Use of Chlorofluorocarbons as Internal Standards for the

Environ. Sci. Technol. 2003, 37, 1002-1007

Use of Chlorofluorocarbons as Internal Standards for the Measurement of Atmospheric Non-Methane Volatile Organic Compounds Sampled onto Solid Adsorbent Cartridges CHRISTINE M. KARBIWNYK, CRAIG S. MILLS,‡ D E T L E V H E L M I G , * ,§ A N D J O H N W . B I R K S Department of Chemistry and Biochemistry and Cooperative Institute for Research in Environmental Sciences (CIRES), University of Colorado, Boulder, Colorado 80309-0216

Solid adsorbents have proven useful for determining the vertical profiles of volatile organic compounds (VOCs) using sampling platforms such as balloons, kites, and light aircraft, and those profiles provide valuable information about the sources, sinks, transformations, and transport of atmospheric VOCs. One of the largest contributions to error in VOC concentrations is the estimation of the volume of air sampled on the adsorbent cartridge. These errors arise from different sources, such as variations in pumping flow rates from changes in ambient temperature and pressure with altitude, and decrease in the sampling pump battery power. Another significant source for sampling rate variations are differences in the flow resistance of individual sampling cartridges. To improve the accuracy and precision of VOC measurements, the use of ambient chlorofluorocarbons (CFCs) as internal standards was investigated. A multibed solid adsorbent, AirToxic (Supelco), was chosen for its wide sampling range (C3-C12). Analysis was accomplished by thermal desorption and dual detection GC/FID/ECD, resulting in sensitive and selective detection of both VOCs and CFCs in the same sample. Longlived chlorinated compounds (CFC-11, CFC-12, CFC-113, CCl4, and CH3CCl3) banned by the Montreal Protocol and subsequent amendments were studied for their ability to predict sample volumes using both ground-based and vertical profiling platforms through the boundary layer and free troposphere. Of these compounds, CFC-113 and CCl4 were found to yield the greatest accuracy and precision for sampling volume determination. Use of ambient CFC-113 and CCl4 as internal standards resulted in accuracy and precision of generally better than 10% for the prediction of sample volumes in ground-, balloon-, and aircraft-based measurements. Consequently, use of CFCs as reference compounds can yield a significant improvement of accuracy and precision for ambient VOC measurements in situations where accurate flow control is troublesome. * Corresponding author e-mail: [email protected]. ‡ Current address: Physical and Theoretical Chemistry Laboratory, Department of Chemistry, Oxford University, Oxford, UK OX1 3QZ. § Current address: Institute of Alpine and Arctic Research (INSTAAR), University of Colorado, Boulder, CO 80309-0450. 1002

