Water Disinfection and Natural Organic Matter - ACS Publications

Chapter 7

Downloaded by KTH ROYAL INST OF TECHNOLOGY on August 24, 2015 | http://pubs.acs.org Publication Date: November 19, 1996 | doi: 10.1021/bk-1996-0649.ch007

Application of Product Studies in the Elucidation of Chloramine Reaction Pathways Peter J. Vikesland, Richard L. Valentine, and Kenan Ozekin

1

Department of Civil and Environmental Engineering, 122 Engineering Research Facility, University of Iowa, Iowa City, IA 52242 Chloramines are commonly used for drinking water disinfection in systems where it is difficult to maintain a free chlorine residual or where concern over DBP formation exists. They are however, intrinsically unstable and also are lost via an autodecomposition reaction resulting in nitrogen oxidation. This work discusses the measurement of chlorine and nitrogen containing species as a tool in understanding chloramine decay mechanisms and reaction pathways. In order to account for the formation of nitrogen gas by monochloramine decay, a N isotope technique was developed and is discussed fully. 15

2

Chloramines have long been used to provide a disinfecting residual in distribution systems in which it is difficult to maintain a free chlorine residual. This is because chloramines are generally less reactive than free chlorine with constituents such as dissolved organic matter (DOM). The EPA has suggested the use of chloramines to replace free chlorine as a disinfectant because they are believed to produce fewer trihalomethane (THM) disinfection by-products (DBPs) (7). A number of studies, however, have shown that a variety of halogenated and non-halogenated DBPs are produced in chloraminated water containing D O M (2-7). While chloramines are generally believed to be less reactive than free chlorine, they are, however, inherently unstable even in the absence of organic matter. This is because a complex set of reactions occur, ultimately resulting in the oxidation of ammonia and reduction of the active chlorine. The rate of the reactions depends on the ratio of chlorine to ammonia nitrogen (C1:N ratio) as well as on pH. In general, the greater the ratio of chlorine to nitrogen, the faster the oxidation of ammonia occurs. However, at dose ratios where monochloramine is the dominant cWorarnine in solution, the redox reactions take hours to days. Recent work indicates that ammonia nitrogen may be oxidized to N , N0 " , and at least one unidentified product (8). During these reactions, chlorine goes from the +1 valence state in monochloramine to -1 in chloride. The net reaction is also expected to increase the free ammonia concentration by an amount which depends on the specific products formed, as is shown for the formation of nitrate and nitrogen gas: 2

3

1

Current address: 5739 South Andes Street, Aurora,CO80015 0097-6156/96/0649-0105S15.00/0 © 1996 American Chemical Society

In Water Disinfection and Natural Organic Matter; Minear, R., et al.; ACS Symposium Series; American Chemical Society: Washington, DC, 1996.

106

WATER DISINFECTION AND NATURAL ORGANIC MATTER

4 NH C1 + 3 H 0 -> 4 CI" + 3 NH + N0 " + 5 H 3 NH C1 - » N + NH + 3 CI" + 3 H 2

2

3

+

3

+


2

2

3

(1) (2)

