Environ. Sci. Technol. 1982, 16, 866-872
Copper( I I ) Complexing Capacities of Natural Waters by Fluorescence Quenching Davld K. Ryan? and James H. Weber" Chemistry Department, Parsons Hall, University of New Hampshire, Durham, New Hampshire 03824-3598
The natural fluorescence of humic substances in both fresh waters and marine samples is quenched upon complexation to Cu2+ion. Binding curves obtained from titration data are mathematically modeled and complexing capacities (CL)determined by computer curve fitting. The results indicate that the fluorescence technique directly measures organic matter complexation only, and CLvalues are unaffected by hydrolysis. Filtered water samples are analyzed at their natural pH under N2 but are otherwise unaltered. Rayleigh scattering, measured along with fluorescence, signals precipitate formation and was used to indicate a suitable stopping point for titrations. Scattering and fluorescence trends were similar for all samples except an estuarine sample. A multiple correlation study of several commonly measured water characteristics and the titration parameters showed a statistically significant trend between UV absorbance and CL values. Introduction Nearly all methods used to study the complexation of metal ions by humic materials in natural systems (1, 2) have one thing in common. A separation or direct measurement of "free" metal ion is performed and subtracted from the total metal ion concentration in order to determine the extent of organic matter complexation. This indirect manner of determining the extent of binding of natural organic ligands is prone to error from several sources. Metal ions may be complexed by inorganic species in the sample and adsorbed to colloidal particulates that are initially present or that form as a result of coagulation of the organc matter (3-7). When the organic ligand is in excess, the free metal ion concentration is extremely low and difficult to measure. The measurement technique near ita detection limit will give unreliable data and cause error in the calculation of bound ligand. When metal ion is in excess, on the other hand, an error can be introduced when taking the small difference of two large numbers. Free and total metal ion concentrations may be orders of magnitude higher than the micromolar and submicromolar ligand concentrations encountered in natural systems. A more direct method of studying humic material binding is to measure a property of the organic matter itself that changes with metal ion complexation. The natural fluorescence of humic materials is such a property (6,8-IO). In a previous paper (6)we derived an expression relating fluorescence intensities, measured during titration of a fluorescing ligand with paramagnetic Cu2+ion, to the fraction of the ligand bound ([ML]/CL): [ML] IL - I
- - --
'Present address: Harold E. Edgerton Research Laboratory, New England Aquarium, Boston, MA 02110. Envlron. Sci. Technol., Vol. 16, No. 12, 1982
Theoretical Section Use of a conventional conditional stability constant to describe the binding ability of the complex mixture of ligands making up humic matter is not the most desirable approach ( I I , 1 2 ) . MacCarthy and Smith (12) have addressed this problem and proposed a stability function or product (S,) to describe the binding of a metal ion to N ligands in a multiligand mixture. For 1:l binding this product takes the form
S [M(Li)I
s 1=
i=l
(2)
5
[MI [ L ~ I i=l
where N is the number of ligand species present in the mixture, [Li] and [M(Li)] are the equilibrium concentrations of the ith ligand and its metal complex, respectively, and [MI is the free metal ion concentration. Modifying our eq 1to be consistent with the stability product approach results in N
(3)
where N
CL = C[LJ i=l
N
N
i=l
i=l
+ C[M(LJ] = C([Lil + [M(Li)I)
(4)
N
IL=
(1)
CL IL- IML The quantity [ML] is the equilibrium concentration of metal ligand complex, CLis the total ligand concentration or complexing capacity, I is the fluorescence intensity at any point in the titration, and I L and Im are fluorescence intensities for free or bound ligand, respectively. This
866
equation describes a simple system quite readily as demonstrated with L-tyrosine (6).Its application to complex natural system such as the fulvic acid ligand requires certain assumptions that are somewhat less than ideal. However, it allows calculation of CLvalues, which generally agree well with those from other techniques, and gives a 1:l conditional stability constant. In this paper we describe the determination of CLvalues of natural water samples by Cu2+ titrations using fluorescence quenching. Six natural waters, including both fresh water and marine samples, were titrated while their fluorescence and Rayleigh scattering were monitored. From this data we calculated CLvalues and average conditional stability products. In addition various commonly measured water characteristics for the samples are correlated with each other and the titration results in order to show statistically significant trends.