9

ENVIRONMENTAL SCIENCE & TECHNOLOGY / VOL. 37, NO. 5, 2003

Introduction The measurement of anthropogenic and biogenic VOCs in the troposphere is essential for determining the strengths of emission sources and their potential for contributing to tropospheric ozone formation. Ground samples have been collected and analyzed extensively from various sites, but they do not accurately assess the VOC loading of the atmosphere (1, 2). Vertical profiling of the troposphere for VOCs and other species has recently garnered much attention (1, 3-10). Vertical profiles show the effects of transport phenomena coupled with a general decrease in VOC mixing ratios with increasing altitude as these compounds undergo reactions with ozone and the OH radical (11). Because of the heterogeneity of sources, the true VOC loading of the atmosphere can only be determined from vertical profiles (12). Kites and balloons are often used for vertical profiling (3, 13-16), but these platforms have severe limitations with respect to size and weight of the sampling package. Solid adsorbent cartridges allow small, lightweight sampling packages with the advantage that many packages can be lifted and many samples collected per profile. However, inaccuracies can result from variability in the flow resistance of solid adsorbent cartridges. Ideally, flow calibrations at the ground for each individual cartridge/pump combination are needed; however, this approach is not feasible because this procedure would result in a sample loading and potential contamination of the cartridge. Also, the change of temperature and pressure with altitude often results in volume errors as large as 20% (3). The use of a flow controller with each sampling cartridge greatly increases the package weight and therefore reduces the number of samples that can be obtained for each profile. CFCs are well mixed in the atmosphere due to their long lifetimes and the phase out of their use since ratification of the Montreal Protocol (17, 18). In fact, atmospheric measurements of CFCs have demonstrated a low variability of their mixing ratios on both spatial and temporal scales (19, 20). The standard deviation of the ln of the compound mixing ratios (slnX) has been used as a quantitative measure for description of ambient mixing ratio variability (20, 21). Aircraft data from several large campaigns with VOC and CFC measurements in the troposphere have shown that the mixing ratio variability is inversely correlated with the tropospheric lifetime (20). Consistently, in these data sets, lowest and very similar slnX values (∼0.01-0.025) were observed for CFC-11, CFC-113, CFC-12, and CCl4 (20). Thus, we hypothesized that these CFCs could be used as internal reference compounds for samples collected onto solid adsorbent cartridges. The observation of CFC in VOC samples collected on solid adsorbents has previously been used for general consistency tests of sampling procedures (22). Similarly, this idea has recently been shown to be of value in gas standard preparation, where CCl4 was used as an internal reference compound (23). The research reported in this paper expands on this principle by utilizing ambient CFCs for calibration purposes. This study addresses the need to measure accurately the sample volume that is collected on the solid adsorbent and describes an internal standard method that makes use of well mixed ambient CFCs for the accurate determination of sample volume.

Experimental Section Materials Used. Pure gases were purchased from Airgas Houston, Houston TX: helium (99.999% purity), hydrogen 10.1021/es025910q CCC: $25.00

 2003 American Chemical Society Published on Web 01/28/2003

TABLE 1. Comparison of Chlorofluorocarbon Mixing Ratios in the Niwot Ridge Standard and Global Mean Data (21)a compound

cylinder mixing ratio (ppt)b

global mean mixing ratio (ppt)c

uses

phased out?

lifetime (years)

CFC-12 CFC-11 CFC-113 CH3CCl3 CCl4

546 267 83.6 67.0 91.2

529 266 81.8 65.3 102

refrigerant, vehicle air conditioning centrifugal chillers foam blowing commercial and industrial air conditioning degreasing agent pesticide, cleaning fluid, spot remover

yes yes yes yes yes

102 50 85 5 42

a CFC use characteristics, Montreal Protocol status, and atmospheric lifetime also are given. values.

