Document dYEGV74kDmybVDj3q71ndZn3G

V >. TECHNICAL REPRINT: GPC-JGC-568 ^ DATE REPRINTED: SEPT/1968 WATERS ASSOCIATES GAS CHROMATOGRAPHIC DETERMINATION . ; OF - A; - TO C|2 HYDROCARBONS IN AUTOMOTIVE EXHAUST Fcipa, L. J., Dinsel, D. L., and Harris, W. C; (E. 1. duPont de Nemours 8r Company) .--v REPRINTED FROM THE JOURNAL OF GAS CHROMATOGRAPHY; v; -> rU;,,... . ... . - MAY, 1968 WITH THE PERMISSION OF THE COPYRIGHT -r ' :: . OWNER, PRESTON TECHNICAL ABSTRACTS COMPANY - ^ ^'IWMMSOCIMB INC./81 FOUNTAIN ST./FRAMINGHAM. MASSACHUSETTS 01701 'wr.v. DSW 298102 STLCOPCB4068013 Gas Chromatographic Determination of C, to C12 Hydrocarbons in Automotive Exhaust* by Louis J. Papa, Donald L. Dinsel, and William C. Harris, E. I. du Pont de Nemours & Company, Organic Chemicals Department, Petroleum Laboratory, Wilmington, Delaware 19898 Abstract A gas chromatographic method is presented for the deter mination of C, to CI2 hydrocarbons in automotive exhaust. The minimum detectable concentration of each hydrocarbon is one part per billion (v/v). The method has detected about 200 individual peaks in chromatograms of automotive exhaust which represent well over 200 paraffinic, olefinic, and aro matic hydrocarbons. The total analysis time is 25 to 30 min utes. Because of its high sensitivity and excellent resolution, the method should have a wide range of application for auto motive exhaust including exhaust from vehicles equipped with emission control devices and in air pollution work; e.g., fuel analysis, atmospheric analysis, and smog chamber re action studies. Introduction Automotive exhaust has been reported to be a sig nificant contributor to air pollution, especially in urban areas, and intensive studies are in progress to mini mize pollution from this source. These studies have necessitated the development of analytical techniques capable of obtaining definitive information on the com position of exhaust. A knowledge of the hydrocarbon composition in exhaust is needed because hydrocar bons are known to take part in the formation of photo chemical smog (1,2). Individual hydrocarbons can have widely varying reaction rates of the type gen erally associated with smog formation (1-8). Hence it is important to have analytical capabilities for de termining as many of the individual hydrocarbons as possible. Several reactivity scales have been proposed for calculating a "reactivity index" of exhaust which is indicative of the potential of the exhaust to form smog. All of these reactivity scales require an analyt ical technique which can determine individual hydro carbons. Many gas chromatographic methods for determin ing hydrocarbons in automotive exhaust have been re ported in the literature (9-16). Only the methods of Jacobs (13) and McEwen (16) provide sufficiently complete analysis to be considered. The former (13) gives excellent resolution of components but is lacking in sensitivity and reproducibility. The latter (16) has sufficient sensitivity but exhibits less resolution than that reported here, especially in the C;-C1= region. The method described in this paper is a much im proved version of that previously reported by Jacobs (13). The sensitivity of Jacobs' original method was increased five-hundred-fold (to one part per billion) and the precision four-fold (to 1.5% relative) by several modifications. These are: 1. Replaced the 0.01 in, i.d., capillary column with a 0.02 in capillary column. 2. Lowered the initial temperature to --70C to effect a cold-trapping of the injected sample at the column inlet and hence keep the injection profile as a narrow band (for best resolution). 3 Removed sample splitter (gives increased sen sitivity and precision). 1. Haagen-Smit, A. J., Scientific American 210, 25 (1964). 2. Leighton, P. A., "Photochemistry of Air Pollution," Academic Press, New York, 1961. 3. Altshuller, A. P., and Bufalini, J. J., Photochemistry and Photobiology 4. 97 (1965). 4. Altshuller, A. P., J. Air Pollution Control Assoc. 16, 257 (1966). 5. Glasson, W. A., and Tuesday, C. S., Hydrocarbon Re activity in the Atmospheric Photooxidation of Nitric Oxide, Presented at 150th National Meeting, ACS, At lantic City, New Jersey, September 12-17, 1965. 6. Jackson, M. W,, Effects of Some Engine Variables and Control Systems on Composition and Reactivity of Exhaust Hydrocarbons, Presented at Mid-Year Meet ing, Society of Automotive Engineers, Detroit, Mich igan, 1966. 7. Wayne, L. G., The Chemistry of Urban Atmospheres, Technical Progress Report III, Los Angeles Air Pollu tion Control District, December 1962. 8. Wimmer, D. B., Coyner, H. M., and Gragson, J. T., J. Am. Chem. Soc. 10, No. 2: B-5 (1965). 9. Ferrin, C. R., Chase, J. O., and Hurn, R. W., "Gas Chromatography," N. Brenner, J. E. Cullen, and M. D. Weise, Eds., Academic Press, New York, 1962, p. 423. 10. Heaton, W. B., and Wentworth, J. T., Anal. Chem. 31, 349 (1959). 