Anal. Chem. 1984, 56,793-798
793
Sampling of Organic Compounds in the Presence of Reactive Inorganic Gases with Tenax GC Edo Pellizzari* a n d Barbu Demian Analytical and Chemical Sciences, Research Triangle Institute, Research Triangle Park, North Carolina 27709 Kenneth Krost
US. Environmental Protection Agency, National Environmental Research Center, Research Triangle Park, North Carolina 27711
The sampllng of vapor-phase organics In the presence of reactive Inorganic gases with the Tenax GC sampllng cartridge was statlstlcally evaluated. A factorial experlment and the use of toluene-d,, styrene-d,, and cyclohexene-d,, coupled with GC/MS analysis provided insight to those reaction products from the sorbent itself and those from adsorbed analytes. The removal of oxidants prior to capturing organic vapors was lnvestlgated by use of mild reduclng agents Impregnated In filters for particulate. Sodium thlosulfate Impregnated glass fiber and Teflon filters quantltatlveiy decomposed ozone.
During the past 2 decades few advances have been made in developing collection methods for characterizing and measuring extremely minute amounts of hazardous vaporphase organics in ambient air. Collection methods now available include sorbents (1-4), cryogenics (5-9), and containers which capture of parcel of air (10-12). Techniques have also been investigated for transferring and analyzing collected substances. There is, however, a paucity of experimental research addressing the question of whether decomposition products or artifacts occur during the collection of vapor-phase organics. Isolated reports do suggest that serious problems could exist with cryogenic traps (13-15),containers (11,16)and sorbent traps (17). With the observation of nitrosamines in ambient air, the potential in situ formation of these substances from secondary amines and nitrogen oxides was of some concern in cryogenic traps (13,14) and on sorbents (17,18). In fact, a metamorphosis of collection methods for nitrosamines occurred leading eventually to a sorbent cartridge impregnated with sulfamic acid to inhibit nitrosation (18). Thus, there is no reason to believe, a priori, that any collection method including cryogenics is exempt from potential problems associated with oxidants in the ambient air. Even though the Tenax GC cartridge does not appreciably concentrate reactive inorganic gases, i.e., ozone, nitrogen oxides, molecular halogens, etc. (191,relative to organics, in situ reactions may occur with the sorbent itself or with certain chemical classes of compounds, under certain conditions. Under laboratory simulation of atmospheric conditions, we have investigated some potentially troublesome chemical groups which could exhibit reactivity with commonly occurring inorganic gases in ambient air. Specifically, the Tenax GC sampling cartridge was tested to gain insight into potential problems which may be encountered during field sampling. EXPERIMENTAL SECTION Materials. Tenax GC (60/80 mesh) was purchased from Applied Science, State College, PA. Distilled in glass solvents were from Burdick & Jackson (Muskegon,MI). Ultrazero air and chlorine were from Matheson, Coleman and Bell, Norwood, OH.
