You are on page 1of 5

Journal of Chromatographic Science, Vol.

27, January 1989

Determining Chlorine Concentrations in Air and


Water Samples for Scrubbing Studies Using
Ion Chromatography
William C. Askew a n d Stephen J. Morisani
D e p a r t m e n t of C h e m i c a l E n g i n e e r i n g , U n i v e r s i t y of S o u t h A l a b a m a , M o b i l e , A l a b a m a 3 6 6 8 8

and a hypochlorite ion to an extent dependent on the amount


Abstract of chlorine dissolved:

An ion chromatographic method is developed to measure Eq. 2


chlorine concentrations in air and water samples for As equation 1 indicates, the p H of a chlorine/water solution
scrubbing studies in packed towers. The method uses a
at equilibrium is a measure of the total chlorine in solution that
sodium sulfite solution as a trapping agent for converting
chlorine to chloride ion in both air and water samples. The is hydrolyzed in the absence of external p H control and of ap­
method also features short analysis time and long shelf life preciable H O C l ionization. For example, at p H of 3 and 25°C
– 4
for collected samples awaiting analysis, but is subject to ( K = 3 . 9 4 4 × 1 0 ( 9 ) ) , more than 99% of the chlorine in solu­
1

chloride peak interference. tion is hydrolyzed, whereas at a p H of 1, only 4% of the chlorine


in solution is hydrolyzed. Ionization of H O C l in water at low
p H is not significant. Only 0.01% of the total H O C l in solution
– 8
is ionized when the p H is 3 ( K = 2 . 8 9 8 × 1 0
2 at 2 5 ° C ( 1 0 ) ) .
Introduction Moreover, for a chlorine/water solution with a p H of 6 at 25°C,
the percent H O C l in solution that is ionized increases only to
Chlorine (Cl ) and chlorine dioxide (ClO ) are important
2 2 about 3%. In this work, the p H of chlorine/water samples to
chemicals used in the bleaching operations of many paper mills. be analyzed was usually below 3. Therefore, the percent of the
Because of their hazardous nature, waste gas streams containing C l hydrolyzed in such samples might be significant, whereas
2

C l and C l O must be treated before discharge to the atmosphere.


2 2 the percent of the H O C l ionized would be small.
This treatment must provide a safe environment for workers, Although a number of methods for determining chlorine were
consistent with government regulations. Reports of C l and C l O 2 2 under consideration, one using ion chromatography (IC) was
scrubbing operations in industry indicate a need for continued considered to have the greatest potential because chloride ions
research in this area (1–4). -
( C l ) in solution are quite stable and are easily measured by
Recently, research was initiated at the University of South IC (11). In order to use I C , therefore, an appropriate trapping
Alabama to study the scrubbing of C l and C l O . Water scrub­
2 2

agent for converting chlorine to CI was needed for both water
bing of C l air streams in packed towers to study the effectiveness
2 and air samples. A search for possible trapping agents showed
of different packings was the starting point for the research. that a number of reducing agents could be used for converting
Feed gas compositions below 1000 ppm C l were used in this 2

chlorine to CI (12). Examples of such agents are:
study so results could be compared to published work. A s part
of this work, a method was needed to analyze air and water Eq. 3
samples for C l . Along with accuracy, precision and simplicity,
2 Eq. 4
it was desirable that the method be essentially the same for both
phases and more importantly that it allow samples to be taken Eq. 5
but not immediately analyzed. A long shelf life for samples Other possible trapping agents are sulfur dioxide, sodium bi­
would allow a greater number of operating conditions to be sulfite, sulfamic acid, and sodium hydroxide.
tested each day. A review of the literature indicated that no such
method was available (5,6).
The chemistry of chlorine dissolving in water is not simply
Experimental
a case of physical absorption. U p o n dissolving in pure water,
molecular chlorine hydrolyzes quickly to establish equilibrium Reagents
with hydrochloric and hypochlorous acids (7,8): Reagent-grade chemicals were used in this work. Freshly
Eq. 1 distilled water was used for solution preparation and dilutions.
The HCl standards were prepared from 0.1 Ν HCl acculute
In addition, hypochlorous acid can ionize into a hydrogen ion (Anachemia).