11. Hurn, R. W., et al., Proc. Am. Petrol. Inst. Ill 42, 657 (1962). 12. Innes, W. B., Bambrick, W. E., and Andreatcli, A. J., Anal. Chem. 35, 1198 (1963). 13. Jacobs, E. S., Anal. Chem. 38, 43 (1966). 14. Jones, J. L., et al., J. Air Pollution Control Assoc. 13, 73 (1963). 15. Mayrsohn, H,, and O'Neal, Q., Division of Water and Waste Chemistry, 148th National Meeting, American Chemical Society, Chicago, Illinois, August 30 to Sep tember 4, 1964. 16. McEwen, D. J., Anal. Chem. 38, 1047 (1966). *Rescarch and Development Division Contribution No. Hi. 270 JOURNAL OF GAS CHROMATOGRAPHY VOL. 6 MAY 1968 DSW 298103 STLCOPCB4068014 4. Incorporated an auxiliary detector flow <ADF) of nitrogen at the column exit to increase the flame sensitivity and simultaneously increase the resolution by acting as a "sweep gas." 5. Changed to on-column injection. The modifications mentioned above provided the desired increase in sensitivity without a loss in resolu tion except for the C, and C. hydrocarbons which are unresolved. Therefore, it was necessary to determine these compounds on a separate column. This was ac complished on a packed column. The two columns (capillary and packed) were valved together (sharing a common flame detector) via a six-port valve in such a manner as to allow the C.-C- determination to be performed either while the capillary column was being cooled or after the determination was com pleted. Alternatively, the two columns can be and have been used with two separate instruments or with two flame detectors in a differential or dual unit. 3 SAMPLE SPLITTER 4 CAPILLARY COLUMN 5 OVEN FAN 6. "SWAGEL0K" "TEE"(FOR ADF I 9 SIX-PORT VALVE ASSEMSLT ( SEE FIG 3) 10 PACKED COLUMN II TRAYS ( ALUMINUM) OF DRY ICE Figure 1. Gas Chromatograph Experimental Apparatus A Perkin-Elmer Model 800 gas chromatograph was used with several modifications (Figure 1). These are: 1. Two gas sampling valves were mounted on the top of the column oven. 2. The inlet end of the capillary column was run through the top of the column oven (V't in hole drilled in the top) and attached to one of the gas sample valves. A slotted cork was used to seal the hole. 3. The second gas sample valve was piped with a six-port flow switching valve and the Porapak column (Figure 2). 4. An auxiliary detector flow (ADF) of nitrogen was added at the column outlet by passing the gas through an appropriate restrictor (packed column, valve, etc.) to the six-port valve men tioned above (Figure 2) and then into the de tector oven. 5. The detector oven was insulated from the col umn oven with fiber packing and asbestos sheet ing. 6. The instrument pyrometer was replaced with one (--60 to +250 C) obtained from Assembly Products, Inc., Model No. 429-8396. Gas samples were injected with Perkin-Elmer gas sample valves (part No. 154-0067). A 31/2 ft by 14 in, o.d., column packed with a 1:1 intimate mixture of Porapak Q and Porapak T (80/ 100 mesh) obtained from Waters Associates was used for C,-C. determinations. The columns were packed, precoiled with an automatic pressure type packer. A 150 ft by 0.02 in, i.d., capillary column coated with Dow-Corning DC-200 silicone oil was used for the C.-C,., determinations. This column was purchased from the Perkin-Elmer Corporation. A Leeds and Northrup Type G Speedomax re- = C,-C2 DETERMINATION POSITION ----- BACKFLUSH POSITION . Figure 2. Six-Port Valve Assembly ForC,-C,j Determination corder with a 2 mv range, 14 second full-scale pen re sponse was used to record all chromatograms. The recorder was also equipped with an Insco multispeed chart drive which provides variable speeds from 1 to 1/60 of the basic chart speed of 12 in/min. An Infotronics Model CRS-11HSB extended range digital integrator with heavy duty printer and auto matic electronic base line drift corrector was used for all quantitative peak area measurements. Reagents Gaseous hydrocarbons (pure and diluted with ni trogen) were obtained from Scott Research Labora tories, Matheson Company, and Phillips Petroleum Company. Liquid hydrocarbons (pure and in mix tures) were obtained from Phillips Petroleum Com pany, PolyScience Corporation, Chemical Samples Company, and Microtek (Distanal Inc.). In doubtful cases, time-of-flight mass spectrometry, gas chroma tography, and nuclear magnetic resonance techniques were used to check the identities of pure components and mixtures of hydrocarbons. Seaford grade nitrogen (Airco) was used for pre paring gaseous standards, for the auxiliary detector JOURNAL OF GAS CHROMATOGRAPHY VOL. 6 MAY 1968 271 DSW298104 STLCOPCB4068015 flow, and for the packed column carrier gas. Extra dry electrolytic hydrogen (Matheson) was used as the detector fuel along with compressed breathing air (Linde or Airco). Helium (Airco) was used as the carrier gas for the capillary column. All of these gases were passed through Gas-Dry filter traps (Chemical Research