Nitrogen dioxide and SO2were acquired from Airco, Montvale, NJ. Benzaldehyde, acetophenone and phenol were purchased from Aldrich, Milwaukee, WI. Benzene-& (minimum 99.5% D) was obtained from Stohler Isotope Chemicals, Rutherford, NJ. Styrene-d8, cyclohexene-dloand dioxane-&,(minimum 99% D) were from Merck, Sharp & Dohme, Canada, Ltd. Sodium thiosulfate (anhydrous 99%), sodium metabisulfite (ACS Certified, 99.8%), ferrous sulfate (USP, crystal), sodium lauryl sulfate (USP),and sodium hydroxide (ACS Certified 97% minimum) were obtained from Fisher Scientific, Pittsburgh, PA. Sodium oxalate (analytical reagent, 99.9%) was obtained from Mallinckrodt, St. Louis, MO. Ammonium sulfamate (anhydrous, 98%) was obtained from Sigma Corp., St. Louis, MO. Oxalic acid (98% minimum) was obtained from Baker & Co., Phillipsburg, NJ. Hydrazine hydrate (99% minimum) was obtained from Matheson, Coleman and Bell. Filters used were Gelman (Ann Arbor, MI) borosilicate microfiber glass, 1.25 mm thick, with acrylate resin binder, 25 mm diameter. Schleicher and Schuell (Keene, NH) glass microfibers, type 25,0.365 mm thick, 25 mm in diameter, were used. Teflon (TE-37) filters were Schleicher and Schuell poly(tetrafluoroethy1ene)membrane filters, maximum pore size 1.0 Mm. Apparatus. The apparatus used to synthesize laboratory air containing oxidants for sampling is depicted in Figure 1. U1trazero air was metered through a Teflon membrane filter and a Drierite/molecular sieve trap, supplied to a glass permeation chamber, humidifier, and ozone generator, and controlled and metered with a Tylan flow system. The permeation chamber was thermostated at 30 f 0.1 “C with a Haake constant temperature bath. The glass humidifier system consisted of two 500-mL gas washing bottles, the first containing distilled water and the second 3-mm glass rings. Ozone was generated by irradiating with a UV lamp the air passing through a quartz tube. All lines leading to or from the devices were made of Teflon. Nitric oxides were metered into the stream with a Gilmont microflowmeter (Figure 1). The various components entered a glass mixing chamber and then the air was sampled from a glass manifold. Concentrations of NO, NO2, and ozone were monitored with Bendix NO, and Bendix ozone analyzers. The humidity was measured with a YSI hygrometer. Intakes for these instruments were at the same point as the intake for the Tenax GC glass cartridge (6.0 cm bed length X 1.5 cm i.d.) through which air was drawn by a Nutech Model 221-A sampler (Nutech, Durham, NC). Analysis Methods. For qualitative analysis of Tenax GC or analyte degradation products, the previously described thermal desorption capillary gas chromatography/mass spectrometry (CGC/MS) method (2) was used. Quantitative analysis of decomposition products (e.g., benzaldehyde, acetophenone,phenol, and deuterated substances) was performed by external calibration of the thermal desorption capillary GC system (Varian Model 3700 equipped with flame ionization). Tenax GC cartridges were spiked with authentic standards using permeation tubes or a solvent evaporation method (2). Experiments. Effect of O,, NO,, and Humidity on Tenax GC. A factorial experiment (Z3) was designed to assess the effects of air containing Os, NO,, and humidity on clean, “virgin” Tenax GC cartridges (2). The exposure conditions chosen exceeded the National Air Quality standards to provide a severe test of the collection method (Table I). Replicate cartridges were also
0003-2700/84/0358-0793$01.50/00 1984 American Chemical Society
704
ANALYTICAL CHEMISTRY, VOL. 56, NO. 4, APRIL 1984
METERINO VALVE
CHAMBER
-
Table 11. Exposure of Tenax GC to SO,, 0,, NO,, and Humiditya amounts in ppb vol sampled, L sox 0 3 NO, 0
30 30 30 30
30 a
0 0 0 100 100 200
0 0 100 100
100
Air samoled was at 60% relative humiditv.
N D ~ 30 65 105
0 0 150 150 0 0
100
amounts in nglcartridge benzaldehyde acetophenone
40
40
ND ND 20 30 ND 10
phenol ND 20 60 125 20
30
ND. not detected.