CIL,
i=l
(5)
N
IML = CIM(L~) i=l
(6)
N
I = IL+ IML = CUL,+ IM(L,)) i=l
(7)
Equation 3 gives the fraction of the overall mixture bound to the metal ion in terms of fluorescence intensities. Combining eq 2-4 in the manner described previously (6), we derive eq 8, which relates the measured fluorescence
0013-936X/82/0916-0866$01.25/0
0 1982 American Chemical Society
intensity, a function of the total metal ion added, to the stability product and CL. Equation 8 is identical with the previously derived eq 9 in ref 6 except that S1replaces K . Therefore, the fluorescence method allows determination of average stability products as well as CLvalues for complex mixtures of ligands (12). It is not clear from this treatment whether or not the determined Sl value is in the CLASP-1 region (limiting region) defined by MacCarthy and Smith (12). Thus, we do not know if S1 has a constant value or will vary with small changes in the composition of the sample under study. This means, therefore, that the stability products calculated here cannot be used for predictive or comparative purposes as they are not necessarily constant and may vary with CM as well as CL. In this study their use is necessary in the data treatment as a means of calculating complexing capacities (C,) which are constant for the particular sample. Although the concept of St is important to the interpretation of our results, its calcuation in this data treatment is no different than for K . Therefore, in order to be consistent with our previous work, K will be used in further discussions of the stability product. Experimental Section Materials and Reagents. A Perkin-Elmer Model 204 spectrofluorometer was used to obtain fluorescence and Rayleigh scattering data. It is equipped with two grating monochromators, each with fixed 10-nm band-pass, a 150-W Xenon arc source, and an R212 photomultiplier tube. An ORIEL Corp. long-pass filter (Model 5147) further discriminates scattered light in some experiments as noted. The instrument was used in conjunction with a flow-through system consisting of a 10-mm flow-through quartz cuvette (Precision Cells, type 57H), a temperature jacketed titration cell and top (Princeton Applied Research, Models K64 and K66), a peristaltic pump (Sigmahotor, Model T8), and Teflon and Tygon tubing. Inserted in the holes in the top of the cell were a combination pH electrode (Corning, Model 476059), coupled to an Orion Model 701A specific ion meter, a solution inlet and outlet tube, and a N2purge tube. The system was operated under anaerobic conditions for all experiments. A Fisher constant-temperature water bath maintained the titration cell at 25 f 0.1 OC. Solutions were pumped from the titration cell via the peristaltic pump and entered the spectrometer through a port on the side. Tygm was used in the pump head only; all other tubing was Teflon. Reagent additions were made with Gilmont micrometer burrets (Models S3200A and S1100A). An Instrumentation Laboratory atomic absorption spectrometer (Model 951) was used with background correction to determine iron concentrations. Absorbance was measured with a Cary 219 spectrophotometer. Water samples were filtered through 0.4-pm membranes by using a Nuclepore 142-mm Teflon and Plexiglass filter support with Tygon tubing and the peristaltic pump. All materials that came in contact with samples or solutions were thoroughly acid washed with 10% HN03 (13). Copper ion standards were prepared from reagent grade C~(C10,)~.6H~0 in 1 X 10”’ M HClO., and standardized by EDTA titration (14). Natural Water Samples. A variety of types of water samples were taken in southeastern New Hampshire. This research group previously described sampling procedures and most of the sampling sites (15). These include the
Oyster River, Lamprey River, Portsmouth Reservoir, and Barrington Swamp. In addition, we collected samples of the Great Bay Estuary (Jackson’s Landing, low tide) and Great Bay pore water from several gravity cores. No special precautions were taken to keep the pore water anaerobic. All the bulk water samples were collected in 10-L acidwashed polyethylene bottles and immediately taken to the laboratory and put on ice. Filtration was begun and aliquots taken for pH measurements. After filtration both filtered and unfiltered portions of the samples were stored at 4 “C prior to characterization. The pore water was filtered through 0.4-pm filters using a 47-mm support (16). Extreme care was taken in all stages of sampling and storage to avoid contamination of the samples with metal ions. Procedures. Several characteristic parameters of the water samples were measured according to procedures in ref 17. Procedures were the same as described by Truitt and Weber (15) with the following exceptions. Color measurements (UV absorbances) were made at 250 nm with samples adjusted to 0.2 M in H2S04(18). Absorbance were referenced to our well-characterized soil fulvic acid (SFA) analyzed in the same way (19-21). In addition the fluorescence of acidified samples was also measured and referred to SFA fluorescence. Excitation and emission wavelengths were 350 and 430 nm, respectively. Fluorescence quenching titrations were conducted in the flow-through system on 50.0- or 25.0-mL aliquots. Samples were deaerated by purging with moist N2 in order to facilitate pH control and interpretation of CL values. Fluorescence excitation and emission spectra were scanned and wavelengths of maximum intensity determined. The fluorescence-quenching titrations with Cu2+ ion were conducted as follows: (1)The pH was adjusted to the desired value *0.05 units with 0.01 M KOH and/or HN03. (2) The