Services, Inc.), which consist of Molecular Sieve 5A and indicator Drierite, for a final purification. Procedure The gas chromatographic conditions for the deter mination of the C, to Cr. hydrocarbons are listed in Table I. The capillary column oven is cooled with powdered dry ice contained in two aluminum trays (8 x 1.5 x 1.25 in) placed inside the column oven (Figure 1). More than one tray filling is generally necessary to reach --70C. The last filling should be made when the oven reaches --55 to --60C to insure that the temperature will hold at --70C for 5 min. When this time has elapsed, a gas sample is injected into the capillary column. Forty-five seconds after the sample is injected, the trays of ice are removed, the column oven temperature is set at +30C, and full heat is ap plied to the oven. Five minutes after the sample is in jected, the column oven temperature is programmed from +30C to +120C at lO'C/'min. When all of the hydrocarbons have eluted, a second sample is injected into the Porapak column with the six-port valve posi tioned to allow the column effluent to reach the de tector (in this position the ADF goes to vent). The C, and C, hydrocarbons are now allowed to elute through the column to the detector. After acetylene has completely eluted (--5 min), the six-port valve is rotated to backflush the Porapak column (ADF is now routed to the flame), and the analysis is com pleted. Alternatively, the C,-C, determination can be performed during the 5-minute cooling period (see Discussion for details). Calibration The flame conditions were deliberately selected to provide a similar response (area/ppm carbon) for every hydrocarbon determined in the method. This was veri fied by quantitatively preparing gaseous standards (in nitrogen) and liquid standards (see Tables II and III for sample results). The liquid standards were pre pared by weight on an analytical balance. The gas eous standards were prepared by: 1. Volumetric techniques which involved the use of a wet test meter to measure the nitrogen and a gas-tight syringe to measure the hydrocarbon gases. The standards were prepared in Saran bags (30 in square). This technique was used only for 0,-0, hydrocarbons. 2. Partial pressure techniques in which the hydro carbon (s) and diluent (dry nitrogen) were placed in evacuated stainless steel cylinders at known pressures. When the linearity of response was verified, a primary calibration was made by preparing many standards of propane and later the 0,-0, components using the above techniques. After this primary calibration was Table I. Chromatographic Conditions C,-C, Detn. C:,-C,2Detn. Column Substrate Column Length Column Diameter Column Material Column Temp. Start End Program Rate Detector Temp. Carrier Gas Carrier Gas Flow Air Flow (Detector) H, Flow (Detector) N.. (ADF) Flow Sample Size Porapak T and Q DC-200 3i/2 ft 150 ft i/g in, o.d. 0.02 in, i.d. Stainless Steel Stainless Steel Room Temp. Room Temp. 0 150C Nitrogen 35 ml/min --400 ml/min 60 ml/min 5 ml -- -65 to --70C + 120C 10C/min 150 C Helium 12 ml/min .--400 ml/min 60 ml/min 35 ml/min 5 ml Table II. Relative Responses for Gaseous Diluted Hydrocarbons Hydrocarbons Relative Response (Area/ppm Carbon) Instr. No. 1 Instr. No. 2 Methane Ethane Ethylene Acetylene Propane Propylene Cyclopropane n-Pentane n-Heptane Benzene Toluene Ethylbenzene 1.01 1.01 1.01 1.04 1.00 0.99 1.01 0.97 1.00 1.00 1.01 0.97 1.01 1.00 1.01 1.03 0.99 0.98 0.98 0.98 1.00 0.99 0.98 0.97 Table III. Relative Responses for Liquid Hydrocarbons Hydrocarbons Relative Response (Area/m Mole Carbon) Instr. No. 1 Instr. No. 2 n-Pentane n-Heptane Isooctane 2-Methyl-1-Pentene Octene-1 Decene-1 Toluene Ethylbenzene o-Xylene Benzene 0.96 1.00 1.02 0.96 1.03 1.00 1.00 1.00 1.00 0.97 0.97 1.00 1.00 0.97 0.97 1.05 0.96 0.98 0.99 1.00 272 JOURNAL OF GAS CHROMATOGRAPHY VOL. 6 MAY 1968 DSW 298105 STLCOPCB4068016 established, two cylinders of dilute hydrocarbon mix tures were purchased from Scott Research Labora tories. One of the cylinders contained a mixture of cyclopropane, n-pentane, benzene, n-heptane, toluene, and ethylbenzene (25-50 ppm) and the other con tained a mixture of methane, ethane, ethylene, and acetylene (^--1000 ppm). These cylinders were then calibrated from the primary calibration and are used as daily standards. These standards are run twice daily (morning and afternoon) to insure that the operation is functioning properly. Discussion As mentioned previously, the original method de veloped by Jacobs (13) does not have sufficient sensi tivity to determine hydrocarbon concentrations in di luted exhaust samples and in exhaust from engines equipped with emission control devices. Another draw back to the original method is that several 0.01 in, i.d., capillary columns (DC-200) subsequently pur chased from Perkin-Elmer (and made in this labora tory) could not reproduce the