42 Reproduction (photocopying) of editorial content of this journal is prohibited without publisher's permission.
Journal of Chromatographic Science, Vol. 27, January 1989

Chlorine trapping were recorded. With these data, the moles of air sampled were
A suitable trapping agent for chlorine in air or water samples calculated, assuming the air to be an ideal gas saturated with

must convert all of the chlorine to Cl for quantitative analysis water vapor.
by I C . The agent must neither produce nor contain ions that When not sampling air, the bottle portion of the impinger
have long elution times nor produce peaks that interfere with was removed, and a tygon tube connected the impinger nozzle
– –
the Cl peak. Although KI converts all of the chlorine to C l , to a caustic bubbler. This allowed a continuous flow of air

it was rejected because the excess I had a peak elution time through the system to eliminate wall effects on the chlorine.
of 18 min. Sodium thiosulfate was rejected also because the ex­ Prior to sampling, the air flow was momentarily stopped, the
2 –
cess S O had a peak elution time of 21 min. Sulfamic acid
2 3 tygon tube was removed from the nozzle, and the bottle portion
was unsuccessful because its elution time interfered with the of the impinger containing 25 m L of trapping solution was

Cl peak. Sulfur dioxide was not tried because of the difficulty installed.
in using a gaseous reagent. Sodium hydroxide was rejected be­ Water was sampled for chlorine in one of two ways. More

cause it produces more than Cl from its reaction with chlorine. concentrated solutions were sampled by placing approximately
Sodium sulfite worked well as a trapping agent. Sulfite rapidly 5 m L of trapping solution, typically 5 to 15 drops of 1 Μ N a S O 2 3


reduced chlorine quantitatively to Cl under either acidic or trapping agent in 5 m L of distilled water, in a 100–mL volumetric
basic conditions. The mechanism is probably p H dependent. flask. Five to ten grams of water sample were added by weight
In this work, the p H varied from very acidic conditions when and the flask was sealed. The used trapping solution was then
trapping chlorine in water samples to basic conditions for air diluted to 100 m L for later analysis by I C . In weaker solutions,
samples bubbled through sulfite solutions. Standard oxidation- usually the case in scrubbing operations, chlorine was trapped
reduction potentials indicated the likelihood that sulfite, besides by simply adding a few drops of 1 Μ N a S O trapping agent
2 3


reducing C l to C l , could reduce either H O C l under acidic
2 to a minimum 200–mL sample.

conditions or O C 1 " under basic conditions to Cl (13,14). In
addition to discussions of a general nature on chlorine and sulfite Analysis by IC
chemistry (15-17), other evidence indicating the ability of sulfite –
Trapped samples were analyzed for Cl on a Model 2000
to reduce chlorine through its hydrolysis products in acidic (18) i / S P ion chromatograph (Dionex) with a Model SP4200 com­
and basic (19) environments has been reported, although the puting integrator (Spectra-Physics) to record elution times and
mechanisms were not certain. The present work was consistent response areas. The column was an H P I C - A S 4 A anion separa­
with these findings. tor (Dionex) with an H P I C - A G 4 A guard column (Dionex).

Furthermore, the Cl produced by trapping with sodium Eluent (0.00075 Μ N a H C O , and 0.0022 Μ N a C O ) was 2 3

sulfite was easily measured by I C , and all ions had peak elution pumped through the column at 2 m L / m i n . The regenerant for
times of less than 6 min. The produced HCl was very stable. the ASF–1 suppressor cell (Dionex) was 0.025 Μ H S O at 2 2 4

Although sodium bisulfite was not tried as a trapping agent, m L / m i n . The conductivity cell was operated at 100ΜSfor a 1–V
it would probably work similarly to N a S O . A l l chlorine trap­
2 3 signal to the computing integrator. The integrator attenuation
ping solutions were freshly prepared from 1 Μ N a S O trapping 2 3 was 1024 m V for full scale deflection, and the peak threshold
agent to provide excess reagent for the stoichiometric conver­ value was 2000. A stable base line was obtained within 1 h of
sion of chlorine to H C l . startup of the equipment. The sample size was 50 μL injected
by glass syringe.
Sampling The N a S O reagent used in preparing trapping solutions con-
2 3

