C. G.; Slinn, W. G. N.; Kelly, T. J.; Daum, P. H., Delany, A. C.; Greenberg,J. P.; Zimmerman, P. R.; Boatman, J. F.; Ray, J. D.; Stedman, D. H. Science (Washington, D.C.) 1987, 235, 46G-464. Vaughan,G. M. Master’s Thesis, University of Miami, 1989. Kieber, R. J.; Zhou, X.; Mopper, K. Limnol. Oceanogr. 1990, 37, 1503-1515. Zhou, X.; Mopper, K. Environ. Sci. Technol. 1990, 24,
oratory, Upton, NY; National Technical Information Service, Springfield, VA; 1985. Boyle, P. F.; Ross, J. W.; Synnott, J. C.; James, C. L. In Impact of Acid Rain and Deposition on Aquatic Biological Systems; Isom, B. G., Dennis, S.D., Bates, J. M., Eds.;
ASTM: Philadelphia, PA, 1986; pp 98-106.
Operations and Maintenance Manual for Precipitation Measurements Systems; EPA-600/4-82-042b;U.S. Environmental Protection Agency, Research Triangle Park, NC,
1981. Fitchett, A. W. In Sampling and Analysis of Rain; Campbell, S. A., Ed.; ASTM: Philadelphia, PA, 1983;pp 29-40. Cogbill, C. V.; Likens, G. E. Water Resour. Res. 1974,10, 1133-1137. Gorham, E.; Martin, F. B.; Litzau, J. T. Science (Washington, D.C.) 1984, 225, 407-409. Downie, N. M.; Starrie,A. R. Descriptive and Inferential Statistics; Harper and Row: New York, 1977. Dickerson, R. R.; Huffman, G. J.; Luke, W. T.; Nunnermacker, L. J.; Pickering, K. E.; Leslie, A. C. D.; Lindsey,
1864-1869. Robinson, E. In The Acidic Deposition Phenomenon and Its Effects; EPA-600/8-83-016AF;Altshuller, A. P., Linthurst, R. A., Eds.; US.Environmental Protection Agency, Washington, DC, 1984; Volume 1, Chapter 2, Section 2. Charlson, R. J.; Rodhe, H. Nature (London) 1982, 295, 683-685.
Received for review March 19,1991. Revised manuscript received June 26,1991. Accepted June 28,1991. The work was supported by N S F Grant ATM-8813353. I t is Contribution No. 37 from the Center for Marine Science Research at the University of North Carolina at Wilrnington.
Interlaboratory Comparison of Thermospray and Particle Beam Liquid Chromatography/Mass Spectrometry Interfaces: Evaluation of a Chlorinated Phenoxy Acid Herbicide Liquid Chromatography/Mass Spectrometry Analysis Method Tammy L. Jones” and Leon D. Betowski US. Environmental Protection Agency, Environmental Monitoring Systems Laboratory, P.O. Box 93478, Las Vegas, Nevada 89 193-3478
Barry Lesnlk
Department of Energy, Laboratory Management Branch, Germantown, Maryland 20874 Tom C. Chiang and John E. Teberg
Lockheed Engineering and Sciences Company, 1050 East Flamingo Road, Suite 126, Las Vegas, Nevada 891 19 Seven laboratories participated in an interlaboratory evaluation of a liquid chromatography/mass spectrometry (LC/MS) method for the analysis of 10 chlorinated phenoxy acid herbicides. The focus of this evaluation was to test the intercomparability of LC/MS data obtained from two types of LC/MS interfaces [i.e., thermospray (TS) and particle beam (PB)]. Eight simulated sample extracts were sent to each laboratory for LC/MS analysis. There were statistically significant differences between interfaces in the quantitative data for all analytes except 2-(2,4,5-trichlorophenoxy)propanoicacid (silvex). Particle beam exhibited a high positive bias and a low relative standard deviation a t the highest sample extract concentration range, 500 rg/mL, while TS showed a low bias and a low relative standard deviation at the lowest sample extract concentration range, 5 pg/mL. Another factor of this study was to look for any performance differences between interfaces of the same type, but differing manufacturers. A statistical difference was observed, between PB interfaces, for 2-(l-methylpropyl)-4,6-dinitrophenol (dinoseb). Introduction A number of compounds of environmental interest, including many on the US. Environmental Protection Agency’s (US. EPA) Resource Conservation and Recovery Act (RCRA) Appendix IX list, are polar, nonvolatile, and/or thermally labile. Thus they are not amenable to 1880 Environ.