sample was then circulated through the cuvette and back to the titration cell for several minutes. (3) Fluorescence and scattering were measured. (4) Flow was then reversed to return all of the sample to the titration cell. (5) Cu2+titrant was added and the process repeated. Approximately 15 min elapsed between successive additions of titrant. Adequate stirring was maintained in the titration cell with a magnetic stir bar as well as with N2 bubbling. The fluorescence of a buffered portion of the sample was also measured in a separate cuvette to detect any fluctuations in source intensity and allow correction if necessary. All measurements were made at the excitation and emission wavelengths that produced the maximum signal. Scattering experiments were done with both emission and excitation monochromators set to 400 nm. Fluorescence measurements made after scattering doubled its initial value were discarded except as noted below. All measurements were made with the sample solution flowing through the system at a rate of approximately 15 mL/min. any increased noise due to turbulence was minimal. Data Treatment. The fluorescence quenching titrations were conducted in at least triplicate for all the pore water, which was done in duplicate due to limited sample. For a given set of experimental data the curve-fitting procedure gives the best fit values of K , CL,and I M L from eq 8. This was performed by using nonlinear regression analysis by a program called NoNREG, which is part of the University of North Carolina, Chapel Hill, group statistical programs. The program uses simplex optimization to estimate unknown parameters by least-squares analysis. Environ. Sci. Technol., Vol. 16, No. 12, 1982 867
Table I. Average Natural Water Sample Characteristics characteristicsa
Portsmouth Reservoir
Lamprey River
Oyster River
Barrington Swamp
PH hardness, mg/L as CaCO, alkalinity, mg/L as CaCO, conductance, pmho/cm color, mg/L as SFA fluorescence, mg/L as SFA DOC,mg/L as KHP total iron, pM CL, p M b 10'4K M-lb I M L , &b
6.0 13.1 4.3 67.3 10.2 29.7 7.2 1.0 6.6 (2.9) 6.7 (1.4) 34.7 (2.0)
6.0 17.5 4.0 92.5 20.3 43.5 11.5 3.8 17.1 (1.2) 5.3 (0.3) 20.0 (0.5)
6.7 47.9 20.2 174. 11.6 32.7 8.6 1.4 4.1 (1.5) 8.2 (0.7) 29.2 (1.3)
5.8 10.3 3.8 71.7 16.6 29.3 8.3 4.1 14.6 (2.7) 6.1 (0.4) 20.0 (1.0)
a KHP is potassium hydrogen phthalate; SFA is soil-derived fulvic acid.
In order to determine if replicate titrations for the same water sample have the same variance, it was necessary to subject them to Bartlett's test (22). Duplicate pore water experiments were analyzed by an F test (22). Variances are calculated by dividing the residual sum of squares by the degrees of freedom for the experiment. The residual sum of squares is the parameter the NONREG program uses to determine the best fit of eq 8 to the data. The number of degrees of freedom is the number of data points for the titration minus the number of constants in the fitted equation, which is 3 for eq 8. Bartlett's test determines if the variances of the titrations are the same and whether, therefore, they form a homogeneous group. The F test also measures homogeneity of variances where only two cases are involved. At the 0.05 probability level the variances of replicate titrations for the pore water and all the fresh water samples except Barrington Swamp are the same. They therefore form a homogeneous group, and the fitted parameters derived for titrations of a given sample can be averaged. The Barrington Swamp results are the same at the 0.02 probability level. Titration parameters and the various water sample characteristics measured were analyzed in a multiple correlation study to detect trends. The correlation coefficients were checked for significance at the 90% confidence level.
Results and Discussion General Water Sample Characteristics. The water samples studied cover a broad range of types including a fresh water lake (Portsmouth Reservoir), two rivers (Lamprey and Oyster Rivers), a swamp (Barrington Swamp), an estuary (Great Bay), and marine sediment interstitial water (pore water). The first three samples are from local drinking water supplies. Table I lists the various measured parameters of the samples. In general the fresh waters fall in the expected pH range and are low in hardness ions and low in alkalinity with the exception of the Oyster River. The marine samples are very high in Mg2+and Ca2+and alkalinity, as expected. The additional alkalinity may be due mainly to borate for the Great Bay; however, the extremely high value for the pore water is probably caused by several inorganic anions including silicates and other mineral species as well as organic compounds. Freshwater conductivities are low while that for the Great Bay is very high and can be used to estimate the salinity for the sample (23). We obtained a value of approximately 20%0.The small quantity of pore water prevented measurement of its conductivity with available equipment. Measurements of color (UV absorbance at 250 nm), fluorescence, and dissolved orrganic carbon (DOC) give independent indications of the amount of organic material 888
Environ. Scl. Technol., Vol. 16, No. 12, 1982
Pore Water 8.0 3900 3140 46.0 170. 40.8 < 0.2 0.8 (0.4) 25.8 (6.9) 81.5 (1.9)
Great Bay Estuary 8.0 4860 86.4 39 400 3.2 6.9 4.3 W
c
5
W
c L
c (
W 0
E 0
w
W
W 0
70
I W
u
J LL
-
70
w
60-
W
u
0 3
50-
11
d
80
60
-I LL
%
0.0
0.3
401
30 0.0
1.0
ADDED
2.0
C U ~ +
3.0
4.0
CM x
1051
ADDED
Q.6 C U ~ +
1.2
0.8
C M x 1041
Flgure 2. Triplicate fluorescence quenching curves for the Great Bay sample titrated with Cu2+ at pH 8.0: A (A) no filter; 0 (B) no filter; * (C) cut-off filter used.