necessary separation of the Cj and C,, hydrocarbons. Therefore, it became nec essary to modify the method to give increased sensi tivity without the loss of resolution. This was accom plished by increasing the sensitivity of the detection system, using a purge gas at the column outlet, chang ing column diameter, increasing the sample size by removal of the splitter, injecting directly on-column, and changing the temperature program. Detection System The sensitivity of the detector supplied by PerkinElmer (ceramic jet type) was increased by the addi tion of nitrogen as an auxiliary detector flow (ADF) to the air-fuel mixture. The precise mechanism for the increased sensitivity is not known although it is re lated to the thermal conductivity of the added gas (17). Figure 3 is a plot of response vs. the ratio of hydrogen to nitrogen and demonstrates the change in sensitivity due to the added nitrogen. The actual gas flows (Table I) were selected to give the maximum response per carbon while maintaining a linear re sponse per carbon for all of the hydrocarbons regard less of class. The relative responses for several repre sentative hydrocarbons are listed in Table II. The ADF was added at the column outlet with a tee rather than directly into the hydrogen line to serve as a purge gas and hence to increase the resolution by reducing tailing and minimizing band spreading. When this addition was made in the column oven, ex cessive noise resulted. This noise could be eliminated by wrapping the tee with asbestos tape or inserting the tee in the detector oven. The latter approach was pre ferred. The detector should be well insulated to minimize temperature changes during the course of the deter mination. Detector temperature changes lead to re sponse changes (at least for this instrument) for some of the hydrocarbons; acetylene appears to be the worst offender in this respect. The detector oven of this in strument was created by insulating the detector section of the column oven with asbesfos sheeting and fiber glass insulation. The insulation helps to stabilize the temperature but the detector oven still changes during the initial 5-minute cooling period. Because of this ini tial detector temperature change, the C,-C.. determina tion is best performed after the C^-C,- determination is completed. At this point, the detector temperature is stable. When the determination was performed dur ing the 5-minute cooling period, 5 per cent variations were experienced with acetylene. A better-insulated detector or a different instrument with a separate de tector oven would eliminate this effect and allow the 0,-C;. determination to be performed during the 5-min ute cooling period without a response variation. This would result in a 5-minute time savings. An alternative approach would be the use of two separate instruments. Both of these approaches have been tried with success. Carrier Gas Helium and nitrogen were evaluated as the carrier gases for the capillary column (Cr,-C12 determination). Helium was selected because it gave better resolution of some of the individual hydrocarbons and smaller response differences per change in the carrier gas flow rate. This is important because the flow rate of the carrier gas does vary due to the wide range of tem perature in the method. Nitrogen was selected for the Figure 3. Effect of Auxiliary Detector Flow on Response 17. Hoffman, R. L,, and Evans, C. D., Science 153, 173 (1966). JOURNAL OF GAS CHROMATOGRAPHY VOL. 6 MAY 1968 273 DSW 298106 STLCOPCB4068017 C,-Cj determination because it gave adequate resolu tion and an increased sensitivity over that obtained with helium. The flow rate was set to match that used for the ADF in the C:,-C1L. determination. In this man ner, the flame conditions, i.e., the hydrogen, air, he lium, and nitrogen flows, are the same for both deter minations, and the same response factors (area/ppm carbon) are obtained for all of the hydrocarbons. Columns The choice of the capillary column was discussed previously (13). The column diameter was changed to 0.02 in, i.d., to better accommodate a larger sample size by reducing the possibility of overloading. Also, the large diameter column has a longer lifetime. The best column found for the C,-C- separation was a mixed column of Porapak T and Q. This column was much superior to adsorption columns because it provided symmetrical peaks, excellent reproducibility, did not require activation, and could be used at am bient temperatures for this separation. A mixture was necessary because "Q" did not separate ethylene and acetylene at all, and "T" gave incomplete separation of ethylene and ethane. The actual column selected consisted of the two Porapaks intimately mixed in a 1:1 proportion. However, two separate columns in se ries, one containing "Q" and the other "T," can be used. If so, either column may be placed first. The col