Sci. Technol., Vol.
25, No. 11, 1991
conventional gas chromatography (GC) analysis. To address this problem efforts are now underway within the US.EPA to develop suitable techniques for these compounds. Examples of environmental organic contaminants that can be analyzed by liquid chromatography/mass spectrometry (LC/MS) methods are organophosphorus pesticides ( I ) , triazine herbicides ( 2 ) )and the chlorinated phenoxy acid herbicides (3). This latter group of compounds can be analyzed directly as the free acids, as well as the esters, by LC/MS. The chlorinated acid herbicides generally have a low mammalian toxicity, but impurities and high dosages may cause teratogenic effects in rodents ( 4 )*
Currently, US.EPA RCRA Solid Waste-846 (SW-846) methods 8150 and 8151 (5)are approved by the U S . EPA for the analysis of chlorinated herbicides in solid waste under the Resource Conservation and Recovery Act. These methods specify quantitation by GC with electron capture detector (ECD) and optional GC/MS confirmation. Also required are the use of hydrolysis and subsequent esterification of the sample extracts prior to analysis to convert the herbicides to gas chromatographable esters. Disadvantages with this method are that the hydrolysis step is time-consuming, and not always quantitative, and the usual esterification reagent, diazomethane, is potentially carcinogenic and explosive. The analysis of several chlorinated herbicides [specifically (2,4-dichlorophenoxy)acetic
0013-936X/91/0925-1880$02.50/0
0 1991 American Chemical Society
Table I. Precision and Accuracy of Interlaboratory Data particle beam mean % recovery” X ( % RSD)’ ft ( % RSD)d
analytes
X ( % RSD)*
2,4,5-T butoxy 2,4,5-T 2,4-D 2,4-DB dalapon dicamba dichlorprop dinoseb MCPA MCPP silvex
109 (14) 135 (25) 111 (14) 120 (13) ND 95 (24) 111 (13) 63 (13) 111 (20) 107 (17) 122 (20)
63 (33) 79 (39) 85 (36) 72 (30) ND 73 (89) 101 (24) 30 (3) 106 (25) 101 (37) 72 (45)
0 6‘colcH”’olcH2’,cH’ CII,
CI
H3C CCI2COOH
CI
2 . 4 . 5 - T . Butoxyethanol Ether Ester
Dalapon y
0-CH2COOH
CI
2.4.D
3
0-CH-COOH QC‘
CIQ
CI
7%
CI
@
CI
2, 4. 5 - T
H 3 c a
Dinoseb
CI
- TP (Silvex)
“ 3 ‘ 9
Dicamba
1
0-CH-COOH
thermospray mean % recovery” X (% RSD)’ (% RSD)d
x
( % RSD)*
90 (23) 90 (29) 86 (17) 95 (22) 83 (13) 77 (25) 84 (20) 78 (15) 89 (11) 86 (12) 96 (27)
62 (68) 85 (9) 64 (80) 104 (28) 121 (99) 90 (23) 96 (15) 86 (57) 96 (20) 76 (74) 65 (71)
90 (28) 99 (17) 103 (31) 96 (21) 150 (4) 105 (12) 102 (22) 108 (30) 94 (18) 98 (15) 87 (15)
one PB interface has a heated nebulization chamber, while the other does not.
