason concen-
trations reported for periods after 9/13/80 should be considered less reliable.
Total S concentrations were calculated using equation 1, and results
follow:
S (ppmv,
Date Time dry basis)
9/13/80 16:55-18:18 3,700-4,000
9/14/80 07:30-07:51 3,800-3,400
While the monitoring data obtained during this test was meager, the
information gained on instrumental requirements of a practical total S monitor
was substantial. Most failures in the tested system were related to its
6-17
-------
inability to tolerate a realistic retort gas sample. A discussion of the
shortcomings of the system tested during this period and possible solutions
follow:
1. During this test the components exposed to retort gas were traced
with heating tapes to keep them above the dew point of the retort
gas. In view of the amount of condensation which occurred, it would
appear to be necessary to keep all exposed components in a
temperature controlled chamber at 110°C or more, above the boiling
point of condensible materials.
2. To use a mass flow monitor at 110°C, a special high-temperature model
is required. Using a mass flow monitor is questionable since the
sensing element of this device consists of a small diameter tube
through which a portion of the gas must pass. It would seem that
such a sensing element is inherently inappropriate for retort gas,
and that retort gas should be controlled by devices which do not need
small, easily plugged orifices.
3. A better approach would avoid using untreated retort gas either by
developing a method of treating the gas which can be shown not to
cause changes in composition, or by diluting the gas with a pure
carrier to lower its dew point to below room temperature. One
practical approach to the latter would be volumetric diTutors which
are now commercially available but have not been tested under this
program. Such devices operate by adding known, fixed volumes of a
sample gas to a continuously flowing diluent gas stream. Metering of
sample gas by fine control valves, rotameters, or mass flow meters,
all of which are easily fouled by retort gas, would not be required.
6-18
-------
4. The dry test meter shown in the figure for calibrating the mass flow
meter is too easily contaminated and too difficult to clean. A
simpler device, such as a bubble meter, would be more appropriate.
5. The lines for the sample and span gas should be better separated than
in Figure 6-3. This design wastes span gas through the back pressure
regulator and also contaminates the span gas line with retort gas.
The mechanical back pressure regulator did not work entirely satis-
factorily, generating back pressure changes when the flow rate
changed. A column of water would appear to be more reliable and
practical as a back pressure regulator.
6. Condenser #2 in Figure 6-3 should be replaced with a permeable
membrane dryer to avoid the need for a recirculating cooling bath.
7. In spite of the in-line filter and the elevated temperature of the
sampling line, condensed material was found in the sampling line and
filter holder. Future designs should thus include a better means of
protection, such as more efficient filters or traps.
In summary, this preliminary test of the total S analyzer indicated several
mechanical and instrumental problems. In retrospect, these problems appeared
amenable to straight forward solution. More importantly, the basic operating
principles of the system appear sound.
At this point in time it was considered that the principles of a total S
measurement had been established, and further work was not required under this
EPA contract. However, further funding was made available from the Laramie
Energy Technology Center (DOE) to provide total S measurements in support of
their field testing program (Contract No. DE-AS20-81LC0771). This provided an
opportunity to assemble and operate a total S analyzer including some of the
suggested modifications.
6-19
-------
The system designed for this purpose was similar to the original
laboratory model with modifications to improve reliability and portability
(Figure 6-4). Prior to introduction into the instrument, the retort gas was
cooled, dried, and filtered in order to improve ease of handling. The flow of
retort gas and standard was monitored with a rotameter rather than a mass flow
meter, an approach which proved to be considerably more reliable than the mass
flow meter used in the previous experiment. Because the calibration of the
rotameter was expected to be different for the standard and sample gases, an
in-line bubble meter was included for calibration. The calibration of the
rotameter for dried, cooled retort gas from the direct mode processes which
have been monitored has not differed measurably from the calibration of the
standard. However, such would not necessarily be the case for different types
of sample gas. Unlike the laboratory model, this model used a DuPont Model 411
monitor, a uv absorption instrument to detect the S02 produced by the flame.
This instrument has an unexpected advantage in as much as the cell operates at
slightly elevated temperatures (approx. 40°C) so that the condenser need not be
as efficient at removing water vapor as the laboratory model. Unlike the
laboratory model, this model included a cartridge of granular PbC03 which could
be switched into the sample line periodically. The purpose of this cartridge
was to selectively remove H2S to measure "non-H2S" sulfur species in retort
gas. When challenged with several 1,000 ppmv of H2S, this cartridge removed
all measurable amounts. However, its effect on other S species has not yet
been established comprehensively. The pressure switch shown is included to
turn off the methane flow in the event of a failure in the air or power supply.
The main problem with the total S analyzer, as shown in Figure 6-4, is the
need for almost constant operator attention. Because the S02 monitor is being
operated near its detection limit (approx. 5 ppmv), the operator is required to
6-20
-------
Hi!
-------
check the zero and span approximately once per hour, in addition to switching
betv/een the various samples which are being monitored.
In the device shown, the inlet flow rate of sample and standard depends on
the pressure in the sample and standard lines. For best results the operator
must therefore monitor and adjust the flow rate through the sample rotameter.
Pressure fluctuations are naturally occurring in the process line and are
transmitted to the sample line which in turn affects the flow to the analyzer.
This effect has been minimized but not eliminated by using in-line regulating
devices such as mechanical pressure regulators and water columns. Although the
rotameter is more reliable than the mass flow meter, it still must be removed
and cleaned every few days. These problems, while causing an inconvenience,
are easily corrected: the reliability of the gas metering could be improved
through the use of a volumetric dilution system as discussed previously. The
frequent switching between standards and various samples could be performed
easily by an automated sequence of solenoid valves. The drifting of the S02
monitor, which is associated with operation near the detection limit, could be
eliminated by using a more sensitive monitor.
To check the operation of the total S analyzer, total S measurements have
been made simultaneously with the measurement of H2S, COS, CS2, and methyl
mercaptan by gas chromatography at the North Site Test Facility of the Laramie
Energy Technology Center. The sum of these individual gases measurements was
normally slightly lower than the total S measurement. However, total S results
were occasionally up to 10% lower than the summation, indicating a need for
improving the precision of one or both of these techniques. Nevertheless,
total S measurements were clearly achievable with the method discussed and
implementation of the improvements suggested should make this method more
reliable.
6-22
-------
APPENDIX 6A
DERIVATION OF THE CALIBRATION FORM FOR S ANALYZER
Since the combusted gas is dried prior to the S02 analyzer, all results
from the monitor are on a dry basis. In the following discussion, all symbols
for gas concentration will be on a dry basis unless otherwise noted.
The S02 concentration at the analyzer (S02) is given by the ratio of the
flow rate of S compounds into the flame divided by the flow rate of the dry
combusted gas. Consider first the volume of dry gas produced in the absence of
any sample. Because of the combustion of methane,
(1) CH4 + 202 = C02 + 2 H20
the volumetric flow rate of the combusted gas, Q , is given by
C
QC« QA - %2 + Qco = -
2 2
QC =
where Qr = volumetric total flow rate of the combusted gas, dry basis
w - •
Qfl = volumetric total flow rate of air into the combustion chamber
ft
QM = volumetric total flow rate of methane
Qn = volumetric total flow rate of 02
U2
Qrn = volumetric total flow rate of C02
OUg
For the flow rates described in the main text, Qc is therefore approximately
3,900 ml/min. When standard gas is introduced in the form of 4,000 ppmv CH3SH
in N2 at. flow rate QSTD, total flow rate is increased by essentially the same
amount:
<3> QC = QA - QM + QSTD
where STD denotes standard.
6-23
-------
The concentration of S02 in the combusted gas (in ppmv) is then
WS)STD
(4) (S02)
STD
where (S)STD is the S concentration in the standard.
When the sample is introduced in place of the standard, the degree to
which Qp is altered depends somewhat on the composition of the sample. Water
vapor will pass through the flame and be collected in the condenser.
Hydrocarbons and H2 will remove a fraction of the 02 supplied by the combustion
air. N2 and C02 will pass through the flame unaffected. Table 6A-1 summarizes
the reactions which can be expected for typical retort gases, along with the
consumption and production of combusted gas.
For the MIS retort gas shown, the flow rate of dry combusted gas is
therefore:
(5) Qc= QA - QM + (0.965 - 0.130) QRG
where the subscript RG denotes retort gas.
The concentration of S02 in the combusted sample stream is thus:
QRG(S)RG
(6) (S°2>RG = QA + 0.835 QRG - QM
Combining equations 4 and 6 thus gives the concentration of S in the retort
gas.
• (S°2)RG Q + °-835
By adjusting the flow rates QRG and QSTD to approximately the same value, the
last term in equation 7 becomes unity within experimental error, as illus-
trated by considering typical values for the various flow rates:
Q. = 4,000 ml/min; QRG = QSTQ = 100 ml/min; QM .= 100 ml/min. For these rates,
the last term in equation 7 is 0.9959.
6-24
-------
Table 6A-1
PRODUCTION OR CONSUMPTION OF GASES DURING COMBUSTION
Component
%,v/v
Reaction
02*
Consumed
(%,v/v)
Dry Gas*
Produced
(%,v/v)
N2
C02
CO
H2
CH4
C2Hg
C3H6
C4H10
02
H2S
NH3
TOTAL
H2
CH4
CO
H2S
C02
C2H4
C2He
C3H6
CsH8
C4H10
CsHi2
^6^14
C7H16
Cs^is
TOTAL
57.7
32.0
1.0
5.5 '
1.5
0.4
0.3
0.4
0.1
0.3
0.8
100.0
22.5
15.2
3.8
4.3
21.4
5.4
10.4
3.7
4.0
4.2
2.5
1.5
0.8
0.3
100.0
—
—
CO + 1/2 02 = C02
H2 + 1/2 02 = H20
CH4 + 2 02 = C02 + 2 H20
C2H6 + 7/2 02 = 2 C02 + 3 H20
C3H6 + 9/2 02 = 3 C02 + 3 H20
C4H10 + 13/2 02 = 4 C02 + 5 H20
—
H2S + 3/2 02 - S02 + H20
NH3 + 02 = 1/2 N2 + 3/2 H20
DCTODT r*AC — _.._......___
H2 + 1/2 02 = H20
CH4 + 2 02 = C02 + 2 H20
CO + 1/2 02 = C02
H2S + 3/2 02 = S02 + H20
--
C2H4 + 3 02 = 2 C02 + 2 H20
C2H6 + 7/2 02 = 2 C02 + 3 H20
C3H6 + 9/2 02 = 3 C02 + 3 H20
C3H8 + 5 02 = 3 C02 + 4 H20
C4H10 + 1/3 02 = 4 C02 + 5 H20
CsH12 + 8 02 = 5 C02 + 6 H20
C6H14 + 9.5 02 = 6 C02 + 7 H20
C7H16 + 11 02 = 7 C02 + 8 H20
C8H18 + 12.5 02 = 8 C02 + 9 H20
0.00
0.00
0.5
2.8
3.0
1.4
1.4
2.6
—
0.5
0.8
13.0
22.5
30.2
1.9
6.5
0
16.2
36.4
18.5
20.0
27.30
20.00
14.3
8.8
3.8
226.4
57.7
32.3
1.0
0.0
1.5
0.8
0.9
1.6
0.1
0.3
0.4
96.50
•
0
15.2
3.8
4.3
21.4
10.8
20.8
11.1
12.0
16.8
12.5
9.0
5.6
2.4
145.7
volume consumed or produced per 100 ml of dry retort gas
6-25
-------
Thus, for the MIS gas, the concentration of total S on a dry basis is
given, within experimental error, by
(S02)RQ QSTD
(8)
oTn fen \
STD (S02)
2STD RQ
Since only relative concentrations of S02 are important, the S02 analyzer need
not be calibrated separately and a single standard of, for example, CH3SH in N2
is adequate.
Equation 8 remains an accurate simplification of equation 7 as long as the
standard is reasonably similar to the sample gas. For example, for the field
test described in this chapter, CH3SH in N2 proved adequate for the measurement
of MIS retort gas. However, significant error may be introduced by using an
entirely inappropriate standard.
Suppose, for example, that the standard of CH3SH in N2 were used for
measurement of a high-heat-content gas, such as that produced by the TOSCO
process. In this casejsee Table 6A-1) the last term in equation 7 becomes:
- 0.807 QRG - QH
(9,
C)
- "M
assuming that the standard is still contained in N2. Using the same flow rates
as above, R = 0.95, equation 8 could not be used without introducing a 5%
error. If the gas composition is known, R could be calculated. However, a
more reliable approach would be to use a standard which more closely
approximates the sample. For TOSCO retort gas, for example, one could use H2S
in methane, in which case R becomes:
QA - 0.807 QRG - QM
A-QSTD-V .
For the flow rates used above, R then becomes 1.005 and equation 8 can be
used without loss in accuracy.
6-26
-------
In summary, equation 8 can be used without the loss of accuracy as long as
the standard is vaguely similar to the sample. The effect of dissimilar
standards can be calculated from equation 9. For the examples considered, an
error of up to 5% could be introduced by use of an entirely inappropriate
standard. Equation 8 also assumes that Q. > 40 QRG, which corresponds to
approximately a two fold excess of air over what is needed for stoichiometric
combustion of the sample and methane. Equation 8 is valid only if Qft and QM
remain unchanged between sample and standards.
