these metals is association with fine particles which can be expected to be deposited and resuspended many times by tidal currents. In the area of direct, alkaline, waste discharge Cd is less mobile than Ni, apparently because of the formation of an insoluble CdC03 phase. Baseline data are provided on the distribution of Cd and Ni in the sediments of a cove, which, because of localized, industrial pollution, can serve as a useful environment for the study of Cd and Ni biogeochemistry. The toxicity and potential health hazards to humans have been the subject of much recent study (15,16) and are most dramatically expressed in the outbreak of “itai-itai” syndrome in Japan (17, 18). While the results of this study do not prove the existence of a public health hazard due to Cd contamination, they do show that the potential for such a hazard exists, should be studied, and that, due to the mobility of Cd, concern for such a hazard must include the western as well as the eastern part of the cove.
(5) “United States District Court Records”, “United States of America vs. Marathon Battery Company”, Civil Suit 4110, Southern District of New York, 1970. ( 6 ) Bondeitti, E. A., Sweeton, F. H., Tamura, T., Perhac, R. M., Hulet, L. D., Kneip, T. J., “Characterization of Cadmium and Nickel Contaminated Sediments from Foundry Cove, New York”, Proc. of the National Science Foundation Trace Contaminants Conf., 1973. (7) Thomas. R. L.. Can. J . Earth Sci.. 10.194-204 11973). (8) Simpson, H. J., Olsen, C. R., Trier, R. hi.,Williams, S. C., Science, 194,179-83 (1976). (9) Iskander, J. K., Keeney, D. R., Enuiron. Sci. Technol., 8,165-70 (1974). (10) Tourtelot, H., Huffman, C., Rader, L., “Cadmium in Samples of the Pierre Shale and Some Equivalent Stratigraphic Units, Great Plains Region”, U.S. Geological Survey Professional Paper 475-D, 1964. (11) Edgington, D. N., Robbins, J. A., “The Behavior of Plutonium and Other Long-Lived Radionuclides in Lake Michigan”, in “Impacts of Nuclear Releases into the Aquatic Environment”, IAEA, Vienna. 1975. (12) Davis, J. C., “Statistics and Data Analysis in Geology”, Wiley, New York, N.Y., 1973. (13) Hem, J., Water Resour. Res., 8,661-79 (1972). (14) Garrels, R. M., Christ, C. L., “Solution, Minerals, and Equilibria”, Freeman, San Francisco, Calif., 1965. (15) Flick, D. F., et al., Enuiron. Res., 4, 71-85 (1971). (16) Fleischer, M., et al., Enuiron. Health Perspect., May, 253-323 (1974). (17) Kobayashi, J., “Relation Between the ‘Itai-Itai’Disease and the Pollution of River Water by Cadmium from a Mine”, Proc. of the 5th Int. Conf. on Water Pollution Res., 1970. (18) Namagata, N., “Manifestation of Cadmium Poisoning in Japan and Geochemical Approach to Environmental Pollution”, in Proc. of Symp. on Hydrogeochemistry and Biogeochemistry,E. Ingerson, Ed., pp 330-6, Tokyo, Japan, 1973.
Acknowledgment We thank H. Feely who served as thesis advisor for the senior author, R. Bopp and C. Olsen for providing some of the analytical data, and R. Cobler for help in field sampling.
Literature Cited (1) Kneip, T . J., “Cadmium in an Aquatic Ecosystem: Distribution
and Effects”, 1st Annual Progress Rep. to National Science Foundation, New York Univ. Inst. for Environmental Medicine, New York, N.Y., 1974. (2) Kneip, T. J., ibid., 2nd Annual Progress Rep., 1975. (3) Kneip, T . J., Hernandez, T., Re, G., “Cadmium in an Aquatic Ecosystem: Transport and Distribution”, in “Trace Contaminants in the Environment”, Proc. of the 2nd Annual NSF-RANN Trace Contaminants Conf. Asilomar, Pacific Grove, Calif., 1974. (4) Buehler, K., Hirshfield, H. I., “Cadmium in an Aquatic Ecosystem: Effects on Planktonic Organisms”, ibid.
Received far review December 15, 1976. Accepted December 23, 1977. Financial support provided under Environmental Protection Agency contract (R803113-01, 02,03).