Experimental Section A concentrated stock standard was prepared containing 11 analytes, listed in Table I, each at 1000 pg/mL, in acetonitrile. One milliliter of the stock standard solution was sent with each sample set for instrument calibration and analyte quantitation. Because methanol can methylate free acid herbicides, or tranesterify herbicide esters, all standards and extracts were prepared in acetonitrile. Each laboratory was provided with eight simulated sample extracts and one blank. The simulated sample extracts were prepared by dilution from the stock standard solution and sealed in 1.6-mL screw-cap, Teflon septa sealed, glass vials. The stock standard was used for preparation of simulated sample extracts, so that all chemicals used were traceable to a single original source. The simulated sample extracts consisted of duplicate extracts, 1 mL each, at four different concentration levels. Each laboratory was required to perform triplicate analysis on the duplicate sample extracts. The laboratories involved utilized three TS-LC/MS (one Finnigan, one Hewlett-Packard, and one Vestec) and four PB-LC/MS instruments (two Extrel and two HewlettPackard). A method blank extract was shipped with each sample set. The blank was prepared by extracting tap water, using the same extraction procedure as outlined in SW-846 method 8150. Because of the complex nature of LC/MS operation, specific operating parameters were not given to the participants concerning instrument (interface and MS) tuning and calibration. The laboratories were advised to follow the instrument manufacturer’s specifications for optimal interface performance. Instructions, in the form of a simplified version of the TS-LC/MS protocol for the analysis of chlorinated phenoxy acid herbicides, were sent to each participant. It was recommended that those laboratories using TS utilize the negative ionization mode for detection and quantitation (3). Quantitation for both interfaces was achieved by using the external standard method. Recommended analytical column: 15 cm X 2.1 mm id., (2-18 reverse phase, 0.5-pm particle size, and use of a guard column. HPLC gradient elution conditions: time 0 min, 50% water (with 1%acetic acid)/50% methanol (with 1% acetic acid); hold for 2 min; time 12 min, 40% water/60% methanol; time 18 min, 100% methanol; hold for 10 min; return to 50% water/50% methanol in 10 min; hold for 5 min before starting next analysis. It is necessary to have 1%acetic acid in the mobile phases in order to keep the acid analytes stabilized in their acid form.
cleH.
CI
2 , 4, 5
y
O-CH2COOH
7%
NO2
Dichlorprop
0-CH-COOH
c l ~ H CI 2 c o o H
OH
O~@CH-CH~-CH~
223 (2.1) 240 (28) 270 (30) 207 (14) ND ND 323 (19) ND 280 (10) 290 (8.3) 220 (6.5)
X
0--ICH,I,-COOH QC1
CI
CI
CI
MCPA
MCPP
2.4-DB
Figure 1. Structures of ten chlorinated phenoxy acid herbicides and one ester.
acid (2,4-D),3,6-dichloro-2-methoxybenzoic acid (dicamba), acid (MCPP)] and 2-(4-chloro-2-methylphenoxy)propanoic by reverse-phase high-performanceliquid chromatography with ultraviolet (HPLC/UV) detection has been reported in the literature (6). A multilaboratory collaborative study of the HPLC/UV method has been conducted and published (7).The use of LC/MS not only eliminates the need for the hydrolysis and esterification steps, but also provides a single-step analysis with selective mass spectral detection. Recent developments in LC/MS interfaces led to the initiation of this project. The purpose of this study was to compare and evaluate LC/MS interface devices [e.g., thermospray (TS) and particle beam (PB)] for their applicability to the analysis of acid herbicides and their esters. The TS (8) interface allows all of the LC effluent to enter into the mass spectrometer source, where ionization occurs by ion evaporation (buffer assisted) in the positive ionization mode or with the assistance of a discharge electrode in the negative ionization mode. The PB (9) interface relies on the principle of particle separation; the LC effluent enters into a nebulization chamber and then a desolvation chamber, where the lighter solvents are pumped away, and the heavier analytes are condensed into a analyte-enriched particle beam that enters directly into the MS source. Once inside the MS source “traditional” electron ionization occurs, i.e., filament-assisted ionization. Two types of PB interfaces were examined in this study;
Environ. Sci. Technol., Vol. 25, No. 11, 1991
1881
Separate HPLC column flow rates were recommended for the different LC column to interface plumbing configurations: for those laboratories using TS interfaces and postcolumn additions of 0.1 M ammonium acetate, column flow rates of 0.4-0.6 mL/min, with 0.8 mL/min postcolumn flow; for laboratories with TS interfaces, but without postcolumn addition, total column flow rates of 1.0-1.2 mL/min; and for those laboratories with PB interfaces, total column flow rates of 0.4-0.6 mL/min.
