Table V. Nonionic Surfactant Concentrationa (ppm) of Sewage Treatment Plant Influent Lab
Llght rolldr
Heavy rolldr
1
4.92 4.98 5.16 4.18 3.90 3.76 4.88 5.13 5.53 4.94 4.94 5.10 5.09 5.52 5.37 4.89 f 0.53
6.30 6.24 6.42 5.88 4.26 3.96 5.19 5.08 5.17 6.30 5.50 5.74 6.96 7.03 6.96 5.79 f 0.91
2
3
4
5 A v f l
(r
9econd decimal figures are presented for statistical considerations only. In no way does it indicate the number of analytically significant figures. One figure after the decimal point is recommended.
The results of this investigation are presented in Table V. The interlab relative standard deviations were f l l %for the “light solids” solution and f 1 6 % for the “heavy solids” solution. The intralab relative standard deviation was better than f10% in almost all cases. Recovery of nonionics was 90% with four sublations. This observation is based upon an additional 10% recovery with two more sublations (a total of six). Results with this precision and accuracy are very acceptable for environmental monitoring and for biodegradation studies. The
concentration differences noted for “light” and “heavy” solids demonstrate the loss in CTAS that occurs if one filters a sample before analysis. Conclusions An improved analytical method for nonionic surfactants has been developed and is recommended for biodegradation and environmental studies. An improved sublation procedure and an ion-exchange step have expanded the method applicability to virtually any aqueous sample. The CTAS analysis was demonstrated to be a reasonable and acceptable alternative to the DAS analysis. The method possesses acceptable accuracy and precision and has the proven advantage of simplicity and applicability to biodegradation and environmental samples.
Literature Cited (1) SDA Scientific and Tech. Rep. No. 6. “The Status of the Biodegradability Testing of Nonion 2 pm) by incorporating a third log-normal function into the size distribution model g(D,,2 ) and by using sensors responsive to the coarse particle size range. The addition of a coarse particle size distribution mode represents a simple extension of this measurement concept because of the observed independence of the two fine particle size distribution modes from the coarse particle mode ( I , 7, IO). The coarse particle mode has not yet been characterized sufficiently to allow both the mean size and geometric standard deviation to be fixed with confidence. Thus, at least two additional aerosol sensors would be needed to characterize this mode. A variety of techniques are possible. One possibility is the use of two mass monitors (19)coupled with impactors for size discrimination. The size distribution of Figure 4 was measured aboard an aircraft in the Labadie power plant plume 10 miles from the stack west of St. Louis, Mo. For this aerosol N T was 62 600 cm-3, ST was 1200 l.tm2/cm3,VT was 41.5 pm3/cm3, and b,, was 2.83 X 10-4 m-1. ST from the inversion was 9.56%higher than from the direct measurement with an uncertainty range of 13.3%. VT was 7.39% higher with an uncertainty range of 14.9%. The sensitivity plot of Figure 4 (c) indicates that the size range 0.05-0.4 pm contributed the most to the value of the aerosol charger output, whereas in Figure 3 (c) the size range 0.01-0.04 pm contributed the most to the charger output. This reflects the relative importance of aerosol surface in these size ranges for the two distributions as can be seen in Figures 3 (b) and 4 (b). The size distribution of Figure 5 was sampled two blocks
tI
1i
Figure 3. Comparison of number (a), surface (b), and volume (d) weightings of submicron size distribution obtained with parametric and direct measurement Histograms obtained with EAA and optical particle counter. Smooth curves obtained by parametric measurement.Sensitivity of three aerosol sensors also shown (c). Distribution representative of background urban conditions
*lo( f 3
t
1
I
0
W
- WPOMElER5
DP
- flIcROnErER5
Figure 4. Comparison of results of parametric and direct size distribution measurements in power plant plume
south of downtown Minneapolis. A strong dry wind blew dilute combustion aerosol from the downtown area toward the instrumentation. For this aerosol NT was 746 000 ~ m - ST ~ , was 422 prn2/cm3, VT was 6.40 pm3/cm3, and b,, was 0.312 X 10-4m-1. ST from the inversion was 14.9% high with an uncertainty range of 3.51%. V T from the inversion was 0.63% high with an uncertainty range of 3.88%. For aerosols strongly influenced by combustion sources, the CNC and aerosol charger sensitivities peaked in the 0.008-0.02 pm size range leaving a gap in terms of peak sensor response in the intermediate 0.02-0.1 pm size range [e.g., see Figure 5 (c)]. Since most of the information from the three sensors was for the
smaller sizes for this type of aerosol, the mean size of the accumulation mode was held fixed during the minimization procedure. The mean size of the nuclei mode was varied. The ability of this parametric measurement method to produce results of the quality shown in Figures 3-5 for three submicron size distribution weightings is due to the choice of an appropriate model distribution g(D,, j i ) and aerosol sensors. The composite distribution of two log-normal components is able to describe the number, surface, and volume weightings of the size distribution. The aerosol sensor response was matched to the requirements of the problem. The CNC, electrical aerosol charger, and integrating nephelometer, Volume 1 1 , Number 13, December 1977
1175
Xld.6,
II.