Seasonal Trends in Denver Atmospheric Lead Concentrations Harry W. Edwards’ and Harovel G. Wheat Department of Mechanical Engineering, Colorado State University, Fort Collins, Colo. 80523
During the 42-month period from January 1972 to June 1975, monthly average atmospheric lead concentrations at five metropolitan Denver sampling sites displayed maxima during winter months that coincided with minima in mixing heights. Monthly atmospheric lead inputs estimated on the basis of city-wide consumption of leaded gasoline showed a general downward trend with maxima during summer months. Atmospheric lead concentrations correlated well with the dispersion factor, the product of the mixing height and the mean wind. Although the data suggested a long-term declining trend a t three sites, monthly atmospheric lead concentrations did not correlate with estimated monthly atmospheric lead inputs. The major source of lead in urban atmospheres is combustion of leaded gasoline ( I ) . Although there is a divergence of scientific opinion regarding some of the possible environmental and health effects of automotive lead (2, 3), there is widespread agreement that a general trend toward rising urban atmospheric lead exposures would be undesirable. The National Academy of Sciences Panel on Airborne Lead (1) 0013-936X/78/0912-0687$01.00/0
@
reported in 1972 that urban atmospheric lead concentrations in the United States had increased very little over several decades in spite of substantial increases in the consumption of lead antiknock additives. Shorter-term increasing trends have been reported for certain sites, however ( 4 , 5 ) . The general implication is that meteorological factors, urban geometry, and traffic routing may enter strongly in determining urban atmospheric lead concentrations. However, very little quantitative information has been compiled concerning the relative sensitivity of atmospheric lead concentrations in a specific urban center to changes in local lead additive consumption. The need for quantitative information has assumed major importance since regulations have been adopted to reduce ambient atmospheric lead exposures in the United States by phasing down the lead content of the total gasoline pool (6). These regulations are based upon a linear rollback formula, Le., changes in atmospheric lead concentrations are directly proportional to changes in lead additive consumption. While no totally satisfactory mathematical model has been developed for predicting pollutant dispersion within an urban complex, the atmospheric concentration of a pollutant emitted from an area source is often proportional to the emission rate and inversely proportional to the product of the mean wind
1978 American Chemical Society
Volume 12, Number 6, June 1978
687
and the vertical mixing height ( 7 ) .This relationship serves as the basis of the linear rollback model and is often useful for correlating trends in pollutant concentrations a t specific sites with trends in emission rates and meteorological parameters. Even though fine resolution in time and space is not always present, the model has been justified on the basis of simplicity and predictive success. Application of the linear rollback model to atmospheric lead presents a t least three difficulties. The first is that of accurately determining the lead source strength. Both the rate and size of lead particles emitted are markedly dependent upon the mode of operation and history of the vehicle (8-10). These studies indicate that the amount of lead emitted can range from 10 to 2000% of the lead burned. The latter value corresponds to full-throttle acceleration, which apparently can dislodge lead particles previously deposited in the exhaust system. Changing traffic patterns may also affect the spatial and temporal characteristics of the lead source strength. The second difficulty is that because lead is emitted primarily in particulate form, some gravitational settling of the larger particles takes place. This is evident from field studies which show that when samplers are placed a t least 100 m from a traffic source, further dispersion of the lead plume can be described by gaseous pollutant transport relationships ( 1 1 ) . Closer to the roadway, an empirical approach is needed to deal with the apparent weakening of the source strength with distance due to particle depletion (12). The third difficulty is that urban geometry can profoundly influence pollutant dispersion. Both field and wind tunnel studies demonstrate the inhomogeneous mixing of automotive emissions within a city street canyon ( 1 3 ) . In this paper we present data that show definite seasonal patterns in monthly average atmospheric lead concentrations in metropolitan Denver, Colo., during the 42-month period from January 1972 through June 1975. These data are examined in terms of the existence of statistically significant correlations between atmospheric lead concentrations, computed atmospheric inputs of automotive lead, and meteorological parameters. The metropolitan Denver area was selected for this study largely on the basis of its virtual isolation in terms of air pollution sources (14).In 1974 there were approximately 1.5 X 106 inhabitants and 1.1 X lo6 registered automobiles in the Denver metropolitan area. In addition to the automobile, restrictive meteorology enters strongly in the Denver air pollution situation, particularly during winter months when thermal inversions are common (15).There is no primary lead smelting in the Denver metropolitan area, and private incineration of solid waste has been prohibited. Natural gas is the dominant fuel for space heating. Electrical power is generated locally, largely by coal combustion.