5.0
Flgure 1. Triplicate fluorescence quenching curves for the Oyster River sample titrated with Cu2+ at pH 6.7: A (A); 0 (B); (C).
200,
however, appeared at 410-420 and 440-450 nm and were of almost equal intensity in making up the major band. Zepp and Schlotzhauer (25)studied several natural waters and found emission maxima ranged from 430 to 470 nm with shoulders at 400-420 and 500-520 nm. The additional peaks in our spectra were only slightly less intense than the maxima and may be indicative of some fundamental differences in the marine organic matter compared to that from fresh water sources (28,29). However, the overall intensity of the fluorescence spectrum of the Great Bay sample was lower than the other samples by at least a factor of 4. The additional peaks may be due to trace component of the organic matter only measurable at the higher instrument sensitivities required for the sample. This trace component might also be present in the other samples, especially the pore water, but may not be discernible by being under the large and rather broad major peak. Fresh Water Cu2+Titrations. Fluorescence quenching titrations of the water samples with Cu2+ion proceeded much as our previous experiments with SFA (6). Figure 1 shows three replicate titration curves for the Oyster River. These data, although they are for the fresh water sample with the least binding ability, are representative of the results for all the fresh water samples. The reproducibility of replicate curves appears to be quite good, better than titrations of SFA with Cu2+(6). The reason for this may be the superior homogeneity of the bulk water samples compared to the SFA solutions. SFA experiments were conducted on solutions made fresh each day from powdered material that was probably heterogeneous, because solids are almost impossible to homogenize. The line through the points in Figure 1 is the best fit curve from the NONREG program using eq 8. The titration parameters (CL, K , and Im), averaged for the three runs, are listed in Table I along with the results for the other samples. The Great Bay sample did not give analyzable results as the NONREG program was unable to converge on the best fit parameters (see below). Analysis of the data in Table I requires consideration of the compositions of the samples. The CL values represent the potential complexing ability of the sample for Cu2+ion (15).The samples contain appreciable amounts of metals such as iron (Table I), which expend some of the ligands' total complexing capacity. Cu2+added to these samples will complex unoccupied binding sites on the
A
A H
v)
L
A
W ; 120 A
0
8ol
A
?i W
tc
A
A
A A
A
0
,
0.0
0.3
0.6
ADDED
C U ~ +
0.8
1.2
tn x 1041
Figure 3. Rayieigh scattering curve measured slmuitaneously with fluorescence titration B of the Great Bay sample with Cu2+ (see Figure 2).
