EPA/600/A-92/095
COMPARISON OF METHODS FOR DETERMINATION OF
DISSOLVED INORGANIC CARBON (DIC)
Michael R. Schock
Drinking Water Research Division
U. S. Environmental Protection Agency
Cincinnati, Ohio 45268
Gregory K. George
Technology Applications Incorporated
Cincinnati, Ohio 45268
INTRODUCTION
The accurate determination of the concentration of dissolved inorganic carbon
(DIC) is important in many areas of drinking water treatment and aquatic chemistry.
For example, when adjusting the alkalinity of water during some forms of treatment
(corrosion control, softening, etc.), several chemicals are widely used: sodium (or
potassium) hydroxide, lime, slaked lime, sodium bicarbonate, sodium carbonate (soda
ash), and sodium silicate. Of these chemicals, only sodium carbonate (NaiCOj) and
sodium bicarbonate (NaHCOj) actually affect the total carbonate concentration.
Controlling the saturation state of calcium carbonate is also very important, in
both lime softening and corrosion control treatment processes. DIC plays an integral
role in these treatments, because it impacts the buffering intensity of the water1: 3'4.
It also controls, or at least affects, the deposition potential of important scale-forming
solids such as calcium or ferrous carbonate. This is clearly seen by examining the
simple saturation index relationship for CaC03,
(Ca2 -} {CO,}
^1calcica ^°9l0 y ^ '
sp
where {} denote activity of the aqueous species. This is similar in concept to the
Langelier Index, although the original Langelier Index derivation used alkalinity
instead of carbonate ion activity. In addition, a correspondingly different equilibrium
constant expression was used. The solubility of calcite is shown in Figure 1. The
numbers labeling each solubility curve represent DIC in mg C/L. The importance of
DIC in controlling calcium solubility is evident.
Beyond the traditional calcium carbonate equilibria, DIC directly controls the
solubility of metals and corrosion by-products such as lead, copper, and zinc. Figure
2 shows the importance of dissolved carbonate species, that make up significant
fractions of the dissolved lead concentration in the pH range of 7 to 10. even with a
DIC concentration of 3 mg C/L (2.5 x 10'4 mol-'L) Higher concentrations of DIC
complex even more lead5 6 7 8. Even when corrosion inhibitors containing
orthophosphate are added to the water, DIC influences lead (as well as copper and
zinc) solubility. Figure 3 illustrates this for pH 7.5.
SCHOCl} 299
-------
Another important impact of DIC is the indirect role it plays in limiting the
solubility of zinc, which may be applied as a constituent of corrosion inhibitor
formulations. This is shown in Figure 4. The effect upon zinc solubility is
dependent upon pH. Zinc dosages beyond the solubility levels can result in turbid
water and loss of inhibitor effectiveness. Similar to what was illustrated previously
for lead, DIC also influences the solubility of zinc orthophosphate passivation films
on galvanized pipe or possibly brass.
This paper has three main objectives. First, an overview will be given of
ways to either directly analyze DIC, or to compute it from data thai can be readily
acquired by most utilities or environmental laboratories. Second, a description of a
direct analysis method for DIC or TIC (total inorganic carbon) will be given, as has
been practiced by the authors using commercially-available instrumentation. Several
considerations of important analytical technique will be described. Finally, the direct
analysis and ngorous computational procedures will be compared under conditions of
practical'application using data collected in a ground water well incrustation sampling
project, in a corrosion control pipe loop experimental project, and in further analytical
and computational studies at the USEPA laboratory of the authors.
APPROACHES TO DETERMINE DIC
In a fashion analogous to differentiating "dissolved" metal concentrations from
"total" metal concentrations, "DIC" is differentiated from "total inorganic carbon
(TIC)" operationally by filtration or ultracentrifugation. Typically, a 0.4 ^m or 0.22
fim membrane filter is used for the differentiation. To accurately preserve the
concentration of DIC in the original sample, the filtration must be carried out
essentially in-line from the sampling device or port, so that all air exchange and
opportunities for pressure-drops are eliminated Being a volatile gas, CO, can either
be absorbed from the air, or lost, thus changing DIC (even if total alkalinity does not
change). Lack of air contact must be maintained through the actual analysis steps for
accurate analytical DIC or TIC values to be obtained.
Many papers have been published in various scientific and engineering fields,
that describe several different instrumental or wet chemistry procedures to quantify
DIC. The methods and instrumentation cover a vast range in cost, complexity, and
reliability. Table 1 is not intended to be a comprehensive bibliography, but is
intended to give a sampling of the variety of techniques, and a list of appropriate
references for their basis and use.
This paper focuses on important aspects of the calculation of TICDIC from
the analyses of pH and total alkalinity, because that is the most likely approach that
will be convenient to utilities and most water researchers. Additionally, the
determination of TIC/DIC by direct coulometric titration will be covered, which can
provide the most precise and accurate results in most cases for laboratories having
more sophistication and bigger instrumentation budgets. The versatile coulometric
titration method can also easily be used to analyze the carbonate content in solids,
such as corrosion deposits, pipe or filter scales, soils, and other materials. The
coulometric titration method provides a much more reliable alternative to traditional
C02-trapping gravimetric procedures.
CALCULATION OF DIC FROM pH AND ALKALINITY
300
-------
The fundamental principle of the computational method from pH, total
alkalinity, and ionic strength data is the general definition of total alkalinity:
TALK = 2[,CO^ ]*[HCO)]*[OH~]-[tf*] (2)
This expression is derived from the electroneulrality condition for water. Interested
readers should consult some of the many books and articles in which a variety of
authors explain the development of this concept in considerable detail. Numerous
references derive and explain many of the other important relationships among pH
and dissolved carbonate species1"4'9"12-.