Atmospheric Lead Inputs Monthly atmospheric inputs of automotive lead in the Denver metropolitan area were estimated on the basis of monthly consumption of leaded gasoline. Through its Oil Inspection Section, the Colorado Department of Labor and Employment maintains monthly records for tax purposes of the total consumption of all grades of gasoline sold a t retail in the Denver metropolitan area (16). Gasoline transported out of the Denver metropolitan area for resale elsewhere is not included. Gasoline sold in Boulder, Colorado Springs, Longmont, Greeley, Fort Collins, and other nearby cities outside of the Denver metropolitan area is also not included. Information concerning the average lead content of the total gasoline pool sold a t retail in the Denver metropolitan area was obtained from the Ethyl Corp., a major local supplier of lead additives (17). The average lead content of the total gasoline pool, including low-lead, unleaded, and all grades of leaded gasoline, 688
Environmental Science & Technology
is given in Table I. Lower lead contents during winter months reflect in part the effects of reblending gasoline feedstocks for improved winter starting characteristics. The product of the monthly gasoline consumption and the lead content of the gasoline pool thus yields the total monthly lead additive consumption. Data reported by Hirschler et al. (10)indicate that averaged over 50 000 miles of consumer driving conditions, an automobile typically exhausts 78% of the lead burned and retains 22% of the lead burned. Some of the exhausted lead is associated with large particles that would settle rapidly in still air but may be temporarily suspended by traffic-induced turbulence. Multiplication of the above product, e.g., monthly consumption of lead additives, by 0.78 yields the estimated monthly atmospheric input of automotive lead. These data are shown in Figure 1 and reflect a downward trend in lead additive consumption during the 42-month period of the study. The downward trend is consistent with increasing consumption of unleaded gasoline since introduction of catalyst-equipped automobiles in the fall of 1974. Computed atmospheric inputs of automotive lead in the Denver metropolitan area show peaks during summer months and minima during winter months. For purposes of comparison, monthly atmospheric inputs of lead from combustion of coal for electrical power generation were also estimated. Electricity consumed in the Denver metropolitan area is generated by the Public Service Co. of Colorado a t three plants, two of which, Arapahoe and Cherokee, are fueled by coal. The Zuni plant is fueled by natural gas on an interruptive basis with oil serving as the back-up fuel. The Zuni plant therefore does not enter into the computations of atmospheric lead inputs. For the coal-fired plants, monthly coal consumption and lead-content data were ob-
Table 1. Lead Content of Gasoline Sold in Denver, g/gal January-March April-September October-December
1972
1973
1974
1975
1.75 2.05 1.75
1.75 2.05 1.75
1.45 1.75 1.45
1.40 1.70
...
I . , . . . . . . . . . I .I..... JFMAMJJASONDJFMAMJJASONDJFMAMJJASONDJFMAMJ 1972 1973 1974 I975 .
.
.
.
.
.
.
I
.
.
Month
Figure 1. Computed atmospheric lead inputs in Denver from January 1972 to June 1975
tained on a unit-by-unit basis from the Public Service Co. of Colorado ( 1 8 ) .The ash content of the coal was 6-8%, and the average lead content of the ash was 0.0031%. These data then permitted cornputation of the monthly lead input to each of the coal-fired units. The particle collection efficiencies (mass basis) were obtained for each of the gas cleaning devices on a unit-by-unit basis. The collection efficiencies ranged from 90 to 99% depending on the type of particulate removal devices employed. Studies by Davison e t al. (19),Lee et al. (20),and Klein e t al. (21)provide ample evidence that the smaller fly ash particles which escape collection in a coal-fired power plant are markedly enriched in lead and other toxic elements. Using these studies as a guide, we have employed a factor of 10 for the lead enrichment ratio. The estimated monthly lead release by each coal-fired unit is then obtained as the product of the monthly ash production, the lead content of the ash, one minus the collection efficiency, and the lead enrichment factor of 10. Results of these computations are also shown in Figure 1 and indicate that the computed monthly atmospheric lead input from coal combustion is typically several hundred times smaller than the computed atmospheric lead input from gasoline combustion in the Denver metropolitan area. Even though great accuracy is not claimed for these estimates, the computations indicate that coal combustion contributed negligibly on a relative basis to Denver atmospheric lead burdens during the 42-month period of this study.