6-27
-------
APPENDIX 6B
CALIBRATION OF THE MASS FLOW METER
The response, r, of the mass flow meter is given by:
(1) P = kCp Q
where k = constant depending on the meter design
C = specific heat at constant pressure (cal g"1)
p = density of gas (g/1)
Q = flow rate under standard conditions (ml/min)
The meter has been calibrated for N2, so that k = (1/C p )N2 where p is the
density of N2 at standard conditions. Thus, for N2
(P0)
The reading of the mass flow meter thus gives the flow rate of the span gas
(CH3SH in N2) in ml min"1 adjusted to standard conditions.
The sensitivity of the mass flow meter for retort gas can be calculated
from a knowledge of its composition. Note that the term C p is the heat
capacity in cal deg"1 liter""1. For a gas mixture under standard conditions,
C p the heat per liter is given by:
CPPQ = lOoV Z Cp V
where V = molar volume at standard conditions = 22.41
C = molar heat capacity (cal deg"1 mole"1)
V = concentration (vol %)
6-28
-------
The concentrations of the major gas components were provided by personnel
from LETC, and the term C is calculated as follows:
Component (vol %)
H2
N2
CO
CH4
C02
H20
other
4.2
46.1
1.8
1.1
20.5
25.0
1.3
6.78
6.94
6.94
8.45
8.75
8.68
11.58
0.289
3.198
0.125
0.093
1.794
2.169
—
7.667
(C pQ) = 0.342 cal deg"1 I"1
The response of the mass flow meter to retort gas is thus given by:
(P) . _ (.342) (o)
RG ~ (CppQ)N2 (P()) G (.310) (oo)RG RG ' (pQ). !
where QAr is the flow rate on a wet basis. Since the flow rate on a dry basis
Ku '
is needed,
In summary, considering only flows at standard conditions, as denoted by
the "o" superscript,
and
where the subscript STD and RG denote the standard and retort gas, and all
flows are on a dry basis.
6-29
-------
SECTION 7
SULFUR SPECIES IN RETORT GAS
Several writers have suggested that atmospheric sulfur emissions could be
the environmental factor which limits the eventual size of the oil shale
industry. Sulfur control technology is thus crucial to the oil shale industry,
and it is advantageous to consider various types of control technologies.
However, proper selection of a sulfur centrol technology requires a knowledge
of the S species which are present in the retort gas. For example, H2S is
normally removed by oxidizing processes, such as the Stretford process, which
are ineffective with other types of sulfur compounds. S02 is most frequently
removed by alkaline agents, which are not necessarily effective for other
components. Reliable methods of analysis for sulfur species in retort gas are
clearly required.
In order to understand the role of sulfur speciation in the development of
sulfur control technology, the overall plant design must be considered.
Present development plans call for combustion of the retort gas to produce
heat. As such, various sulfur species would be emitted as their combustion
product, S02. Thus, one option to sulfur control is the use of flue gas
desulfurization systems, such as alkaline scrubbers. However, it is usually
preferable to remove the sulfur from the lower volume uncombusted fuel gas,
where higher control efficiencies can often be obtained at lower cost. In this
context methods for sulfur speciation are required, not for regulatory
purposes, but for the selection and testing of retort gas desulfurization
systems.
The objective of the efforts described in this chapter is thus to develop
a method or methods appropriate for determining major and minor sulfur species
7-1
-------
in retort gas. These methods must necessarily be sufficiently rugged for use
on site in a field laboratory and must yield results rapidly enough to permit
the operator to optimize plant conditions.
The characteristics of the gas which must be analyzed depends principally
on the retorting process. Two major categories of retort gases are recognized:
(1) low heat-content gases, which are produced by processes which introduce air
directly into the retort (direct processes), and (2) high heat content gases,
which are produced by retorts which exclude air from the retort. The indirect
process gases consist primarily of H2, CH4, C02, CO, and C± to C4 hydrocarbons.
The direct process gases are of similar composition but are diluted with
approximately 60-70% N2 and 20% C02. Both gases are saturated with water vapor
and hydrocarbons and may contain suspended oil and water droplets and
particulate matter. NH3 may be present in amounts which depend on the
retorting process. At the start of this project it was recognized that H2S was
a major sulfur species, although the contribution of other species such as COS,
S02s CS2, and CH3SH was uncertain.
It was clear at the outset of this project that whatever method was
developed for sulfur speciation must be tolerant' of high levels of the
contaminant gases and water vapor. While it must be capable of monitoring H2S
routinely, it must also be flexible enough to monitor the several other sulfur
species which had not yet been determined.
The approach used for this problem was gas chromatography with a flame
photometric detector (GC/FPD). Several columns were evaluated and are
described below in order that the analyst can select the most appropriate
columns as he gains more information regarding the most important sulfur
species.
7-2
-------
GC/FPD has been applied widely and is much discussed in the literature.
Here we mention only the most relevant citations, discussing first the FPD and
then the columns. A more complete survey has been provided by Wallace (1981).
The characteristics of the FPD have been reported extensively in the
literature (Stevens et a!., 1971; Lucero and Palzing, 1974; Bradey and
Chaney, 1966). Salient features include
1. A dynamic range of 1CT10 to 10~7 g/sec
2. A response proportional to S , where n is normally close to 2.0 but
depends on the molecular form of the S
3- Freedom from reagent requirements except for compressed gases
4. Fast response compared to most other S detectors
5. Instability of the flame upon exposure to combustible gases such as
H2, CO, and hydrocarbons, or upon sudden changes in gas flow
6. Selectivity ratios for S compounds (as compared to hydrocarbons) of
up to 104
7. Suppression of the S signal by C02 or hydrocarbons
Several of these features are distinct advantages for field monitoring of
S species. For example compressed gas requirements are normally minimal and
could, in fact, be provided by an air compressor and H2 generator in
environments where compressed gases are not permitted for safety reasons. The
fast response (item 4) means that this detector, unlike other S-selective
detectors can be used with capillary columns. The detectability (item 1) is
more than sufficient for retort gases which typically contain H2S in the range
1,000 to 50,000 ppmv. The dynamic range translates to approximately 500-20,000
ppmv with a 1 ul sample injection volume and a packed column. Present
technology permits direct injection of sample volumes between 0.25 to 3,000 ul,
enabling use of the FPD conveniently for S concentrations in the range 0.2 ppmv
7-3
-------
to 8% v/v. However, for any given sample volume the dynamic range is
approximately 500:1 and a wider range would be more convenient when trying to
cover a range of concentrations with a single instrument.
The dependence of the FPD response on molecular form means that the GC
must be separately calibrated for each gas. Especially in a field laboratory
the preparation and handling of numerous gas standards is often more difficult
than the analyses proper, and this feature of the FPD means that most analyses
are carried out on only a few S species. Varian Associates has attempted to
minimize this inconvenience by placing two flames in series, the first to
combust all forms of S to common fragments and the second to provide the
fluorescent signal. This arrangement was also designed to minimize flame
extinction and hydrocarbon quenching. However, even with the dual flame .
detector, Gangwal et al. (1979), report that the FPD remains measurably
sensitive to molecular form.
The instability of the flame arises from the requirement of operating the
FPD with a fuel rich flame, often verging on extinction. After extinction the
FPD typically must be recalibrated, resulting in the loss of valuable data.
During the initial stages of this investigation it was uncertain whether retort
gases contained sufficient combustible gases to extinguish the flame. This did
not prove to be a problem. However, flame-out did often occur during valve
switching until a flow regulator was installed between the column and detector.
The problem considered in most detail in this study was the possibility of
the S signal being affected by co-eluting H2, C02, or hydrocarbons found in
retort gas. This effect has been reported in the literature, although not
necessarily in terms that permit direct application to retort gas. For
example, Von Lehmden (1978), reported up to 40% suppression of the FPD response
when C02 was present in a thousand fold excess. The effect of C02 in retort
7-4
-------
gas, which may reach levels of 20% v/v, was uncertain. Fluoresence quenching
by hydrocarbons has been discussed by Patterson (1978), Sugiyama et al. (1973),
and Thompson (undated). They observed that major amounts of hydrocarbons could
entirely suppress the signal from a minor S species, (e.g., S in natural gas),
although quantitative data for commercially available detectors was not
available. The dual flame detector manufactured by Varian (Patterson, 1978;
Thompson, undated) apparently minimized the quenching effect, although data was
not available to quantitate its importance in retort gas.
Because some overlap of hydrocarbons and sulfur species inevitably occur,
it v/as necessary to measure quantitatively the quenching effect of co-eluting
hydrocarbons. For example, would 3% ethane in retort gas significantly distort
the measurement of 300 ppmv H2S? Because the dual flame detector offers
apparent advantages while the single flame system is more common, both types of
detectors were investigated and compared.
As with the detector, column material and design are also widely discussed
in the literature. By way of summary, numerous column packings have been
designed, each of which have been applied to a relatively small group
(~ie, 2-4) of S compounds. However, in the absence of previous data the
analysis of retort gas was complicated by the possible existence of over 20
possible compounds and isomers with 4 or less carbon atoms.
Packings described in the literature include silicone rubber (Martin and
Grant, 1965), polyphenylether +H3P04 on teflon (Stevens and O'Keefe, 1970),
acetone-washed Porapak QS, one of the most popular columns (De Souza and
Bhatia, 1976), silicone QFI 6,500 on Porapak QS (Pearson and Hines, 1977),
Tenax GC (Walker, 1978), treated silica gel (Stetter, 1977), and Chromosil 107
(Sklarlew et al., 1981). The limitations of packed columns are typified by the
procedures described by Pearson and Hines (1977), which they recommend for the
7-5
-------
determination of H2S, COS, S02, and CS2 in hydrocarbon gases. For example, the
total analysis requires 4 different columns. An alternative approach with a
wall cooled open tubular column (WCOT) was described by Farwell (1979) which
could separate most compounds with boiling points below 200°C. Disadvantages
of the WCOT column, however, included the loss of S02 (one of the species of
interest in retort gas), questionable separation of COS from large excesses of
H2S, and cryogenic column temperatures, which results in somewhat more
complicated equipment.
Gas chromatography has also recently been applied to oil shale process
gas. For example, Goodfellow and Atwood (1974) analyzed the gas from a Fischer
assay for H2S using a GC technique. They achieved 95% elemental mass closure,
which suggested that major analytical errors did not occur. Skogen (1980) has
employed columns of 5% QF-1 and polyphenylether on Porapak QS for monitoring a
modified in situ retort. While these reports deal with application of GC/FPD
to retort gas, they do not include in-depth analyses of interferences and
suppression effects.
Because of the detector characteristics, it is preferable that the GC
column be capable not only of separating the constituent sulfur species from
each other but also from the hydrocarbons in retort gas. Because of the
numerous possible S species (greater than 20 with less than 4 carbons) which
could conceivably occur in retort gas, and because of the lack of previous
knowledge regarding the species present, several columns were investigated with
regard to their separation ability. These included a acetone-washed Porapak
QS, Chromosil 310, Chromosil 330, Carbopack B HT 100 and Carbopack B/l.5% XE 50
+ 1% H3P04, and an OV-210 treated Porapak QS. An SP-2 100 capillary column was
also considered intitially but was discontinued due to its cryogenic
requirements and its inability to separate H2S and COS.
7-6
-------
In summary, the ideal column for analyzing retort gas must possess these
characteristics:
• Separation of all possible S species (which could in theory number
over 20)
• Separation of S species from C±, -C4 hydrocarbons
• Resistance to water vapor and other components in retort gas
• Reproduci bi1i ty
• Complete an analysis in 30 minutes or less for the rapid testing of
control equipment
A simple protocol was developed using a packed column for H2S, COS, CH3SH,
CH3CH2SH, and CS2. S02 can also be measured depending on the relative size of
the neighboring peaks and operations. While capillary columns thus appear
promising for future work, they were not persued in depth at this time. It was
soon obvious that no packed column could satisfy all criteria, although through
use of a selection of columns, several species can be adequately separated,
albeit in separate runs.
EXPERIMENTAL '
Reagents
Several columns, each six feet long, were packed in 1/8" OD teflon tubing
(Supelco, Bellefonte, PA) and tested. Column packings tested were Porapak QS
(Waters Associates), OV-210 (Pierce Chemical Co. , Rockford, 111.) on Porapak
QS; Carbopack B HT 100 and Carbopack B/l.5% XE 60/1% H3P04, both from Supelco
(Bellefonte, PA); and Chromosil 310 and Chromosil 330, both of which are
treated silica gel packings from Supelco.
Gases were purchased either from Linde Specialty Gases or from Matheson
(Joliet, 111.). The flame photometric detector was operated with "zero" grade
gases from Matheson. Two hydrocarbon mixtures containing Cj to C6 alkanes and
7-7
-------
C2 to C6 n-alkenes were obtained from Scott specialty gases (Plumstead-
ville, PA). Liquid mercaptans and thiophenes were obtained from Eastman
Organic Chemicals (Rochester, NY). Reagent grade acetone was purchased from
VWR Scientific (Denver, CO).
Column Packing Process
Each column was prepared in a six feet length except the acetone-washed
Porapak QS, which was prepared in an 18 inch length. A plug of silanized glass
wool was placed in one end, which was attached to a vacuum source. The column
support was poured into the tubing and compacted with the aid of a vibrator.
All materials were used as received from the manufacturer with the exception of
Porapak QS, which was first acetone washed, and the Chromosils, which were
purchased prepacked.
Notes on the Individual Column Materials Follow—
Porapak QS. Acetone-washed Porapak QS was prepared according to the
•procedure described by De Souza et al. (1975). 5.0 g of Porapak QS was placed
in a Buchner funnel on a' piece of filter paper. The Porapak was then washed
with 50 ml of reagent grade acetone and allowed to stand for five minutes. A
vacuum was then applied for 20 minutes to dry the packing. The column was then
packed and then conditioned overnight at 240°C with a carrier flow rate of
55 ml min"1.