(99.999% purity), nitrogen (99.999% purity), ultra zero grade air (THC < 0.1 ppm, CO < 1 ppm, CO2 < 1 ppm, H2O < 5 ppm); and electron capture detector (ECD) makeup gas, 5% methane (99.97% purity), 95% argon (99.999% purity). In addition, ambient air standards were collected into pressurized cylinders with an oil-free compressor on June 6, 1996 between 5:30 and 6:30 pm in Boulder, CO and on March 6, 1998 at Niwot Ridge, CO. AirToxics adsorbent cartridges were purchased from Supelco, Bellefonte, PA, and were used for all sample collections. The cartridges are stainless steel, 88.9 mm in length with a 4.8-mm inner diameter and contain adsorbent beds of 35-mm of Carbopack B plus 10-mm of a proprietary molecular sieve separated by glass wool. Supeltex M-2A Vespel ferrules were used with Swagelok fittings to seal the cartridges. Sampling Cartridges. AirToxics adsorbent cartridges were conditioned by flowing through the cartridges approximately 90 mL min-1 (standard temperature and pressure [STP]) of N2 that had passed through oxygen traps and a hydrocarbon trap. Flow was in the desorb direction for 60 min while heating at 350 °C. The cartridges were then capped with Swagelok fittings and stored in sealed glass jars until use. Sampling cartridges were stored in an ice chest before and after sample collection (24). Sample cartridges were weighed before and after sampling to determine the atmospheric water uptake. 10-Port Sampler. Samples were collected onto the solid adsorbent cartridges by use of a sampler constructed in our laboratory that maintained a constant flow rate via a mass flow controller (MFC, Tylan model FC-280SAV). The sampler accommodated 10 sampling cartridges connected by Swagelok fittings. All tubing upstream of the sampling cartridge was silica-lined stainless steel (Silcosteel, Restek, Bellefonte, PA). The sampling flow was controlled with two electrically actuated valves (4-port switching valve and 10port sampling valve) (VICI, Houston, TX). The sampler valves were timed and switched by a computer running a BASIC program, via communication through a serial port. Light Sampling Packages. Battery-powered (0.51 kg), programmable, flow-controlled air sampling pumps (Genie LowFlow, Buck, Inc., Orlando, FL) were used to pull sample air first through a sodium thiosulfate ozone filter and then through the sampling cartridge. Typical collection conditions were 300 mL min-1 with 20 min sampling times. The sampling packages have been illustrated and described in detail previously (3). As an independent means of measuring the sampling volume, the pump vent was collected into a Tedlar sampling bag after passing through the sampling cartridge. The volume of air inside the Tedlar bag was measured at the ground by pumping the air through a wet test meter. Teflon tubing and Teflon PFA compression fittings (Cole-Palmer, Vernon Hills, IL) were used throughout. The flow rate of the pump, which used a differential pressure sensor, an rpm sensor, temperature sensor, and flow sensor to calculate the sample volume collected, was calibrated on the ground. It was programmed to turn on and off at specified times so that all samples could be collected simultaneously at each altitude along the tether line.

b

Collected 1998. c Estimated 1998 global mean

Desorption and Cryogenic Focusing. Using a PerkinElmer ATD-400 autosampler, samples were desorbed at 300 °C for 15 min with UHP helium that had passed through oxygen traps and a hydrocarbon trap. Desorbed compounds were focused onto an adsorbent-filled (AirToxics, PerkinElmer) microtrap held at -25 °C. With the GC oven at 0 °C, samples were injected on the GC column by rapidly heating the cold trap to 325 °C and holding at that temperature for 5 min. Samples were transferred from the cold trap to the GC column by a 0.53 µm ID deactivated fused silica transfer line maintained at 150 °C. GC Separation and Detection. Analytes were separated on a 30 m × 0.32 mm DB-1 column with 5 µm film thickness (J & W Scientific, Folsom, CA). Liquid nitrogen was used to maintain the GC oven at 0 °C for 5 min. The oven temperature was then ramped at 6 °C per minute to 180 °C and held at that temperature for 5 min. Following chromatographic analysis, a second ramp of 30 °C per minute to 250 °C was used to clean the column prior to cooling the column to 0 °C for the next sample. Total sequence time was 42.3 min per sample. The end of the GC column was split into two flows by a glass Y connector. The split ratio was determined by the proper selection of two flow resistors made of capillary column. A deactivated fused silica capillary with a 0.18 mm internal diameter, 100 cm in length, directed 5% of the effluent to an ECD. Another deactivated fused silica capillary that has an internal diameter of 0.32 mm and 53 cm length, directed 95% of the effluent to a flame ionization detector (FID). The reproducibility of the split ratio was found to be equal or better than 1.2% relative standard deviation (see results section). The ECD was operated at 375 °C with 40 mL min-1 of 95% argon, 5% methane make up gas. The FID was operated at 300 °C with flows of 350 mL min-1 of air and 45 mL min-1 of hydrogen. Niwot Ridge Standard Collection. The University of Colorado Mountain Research Station maintains research equipment at several levels of Niwot Ridge Mountain, which is located approximately 35 km west of Boulder, CO. A gas cylinder (11 L, AccuLife- treated by Scott Specialty Gases, Longmont, CO) was filled to 2050 psi with an oil-less compressor on March 6, 1998 at the C-1 station. The station is surrounded by subalpine vegetation and located at an altitude of 3022 m. Wind condition during the collection was upslope at 0-3 mph from the E/SE. The air in this cylinder is referred to as the “Niwot Ridge Standard” throughout this paper. It served as a source of stable, relatively clean ambient air that was used in analytical experiments. The Niwot Ridge Standard was analyzed for CFCs in the Climate Monitoring and Diagnostic Laboratory (CMDL) of the National Oceanic and Atmospheric Administration (NOAA), Boulder, CO. CFC mixing ratios, given in Table 1, are reported at part per trillion (ppt) level dry mole fractions. The certification of the CFCs resulted in a standard that was then used to calibrate the ECD response. The cylinder mixing ratios were compared to estimated global mean values, also given in Table 1. VOL. 37, NO. 5, 2003 / ENVIRONMENTAL SCIENCE & TECHNOLOGY