Table II. Average Daily Traffic Volume, Cars/Day Location
1971
State Health Dept. School Admin. Bldg. Cherry Creek Dam Arvada Adams City Englewood
2 150 a goo 1000 5 650 8 580 25 000
1975
2 350 9 600
1922 5 000
a ooo 24 400
State Health Department 2 .o
30
____
c
School Administration Building
Cherry Creek Dam
Atmospheric Lead Concentrations Monthly average atmospheric lead concentrations in metropolitan Denver were determined by the Colorado Department of Health (22)as a part of a broad air quality monitoring program. The six sites employed in this study provide a variety of traffic conditions, urban geometries, and locations. Approximate daily traffic volumes 011 streets nearby each site were compiled by the Colorado Department of Highways and are given in Table 111. These data bracket the 42-month period of the study and indicate that, with the exception of Cherry Creek Dam, traffic flow rates a t the beginning and end of the study were comparable. The Cherry Creek Dam sampling station is located in an outlying state recreation area where traffic volume is relatively light. Because of the prevailing southwesterly winds during daylight hours, the Cherry Creek Dam station is not normally in the path of the major pollutant plume from the city center. The School Administration Building is located in the central, downtown area. The State Health Department is located in a largely commercial area. The southern Englewood site represents a suburban area with relatively heavy traffic. The northern Arvada and Adams City sites represent suburban metropolitan Denver regions with moderate traffic volume. A t each site, conventional high-volume air samplers equipped with No. 25 Schleicher and Schuell, Inc. (Keene, N.H.), glass-fiber filters were operated for continuous 24-h periods every fourth day. Monthly composites for each site were prepared and analyzed for lead by atomic absorption spectrometry in accordance with the procedure reported by 'I'hompson et al. (23).Atmospheric lead measurements apply only to the filterable, particulate portion of the total lead content and do not include that fraction that can occur as an organic vapor ( 2 4 ) . Monthly average atmospheric lead concentrations a t each of the six sites are shown in Figure 2 for the 42-month period. With the exception of Cherry Creek Dam, monthly average atmospheric lead concentrations show definite maxima in winter months, the period during which computed atmospheric inputs of automotive lead show minima. Monthly average CO levels also show peaks during winter months ( 1 5 ) . This is also in sharp contrast with peak metropolitan Denver
Arvada
X
25
051 0
Adams City
25 -
20. IS-
10-
05-
1
Englewood
3.0
2.5
0
0'
i F M A M i J Asoirb;
i M A M i i ASOND:
1972
1973
FMAbi i A S O N D; 1974
F MAMJ
1975
Month
Monthly average atmospheric lead concentrations at six metropolitan Denver sites from January 1972 to June 1975
Figure 2.
gasoline sales during summer months. While peak atmospheric lead concentrations measured a t the School Administration Building and Eriglewood sites during 1974-1975 winter months are definitely lower than in the previous two winters, peak levels a t the Arvada, Adarns City, and State Health Department sites are not greatly different. If the concentrations are averaged on a yearly basis, declining trends are apparent a t the School Administration Building, Arvada, and Englewood. At the Cherry Creek reference site, a definite upward trend in monthly average atmospheric lead concenVolume 12, Number 6, June 1978 689
trations is apparent. This may reflect the effect of increasing traffic volume associated with heavier use of the recreation facilities.