Equation 2 is made up of two main sets of species, those attributable to DIC
(C03:\ HCO3 ), and those attributable to the water itself (OH , H + ). Thus, total
alkalinity is not independent from pH. Changes in pH will affect the total alkalinity
of water, but will only impact the total carbonate concentration indirectly by affecting
the solubility of C02 gas in water (which generates the other carbonate species
dissolved in the water). In a closed system, however, where there is no gain or loss
of CO; gas, raising or lowering pH (such as by adding acid or caustic) will not affect
DIC, but will raise or lower alkalinity . Figure 5 depicts this interdependence of pH
and total alkalinity for the case of a water at 15° C, and an ionic strength of 0.01
(TDS of approximately 350-450 mg/L). The straight lines show the example of
constant DIC concentration, regardless of pH. For the corresponding DIC
assumptions (3, 30, and 60 mg C/L), the curved lines show how total alkalinity vanes
as pH is changed.
Figure 6 illustrates an additional important aspect of this point. The total
alkalinity/DIC/pH relationship (from equation 2) can be linearized2-3,8,14,15. This
transformation enables DIC to be easily read from a graph for appropriate chemical
conditions (temperature, ionic strength), if pH and total alkalinity are measured. In
the example here, two waters are compared. Both have a total alkalinity of 25 mg
CaCOj/L, but one has a pH of 10 (the top line) and one has a pH of 6 (the bottom
line). The actual carbonate concentrations are 3.4 mg C/L and 18.6 mg C/L,
respectively. An important application of this kind of understanding is that just
because a water has a "low-" alkalinity, it doesn't necessarily have a low DIC content.
If water treatment is planned, for instance to try to reduce corrosivity by forming
passivation films on the pipe, the water may end up having an unexpectedly high
alkalinity by the time the pH was raised to 8 or so. The higher alkalinity will result
in a higher buffering intensity (resistance to pH change, which is mathematically
directly related to pH and TALK). That will affect the size of lime, caustic, or other
chemical dosages used to raise the pH. It can also increase the solubility of metals
such as lead and copper through the formation of soluble complexes with the
corrosion byproduct metals and the pipe surfaces5 61 *.
In many cases, equation 2 accurately represents the chemistry of waters
without additional modification, and can be used to compute solid saturation states,
metal speciation for aquatic toxicity studies, treatment chemical dosages, etc.
However, by the most rigorous definition, alkalinity is really the sum of titratable
proton-accepting species in the water, which encompasses more than just the
carbonate and water system. Other "weak" acids (eg. HF°, HOC10, HP04: ) and
"weak" bases (eg. NH3) contribute to alkalinity when present. These species, and
others, can be present naturally or introduced by treatments such as chlorination,
301
-------
ozonation, fluoridation or the addition of corrosion inhibitors. The resulting alkalinity
expression, when these additional species are considered, is illustrated in Table 2. If
a conventional alkalinity titration is done, the acid titrani will be consumed by species
other than C032', HC03', and OH'. When D1C is then computed from the water pH
and the titration alkalinity, it will be overestimated by the amount of the other
constituents present Figure 7 gives diagrams for four important acids and bases that
can complicate the derivation of DIC from alkalinity titration data. The diagrams
show at what approximate pH values the dissociated and reactive species become
significant. The dissociations depend upon both temperature and ionic strength, but
these figures will give reasonable illustrative representations. Depending upon the
total concentrations of these secondary acid or base systems, they may or may not
significantly bias carbonate alkalinity or DIC determinations.
If a reasonably complete water analysis is performed, and the total
concentrations of these other acid-consuming species determined, the equivalent
amount can be subtracted from the alkalinity to enable an accurate computation of
DIC. Methods for the calculation of these alkalinity-consuming species are well-
known, and are already given in many texts and articles1,3 ll M I5. Conceptually, five
steps must be followed to make the corrections.
~ Analyze the total alkalinity to the carbonic acid equivalence point. Convert the
alkalinity to eq/L or meq/L units (divide by 50044.5 for eq/L).
c Analyze the total concentrations of other contributors, eg. free chlorine, ammonia,
total onhophosphate, silica), and convert to mol/L or mmot/L units (to
correspond to the alkalinity units).
~ Use ionization fraction equations or diagrams such as Figure 7 for the appropriate
temperature and ionic strength of the water to distribute the total
concentrations amongst the different species that will exist at pH values above
the carbonic acid equivalence point (eg. HPOa:', P043 , NH3, OCT).
~ Multiply each of the individual species' concentrations by the number of protons
that each would consume during the titration (eg. 1 for NH3, 1 for OC1", 2 for
P043 ), and add them up.
c Subtract the sum from the previous step from the total alkalinity in eq-'L or meq/L
units at the end of the first step to get the DIC concentration. To get DIC in
mg C/L units from eq/L units, multiply by 12011 (atomic weight of C times
103).
In the most rigorous sense, all ion pairs and complexes capable of consuming
protons from an alkalinity titrant are contributors to the total alkalinity. Examples of
these additional species would be: CaHC03 + , CaS04°, MgC03°, Al(OH)3°,
FeHP04°, Cu(OH)3", etc. Table 3 shows part of such a complete expression defining
alkalinity. Because the formation of these various aqueous species results in a
different charge distribution in the water, they do have a possible impact on the
observable alkalinity of the water (and hence, the computation of DIC). For almost
all practical purposes, expressions of greater complexity than that represented in
Table 2 are unlikely to provide any important gain in accuracy. Usually, the most
complicated and rigorous definition expressions for alkalinity are only employed in
equilibrium chemical speciation computer programs, such as WATEQX, WATEQ4F,
PHREEQE, M1NTEQA2 and others414 37.
302
-------
From equation 2, there is an additional property of the carbonate system in
water that complicates the determination of D1C or anything computed using alkalinity
values. Though the endpoint pH for an alkalinity titration has historically been
selected as 4.5, the true equivalence poiru for the titration is more correctly where
HC03' is converted to H;C03" (H^CO,0 + CC^aq))1'14'15. Figure 8 illustrates how
the pH of this point shifts as the concentration of DIC in the water differs for a
system containing only carbonated species and water. For waters with low DIC, the
equivalence point is at a higher pH. For high-DIC waters, the equivalence-point pH
can be significantly less than the operational 4.5 value. In many editions of Standard
Methods10, different color changes have been suggested for indicators to attempt to
conect for this. Other contributing acid or base species will also influence this shift
ore way or another, A significant bias can be produced, especially in "soft" and low-
alkalinity waters by selecting an erroneous position for the endpoint of the alkalinity
titration.