VI\\ Figure 5.
I
I
\
\
i
Comparison of results of parametric and direct size distribution measurements for aerosol heavily influenced by automotive combus-
tion which the calibrations and sensitivity analysis showed to be most sensitive to the number, surface, and volume weightings, respectively, were used to monitor the submicron aerosols in the size ranges containing peak number, surface, and volume concentrations, respectively.
Acknowledgment The helpful discussions with B. K. Cantrell, B.Y.H. Liu, J. C. Wilson, D.Y.H. Pui, and J. L. Wolf are sincerely appreciated.
Summary A parametric aerosol size distribution measurement system was developed. Using this measurement method the number, surface, and volume weightings of 15 submicron aerosol size distributions were determined from the output of three continuous, integral aerosol sensors and the relative humidity. The data inversion procedure involved calculating the theoretical sensor responses based upon a size distribution model. A composite distribution made up of two log-normal function components was selected because of its ability to model the three weightings of the size distribution. The parameters of the model distribution were then determined by adjusting them to minimize the sum of the relative differences between the experimental and theoretical sensor outputs. The results of the 15 measurements demonstrate that the parametric measurement gives as good results for NT, ST, and VT, and for b,, in the submicron size range as does a direct size distribution measurement using an EAA and OPC. NT and b,, were identical for the two methods of measurement. ST and VT were 6.19 and 12% higher, respectively, for the parametric measurement with relative standard deviations of 11.7 and 16%. This is well within the estimated uncertainty of 30% ( I ) for VT from a direct measurement. Because this parametric method of submicron size measurement uses only continuous sensors, it should be useful for submicron aerosol monitoring applications in which fast time response is necessary. It is ideally suited for use in aircraft sampling. This study utilized commercial sensors to show the applicability of the parametric measurement method. Less costly sensors, when properly calibrated, could be used. Such an instrument package would be highly desirable in submicron aerosol studies for which the direct measurement methods are too complex or expensive.
Literature Cited
1176
Environmental Science & Technology
(1) Whitby, K. T.,Clark, W. E., Marple, V. A., Sverdrup, G. M., Sem, G. J., Willeke, K., Liu, B.Y.H., Pui, D.Y.H., Atmos. Enuiron., 9,463 (1975). (2) Dzubay, T. G., Stevens, R. K., Enuiron. Sei. Technol., 9, 663 (1975). (3) Middleton, W.E.K., “Vision Through the Atmosphere”, University of Toronto Press, Toronto, Ont., Canada, 1958. (4) Sverdrup, G. M., PhD thesis, University of Minnesota, 1977. (5) Jaenicke, R., J . Rech. Atmos., 8,723 (1974). (6) Heintzenberg, J.,J . Aerosol Sci., 6,291 (1975). (7) Whitby, K. T., “Modeling of Atmospheric Aerosol Particle Size Distributions”, Progress Rep. EPA Grant No. R800971, Dept. of Mechanical Eng., University of Minnesota, 1975. (8) Chandler, J. P., “SUBROUTINE SIMPLEX”, available from Quantum Chemistrv Program Exchange, Dept. of Chemistry, In&ana University, 1965. (9) Nelder, J . A,, Jr., Mead, R., Comput. J., 7,308 (1965). (10) , , Whitbv. K. T.. Kittelson. D. B.. Cantrell. B. K.. Barsic. N. J., Dolan, D,“F.,Tarvestad, L. D., Nieken, D. J.,’Wolf,J. L., Wood, J. R., “Aerosol Size Distributions and Concentrations Measured During the GM Proving Grounds Sulfate Study”, Particle Technol. Lab. Publ. No. 286, University of Minnesota, 1976. (11) Liu, B.Y.H., Pui, D.Y.H., J . Colloid Interface Sei., 47, 155 (1974). (12) Liu, B.Y.H., Pui, D.Y.H., J . Aerosol Sci., 6,249 (1975). (13) Mie, G., Ann. Phys., 25, Series 4, 377 (1908). (14) Dave, J. V., Subroutines for Computing the Parameters of the Electromagnetic Radiation Scattered by a Sphere,” IBM Rep. No. 320-3237, IBM Scientific Center, Palo Alto, Calif., 1968. (15) Winkler, P., Junge, C. E., J . Rech. Atmos., 6,617 (1972). (16) Hanel, G., Tellus, 20,371 (1968). (17) Hanel, G., Beitr. Phys. Atmos., 44, 137 (1971). (18) Tuomi, T. J., Contrib. Atmos. Phys., 48,159 (1975). (19) Lundgren, D. A., Carter, L. D., Daley, P. S., “Fine Particles,” B.Y.H. Liu, Ed.. p 485, 1976. I
_
Received for recieu: March 8, 1977. Accepted June 24, 1977. Work supported by EPA Grants R-800971-01 and R-803851-02 and by a grant from the University of Minnesota Computer Center.