9

1003

Calibration of CFC Response Factors. Compounds persistently present in the ECD chromatograms of ambient samples were CFC-12, CFC-11, CFC-113, chloroform, methylchloroform, carbon tetrachloride, and tetrachloroethylene. Blank samples occasionally showed contamination peaks of CFC-11 and tetrachloroethylene. The source of this contamination could not be determined but was assumed to be an unidentified laboratory source. CFC-11 (and to a lesser extend CFC-12 under some conditions) suffered from breakthrough at the sample volumes and collection temperatures used throughout this study. Therefore, this study focuses on the compounds CFC-113, carbon tetrachloride, and methylchloroform. CFC calibration was performed using the 10-port sampler to load the solid adsorbent cartridges with 1.5-6.6 L samples of the Niwot Ridge Standard. CFC peak areas were plotted against sample volume (STP liters). CFC-113 calibration had a relative standard deviation (n ) 3) of less than 1% for each sample volume with an r 2 value of 0.9992. The relative standard deviation for the methylchloroform calibration varied from 3 to 13% (n ) 3) with an r 2 value of 0.9966. The carbon tetrachloride calibration had an r 2 value of 0.9962 with relative standard deviations of 1-2% for each mixing ratio, except for one concentration, which had a 7% relative standard deviation (n ) 3). These calibration curves were used to determine sample volumes from ECD chromatograms of air samples. Ambient sampling took place between November 16, 1999 and August 1, 2001. Because of the reduction in CFC release rates and their gradual atmospheric depletion their ambient levels have further decreased since the collection of the calibration reference standard. For instance, tropospheric mixing ratios of CFC-113, methylchloroform, and carbon tetrachloride have been decreasing at rates of approximately 0.5, 10, and 1 ppt per year, respectively (17, 18). Also, the quantitative analysis of CFCs in the Niwot Ridge standard revealed that CFC mixing ratios differed (up to ∼10%) from estimated, global average background levels at the time of the collection of this standard (Table 1). The mixing ratio offsets for each of these CFC were corrected in the use of this standard for the analysis of ambient air in the experiments described below. Second, published ambient CFC decline rates and the times that had passed between the standard collection and the experiment were used to correct for the changes in ambient CFC mixing ratios relative to the Niwot Ridge Standard. Ground Sample Collection at Niwot Ridge, CO. An experiment was carried out at a “clean air” site where influences from point sources were minimal. A total of 27 ambient air samples were collected from a clearing near the C-1 met station site at Niwot ridge on June 18, 2001 using the 10-port sampler between 10:50 and 15:00 MST. The sample inlet line was 1/8-in. silcosteel (Restek) and extended 1.22 m above the ground. Both flow rate and sample time were varied to load different sample volumes onto the solid adsorbent cartridges. Three replicates were collected at each volume. The relative humidity was ∼17% during the sampling period with an ambient temperature of ∼20 °C and a barometric pressure of 537 Torr. Wind speed varied from 1 to 4 m s-1 and was predominately from the west with a short burst of air from the north. Vertical Profiles from an Aircraft Sampling Platform. A Cessna 182 airplane was used to collect vertical profile samples within and above the convective boundary layer to an altitude of 4 km above mean sea level. This experiment served to test the ability of CFCs to predict volumes in an urban area and over a range of altitudes where pressure and temperature vary. The plane was loaded with sampling equipment at the Boulder Municipal Airport before each flight. Samples were collected at constant altitude by circling the Oklahoma Reservoir, located between Greeley and Fort Collins, CO and approximately 4 km east of the Interstate 25 1004

