(15) Pell, E. J., Weissberger, W. C., Speroni,J. J., The Pennsylvania
(1970). (10) Kissmeyer-Nielsen,E., Weckel, K. G., Eur. Potato J., 10, 312 (1967). (11) Speroni, J. J., M.S. Thesis, The Pennsylvania State University, Universitv Park. 1979, D 31. (12) Shaw,R.,Am. Potatb J., 46,201 (1969). (13) Seem, R. C., Ph.D. Thesis, The Pennsylvania State University, University Park, 1976, p 120. (14) Sweeley, C. C., Bentley, R., Makita, M., Wells, W. W., J . Am. Chem Soc , 85,2497 (1963).
State University, unpublished data, 1978. Received f o r review October 12, 1979. Accepted February 4, 1980. Contribution No. 1107 from the Department of Plant Pathology, Journal Series Paper No. 5843, The Pennsylvania Agricultural Experiment Station, and Contribution No. 544-79 from the Center for Air Enuironment Studies, Department of Energy No. COO/ 4331-05. This research was supported i n part by Department of Energy Contract No. 79EV04331.000.
Reaction of Nitrilotriacetate with Ozone in Model and Natural Waters Larry M. Games* and Jean A. Staubach Procter and Gamble Company, lvorydale Technical Center, Cincinnati, Ohio 452 17
The kinetics of the reaction between ozone and nitrilotriacetate (NTA) in water and the effects of pH, metal content, and natural organics on the rate and extent of degradation were measured. The rate constant for the primary reaction was 83 f 9 M-l s-l a t pH 2. Conversion of NTA to C02 was determined using uniformly 14C-labeledmaterial. Both the rate constant for the primary reaction and the rate and extent of ultimate degradation to C02 were affected by pH and NTAmetal speciation. The effect of these variables was difficult to differentiate, since pH changes can affect the NTA-metal speciation. However, the effect of both parameters was clearly more important for ultimate degradation than for primary degradation. Rapid ultimate degradation of NTA during ozonation did occur in a natural water. We conclude that NTA will undergo substantial ultimate degradation in a drinking water plant practicing ozone treatment. Ozonation has been used as a water treatment process since the early 1900s in Europe, but only recently has its introduction been contemplated on a large scale in North America. I t is receiving attention now because of the discovery of chlorinated hydrocarbons in drinking water as a result of chlorine disinfection. In addition to replacing chlorine for disinfection, ozone has many other beneficial chemical properties in water treatment ( I ) . I t is inferior to chlorine, however, in two important respects. First, ozone degrades quickly and completely in water, so that a residual concentration of it cannot be maintained in water distribution lines. Second, since ozone is a more powerful oxidant than chlorine, it forms more oxidation products. The latter factor has been a major rationale and for the reluctance to adopt ozonation widely in the U.S., most reviews of the subject suggest further research before introducing it (2). In this study, we have investigated the interaction of ozone with trisodium nitrilotriacetate (NTA) in water in a laboratory scale system designed to model a full-scale treatment plant. NTA is an organic complexing agent whose properties make it useful as a builder for detergents. The effects of pH, metal content, and other organics were investigated since they were expected to influence the ozonation process. The pH can affect the ozonation process by influencing both the stability of ozone and the ionization state of the organic compounds. Several investigators have demonstrated the effect of pH on the chemistry of ozone in water ( 3 - 5 ) . At low pH, ozone is relatively stable in water and it reacts directly with organic material. A t high pH, ozone decomposes rapidly to form radicals. These radicals can then react with viruses and bacteria or, more likely, dissolved organics and metals. A t intermediate pHs, both reactions can occur. The pH range 0013-936X/80/0914-0571$01.00/0