Chromosil 310 and Chromosil 330. These column materials can be obtained
only as pre-packed columns from Supelco. They were selected because of their
ability to elute COS before H2S, a feature which is advantageous when trying to
determine COS in a large excess of H2S, the situation expected in retort gas.
It is also recommended because of its ability to separate higher molecular
weight sulfur species (Supelco Bulletin 722G). This column was conditioned at
70°C under a flow rate of carrier gas (He) of 20 ml/min.
7-8
-------
Porapak QS/OV 210. This column was prepared by coating acetone-washed
Porapak QS, prepared as described above, with 0V 210, a trifluoropropyl methyl
silicone. To 5.0 g of acetone-washed Porapak QS was added 5 ml of 0.1% v/v 0V
210 in reagent grade acetone. The support was coated by swirling the beaker
until the acetone evaporated. If the packing was stirred, it tended to break
apart, so this procedure was avoided. When the support was dry, the column was
packed and then conditioned at 25°C with a carrier flow rate of 40 ml/min for
8 hours.
Carbopack B HT 100 and Carbopack B/l.5% XE 50/1.0% H3P04. These columns
are graphitic solids supplied by Supelco for the separation of sulfur
compounds, especially H2S, S02, COS, and CH3SH. In addition, the acidified
column is optimized for the separation of the isomers dimethyl sulfide and
ethyl mercaptan. Since the properties of these two columns were not
sufficiently different, and since the acidic column is more difficult to
prepare, the latter was not investigated in detail. The Carbopack B HT 100
column was prepared in a manner similar to the other columns. However, because
of its fragility the column was not vibrated during the filling process, but
was agitated gently by tapping the side with a spatula.
Equipment
Two separate gas chromatography systems were used during this
investigation. The first consisted of a Hewlett Packard model 5830A gas
chromatograph equipped with a model 18850A integrating terminal, a thermal
conductivity detector, and a Varian Associates (Sunnyvale, CA) dual flame FPD
which had been specially adapted to the Hewlett Packard gas chromatograph. The
second system consisted of a Baseline Industries (Lyons, CO) model 1030A gas
chromatograph equipped with a Meloy, single flame FPD (CMS, Austin). This
instrument was modified to permit the FPD signal to be recorded by an external
7-9
-------
integrator. This FPD is the mechanical design sold by most vendors in this
country, although many manufacturers supply their own electronics. The valving
and column arrangement of the Baseline System was designed to permit
backflushing two columns to the detector without extinguishing the flame and is
described in the "Protocol" section of this report. The Hewlett Packard system
was configured similarly except that only one column was installed.
Retention times were measured with the Hewlett Packard system using the
thermal conductivity detector. This permitted the simultaneous determination
of the system dead volume which was helpful in presenting the data in a
coherent manner. The quenching effects of the single flame detector were
determined by monitoring the Meloy detector with the Hewlett Packard Model
18850A integrating terminal; the response of the dual flame detector was
determined using the Hewlett Packard gas chromatograph and integrating
terminal. Retort gas samples were analyzed using only the Baseline system in
conjunction with a Hewlett Packard model 3390 peak integrator.
Preparation of Gas Standards
All quantitative standards were prepared by blending continuously flowing
gas streams. .Most such standards were prepared by blending pure gases or gas
mixtures with mass flow meters manufactured by Tylan Corp. (Carson, CA).
Hydrocarbon components of the gas mixtures were introduced with Matheson model
7642T rotameters. Both rotameters and mass flow meters were calibrated for
each gas with dry test meters and bubble meters, both of which were calibrated
in reference to NBS standards. Qualitative mixtures (e.g. for the determina-
tion of retention times) were frequently prepared in static systems such as
teflon bags and glass gas bulbs which had been equipped with teflon-lined
septa. The latter was especially convenient for the preparation of qualitive
standards of the sulfur species which are normally liquid at room temperature
7-20
-------
(such as thiophene). In such cases a few microliters of pure liquid was simply
injected into the bulb with gentle heating if necessary.
Procedure for Measuring Signal Quenching
Two different approaches were used for measuring the quenching of the
fluorescent signal by co-eluting hydrocarbons. In the first approach
calibration curves were obtained for H2S by varying the amount of H2S while
keeping the concentration of ethane at a constant level of 20% v/v. In the
second approach the concentration of H2S was kept constant at a level
corresponding roughly to the middle of the FPD range (~5 ng) while the
concentration of ethane was varied. Identical experiments were conducted for
methyl mercaptan in the presence of varying amounts of n-butane. All
suppression experiments were conducted using an acetone-washed Porapak QS
column, six feet long by 1/8 inch od operated at 70°C. The flow rate was set
at 30 ml/min for all experiments and the injection volume was 100 ul. This
column was selected for the suppression studies because of the naturally
occurring peak overlap of ethane with H2S and methyl mercaptan with butane (see
below). However, to further enhance the overlap of the sulfur gas and
hydrocarbon, measurements were made during the backflush mode of the column.
That is, the sulfur gas/hydrocarbon mixture was injected and allowed to proceed
in the normal forward direction for one minute, at which time the flows were
reversed so that the sulfur species and hydrocarbon would co-elute in two
minutes. Comparison of the retention volumes obtained under backflush
conditions indicated essentially indentical retention times. These experiments
thus indicate the approximate signal suppression that would be expected due to
exactly overlapping peaks which elute in approximately 2 minutes in the forward
direction for the type of columns used in this experiment.
7-21
-------
The single flame detector was operated with a flow rate of 180 ml/min of
air and 140 ml/min of hydrogen. These flow rates were selected by varying the
flows of both gases until a plateau was reached in the sensitivity of the
response. This detector temperature was set to 140°C, although there was no
way of confirming its actual temperature. The single flame detector was
operated in two different modes. In the first mode the hydrogen mixed with the
carrier gas was introduced concentric and interior to the flow of air. This is
how the detector is operated in most cases, and will be referred to as the
"normal" mode in this text. In the other mode, referred to here as the
"reverse" mode, the air was introduced interior and concentric to the flow of
hydrogen.
The dual flame detector requires two air flows and one hydrogen flow to
maintain the two flames. Gas flows were as follows: Hydrogen: 140 ml/min;
air 2 (to the first flame), 80 ml/min; and air 1 (to the fluorescent flame),
170 ml/min. The temperature of this detector was set to 155°C.
Procedure for Measuring Retention Times
The retention times of each gas were found by injecting one ml of
approximately 1,000 ppmv gas and monitoring the response with a thermal
conductivity detector. (Lower concentrations were required for the less
volatile gases such as thiophene.) Column temperatures were varied to permit
elution of the gas of interest in a reasonable time. The carrier gas flow rate
was constant at 30 ml/min.
Procedure for Measuring Actual Retort Gas
Samples of actual retort gas were obtained from the pilot scale oil shale
retorts located at the North Site of the Laramie Energy Technology Center
(DOE). These included the low void volume retort and the 150 ton retort at
this site. During routine monitoring the gas was cooled by passage through a
7-22
-------
condenser kept in an ice bath to remove excessive water. Selected samples,
however, were analyzed from warmer sections of the process without pretreating
or cooling the gas for testing the tolerance of the column to the myriad
materials and water vapor expected in retort gas.
RESULTS AND DISCUSSION
Flame Stability
One of the supposed problems with the FPD is that the flame is easily
extinguished by the elution of combustible gases, such as solvents. This
occurs because the flame operates with approximately 100% excess hydrogen over
that required for stoichiometric combustion and is thus easily extinguished by
additional fuel. An initial major concern was the stability of the flame in
the detector upon the elution of combustible components in retort gas. This
concern proved groundless as the flame once lit, proved entirely stable during
the analysis of retort gas once it was lit. However, other aspects of flame
stability proved of more concern.
Because the composition of retort gas was likely to include a number of
heavier components, including sulfur species, the gas chromatographs were
plumbed to permit the column to be backflushed to the detector. With this
arrangement it was hoped to estimate the sulfur species not analyzed in forward
mode and to preserve the column from degradation during repetitive analyses.
However, it was initially observed that the flame would often be extinguished
when the backflush valve was activated. This would occur because the flow of
carrier gas to the detector would be momentarily disrupted when the valve was
between positions; in addition, when the switching was complete, the previously
high-pressure end of the column would be on the exhaust end, thereby leading to
a sudden burst of gas to the detector. This problem was elimated by installing
the flow regulating arrangement shown in Figure 7-1. In this figure the column
7-13
-------
01—
rtLU
&%'
CO_|'
LULU
ce
v
IT
«
'////////////A
'///////////A
CCLU
23
o5
=>
rcr
.£
a-
o
-P
ny
O
-Q
ta
a>
z
I
r»
UJ
CO
CC
2
I
O)
o
o
CD
-a
c
ny
7-14
-------
exhaust does not go directly to the detector, but first passes through a "T"
and a resistance element, R. Excess pressure at the "T" is relieved by the
back pressure regulator, BPR. In practice the pressure regulator is simply a
column of water contained in an acrylic tube into which is submerged a glass
tube. The "T" is connected to the glass tube and the height of the tube is
adjusted with the carrier gas flowing until the back pressure at the "T"
almost, but not quite, causes the gas to escape against the column of water.
The resistance element consists of approximately 12 inches of 1/16 inch od
stainless steel tubing which is adjusted in length and inside diameter until it
causes the required back pressure. Valve VI in the figure is adjusted to
approximately 2 ml/min and is included to avoid the accumulation of sulfur
species in that portion of the system. In practice, when the backflush valve
is actuated, the column of carrier gas contained in the glass tube first flows
backwards to compensate for the interrupted flow, and then bubbles to vent the
excess gas corresponding to the previously high pressure end of the column.
When operating this system for a period of several days it is necessary to
monitor the height of the carrier gas column in the back pressure regulator,
since the resistance of the element R apparently changes slowly over time, and
carrier gas escaping from the BPR would obviously change the calibration of the
system.
The flow regulating system shown in Figure 7-1 has been used during the
routine analysis of retort gas samples over a period of several days. During
this time the flame was not extinguished during valve actuation, nor during the
elution of the sample components. In this sense the flame can be said to be
stable during the routine analyses of retort gas.
However, the ease with which the flame could be lit varied markedly with
the type of detector and whether the single flame detector was used in the
7-15
-------
normal or reverse mode. The single flame FPD, referred to hereafter as the
SFPD, could normally be lit on the first attempt when used in the normal mode,
although a certain amount of trial and error was required to arrive at a
reliable procedure. This procedure consisted of setting the air and the
carrier gases to their usual levels with no hydrogen flow. The hydrogen was
slowly introduced while the ignition button was depressed, activating a hot
filament in the vicinity of the burner head. As the hydrogen flow was
gradually increased with the ignition on, the flame would eventually ignite, as
evidenced by the appearance of steam in the exhaust from the detector or the
deflection, of the strip chart recorder. The ignition was maintained while
increasing the hydrogen flow until a self-sustaining flame could be obtained,
usually in the range of 50 to 80% of full flow. The hydrogen flow was then
slowly increased to the correct level. Experience showed that the SFPD in the
reverse mode could best be lit in a similar manner, except that the carrier gas
was initially turned off, and the hydrogen flow was brought up more slowly and
cautiously. After ignition the flame is allowed to stabilize for 5 minutes,
after which the He carrier gas is slowly added. In practice the SFPD in the
reverse mode was often very difficult to light, and attempts to do so sometimes
had to be abandoned. Numerous lighting procedures were attempted, although
none proved entirely satisfactory. The authors consider this feature of the
SFPD a major impediment to its use in the reverse mode.
In every case the dual flame FPD, referred to hereafter as the DFPD, was
very easy to light following the manufacturers instructions: both the air and
the carrier gases were set to their final values, and the igniter was activated
as the H2 was introduced. The ease of ignition was a definite attractive
feature of the DFPD. Although the DFPD was designed, in part, to be less
7-26
-------
readily extinguished, it nevertheless required the flow regulation system shown
in Figure 7-1 when used in the backflush-to-detector mode.
For each detector the response was less reproducible when the flame was
first lit, and a warm-up period of 24 hours was required for best results.
This was allowed in every case when quantitative data were required. After
such a warm-up period the response to standards was reproducible to 0.5-1.0%
relative under laboratory conditions.
Fluorescence Quenching
Figures 7-2 and 7-3 show calibration curves for H2S obtained on the DFPD
as a function of peak area and peak height respectively. The injection volume
for each curve is 100 microliters, so that the range of injected H2S varies
from approximately 7-70 ng. Also shown in these figures are the calibration
relationships obtained under identical conditions but with the standard
containing 20% ethane,, corresponding to approximately 2 micrograms per
injection. -
In both figures no effect due to the added ethane can be observed. For
the injection volume employed (100 (jl), it can be concluded that 7-70 ng of H2S
(60-600 ppmv) can be determined without interference in the presence of 20 (jg
(20%) of co-eluting ethane. Since the peak shape is not affected by the
injection volume, the same statement applies to the determination of
600-6,000 ppmv of H2S in pure ethane with a 10 (jl sample volume, or to the
*
determination of 6-60 ppmv of H2S in the presence of 2% ethane with a 1 ul
injection volume.
Based on the presumption that other gases behave similarly to ethane and
H2S, similar statements can be made for other sulfur species coeluting with
other hydrocarbons. Most retort gases contain much less than 20% v/v ethane,
and this concentration was chosen as an extreme example. In fact, 2% v/v would
7-17
-------
10
20% ETHANE
NO ETHANE
LU
10"
Ui
Q.