A i r was sampled for chlorine by passing 1 L through a midget


impinger containing 25 m L of trapping solution, as shown in
Figure 1. A l l air samples were taken from sources above at­
mospheric pressure, which provided the approximately 15 i n .
of H O pressure (gauge) needed to drive the air sampling system.
2

A typical trapping solution for air sampling was 6 drops of 1


Μ N a S O trapping agent added to 25 m L of distilled water
2 3

giving a N a S O concentration of about 0.012 Μ and a p H of


2 3

about 8.7 in the midget impinger. Sample volume was controlled


by displacing 1 L of water from an intermediate vessel, initially
filled to overflowing, into an empty 1–L volumetric flask. When
air flow in the sampling system was stopped, the vacuum breaker
on the intermediate vessel eliminated suction on the water in
that vessel, which prevented overfilling the 1–L flask. A single
impinger was sufficient because a second impinger with trapping
solution in series showed no chlorine breakthrough upon analy­
sis. The rate of air sampling, controlled by a Teflon stopcock
upstream of the impinger, ranged from 70 to 250 m L / m i n with
no difference in the results.
The used trapping solution in the impinger was then carefully
washed with distilled water into a 100–mL volumetric flask. The
sample was then diluted to 100 m L for later analysis by I C . A l l
surfaces upstream of the impinger were either glass or Teflon.
The ambient temperature and pressure and the resulting water
Figure 1. Gas sampling apparatus.
head pressure on the air contained in the intermediate vessel

43
Journal of Chromatographic Science, Vol. 27, January 1989

tained an unknown impurity that appeared on chromatograms Gas samples were collected on a volume basis, and the chlorine

at the F elution time. This counteracted the water dip, which mole fraction was calculated by converting the trapped mass

had about the same elution time. A sample strength of about of Cl into equivalent moles of chlorine and dividing by the

0.003 Μ N a S O resulted in an impurity peak size that effec­
2 3
moles of gas sampled. The trapped mass of Cl was the pro­

tively nullified the water dip and gave a straight base line from duct of the measured Cl concentration of the diluted trapped

injection to the Cl peak. A l l samples, including standards, sample multiplied by its volume.
were dosed as needed with N a S O to a concentration of 0.003
2 3
For water samples collected on a weight basis, the trapped

Μ to eliminate the water dip and give consistent base lines on mass of Cl was determined as it was for air samples. This

all chromatograms. trapped mass of Cl was then divided by the sample volume
Ten HCl standards were prepared over the concentration taking the density of the solution as 1 g / m L to give chlorine
range 18.23 to 1.823 m g / L . This range was selected because, concentration. When the chlorine in water samples was trapped

on an equivalent Cl basis, it matched Cl concentrations ex­ –
by adding a few drops of 1 Μ N a S O to a minimum 200–mL 2 3


pected in trapped samples. Seven successive analyses of an in­ sample, the chlorine concentration was equal to the Cl con­
dividual standard gave area responses with a coefficient of varia­ centration corrected for dilution.

tion (RSD) of 0.4%. This value was well within the chromato­ The column gave excellent separation of Cl from other
graph manufacturer's operational limit of an R S D value of 3% anions. Chromatograms of trapped chlorine solutions clearly
– 2 – 2 – 2 –
(20). Table I gives the instrument calibration data. Only two showed C l , excess S O , and produced S O . The S O
3 4 3

2 –
measurements of each standard were made because all R S D to S O conversion could not be correlated quantitatively
4

2 – 2 –
values of standards were within 3 % . Although area response because S O slowly forms S O
3 in water in the absence of
4

to standards could vary up to 2% over an 8–h period, samples chlorine. Analysis of trapped samples repeated three weeks after
and standards were usually analyzed within 10 min of each an initial analysis gave identical results showing at least a three-
other. A typical chromatogram is shown in Figure 2. week shelf life for trapped samples.
The standard concentration vs. area curve for the calibration
data in Table I deviates slightly from linearity at the low end Testing the method
of the concentration range. This was possibly the result of a The I C method for analyzing chlorine in gas samples was
large peak threshold value eliminating a larger percentage of tested by using it to measure the composition of a certified
a smaller peak. C l / N gas mixture of 530 ppm ±2% (cylinder pressure was
2 2

Trapped samples (unknowns) were analyzed for Cl by –


2000 psig) purchased from Linde. To verify the method for
proper dilution with 0.003 Μ N a S O so that the Cl concen­
2 3

determining chlorine in water samples, a comparison was made
tration was within the range of the calibration data while keeping

the Cl peak as large as possible, but still on the recorder

scale. A standard was then analyzed that gave a Cl peak area
as close in size as possible to the unknown. With the known
– –
standard concentration on a Cl basis, the Cl concentration
(mg/L) of the unknown was calculated using a straight line slope
for the standard concentration vs. area curve at this point and
the responses of the standard and the unknown. The slope was
based on the two standards closest in concentration to the
unknown.