3400
E 3200
-
R
Meteorological Data Monthly mean afternoon atmospheric mixing heights for the Denver metropolitan area were obtained from the Colorado Department of Health ( 2 5 ) .Soundings were made in accordance with the method reported by Holzworth (26).In Denver the mixing height typically rises continually from morning through afternoon and decreases with nighttime cooling. The monthly mean morning mixing height shows very little seasonal variation and averages about 250 m. In contrast, the monthly mean afternoon mixing height shows a pronounced seasonal pattern as shown in Figure 3. For the purposes of this study, the afternoon mixing height is used as an indicator of the dilution volume, and no assumption is made concerning uniformity of mixing below the inversion ceiling. The general trend emerging from these data is that afternoon mixing heights show pronounced minima during the winter, the period during which metropolitan Denver atmospheric lead concentrations tend to show maxima. Correspondingly, afternoon mixing heights reach maximum values during summer months, the period during which minimum atmospheric lead concentrations are observed. Monthly mean wind speeds were obtained from data compiled by the National Oceanic and Atmospheric Administration (27) and are shown in Figure 4. Monthly mean wind speeds are fairly steady throughout the year. Wind measurements were made a t Stapleton International Airport located on the eastern periphery of the Denver metropolitan area. Because of the effects of topography, urban heating, and urban geometry, circulation patterns may differ significantly throughout the Denver metropolitan area. Nevertheless, the mean wind speed determined a t Stapleton International Airport serves as a useful overall indicator of pollution dispersion conditions. The product of the afternoon mixing height (m) and the wind speed (m s-1) is employed as an air pollution dispersion factor by the Colorado Department of Health for the Denver metropolitan area (25).Pollution dispersion conditions are considered excellent when the dispersion factor is a t least 60 X lo2 m2 s-l and very poor when the value of the dispersion factor is less than 20 X l o 2 m2 s-l. Kneip and coworkers (28-30) used a similar dispersion factor to correlate atmospheric trace element and suspended particulate concentrations in New York City with meteorological conditions. Statistical Analysis The method of partial correlations (31) was employed to quantitatively evaluate correlations among the three variables, e.g., monthly atmospheric lead concentrations, computed monthly atmospheric inputs of automotive lead, and monthly dispersion factors. This technique permits quantitative evaluation of the association between the dependent factor, atmospheric lead concentrations, and each of the two independent factors while eliminating any tendency of the remaining independent factor to obscure the relation (32).In other words, this technique permits the correlation between atmospheric lead concentrations and computed automotive lead inputs to be evaluated under conditions of constant dispersion conditions. Both the simple and partial correlation coefficients between monthly atmospheric lead concentrations a t each of the six sites and computed atmospheric inputs of automotive lead are given in Table 111. The correlation analysis was repeated for a pool-averaged atmospheric lead concentration. This quantity was computed by first excluding Cherry Creek Dam as an outlyer and then averaging the remaining five sites for a single atmospheric lead value each 690
Environmental Science & Technology
2 -x
z
5
,,,,,,,,, 800 600
P
: .; ,,,,,,,,
, , , , , , , , , , ~, , , ,
5 400 200 JFMAMJJASONDJFMAMJJASONDJFMAMJJASONDJFMAMJ 1972 1973 1974 1975
Month
Figure 3. Monthly mean afternoon mixing heights at Stapleton International Airport from January 1972 to June 1975
1972
1973
1974
1975
Month
Figure 4. Monthly mean wind speeds at Stapleton International Airport from January 1972 to June 1975 month. The negative values of the simple correlation coefficients in part reflect the confounding effect of varying meteorological parameters. When the dispersion factor d is held constant, monthly average atmospheric lead concentrations still do not correlate with computed monthly atmospheric inputs of automotive lead during the 42-month period of this study. The dominant effect of meteorological parameters is apparent when the correlation between atmospheric lead concentrations and dispersion factors is examined. These data are given in Table IV and demonstrate that the relationship between atmospheric lead concentrations and dispersion factors is highly significant. When the computed lead release Q(Pb) is held constant, correlation coefficients a t the five urban sites (excluding Cherry Creek Dam) range from -0.66 to -0.80. A t these five sites, there is less than one change in one thousand that the correlation is not significant; 44-64% of the variability in atmospheric lead concentrations is attributable to variability in dispersion factors. When the data for the five sites are pooled, variability in the dispersion factor alone accounts for 69% of the variability in the aggregated atmospheric lead concentration. Discussion Metropolitan Denver atmospheric lead concentrations show a definite seasonal pattern a t five of the six sites and correlate
Table 111. Atmospheric Lead Concentrations and Automotive Lead Release: Correlation Coefficients
a
Location
r
Partlal r. d const
State Health Dept. School Admin. Bldg. Cherry Creek Dam Arvada Adams City Englewood Pooled dataa
-0.2449 -0.1556 -0.3764 -0.3439 -0.3537 -0.2388 -0.3014
-0.0218 0.1079 -0.3148 -0.1804 -0.2095 0.0126 -0.0895
Excluding Cherry Creek
Dam.