umn order is not important for this separation although Hildebrand and Reilly (18) have shown that it can be very important for many separations. Temperature Program The temperature program begins with a 5-minute hold period at --70C before sample injection. This period assures that the column attains a low enough temperature to perform the desired separation. Fur thermore, the column attains a sufficiently low temper ature to act as a "cold trap" on the injected sample. This cold-trapping allows a relatively large sample (5-15 cc) to be condensed into a rather narrow band at the head of the column. Thus good resolution can be achieved for large sample volumes without the need for a sample splitter. The removal of the sample split ter greatly increased the sensitivity (150-foid) and precision (3-4-fold) of the method. With the sample splitter gone, the sample is now injected directly onto the head of the column by attaching the column di rectly to the gas sample valve, external to the column oven (Figure 1) rather than passing the sample through the entire inlet system provided with the chro matograph. When the sample is injected in this man ner, the resolution of components is markedly increased especially in the Ch-C12 region. This increased resolu tion is illustrated in Figure 4 which shows the back section of two chromatograms obtained with and with out on-column injections. It can be seen that in the on-column injection case the peaks are much sharper, exhibit a marked decrease in tailing and are much better resolved. This contributes to the increased sen sitivity and precision. All of the improvements men tioned above are a direct result of the initial cold pe riod. It is perhaps noteworthy that if a small section of the column inlet (about 1 foot) is immersed in a OJi V) low temperature coolant (external to the column oven) such as liquid nitrogen, a much more efficient trapping will result which will permit larger samples to be in jected and hence give a further increase in sensitivity. This technique has allowed us to inject samples as large as 50 ml, thus giving another tenfold sensitivity increase. This technique can be very useful in atmos pheric analysis and smog chamber work where the hydrocarbons are present in very small quantities. Further increases in sensitivity can probably be ob tained by injecting still larger samples. The upper limit on sample size has not yet been determined. Cold trap ping in this manner also separates methane from the C. hydrocarbons. The sample is injected after the initial hold period at --70C. The temperature is then brought up to 120C in three steps as described in the procedure. This mode of programming was selected to give the desired separations in the simplest manner possible. The upper temperature was limited to +120C to help prolong column life. Chromatographic Sampling Liquid samples, such as prepared standards and gasolines, were injected (--0.2 gl) with a Hamilton microliter syringe using a different chromatographic setup than was shown for the gaseous determinations 18. Hildebrand, G. P., and Reilley, C. N., Anal. Chem. 36, 47 (1964). 274 JOURNAL OF GAS CHROMATOGRAPHY VOL. 6 MAY 1968 . DSW298107 STLCOPCB4068018 (Figure 1). In this ease the column was reattached to the injection port (heated to 16(T0) and a 16:1 sam ple splitter was used. All other conditions (tempera ture program, detection system, etc.) were the same as reported above. The results were calculated as area per cent which is directly related to mole per cent because of the linearity in responses for the various hydrocarbons (see Tables I and II). Thus, the in jected volume is not critical as long as the column is not overloaded and all of the components elute. Gaseous samples are injected with a gas sample valve using the modified apparatus previously de scribed (Figure 1). The sample is introduced into the loop from a plastic bag in either of two methods: flow through or evacuation. The first method is performed by squeezing the bag and purging the loop. The gas is vented through an exit line which is immersed in water to provide pressure equilibration and to prevent air from backing up into the loop. The second is ac complished by evacuating the loop (via appropriate hardware) and then opening it to the sample bag to allow the loop to fill to atmospheric pressure. When a valve is in good operating condition, i.e., clean and leak-free, these methods are equivalent. The former is preferred because of its simplicity. However, when valves are in constant use, they tend to become "dirty" (although not visible to the eye), and adsorption oc curs in the valve body. This leads to response factors (area/ppm carbon) which decrease with increasing molecular weight when the evacuation technique is used and increase with increasing molecular weight when the flow-through technique is used. The increases