Table I. Data for Standard Curve for Chloride Analysis


Cl Peak areas –

(integirator counts χ 10 ) –6

HClConcn Standard
(mg/L) Sample Mean deviation

18.23 529.204 529.123 0.114


529.042
16.41 473.190 473.594 0.571
473.998
14.58 413.939 414.886 1.339
415.833
12.76 358.704 360.083 1.950
361.462
10.94 303.384 303.452 0.095
303.519
9.115 249.378 248.708 0.948
248.037
7.292 197.115 196.992 0.173
196.870
5.469 143.143 142.990 0.217
142.836
3.646 94.399 94.572 0.244
94.745
1.823 47.861 46.888 1.376
Figure 2. Typical chromatogram of a trapped air or water sample.
45.915

44
Journal of Chromatographic Science, Vol. 27, January 1989

between the method of this work and a standard iodometric Results and Discussion
method (wet test) (21) by using both procedures to measure the
concentration of chlorine/water solutions. The iodometric The I C method of analysis for C l in air was tested on the
2

method was chosen over others because it is a well-established certified C l / N mixture of 530 ppm ±2%. Analyses for five
2 2

procedure for determining chlorine over a wide range of con­ samples taken directly from the cylinder spanning five days gave
centrations and it requires only standard laboratory equipment an average composition of 538 ppm C l with a standard devia­
2

(6,22). tion of 11 ppm C l . 2

The comparison of the water analysis method with the wet Agreement between the wet test and the I C method for the
test was made at four different chlorine/water concentrations. analysis of chlorine in distilled water was also very good, as
Test chlorine/water solutions were prepared by bubbling both shown in Table II. Trials 1, 2, and 3, representing measurements
pure chlorine from a standard 150-pound cylinder and the cer­ made on three different days, showed a percent difference be­
tified C l / N mixture mentioned above through distilled water
2 2 tween the two methods of less than 1%. In addition, standard
in an equilibrium cell at 20°C and at 30°C. In each case, water deviations of concentrations determined by wet test and I C for
samples were taken at equilibrium. Preliminary experiments in­ the same trials were, respectively, 0.6 and 1.4 m g / L . O f the re­
dicated that bubbling pure chlorine through water at 600 m L / m i n maining three trials for different operating conditions, only Trial
for approximately 45 min would produce a saturated solution; 4 showed a larger difference between the methods than was ex­
however, to insure equilibrium in the cell, pure chlorine was pected. The difference found for Trial 4, more than 4%, was
bubbled for two hours before sampling. Equilibrium was con­ probably because the equilibrium cell had not been adequately
firmed when hourly I C analysis of water samples from the cell, flushed between Trials 3 and 4, which were made sequentially
made after two hours of bubbling, showed no change in com­ on the same day. A s a result, the distilled water charged to the
position. When bubbling with the certified C l / N mixture, the
2 2 equilibrium cell for Trial 4 probably became contaminated with

liquid was assumed to be saturated with chlorine when hourly a small amount of Cl remaining in the cell from the previous
analysis of the exit gas of the cell by our I C method was un­ trial. This would cause the Trial 4 analysis by IC to be larger
changed. For this case, analysis of the exit gas showed no change than that by the wet test. The chlorine concentration range
in composition after bubbling for approximately 4 h at 600 covered for these samples was 160 to 7000 m g / L .
m L / m i n . In either case, once equilibrium between gas and liquid As part of the comparison of the water analysis method with
had been reached, water samples were taken in duplicate for the wet test, the exit gas of the equilibrium cell was analyzed
analysis, one by I C and the other by wet test. by the I C method to determine equilibrium when bubbling with
The wet test for chlorine is based on the formation of iodine the certified C l / N mixture. This was approximately one year
2 2