Table IV. Atmospheric Lead Concentrations and Dispersion Factors: Correlation Coefficients
a
Location
r
Partial r , O(Pb) const
State Health Dept. School Admin. Bldg. Cherry Creek Dam Arvada Adams City Englewood Pooled dataa
-0.7657 -0.7450 -0.2975 -0.7854 -0.6950 -0.8130 -0.8468
-0.7484 -0.7410 -0.2080 -0.7614 -0.6596 -0.8003 -0.8314
Excluding Cherry Creek
Dam
well with meteorological parameters affecting air pollution dispersion. At these five urban sites, winter maxima in monthly average atmospheric lead concentrations coincide with minima in mixing heights. The statistical analysis indicates that meteorological parameters enter strongly, with about half of the variability in atmospheric lead concentrations associated with variability in the dispersion factor. The relationship is slightly stronger when Cherry Creek Dam is excluded and the data from the five urban sites are pooled. In terms of long-term trends, yearly averages suggest that atmospheric lead concentrations are declining a t three of the sites but rising a t the s u b u r b a n reference site. The trend toward rising concentrations a t the Cherry Creek Dam sites appears linked to an increase in nearby traffic volume. The effects of the downward trend in lead additive consumption are thus not uniform throughout the Denver metropolitan area. The data of this study suggest a definite limitation of the linear rollback model as a method for predicting month-tomonth changes in urban atmospheric lead concentrations on the basis of short-term changes in local lead additive consumption. Monthly average atmospheric lead concentrations do not correlate with a simple linear proportion of monthly lead additive consumption, even when effects due to variability in the dispersion factor are statistically removed. The existence of such a correlation would have been useful for predictive purposes, particularly for anticipating the shortterm rate of reduction in urban atmospheric lead exposures in response to a given gasoline-lead reduction schedule. The absence of a significant short-term relationship implies that the short-term variability in other parameters, including actual automotive lead emission factors and traffic patterns, masks the variability due to changes in lead additive consumption during the study. However, this does not preclude the possibility that the linear rollback model may still apply to automotive lead over the long term of several years to decades.
Acknowledgment The authors acknowledge the cooperation and assistance of the Colorado Petroleum Council; Ebhyl Corp.; Colorado Departments of Health, Labor and Employment, and Highways; Public Service Co. of Colorado; and the Denver Chamber of Commerce. The valuable advice and assistance provided by T. J. Boardman and M. C. Bryson of the Colorado State University Department of Statistics are also acknowledged. Literature Cited (1) National Academv of Sciences. “Airborne Lead in Persuective”.
NAS, Washington, D.C., 1972. (2) Greenfield, S. M.. Bridbord, K., Barth. D.. Enael. K.. “The Changing Perspectives Regarding Lead As An Environmental Pollutant”, in Proceedings of the Int. Symp. on Environmental Health Aspects of Lead, pp 19-26, Commission of the Eurupean Communities, Luxembourg, 1973. (3) Dinman, B. D., “Airborne Lead in Perspective”, ihid., pp 45 57. (4) Chow, T . J., Earl, J. L., Science, 169,577-80 (1970). (5) Tepper, L. B., Levin, L. S., “A Survey of Air and Population Lead Levels in Selected American Communities”, PB-222-459, National Tech. Information Service, U.S. Dept. of Commerce. Springfield, Va., 1972. (6) Fed. Regist., 38 (2341, 33734-41 (1973). (7) Wanta, R. C., in “Air Pollution”, A. C. Stern, Ed., Vol I , 2nd ed., p 216, Academic Press, New York, N.Y., 