Results and Discussion Data were collected from four laboratories using the PB interface and from three laboratories using the TS interface. The data collected were used to compare and evaluate the two LC/MS interface devices for their relative strengths and weaknesses for herbicide acid and ester quantitation. A Statistical Analysis Systems (SAS) software program was employed to statistically treat the laboratory data. A probability ( P )value (IO)was calculated to determine if significant differences existed within the different data sets. For the data sets to be significantly different, with 95% confidence, the P value must be less than 0.05. Initially, duplicate extracts at the same concentration level were evaluated separately to determine if there were statistical differences in the results between the two samples. Statistical analysis on the triplicate LC/MS data from the same laboratory from the duplicate extracts indicated no significant difference. This was not unexpected, because these extracts were identically prepared. It was important, however, to demonstrate that statistically the procedure showed no within-laboratory bias between the duplicate extracts. Because the data indicated no within-laboratory bias, the m a l m recovery data from each laboratory was pooled. However, there was an indication of a sample preparation error for the concentration level 250 pg/mL; therefore, the data collected from this concentration level were considered as outliers and were not used. Two different statistical approaches were used to examine the remaining data for similarities and/or differences between the P B and TS interfaces. First, a probability (P)value was calculated by pooling the mean percent recovery from each interface type a t each concentration level (Le., 500, 50, and 5 pg/mL) and comparing the interface results. As an example, the analyte (2,4-dichlorophenoxy)acetic acid (2,4-D) exhibits a statistical difference between PB and TS at the theoretical concentration of 500 pg/mL. The calculated P value is 0.0433, indicating a significant statistical difference between the two sets of pooled data a t that concentration level. At the lower individual 2,4-D concentration levels (e.g., 50 and 5 pg/mL), there was no indication of a statistical difference between the two interfaces (PB and TS). Figure 2 is a bar graph showing the differences between the P B and TS, using normalized P values, P’, (where P’ = P - 0.05, and P’ < 0 indicates a significant statistical difference) at the individual concentration levels. No other analytes exhibited a significant statistical difference between the interfaces at the individual concentration levels. It should be noted that, while not giving a P value less that 0.05, the analytes 2,2-dichloropropanoic acid (dalapon), a-(l-methylpropyl)-4,6-dinitrophenol(dinoseb), and 3,6-dichloro-2methoxybenzoic acid (dicamba) were not detectable by the PB interfaces a t the low concentration level, 5 pg/mL. 2,2-Dichloropropanoic acid (dalapon), an aliphatic acid, was not detected at any concentration level by the PB interfaces. This is probably due to its high volatility; it is either vaporized along with the solvent in the desolvation 1882
Environ. Sci. Technol., Vol. 25, No. 11, 1991
I *1i.
241 .B
i
4
3
2 4
08 O*L*PONDlCI\MB‘
“lC3.OH
3”05EB
m
50ug WL
500ugrL
MCPA
UC”P
I
1 51 / I
x
54nt
Figure 2. Probability values for individual concentrations, PB vs TS.