organic matter and will also probably displace more weakly bound metal ions such as Zn2+,Cd2+, Ca2+,and Mg2+. More strongly binding ions such as A13+,Fe3+,and Pb2+ may compete successfully with Cu2+for sites. Near the end of the titration when Cu2+is in the large excess, A13+ and Fe3+may be the only metal ions capable of still competing. Marine Samples. Titrations of the Great Bay sample with Cu2+resulted in some significant differences from fresh water samples. In order to better characterize this sample's behavior, we monitored both emission wavelengths (398 and 430 nm) with excitation at 350 nm. Quenching occurred at both wavelengths but was considerably greater at 430 nm. Rayleigh scattering became significant very early in the titration in a region where fluorescence quenching with each successive addition of Cu2+was still relatively large. For this reason we titrated the sample beyond the region where scattering was double the Cu2+-freevalue. Figure 2 shows results for three titrations of the Great Bay sample and Figure 3 shows the scattering observed during titration B. Curves A and B in Figure 2 exhibit a marked increase in the measured signal ater going through a minimum. This behavior was not expected on the basis of our previous experience with humic material fluorescence (6, 8). In addition, our mathematical model (eq 8) is invalid for this situation. Environ. Sci. Technoi., Vol. 16, No. 12, 1982 860
Table 111. Correlations of Natural Water Characteristics‘ hardness
PH hardness alkalinity conductance color fluorescence DOC tFe1 CL
0.989
alkalinity
conductance
0.976 0.986
0.967 0.993 0.973
color -0.461 -0.361 -0.457 -0.258
fluorescence 0.017 0.049 -0.113 0.111 0.730
DOC
Pel
-0.041 0.028 -0.118
-0.595 -0.476 -0.510 -0.368 0.918 0.400 0.601
0.113 0.860 0.963
CL - 0.7 24 - 0.641 -0.709 -0.552 0.946 0.553 0.678 0.936
K
K 0.857 0.830 0.901 0.778 -0.778 -0.500 -0.553 -0.625 -0.914
IML 0.402 0.271 0.319 0.158 -0.930 0.553 -0.695 -0.975 -0.875 0.604
’Coefficients set in italics are statistically significant at the 90% confidence level. Careful comparison of Figures 3 and 2 shows that Rayleigh scattering is about 4 times its initial value in the region of the titration where the fluorescence signal goes through a minimum and then begins to increase. We suspected that the increase in signal after 30 pM total Cup+was due to scattered radition not rejected by the emission monochromator. To prove this and hopefully eliminate this interference, we conducted a third titration (C) using a 400-nm cut-off filter placed between the sample and the emission monochromator. The results shown in curve C of Figure 2 demonstrate that the cut-off filter successfully blocked the scattered radiation and gave the expected curve. However, the NONREG program did not give reasonable parameters for any of the replicate titrations conducted with the filter. There are several possible reasons for the distinctly different behavior of the Great Bay estuarine sample compared to the others. The organic matter concentration, as measured by DOC, color, and fluorescence, is extremely low, indicating a very small complexing capacity that may be unmeasurable by this technique or any other. The composition of this material may also be significantly different. Its origin may be predominantly from marine organisms compared to the terrestrial origin of fresh water organic matter (28,29). High concentrations of Ca2+and Mg2+in the Great Bay sample probably compete with Cup+ for the relatively few binding sites available. In addition, the high salt content and high pH aid flocculation and precipitation of humics as Cu2+is added (30). This causes the observed early onset of scattering. The pore water sample had a very low CLvalue, yet its organic matter concentration is the highest by far of all the samplex as measured by solution color, fluorescence, and DOC. The most likely reason for ita low CLvalue and high organic carbon content is the presence of nonbinding organic matter. It has been shown that marine sediment organic matter contains anywhere from 30 to 70% humic substances (31),and much of this may be associated with naturally present trace metals. Another interesting parameter for this sample is its unusually high I M L value (Table 1). The Imvalue is the lower limit of the sample’s fluorescence when all the available ligand is bound and its fluorescence quenched. This limiting fluorescence could be due to material in the sample that fluorescences but does not bind and is therefore not quenched (6). In these samples it may also be due to fluorescing complexes of strongly binding diamagnetic metal ions such as Pb2+and AI3+,which do not quench the fluorescence of humic materials as effectively as paramagnetic metal ions (8). The INILvalues for all the other samples were in the range 20-35%. This is similar to the results obtained for Cup+-SFA (6) and is probably due to the lower trace-metal loading of these samples. 870
Environ. Sci. Technol., Vol. 16, No. 12, 1982
Experiments carried out above pH 6 should probably be corrected for hydrolysis of Cup+. This is done by using the “side reaction coefficient” a (32), which is a measure of the extent of reactions other than the principal one for the species of interest. For hydrolysis a = 1 + [OH-]@, [OH-I2P2 (9)
+
where and P2 are the overall formation constants for the mono- and dihydroxy Cu(I1) species and [OH-] is the molar concentration of hydroxide ion. For our titrations, conducted at fixed pH values, a is a constant that, when carried through the derivation of eq 8 (6) gives the following equation: [CuL]/CL = c~K[CU~+]/(CUK[CU~+] + 1) (10) Thus, when corrected for hydrolysis of Cup+,eq 8 wczuld have an a K wherever K appears. This demonstrates that the C L values (Table I) obtained from the fluorescence technique are not affected by hydrolysis. Only the K values in Table I would be divided by a! to make the correction. This was not done because significant differences exist in literature values of PIand & from different sources (32-34). For this reason and the fact that the ionic strength of the samples is not known, the hydrolysis correction is only an approximation. Correlations. Each of the measured water sample characteristics listed in Table I was correlated with the other characteristics in a multiple correlation study. Table 111lists the correlation coefficients for the four fresh water samples only. The coefficients set in italics are statistically significant at the 90% confidence level. Among the significant correlations are those of pH, alkalinity, hardness, and conductance,which all correlate with each other. This behavior is predicted since these parameters measure the inorganic species in the samples. Higher alkalinity is expected to give higher pH and possibly increased conductivity. The presence of significant levels of the hardness ions, Ca2+and Mg2+,would be expected to contribute to the conductance. The origin of these ions from carbonate minerals gives rise to their correlations with alkalinity. Other correlations that are of importance include significant correlations between color, CL,and [Fe]. The color in these samples probably comes from two sources, organic matter and certain Fe(II1) species. Higher organic matter concentrations presumably give rise to increased CL values. The correlation between [Fe] and C L may indicate that much of the binding organic matter is associated with Fe as discussed by Giesy et al. (35) and Florence and Batley (1).