Loss of chloramines in a distribution system is therefore attributable to both the normally occurring redox reactions involving ammonia production (auto-oxidation which is presumably the good pathway), as well as to reactions which involve organic and inorganic species, some of which might lead to DBP formation (possibly bad pathways). The relative importance of the two general decomposition pathways, i.e ammonia oxidation vs. oxidation/substitution reactions involving organic and other matter, is important because it relates chloramine loss to the potential formation of DBPs. It is the two concerns, DBP formation and disinfection, which require that chloramine dosages be small enough to minimize DBP formation and yet still provide an adequate disinfecting residual throughout the system. Product studies incorporating all known inorganic products may be a very useful tool in understanding the fate of monochloramine in distribution systems, especially in differentiating between auto-decomposition and DOM reaction pathways. This provides an understanding of the limits on the formation of halogenated and nonhalogenated DBP formation and monocWoramine loss mechanisms which govern the stability of disinfectant residuals. To date, work has not been done to simultaneously measure all known inorganic species which include chloride, nitrate, nitrogen gas, and ammonia. This paper discusses our approach to making detailed analyses of all major products and presents results at two different reaction conditions. This was accomplished by utilizing nitrogen isotope techniques for the measurement of nitrogen gas, by using ion chromatography for the measurement of chloride, nitrate and total ammonia, and by quantifying the monochloramine concentration with the DPD-FAS titrimetric method. Experimental Materials and Methods. The water used for all experiments and for the cleaning of glassware was at a minimum deionized using a Barnstead ULTRO pure water system. For the mass balance experiments which required highly purified water, a Barnstead OrganicPure system was employed. All chemicals used in these experiments were analytical laboratory grade or better. The DPD-FAS method was employed for the measurement of free chlorine and cMoramine residuals (9). A Dionex 20001 IC, equipped with an AS4A anion separatory column and an AS4G guard column was used for the chloride and nitrate measurements. Ammonia concentrations were measured using a Dionex 40001 IC equipped with an AS 10 cation separatory column and an AGIO guard column. To eliminate any possible interactions which might occur between the cation resin and monochloramine, all of the samples were reduced with sodium sulfite prior to injection. The pH was measured with a Fisher Model 420 meter after appropriate calibration. In order to obtain accurate chloride measurements it is essential that a low background chloride concentration be obtained in the free chlorine stock solution. To this end, the sodium hypochlorite solution used for the preparation of the monochloramine samples was produced using a modification of a method developed by Reinhard and Stumm (10). This method involves the dissolution of chlorine gas in an aqueous suspension of mercuric oxide (HgO). After approximately one hour, the solution was distilled using a Buchler rotary evaporator. The resultant solution had a HOC1 concentration of approximately 8000 mg/L, and the chloride concentration in the sample was subsequently determined to be 500 mg/L. This corresponds to a 7.89:1 C1/C1" molar ratio. For comparison the C1 /CF molar ratio in a commercially available HOC1 solutionfromFisher is approximately 1:2. 2

2


7. VIKESLAND ET AL.

Chloramine Reaction Pathways

107

To facilitate the mass balance experiments detailed herein it was also necessary to eliminate any chlorine demand which might exist between compounds adsorbed onto the glassware and the free and/or combined chlorine found in solution. This was done by placing all glassware into a concentrated chlorine bath (-5,000 mg/L Cl ) for a period of at least 24 hours. Adsorbed compounds react with the chlorine and exert any chlorine demand which they may have. After the glassware was allowed to soak, it was thoroughly rinsed with chlorine free OrganicPure water and then allowed to dry. The experimental cWorarnine solutions were prepared with the following conditions: pH 6.5 and 7.5, 4 mM NaHC0 , incubation at 25 °C in darkness, and [NH C1] = 0.25 mM (17.75 mg/L). These solutions were prepared using a preformed monocMoramine stock solution, using ( NH4) S0 for the production of the ammonium stock instead of unlabeled ammonium sulfate (11). Aliquots of 100 mL were then placed into 120 mL septa capped vials (Supelco #3-3 111), leaving a 20 mL headspace. These vials were capped using Teflon coated silicone septa (Supelco #2-7236) and were then placed upside down in a 25 °C incubator. In order to ensure that mass transfer effects were rrimimized, the vials were periodically shaken. At the beginning and at each point during the experimental trial where the system was analyzed, the following measurements were taken: CI", N0 ", total ammonia, the total oxidant residual, and the N gas concentration in the headspace.


2

3

2

0

15

2

4

3

2

GC/MS Analysis for N . The GC/MS system employed for these experiments was composed of a Hewlett Packard 5890 gas chromatograph containing a J&W Scientific DB-1 column (0.32 mm χ 30 m). This GC was used as the inlet to the mass spectrometer and was operated isothermally at 50 °C throughout the experiments. A VG TRIO-1 single quadrapole mass spectrometer with electron impact (EI) ionization was used for the mass spectra measurements. The standard operating parameters for the GC/MS, are given in Table I. The system was calibrated using N standards which were produced by removing aliquots of 98% N (Cambridge Isotope Labs, Lot ED-314) from a two liter gas cylinder and then transferring them to Mrninert capped vials of known total volume using gas tight syringes. The prepared nitrogen gas standards were then allowed to equilibrate for one hour prior to analyzing them by injecting a 100 μΐ. aliquot of gas sample into the GC/MS. Each sample was measured in at least quadruplicate in order to ininimize the effect of sampling and analysis errors. By monitoring the peak area of the mass spectral peak at 30 m/z for a number of different concentrations it was possible to construct a calibration curve for the instrument (Figure 1). This calibration curve is given in terms of nmol N injected in order to make it applicable to other injection volumes. Using the calibration curve it is then possible to calculate the mass (nmol) of N injected for each of the monocMoramine samples. This value was then converted into an equivalent water concentration by taking into account the injection volume and the partitioning between the solution and the headspace. 1 5