10
-I—I I » I 11
-I—I—II I til
10
ioo 1000
H2S CONCENTRATION (ppmv)
FIGURE 7-2
H2S calibration curves obtained with the Dual Flame
Photometric Detector as a function of peak area
10,000
7-18
-------
10
10°
LU
10*
10
O H2S WITH 20° ETHANE
X H2S ONLY
1.0
J L
t i I
-i—i—i i i i i 1
10
100 1000
H2S CONCENTRATION (ppmv)
FIGURE 7-3
oS calibration curves obtained with the Dual Flame
Photometric Detector as a function of peak height
lopoo
7-19
-------
be a more likely upper limit, indicating that the DFPD should be essentially
free of quenching effects during the analysis of retort gas.
Figures 7-4 and 7-5 show calibration curves for H2S obtained on the SFPD
operated in the reverse mode as a function of peak area and peak height
respectively. The injection volume for each curve was 100 microliters, as for
the DFPD. Also shown in these figures are the calibration relationships
obtained under identical conditions but with the standard containing 20%
ethane, corresponding to approximately 20 micrograms of ethane per injection.
Although not clear in this figure, other calibrations of the SFPD have shown
its dynamic range, operated under these conditions, to be approximately
1-100 micrograms per injection, somewhat lower than the DFPD.
The effect of the added ethane is clearly visible in either figure, as is
the observation that the suppression generally increases with decreasing sulfur
content. Referring to Figure 7-4, it is apparent that for a peak area count of
10,000 in the presence of ethane the H2S concentration would be underestimated
by 75% under the chromatographic conditions employed for this experiment.
Measurements obtained by peak height would be in error by similar amounts.
However, in the peak height mode both signal suppression and enhancement occur
in the presence of ethane, a result which, to the authors' knowledge, has not
been previously reported in the literature.
Twenty percent ethane exceeds the amount expected in retort gas and such
extreme errors would not normally be expected when analyzing actual samples.
However, the SFPD is clearly more prone to signal supression by co-eluting
hydrocarbons than is the DFPD. Thus, while the DFPD could be used with confi-
dence for essentially any retort gas sample, the SFPD must be applied more
cautiously.
7-20
-------
IxJ
•
O
o
o
o
o
o
cf
o
ei~'
£-
r(->
O)
§
•!->
O
Oi
OJ
(O
OJi UJ
•I--OJ
t^ Q..
O
ns
(O
(/>' O
§•
O
O
S-
-------
UJ
z:
LU
o o
CM Z
O X
.05
J.r~
•cu
to
O)
Q-
£=
O
O
c:
re
s-
o
O
-S
-------
Following the same reasoning as applied to the DFPD, one would expect
quenching effects similar to those shown in Figures 7-4 and 7-5 if the sample
gas contained 2% ethane with a 1 ml sample injection volume, corresponding to a
sulfur range of 1-100 ppmv. Slightly less quenching would be expected with a
10 micro!iter sample injection (100-10,000 ppmv) if the sample consisted of
100% ethane. These results are applicable only to the chromatographic
conditions applied in this experiment, consisting of two exactly overlapping
peaks eluting from a packed column in approximately 2 minutes. Since the FPD
is a flow sensitive detector, similar quenching would be expected under similar
flow rates of hydrocarbon and S into the detector. For the experimental
conditions discussed, the mass flow rates can be estimated from the peak shape
and elution volume. (Based on response of the thermal conductivity detector in
this experiment the full peak width at half height was 0.2 min at a flow rate
of 30 ml/min.)
Figure 7-5 suggests that the effect of hydrocarbons on the S emission is
more complex than a uniform supression of signal, and this is supported by
examination of the peak shapes which occur in the presence and absence of
hydrocarbons. In the absence of hydrocarbon the H2S peak is symmetric and
normally shaped. Upon the addition of 3% ethane the peak sometimes became
slightly higher and often three times narrower. Upon the addition of 60-100%
ethane the center of the peak was suppressed more than the edge causing the
original S peak to divide into two separate peaks.
The second set of experiments measured signal suppression by injecting a
constant mass of S while varying the amount of co-eluting hydrocarbon.
Figures 7-6 and 7-7 were obtained with a SFPD operated in the normal and
reverse mode, respectively and show the effect of coeluting ethane on the H2S
peak area and height. The indicated points, if taken in the direction of
7-23
-------
1H9I3H MV3d 3AI1V13M
O
O
O
CO
O
evi
O
O
O
OL
Q)
03.
CJ
ED
E
in 0}
-a
01 o
i= s
-t->
S-'cO
O) C
£T
njd)
IO -i—O
I- -t-> O>
Hj LU O) i.
O °^ I OJ
z r> o-cx.
o J5 °°
z p_-j
O)
CDOJ
•ir-'Q
i/r
o
t/>'i—
CMS-
rc-u
OJ
cu E
O
-------
o
r
O
O
J.H9I3H W3d 3AI1V13N
o
CO
§
o
OJ
o
o
to
CVJ
UJ
CO
ae
ui
LU
oe
o
ci
o
o
o
O)
> S-
-Q O
4->
•— O
as O)
£= -M
cn a>
•r--Q
to
o
e/> -i—
OJS-
ac +J
O)
o
o a.
O5
•r-'
.£=
O
CD
CD
«\J
>IV3d 3AI1V13M
7-25
-------
increasing concentration of added ethane, correspond to 0, 0.1, 0.3, 1.0, 3.0,
10.0, 30.0, and 100.0% added ethane when a 100 microliter injection volume is
employed.
The plot of peak area (Figure 7-6) shows a monotonic decrease with added
ethane, the so-called fluourescence quenching effect. The remaining three
curves are more complex, showing both attenuation and enhancement of the
signal. It is difficult to predict from the data in these figures the extent
to which fluourescence quenching would be a problem. Of the curves shown, the
plot of peak area for the SFPD operated in the normal mode resulted in the
largest errors, while the plot of peak height for the SFPD operated in the
reverse mode resulted in the smallest errors. For example, for the former plot
(peak area, normal mode) an error in the signal of 10% occurs at 1% added
ethane, while the latter plot (peak height, reverse mode) does not deviate 10%
over the entire range of 1-100% ethane. These data were obtained with a
100 micro!iter injection volume, and corresponding less quenching would be
expected as the injection volume is decreased. Thus, if H2S were to be
measured at a concentration of 5,000 ppmv with a 1 microliter injection volume,
the signal quenching would be less than 10% even in pure ethane. Response of
the FPD varies with the square of the S concentration, so that a 10% signal
suppression leads to a 5% error in concentration.
Experiments identical to those described in the previous paragraphs were
performed for methyl mercaptan co-eluting in the presence of n-butane. These
data are shown in Figures 7-8 and 7-9 for the SFPD in the normal and reverse
mode respectively. These figures illustrate that signal suppression is more
pronounced than for the experiments with H2S. For example, for the plot of
peak area in the normal mode the signal is attenuated 22% upon the addition of
only 0.1% butane. For the plot of peak height obtained from the normal mode,
7-26
-------
1H9I3H >!V3d 3AllV13y
O)
4->
S_ O
, g
O)
r— Q.
rO O
E
O> S-
•r- O
CO 4->
O
S= 01
CO 4-5
4-> O)
O.O
CO
O O
01 sL
£ 4->
>. o
-C •)->
4J O
a> ^:
E Q-
a> ai
o
a>
en,—
c a>
•r- c:
O
-------
1H9I3H MV3d 3AIJLV13M
IV3d 3AI1VT3H
7-28
-------
the addition of 0.1% butane results in an attenuation of 15%. When the
detector is operated in the reverse mode, the same addition of butane results
in a signal enhancement of 5% in terms of peak area and of 4% in terms of peak
height. When the detector is. operated in the reverse mode, the peak area is
within 10% of the correct value until 0.6% butane is added, but then declines
to 90% of the original value upon the addition of 30% butane. Examining the
plot of peak height for the reverse mode indicates that the signal is within 6%
of the correct value until 20% butane is added. While peak attenuation is
observed in all figures, it is apparent that hydrocarbon interference is
minimized by operating the SFPD in reverse mode and by using peak height as the
calibration parameter.
In relating this data to the analysis of retort gas, recall that an error
of 10% in the FPD signal generates a 5% error in concentration. According to
i
these data, then, 95% accuracy could be achieved in the presence of 0.6% butane
by measuring peak area in the reverse mode, or in the presence of 20% butane by
measuring peak height. These results apply to a 100 microliter sample
injection, which corresponds to a concentration range of approximately
10-1,000 ppmv of S, and should be adjusted accordingly for different concen-
tration ranges. Thus, assuming the range of interest for S compounds in retort
gas is above 10 ppmv, the data suggest that determinations can be obtained with
a 95% accuracy by using the SFPD in the reverse mode and calibrating against
peak height.
Identical experiments were carried out with the DFPD, and .these results
are shown in Figure 7-10 for H2S in the presence of ethane and in Figure 7-11
for methyl mercaptan in the presence of n-butane. In the case of H2S/ethane
attenuation is not observed except at the highest concentrations. The same is
true of the methyl mercaptan/butane results. The DFPD all but eliminates the
7-29
-------
JLH9I3H XV3d 3AI1V13H
O
1
cvl
O
in
CM
Q."
O.
O
O
-»
UJ
OC
p
O
O
O
O
O)
OJ
10
4J S-
O) O
ii cno
o
CD
3 Q
a> o
o •»->
O)
>» E
^j C3
r— O
ta jz
Sw f^ u
Ol
•r- O)
w S
to
«/> 1—
CM4-
O) ra
u
E
(1)
—'o
o
>»V3d 3AI1V13H
-------
CO
o
1H9I3H MV3d
co
O
O
1
tn
o
Q.
LU
CC
^
^>
g
O
O
O
O
C3
I
OJ
E
as
•3
cn
=s s-
i— O
o> -i-»
i o
O d)
O -M
CD
2 ^ -5
U
tT
o
LU
m
LU (/) O
=3 CO
CEJ (O .^
4-> Q.
Q-
fC O)
.0 E
S- (O
0) r—
E U-
£§
•<-> o
O)
-------
interference effect except possibly at concentrations of added hydrocarbons
which are greater than would be encountered in retort gas.
Each of these detectors has its limitations. The DFPD appeared to be more
sensitive to variations in flow rate of the carrier gas and is more prone to
develop light leaks which have the effect of degrading the dynamic range to a
point of marginal utility. This requires occasional replacement of light
seals. However, the SFPD is more prone to quenching, an effect which is
minimized by calibrating with peak height and operating in the reverse mode.
However, in the reverse mode the SFPD flame is difficult (sometimes impossible)
to ignite.
The mechanism(s) of signal quenching has not been addressed in this
report. However, several types of mechanisms have been proposed including
(1) interference in the formation mechanism of the activated S species;
(2) reaction with the activated S species; (3) absorption of the emitted
radiation by hydrocarbon decomposition products; and (4) a change of the
position of the H-rich emission region in the flame. Further information is
present in the literature (Gilbert, 1970; Patterson et al., 1978).
Uniformity of Response
In theory one could quantitate the various S species by calibrating the
FPD separately for each species. In practice, it becomes .prohibitive to be
prepared to calibrate for every S species, especially in a field laboratory
with limited space and facilities. It would obviously be preferable to
calibrate with only a few species and then infer from their responses the
calibration of the remaining gases.
For this reason the relative responses of four S species—H2S, S02, CS2,
and thiophene, representative of different compound classes, were determined.
Standards of these gases were injected in one microliter quantities onto a
7-32
-------
Chromosil 310 column at 70°C; detection was by SFPD operated in the normal
mode. In this experiment the current from the detector was recorded on a strip
chart recorder and entered in digital form with a model 9874A digitizer into a
model 9845B desk top microcomputer (Hewlett Packard). The linearized peak
area, A1, was then calculated according to the expression
(1) A'. = Ji^dt'
where i = detector current
t = time
k = arbitrary constant
To the extent that the FPD responds to the mass flow of S independent of the
molecular form, the quantity A1 should be independent of the species which are
injected, as well as elution time and peak shape.
Table 7-1 lists the relative responses obtained for these four compounds
in terms of linearized peak area per ng of S. Relative responses of these
compounds, which represent a disparate group, vary from 1.1 to 5.6. This
suggests that relative responses of various S species would fall within a
factor of three of an average value, although such a conclusion would require
additional testing. Based on the data presented in the previous section, the
DFPD or the SFPD operated in the reverse mode would presumably demonstrate less
variation in relative responses. However, time and budget did not permit
further investigation of this possiblity, and until such testing is performed,
it is evident that the FPD of either variety must be calibrated with the gas of
interest to obtain quantitative results.
Column Characterization
Retention data for the columns investigated are grouped as follows: the S
gases of most interest—H2S, COS, CS2 methyl mercaptan, and S02—are grouped
together and are referred to as the primary S gases. The S gases of less
interest—dimethyl sulfide, dimethyl disulfide, ethyl mercaptan•, thiophene,
7-33
-------
Table 7-1
RELATIVE RESPONSE OF THE FPD TO FOUR SULFUR GASES
WITH THE SINGLE-FLAME FPD OPERATED IN THE NORMAL MODE
Gas
H2S
S02
CS2
Thiophene
Concentration
(ppmv)
4,927
15,000
5,010
4,544
A1 = Ji^dt
(arbitrary units)
12.5
15.69
53.6
15.89
A '/mass of S
(ng-1)
2.68
1.10
5.67
2.83
7-34
-------
2-methyl thiophene, 3-methyl thiophene, and t-butyl mercaptan—are also grouped
together and will be referred to as the secondary S compounds. The
hydrocarbons examined include methane, ethane, ethene, propane, propene, and
n-butane; data for these species are also grouped together. Retention time
data are also included occasionally for water and ammonia. For each column the
adjusted retention time, tr, which is defined as the absolute retention time
less the dead volume elution time, is plotted as a function of column
temperature. Such plots are normally straight lines over the temperature range
of interest for this study, which simplifies data collection and presentation.