from the reaction of iodide ion with chlorine: after the original analysis of the certified C l / N mixture, and
2 2

cylinder pressure at this time was about 1400 psig. Eight analyses
Eq. 6
were made of the exit gas (i.e., the certified gas mixture after
The iodine is then reduced by sodium thiosulfate with starch having been bubbled through the cell for equilibrium conditions
as an indicator: at 20°C and 30°C). One analysis of a sample taken directly from
the supply cylinder of the certified gas mixture also was made.
Eq. 7 These nine analyses spanning three days gave an average com­
position of 512 ppm chlorine with a standard deviation of 3
The C l concentration is given by:
2
ppm C l . The analysis of the sample taken directly from the
2

cylinder was 511 ppm. The higher precision of these analyses


compared to the original analysis of the certified mixture was
Eq. 8 probably due to better purging of the gas sampling apparatus
The procedure follows: before sampling. The small change in chlorine concentration
from 538 to 512 ppm is easily possible, because C l / N mix­ 2 2

1. Ten milliliters of 2% by weight K I , 2 m L of glacial acetic tures stored in steel cylinders are known to change concentra­
acid, and approximately 1 m L of a standard 1% starch tion by adsorption or desorption of chlorine as cylinder pressure
solution were added to a 100-mL volumetric flask which changes. The 512 ppm C l concentration is included as operating
2

was tightly capped and tared. information in Table II.


2. A chlorine/water sample was introduced into the flask,
The IC method of this work was also used on a pilot 4-in.
which was immediately capped and weighed. Sample sizes
packed tower that scrubbed chlorine from air with city water
varied from 10 to 15 g.
at 30°C. Analysis of air samples taken from the scrubber covered
3. The sample was then titrated using either 0.1 or 0.005 Ν a concentration range of 50 to 800 ppm. Additional H C I stand­
N a S O depending on the expected chlorine concentration.
2 2 3
ards more dilute than those given in Table I were used when
For the comparison involving the water analysis, all chlorine/ analyzing air samples as low as 50 ppm, because the resulting

water solutions were prepared with freshly made distilled water trapped samples gave Cl concentrations below the range
stored in a nonmetal system. For wet test analyses, chlorine/ shown in Table I. For water samples taken from the scrubber,
water solutions made with pure chlorine were titrated with 0.1 the concentration varied from 8 to 65 m g / L . The city water

Ν N a S O , whereas 0.005 Ν N a S O was used for titration
2 2 3 2 2 3 already contained some chlorine as well as Cl from other
of much weaker solutions made with the certified C l / N mix­ 2 2 sources; therefore, the chlorine concentration could not be deter­
ture. Stock 0.1 Ν N a S O was diluted to make 0.005 Ν N a S O .
2 2 3 2 2 3 mined by the I C method. The method was used, however, to
The strength of the N a S O was checked by a standard method
2 2 3 measure the chlorine added to the scrubbing water in the packed
using titration with potassium iodate (23). The entire process tower. This application of the I C method was satisfactory
from loading the flask to completing the titration took approx­ because chlorine input-output balances for 10 trials were on
imately 30 min for three samples analyzed by wet test. average within 3%, with a maximum difference of only 5°7o.

45
Journal of Chromatographic Science, Vol. 27, January 1989

Table II. Comparison of IC and Wet Test Analysis of 2. J.L. Morrison. Recovery scrubber for waste chlorine gas. Tappi
Chlorine/Water Samples from Equilibrium Cell 51(12): 124A–25A (1968).
Cell Cl con centratioms 3. G.Y. Pan, J.F. DeGraw, and J.J. Renard. Tappi Proceedings, 1982
2

Pulping Conference, pp. 365–70.