exposed for identification of decomposition products by CGC/MS. Effect of Os,NO,, and SO, on Tenax. Tenax GC cartridges were exposed to 30 L of air containing 03,SOz, and NO,, and humidity at different levels as indicated in Table 11. Qualitative and quantitative analyses were as described above. Effect of Molecular Chlorine. The experimental conditions of exposure of Tenax GC to Clz, O,,and humidity are given in Table 111. Reaction of Adsorbed Analytes with Oxidants. Perdeuterated toluene, styrene, and cyclohexene were singly introduced onto
a Tenax GC sampling cartridge as a discrete band (1-2 yg) at one end of the sorbent bed either from a permeation tube or by the solvent evaporation method (2). In the first experiment each cartridge containing the deuterated substance was exposed to 200 ppb O3 from a flowing air stream (0.5 L/min) for 30 min (Figure 1). Each sampling cartridge was then qualitatively analyzed by thermal desorption CGC/MS. In a second experiment the cartridges were exposed to air containing high levels of ozone (1000 ppb), NO, (500 ppb), S02/S03 (400ppb), ClZ(564 ppb), and 50% relative humidity
ANALYTICAL CHEMISTRY, VOL. 56, NO. 4, APRIL 1984
passed through the filter, the air was directed to 0s and NO, monitors. The analog signal from the monitors was recorded on a Fisher Recordall 4000. Glass fiber and Teflon filters were tested for absorption of styrene. Air was passed through the permeation chamber (100 mL/min) containing a styrene permeation tube (17.5 ng/min at 30 "C). The air was theh passed through a Tenax GC sampling cartridge with and without impregnated filters upstream. After 30 min of sampling the Tenax GC cartridges were analyzed for styrene. Additional testing of G-2 and G-7 filters for decomposition of Tenax GC and/or styrene was performed. Sampling cartridges were exposed for 30 min to flow of air at 1L/min containing 200 ppb 03, 400 ppb NO,, and 50% humidity, with and without impregnated filters upstream. After sampling, an internal standard, 2-fluorobiphenyl, was added to each cartridge and quantification of products performed.
Table 111. Exposure of Tenax GC to Ozone and Chlorine' amounts in ppb chlorine ozone 0 0
0
117 117 173
150 200
amounts in nglcartridge acetobenzaldehyde phenone phenol -
N D ~ ND ND 6
ND ND 5 40
ND ND ND 10
Volume air sampled was 30 L, relative humidity was 60%. ND, not detected. at 25 "C (Figure 1). Again qualitative analysis was performed by CGC/MS. Inhibition of in Situ Reactions on Tenax GC. Gelman borosilicate microfiber glass filters (1.25 mm thick, 25 mm diameter) with acrylic binder were purified by repetitive 24-h Soxhlet extraction with distilled-in-glassmethanol and n-pentane and dried under nitrogen flow (24 h) and then in a vacuum oven at 200 OC (48 h). Schleicher and Schuell glass microfiber filters type 25 (0.36 mm thick, 25 mm diameter) without binder were used as received. Schleicher and Schuell poly(tetrafluoroethy1ene)membrane filters type TE 37 (24 mm diameter, 1.0 pm pore size) backed with nonwoven polypropylene were also used as received. Filters were impregnated with the chemicals listed in Table IV by submerging for 1.5-2 h in a 50-mL aqueous solution, air-dried 0.5-1 h, and then dried in a vacuum oven at 50 "C overnight. Each filter was weighed and tested for pressure differential under flow with a manometer before and after treatment. The pressure drop was relatively linear with flow rates between 0.2 and 1.2 L/min. The effectiveness of filters impregnated with chemicals in destroying oxidants was tested by use of the experimental apparatus in Figure 1. Air flows were 1 L/min and 0.5 L/min for glass and Teflon filters, respectively,while the air contained 200 ppb 03, 400 ppb NOz, and 50% relative humidity. Aftr air was