and decreases are most marked with the aromatic hy drocarbons. It is believed that a contaminant film, per haps of high molecular weight hydrocarbon i s), ac cumulates in the valve body to cause these effects. Thus, in the evacuation technique, hydrocarbons are lost by adsorption into the contaminant film leading to a response decrease. On the other hand, the flow through technique leads to a build-up of hydrocarbons in the film. Injecting the sample causes the valve to be purged by the carrier gas which now sweeps the excess hydrocarbons out of the film and into the column thus causing the increasing response factors. Verifications for this belief are: 1. The longer the loop is purged (on a dirty valve) during the flow-through technique, the larger are the increases in the response factors. 2. The valves can sometimes be restored to their original operating condition by heating at 100C while under vacuum. "3 3. Thus far the valves have always been restored to their original operating condition by cleaning in an ultrasonic bath containing "Freon-113" Trichlorotrifluoroethane (one to two hours in each valve position). It is noteworthy that not all new valves were found to operate properly without prior cleaning. The adsorp tion effect must, of course, be watched closely if ac curate determinations are to be made (see Standard ization section). Calibration gases are transferred from pressurized cylinder to plastic bags using a needle valve as an ex pansion apparatus to prevent fractionation. In prin- Figure 5. Exhaust Sample Collection Apparatus ciple, this is similar to the apparatus described by McEwen (16). The gas is introduced into the needle valve at cylinder pressure, and the cylinder valve is closed. The entire gas sample in the needle valve is then discharged into the plastic bag. This process is repeated until the bag is filled. Exhaust Sampling Exhaust samples are collected by the California cycle bag procedure described by Zelson (19). The chromatographic sample is then transferred as quickly as possible to a smaller bag (18 in square) in the di luted form by partially prefilling the smaller bag with nitrogen and then admitting the raw exhaust. This is accomplished by placing the prefilled bag into a suit able container (a Fibrepak drum was used in this work) and making a short, direct connection from the bag to the exhaust source through the lid of the con tainer (see Figure 5). Air is then drawn from the con tainer (around the outside of the bag) via a pump, and exhaust enters the bag to equalize the pressure. In this manner, raw exhaust enters the bag having only passed through a small amount of tubing (stainless steel and/or "Teflon" TFE Fluorocarbon Resin). The exhaust is diluted, thereby preventing water conden sation and minimizing hydrocarbon losses which can occur by reaction, adsorption, etc. Simultaneously, an empty bag is filled with raw exhaust which has passed through two ice-cooled traps in series. The carbon di oxide content is determined for each of these bags by another gas chromatographic procedure. These carbon dioxide values are then used to calculate the dilution ratio in the sample bag. A dilution ratio of 3-4:1 is used for exhaust from normal vehicles and of 2-3:1 for exhaust from vehicles equipped with air-injection emis sion control devices. Bag Materials The material of which the chromatographic sample bag is constructed is very important. These small 19. Zelson, J., California Cycle Bag Technique For Ve hicle Emissions Testing, Du Pont Technical Memo randum 10,006, March 1, 1966. JOURNAL OF GAS CHROMATOGRAPHY VOL. 6 MAY 1968 275 DSW 298108 STLCOPCB4068019 Figure 6. Hydrocarbon Losses In Polyethylene Bag RESIDENCE TIME IN BAG (MIN.) Figure 7. Ethylbenzene Losses In Several Plastic Bags bags, unlike the large bag used for the total ex haust collection, have a high surface to volume ratio (--'10 ft-/ft3), unconditioned walls, and no water film. Therefore, hydrocarbons are much more likely to be lost by adsorption, permeation, etc. Several bag mate rials were studied in an effort to select one for the ex haust collection. Decay curves were constructed for several hydrocarbons in various plastic bags. A typical example is shown in Figure 6 for six hydrocarbons in a polyethylene bag. Ethylbenzene exhibited the most rapid disappearance in the polyethylene bag as well as in the other bags tested ["Tedlar" PVF Polyvinylfluoride Film, "Teflon" FEP Fluorocarbon Resin, "My lar" type 75M22 (coated "Mylar")]. A plot of ethyl benzene concentration vs. residence time, for all of the bags tested, is shown in Figure 7. "Teflon" and "Ted lar" had essentially no effect on the composition of the sample over a 3-hour period. On the basis of these tests, it is recommended that "Teflon" or "Tedlar" bags be used for exhaust sampling. Studies conducted on sampling line materials yielded similar results. Nylon is better than polyethyl ene, but neither is considered satisfactory. Sampling lines of "Teflon" and/or stainless steel are recom mended. Needless to say, it is necessary for all mate rials that contact the sample to be clean. Standardization As mentioned previously, it is desirable to calibrate twice daily to be sure that everything is functioning properly. The C,-C. method is calibrated with a mix ture composed of methane, ethane, ethylene, and acetylene. The C.