ANALYTICAL PROCEDURES FOR pH AND ALKALINITY
The data used in this paper were mostly generated in two Illinois State Water
Survey (ISWS) projects that include fairly complete discussions of all of the analytical
procedures used8 ~8. Only the most significant points relevant to this report will be
repeated here.
In the well encrustation study, pH measurements were directly obtained from
an electrode mounted in a specially-designed measurement cell, attached in-line from
the bladder pump used for sampling:®. Calibration was done with pH buffers cooled
to the sample water temperature. A 2-point standardization was performed at each
sampling site, using a digital pH meter with independent slope and temperature
compensation.
The pH measuring procedure involved using a rubber stopper around the
electrode to lightly cover the top of the sample bottle. This enabled measurement to
be done with minimal atmospheric interference. Very slow stirring was done to avoid
streaming potential and other stirring-induced pH bias. The general technique has
been described previously30 31. A digital pH meter was used, along with either
buffers prepared directly from NIST* salts, or from commercial high-precision
buffers traceable to NIST materials and calibration.
Alkalinity titrations were performed immediately in the field on duplicate
samples, using a digital buret for dispensing 0.2-N hydrochloric acid. The procedure
followed was generally the Gran plot technique described by Kramer3- Calculations
were done using either a hand-held programmable calculator, or a Lotus 1-2-3®
spreadsheet1
The Gran plot procedure warrants additional discussion, particularly because it
has received little attention in the water treatment and distribution field. While most
ext developments and presentations on it are confusing and intimidating, there are
'National Institute of Standards and Technology, Gaithersburg. MD.
''Lotus Development Corporation, Cambridge, MA.
303
-------
many cases where the procedure is simpler and less time consuming than most
methods of less accuracy. Gran plots work with a linear function in a pH range
beyond (lower pH) the equivalence point of the titration, and they extrapolate back to
locate it. Many functions relating to Gran plots and analyzing a single or mixture of
weak acids or bases, and extrapolate back to get the endpoint. The Gran plot
approach to alkalinity equivalence point determination has several advantages over
other, more-"standard" methods. A major advantage is that it is often faster than
other potentiometric techniques because the data points are not taken at a specific pH
or mL readings. Analysts who try to precisely obtain pH 4.50 endpoints in their
titrations have to slow down and reduce the size of the titrant increments as the
endpoint is approached, which consumes much time. Further, Gran plots determine
the carbonic acid equivalence point better than fixed-pH endpoint techniques, whether
automated or manual, within certain limitations that also plague other procedures" •3:.
Another advantage is that extremely accurate calibration of the pH electrode is not
necessary when using Gran plots. Finally, Gran plot regressions are simple to
program on scientific calculators, and personal computers using software readily
available in popular commercial packages. Actually, Gran plots can even be drawn
adequately by plotting the data on graph paper.
The simplified field method used in the encrustation study discussed in this
paper, is as follows. First, two or three 50-mL duplicate samples are taken. To each
is added 0.5 mL of 1 M NaCl to buffer ionic strength. Then, the samples are titrated
with the standard acid titrant, usually either 0.2 or 0.02 N HC1. Sulfuric acid should
not be used3:, in case the pH after acid addition gets too close to the point where
sulfate ion pairs are formed, along with HS04'. At least 3 stable pH and mL readings
should be taken below a pH of 3.5 or 4 (eg. in the range of pH 3-3.5). Higher
alkalinity samples require going to lower pH values, to overcome the fact that their
equivalence points occur at a lower pH. The best range is usually determined by
trial-and-error to some degree. Blank titrations should also be conducted for the most
accurate results.
After the 3 (or more) stable readings are taken, the Simplified Gran Function,
(F,).
F, = (V, + V,) • 10"pH
should be plotted on the "Y" axis, versus the mL of acid titrant added V, represents
the initial volume (generally 50.5 mL in this study), and V, represents the volume
after addition t to the sample. The final steps are to perform the least-squares linear
regression with the calculator or spreadsheet, and obtain the total alkalinity (TALK)
from extrapolation to get the equivalence point as follows.
V„(JILL) N 50044 .5
TALK*—_ 1 (3)
(V./10)
Figure 9 shows a compressed view of one of the Gran plot titration calculation
worksheets for total alkalinity determinations developed for the well encrustation
study, and that is generally also applicable to laboratory use. The linear regression
calculation and graph is invoked by two macros executed by the user.
Alkalinity titrations for both the pipe loop study and for relevant recent
USEPA investigations were done by titrating with 0.02 N H-S04 to the carbonic acid
304
-------
equivalence point using an automated Metrohm E-636 Titroprocessor®f. The
equivalence point was determined using the preprogrammed internal instrumental
algorithm. Pooled data from 45 alkalinity analyses of known reference samples
yielded 100.8 % recovery for the USEPA instrument and laboratory personnel.
Ironically, many quality control standards for alkalinity have "true" or reference
values given only presuming titration to a fixed-pH (4.5) endpoint. Thus, the
instrument is generally more accurate than the QC reference solutions available to
check its performance.
DIRECT D1C/TIC ANALYSIS BY COULOMETR1C TITRATION
DIC samples were taken by in-line filtration through a 0.4 /im 47 mm
diameter polycarbonate membrane filter in a Teflon® holder. The DIC samples were
directly fed through the filters into 20-mL plastic syringes. TIC samples were also
taken, which were done the same way except in bypassing the filters.