RESULTS AND DISCUSSION Reactions with Tenax GC. Effect of 03,NO,, and Humidity. Tenax GC cartridges used in sampling low levels of oxidants (Table I) were analyzed by CGC/MS. Three major products-benzaldehyde, acetophenone, and phenol-were identified. a-Hydroxyacetophenone and ethylene oxide were also detected in trace quantities. In order to quantitatively determine the effect of oxidants on Tenax GC itself, a factorial experiment (23)was designed. Table V presents the results of this test. No statistically significant difference was found between blank-exposed and blank-unexposed sampling cartridges; thus, the experimental system was reasonably free from contamination. The same test was applied to compare blank sampling cartridges (exposed and unexposed) and sampling cartridges (duplicates) exposed to the lowest level of oxidants and humidity (Table V). The difference between blank and exposed
Table IV. Properties and Effectiveness of Impregnated Filters in Removing Oxidants filter wt gain, mg i: std dev filter code G-1' G-2 G-3 G-4 G-5 G-6 G-7
10%Na,S,O, 1.5%Na,S,O, 2.5% FeSO, t 5% NaOH 10% (COOH), 10%ZN,H;(COOH), 10% NH,SO,NH, 2.5% FeSO, t 1%Na,SO, t 5% N,H, + 2% NaOH 5% (COOH), + 0.1% NaLS 5% NH,SO,NH, t 0.1% NaLS 5% Na,C,O,
G-8 G-9 G-10 G-11
pressure drop, mm H,O i: std dev ( N ) a t l L/min
63.7 i: 0.9 ( 2 ) 6.0 i 0.06 (5) 87.8 i: 1.00 (2) 76.6 * 1.01 (2) 51.5 * 0.56 (2) 78.4 i: 0.90 ( 2 ) 38.2 * 0.35 (4) 1.1i: 0 (5) 71.8 i 1.46 (5) 21.6 i: 0 (5)
5% Na,S,O, 5% Na,S,O,
TE37-1 TE37-2 TE37-3 TE27-4
+ 0.2% NaLS
31.1 i: 1.19 (5) 49.5 i 0.03 (5)
5% Na,S,O, + 0.2% NaLS 10% Na,S,O, t 0.1% NaLS 1 0 % Na,S,O, + 0.2% NaLS
6.1 i: 0.10 ( 5 ) 9.4 * 0.35 ( 5 ) 10.7 i: 0.35 (5)
' Gelman glass fiber filter. measured.
__ d 35.5 + 1 . 3 (3) -
-
e
Schleicher & Schuell glass fiber filter. Extrapolated value,
___ozone min, ppb
ppb/ min
2.5 0 96 190 156
7.5 0 -
-
-
0
0
-
33.2
i:
11 ( 3 )
-
155
-
-
135
-
35.8
S&S 25-1 S&S 25-2 S&S 25-3 S&S S&S S&S S&S
(N)
agent impregnated
i
4.3 (3)
90.6 * 6.4 ( 3 ) 86.7 i 7.3 (3) 40.3 i 0.6 (3) -
315" 348 217
-
1.3
-
4
-
0 0
0 0
-
-
26.5 26.5
0.37 0.25
11
-
-
-
Schleicher & Schuell Teflon filter.
-, not
Table V. Blank and Low Level Oxidant Exposed Tenax GC Cartridges' benzaldehyde blank, unexposed ( n = 8)' 0.14 blank, exposed ( n = 2 ) = 0.12 blank ( n = 5 ) b 0.14 low level ( n = 2 ) b 0.90 ' Paired experiment. Paired experiment.
i: i i i
0.210
acetophenone
0.085 r 0.092 0.075 i: 0.049 0.078 t 0.058 0.28 * 0.120 Values in pg/m3.
0.110 0.10 0.53
795
phenol 0.025 * 0.050 f 0.040 i 0.650 i:
0.007 0.035 0.028 0.520
all three summed 0.28 i: 0.24 0.24 * 0.18 0.26 i 0.18 1.8 * 1.2
796
ANALYTICAL CHEMISTRY, VOL. 56, NO. 4, APRIL 1984
Table VI. Data Matrix for Factorial Experiment ( 0 3NO,, , H,O Vapor)a parameter YI Y1 Y3 Y4
NO,
t t
-
YS
-
Y6
t -
Y7
+
Y8
a
0,
Values in pg/cartridge
H,O
vapor
t t -
-
0.90
-
4.00
-
-
t t
t
t
+ i
benzaldehyde
+
i
0.37
0.67 11.70 11.60 5.80 i 1.20 14.00 i 1.00 8.60 i 5.00
acetophenone
phenol
all three
0.28 i 0.085 1.00 0.30 3.90 3.20 6.30 -t 3.80 4.40 i 0.38 3.20 i 1.80
0.64 i 0.37 3.50 0.49 10.00 9.80 7.70 i 2.60 13.00 i 0.45 6.80 i 4.80
8.50 1.50 26.00 25.00 20.00 i 5.20 31.00 i 1.80 19.00 i 12.00
1.80 i 0.83
standard deviation.