Dissolved organic carbon (DOC) correlated significantly with sample fluorescence. Smart et al. (27) found DOC measurements gave statistically significant correlations with fluorescence for natural water samples. Others (36)
have used this as the basis of a method to analyze humic materials. However, Ghassemi and Christman (37) found fluorescence to DOC ratios were not constant. Neither fluorescence nor DOC showed significant trends with color or CLvalue. These results are similar to those of Truitt and Weber (15) for DOC, color, and Cu2+CL values obtained from dialysis titrations of natural waters. The CL values represent the potential binding ability of the sample, but its correlation with parameters that measure organic matter content can be affected by many factors. The most important of these are probably trace-metal loading of the sample (15), the presence of inorganic ligands, and the variable amount of complexing organic matter making up the total organic composition of samples from different sources (38). The nonsignificant correlation between DOC and color that we observed is contrary to a previous report (18)using the same analytical methodology for measuring color. This may be due to our limited sampling. However, Stewart and Wetzel(39) concluded that DOC, absorbance, and fluorescencewere not linearly related in either isolated organic matter or the natural material found in 55 lakes. One reason for the absence of a significant correlation between the two optical parameters in both their study (39) and ours is that color was measured at 250 nm while fluorescence was excited at 350 nm. Greater sensitivity is achieved under these conditions, but the data would probably agree better if 350 nm were used for both. The lack of consistent results in the literature demonstrating relationships between DOC, fluorescence, color, or CL values (15,38,40) indicates that they do not measure the same properties. Therefore, simple generalizations are impossible. Certain correlations were unexplainable on the basis of our present knowledge of the natural water systems under study. In these cases statistical significance may be coincidental due to the limited number of samples used, There include the correlation of alkalinity and K and the negative correlation of color and Im Since all the samples were thoroughly purged with N2 before and during fluorescence titrations and acid was added to keep them at their original pH, it is difficult to explain relationships between any titration parameters and alkalinity. The complexing capacity titrations should indicate the behavior of the organic matter alone in the fresh water samples. The results for the pore water sample may be influenced significantly by borate ion, silicates, and other inorganics. The negative correlation of K and CL cannot be explained chemically but may be an artifact of the data treatment. From eq 8 it is obvious that if K gets smaller CLincreases and vice versa for a given set of data. This interrelationship may give rise to the correlation. The data treatment in this paper is a rather simple one from a theoretical standpoint and has several limitations when applied to this complex system. We propose it as a first attempt to describe fluorescence data. The same correlations discussed above and listed in Table I11 were made for all the data including the two marine samples. This gave four degrees of freedom (six cases) for most parameters but only three degrees of freedom for some, due to incomplete data for the Great Bay sample. Generally the same parameters were significant at the 90% confidence level with a few additions. The additional significant correlations were felt to be somewhat artificial due to the distinct differences between fresh water and marine samples. For example, the relatively high alkalinities of the marine samples and the high K value for the pore water caused these two parameters to correlate with nearly every other property. This effect
is not considered to have any chemical significance and was therefore ignored.
Conclusions The results described here demonstrate that the fluorescence quenching technique allows determination of Cu2+complexing capacities of natural water samples at the micromolar and submicromolar level. Our previous work indicates that these values are consistent with those measured by other techniques. The water samples offer a significantly greater challenge than laboratory prepared solutions of isolated SFA. The natural samples are more complex with respect to their inorganic constituents, and trace metals block and/or compete for binding sites. The natural waters had lower CLvalues than our 10 mg/L SFA; however, the fluorescence method is extremely sensitive and can be applied even into the submicromolar CLrange. The experiments described here were done under N2 in order to make pH control easier. This, however, is not a requirement. Neither C02 nor O2 interfere with the fluorescence (7). It is not necessary to alter the sample in any way, as in adding an electrolyte for example. Filtering is not critical if proper means are used to reject scattered radiation. Acknowledgments We thank William H. Orem for supplying the pore water, Lyndon Marble for conducting the water sample characterization, and C. L. Grant for assistance with statistical analysis.