To some extent the division of the S species into primary and secondary
groups reflects the philosophy behind this effort, which was to develop a
simple analytical protocol for the most important S species in retort gas. At
the beginning of this project, however, it was not clear which compounds should
be considered the most important, although it has long been considered that H2S
is the predominant S species in retort gas. Preliminary reports have also
/
noticed the presence of CS^, methyl mercaptan, and COS in retort gas and other
synthetic fuel gases (Sklarew et al., 1981), and these compounds were therefore
included in the primary group. Although S02 has not been reported in retort
gas to the authors' knowledge, it was nevertheless of interest due to its
common occurrence in combustion gases. The presence of S02 has also been
proposed as a precurser to various S species of intermediate oxidation state
which are found in the condensate from retort gas (Stuber et al., 1978). The
objective of this work thus became development of a procedure for determining
the primary S species in retort gas in a manner which was free from
interferences by other S species, that is, free from interferences by the
secondary S species and also free from significant quenching by co-eluting
hydrocarbons. In final analysis, this division into primary and secondary
7-35
-------
groups is somewhat arbitrary, since some secondary S species may eventually be
found at higher concentrations than primary S species.
Retention times for the Porapak QS column are shown in Figures 7-12, 7-13
and 7-14 for the primary S species, secondary S species, and hydrocarbons,
respectively. Although an adequate separation of the primary S species can be
obtained, for example at 90°C, all secondary species elute so closely as to be
essentially inseparable. The secondary species elute on top of the primary
species COS, S02, and methyl mercaptan, making this column useless for the
analysis of retort gas except for H2S and possibly CS2. Indeed, the same
comment would apply for any type of sample gas except those known not to
contain any of the secondary S species. Ethane and ethene both elute
essentially simultaneously with H2S, precluding the use of this column for the
determination of H2S in gas streams containing high levels of either of these
species. More than one batch of acetone-washed Porapak QS was prepared during
this experiment, and it was observed that the retention times varied widely for
this column. In addition, although S02 tailing was not a problem with freshly
prepared columns, it tailed for several minutes on columns which had been used
for longer periods. On the other hand, one major advantage of this column,
which consists of a styrene-divinyl benzene polymer, is its resistance to water
vapor, which is often one of the major components of retort gas. Thus, while
this column packing appears to be of little use for other species, it may be
worthwhile for H2S, as long as the levels of ethane and ethene are sufficiently
low. This column was also evaluated during a run of several days using actual
retort gas, as discussed below.
Figures 7-15 to 7-17 illustrate the retention times that were obtained
with 1% 0V 210 on Porapak QS. Among the primary group, H2S, CS2, and methyl
mercaptan are well separated, although COS and S02 are not. The former
7-36
-------
lOO.Or
ACETONE-WASHED
PORAPAK QS
30
40
50
60 70 80
TEMPERATURE (°C)
FIGURE 7-12
90
100
Retention times for the primary sulfur compounds with acetone-washed Porapak QS
7-3 7
-------
100.0 r
10.0-
1=
UJ
•z.
o
LiJ
O
Ul
co
:=>
o
ACETONE-WASHED
PORAPAK QS
2-Me + thiophene
0.1
30
40
50
80
90
60 70
TEMPERATURE (°C)
FIGURE 7-13
Retention times for the secondary sulfur compounds
with acetone-washed Porapak QS
7-38
-------
lO.O
1.0
o
LU
UJ
OC
a
to
13
O
0.1
.0
ACETONE-WASHED
PORAPAK QS
Ethene
.Methane
40" 50 60 70 80 90
TEMPERATURE (°C)
FIGURE 7-14
Retention times for hydrocarbons with the
acetone-washed Porapak QS
100
7-39
-------
lOO.Or
1% OV-210 on
PARAPAK OS
40
50
60
90
100
70 80
TEMPERATURE (°C)
FIGURE 7-15
Retention times for the primary sulfur compounds
with OV-210 on Porapak QS
110
7-40
-------
100.0
1% OV-210 on
PARAPAK QS
10.0
LU
LU
CC
V)
GH36H
1.0
S02
0.1
_L
J_
100 110 120 130 140
TEMPERATURE(°C)
150
160
170
FIGURE 7-15(a)
7-41
-------
100.0
1% OV-210 on
PARAPAK QS
Thiopene
Me9S.
10.0
e
UJ
p
o
t—
0
LU
Me2S2
EtSH
1.0
0.1
110 120 130 140 150
TEMPERATURE (°C)
FIGURE 7-16
160
170
180
Retention times for the secondary sulfur compounds
with OV-210 on acetone-washed Porapak QS
7-42 '
-------
100.0r
1% OV-210 on
PARAPAK QS
10.0-
Ul
Ul
or
40
50
60
70 80
TEMPERATURE(°C)
FIGURE 7-17
90
100
no
Retention times for hydrocarbons with OV-210 on
acetone-washed Porapak QS
7-43 ,
-------
compounds are also well separated from the secondary S compounds, indicating
that this column may be useful for H2S, COS, and methyl mercaptan. Preliminary
examination of Figure 7-16 suggests that this column may have some utility for
the separation of the secondary group. For example, at 165°C the compounds
thiophene, n-butyl mercaptan, and dimethyl disulfide are separated. • However,
the isomers ethyl mercaptan and dimethyl sulfide are not separated, and it is
likely that the similar isomer pairs for propyl and butyl mercaptan would
similarly co-elute. Further consideration indicates that there are several
additional possible S compounds which should elute between ethyl mercaptan and
dimethyl disulfide in the figure. These include n-propyl mercaptan, isopropyl
mercaptan, the iso- and tert-butyl mercaptans, and the various methyl
thiophenes. It thus appears that in order to apply this column to the
determination of the secondary S species, one would have to know beforehand
which species to expect.
Comparison of Figures 7-15 and 7-17 indicates that H2S is well separated
from the C2 and C3 hydrocarbons. This suggests that for H2S in high levels of
C2 or C3 hydrocarbons, the OV-210/Porapak QS column is preferred over the
untreated Porapak QS column.
While this column might be useful for the determination of H2S, it was a
difficult column to prepare and stabilize, repeated large injections of H2S
being required for this purpose. Another undesirable feature of this column
was that the retention times were too long to permit routine, repetitive
analyses every 30 minutes, which was one of the criteria of this analysis. For
these reasons as well as its inability to separate COS and S02, this column was
not tested extensively with retort gas.
Figures 7-18 to 7-20 illustrate the retention times obtained with the
Chromosil 310 column. Dimethyl disulfide and t-butyl mercaptan are not shown
7-44
-------
100.0r
CHROMOSIL 310
10.0
e
Ul
UJ
OC
Q
UJ
CO
=3
1.0
CH3SH
COS
0.1
30
40
50 60 70 80
TEMPERATURE (°C )
FIGURE 7-18
90
100
Retention times for the primary sulfur compounds with
Chromesil 310
7-45
-------
100.0
10.0
tu
3
I-
o
UJ
as
UJ
O
** 1.0
0.1
30
40
TBuSH
CHROMOSIL 310
thiophene
50 60 70
TEMPERATURE (°C)
FIGURE 7-19
80
90
100
Retention times for the secondary sulfur compounds
with Chromosil 310
7-46
-------
100.0
CHROMOSIL 310
Hexane
0.1
30
40
50
60 70
TEMPERATURE (°C)
80
FIGURE 7-20
Retention times for hydrocarbons with Chromosil 310
-------
because they eluted too late, over an hour at the temperatures shown.
Figure 7-18 illustrates that most primary S compounds were separated with the
possible exception of COS and H2S. Comparing Figures 7-18 and 7-19 indicates
that primary compounds are separated from the secondary group with the possible
exception of methyl mercaptan and thiophene. However, comparing Figures 7-18
and 7-20 indicates that H2S falls between propane and propene, COS co-elutes
with propane and ammonia, and S02 co-elutes with pentane.
As with the previous column it is not clear whether this column would be
useful for the secondary group of compounds until additional possible compounds
are considered. In addition, hydrocarbon interferences demand consideration
before application of this column to the primary group analysis. However, on
this column COS elutes before H2S, a feature potentially helpful when measuring
small amounts of COS in large amounts of H2S. For this reason, this column was
injected with actual retort gas to evaluate its usefulness for COS. However,
with the ratios of COS to H2S found in retort gas, the two peaks could not be
resolved.
Besides its inability to achieve the desired separation, this column also
had other serious drawbacks. Its temperature limit (less than 70°C), can only
be exceeded occassionally. Another problem is its sensitivity to water, which
is abundant in retort gas. (The support for this packing is a silica gel.)
The next two columns to be evaluated were the Carbopack B HT 100 and the
Carbopack B/1% XE 1.5% H3P04. Of .these two columns the latter packing is mbre
difficult to prepare and, according to preliminary experiments, did not differ
enough from the former column to warrant a separate investigation. This
section therefore reports data only for the Carbopack B HT 100 column.
Retention time data for this column are shown in Figures 7-21 to 7-23. Of
the columns examined thus far, this column gave the best separation of primary
7-48
-------
100.0r
10.0
E
UJ
LU
Ut
a:
=5
—3
o
•* 1.0
0.1
CARBOPACK BHT 100
J_
_L
30
40
50 60 70
TEMPERATURE (°C)
FIGURE 7-21
80
CS,
90
100
Retention times for the primary sulfur compounds
with Carbopack BHT 100
7-49
-------
100.0
llhiophenes
CARBOPACK BHT 100
10.0
ee
o
LU
CO
=3
a
1.0
O.T
'40 50 60 70 80
TEMPERATURE (°C)
FIGURE 7-22
90
100
110
Retention times for the secondary sulfur compounds
with Carbopack BHT 100
7-50
-------
100.0r
CARBOPACK BHT100
10.0
E
LU
O
H-
UJ
1—
ut
ce
Butane
1.0
0.1
Ethane
30
40
50
80
90
60 70
TEMPERATURE (°C)
FIGURE 7-23
Retention times for hydrocarbons with Carbopack BHT 100
100
7-51
-------
compounds. The wide separation of COS and H2S is especially encouraging in
light of the large excess of H2S which is found in retort gas. H2S essentially
co-elutes with the C2 hydrocarbons, although this is not necessarily a problem
in retort gas which contains high levels of H2S in the presence of small levels
of ethane and ethene. S02 also elutes very close to the C3 hydrocarbons and in
the event that S02 is found in retort gas, it would be necessary to consider
possible quenching effects on this species. Thiophenes, and likely other
aromatic S species, do not elute from this column in less than one hour, and
would therefore require a different column. The determination of CS2 should
also be approached cautiously with this column because of the proximity of
dimethyl sulfide.
This column has also demonstrated other desirable characteristics, notably
ease of packing, a wide temperature range, and tolerance to water. Water
tolerance was examined by injecting 10 mg of liquid water onto the column. The
column was "reconditioned" by temporarily raising the column temperature above
the boiling point of water. When the temperature was again decreased to normal
operating range, the retention times' of COS and H2S were observed to have
changed less than 0.3%, an insignificant amount. Considering that this amount
of water would not normally be injected during the analysis of 1,000 retort gas
samples, it was concluded that this column possessed more than sufficient
tolerance towards water. Of the columns examined, the Carbopack B HT 100
column is clearly the best suited for the determination of the primary S
species in retort gas.
Application to Retort Gas
The final test of any analytical method occurs when it is applied to an
actual sample. Various columns and protocols were tested during operations of
7-52
-------
oil shale retorts at the Laramie Energy Technology Center. Experiments at that
site included:
1. A test of the acetone-washed Porapak QS column and protocol for the
monitoring of H2S. The objective of this effort was to determine the
column longevity and stability, and to determine sampling problems.
2. Qualitative test of the remaining columns. The objective during this
period was to determine which compounds could be observed with the
columns of interest.
3. Quantitative monitoring of primary S species.
4. Quantitative monitoring of primary S species in support of the
evaluation of a pi lot-scale venturi scrubber during the May, 1982
operation of the 150 ton retort.
These tests were supported by the DOE (Contract Numbers DE-AS20-81LC10771 and
DE-AS20-81LC10845), but are mentioned in this report because of their relevance
to the subject at hand (Wallace and Sexton, 1982). During the first test, it
was observed that H2S could be monitored routinely with the Porapak QS column.
Retention times were stable, the flame in the FPD remained lit, and standards
run during the period of the experiment varied less than 10% relative once the
system was warmed up. These measurements were conducted with the valving and
column arrangement shown in Figure 7-1 and described in more detail in the
Protocol section of this chapter. Of particular importance was the observation
that the column did not degrade even when exposed to uncooled (32°C) retort gas
saturated with oils and water at that temperature (Wallace, 1982). The main
problem encountered during this period was with the sampling system which
permitted fluctuations in the process line pressure to affect the pressure and
flow of sample at the chromatograph.
7-53
-------
The results of the qualitative tests have been discussed in the column
evaluation section of the preceeding text. In summary only the Carbopack B HT
100 column was capable of separating the primary S compounds. For most columns
the separation of COS from a large excess of H2S was especially difficult.