Table VII. Main Effects and Interactions Calculated for Factorial Experiment ( O s ,NO,, H,O Vapor)a main effect ozone NO, humidity interactions ozone x NO, ozone x humidity NO, x humidity ozone x NO, x humidity a
Values are in pglcartridge
i
benzaldehyde
acetophenone
0 8.20 i 1.20 5.70 i 1.20
1.50 i 0.45a 0 2.90 i 0.45
0
-6.30
i
1.20
N = number of runs
(1)
phenol sum of three benzaldehyde acetophenone -~ 1250 445 1260 2945 (ng/
humidity = 7 4 b 5 Interactions:
+ Y6 + y7 + y8 - y1 - Y Z - y3 - ~
ozone X NO, = 74(Y1 + ~4 + ~5
4 )( 5 )
+ YS - Y Z - Y3 - Y6 - Y7) (6)
ozone X humidity = 1
2.20 5.60
0
0
-5.20 0 -2.70
i t
0.45
0.45
all three 0
i i
1.20 1.20
i
1.20
5.60 i 2.90 1.40 i 2.90 0 -12.00
i
2.90
i
2.90
0 i
1.20
-6.40
standard deviation.
cartridges was not statistically significant; however, a trend is apparent in the data. Table VI lists the matrix of data from the factorial experiment. An estimate of variance (20) in the data was made, even though not all measurements were in replicate. The variance and standard deviation of the study results were calculated in terms of observed effect as follows:
4 Veffect= =S2
0
0
-1.20 -1.80
0 0
phenol
hb1 + Y3 + Y6 + YS - YZ - Y 4 - Y5 - Y7) ( 7 ) ozone X NO, X humidity = 7 4 b Z + Y 3 + Y5 +Y2 - Y8- Y 4 - Y6 - Y7) (8) The main effect contribution to an artifact was calculated as the difference between two averages main effect = y+ - 9(9)
where 9’ is the average response for the (+) level of the variable and 9- is the response for the (-) level of the variable. Furthermore, the interactions are measures of nonadditivity of the main effects. These interactions are differences between averages of one parameter effect at level (+) and (-) for the second parameter. Table VI1 gives the effects and interactions for the factorial experiment. The experimental conditions revealed a region in which a combination of high concentration of oxidants and high relative humidity appears to reduce the amounts of the three decomposition products, by decomposing them faster than they are generated from the adsorbent (minus sign for all significant interactions). Also, humidity played an important role (Table VII). These data suggest that benzaldehyde, acetophenone, and phenol are more prone to increase in magnitude when sampling ambient air with a high relative humidity. The ozone and NO2 levels behaved additively, Le., the total concentration of oxidants in air reaching the Tenax GC sorbent was important in producing benzaldehyde, acetophenone, and phenol (Table VII). The three major decomposition products of virgin Tenax GC were easily detectable even in freshly prepared cartridges. The analysis for these compounds in ambient air is not possible without special precautions. Effects of Sulfur Oxides. In addition to the oxidants discussed above, the effect of sulfur dioxide of Tenax GC was examined. The major substituents found were again confirmed by CGC/MS/COMP to be benzaldehyde, acetophenone, and phenol. Neigher their increase nor compounds specific to the presence of sulfur oxides could be detected under the conditions studied. Effects of Chlorine. Table I11 gives the experimental parameters and the quantities of benzaldehyde, acetophenone, and phenol formed when Tenax GC cartridges were exposed to ozone, chlorine, and humidity. The major characteristic change observed was an increase in acetophenone relative to the other products. Effects of Aging Tenax GC. For all data derived from the SO, and Clz experiments, substantially less of the decomposition products were found. This effect was not related to the nature of the reactive inorganic gases but to the “history” of
ANALYTICAL CHEMISTRY, VOL. 56, NO. 4, APRIL 1984