2

15

15

2

2

15

15

1 S 15

2

2

M(umol) F 1 , x r£-x— fe

N Concentration (mM) = 2

(3)

Where M = mass N injected (μηιοί), = injection volume, = volume of reactor headspace, V = monocMoramine solution volume, and f = the fraction of nitrogen which partitions into the headpace. In order to calculate f it is necessary to 15

2

w

hs

te


108



Table I. Operation parameters for the VG TRIO-1 GC/MS. Parameter Ion Repeller

Set Point 4.0 V

Electron Current

150 mA

Electron Energy Focus 1 Focus 3 Ion Energy

70 eV 15.0 V 120.0 V 4.0 V

Parameter Low Mass Resolution High Mass Resolution Multiplier Focus 2 Focus 4 Ion Energy Ramp

Set Point 20.0 12.5 700 V 8.0 V 25.0 V 3.0 mV/ amu

1500000

Mass N2 injected (nmol) A:)

Figure 1 Example calibration curve for the measurement of N via GC/MS. 100 mL injection volume. 15

2

utilize the headspace volume, the Henry's constant for N at the temperature of interest, and the ideal gas law (11). 15

(

I

fhs = 1 +

2

r

1

ν

HV,J

(4)

Where V = volume of headspace, V = volume of aqueous solution, and H= dimensionless Henry's constant (60 @ 25 °C). hs

w

Reactor Integrity. Because the sample protocol used for the measurement of nitrogen gas developed in this paper involves the use of septa capped serum vials it


7.

109


VTKESLAND ET AL.

is vital that the reliability of the seals and the GC/MS system be determined. Without this information it is impossible to say with any validity that the measured N concentrations are correct. In order to make sure that the sample vials did not have any systematic leaks, the following procedure was used: Using the same 120 mL sample vials used for the monochloramine decay experiments, a set of standards containing 100 mL of water with a headspace of 20 mL was produced. These vials were then sealed using the same silicone-Teflon septa and crimptop seals as were used for the decay experiments. A sample aliquot (varying from 0.5 to 6.0 mL) of N , gas was taken from the N gas cylinder and then transferred to each vial using a gastight syringe. Once the gas had been added to each vial, the hole which had been punctured in the septa was sealed by applying a thin coat of silicone sealant to each septa. The silicone was used to cover the hole because it was desired that the control experiments be conducted in vials that approximated as closely as possible the experimental reactor vials. Since it was not possible to add N gas to the control vials without puncturing the septa, it was necessary to add the external sealant in order to approximate the virgin septa used in each of the decay experiments. Once the control experiment vials had been sealed, they were treated in the same manner as the monochloramine decay experiments. The controls were stored in this manner for thirteen days after which they were removed from the incubator and analyzed for their N content using GC/MS. On the same day that the stored controls were analyzed, a second set of controls was produced. These controls were produced using the exact same procedure and concentrations as had been used to produce the original set of controls. This second set of controls was then analyzed at the same time as the original set. By comparing the measured N concentrations for the two sets of controls it was possible to ascertain if any leakage of N gas occurred. A 'two-sample* t-test was utilized to determine if the measured N concentrations in the two sets of vials were statistically indistinguishable. For a situation like this one, where the absolute variance (σ) of the concentration measurements is unknown it is appropriate to assume that the variances of the two data sets are equal (12). By making this assumption, it can be shown that if the means (μ) of the two data sets are equal (i.e. null hypothesis ( H ): μι = μ ) then the following statistic has a i i+n2-2 distribution: 1 5

2


15

1 5

2

1 5

2

15

2

1 5

2

1 5

2

15

2

0

2

n

Y1-Y2

(5)

s.