Literature Cited Florence, T. M.; Batley, G. E. CRC Crit. Rev. Anal. Chem. 1980,9, 219-96. Hart, B.T. Environ. Technol. Lett. 1981,2, 95-110. Saar, R. A.; Weber, J. H. Environ. Sci. Technol. 1980,14, 877-80. Saar, R. A.; Weber, J. H. Geochim. Cosmochim. Acta 1980, 44,1381-83. Pagenkopf, G. K.;Whitworth, C. J. Inorg. Nucl. Chem. 1981, 43, 1219-22. Ryan, D. K.; Weber, J. H. Anal. Chem. 1982,54,986-990. Sholkovitz, E. R. Geochim. Cosmochim. Acta 1976, 40, 831-45. Saar, R. A.; Weber, J. H. Anal. Chem. 1980,52,2095-100. Schnitzsr,M.; Ghosh, K. Soil Sci. SOC.Am. J. 1981,45,25-9. Cline, J. T.; Holland, J. F. In “Biological Implications of Metals in the Environment”,Proc. Fifteenth Hanford Life Sci. Symp., Richland, WA, 1975, Tech. Info. Ctr. Energy Res. Development Admin., 977, pp 264-79. Gamble, D. S.;Underdown, A. W.; Langford, C. H.Anal. Chem. 1980,52, 1901-08. MacCarthy, P.; Smith, G. C. In “Chemical Modeling in Aqueous Systems”; E. A. Jenne, Ed.; ACS Symp. Ser. No. 93; American Chemical Society: Washington, D.C., 1979; pp 201-22. Laxen, D. P. H.; Harrison, R. M. Anal. Chem. 1981,53, 345-50. Flaschka, H. A. “EDTA Titrations: An Introduction to Theory and Practice”;Pergamon Press: New York, 1964; P 82 Truitt, R. E. Weber, J. H. Environ. Sci. Technol. 1981,15, 1204-8. Truitt, R. E.; Weber, J. H. Anal. Chem. 1979,51,2057-9. ”Standard Methods for the Examination of Water and Wastewater”, 14th ed.; American Public Health Association: New York, 1975. Reid, J. M.; Cresser, M. S.; MacLeod, D. A. Water Res. 1980, 14,525-9. Weber, J. H.; Wilson, S. A. Water Res. 1975,9,1079-84. Wilson, S.A.; Weber, J. H. Chem. Geol. 1979,26,345-51. Wilson, S.A.; Weber, J. H. Chem. Geol. 1977,19,285-93. Envlron. Sci. Technol., Vol. 16, No. 12, 1982
871
Envlron. Sci. Technol. 1982, 16, 872-880
Volk, W. “Applied Statistics for Engineers”, reprint ed.; McGraw-Hill: New York, 1980; pp 159-71. Grasehoff, K. “Methods of Seawater Analysis”; Verlag Chemie: New York, 1976; pp 31-34. Reuter, J. H.; Perdue, E. M. Geochim. Cosmochim. Acta 1977,41, 325-34. Zepp, R. G.; Schlotzhauer, P. F. Chemosphere 1981, 10, 479-86. Larson, R. A.; Rockwell, A. L. Arch. Hydrobiol. 1980,89, 416-25. Smart, P. L.; Finlayson, B. L.; Rylands, W. D.; Ball, C. M. Water Res. 1976,10, 805-11. Stuermer, D. H.; Payne, J. R. Geochim. Cosmochim. Acta 1976.40. 1109-14. Hatc‘her,’P. G.; Rowan, R.; Matingly, M. A. Org. Geochim. 1980.2. 77-85. Sholkoktz, E. R.; Copland, D. Geochim. Cosmochim. Acta 1981,45, 181-9. Nissenbaum, A.; Kaplan, I. R. Limnol. Oceanogr. 1972,17, 570-82. Ringbom, A.; Wanninen, E. In “Treatise on Analytical Chemistry”;Kolthoff, I. M., Elving, P. J., Eds.; Interscience:
New York, 1979; Vol. 2, Part I, p 461. (33) Lindsay, W. L. “Chemical Equilibria in Soils”, Wiley-Interscience: New York, 1979; p 224. (34) Baes, C. F., Jr.; Mesmer, R. E. “The Hydrolysis of Cations”; Wiley-Interscience: New York, 1976; pp 267-74. (35) Giesy, J. P.; Briese, L. A.; Leversee, G. J. Environ. Geol. 1978,2, 257-68. (36) Brun, G. L.; Milburn, D. L. Anal. Lett. 1977,10,1209-19. (37) Ghassemi, M.; Christman, R. F. Limnol. Oceanogr. 1968, 13, 583-97. (38) Langford, C. H.; Khan, T. R.; Skippen, G. B. Inorg. Nucl. Chem. Lett. 1979, 15, 291-5. (39) Stewart, A. J.; Wetzel, R. G. Limnol. Oceanogr. 1981,26, 590-7. (40) Shuman, M. S.; Woodword, G. P., Jr. Environmen. Sci. Technol. 1977, 11, 809-13.