J
0 3 2 4
>”
00
I
-
9
006-
1 2 4 5
?lSTB
2 r O
2rOB
,
,
I
3 4 L A P O N 3 C A M B A ):hL093NOSEB
I
UCPd
HCPP
Sir
x
Figure 3. Probability values for mean percent recovery, PB vs TS.
chamber and pumped away or it does not form particles after the solvent is removed. The second method for testing the data was to examine the pooled mean recoveries from each interface over all the concentration levels and examine the results for a difference between the interfaces. This difference between the interfaces over all the concentration levels gave more interesting results than the first statistical method. What was surprising is that while most compounds did not exhibit a statistical difference at the individual concentration levels, the combined concentration data gave an indication of fundamental differences between the interfaces (PB and TS). Figure 3 is another bar graph showing the differences between the PB and TS with normalized P values, P’, (P’ = P - 0.05, and P’ < 0 is considered a significant statistical difference) over all the concentration levels. As an example, the analyte (2,4,5-trichlorophenoxy)acetic acid (2,4,5-T) gives no indication at the individual concentration levels that there is a statistical difference between the interfaces. However, when the P value from overall concentration levels is examined, there is an indication of a difference between the two interfaces (Le., P‘ = -0.0353 < 0). What is occurring is that a t 500 and 50 pg/mL the two interfaces behave similarly; however, at 5 pg/mL the PB interface gives a very high mean percent recovery, 223%, while the TS interface has a mean percent recovery of 90%. This overall difference gives the low P value. All of the analytes show this same PB vs TS deviation, where the two interfaces behave similarly at the high and medium concentration levels, but depart dramatically a t the low concentration level, giving the low P values. Another objective of this study was to compare the performance of the differing manufacture types of LC/MS interface devices to each other (i.e., P B to PB and TS to
l160 6O1 140
-
;
120
-002 -003 -004 -005 -001
0 0 6 ' 1 215T
1
I
245 iB
14 0
, 1 4 08
P
I
g
100
U
8
60
60 40 20
,d
O A U W N DICAuBA OlCHLOR OINOSEB
UCPL
MCPP
0
SILVEX
A
Flgure 4. Probablllty values for comblned concentrations (PB): manufacturer A vs manufacturer B.
TS). The experimental design for making such an evaluation is influenced by the number of participating laboratories and their instrumentation configurations. There were only three laboratories with TS interfaces and each was from a differing manufacturer; therefore no statistical evaluation was possible. Four sets of PB data, produced by using interfaces made by two different instrument manufacturers (A and B), were received. This allowed for a limited comparison of the PB results. The P test was applied, as previously discussed, by combining the data from the two laboratories using PB from manufacturer A and comparing it with the combined data from the two laboratories using manufacturer B. There was no indication of significant statistical differences between the two data sets; see Figure 4. However, 2-(1methylpropyl)-4,6-dinitrophenol(dinoseb) was not detected at any level when analyzed by interface A and only by one laboratory using interface B. A comparison of the overall precision and accuracy between PB and TS is shown in Table I. PB has a tendency to give bias high results, at 500 pg/mL (average percent mean bias is +8%), and TS tends to give bias low results (average percent mean bias is -13%) when compared to the true value. The tabulated results for the precision data indicate that PB, at 500 pg/mL, gives slightly better precision (average % RSD = 17%) than TS (average % RSD = 19%). At the medium concentration level, 50 pg/mL, there was no clear difference between the results obtained from PB and TS, although both are biased low compared to the true value and both exhibit poor precision, average % RSD is for PB 36% and 49% for TS. Only one P B laboratory could detect analytes at the 5 pg/mL concentration level. The results reported by this laboratory were nearly twice as high as the true value. This strongly indicates that the detection limit for these compounds by PB is above 5 pg/mL. Figure 5 compares the percent recovery of 2,4,5-T at each concentration level for the four PB-LC/MS interfaces. Two of the three laboratories using TS reported values for the 5 pg/mL concentration level, and the results show a very low bias (average percent bias is 3 % ) and an average % RSD of 19%; see Table I. This is a strong indication that TS provides better sensitivity, with greater accuracy, than PB in detecting low levels of these compounds, although with moderate precision. An example of the percent recoveries for 2,4,5-T at each concentration level for the three TS interfaces is shown in Figure 6. Conclusion Although the data collected were from a limited data set (Le., seven laboratories, and one type of extract) the
A 500NGlUL
B
0
50NGlUL
SNGIUL
Flgure 5. Comparlson of the percent recovery of 2,4,5-T between fow PB-LCIMS interfaces. 110
,
I
100
90
60 70
60
H
'
50 40
30
20
IO
A
5WNGlUL
C 50 NGIUL
0 5NGIUI
Flgure 6. Comparison of the percent recovery of 2,4,5-T between three TS-LC/MS interfaces.