-C,... method is calibrated with a mix ture composed of cyclopropane, n-pentane, benzene, nheptane, toluene, and ethylbenzene. These compounds were selected for three reasons: 1. They cover most of the retention time range of normal exhaust chromatograms. 2. It is desirable to have more than one hydrocar bon class present to insure that the detector response has not changed for a given class. That is, if the flame conditions are changed, the re sponse of one class, aromatics for example, might change significantly while the response of paraffins does not. 3. If any of the sample handling equipment, such as the gas sampling valve, becomes dirty, ad sorption of hydrocarbons can occur causing er roneous results. When this occurs, the aromatics are generally affected to the greatest extent. Significant changes in the response factors for toluene and ethylbenzene will pinpoint this effect. Data Handling The chromatographic signal is fed to an Infotronics Model CRS-11HSB extended range digital integrator to integrate each of the peak areas as the peaks emerge from the column. The peaks are simultaneously dis played on a strip chart recorder. The electrical connec tions to both the strip chart recorder and the integra tor are made from the 0-5 mv taps located behind the gas chromatograph. This setup allows peaks with 8 mv heights (4 times larger than full scale) to be ac curately integrated before a nonlinear condition is reached due to saturation of the amplifier. Thus the attenuation must be selected to satisfy this require ment. When it is desired to accurately integrate the large peaks and also detect and integrate the very small peaks, the attenuation should be adjusted during the course of the determination. The peaks are then 276 JOURNAL OF GAS CHROMATOGRAPHY VOL. 6 MAY 1968 DSW 298109 STLCOPCB4068020 identified by peak number from a "standard" chro matogram and these peak numbers, peak areas, and attenuation levels of the signal are keypunched onto IBM cards along with the response factor (integrator counts per ppm carbon). The cards are then fed into a computer which is programmed to print out concen trations and reactivity indices. Identification of Hydrocarbons The hydrocarbons in the exhaust chromatograms were identified by several techniques. Known, pure hy drocarbons were added to previously run exhaust sam ples so that the peak number could be determined. Also, time-of-flight mass spectrometry and gas chro matography (in combination) were used to positively identify many of the components. Many exhaust sam ples were run through the chromatographic procedure as described and with a mercuric perchlorate precol umn used to remove olefins and aromatics. This aided in the identifications and pointed out several cases of overlap between paraffins and non-paraffins. Because there are relatively few cases of this type of overlap, Figure 8. Gas Chromatogram of Typical Exhaust Sample JOURNAL OF GAS CHROMATOGRAPHY VOL. 6 MAY 1968 277 STLCOPCB4068021 it was found that using an average reactivity index for the overlapping components did not significantly alter the final reactivity index for the exhaust sample. Thus it was unnecessary to make two runs on each ex haust sample, i.e., one regular and one with a subtrac tive column. A gas chromatogram of a typical exhaust sample is shown in Figure 8. This particular sample contained approximately 1200 ppm (mole basis) hydrocarbons. This figure illustrates the high degree of resolution ob tained with this method. The sensitivity which can be achieved with this method can be seen from peak No. 48, 2,4-dimethypentane, which is 1.0 ppm in this par ticular sample (run at 200X attenuation). A complete list of all of the hydrocarbon peaks that have been detected to date is shown in Table IV. The peaks are listed by peak number. As can be seen, many are still unknown but, in general, these amount to less than 4 per cent of the total moles of hydrocarbon in automotive exhaust. Table IV. Identification of Hydrocarbon Peaks Peak Number Hydrocarbons Peak Number Hydrocarbons 1 Methane 2 Ethane 3 Ethylene 4 Acetylene 5 Propylene 6 Propane 7 Cyclopropane 8 Propadiene 9 Methylacetylene 10 Isobutane ; 11 Isobutylene and/or 1-Butene 12 1,3-Butadiene 13 n-Butane 14 trans-2-Butene 15 Unknown 15 16 cis-2-Butene 17 Unknown 17 18 3-Methyl-l-butene 19 Isopentane 20 1-Pentene 21 2-Methyl-l-butene 22 n-Pentane 23 2-Methyl-1-3-butadiene 24 trans-2-Pentene 25 cis-2-Pentene 26 2-Methyl-2-butene 27 Unknown 27 28 2,2-Dimethylbutane 29 Unknown 29 30 Unknown 30 31 Cyclopentene 32 4-Methyl-l-pentene and/or 3-Methyl-l-pentene 33 Cyclopentane 34 2,3-Dimethylbutane 35 2-Methylpentane 36 4-Methyl-cis-2-pentene 37 3-Methylpentane 38 2-Methyl-l-pentene and/or 1-Hexene 39 2-Ethyl-1 -butene 40 n-Hexane 41 trans-3-Hexene 42 trans-2-Hexene 43 2-Methyl-2-pentene 44 cis-3-Hexene 45 cis-2-Hexene 46 3-Methyl-trans-2-pentene and/or 3-Methyl-cis-2-pentene 47 Methylcyclopentane 48 2,4-Dimethylpentane 49 2,2,3-Trimethylbutane 50 3,4-Dimethyl-l-pentene 51 4,4-Dimethyl-cis-2-pentene - 52 3,3-Dimethylpentane 53 Benzene 54 Cyclohexane . 