1968. (8) Hirschler, D. A., Gilbert, L. F., Lamb, F. W., Niebylski, L. M.,Irid. Eng. Chern., 49,1131-42 (1957). (9) Ter Ham, G. L., Lenane, D. L., Hu, J. N., Brandt, M., J . Air Pullut. Control Assoc., 22, 39-46 (1972). (10) Hirschler, D. A., Adams, W. E., Marsee, F. J., “Lean Mixtures, Low Emissions and Energy Conservation”, AM-73-15, National Petroleum Refiners Assoc., Washington, D.C., 1973. (11) Katen, P . C., “Intermediate Range Transport of Lead from Automotive Sources”, in Proceedings of the Second Annual NSF-RANN Trace Contaminants Conf., Lawrence Berkeley Lab Publ. 3217, pp 89-94,1974. (12) Katen, P . C., “Modeling Atmospheric Dispersion of Lead from Automotive Sources”, in Proceedings of the First Annual NSF Trace Contaminants Conf., Oak Ridge National Lab Publ. 730802. pp 298-313,1973. (13) Lombardi, D. J., Thompson, R. S., Cermak, J. E., “Physical Modeling of Automotive Emissions in a City Street Canyon.” in Proceedings of the Second Annual NSF-RANN Trace Cuntaminants Conf., Lawrence Berkeley Lab Publ. 3217, pp 116-22, 1974. (14) Riehl, H., Herkhof, D., “Weather Factors in Denver Air Pollution”, Paper No. 158, Dept. of Atmospheric Science, Colorado State Univ., Ft. Collins, Colo., 1970. (15) Air Pollution Control Division, Colorado Dept. of Health, “Assessment of Air Quality-Metro Denver Region”, Denver, Colo., 1975. (16) Koch, Oliver, Oil Inspection Section, Colorado Dept. of Labor and Employment, Denver, Colo., private communication, 1975. (17) Laubach, J. H., Ethyl Corp., Denver, Colo., private coniniunication, 1975. (18) Brines, H. G., Public Service Co. of Colorado, Denver, Colo., private communication, 1975. (19) Davison, R. L., Natusch, D.F.S., Wallace, J. R., Evans, C. A,. Jr., Enuiron. Sci. Technol.. 8. 1107-13 (1974). (20) Lee, R. E., Jr., Crist; H. L., Riley, A. E., MacLeod, K. E., ihld , 9,643-7 (1975). (21) Klein, D. H., Andren, A. W., Carter, J. A., Emery, J. F., Feldman. C., Fulkerson, W., Lyon, W. S., Ogle, J. C., Talmi, Y.. Van Hook, R. I., Bolten, N., ibid., pp 973-9. (22) Reynolds, L. T., Air Pollution Control Div., Colorado Dept. of Health, Denver, Colo., Drivate communication. 1975, (23) Thompson, R. J., Morgan, G. B., Purdue, L. G., A t Ahsurp. Newsl., 9,53-7 (1970). (24) Purdue, L. J., Enrione, R. E., Thompson, R. J.. Bonfield, B. A,, Anal. Chern., 45,527-30 (1973). (25) Retallack, W. G., Air Pollution Control Div., Colorado L)eut. of‘ Health, Denver, Colo., private communication, 1975. (26) Holzworth, G. C., J . A p p l . Meteorol., 6, 1039-44 (1967). (27) National Oceanic & Atmospheric Admin., Environmental Uata Service, U.S. Dept. of Commerce, “Local Climatologicdl Data, Denver, Colorado”, National Climatic Center, Asheville, N.C., 1972-75. (28) Kleinman, M. T., Kneip, T. J., Eisenbud, M., “Meteorological Influences on Airborne Trace Metals and Suspended Particulates”,
Volume 12,Number 6,June 1978 691
in “Trace Substances in Environmental Health-VII”, D. D. Hemphill, Ed., pp 161-6, Univ. of Missouri Press, Columbia, Mo., 1973. (29) Kleinman, M. T., Bernstein, D. M., Eisenbud, M., Kneip, T. J., “Seasonal and Source Relationships for Urban Suspended Particulate Matter and Trace Element Concentrations in New York City”, Paper No. 75-14.4,presented at 68th Annual Meeting of the Air Pollution Control Assoc., Boston, Mass.,June 1975. (30) Kleinman, M. T., Kneip, T. J.,Eisenbud, M., Atmos. Enuiron., 10,9-11 (1976).
(31) Snedecor,G. W., Cochran, W. G., “Statistical Methods”, 6th ed., p 400, Iowa State Univ. Press, Ames, Iowa, 1967. (32) Ezekiel, M., Fox, K. A., “Methods of Correlation and Regression Analysis”, 3rd ed., p 192, Wiley, New York, N.Y., 1959.
Receiued for reuieui June 28, 1976. Accepted December 27, 1977. Study supported by the National Science Foundation under Grant AEN74-17624 A01.