In the corrosion pipe loop study8, TIC samples were taken by filling 25-mL
glass bottles having caps with conical polyethylene liners from below the surface, and
quickly capping them. pH samples were taken in similar bottles, and analyzed as
soon as all samples were collected.
In the ISWS studies, TIC and DIC were analyzed using a Coulometrics,
Incorporated" Model 5010 CO: Coulometer with model 5030 Carbonate Carbon
system. The procedure was similar to that described by ASTM9. Aqueous samples
in their syringes were weighed before and after injection into the instrument, to obtain
accurate sample volumes for concentration calculation. The total number of
micrograms of evolved carbon from the samples was divided by the sample volume
(derived from weights) using the appropriate conversion factors to get the
concentrations of DIC or TIC. Samples were analyzed no later than 1 day after
sample collection, immediately after their return to the laboratory in coolers
containing ice.
The principles of the coulometric analysis technique embodied in commercial
instrumentation has been described by several researchers9 24 :5'3\ and will only be
briefly summarized here.
The main reaction cell of the instrument consists of a platinum electrode to
generate OH ions as a cathode, and a silver electrode as the anode. The main
solution in the cell is a proprietary mixture containing ethanolamine as the trapping
agent for CO; gas liberated from the sample. The ethanolamine is in a dimethyl
sulfoxide substrate, that also has some thymolphthalein pH indicator present. The
thymolphthalein absorbs most strongly at 612 nm. The cell is essentially held in a
beaker in the path of a fine beam of white light.
The aqueous sample is injected into a reaction cell (that can be heated) that has
been purged with a C02-free dry gas, such as processed air or N;. A 45% KOH
"Brinkmann Instruments, Westbury, New York.
"Wheat Ridge, Colorado. Now, UIC, Inc., Joliet, Illinois.
305
-------
scrubber is normally used to keep the gas CO;-free. Acid is added to the sample by a
repipet. and the carrier gas strips the evolved CO2 gas from the acidified sample
solution. All studies reported in this paper used 20% HCI04 for sample acidification.
The CO; passes through a scrubber or series of scrubbers (frequently solutions
of pH 3 saturated with Ag:S04, and with HXK added) to remove potentially-
interfering acidic gases, such as H2S or HC1. After passing through the final
scrubbers, the C02 gas becomes trapped in the coulometer cell. This lowers the cell
solution pH, which in tum decreases the absorbance by causing the color of the
thymolphthalein (initially a medium blue) to fade. This triggers the generation of
current into the Pt electrode, to produce hydroxyl ions This continues until the
initial absorbance of the light beam is restored. Because both current and time can be
measured very accurately and precisely, the amount of DIC in the sample can be
determined to a high degree of certainty by relation to the micrograms of CCX
evolved from the sample and trapped in the cell.
The accuracy of the coulometric procedure is essentially the same, whether
aqueous or solid samples are analyzed. Differences arise mainly from sample
handling biases. In two different studies documented on a similar instrument to that
of ISWS in the USEPA laboratory yielded 99% recovery for 31 determinations on
N1ST CaC03 standards, and 100% recovery for 16 determinations on other CaC03
standard reference materials (including N1ST ones). Duplicate and triplicate analyses
of TIC and D1C ground water samples dunng the course of this reported study
yielded pooled standard deviations of 0.37 mg C'L and 0.51 mg C/L, respectively-8.
These values correspond to relative standard deviations on the order of + 0.2 to 0.5
%.
Generally, the coulometric instruments do not need calibration curves to be
run, as with spectrophotometers, for example. The reason is because the coulometers
work in very fundamental properties (current, time), and there is a broad linear
operational range. In fact, because of CO; gas exchange problems, the preparation of
accurate aqueous DIC samples over a wide (but practical) concentration range is
highly problematic.
The sample collection process, injection technique, and selection of container
for DIC/'TIC samples are all very important. The ISWS work indicated that once
laken, samples could be held in refrigerated syringes for only 1-2 days before
statistically-significant low bias was observed, resulting from CO- gas loss from the
ground water samples. Dunng the pipe loop study, several sampling protocols and
containers were investigated. These tests were repeated on Oakwood. Ohio tap water
in the USEPA laboratory, using 4 types of containers: 1. Conventional 250-mL high-
density linear polyethylene bottle (HDPE), 2. Plastic 20-mL syringe; 3. Glass 25-
mL vial with Teflon'-coated septum; and, 4. Glass 25-mL vial having cap wish
conical polyethylene liner. Previous studies indicated that Oakwood tap water (TALK
approximately 370-390 mg CaCOj/L) was prone to exsolve carbon dioxide into the
air, and would therefore make a viable test case. All containers were filled quickly
from a single transported and sealed container containing a homogenous batch of the
water to be tested. Previously-unused triplicate individual samples were analyzed
immediately, and after being refrigerated for 1,4 and 11 days.
306
-------
The different containers required different procedures to get the samples
injected into the coulometer instrument reaction cell. Samples were withdrawn from
the LPE bottles with the same kind of plastic syringe that was used in the field
sampling. With the vials having the Teflon*-coated septa, samples were withdrawn
by syringe, with a second needle used to vent the bottle and equalize pressure to
reduce vacuum errors. The syringes were simply used as-was, for direct injection.
To minimize air contact in the other set of vials (Poly-seal® caps), the injection
needles were not put on the syringes. The bottles were opened, and the heads of the
syringes were put into the necks of the bottles. The bottles were then inverted, and
the water quickly drawn into the syringe. Air was let in around the neck and sides of
the bottles, but there was little mixing with the water flowing into the syringe.
Figure 10 shows the results of this stability study. The calcium carbonate
reference standard data (as % recovery, not mg C/L) is also shown for comparison.
There was no statistically-significant drift in the standards from day to day.
The syringes continually lost DIC from the start. The septum-capped vials
generally performed well over all 11 days, but there was one random failure. Similar
problems were observed with the same kind of caps in earlier work on pH stability3-.
The HDPE bottles gave consistent results, but showed a low bias relative to both
types of glass vials.