Table VIII. Effect of Aging Tenax GC’ cycle no. 1
5
benzaldehyde 670 65
acetophenone 300 20
Values in nglcartridge. intrusion porosimetry. a
phenol 490
surface area, b m’/g
Table IX. Adsorption of Styrene to Impregnated Filters‘ filter
styrene, srbitrary units
none G-2 G-7 S&S 25S&S 25S&STE37S&STE37-
499 467 401 390 513 452 452
69
60
63
Determined by mercury
the Tenax GC cartridge. In the SO, and C12 experiments, recycled Tenax GC cartridges were used, and thus the hypothesis that ozone degradation of the sorbent was not constant was tested. The same sampling cartridges were repeatedly exposed to Os,NO,, and 60% relative humidity and the quantities of degradation products were measured. Table VI11 lists the levels of benzaldehyde, acetophenone, and phenol and surface area measurement for “virginnsorbent and sorbent recycled five times. The quantities decreased by an order of magnitude between the first and fifth cycle. The specific surface area decreased slightly (not statistically significant); however, this increase cannot explain this phenomenon. A more intriguing hypothesis is that the sorbent surface was depleted in readily degradable oligomers which are impurities in the sorbent, the repeated oxidation/thermal desorption. Tenax GC itself does not contain the aliphatic carbons which must be present to account for some of the observed decomposition products. Decomposition of Adsorbed Analytes. The effect of oxidants on organic vapors adsorbed to Tenax GC was investigated by spiking sampling cartridges with perdeuterated compounds and then sampling. The use of perdeuterated compounds provided a means of differentiating decomposition products occurring from Tenax GC itself (as undeuterated substances) and those formed by reaction of oxidants with adsorbed species. Thus, the origin of benzaldehyde as a decomposition product from Tenax GC, absorbed organics or both was elucidated. Tenax GC cartridges spiked with cyclohexene-dlo and used in sampling O3were found to produce cyclohexadiene-d8,three isomers of C6D1,0, and benzene-& Because capillary GC was employed as the separation method, it is also conceivable that more polar products were produced which did not pass through the column. In addition, when air containing Clz was sampled 2-chlorocyclohexanol-dlo and two isomers of dichlorocyclohexane-d,, were identified. Approximately half of the parent compound was recovered, whereas only cyclohexene-dlo was detected in control cartridges which sampled air without oxidants or molecular chlorine. Benzaldehyde-d, and benzoic-de acid were identified in sampling cartridges spiked with styrene-de. More than 80% of the parent compound was recovered. When air also containing Cl2 was sampled, styrene-d6dichloride and two isomers of a-chlorostyrenes-d, were found. Only styrene-d6 was detected on control sampling cartridges. Traces of benzaldehyde-d, were detected when toluene-d6 was the adsorbed analyte and cartridges were exposed to oxidants. No deuterated decomposition products were detected on control samplers. The results of these experiments clearly demonstrate that the presence of oxidants in ambient air can produce decomposition products both from the Tenax GC sorbent and from analytes adsorbed to the sorbent. Because of the susceptibility of olefins and some oxygenated substances to ozone, we then turned to an investigation of inhibiting these reactions. A method of circumventing this decomposition problem could be applied to any sampling method (cryogenics, impingers, etc.) if it were capable of removing oxidants before capturing the organic vapors of interest.
797
a
x:
n
S
t
1-a
32 31 79 200
7 3 3 2 1 2 2
31.7 48.2 81.2 32 29.9 30.1
1.46 2.96 1.67
0.95