Isolation and Identification of Some Thermal Energy Analyzer (TEA) Responsive Substances in Drinking Water Tsai Y. Fan*, Ronald Ross, and David H. Fine Thermo Electron Research Center, 85 First Avenue, Waltham, Mass. 02154
Lawrence H. Keith’ and Arthur W. Garrison Environmental Research Laboratory, Environmental Protection Agency, Athens, Ga. 30605
Thermal Energy Analyzer (TEA)-responsive substances were found in the methylene chloride extracts of finished water samples from the water treatment plants in Cincinnati, Washington, and Philadelphia when they were analyzed by high-pressure liquid chromatography. The TEA-responsive substances, one from Cincinnati, two each from Washington and Philadelphia, were isolated and characterized by various instrumental techniques. Ethylene glycol dinitrate was identified in the isolates from the finished water in all three cities. The results of this study suggest that TEA response in any given sample should not be taken as presumptive evidence for the presence of a n N-nitroso compound. Independent identification techniques are needed t o confirm the identity of the suspected compounds. The Thermal Energy Analyzer (TEA) is a n N-nitroso detector that is based on the catalytic cleavage of the thermally labile N-NO bond and the subsequent detection of the nitrosyl radical by its reaction with ozone ( I ) . T h e TEA can be interfaced to a gas chromatograph (TEA-GC) for volatile N-nitroso compound analysis ( 2 ) or t o a high-performance liquid chromatograph (TEA-HPLC) for analysis of both volatile and nonvolatile N-nitroso compounds ( 3 ) .Although not exclusively specific for N-nitroso compounds, only N nitroso and some C-nitroso compounds, organic nitrites, and organic nitrates generally exhibit cleavage of the nitrosyl radical under the conditions of the TEA detector. T h e detector is also highly sensitive. Determination of volatile N nitrosamines in water a t the part per trillion level has been demonstrated ( 4 ) . Dimethylnitrosamine (DMN), probably of industrial origin, has been detected at levels of 0.08-2.7 pg/L in Curtis Bay, Baltimore by TEA-GC ( 5 ) . Many naturally occurring and man-made precursors of N-nitroso compounds, i.e., amines and nitrite, are present in water where N-nitroso compounds may be formed. TEAHPLC revealed the presence of at least 24 TEA-responsive compounds in drinking water from New Orleans, La., although no responses were obtained with this water a t a detection level of 0.002 wg/L using TEA-GC (6).The present communication reports the isolation and identification of several TEA-responsive compounds from drinking water in Cincinnati, Washington, D.C., and Philadelphia. Present address, Radian Corp., Austin, Tex. 78766. 692
Environmental Science & Technology
Experimental Apparatus. The high-performance liquid chromatograph (HPLC) was constructed by combining a high-pressure pump (Model 6000A, Waters Associates, Milford, Mass.) with an injector (Model U6K, Waters Associates). pPorasil or pBondapak CN columns were used (30 cm X 4 mm, Waters Associates). The effluent from the column was connected to either a UV detector (Model 440, Waters Associates) monitoring absorbance a t 254 nm or a Thermal Energy Analyzer (Model 502LC, Thermo Electron, Waltham, Mass.). Solvents. Glass-distilled solvents (Burdick and Jackson, Muskegan, Mich.) were used throughout this study. Water Sources. Drinking water samples a t various treatment stages were collected a t the water treatment plants in Cincinnati (May 4-7, 1976), Washington, D.C. (May 11-14, 1976) and Philadelphia’s Torresdale, Belmont, and Queen Lane plants (May 17-19, 1976). The Ohio River, Potomac River, and both the Delaware and Schuylkill Rivers supply raw water to treatment plants of Cincinnati, Washington, and Philadelphia, respectively. Although there is some variation in each plant, the method of water treatment is similar. The water samples were defined as follows: raw water is water pumped from the river to a reservoir for treatment. I n f l u e n t is raw water that has been chlorinated and flocculated and is t o be filtered. E f f l u e n t is influent that has been filtered through beds of sand and gravel. Finished water is effluent that has been treated chemically to be suitable for public consumption. The treatment may include chlorination, metaphosphate treatment, fluoridation, and ammonia treatment. T a p water is finished water collected a t consumer outlets. Analysis. Water samples (1L) were extracted three times with 50 mL of dichloromethane (DCM). The DCM layers were drained, combined, and concentrated in a Kuderna-Danish evaporator to less than 1 mL. The concentrated extracts were analyzed by TEA-HPLC using a pPorasil column with acetone/hexane (5/95) as the elution solvent a t a flow rate of 2 mL/min. Preparative Isolation of TEA-Responsive Substances Preparation of XAD Resin Cartridges. XAD-4 resin (Rohm and Haas, Philadelphia, Pa.) required extensive washings and sizing before it could be used. A slurry of XAD-4 resin in 5% aqueous sodium carbonate was allowed t o stand a t least 4 h. The slurry was filtered, and the resin rinsed suc-
0013-936X/78/0912-0692$01 .OO/O @ 1978 American
Chemical Society