The best results seemed to come from the glass vials with the Poly-seal® caps.
They showed excellent stability for 4 days, and possibly only a minor (< 1%) loss
after 11 days. They were also the easiest to manipulate for sample extraction and
analysis. Normally, samples are analyzed within a day of receipt, so a wide safety
margin is provided by this type of sample container.
EXPERIMENTAL DATA ANALYSIS
The comparability of the computational procedure to direct analysis was tested
on 31 samples from the encrustation study, and 68 samples from the pipe loop study.
Complete water analyses were initially checked for internal consistency, and for gross
errors detectable through ion balance error calculations. These are very useful, even
if not always unambiguous34. Calculations were made using the complete water
analytical data input into the WATEQX aqueous chemistry modeling program27, run
under Microsoft FORTRAN 5f on two IBM-compatible 486/25 MHz-processor
personal computers. Complete aqueous speciation, major mineral saturation indices,
and ion balance errors were computed. Two runs were made. The first sample set
used total alkalinity input, and was corrected for all possible alkalinity-producing
species present in the analyses except OC1'. The second sample set used the directly
analyzed DIC concentrations instead of total alkalinities.
The ion balance errors for the field study had a mean of -1.5 % for both DIC
and total alkalinity input, with standard deviations of 1.8 % and 2.5 % respectively,
for alkalinity and DIC input. For the pipe loop data, the mean ion balance error
was -3.3 % with a standard deviation of 2.9 % for total alkalinity input. With DIC
Microsoft Corporaton, Redmond, WA.
1
307
-------
input, the mean was -3.6 % and the standard deviation was 3.0 %. A least-squares
linear regression fit using the TableCurve*t software package was made for the
computed D1C (DIC^ concentration versus that directly analyzed (DICM). The data
are shown in Figure II. The equation and correlation coefficient (r) for the fit of
the line are:
DlCc = 1.0032 DICm - 0.3431
r2 = 0.991
The uncertainty in the slope of the regression line that was statistically determined in
the curve fitting procedure indicated that the true slope lines in the range of 0.99 to
1.01 with 95 % confidence. This range indicates no observable difference in
measured and calculated D1C. Therefore, under both field and laboratory conditions,
excellent agreement can be found between directly analyzed D1C values and those
computed from total alkalinity, coupled with complete major chemical constituent data
and the use of rigorous computational procedures. Consequently, similarly excellent
agreement can be found between saturation indices or other computed water chemistry
parameters relying on DIC and pH.
CONTROLLING SOURCES OF ERROR IN PRACTICE
A detailed sensitivity examination of the alkalinity equation and solubility
expressions (not included here) indicates that the computation of DIC is much more
sensitive to imprecision or bias in pH than in total alkalinity. Therefore, careful
attention to pH calibration and analysis methodology will considerably improve the
reliability of DIC estimation from alkalinity titrations. Figures 12, 13, and 14 present
some examples of the precision thai can be expected in pH and alkalinity analysis.
The data is taken from reports of results from the U. S. Geological Survey Standard
Reference Water Sample Program, plus precision calculations from this study. The
equivalence point method using the Turoprocessor® (Figure 13, last two entries)
attained almost ten-fold lower standard deviations compared to the pooled data from
the other laboratories. The precision of the DWRD and 1SWS samples was relatively
independent of concentration. Statistically-processed "most probable value"
concentrations are given for reference. The concentrations represented for the ISWS
and DWRD data are approximations of the median alkalinities of the waters involved
in the study. Colorimetric alkalinities (Figure 12) yielded similar data to the
potentiometric alkalinities, with a tendency towards higher standard deviations. The
precision of the ISWS pH analyses are also nearly a factor of 10 better than the
reference data (Figure 14), which substantially contributed to the excellent
comparability noted m this study.
These investigations demonstrate that in a research situation, under true
equilibrium conditions, virtually equivalent results can be obtained for directly
analyzing or computing DIC. However, the calculations are much more complicated
than would be desired by many utilities, consultants or laboratories for routine use in
such applications as chemical feed rate calculations to achieve a certain pH or
Langelier Index value. Reliability of DIC determination by computation can be
substantially improved by some simple analytical procedure adjustments, analyses of
Handel Scientific, Cone Madera, CA.
30B
-------
secondary proton-consuming constituents to provide necessary analytical corrections,
and using good computational techniques with up-to-date equilibrium constants for the
calculations.
The greatest limitation to the accurate determination of DIC by pH and
alkalinity is the state of equilibrium of the system. If the system is not at
equilibrium, the constants used in the chemical equations can not be validly applied.
Therefore, in cases where process control troubleshooting must be done, direct
determination of D1C/TIC is preferable.
When a laboratory is also interested in solids analysis, instrumentation for
directly analyzing TIC becomes much more attractive as well. Coulometnc CO-,
analyses has been well-documented for straightforward application to soils, sediments,
scales, rocks and minerals35-36'37.
ACKNOWLEDGEMENTS
The field and pipe loop corrosion studies cited in this report were conducted
by the Aquatic Chemistry Section, Illinois State Water Survey, Champaign, Illinois.
Funding was provided by the Illinois Department of Transportation, Springfield,
Illinois, and the American Water Works Association Research Foundation, Denver,
Colorado. In addition to the authors, analytical credits belong to Sarah H. Smothers,
Edward Garske, Osia Smith and Joseph Kamy of the Illinois State Water Survey. Dr.
Thomas R. Holm, also of ISWS, provided assistance with the development of the
Lotus 1-2-3® spreadsheet for the Gran plot alkalinity determinations.
DISCLAIMER
The names of all manufacturers, software developers, and vendors identified in
this paper are given for explanatory purposes, and does not constitute an endorsement
by the United States Government or by Technology Applications, Incorporated.
REFERENCES
1. Snoeyink, V.L. & Jenkins, David. Water Chemistry, John Wiley & Sons,
New York (1980).