Received for review February 24,1982. Accepted July 16, 1982. The National Science Foundation partially supported this work under Grant OCE 79-10571 and Grant CHE 79-08399 for the Cary 219.
Distribution of Selected Gaseous Organic Mutagens and Suspect Carcinogens in Ambient Air Hanwant B. Slngh,” LOUISJ. Salas, and Robln E. Stiles SRI International, Menlo Park, California 94025
An on-site field data collection program, based on short-term studies, was conducted in seven U.S. cities. Atmospheric concentrations, variabilities, and diurnal behaviors of 20 gaseous organic bacterial mutagens or suspect carcinogens are described. Except for benzene and formaldehyde, average concentration levels for all chemicals measured were in the 0-1-ppb range. Benzene and formaldehyde average levels were in the 14 and 10-20-ppb range, respectively. Typical diurnal profiles show highest concentrations during nighttime or early morning hours, with minimum concentrations in the afternoon hours; chemistry plays only a nominal role in defining this diurnal behavior in most cases. It is concluded that organic mutagens have always existed in the atmosphere (and the ocean), although at relatively low background concentrations. Our measurements for this group of 20 chemicals show that in the cleanest environments the present exposure is more than twice the natural background, whereas in the U.S. cities we studied exposure may be 15-30 times greater. W
Introduction In a recent report the surgeon general stated that ”Toxic chemicals are adding to the disease burden of the United States in a significant, although as yet not precisely defined way” (1). Estimates suggesting that 50-90% of human cancer may be of chemical origin persists (I,2). The degree to which synoptic and macro- and microenvironments individually contribute to human cancer is a matter of ongoing research and debate ( 3 , 4 ) . Although the risks may be highly uncertain, there is little doubt that significant quantities of a growing number of synthetic organic chemicals have been released into the ambient enviornment during recent decades. In many cases, virtually the entire quantity of the chemical manufactured is released into the environment as a necessary outcome of use (5,6). A key parameter in assessing risk from ambient exposure entails the characterization of ambient atmospheres in 872
Envlron. Scl. Technol., Vol. 16, No. 12, 1982
which the affected population resides. Because of the relatively recent interest in ambient hazardous chemicals, the atmospheric abundance, sources, and sinks of this group of pollutants are poorly understood. Although environmental episodes (e.g., the “Love Canal” incident) have received considerable attention (I), the extent of human exposure to chemicals in normal ambient atmospheres remains relatively poorly determined. The present study was initiated to measure selected organic chemicals in several US. cities. Although we measured 44 organic chemicals, results presented here are limited to 20 bacterial mutagens and suspect carcinogens. Table I lists chemicals that have been defined as bacterial mutagens or suspect carcinogens. References 8-14, shown in the last column of Table I, often refer to additional studies that support their findings. Other chemicals for which concurrent ambient data were collected but not included here are fluorocarbons F 12, F 11,F 113, and F 114, ethyl chloride, 1,l-dichloroethane, 1,1,1,2-tetrachloroethane, 1,2-dichloroethylene, monochlorobenzene, o-dichlorobenzene, m-dichlorobenzene, 1,2,4-trichlorobenzene, toluene, ethylbenzene, m- and p-xylenes, o-xylene; 4-ethyltoluene, 192,4-trimethylbenzene,1,3,5-trimethylbenzene, acetaldehyde,phosgene, peroxyacetyl nitrate, and peroxypropionyl nitrate. These excluded chemicals are not considered to be mutagenic or carcinogenic at the present time. The data can be found in ref 15. Empirical tests have shown that nearly 90% of tested animal carcinogens are also bacterial mutagens, while an equal percentage of noncarcinogens are nonmutagens (7). Bacterial mutagenicity tests are simple and direct and provide a useful screening test for carcinogenicity. The carcinogenicity information is based on tests involving epidemiology and a critical and comprehensive evaluation of carcinogenicity, mutagenicity, and other toxicological data (8-10). The terms ”bacterial mutagens” (BM) and “suspect carcinogens” (SC) as used here do not imply that a proven human health hazard exists; however, these chemicals are
00 13-936X/82/0916-0872$01.25/0
0 1982 American Chemical Society