9

ENVIRONMENTAL SCIENCE & TECHNOLOGY / VOL. 37, NO. 5, 2003

highway. Flights were generally begun around noon under conditions of a well developed convective boundary layer. Ambient samples were loaded onto AirToxics solid adsorbent cartridges using the 10-port sampler with a 1/8in. Silcosteel (Restek) inlet line protruding from the cabin air inlet. The inlet line was bent away from the direction of flight to eliminate the effect of ram air on the sampling rate. The sampler was situated in the back seat of the plane and powered by a 12 V battery and power inverter. Simultaneous measurements of temperature, pressure, and percent relative humidity were recorded with a meteorological sonde (Vaisala RS-80) or thermohygrometer probe (Cole Palmer, Vernon Hills, IL). Ozone data also were collected during one flight using a portable UV absorbance instrument (2B Technologies Inc., Golden, CO) with a 1/4-inch Teflon inlet line protruding through the right wing of the plane. These measurements helped to identify the location of the boundary layer for evaluation of the VOC and CFC vertical profile data. The inlet line and sample valves were flushed for approximately 3 min prior to sample collection. Glass fiber filters were soaked in a 12% (w/v) solution of sodium thiosulfate and dried under UHP nitrogen flow before being placed in the inlet line upstream of the sampler to scrub ozone from the sample stream (25). The mass flow controller was set to approximately 775 mL min-1 (STP), and samples were collected for 5 min while flying in a large circle at each altitude. Vertical profiles were collected on November 16, 1999, February 5, 2000, March 4, 2000, and April 8, 2000. Vertical Profiles from a Balloon Sampling Platform. A single ply, urethane balloon (Sky-Doc (26), Floatograph Technologies, Napa, CA), 5.5 m in diameter, was used to collect vertical profiles of ozone, meteorological parameters, and VOCs. A forest clearing, located at 45:33:05 N, 84:46:59 W within the State of Michigan, served as the balloon launch site. Vegetation consisted of beech-maple and successional stages of aspen, oak, and pine trees. Canopy height was estimated at between 12 and 15 m. Lakes were especially numerous in the region, and relative humidity levels during sampling were 75-100%. The region was heavily forested, and, as such, biogenic emissions were expected to dominate the ambient samples collected from the site. Each VOC sampling package consisted of its own adsorbent cartridge, pump, and Teflon bag. Samplers were placed approximately 30 m apart on the sampling line. The pumps were programmed on the ground to start sampling at a specified time so that all the samples were collected simultaneously. It took 15-20 min to get five sampling packages attached to the sampling line and raised to the proper altitude. The balloon was tethered at ∼150 m above ground level. Temperature, pressure, and relative humidity were recorded by a Vaisala RS-80 radiosonde in addition to wind speed and wind direction, which were recorded with a 3-vane cup anemometer and magnetic compass (Vaisala TSP-5A-SP tether sonde). A portable UV ozone detector (2B Technologies) was used for the measurement of ambient ozone mixing ratios. Vertical profiles were collected on August 1, 2000. All balloon sampling pumps were flow-calibrated on the ground (0.24 km ASL) and set to collect 7.5 L samples over 25 min sampling times. Sample volumes were measured after the flight on the ground by pumping the air from each Tedlar bag through a wet test meter. The CFC peak areas from each sample were used to determine sample volume based on the previous calibration of the Niwot Ridge standard. In this manner, three different volume measurements were evaluated as a function of altitude: the integrated sampling volume recorded by the pump, the measured bag volume, and the CFC determined sample volume. A fairly large discrepancy was found between the volumes registered by the pump and the measured bag volumes. Figure 1 shows the measured