2. Loewenthal, RE. & Marais, G.V. Carbonate Chemistry of Aquatic Systems;
Theory & Application, Ann Arbor Science, (1976).
3. Butler, J.N. Carbon Dioxide Equilibria and Their Applications. Addison-
Wesley publishing Co., Reading MA (1982).
4. AWWA Joint Task Group on Calcium Carbonate Saturation. Suggested
Methods for Calculating and Interpreting Calcium Carbonate Saturation
Indices. Jour. A WWa,82:7:71 (1990).
5. Schock, M R. Response of Lead Solubility to Dissolved Carbonate in Drinking
Water. Jour. AWWA, 72:12:695 (1980).
>
309
-------
6. Schock, M.R. & Gardels, M.C. Plumbosolvency Reductions by High pH and
Low Carbonate - Solubility Relationships. Jour. AWWA, 72:2:87 (1983).
7. Schock, M.R. & Wagner, I. The Corrosion and Solubility of Lead in Drinking
Water, Ch. 4 in Interna! Corrosion of Water Distribution Systems,
AWWARF/DVGW Forschunggtelle (1985).
8. Lead Control Strategies. American Water Works Association Research
Foundation, Denver, CO (1989).
9. 1983 Annual Book of ASTM Standards. Standard D513-82, American Society
for Testing and Materials, Vol. 11.01 Philadelphia, PA (1983).
10. Standard Methods for the Examination of Water and Wastewater. 17th Edition,
AWWA-WPCF-APHA, (1989).
1). Loewenthal, R.E., Ekama, G.A., & Marais, G.V.R. Mixed Weak Acid/Base
Systems; Part 1 - Mixture Characterization. Water SA, 15:1:3 (1989).
12. Schock. M.R. Internal Corrosion Deposition Control. Ch. 17 in Water Quality
and Treament, 4th Edition, AWWA/McGraw-Hill (1990).
13. Noll, C.A. & Polsky, J.W. Determination of Carbon Dioxide in Water by
Conductivity Measurements, TAPPl, 56:39:1 (1956).
14. Faust, S B. & Aly, O.M. Chemistry of Natural Waters, Ann Arbor Science
(1981).
15. Stumm, W. & Morgan J.J. Aquatic Chemistry, 2nd Edition, Wiley-
lnterscience, New York (1981).
16. Kramer, James R. Alkalinity arid Acidity. Ch. 3 in Water Analysis, Vol. 1,
Inorganic Species, Part 1, R.A. Minear and L.H. Keith editors, Academic
Press, New York (1982).
17. Salonen, Kalevi. Rapid and Precise Determination of Total Inorganic Carbon
and Some Gases in Aqueous Solutions. Water Research, 15:403 (1981).
18. Scarano, E & Calcagno, C. High Sensitivity Carbon Dioxide Analyses. Anal
Chem. 47:7:1055 (1975).
19. Kanamori, Satoru Shipboard Calibration of an Infrared Absorption Gas
Analyzer for Total Carbon Dioxide Determination in Sea Water. Jou,.
Oceanog. Snc. Japan, 38:131 (1982).
20. Riano, M.D., et al. Determination of Total Carbonate by Ligand Exchange.
Analyst, 115:975 (July 1990).
21. Crowiher. Joan & Moody, W.B. Automatic Determination of Inorganic
Carbon in Surface Waters. Anal Chim. Acta, 120:305 (1980).
310
-------
22. Amundson, Ronald G., et al. A Rapid Method of Soil Carbonate AnaJysis
Using Gas Chromatography. Soil Sci. Soc. Am. J 52:880 (1988).
23. Pigott, John D. Coupled Ion-Selective Electrode Measurement of Aqueous
Carbonate and Bicarbonate Ion Activities. Anal. Chem. 61:638 (1989).
24. Johnson, K. M., et al. Coulometnc TCO; Analyses for Marine Studies;
An Introduction. Marine Chem. 16:61 (1985).
25. Johnson, K. M., et al. Coulometnc Total Carbon Dioxide Analysis for Marine
Studies: Automation and Calibration. Marine Chem. 21:117 (1987).
26. Homer, S.M.J. & Smith, D.F. Measurement of Total Inorganic Carbon in
Seawater by a Substiochiometric Assay Using NaHl4C03. Limnol <4 Oceanog
27:5:978 (1982).
27. vanGaans, P.F.M. WATEQX - A Restructured. Generalized, and Extended
FORTRAN 77 Computer Code and Database Format for the WaTEQ
Aqueous Chemical Model for Element Speciation and Mineral Saturation, for
Use on Personal Computers or Mainframes. Computers & Geosci. 15:6:843
(1989).
28. Sanderson, E.W., et al. Dewatering Well Assessment for the Highway
Drainage System at Four Sites in the East St. Louis Area, Illinois (Phase 2),
Plus Appendices. Ilinois State Water Survey Contract Report 424 (March
1987).
29. Garske. E.E. & Schock, M.R. An Inexpensive Flow-Through Cell and
Measurement System for Monitoring Selected Chemical Parameters in Ground
Water. Ground Water Monit. Rev, 6:79 (1986).
30. Kramer, J.R. Precise Determination of Low Alkalinities Using the Modified
Gran AnaJysis, An Inexpensive Field Procedure. McMaster University
Department of Geology Environmental Geochemistry Report. (March 1982).
31. Schock, M.R., et al. Laboratory Techniques for Measurement of pH for
Corrosion Control Studies and Water not in Equilibrium with the Atmosphere.
Jour. A WW A 72:5:304 (1980).
32. Schock, M.R. and Schock, S.C. Effect of Container Type on pH and
Alkalinity Stability. Water Res. 16:1455 (1982).
33. Huffman, Edward W.D. Jr. Performance of a New Automatic Carbon Dioxide
Coulometer. Microchem. Jour. 22:567 (1977).
34. Merino, Enrique. Internal Consistency of a Water Analysis and Uncertainty of
the Calculated Distribution of Aqueous Species at 25°C. Geochim. Cosmochim.
