Chromatographic Monitoring of Hydrocarbons in Ambient Air Edgar R. Stephens* and Oscar P. Hellrich Statewide Air Pollution Research Center, University of California, Riverside, Calif. 92521
An automated gas chromatograph utilizes an eight-port, two-position sample valve with a 4.4-mL room temperature sample loop to measure methane, the three two-carbon hydrocarbons, and, by backflush, higher hydrocarbons. For the two-carbon hydrocarbons, minimum detectable concentrations were about 1-2 pg/m3. For higher hydrocarbons, about 100-200 pg/m3 was detectable. Measurement of higher hydrocarbon by backflush eliminates the error magnifying step of methane subtraction in the determination of nonmethane hydrocarbon. Analysis of polluted air for hydrocarbons is necessary to guide control strategy for this important precursor of atmospheric oxidant. Gas chromatography with flame ionization detection is the principal method. I t has been used in two forms: with preconcentration by freeze out to estimate a long (but still incomplete) list of hydrocarbons down to sub-ppbv levels (1-3), and with a partial separation to measure total hydrocarbon (THC) and nonmethane hydrocarbon (NMHC). Methane is excluded for three reasons: (i) It is present in quite high concentrations even in clear air due to natural sources (1.4 ppmv). Local natural gas emissions add varying amounts to the background. (ii) It is quite unreactive and so plays little role in polluted air chemistry. (iii) Its presence can mask the variations in more reactive hydrocarbons due to human activities. Data on the ambient concentrations of NMHC are needed for comparison with air quality standards (160 pg/m3 or 0.24 ppmC) and with both laboratory studies and computer modeling of photochemical smog. Detailed analysis of atmospheric hydrocarbons by freeze out concentration has required samples of 0.1 L or more and tedious data reduction procedures. Determination of total hydrocarbon by such a procedure would require summation of scores of very small concentrations and could never be complete. On the other hand, the direct methods used in commercial instruments measure total hydrocarbon and methane separately and determine NMHC by subtraction. Errors are magnified by this subtraction of two large numbers to obtain the required NMHC concentration. Urban air may contain a few hundred micrograms/cubic meter of “natural gas” methane in addition to 870 pg/m3 of background methane. This can easily be 8 to 10 times larger than the 160 pg/m3 concentration of all higher hydrocarbons a t the air quality standard. The accuracy is further jeopardized by the use of two separate samples to measure the total hydrocarbon and the methane. If the flame calibrations for methane and for THC are not accurately correlated or if the two samples have methane concentrations differing by even a small amount, large errors are possible in the indicated NMHC. The “apparent” NMHC may even become negative ( 4 ) ! In the method reported here the subtraction of methane (as well as the two-carbon hydrocarbons) was carried out chromatographically, rather than electronically, by backflushing the hydrocarbons with three or more carbon atoms into the flame detector after separation of the one- and two-carbon hydrocarbons in the direct elution mode. This permitted measurement of five components on one sample: methane, ethane, ethene, acetylene, and C3+ hydrocarbons. I t is desirable to measure the two-carbon unsaturates because they are closely related to auto exhaust, and the ratio of ethene to acetylene gives a measure of the extent of photochemical ox836 Environmental Science & Technology
idation in the air, since ethene is several times more reactive than acetylene ( 2 , 3 ) . The backflush, of course, combines hydrocarbons, indeed organics, of many kinds into one peak, which raises problems of interpretation and calibration. Since the flame detector responds approximately according to carbon content and a large portion of the backflush peak is gasoline type hydrocarbon, it can be calibrated with gasoline to provide a peak of similar shape. Expression of the results as micrograms/cubic meter has three advantages: (i) no assumptions about average molecular weight of the peak need be made; (ii) direct comparison with emission standards expressed in weight units can be made (for example, gramdmile or tondday); (iii) auto hydrocarbon emissions are measured with a flame detector. Experimental T o permit elution of both direct injection and backflush peaks, an eight-port two-position value was used (Carle No. 2012). I t was connected to the chromatograph (Varian HiFi Model 600) in the manner shown in Figure 1. In the backflush-sample mode, the pump flushed a sample of ambient air through the 4.4-mL sample loop while the column was backflushed into the detector (solid lines on the valve). Sample injection and column reversal occur simultaneously as the valve is rotated ‘/s turn. Elution of the one- and two-carbon hydrocarbons occurs within 3 min with this fairly short column operated a t 64 “C. After this is complete the column is reversed to backflush higher hydrocarbons into the flame detector. These appear as a single rather broad peak. An additional 10 min is allowed to complete the purge. One important problem that required solution was the snuffing of the flame by the surge of carrier gas when the column was reversed. To prevent this, a variety of short buffer columns were inserted between the valve and the detector. The most successful of these was a 7.5-cm column containing the same packing as the separating column. Unattended operation was provided by using a “valve minder” (Carle 4100) and actuator (Carle 4200) to operate the valve on the schedule just described. This also caused trouble. Sometimes the signal would drift up or down the chart paper for unknown reasons, and sometimes column reversal would cause a major offset of the signal level. The latter problem was caused by failure of the valve to seat in perfect alignment, so that the carrier gas flow in the two directions was not identical. Freeze out techniques have permitted measurement of the
h
2 16 mmid CARRIER 100ml N2/rnin.
90 ml 410 ml
SAMPLE BUFFER COLUMN (7.5cm x 2.16 rnmid) 50/80 MESH POROPAK N
I
p”Mp{
44r;b&LE
Figure 1. Eight-port, two-position sample backflush value
0013-936X/80/09 14-0836$01.00/0 @ 1980 American Chemical Society
0715
0700
0730
0745
1300
1245
1315
I330
1345
1800
I815
I030
1845
PACIFIC STANDARD TIME, 15 JULY 1977
Figure 2. Sample chromatogram from automated operation. Upper trace shows the two-carbon hydrocarbons and the full sensitivity. Lower trace shows the methane peak at 20-fold attenuation
C3+ backflush peak at
Table 1. Calibration with Light Hydrocarbons compd
ppb a
WmJ
peak helght scale, dlvlslons
methane methane ethane ethene acetylene propane propane
1160 58.8 49.8 51.0 55.0 46.2 607
757 38.3 60.9 58.3 58.3 82.9 1089
34 35.7 25.5 32.2 26.8 6.2 46
concn
a
Parts per billion b y moles.
an.
response,
X
mV
20 1 1 1 1 1 2
K x
103,
mV pg-1 m--3
6.80 0.357 0.255 0.322 0.268 0.062 0.920
8.98 9.32 4.19 5.53 4.60 0.748 0.844
3dlvlslon
1.11 1.07 2.39 1.81 2.17 13.4 11.8
Backflush mode.
Table It. Joint Distribution of NonmethaneiEthane Organics with Acetylene C2H2,pg/m3
0
500
(C3' 2000
1500
1000
NO
0-