for each reaction is loosely defined and depends on the relative rates of the direct reaction and the radical reaction for individual species. For most organic compounds, the radical reaction is several orders of magnitude faster than the direct reaction (6). The pH also affects the ozonation reaction for species such as amines and carboxylic acids because their reaction rates with ozone depend on their ionization states. Amines react only as free amines, and carboxylic acids react significantly only as free anions. This presents something of a dilemma for zwitterions such as alanine or for NTA a t pHs where both the protonated amine and deprotonated acid exist. However, as expected, the reaction of alanine appears to proceed as if the unreacting functionality is not present (7). The effect of metals on the ozonation of NTA and related compounds has not been defined. Shambaugh and Melnyk (8)found that “the rate of destruction of metal-EDTA complexes is an order of magnitude faster than the simple attack of ozone on EDTA alone”. Similarly, NTA might be expected to react with ozone a t a different rate when complexed with metals. Stolzberg and Hume (9, 10) studied the photodegradation of NTA, and their data suggest that the kinetics of NTA photooxidation is influenced by the formation of metal complexes. In the work described here, NTA was added to water samples that ranged from highly purified distilled water to Ohio River water at pHs from 2 to 9.65. NTA concentrations varied from 35 to 350 ppb, concentrations that are about in the range that has appeared in Canadian river waters since NTA was introduced into detergents there ( 1 1 ) . (These numbers, as do all others in this report, refer to the free acid.) We used uniformly 14C-labeled NTA in our ozonation experiments so that both primary and ultimate degradation rates could be established. Ozone dosages corresponded as closely as possible to practical use levels, and the ozone contact chamber was a laboratory scale version of a pilot plant system described in the literature (12). All variables studied were measured directly except the complexation state of NTA, which was calculated by use of the REDEQLZ computer program for complexation in aqueous systems ( 1 3 ) .
@ 1980 American Chemical Society
Experimental Materials. Uniformly I4C-1abeled trisodium nitrilotriacetate monohydrate was purchased from Amersham Searle. Its specific activity was 11.6 mCi/mmol, and reverse isotope dilution analysis in our laboratory indicated its purity was greater than 97%. Nonlabeled nitrilotriacetic acid (97%purity) was purchased from Eastman Kodak and used without further purification. All other chemicals were reagent grade from Fisher or Matheson Coleman and Bell. Volume 14, Number 5, May 1980
571
Teflon C a p
Table 1. Concentrations of Chemicals Present in Model River Water during Ozonation of NTA a reagent
concentration (as metal)
Ca as CaCI2.2H20 Mg as MgCIz-6H~0 Cu as CuS04 Zn as ZnClz Ni as NiS04.6H20 HC03 as NaHC03 a
67.4 mg/L 9.97 mg/L 7.8 Fg/L 36.9 Fg/L 5.7 Fg/L 163 mg/L
COPTraps
pH of the starting solution ranged from 7.2 to 8.3
Ozone was generated with a Welsbach T-23 laboratory ozonator from tank dry air that was additionally purified by passing through DrieriteB and molecular sieves. The flow rate was 1 L/min and the pressure 9.3 psig. Voltages of 65 and 110 V were used to produce 7.5 and 37 mg/L ozone concentrations in the air, respectively. This resulted in ozone residuals in water of approximately 1 and 8 mg/L, respectively. For exposure to ozone, NTA was dissolved in several different aqueous media. Specially purified laboratory distilled water with greater than 18 MQ resistivity and less than 0.5 mg of C/L was the basis for most of the media. In several experiments NTA was dissolved in this water alone. In one experiment, 67.4 mg of Ca/L as CaC12.2H20 was added. In most experiments, metal salts were added to the solutions to simulate the water of rivers with moderately high concentrations of metal ions. Concentrations in this model river water are shown in Table I. In addition to these model systems, two natural waters were used. One was Ohio River water collected from the shore a t the Cincinnati Water Treatment Plant intake. The other was water from the cistern of a home that collects rainwater for general use. NTA Complexation State. The complexation state of the NTA in each of the systems studied was calculated by the July 1975 version of the REDEQLZ computer program (13).This program is designed to calculate equilibrium states for ionic species in water from their stability constants and their concentrations. The anions considered in the calculations were C1-, S04*-, C03*-, P043-, and NO3-, as well as NTA. Competition for other organic ligands was not considered. Metal concentrations in the Ohio River water and cistern water samples were measured by