Acta, 43:1533 (1979).
35. Engleman, E.E., et al. Determination of Carbonate Carbon in Geological
Materials by Coulometric Titration. Chem. Geol. 53:125 (1985).
311
-------
36. Chan, Chris C.Y. Determination of Carbonate Carbon in 41 Internationa]
Geochemical Reference Samples by Coulometric Method Geostandards
Newslett. 12:1-.39 (1988).
37. Cahill, R.A. & Autrey A.D. Total and Inorganic Carbon Content of Eighteen
National Bureau of Standards and Four Canadian Certified Reference
Materials. Geostandards Newsleit. 12:1:39 (1988).
312
-------
TABLE 1
Brief Listing of Representative References for the
Determination of Dissolved Inorganic Carbonate in Water
Method Principle Example References
Computed using total alkalinity, pH, 6, 9, 10, 11, 12, 13, 14, IS, 16
temperature and ionic strength from
conductimetric or potentiomeiric
titrations
Infra-red gas analyzer 17, 18, 19
Indirect colorimetric, potentiometric 20,21
Gas chromatography 22
lon-selective electrode (COj) 23
Coulomeuic titration 9, 24, 25
Miscellaneous Isotopic 26
TABLE 2
Alkalinity Expression More Typical of a Treated
Drinking Water than Equation 2
Titration Alkalinity
Simplified for Drinking Water
TALK = 2 [CO,2-] + [HCO,-] + [OH ] +
+ [OCI] + [NH,] + [SiO(OH),'] +
2 [P04a] + [HP04»] - [H+]
313
-------
TABLE 3
Complete Alkalinity Expression for Potable Waters,
Including Ion Pairs, Complexes, and Other Proton Acceptors
Titration Alkalinity
Complete, eg. WATEQX
TALK = { 2 ((CO,1] + [CaCO,°] + [MgCO,0l +
+ [NaCO,] +" [PbCO,0] + (Al(OH)/] + ...) +
([HCO,"] + [HPO/ l + [OCL'l + [CaHCO, ] +
(SiO(OH),-] + [NHJ + [OH] + [F] + ...) +
[H*J + [HS04] + [HF°] + ... >
FIGURE 1
Effect of D1C on Solubility of Calcite (CaC03)
314
-------
FIGURE 2
Distribution Diagram Showing the Change in Predominance
of Dissolved Lead (II) Species with Changing pH
in the Presence of 3 mg C/L DIC
LEAD SPECiiON cOR 25 C. 1=0 01
D l C= 3 mg/L
-------
FIGURE 4
Effect of DIC on Zinc Solubility at Different
pH Levels
Zinc Solubility
I = 0.01, 25°C
(Mryiftic
Cartoon
' *1 mg/L
~fl mg/L I
~ 20 mg/L 1
* *W mg/l
FIGURE 5
Illustration of Changes in Total Alkaltniiies as
pH Changes for Three Waters with Constant DIC's
Relationship of DIC to Alkalinity
16° C, 1-0.01
o.oot
O-
¦ MO
U
I
f"°
S
100
jc
DIC -60
t
xa&s&ss
f
/
W'
/ >
..1. - -. J\
DIC ¦= 30
: I
/-I
0 006
o
o.oo4 n
o oos —
>
/ ¦
- - -f/-•
.V
»
0.001
i • 7 ¦ • 10 11 II
pH
o.ooo
-------
FIGURE 6
Graphical Calculation of D1C from pH and Tola! Alkalinity
Data Using the Linear Equation Transformation.
(pH 10 is the top line; pH 6 the bottom line)
Alkallnlty/DIC Relationship
25-C, 1 = 0.005
mq C/. Inofgontc Coffren
317
-------
FIGURE 7
Distribution Diagrams for Four Weak Acid and Weak Base Systems
Commonly Encountered in Drinking Water Systems, that Can Interfere
With the Calculation of DIC from pH and Alkalinity Data
DtRrtbutlon Diagram for Carbonic Add
i • rov a* e
PlWfarton Dhgnm for HypocNorou* Acid
la 0 01V C
HOD8
PH
ot^rw* for Onhophospheht Add Distribution Diagram for Ammonia
c i - o.oi. »• c
N
J
\ . .1.. .1.. J.
I
\ i : /
hw!
y : 7
! POJ
-L -t r
|'
-vi il
t: 1
... ..
¦¦ 1 1
h
. ../l 1: . .
i\ j I /i
1
1 \ M 1 \
r v
• \ 1
/ .A
S 1 1 •
t M ) ( t • M i <0 n mj
318
-------
FIGURE 8
Relationship of Equivalence Point pH to the DIC Concentration
in a Pure Water/Carbonate System at 10° C and 1=0.005
Alkalinity Equivalence Point pH
10°C, 1=0.005
».oo:
5.50
Z
a
8.00
4.50-
4.00
100
25
50
125
150
mg C/L DiC
-------
FIGURE 9
Example of Computer Spreadsheet Used to Determine the Total Alkalinity
Equivalence Point by a Gran plot Method
ORAM PLOT TITRATION WORKSHEET
Number
Well Number
5*mp* Vokjme (mi)
NaCt (mi)
IthaI vokjme
Ttrarit NomaMy
V 0
N
60 00 Enapotitmi
O.W mQC*CQ3/l TAlfi
60.05
0.2000 ft-tqutna
JL fttL
LL.
VCOR-F1
4.10 3 10
004X1
0.04301
4 16 3 00
006421
0.05361
4.20 2.96
0 OS 946
0 06066
4.24 2.9
0.06*35
0-06775
Rtgrestion Output:
Constant
-0 68^4013
$10En ol YEst
000103611
R Squared
0 99361343
No ct Ooaervcions
4
Degrees of Freedom
2
X CoefTcient(s)
017660616*
Sid Eft of Coef
0-010016422
3 657
771.23
C
.2 !
£ 0 06 -
u
u
CC 0 05! -
Ll !