statistical data showed some interesting and informative results. A low molecular weight chlorinated aliphatic acid, 2,2dichloropropanoic acid, was not detected by PB, presumably because it is too volatile to be transmitted through the P B interface. The substituted phenol, 2-( l-methylpropyl)-4,6-dinitrophenol,responds poorly to PB; even at the highest concentration level, 500 pg/mL, only one of the four P B laboratories reported values for 2-(1methylpropyl)-4,6-dinitrophenol. 3,6-Dichloro-2-methoxybenzoic acid did not respond well with PB, especially at the lower concentration levels. PB generally gives better precision than TS, particularly at the high concentration level (500 pg/mL). This is indicated by the lower % RSD values, shown in Table I. Since TS is more sensitive in detecting the target analytes, the extracts often had to be diluted prior to injection in order to be within the linear response calibration range of the instrument. Therefore, dilution errors may have contributed to the poorer % RSD observed for TS. However, it can be assumed that part of the precision difference is due to the fundamental differences in the operating principles of the two interface systems. The choice of TS or PB interfaces will depend on the type of analytes and the analytical requirements of the data user. From this study one can conclude that for the analysis of low-level samples, TS, with negative ion detection, would be preferred for phenoxy acid herbicides. For the analysis of high-level samples in which identification of the analytes is essential, PB, with electron ionization, might be preferred. Further work should delineate the general applicability, if any, of these conclusions. Environ. Scl. Technol., Vol. 25, No. 11, 1991
1883
Cremlyn, R. Pesticides: Preparation and Mode of Action; John Wiley and Sons: Chichester, England, 1978; p 142. Test Methods for Evaluating Solid Waste, 3rd ed.; Office of Solid Waste and Emergency Response, U.S. Environmental Protection Agency: Washington DC, 1986; Vol. 1B. Grorud, R. B.; Forrette, J. E. J . Assoc. Anal. Chem. 1983, 66, 1220-1225. Grorud, R. B.; Forrette, J. E. J. Assoc. Anal. Chem. 1984, 67,837-840. Blakely, C. R.; Vestal, M. L. Anal. Chem. 1983,55, 750. Willoughby, R. C.; Browner, R. F. Anal. Chem. 1984,56, 2626-2631. SAS Institute Inc. S A S I S T A p M Guide for Personal Computers, Version 6 Edition; SAS Institute Inc.: Cary, NC, 1987; p 1028.
Acknowledgments We thank the individuals and laboratories who volunteered their time and efforts to this study: Dr. Robert Betham; Dr. Mark Brown, California Department of Environmental Services; ENSECO; Dr. Thomas Behymer, U S . EPA, EMSL-Cincinnati; Dr. Paul Goodley, Hewlett-Packard Corp.; Dr. Mark Roby and Chris Pace, Lockheed Engineering and Sciences Co.; and Dr. Jack Northington, West Coast Analytical Services. Registry No. 2,4-D, 94-75-7;2,4,5-T, 93-76-5;MCPA, 94-74-6; MCPP, 7085-19-0; 2,4,5-TP, 93-72-1; 2,4-DP, 94-82-6; dalapon, 75-99-0;dichloroprop, 120-36-5;2,4,5-T butoxyethanol ether ester, 2545-59-7; dinoseb, 88-85-7; dicamba, 1918-00-9.