atomic absorption spectrophotometry. Metal concentrations in the model river water were assumed equal to the concentrations added, and no metals were assumed present in the distilled water. The pH values in the program were adjusted to match the pH values seen in various experiments. Primary Degradation. The rate of primary degradation was determined using pseudo-first-order conditions with either NTA or ozone in excess. When ozone was in excess, NTA was monitored polarographically (14).These experiments are described later. When NTA was in excess, the initial rate of NTA ozonation was measured spectrophotometrically. The initial rate was measured to minimize the effects of ozone reaction with products formed in the initial reaction. The experiments with NTA in excess were done a t pH 2. The disappearance of ozone was followed spectrophotometrically at 258 nm with a Beckman Model 26 spectrophotometer. We found the molar absorptivity of ozone a t 258 nm to be 2800 L M-l cm-'. Hoigne and Bader (5) reported 2900 L M-l cm-l. Absolute ozone concentrations were measured by standard methods (15). The experiments were modeled after work done by Hoigne and Bader (5).A cardboard cover was fitted over the top of the light-path chamber of the spectrophotometer so that a solution could be injected directly into the cuvette without allowing light to enter. Model river water was adjusted to pH 572
1 1 -
Environmental Science & Technology
Porous Frit
N2
7g-03
-
Vent Vent
Figure 1. Diagram of the ozonation chamber used in the studies of ultimate degradation
2 and ozonated for 10 h to oxidize completely any substances present. After ozone was removed from a portion of this water by purging, it was used to fill the reference cell of the instrument. Another portion of ozone-saturated water was placed in the sample cell to establish base-line absorbance. Excess NTA a t varying concentrations was injected into the cuvette, and the reaction rate was calculated from the initial rate of absorbance decrease. A blank of distilled water was also run a t each concentration. This allowed correction of the ozone concentration decrease for the dilution that was caused by the addition of the sample. To determine whether complete mixing occurred in the cuvette, a concentrated solution of Rhodamine B dye was injected into the cuvette. Observation of dye distribution indicated that mixing was essentially instantaneous and complete. Monitoring the absorbance when no sample had been added showed that ozone was stable during the time periods of interest. Ultimate Degradation. The ultimate degradation of NTA was measured by observing the rate of formation of I4CO2 from [I4C]NTA when it was exposed to a constant concentration of ozone. The reaction vessel used for these experiments was modeled after a pilot plant design (12)and is diagrammed in Figure 1. It permitted the reactants to contact no materials other than glass and Teflon. The air/ozone mixture enters the bottom of the 116 X 6 cm diameter glass chamber through a 6-cm medium-porosity glass frit. Nitrogen can be introduced through the frit to purge C02 from the chamber. The chamber is vented through two gas-washing bottles in series containing 1.5 N NaOH to trap the CO2. In tests with 14C-labeledHCOB-, 99.8%of the COSwas trapped in the first bottle within 20 min. An injection port in the side of the chamber permits the addition of H2S04 to release COSor NazS203 to decompose residual ozone. A 10-mL buret permits rapid, volumetric removal of samples. Sufficient initial volume is discarded between samples to account for the volume in this buret which remains from the previous sample. Ozonation Procedure. For each experiment the reaction chamber was filled with a 3-L sample of one of the test solutions containing 35,140, or 350 pg of [14C]NTA/L. Ozone was bubbled through the solution a t 1 L/min for 45 min. Samples were taken from the ozonation chamber at 5-min intervals during ozonation and were analyzed for volatile and nonvol-
07
NO lniection
H 2 0 lnlection
05
E
NTA-03Direct Reaction PH 2
W
N 0
9
4
04-
8
+ 10 mi. Sample
Test
I
250 ml. Flask
Figure 2. This biometer was used to separate the volatile and nonvolatile
14C in the various experiments
02-
01 -
I
Tube
0 3 -
0 sec
L
10 sec
2 0 sec
Time
Figure 3. A typical curve for ozone decomposition. The upper curve
demonstrates the stability of the ozone and the middle curve shows the blank used for correction for dilution