Where Y\ and ζ are the sample means determinedfromtwo sets of samples, s is the pooled variance of the two samples, and ni and n correspond to the number of individual trials in each experiment. The t-statistic calculated in this manner can then be compared to the t-statistics tabulated in any statistics handbook. The calculated average and standard deviations for ten measurements of each experimental vial are tabulated in Table Π along with the corresponding statistical information. Because the t-statistic obtained for the two data sets, -0.583, is less than the critical 2-sided t-statistic, -2.1, the two data sets are statistically indistinguishable at the 95% confidence level. This suggests that it can be said with a 95% confidence level that the experimental reactor vials do not leak. p

2


110


Table II. Two sample t-test statistics for ten measurements of each standard. (100 μL injection volume)


N

2

Fresh Vials

Thirteen Day Old Vials

Mean Variance Observations Pooled Variance Hypothesized Mean Difference df tStat P(Tfiee

Table IV. Observed ammonia production relative to monochloramine decay values. [NH C1] = 0.25 mM, [ H C 0 ] = 4 mM. 2

PH 6.5 7.5

0

3

Time (day)

NH C1 (mM)

ANH C1 (mM)

Ν(ΠΙ) (mM)

0 9 0 9

0.256 0.020 0.256 0.164

0 0.236 0 0.092

0.375 0.210 0.375 0.314

2

2

NH

3tfte

e

(mM)

ANH ,f 3

(mM)

0.119 0.190 0.119 0.150

ree

0 0.071 0 0.031

Nitrate Production. No nitrate was measured at any time during the experiment for pH 6.5. However, at pH 7.5, approximately 0.001 mM nitrate was measured after 35 days. This value is quite low and is quite near the MDL for nitrate analysis on the Dionex ion chromatograph which was used (MDL = 0.0008 mM). Therefore it is quite possible that any nitrate produced at pH 6.5 was below the detection limit for the instrument. Leung (16), using an initial monochloramine concentration of 1.51 mM (at pH 7.5 and Cl/N = 0.1) measured 0.0222 mM nitrate for a 98% decay in monochloramine. This corresponds to approximately 1.5% of the nitrogen in monochloramine going toward the production of nitrate. For the pH 7.5 experiment described here, approximately 0.45% of the monochloramine nitrogen ended up as nitrate. Nitrogen Gas Production. Nitrogen gas has been theorized to be a product of monochloramine decomposition since the early decay studies of Chapin (75), but it has never actually been measured. The results presented in Figure 3 give conclusive evidence that nitrogen gas is produced. It is apparent that nitrogen production occurs more rapidly in the pH 6.5 sample than in the pH 7.5 sample. This finding is not surprising due to the increased rate of monochloramine decay at lower pH. What is surprising is the fact that for roughly the same level of monochloramine decay in the two samples on day 35, the observed nitrogen gas concentration is much lower in the pH 7.5 sample. This indicates that the monochloramine decay product distribution is affected by the pH of the solution, and that the difference in the observed decay rate at each pH may be due to the corresponding change in reaction mechanism. Nitrogen Mass Balance. The measurements for all of the nitrogen containing species were tabulated in order to derive a nitrogen mass balance for monochloramine decay (Table V). This mass balance is symbolized in the following equation where each species is given in mM (as N): Ν Balance = Δ NHjCl - (ANH

3ftee

+

ΔΝΟ3-

+ 2AN )


2jee

(8)

7.

VIKESLAND ET AL.


113


0.08

Time (days) 15

Figure 3 N production due to monochloramine decay. Error bars indicate the pooled standard deviation of five injectionsfromthree separate experimental vials. [NH C1] = 0.25 mM, [HC0 ] = 4 mM, pH 6.5 and7.5 2

2

0

3

By subtracting the sum of the ammonia production, the nitrate production, and the nitrogen gas production (as N) from the measured monochloramine decay it is possible to get a mass balance for nitrogen. The mass balance information tabulated in Table V indicates that the sum of the measured nitrogen containing product concentrations correlate reasonably well with the measured monochloramine decay. For the pH 6.5 experiment the nitrogen balance is within 16% and for the pH 7.5 experiment the balance is within 7.6% of closure. Based upon these measurements it is apparent that most of the nitrogen released by monochloramine decay may be accounted for. Table V* Nitrogen mass balance derivation. [NH C1] = 0.25 mM, [HC0 ] = 4 mM, pH 6.5 and 7.5. Ν Balance = Δ NUCl - ( Δ Ν Η + Δ Ν 0 + 2ΔΝ , ) 2