The third experiment was conducted at the Low Void Volume Retort at the
Laramie North Site facility. During this experiment the retort gas was
analyzed simultaneously for S species by a GC/MS and for total S with the total
S analyzer described in Chapter 6 of this report. Major species and
hydrocarbons were also measured by a gas chromatograph, data which permitted
the estimation of quenching effects whenever required. The primary S species
were determined using the protocol described in the next section of this report
with a Carbopack B HT 100 column. The objective of this test was to compare
the results obtained by the different methods to determine the efficacy of
each.
The objective of the test carried out during the operation of the 150 ton
retort during May 1982 was to provide monitoring data to determine the
efficiency of a venturi scrubber for the removal of H2S and other primary S
species. This provided an opportunity to observe the performance of the
sampling and analytical system described below. During this test, the primary
sulfur species were monitored using the procedures described in the Protocol
section of this chapter.
Figure 7-24 shows a chromatogram of retort gas obtained during this test,
and Table 7-2 describes the detailed chromatographic conditions and integrator
routines. Two injections were made to keep the compounds of interest within
the dynamic range of the detector. The first was of 1 microliter for H2S
10 seconds before the integrator was started, and the second was of 100 micro-
liters 7:00 minutes later. (Several changes in attentuation and chart speed
7-54
-------
o
ro
O
^
d
256
ATTENUATION
V)
FIGURE 7-24 •
Chromatogram of retort gas sample: 6' x
Carbopack BHT 100 at 65°C isothermal
1/8"
7-55
-------
Table 7-2
INTEGRATION AND VALVE SEQUENCING PROGRAMS FOR FIGURE 7-24
INTEGRATOR (HEWLETT PACKARD MODEL 3390)
Time (min.) Function
0.00 Attenuation = 8
Chart speed =1.5
Peak width = 0.16 min.
Threshold = -1
Area reject = 500
0.10 Intg. #5 (extend baseline horizontally)
0.10 Att 2t2 (attenuation = 4)
1.20 Chart speed
5.00 Disable horizontal baseline extension
6.75 Attenuation = 256
Chart speed = 1.5 cm/min
7.39 Intg. #4 (disable automatic testing for solvent peak)
7.87 Attenuation =16
8.75 Chart speed =0.5 cm/min
Attenuation =0
9.10 Intg. #2 set baseline at all valley points
12.40 Disable previous instruction
14.00 Chart speed =0.3 cm/min.
7-56
-------
Table 7-2 (cont.)
PROGRAM ON CHROMATOGRAPH (BASELINE MODEL 1030)
Item #
01
02
02
03
04
05
06
07
08
09
99
Time (min:
00:01
00:10
00:11
01:00
07:00
12:00
18:00
18: 01
29:45
00:30
sec) Code
13
23
24
. 14
03
04
23
24
99
00
CHROMATOGRAPHIC CONDITIONS
Operation
Inject 1 (jl sample onto
column 1
Start integrator at
11 seconds
Backflush column 1
Inject 100 nl sample
onto column 2
Backflush column 2
Stop integrator
Repeat every 30:00
minutes
Columns: Carbopack B HT 100, 6' x 1/8" OD Teflon
2 each
Carrier flow rate: 25 ml/min in each column, He
Column temperature: 65°C
H2 flow to detector: 140 ml/min
Air flow to detector: 180 ml/min
Detector: Flame photometric operated in reverse mode
7-57
-------
occur to preserve the visual appearance.) H2S, COS, methyl mercaptan, and CS2
are plainly visible in the figure and occur well above the detection limit.
Identification of these compounds was confirmed on site by injection of
standards which yielded identical retention times. Their identities are also
confirmed by the data in Figures 7-21 and 7-22 which indicate an absence of
interfering compounds. Were it present, S02 would occur in the window between
COS and methyl mercaptan; it is apparently absent in appreciable amounts. The
peak labeled "CH3CH2SH?" was not confirmed by the injection of standards on
site. However, it is the only known compound eluting during this time period,
according to the work described earlier in this report, and is therefore very
likely identified correctly in the figure.
Figure 7-24 illustrates that COS and to a lesser extent methyl mercaptan
elute over a sloping baseline due to H2S, and both the latter compounds must be
determined with a tangent skimming procedure. For the chromatogram shown»H2S,
COS, methyl mercaptan, and CS2 occurred at concentrations of 3300, 62, 32 and
10 ppmv respectively, and should the proportion of H2S increase significantly,
COS could not be determined under these chromatographic conditions. Although
the separations during the May 1982 burn were conducted isothermally at 65°C, a
better separation of H2S and COS can be effected by temperature programming.
For example, through temperature programming between approximately 40°C and
120°C, the separation of COS can be improved significantly using an otherwise
identical chromatographic system (Wallace, instruction manual prepared for the
Laramie Energy Technology Center, 1982). However, temperature programming
requires a considerable increase in instrument cost and complexity compared to
isothermal systems. After due consideration of the types of samples expected,
the analysts must therefore compromise between the better separation achieved
with temperature programming and the lower costs of isothermal separation.
7-58
-------
The sampling system which was used during the May 1982 test is shown in
Figure 7-25, which also illustrates the auxiliary gas supplies and a system for
generating and supplying calibration gas mixtures. This system was designed to
provide a sample gas to the analyzers at a constant pressure and flow rate, and
to remove excess water vapor and entrained material to prevent fouling of the
analytical instruments. While it is likely that the system shown will be
modified by the analyst to meet specific requirements, it is worthwhile to
discuss briefly the functions of the various major components shown in the
figure in order to illustrate the requirements of a sampling system.
Two sample conditioning and delivery systems, one for the inlet and the
other for the outlet of a hypothetical treatment process, are shown in the
upper left of Figure 7-25. The functions of the various components can be
illustrated by considering only the top sampling train: the sampling probe.,
PI, consists of a 3/8 inch stainless steel tube which passes into the center of
the process line to avoid collecting condensation which occurs on the duct
walls. The first valve, VI, is included to permit the remainder of the
sampling system to be removed and leak checked without breaking the seal to the
process line. The condensers consist of two stainless steel vessels containing
glass wool and immersed in an ice bath. Their function is to remove excess
water vapor and most particulate matter. Additional filtering is provided by
filters HI and H2. All components discussed thus far are located next to the
process line to avoid fouling the sample line which runs to the sampling
trailer. During various tests it was discovered that pressure in the retort
gas process lines often varied and sometimes became sub-ambient, a situation
which made it impossible to rely on the process line pressure to force the gas
through the sample line. For this reason a leakless, aluminum-head sampling
pump, PI, was included to provide a constant flow. Valve V4 was included as an
7-59
-------
IM
>>
^-
ro
c
rcf
CO
O
'£
ITS
CO
OJ
(8
O)
O
Q.
Q.
3
CO
• CO
•a
LO S-
CM (O
i -a
UJ 4->
a; to
as
CD
CD
a.
a.
3
CO
•a
ca
o
•o
o
o
-------
aid in leak checking the downstream portion of the sampling system. Check
valve V15 was included to prevent backwards flow during periods of negative
pressure in the process line. The back pressure regulators shown consist of
acrylic tubes containing a column of water through which the sample gas must
pass before exhausting to the atmosphere. Although mechanical back-pressure
regulators were also tested, they did not prove reliable. Two main problems
became apparent with the back pressure regulators shown in the figure. First,
if a vacuum were inadvertently applied to the sample line, (e.g., due to pump
failure), the water in the regulators would be pulled back through the entire
system. This problem was solved by including two acrylic columns in line, the
upstream column acting as a catch basin for any overflow. Second, pressure
fluctuations in the process line still affected the analyzers, even with the
pressure regulators in place, albeit to a lesser extent. This phenomenon was
further minimized by including pressure regulator R9 and rotameter Fl for the
purpose of measuring and maintaining a constant flow rate through the sample
line. a
In order of operation, the flow rates are first set by opening valve VI
with the pump running. With valve V7 fully open, valve V5 is adjusted to give
slightly more than the desired flow rate. The flow rate is trimmed by
adjusting regulator R9 to cause a slight lowering of the pressure, and further
adjusted with valve V7 to achieve the desired flow rate. The system is next
leak checked by plugging the inlet of the sample probe. The system is
considered leak tight if no flow is apparent at the flow meter Fl.
If the gas chromatograph is connected directly to the sample line, the
entire pressure drop occurs across the 1/16" tubing which is contained therein.
With this arrangement small changes in the sample line pressure, or small
changes in the resistance of portions of the 1/16" tubing cause pressure
7-61
-------
changes in the sample loop and, therefore, changes in the calibration of the
chromatograph. Such effects are minimized by decreasing the flow through the
chromatograph with control valve V30. The flow rate should be set fast enough
to sweep out the sample delivery valves V17 through V20 and filter holder H5.
The flow through the chromatograph is monitored by a rotameter F8 and should
remain constant for different samples and standards.
The sampling system shown in Figure 7-25 was developed in response to the
practical exingencies of field sampling, and certain questions remain to be
addressed. For example, "What components are lost during the cooling step, at
which point the gases are exposed to an alkaline water solution?" It is also
of interest to consider what fraction of the sulfur pollutants are present in
the particulate as opposed to the' gaseous phase in the sample stream.
Potential methods for improving the reliability and integrity of the sampling
method are presently under investigation.
Figure 7-25 illustrates two approaches to providing standards, a mass flow
system and pre-mixed gas bottles. The former provides a range of
concentrations to bracket the sample concentrations, while the latter provides
a fixed standard for occasionally checking the operation of the analyzers.
This dual arrangement is convenient when operating more than one type of
analyzer, each on a different calibration schedule. For mass flow controllers
with the ranges indicated in the figure, H2S standards can be generated in the
range 500-100,000 ppmv starting with pure H2S. For preparing standards for the
minor gases, it is necessary to start with more dilute mixtures, typically in
the range 0.1-1.0% v/v.
7-62
-------
PROTOCOL FOR DETERMINATION OF HYDROGEN SULFIDE, CARBONYL SULFIDE, SULFUR
DIOXIDE, CARBON DISULFIDE, METHYL MERCAPTAN, AND ETHYL MERCAPTAN IN RETORT GAS
Principle '
The above mentioned sulfur species, referred to hereafter as the primary S
species, are determined by gas chromatography with flame photometric detection.
The concentration range of interest in retort gas is covered by using two
sample injection valves with different injection volumes.
Interferences
Hydrocarbons, which co-elute with the S species of interest, alter the
fluorescent S signal if present in excess. For the procedure described herein,
this interference is addressed by proper selection of column and detector. In
the case of the single flame detector (SFPD) the operating mode also affects
the likelihood of interferences.
Using the column recommended in this section, H2S co-elutes with ethane
and ethene. However, with a one microliter injection volume, 1,000 ppmv of H2S
can be determined in 10% ethane without measurable error by using the single
flame detector in the reverse mode and by calibrating with peak height. Higher
levels of ethane can be tolerated with a dual flame detector (DFPD). If the
sample volume is increased to higher values to detect lower levels of H2S or
other minor S species, less ethane could be tolerated. However, with the
concentrations of H2S normally found in retort gas, it is unlikely that such
interferences would be a problem.
With the column recommended below, S02 elutes very close to the C3
hydrocarbons. In the event that S02 is detected in retort gas this
interference should be considered. However, as of this writing, S02 has not
been found in any retort gas of which the authors are aware.
7-63
-------
In addition to the above-listed compounds, it must be recognized that
there are numerous additional S compounds which have a high enough vapor
pressure to be found in retort gas. Listing the C2 to C4 mercaptans, sulfides
and disulfides, in addition to the thiophene and methyl thiophenes, one can
easily generate a list of 20 such compounds. In general, these compounds will
elute after the above-named compounds. However, CS2 elutes close to dimethyl
sulfide, and the elution times of both compounds should be established with
standards.
Equipment
1. Gas chromatograph. The gas chromatograph should be 'equipped with two
gas sampling valves and a flame photometric detector. The necessary
arrangement of valves and columns is shown in Figure 7-26.
The sample injection volumes must be selected for the
concentrations of S species encountered, and the injection volumes
should be changed on site should unexpected concentrations be encoun-
tered. With a packed column, a 1 micro!iter injection volume covers
the range 500-20,000 ppmv, which is adequate for monitoring H2S in
retort gas from most indirect retorting processes. A 50 micro!iter
sample loop is typically adequate for the range 10-200 ppmv, the
concentrations of minor gases encountered in the same type of gas.
It is most convenient if the sample valves are operated from an
automatic timer.
2. Integrator. If the separation is carried out isothermally, Csee
below) it is essential that the integrator be capable of
tangent-skimming COS from the tale of the H2S peak. Alternatively,
the integrator can be replaced with a dual-pen, dual-range strip
chart recorder. However, a peak integrator has the advantages of a
7-64
-------
CM
LU
Di
cu
if)
O)
in
to
cn
+J
o
4->
CD
to
cu
•r~'
O
QJ
CX
IO
CO
O»
s.
o
s=
o
c.
o
(O
cn
o
o
cu
•a
c
ca
O
O
-------
wider dynamic range, automatic printout, and programmable control of
the valves. ;
3. Columns. Two columns are required, one for each injection valve. A
1/8" od X 6' column of Carbopack B HT 100 is adequate for most
compounds, including H2S, COS, S02, methyl mercaptan, ethyl mercaptan
and CS2. Thiophene does not elute from this column in a reasonable
time and must be determined with a different column (e.g.,
Chromosil 310). It is not clear at this time whether later eluting
< compounds can be uniquely identified on the basis of their retention
time using a packed column.
4. Detector. A flame photometric detector is required to distinguish
among the S species and the other numerous gases found in retort gas.