Cell Liquid, mg/L b

4. I.F. Rees. Bleach plant scrubber for chlorine and chlorine dioxide.
temperature
Trial Gas a
Wet test IC Difference 6 Appita 34(6): 471–74 (1981).
(°C)
5. A laboratory investigation of an iodometric method for determining
1 30 pure 5467 5480 0.24 chlorine and chlorine dioxide in pulp and paper industry workplace
2 30 pure 5390 5407 0.32 atmospheres. NCASI Technical Bulletin 409. September 1983.
30 pure 5481 5429 –0.95 6. G.C. White. Handbook of Chlorination, 2nd ed. Van Nostrand
3

4 30 512 ppm 159 166 4.40 Reinhold, New York, 1986, pp. 214–55.
5 20 pure 7072 6904 –2.38 7. G.C. White. Handbook of Chlorination, 2nd ed. Van Nostrand
CD

20 512 ppm 169 171 1.18


Reinhold, New York, 1986, p. 151.
a This column indicates the gas bubbled through the equilibrium cell. Pure C l gas 2 8. C.W. Spalding. Reaction kinetics in the absorption of chlorine into
contained 0.5% inerts.
aqueous media. AlChE J. 8(5): 685–89 (1962).
b Each reported chlorine concentration for the water in the equilibrium cell, as analyzed
by the method indicated, is the average value for three samples except for Trials 9. R.E. Connick and Y. Chia. The hydrolysis of chlorine and its varia­
1 and 6, which had one and two samples, respectively, tion with temperature. J. Am. Chem. Soc. 81: 1280-84 (1959).
c Based on the wet test.
10. G.C. White. Handbook of Chlorination, 2nd ed. Van Nostrand
Reinhold, New York, 1986, p. 154.
11. C.A. Pohl and E.L. Johnson. Ion chromatography: the state-of-
the-art. J. Chromatogr. Sci. 18: 442–52 (1980).
Further testing of the method at other air and water concen­ 12. W.J. Weber Jr. Physicochemical Processes For Water Quality
trations was felt to be unnecessary because of the good results Control. Wiley-lnterscience, New York, 1972, p. 439.
obtained in the comparison studies and in the scrubber material 13. J.A. Dean. Lange's Handbook of Chemistry, 11th ed. McGraw-
balance calculations. Hill, New York, 1973, pp. 6–2–6–16.
The chlorine trapping method of this work for air and water 14. T. Moeller. Inorganic Chemistry. John Wiley&Sons, New York,
analysis could easily be extended to higher chlorine concentra­ 1952, pp. 536–37.
tions by using a stronger trapping solution followed by greater 15. E.S. Gould. Inorganic Reactions and Structure. Henry Holt, New
York, 1955, pp. 210, 220.
dilution of the trapped samples. Lower chlorine concentrations
16. J.C. Bailar, T. Moeller, and J. Weinberg. University Chemistry. D.C.
in air and water samples could be determined simply by using
Heath, Boston, 1965, pp. 423–24.
more sensitive range settings on the chromatograph and larger
17. G.C. White. Handbook of Chlorination, 2nd ed. Van Nostrand
sample volumes. A l s o , it should be possible to modify the gas Reinhold, New York, 1986, pp. 759–61.
sampling apparatus to accommodate unpressurized gas samples. 18. J . Halperin and H. Taube. The transfer of oxygen atoms in oxida­
tion-reduction reactions. Ill The reaction of halogenates with
sulfite in aqueous solution. J. Am. Chem. Soc. 74: 375–80 (1952).
19. M.W. Lister and P. Rosenblum. Oxidation of nitrite and iodate ions
Acknowledgment by hypochlorite ions. Can. J. Chem. 39: 1645–51 (1961).
20. System 2000 i/SP Ion Chromatograph Operation and Maintenance
Manual, Document No. 32584, Revision 01, 26 March 1985.
This work was done under Grant ARI–108 of the Alabama
Dionex Corporation, Sunnyvale, CA. p. 4–1.
Research Institute whose support is most gratefully acknow­
21. G.C. White. Handbook of Chlorination, 2nd ed. Van Nostrand
ledged. Reinhold, New York, 1986, pp. 225–27.
22. Standard Methods for the Analysis of Water and Wastewater,
14th ed. ALPHA, AWWA, WPCF, 1976, pp. 309–45.
23. J . Rosin. Reagent Chemicals and Standards, 5th ed. D. Van
References Nostrand, Princeton, 1967, pp. 383–84.

1. L.M. Bungey. Bleach plant atmosphere improved at CFP by scrub­ Manuscript received December 29, 1987;
bing off-gases. Pulp & Paper 55(7): 106–11 (1981). revision received July 25, 1988.

46

You might also like