Literature Cited Betowski, L.D.; Jones, T. L. Enuiron. Sci. Technol. 1988, 22, 1430-1434. Parker, C. E.; Haney, C. A,; Harvan, D. J.; Haw, J. R. J. Chromatogr. 1982,242, 77-96. Jones, T. L.;Betowski, L.D.; Yinon, J. In Liquid ChromatographylMass Spectrometry: Applications in Agricultural, Pharmaceutical, and Environmental Chemistry; Brown, M. A., Ed.; ACS Symposium Series 420; Americal Chemical Society: Washington DC, 1990; pp 62-74.
Received for review October 24, 1990. Revised manuscript received April 5, 1991. Accepted April 10, 1991. Notice: Although the research described in this article has been supported by the United States Environmental Protection Agency, it has not been subjected to Agency review and, therefore, does not necessarily reflect the views of the Agency, and no official endorsement should be inferred. Mention of trade names or commercial products does not constitute endorsement nor recommendation for use.
Effect of Temperature on the Formation of Photochemical Ozone in a Propene-NO,-Air-Irradiation System Shlro Hatakeyama, Hajlme Aklmoto, and Nobuakl Washlda National Institute for Envlronmental Studles, Tsukuba, Ibaraki 305, Japan The profile of ozone formation in a propene-NO,-airirradiation system was studied at 20,30,40, and 50 "C in order to clarify the effect of temperature. The higher the temperature, the faster the formation of ozone and the longer the duration of a high concentration of ozone. The former phenomenon was due to the effect of temperature on the stability of HOON02, and the latter was due to faster decomposition of PAN at higher temperatures. Computer simulation was employed to realize the mechanism for those temperature effects. Introduction It is now well recognized that the global warming is a real threat for mankind. The rise of temperature within this century was observed to be -0.6 "C (I). Additional future warming is predicted (2-5) and its effects on global climate as well as on human life are discussed intensively (6). Among the effects of global warming, the effect on the quality of the urban atmosphere is one of the important problems. However, little information is available so far. Only one report by Carter et al. (7) was made as to experimental studies of the effect of temperature on photochemical smog formation. They observed increased rates of formation of ozone, consumption of hydrocarbons and NO,, and conversion of NO to NO2 and, also, a second maximum of ozone at a high temperature, which was ascribed to the thermal decomposition of PAN. Recently, Gery et al. (8) reported the results of modeling studies on the effect of temperature change on ozone formation. An increase of ozone due to an increase of surface temperature was simulated. However, experimental data are not yet sufficient enough to discuss the mechanism of the tem1884
Envlron. Scl. Technol., Vol.
25, No. 11, 1991
perature effect on ozone formation. In the present study we performed smog chamber experiments to investigate the dependence of ozone formation on temperature in the photochemical reactions of propene-N0,-humid air systems. Computer simulation was also done to understand the mechanism controlling the difference in the profile of ozone formation. Although the predicted change in temperature in global warming is less than a few degrees Celsius [l-5 "C by 2050 (9) is one of higher predictions], temperatures with much wider ranges were employed in the present study in order to make the results more visible. Experimental Section All the experiments were carried out in the evacuable smog chamber (6 m2),whose inner surface is coated with tetrafluoroethylene-perfluoroalkyl vinyl ether copolymer (PFA). Details about this chamber were already reported (IO). The chamber wall was temperature-controlledwithin fl "C at 20, 30, 40, and 50 "C. Photoirradiation was performed by 19 Xe arc lamps (light intensity as measured by the NO2 photodissociation constant, k,,was 0.29 min-'). Experiments were performed under 1 atm of air. Propene was chosen as a model hydrocarbon in hydrocarbon-NO,-humid air-irradiation systems. Initial concentrations of reactants were [propene], = 2 ppm, [NO], = 0.26 ppm, [NO,], = 0.48 ppm, and [H2010= 5.8-6.5 Torr. Since we had already found that the formation of HONO by the heterogeneous reaction of NO2 and H20 is very important as an initial source of OH radical (11),care was taken to keep the experimental procedure nearly the same for each run. First purified dry air was passed through a humidifier and introduced into the chamber up
0013-936X/91/0925-1884$02.50/0
0 1991 American Chemical SOCletY