Of the FPD's available, the dual flame model is least prone to
interferences due to co-eluting hydrocarbons. The single flame'model
is more prone to interferences, although it may be applicable for
certain types of retort gas. (See previous text.) Hydrocarbon
quenching effects in the single flame FPD are minimized by reversing
the hydrogen and oxygen flows so that the air is introduced
concentric to and interior to the flow of hydrogen. However, in this
configuration the flame is often difficult if not impossible to
ignite.
5. Mass Flow Controller. Mass flow controllers are needed for blending
gas standards. A set of two controllers which operate in the range
0.5-10.0 SCCM plus one controller in the range 50-1,000 SCCM are
convenient. Connect as shown in Figure 7-25.
7-66
-------
Standards
Standards should be prepared which bracket the concentrations of the S
species found in the retort gas. This is most easily achieved by mixing
standards on site using a set of mass flow controllers.
Calibrate each of the mass flow controllers against a volumetric standard
which is traceable to an NBS standard. The following procedure is applicable
to the 1,000 SCCM mass flow controller. Calibrate the stop watch against an
NBS time base which is available on most telephone exchanges.
A dry test meter is required for calibrating the 1,000 SCCM channel, while
the 10 SCCM channels require a small bubble meter. The following procedure is
written in terms of the dry test meter, although the procedure is identical for
the 10 SCCM channels with the substitute of the bubble meter in place of the
dry test meter:
1. Leak check. Connect a supply of N2 to the inlet of the mass flow
controller and the dry test meter to the outlet. Block the exit from
the dry test meter with a leak-tight seal and pressurize slightly
with N2. Adjust the mass flow controller to its maximum rating.
(CAUTION: Do not exceed the pressure rating of the dry test meter.)
The mass flow controller should read zero flow. Back off the
regulator to a lower pressure. The pressure at the regulator should
remain constant.
2. Calibration. Remove any restrictions from the exit of the dry test
meter. Set the flow rate on the mass flow controller and allow a
reading to stabilize. Determine the volumetric flow rate by
measuring the time required for an integral number of revolutions of
the dry test meter. Adjust to standard conditions. Repeat for
several flow rates spanning the intended range of use.
7-67
-------
It is recommended that flow rate calibrations be conducted with the gas to
be monitored. However, in the case of certain reactive gases such as H2S, this
is not presently possible with available equipment. For example, the fluid in
a bubble meter absorbs a large percentage of the flow of H2S, while dry test
meters are not practical at the low flow rates required for H2S. For such
gases, the mass flow meter must be calibrated with N2, and then converted to
H2S using conversion factors which are calculated from the molar weight and
heat capacity.
Procedure
Typical chromatographic conditions for an isothermal separation are listed
in Table 7-2. However, chromatographic conditions may need to be adjusted
somewhat to take into account differences among samples. In particular,
separation of COS from the tail of H2S can be improved by starting the column
temperature at approximately ambient and increasing after the elution of methyl
mercaptan.
Calibrate the chromatograph over the entire range of interest to determine
the linear range of the system. Plot the square root of the peak height
against concentration.
Inject a sample under identical flow rate and pressure as the standard and
note the peak height of each component of interest. Bracket the peak heights
of the sample by the heights of the standards. Determine the concentration by
comparison to the calibration plot described in the previous paragraph.
REFERENCES
Bradey, S.S, J.E. Chaney. 1966. J. Gas. Chromat., 4:460.
De Souza, T.L.C., S.P. Bhatea. 1976. Development of Calibration and Moni-
toring Systems to Measure TRS and S02 in the PPB Range. Presented at the
1976 Pittsburgh Conference.
7-68
-------
Farwell, S.W., S.J. Cluck, W. L. Barnesberger, T.M. Schutte, D.F. Adams. 1979.
Determination of Sulfur Containing Gases by a Deactivated Cryogenic
Enrichment and Capillary Gas Chromatography System. Anal. Chem., 51:609.
Gangwal, S.K., D.E. Wagoner. 1979. Response Correlation of Low Molecular
Weight Sulfur Compounds using a Novel Flame Photometric Detector.
J. Chromat. Science, 17:196.
Gilbert, P.T. 1970. "Non-Metals" in Analytical Flame Spectroscopy.
R. Marrodineanu, ed. Sprenger-Verlog, NY.
Goodfellow, L., M.T. Atwood. 1974. Fischer Assay of Oil Shale Procedures of
The Oil Shale Corp. Proc. of the Seventh Oil Shale Sym. Golden, CO.
Colorado School of Mines, 69:205.
Lucero, D.P., J.W. Palqug. 1973. Monitoring Sulfur Compounds by Flame
Photometry. ASTM Special Tech. Pub., 74:555.
Martin, R.L., J.A. Grant. 1965. Determination of Sulfur Compound Distribution
in Petroleum Samples by Gas Chromatography with a Coulometric Detector.
Anal. Chem., 37:644.
Patterson, P.L. 1978. Dual Flame Photometric Detector for Sulfur and
Phosphorus Compounds in Gas Chromatography Effluents. Anal. Chem.,
50:345.
Pearson, C.D., W.J. Mines. 1970. Determination of Hydrogen Sulfide, Carbonyl
Sulfide, Carbon Disulfide, and Sulfur Dioxide in Gases and Hydrocarbon
Streams by Gas Chromatography/FPD. Anal. Chem., 49:123.
Sklarew, D.S., J.S. Fruchter, D.M. Schoengold, M. R. Tompkins, E. L. Morris.
1981. Measurement of Sulfur Species in Offgas .from Rio Blanco1s Retort on
Tract C-a Colorado. Preliminary Report to DOE.
Stetter, J.R., J.M. Sedlak, K.F. Blurton. 1977. Electrochemical Gas
Chromatographic Detection of Hydrogen Sulfide at PPM and PPB Levels.
J. Chromatogr. Sci., 15:124.
Steven, R.K., A.E. O'Keefe. 1978. Modern Aspects of Air Pollution Monitoring.
Anal. Chem., 143A.
Stevens, R.K., J.D. Mulik, A.E. O'Keefe, K.J. Krost. 1971. Gas Chromatography
of Reactive Sulfur Gases in the Air at the Parts-per-Billion Level.
Anal. Chem., 43:827.
Stuber, H.A., J.A. Leenheer, D.S. Farrier. 1978. Inorganic Sulfur Species in
Wastewaters from in Situ Oil Shale Processing. J. Environ. Sci. Health.
Part A, A13:663-675.
Sugiyama, T., Y. Suziki, T. Takeuchi. 1973. Interferences of S2 Molecular
Emission in a Flame Photometric Detector. J. Chromatography, 80-61-67.
Thompson, B. undated. Determination of Sulfur Compounds by Gas Chromatography.
Varian Associates.
7-69
-------
Von Lehmden, J.D. 1978. Suppression Effects of Carbon Dioxide in FPD Total
Sulfur Air Analyzers and Recommended Corrective Answer. Joint Conf.
Sensing Environ. Pollutants, 4:360.
Walker, D.S. 1978. Gas Chromatographic Determination of some Sulfur Gases at
the Volume per Million Level in Air Using Tenax GC. Analyst, 103:397.
Wallace, J.R. 1981. The Analysis of Oil Shale Wastes: A Review.
EPA-600/57-81-084.
Wallace, J.R., E. Sexton. 1982. Evaluation of a Gas Chromatographic Method
for Monitoring H2S in Retort Gas. Report to Laramie Energy Research
Institute. See also Instruction Manual for Monitoring S Species During
Test of a Stretford Unit.
7-70
-------
SECTION 8
CONCLUDING REMARKS
In addition to the tests investigated explicitly during this program,
several other measurements have been employed for the routine analysis of
retort effluents, an experience which has provided worthwhile information on
many procedures. While such information does not justify its own chapter, it
is nevertheless helpful to the analyst. During the period of this investiga-
tion, information on additional methods has also be described by contemporary
investigators. The purpose of this section, therefore, is to summarize the
available information on methods of analysis for oil shale effluents which
should be important to the wastewater engineer.
The pH of retort water is determined almost universally by application of
the pH electrode. In the authors' laboratory this method is applied routinely
without major problems. However, certain procedural precautions are necessary
to obtain the best results. First, the glass membrane becomes fouled easily
when working with oily waters in general and retort waters in particular, which
results in drifting and inaccurate response. (The electrode response can be
restored by following the manufacturer's instructions for this purpose, which
normally involves washing the electrode with a strong detergent.) This means
that the calibration and response of the electrode must be checked both before
and after a series of measurements with at least two buffer standards; a
degraded response at the end of the series requires that all analyses be
repeated. It should also be realized that the pH of retort waters is tempera-
ture dependent. For example, the pH of one set of retort waters examined
varied 0.024 pH units/°C. While many pH meters are equipped with a
"TEMPERATURE" control, this typically adjusts only the Nernstian slope and does
8-1
-------
not compensate for the effects discussed here. Thus, in order to obtain pH
values which are accurate to ±0.1 pH unit, the temperature should be
controlled to better than ± 4°C; in order for the reported pH to be meaningful,
the measurement temperature must also be reported.
In the authors' experience, when these precautions are taken pH
measurements on retort water are normally precise to ± 0.1 pH units, which is
the limit of^readibility for the pH meter in use. While the absolute accuracy
of this measurement has not been established by reference to an independent
method, it is noteworthy that when ion balances were calculated on one set of
dilute samples using pH measurements so determined, the balances were within
experimental error. (See below.)
Similar comments can be made regarding the measurement of electrical
conductivity (EC). As with the pH electrode, the conductivity probe is easily
fouled and needs to be cleaned periodically according to the manufacturer's
instructions. In order to assure that the cell has not been fouled during a
series of measurements, it is necessary to analyze a KC1 standard both before
and after and to discard any analyses if the calibration has changed. The
analyst should also be aware that the temperature coefficient for the EC
measurement is different for retort waters than for most surface and ground
waters. For example, for one set of retort waters the EC varied approximately
4%/°C, while the conductivity of most surface waters varies approximately
2%/°C. Many of the commercially available conductivity meters are equipped
with a temperature compensation control which corrects the conductivity 2%/°C,
which would obviously be inappropriate for retort waters. It is therefore
recommended that retort waters be equilibrated to a fixed temperature, say
25°C, before the EC is measured, and that this temperature be reported with the
conductivity results. It should also be noted that the EC does not necessarily
8-2
-------
vary linearly with concentration in high-ionic strength waters so that using EC
as an indication of total dissolved ions is not as accurate for retort waters
as for more dilute surface and ground waters.
Alkalinity, which is defined as the amount of strong acid required to
lower the pH to certain end points, is often interpreted in terms of the amount
of carbonate and bicarbonate. While such an interpretation is usually correct
for surface and ground waters, it is incorrect for most retort wastewaters
because of the presence of NH3, which consumes a portion of the acid titrant.
Examination of a typical titration curve of an oil shale wastewater
(Figure 8-1) indicates the absence of clear end points and emphasizes the
difficulty of determining carbonate species by this method. Instead, it is
recommended that inorganic carbon (carbonate + bicarbonate) be determined by
the instrumental methods that are used for total organic carbon.
Total inorganic carbon (TIC), which in retort waters is equivalent to
carbonate + bicarbonate, is determined instrumentally by acidifying the sample
and purging the resulting C02 into a detector. Total carbon (TC) is determined
similarly by combusting or rigorously digesting the sample to convert all
organic and inorganic carbon to C02. Commercially available instruments
typically combust the sample in an oxygen stream or digest it in strong
chemical oxidants; detection is by coulometry, infra-red absorption, or flame
ionization after conversion to methane. Several variations of these procedures
are available: total organic carbon (TOC) can be determined "indirectly" by
measuring TC and TIC separately and calculating the difference, or "directly"
by acidifying the sample in order to remove the inorganic carbon prior to
injection into the instrument. If the sample is filtered through a
0.45 micrometer filter prior to injection, the resulting measurements are
8-3
-------
CJ
cr>
oo
p
ui
CD
irj
O
rt-
O
CO
O
to
p z
! z
p
o
p
CO
• =5
CS
UJ
o
ex
(t3
rtf
£3
S-
O
O)
>
3
o
o
C3
evj
8-4
-------
referred to as dissolved organic carbon (DOC), dissolved inorganic carbon
(DIG), and dissolved total carbon (DTC).
Although TOC and TIC are important measurements, they have not been
addressed during the experimental phase of this study because of concurrent
investigations being carried out at other laboratories. In the authors'
laboratory TOC and TIC measurements have been carried out routinely in the
support of various pollution control programs, and the quality control
procedures applied during'these tests have provided some insight into the
problems which can occur with these procedures. For example, with an instru-
ment using an infra-red detector the reproducibility for the measurement of TIC
and TC measurements for retort waters analyzed under field conditions was
approxiamtely 10% relative. Thus, when determining TOC by difference, the
errors generated by taking the difference between two large numbers can be
substantial. Applying the common formulas for propagation of error indicated
that for retort waters the error in the TOC measurement varied 15 to 100%,
depending on the relative concentrations of TIC and TC. During this same
investigation, however, no evidence of analytical bias was observed and the
recovery of added TIC was 120%, which is reasonable for the precision of the
technique used (Habenicht et a!., 1980). TIC measurements have also been
performed under similar conditions by coulometric titration with a routine
precision at 1% relative (one a). These data illustrate the importance of
applying quality control and statistical interpretation during the determina-
tion of TOC and TIC, and of reporting the precision of the results along with a
statement regarding the method by which the precision was determined.