0

3

3ίΓββ

3

pH

Time ±NH C\ *' (mM) 2

Κ{ΧΛ

2

ΔΝ0 ΔΝΗ (mM) (mM) 3

3ίκε

888

ΔΝ Ν " %Ν (mM) Balance Unaccounted For

0.0632

0.039

16.5%

0.0269

0.007

7.61%


114



Conclusions The decay of monochloramine was studied at two different pHs, 6.5 and 7.5 and all known inorganic products were measured. The use of GC/MS to measure isotopic nitrogen gas using headspace analysis in batch reactors consisting of septa closed vials was successfully applied. Chloride, ammonia, and nitrogen gas are all significant products of monochloramine decay. At pH 7.5, nitrate was also shown to be a product of the decay process. Based upon the high percent recoveries (> 80%) of chlorine and nitrogen containing decay products it is apparent that under the conditions studied that most of the initial chlorine and nitrogen in monochloramine may be accounted for. Acknowledgments The authors would like to express their appreciation to the AWWA Research Foundation (AWWARF) for financial support. AWWARF assumes no responsibility for the content of the research documented in this article or for the opinions expressed herein. The use of trade names for commercial products does not imply endorsement by AWWARF but is included solely for reference. The authors express their thanks to AWWARF project officer Joel Catlin for his assistance. Literature Cited (1) Cotruro, J.A. ES&T, 1981, 14, 268. (2) Fleischacker, S.J. and Randtke, S.J. J. AWWA, 1983, 75, 1323-144. (3) Amy, G.L., et al. In Water Chlorination: Chemistry, Environmental Impact, and Health Effects; R.L. Jolley, et al. Eds.; Lewis Publishers: Chelsea, MI, 1990, Vol. 6; pp 605-622. (4) Jenson, J.N., et al. In Water Chlorination: Chemistry, Environmental Impact, and Health Effects; R.L. Jolley, et al. Eds, Lewis Publishers: Chelsea, MI, 1985, Vol. 5; pp. 939-950. (5) Arber, Α., et al. In Water Chlorination: Chemistry, Environmental Impact, and Health Effects; R.L. Jolley, et al. Eds, Lewis Publishers: Chelsea, MI, 1985, Vol. 5; pp. 951-964. (6) Kanniganti, R., et al. ES&T, 1992, 26, 1988-2004. (7) Stevens, A.A., et al. In Water Chlorination: Chemistry, Environmental Impact, and Health Effects; R.L. Jolley, et al. Eds, Lewis Publishers: Chelsea, MI, 1990, Vol. 6; pp. 579-604. (8) Valentine, R.L. and Wilber, G.G., In Water Chlorination: Chemistry, Environmental Impact, and Health Effects; R.L. Jolley, et al. Eds, Lewis Publishers: Chelsea, MI, 1990, Vol. 6; pp. 819-832. (9) APHA, ANWA, and WPCF, Standard Methods for the Examination of Water and Wastewater; American Public Health Association, American Water Works Association, and Water Pollution Control Association: Washington, D.C.; 16 Edition (10) Reinhard, M. and Stumm, W. In Water Chlorination: Chemistry, Environmental Impact, and Health Effects; R.L. Jolley, et al. Eds, Lewis Publishers: Chelsea, MI, 1980; Vol. 3; pp. 209-218. (11) Vikesland, P. MS Thesis, The University of Iowa: Iowa City, 1995. (12) Fisher, L. D. and Van Belle, G. Biostatistics: A Methodology for the Health Sciences; John Wiley and Sons, Inc.: New York, 1993. (13) Diyamandoglu, V. Ph.D. Dissertation; University of California-Berkeley,1989. (14) Leao, S. F. Ph.D. Dissertation,: University of California, Berkeley, 1981. (15) Chapin, R. J. Am. Chem. Soc. 1931, 53, 912-921. (16) Leung, S. Ph.D. Dissertation; The University of Iowa: Iowa City, 1989. th


Water Disinfection and Natural Organic Matter - ACS Publications

Recommend Documents