55 3-Ethyl-l-pentene 56 5-Methyl-l-hexene 57 4-MethyI-l-hexene 58 2-Methylhexane and/or 2,3-Dimethylpentane 59 Cyclohexene 60 3-Methylhexane 61 Unknown 61 62 Unknown 62 63 Unknown 63 64 2,2,4-Trimethylpentane 65 1-Heptene 66 Unknown 66 67 trans-3-Heptene 68 n-Heptane 69 cis-3-Heptene and/or 3-Ethyl-trans-2-pentene 70 2,4,4-Trimethyl-l-pentene and/or trans-2-Heptene 71 cis-2-Heptene 72 2,5-Dimethyl-trans-3-hexene 73 Methylcyclohexane 74 Unknown 74 75 Unknown 75 76 2,4,4-Trimethyl-2-pentene 77 4-Methyl-l-cyclohexene 78 2,4-Dimethylhexane and/or 2,5-Dimethylhexane 79 2,2,3-Trimethylpentane 80 Unknown 80 81 4-Methylheptane 82 2,3,4-Trimethylpentane 83 Unknown 83 84 Toluene 85 Unknown 85 DSW298111 278 JOURNAL OF GAS CHROMATOGRAPHY VOL. 6 MAY 1968 STLCOPCB4068022 Table IV. (continued) Peak Number Hydrocarbons Peak Number Hydrocarbons 86 2,3.3-Trimethylpentane 87 2,5-Dimethyl-trans-2-hexene 88 2-Methyl-3-ethylpentane and/or 2,3-Dimethylhexane 89 Unknown 89 90 3,4-Dimethylhexane and/or 3-Methylheptane 91 Unknown 91 92 2,2,5-Trimethylhexane 93 1-Octene 94 trans-l,2-Dimethylcyclohexane 95 Unknown 95 96 n-Octane 97 trans-2-Octene 98 Unknown 98 99 Dimethylheptane 100 cis-2-Octene 101 cis-l,2-Dimethylcyclohexane 102 Unknown 102 103 Ethylcyclohexane 104 Unknown 104 105 Unknown 105 106 Unknown 106 107 Ethylbenzene 108 Unknown 108 109 m + p-Xylene 110 Unknown 110 111 Unknown 111 112 Unknown 112 113 o-Xylene 114 2-Methyloctane 115 Unknown 115 116 Unknown 116 117 n-Nonane 118 Unknown 118 119 Isopropylbenzene 120 Unknown 120 121 Unknown 121 122 Unknown 122 123 Unknown 123 124 n-Propylbenzene 125 l-Methyl-4-ethylbenzene and/or l-Methyl-3-ethylbenzene 126 1,3,5-Trimethylbenzene 127 Unknowm127 128 l-MethyI-2-ethyibenzene 129 Unknown 129 130 Unknown 130 131 t-Butylbenzene 132 1,2,4-Trimethylbenzene 133 Unknown 133 134 Isobutylbenzene 135 Unknown 135 136 Unknown 136 137 sec-Butylbenzene 138 l-MethyI-3-isopropylbenzene 139 n-Decane 140 1,2,3-Trimethylbenzene 141 l-Methyl-4-isopropylbenzene 142 1,3-Diethylbenzene 143 Unknown 143 144 n-Butylbenzene and/or l-Methyl-4-n-propylbenzene 145 1,3-Dimethyl-5-ethylbenzene and/ or 1,2-Diethylbenzene 146 l-Methyl-2-n-propylbenzene 147-153 Unknown 154 Durene 155-172 Unknowm 173 1-Dodecene 174-200 Unknown Applications The method presented has been successfully ap plied to automotive exhaust analysis for some time. Applications have already been reported in which the effect of gasoline antiknock additives (20) and gaso line composition (21) on exhaust emissions has been determined. The technique also finds utility in the study of emission control devices. In studies of this type, it is desirable to determine the concentration of individual components rather than a simple total hy drocarbon level. The more definitive data offer a better method of evaluating and designing a device because effects such as selective oxidation can be uncovered. For similar reasons, the technique is desirable for studying the effects of fuel additives and engine vari ables on the composition and reactivity of exhaust. The applications of the method are not restricted to exhaust and fuel analysis. Due to its high sensitivity and excellent resolution, the method should have a wide range of application in air pollution work, e.g., atmospheric analysis, smog chamber reaction studies, and evaporative emissions. Manuscript received September 8, 1967 Manuscript accepted October 23, 1967 20. Pahnke, A. J., and Squire, E. C., Oil and Gas Journal 64, 106 (1966). 21. Pahnke, A. J., et al., Effect of Gasoline Composition on Exhaust Emissions of a Vehicle With Air Injection, presented at 152nd National Meeting. American Chemical Society, New York, September 15, 1966. DSW 298112 JOURNAL OF GAS CHROMATOGRAPHY VOL. 6 MAY 1968 279 STLCOPCB4068023 6 I; i. W. U Pf %*. t ar. 't DSW 298113 tyr1 : ; -V V.^; >a. "n't* WATERS ASSOCIATES # 61 Fountain Street, Framingham, Massachusetts8170! J (617)879-2000 &.%'/' --'VX'. -- : 1;' '-:-.:'. ' >V\: /,. ;>; -H. / "*': ' : :.; V- '"- V- ' ' - 'X*u- `i*?. * '*--.Vf -fc-f7 ? T1. - ''. --tllvV ' 1 u .=-. , . ..X - - '.<; r- v- ',.. t: . -,. TRR- 968/1 &r .'J.. /v' tL`r'"i <A-'. -1-ii "ViiaWr li*-. vi*.. i..y i. j.,. -... STLCOPCB4068024