Others who have studied the determination of TOC and TIC more thoroughly
have arrived at similar conclusions. For example, an interlaboratory
comparison including 18 laboratories has been carried out using retort and
8-5
-------
synfuel wastewaters along with identical samples which had been spiked with
known quantities of TOC. A variety of analyzers was applied, including those
with infra-red, coulometric, and flame ionization detectors. The results from
this study indicated that there was no significant difference between the
results obtained with the various types of analyzers. The recovery of the
spiked standards was essentially 100%. However, those analyzers equipped with
coulometric detectors were the most precise. Standards of known composition
were also included in this study to determine which laboratories were capable
of measuring TOC in pure solutions at concentrations encountered in retort
waters, and the results of three laboratories were rejected on this basis. The
results from these standards also indicated that many laboratories had
difficulty quantitating high level samples. Thus, those laboratories capable
of-analyzing pure samples were normally capable of analyzing retort waters.
These investigators thus concluded that "most of the problems associated with
the analysis of these samples are due to technique and not the TOG method
itself," and provided a list of 9 suggestions for best results (Jackson,
personal communcation, 1982). These included suggestions for volumetric
dilution, careful application of quality control procedures such as
replications and spikes, vigorous stirring, and practice with very high level
standards.
In a separate study, Langlois et al. (1982) compared the TOC determina-
tions obtained with two different oxidation processes, the first consisting of
high temperature oxidation in 02 and the second consisting of a photochemically
assisted oxidation in persulfate solution. A coulometric finish was employed
in both cases.
Of the 17 pure compounds tested, 15 gave equivalent results by both
methods while two gave incomplete recovery by the photochemically assisted
8-6
-------
procedure. However, for the 9 retort wastewaters examined equivalent results
were obtained, indicating that the most refractory compounds were not signifi-
cant sources of error in retort water. DOC was determined both by difference
and directly, and no bias between the two methods was observed. However, the
direct method was the more precise. The precision of these analyses was
approximately 1% relative (la). Compared to the other finishes available, the
coulometric method was preferred by these authors because of a wider dynamic
range, a higher upper limit, and more stable calibration. In summary, the
avail able'evidence suggests that the existing methods for measuring TOC and TIC
in retort wastewaters are adequate and that acceptable results can be obtained
by considering the guidelines mentioned in these paragraphs. The data
discussed herein also suggest that the coulometric titration is the preferred
finish.
In the authors' opinion there is not yet a proven method for the determin-
ation of total nitrogen in retort water. The same statement therefore applies
to organic nitrogen, which is defined as the difference between total and
ammoniacal nitrogen. This deficiency may be important because organic bases
are abundant in retort water, and it would be helpful to be able to establish
their total concentration as a quality control parameter. Two methods are
presently in use for the total nitrogen measurement: (1) the well-known
Kjeldahl method, and (2) combustion followed by the chemiluminescent detection
of the NO produced. The problem with the first method is that the more
refractory compounds are incompletely digested (although this can often be
alleviated by the addition of catalysts and the use of higher temperatures),
and that the method is slow and cumbersome for field use.
The second method was investigated during the first phases of this
project, and was shown to give different responses for different compounds
8-7
-------
(June, 1980 Progress Report). That is, the conversion efficiency to NO was
compound dependent, especially for azo- and nitre-compounds. Similar findings
have been reported in the literature for this method (Drushel , 1977). Whether
or not this is significant for retort waters has not yet been determined.
In addition to the tests which are performed for the purpose of evaluating
pollution control equipment, certain other tests are helpful to the analyst for
the purpose of checking the correctness and completeness of the individual
determination. One such procedure is the calculation of the ion balance, that
is, the electrical balance between the anions and cations which have been
determined. Here the ion balance, B, is defined
ZC ~ ZA
R = ?
+ Z
ZA
where lc and IA indicate the absolute values of the cationic and anionic
charges respectively. Since the true ion balance is always zero, incorrect or
incomplete analyses are reflected in non-zero values for B. In practice the
major cation (NH|) and anions (COf, HCOi) are not measured directly, but are
calculated based on measurements of total ammonical N, total inorganic carbon,
and pH. Thus, errors in any of these measurements result in non-zero balances.
The principal difficulty with the ion balance calculation is due to the
buffering capacity of retort waters. That is, a small change in pH causes a
large change in ionic concentrations. This problem can be illustrated for a
water containing 100 mM ammonical N and 45.84 mM total inorganic carbon. At a
pH of 9.25, the ion balance would be 0.00, assuming ideal solution behavior.
However, if the pH were measured incorrectly high by 0.1 pH units, which
8-8
-------
represents a reasonable error during routine analysis, the calculated ion
concentrations would change as follows:
NHJ + CO! > NH3 + HCOi
pH = 9.25 50.00 mM 4.17 mM 50.00 mM 41.67 mM
pH = 9.35 44.26 mM 5.13 mM 55.74 mM 40.71 mM
For a measured pH of 9.35 the resulting ion balance would thus be -14%. Thus,
small errors in pH measurement thus contribute to large imbalances.
A second precaution regarding the ion balance is related to the high ionic
strength of many retort waters. The calculations described in the previous
paragraph are strictly correct only at infinite dilution and become less
accurate as the ionic strength increases. In theory, this problem can be
alleviated by simply diluting the sample before measuring the pH which is used
in the ion balance calculation, a procedure which can result in a slightly
different pH than is obtained with an undiluted sample. This is illustrated by
analyses performed in the authors' laboratory on a set of 15 samples. Ion
balances were normally within ±15% for samples containing less than 30 meg/1 of
NH|. For samples containing approximately 100 meg/1 of NHJ, ion balances were
in error by approximately 80%. However, when the latter samples were diluted
and the pH of the diluted solution was used in the calculation, more reasonable
ion balances were achieved (Shaffron and Wallace, 1982).
Although electrical conductivity is often used to check the determinations
of total ionic species, it is obviously subject to the same problems inherent
in the ion balance calculation. While both of these determinations are useful
for retort waters, they are obviously best applied to diluted samples. A
measurement of pH more accurate than the normal ± 0.1 pH units would also
improve the quality of these calculations.
Table 8-1 summarizes the recommendations of this report for methods of
chemical analysis for oil shale wastewaters and product gases. Combusted gas
8-9
-------
Table 8-1
RECOMMENDED PROCEDURES FOR THE ANALYSIS OF OIL SHALE EFFLUENTS
Species
Recommended Methods
WASTEWATERS
pH
Electrical conductivity
Total dissolved material
Sulfide (HS~ + S=)
Total ammoniacal N, (Na,K)c
Aqueous NH3
Inorganic C (HCOi + C0§)
Total organic carbon
Ion Balance
Total S
Anionic species:
S201 SCN~, NOi, SOI,
S0|, P0|-, NOi
Total S
Individual S species:
H2S, COS, CH3SH, CS2, CH3CH2SH,
pH electrode
Conductivity cell
Freezing point depression
Pontiometric titration with Pb(II)
(1) ion chromatography
(2) modified electrode method
UV absorption of headspace
Evolution of C02 from acidified
sample; coulometric finish
Combustion or UV assisted chemical
oxidation to C02;
coulometric finish
Calculate based on pH of diluted
sample
Oxidation with H202 followed by
determination of SOI
Ion Chromatography
RETORT GAS
thiophene
Continuous combustion followed
by measurement of S02
Gas chromatography with flame
photometric detector, using
procedures detailed in text.
When using ion chromatography, Na and K can often be determined simultane-
ously with NH|. However, if NH^ need not be measured, more convenient
methods are available for K and Na (Wallace, 1981).
8-10
-------
streams have not been treated because of their similarity to utility and
industrial stack gases which have already been studied exhaustively. For
information on additional methods, such as minor and trace elements, the reader
is referred to the first report prepared for this project (Wallace, 1981).
While these proposed methods have been tested to varying degrees with actual
retort samples, it must be realized that wider application will inevitably lead
to new information and refinements. The authors of this report would
appreciate such information as it is developed by practitioners.
In conclusion, it is worthwhile to emphasize the importance of quality
control and quality assurance in applying these methods to oil shale effluents.
Although a laboratory may have control charts and statistical interpretations
based on their experience with a variety of surface and ground waters, these
data should not necessarily be applied to retort samples. Statements of
precision and accuracy should be based on experience with actual retort
samples, not on experience with related samples or on the repeatibility of
standard solutions. In order to obtain the best estimate of the precision,
samples should be repeated, not sequentially, but spaced at intervals
throughout the day in order to include contributions from instrument drift (at
least for stable species). Standards should also be run both before and after
a series of analyses in order to be sure that the instrument did not change
during exposure to retort samples. It is, of course, recommended that the
analyst first practice with a previously characterized sample whenever
possible.
Because many of the species of interest in retort wastes are unstable, it
is important that the analyst attempt to establish the integrity of the
sampling methods and not simply rely on published guidelines. For this purpose
it is recommended that the sampling and.preservation techniques appearing in
8-11
-------
the literature be a starting point, but that decay studies be an integral part
of any sampling and analysis program. If the efficacy of the sampling and
preservation techniques cannot be established for the particular samples in
hand, it is recommended that analyses be completed immediately on site.
Finally, the analytical report should include a quality assurance section
specifying not only such parameters as precision and recovery, but the exact
method by which they were determined.
REFERENCES
Drushel, H.V., 1977. Determination of Nitrogen in Petroleum Fractions by
Combustion with Chemiluminescent Detection of Nitric Oxide and Chem,
49(7):932.
Habenicht C.H., A. Jovanovich, M. Shaffron, J. Wallace, E. Hicks, L. Liang,
1980. Steam Stripping and Reverse Osmosis for the Treatment of Oil Shale
Wastewaters: Report of a Field Test. Report to the Laramie Energy
Technology Center (DOE) from the Denver Research Institute, Project
5-31343, November, 1980.
Longlois, G.W., B.M^ Jones, R.H. Sakaji, 1982. Coulometric Quantitation of
Carbon via UV-peroxydisulfate or High Temperature Oxidation. Report
prepared under US DOE Contract DE-AC03-765F00098.
Shaffron, M., J.R. Wallace, 1982. Oil Shale Wastewater Treatment for Transfer
to an Open Pond. Report to Laramie Energy Technology Center prepared
under Contract DE-AS20-81LC0775.
8-22
-------
I. HEPOR1
. . TECHNICAL REPORT DATA
iL'ase read Imtrucnom on the .-fiivre tm'/tjre completing/
3. RECIPIENT'S ACCESSION NO.
Methods of Chemical Analysis for Oil Shale Wastes
5. REPORT DATE
S. PERFORMING OHGAN1ZATI )N CODE
7. AUTHQRtSI —' •
John Wallace, Linda A!den, Francis S. Bonomo,
John Nichols, Elizabeth Sexton
3. PERFORMING ORGANIZATION REPORT NO.
PERFORMING ORGANIZATION NAME AND AQORESS
Denver Research Institute
Chemical and Materials Sciences
Univeristy of Denver
Denver, CO 80208
10. PROGRAM ELEMENT NO.
1 1. CONTRACT/GRANT NO.
12. SPONSC
ME AND ADDRESS
Industrial Environmental Research Laboratory
Office of Research and Development
Environmental Protection Agency
Cincinnati, OH 45268
13. TYPE OF REPORT AND PERIOD COVERED
14. SPONSORING AGENCY CODE
5. SUPPUE
6. ABSTRACT ——————______ ___ •
Several methods of chemical analysis are described for oil shale wastewaters and
•etort gases. These methods are designed to support the field testing of various
'Dilution control systems. As such, emphasis has been placed on methods which are
apid and sufficiently rugged to perform well under field conditions. Ion chromato-
raph has been developed as a=technique for the minor non-carbonate inorganic anions
n retort water, including S04, NOg, S20~3, SOT, and total S. Acetate, CT, N02, SOg
nd P0|~ can be observed with this technique but cannot necessarily be separated if
resent simultaneously^ The method recommended for sulfideis a potiometric titration
1th Pb(II). SCN", S20s, SCC, CT, C03, NH3 , and OH" were shown to not interfere with
his technique. The freezing point depression is used to determine the total solute
ontent in retort waters, a test which can be considered analogous to the standard
esidue test. Three methods are described for the determination of total ammoniacal
itrogen in retort wastewaters: (1) a modified ion selective electrode technique,
2) an optical absorption technique, and (3) an ion chromatographic technique. The
atter techlnque is recommended for routine monitoring of retort water, although the
elative advantages of each are discussed in the report. Total sulfur in retort gas
s determined by combusting the gas in a continuously'flowing system, whereupon the
esulting sulfur dioxide is determined by S02monitor. Individual sulfur species in re
ort gas including H7S, COS, S02,CS2, CH3CH7SH, and CH3CH?SH, are determined by gas (C
er)
KSY WORDS AND DOCUMENT ANAUYSIS
DESCRIPTORS
ossil Fuels, Oil Shale pollution, chemical
halysis, chromatographic analysis, flame
hotometry, gas analysis, spectroscopic
nalysis, volumetric analysis, water analy-
b.IDENTIFIERS/OPEN ENDED TERMS
sulfate, thiosulfate, ni-
trate, thiocynate, total
sulfur, sulfide, residue
ammonia, hydrogen sulfidt
carbonyl sulfide, carbon
disulfide, methyl mercap-
tan, ethyl mercaptan, pH
19. S6CUHI TY CLASS (Hill Repurt)
70 SECURITY CLASS
e. COSATI ficid/GfOup
NTIS Terms:
97F--Fuel Conver-
sion
97R—Energy, envi-
ronmental studies
99A—Analytical
Chemisti
21. NO. OF PAGES
22. PRICE
EPA Form Z2;0-l :J
------- |