atile 14C (see below). Residual ozone and p H were measured a t 10-min intervals. In a few experiments, samples were taken NTA, unchanged by the reduction step, was analyzed by difto determine the rate of primary degradation by analysis for ferential pulse polarography on a Princeton Applied Research parent NTA. At the end of 45 min, the flow of ozone was Model 174-A polarograph (15). stopped, residual ozone was destroyed by injecting 20 mL of Residual ozone in the ozonation solution was measured by 0.1 N Na2S203, and carbonate was converted to COn by inadding a 10-mL sample to 5 mL of a 20 g/L KI solution. This jecting 30 mL of 1 N H2S04. Nitrogen was then bubbled solution was acidified with 1 mL of 1 N HnS04 and titrated through the system for 30 min to purge C02 from the ozonwith 0.005 N NazSn03 using starch indicator (14). ating chamber. Final samples were then taken from the ozonation chamber and the COn traps. Results and Discussion Analyses. The 14C in the samples taken from the ozonation Primary Degradation. Primary degradation was meachamber was separated into volatilizable (carbonate, bicarsured spectrophotometrically by following ozone decompobonate, carbon dioxide) and nonvolatile fractions in a biomsition in test solutions containing an excess of NTA. Under eter. The biometer is a 250-mL flask with a side arm leading these pseudo-first-order conditions the following equation to a test tube (Bellco Glass Inc., Vineland, N.J.); both sides describes the decrease in ozone: were sealed with rubber stoppers (see Figure 2). Before use, 2 mL of 1.5 N NaOH was placed in the test tube and two -d[03]ldt = h1[03] (1) crystals of Na2S203 were put into the flask. A 10-mL sample where of the ozonated solution was added; the assembly was swirled to speed up ozone decomposition, and 1 mL of 1 N H2S04was h i = h [NTA] (2) added to the flask through a stopcock. The COSthat evolved and thus was captured by the NaOH while the apparatus stood overnight a t room temperature. The acid solution containing -d[03]/[03] = h [NTA] dt (3) nonvolatile 14C and a 1-mL aliquot of the NaOH solution Integrating this equation and then plotting In ([03]containing volatilizable carbon were assayed for radioactivity. [03]inltial) vs. time should give a straight line with slope equal to h [NTA]. The rate constant ( h )is then obtained by dividing Radioactivity was assayed by liquid scintillation counting the slope by NTA concentration. The above equations are on an Isocap/3OO instrument (Nuclear-Chicago).The 1-mL written assuming that the reaction is first order in both NTA samples from the COz traps were mixed with 20 mL of Caband 0:i.The linearity of the plots demonstrated this for ozone. 0-Si1 cocktail for counting. The Cab-0-Si1 cocktail is 64 mL of Spectrofluor PPO-POPOP (2,5-diphenyloxazole-1,4- T h a t the reaction was also first order in NTA was inferred from the finding that h was constant over the range of NTA bis[2-(5-phenyloxazolyl)]benzene) (Amersham Searle Corp.), concentrations used. 68 g of Cab-0-Si1(M-5grade, Cabot Corp.),936 mL of toluene, The rate constants were determined a t an ozone concenand 1000 mL of ethanol (3A grade). All aqueous samples M and a t NTA concentrations of 1.05 tration of 2.25 X (usually 10 mL) were added to 10 mL of Triton X-100 cocktail; X 1.99 X and 3.61 X M. The mean rate conthis cocktail is 64 mL of Spectrofluor PPO-POPOP, 936 mL stants at each concentration were 86 f 10.7 M-l s-l, 90 f 10.1 of toluene, and 1000 mL of Triton X-100 (Amersham Searle). M-' s-l, and 74 f 5.1 M-l s-l, respectively. Rates for each Correction factors were applied to the raw data from the concentration were determined five times, and the mean rate scintillation counter to correct for counting efficiency. constant for all experiments was 83 f 9.0 M-l s-l. Some of the ozonized solutions were measured for NTA. To Figure 3 shows that ozone did not degrade a t a detectable destroy residual ozone in these 10-mL samples, they were rate when NTA was not added. The figure also shows a typical passed through a Jones reductor column made by packing 10 cm of 8-30 mesh amalgamated zinc into a Pasteur pipet. The curve demonstrating that ozone disappears rapidly when an Volume
14, Number 5,May 1980
573
Table II. Rate Constant for the Direct Reaction of Ozone with Organic Substance in Aqueous Solution a compound
HCOOH HCOORCHO CH3COO"3
k , M - I S-1
compound
k, M-I 5-1
0