0 05 -
I
0CM? -
< 05 4 J . « l« < 2 4 25
Volume
320
-------
FIGURE 10
Results of Container Evaluation for DIC Sampling and Preservation
o
B
"tf o
0
s
*
c«co,
o
LPE BotUa
0
Glttt/PSC
A
Syringe
•
GlMt/TS
O
_i i l_
Bapsed Days
FIGURE 11
Comparison of Computed DIC from Alldinity Titration and
WATEQX Equation (DIC<-) with DIC Directly Analyzed by
Coulometric Titration (DICM)
_ 200
Q)
+-»
| 150
o
o
O 100
o
O 50
0)
E 0
0 50 100 150 200
mg C/L DIC measured
321
I . I
-------
FIGURE 12
Summary of Precision of Colorimetric Alkalinity Titrations
Reported in U. S. Geological Survey SRWS Certification Studies.
Reference values are mg CaC03/L.
Performance of Alkalinity Titration
Colorimstric Method
260f ' ¦
<200-
l»j- ; •
' * + <~ *
60- # * ~
J
10 l
5 |
; 3 5 3 1 1
tatowi.' | til i«« a ' a | m ' in in > ib : « < m • . itq
me Om +i o 4 i 18j •« • iij : u ! 4.4 ' i y ' u ' u ! a
8RU
FIGURE 13
Summary of the Precision of Potentiometric Alkalinity Titrations
from USGS SRWS Certification Program, along with Data from 1SWS
and USEPA (this Study). Reference values are mg CaC03/L.
Performance of Alkalinity Titration
Electrometric Method
000 (20 .
660 - 1
600- • {
430- -15'
far -j ¦
I 300- 10 '
1 250 r 4 -t II
160-. . -1- -|5 ,
100- • « + . * ' * | '
M" " + * 1 J I
0 I * i i | I jTTTTn s g 0
] 3 I £
i Pw» «¦ j».r ik li i |a.t >? n n ' * it >: u 1» < ioa u
saw
322
-------
FIGURE 14
Summary of Precision for pH Determinations from USGS
SRWS Certifications Studies, and ISWS Field Data (this Study)
Performance of pH Measurements
f !
3 7-
i '
-0.8 *
5 —
t« i i § i ! r
¦ » * 2 i * i 5
-0.2
I
• • js r« , ip irt ijt nr u* iu ii ij
- on ojt oi eit ' en o— or» ojt eaa
9RM
-------
TECHNfCAL REPORT DATA
(Please read Instructtonf pn the revenc before comple
1 REPORT NO. 2.
EPA/600/A-92/095
3
4. TITLE and subtitle
COMPARISON OF METHODS FOR DETERMINATION
OF DISSOLVED INORGANIC CARBON,"..'
5 REPORT DATE
12/16/91
6 PERFORMING ORGANIZATION CODE
7 AUTHOR(S)
MICHAEL R. SCHOCK*
GREGORY K. GFORGE ^
8. performing organization report NO.
9 PERFORMING ORGANIZATION NAME AND ADDRESS
''"Drinking Water Research Division, USEPA
Cincinnati, 0 45268
2
Technology Applications, Inc., Cinti, OH
10 PROGRAM ELEMENT NO.
1 1 . CONTRACT/GRANT NO.
12. SPONSORING AGENCY NAME AND ADDRESS
Risk Reduction' Engineering Laboratory--Cin., OH
Office of Rese'arch and Development
U.S. Environmental Protection Agency
Cincinnati, Ohio 45268
13. TVPE OF REPORT AND PERIOD COVERED
Piihi i chpH Pflpaf*
14. SPONSORING AGENCY C?ODE
EPA/600/14
15. SUPPLEMENTARY NOTES
PROCEEDINGS, AWWA WATER QUALITY TECHNOLOGY CONFERENCE, ORLANDO, FL.
NOVEMBER 10-14, 1991 p :299-323
16 ABSTRACT
& THE /P>E:s:ENTAT.E0'N S£v;EW ,A-P.P-R0'A«S'-.F,OR' BB?£RIWf'EN6 /D',KS"-S'0L>V-E'D x-I'N OR GAWKS'
.C'A'RBON' (DIC) J'N' „WA'T£R'. EXPZkwmAK/Stm'teXMimRWMFMbORW? AND'
J&mrWJbF DIC .0,E-T:E-RMTN'A^I'.0N ,0B7A$N£B M A'JZm&m&mZ
WATEQX ptyjxjUmW! JWVm'W- . THEVlWB^ME^ODS'^E^EQ.UJWA'bBi?
DIC 'V'/feUES'/ AT^THE 95% 'jtqWOBmjUEMXX. HEWE'AfBR, pH ,AN0 fflXM
,MU5-T^BI'.0:B'T>I^'I'2BD MMtmWL , -aR'.BL'S'.E^b;iiRE'&JvAN'Al^Y:S-tS A#
/Mitfffl. FMxm Jim £mM .go n-t.ro i^-ed , jm .mmzwiw. jvssnm jMwsswaqsjguer
xI0^E'N:A;B>E X50&& DKLj;ECERMiNA«0N/F.0;R^lrAB0R'Af,0RsI'E;S<
17 KEY WORDS AND DOCUMENT ANALYSIS
i. DESCRIPTORS
b IDENTIFIERS.OPEN ENDED TERMS
COSati Field/Group
Water Analysis
Water Treatment
Drinking Water,
Analytical Method,
Dissolved Inorganic
Carbonate,
Corrosion Control ,
A1kalinity Analysi s
18. DISTRIBUTION STATEMENT
RELEASE TO PUBLIC
19 SECURITY CLASS : This Hfporn
UNCLASSIFTFD
21. NO OF PAGES
24
20 5ECUBITY CLASS /Diiipiifc.
UNCLASSIFIED ""
22 PRJCE
EPA Form 2220—1 (Rav. 4—77) previous edit.on is jbsolete
------- |