TABLE 2. Relative Standard Deviation (in %, n ) 10) of the Peak Area Ratio of VOCs (FID Signal) to CFCs (ECD Signal) in an Ambient Air Sample by Solid Adsorbent Sampling (AirToxics Cartridges) and Thermal Desorption GC/FID/ECD Analysis VOCs CFCs

propane

butane

pentane

hexane

benzene

toluene

xylenea

decane

undecane

CFC-11 CFC-12 CFC-113 CH3CCl3 CCl4 Cl2CdCCl2

18.4 2.9 4.5 3.7 3.5 4.1

16.9 2.1 2.1 1.7 2.2 2.4

18.2 2.2 3.1 2.6 3.5 3.9

18.0 0.6 2.3 1.8 2.4 2.3

18.2 1.3 2.1 1.5 2.4 2.8

17.3 1.2 1.5 1.2 1.7 1.5

17.3 1.8 1.4 1.3 1.6 2.0

17.2 1.5 2.0 1.9 2.3 1.4

14.4 6.6 6.1 6.4 5.5 5.3

a

Coeluting m/p-xylenes.

FIGURE 1. Measured bag volumes (corrected to STP conditions) shown as a function of altitude for the different pumps. Pumps are identified in the legend by their serial number. Prior to taking the balloon sample, all pumps were flow-calibrated at the ground with a reference (“dummy”) cartridge to flow rates of 300 mL min-1. The inferred theoretical sample volume is indicated in the figure as a dashed line. bag volumes as a function of altitude for each portable pump. This figure demonstrates the magnitude of sampling volume errors that can occur as the pump experiences the adverse conditions during a balloon sampling protocol, such as the changes in temperature, pressure, and the motion of the balloon and tether line. Sample volume deviations from the set point reach more than 25% in some instances. Extensive studies of the effects of the humid sampling conditions (80-100% RH) on quantification of CFCs and VOCs were published recently (27). Based on the results of that study, slight losses of CFC-113 and methylchloroform were expected to occur under the relatively humid conditions encountered at the Michigan site. In contrast, carbon tetrachloride did not exhibit sampling losses in laboratory studies under comparable humid sampling conditions.

Results In evaluating the best achievable precision with this relative calibration method, a 100 L Teflon bag was filled with ambient air from the CU campus using a Teflon diaphragm pump (KNF Neuberger, Trenton, NJ). The bag was well mixed, and a series of 10 adsorbent tubes was loaded with 5 L sample aliquots using the flow and temperature-controlled 10-port cartridge sampler. Table 2 summarizes the relative standard deviations of the mean peak area ratios of VOC/CFC pairs. These data demonstrate that relative standard deviations for VOC/CFC ratios on the order of 1.5-3% can be achieved in many cases using solid adsorbent sampling, thermal desorption, and this dual-detection method under controlled,

laboratory-sampling conditions. The reproducibility of the column flow split to the detectors is one factor contributing to the overall relative error. Assuming that this error is independent of retention time, it can be deduced that the reproducibility of the split ratio is on the order of equal or better than 1.2% relative standard deviation. In the analysis of field samples, the CFC peak area from each sample was used to calculate the sample volume based on the Niwot Ridge standard calibration, with the corrections that were discussed in the Experimental Section above. The ability of CFC measurements to accurately predict sample volume was evaluated by comparing volumes inferred from the CFC peak areas with volumes measured by physical means; i.e., integrating the mass flow controller readings (ground and aircraft samples) or by collecting the vent air in the Tedlar bags (balloon samples). Figure 2a shows the correlation between volumes based on CFC-113 peak areas and MFC-determined sample volumes for ground samples collected at Niwot Ridge on June 18, 2001, vertical profiles collected from the airplane between November 16, 1999 and April 8, 2000, and vertical profiles collected from the balloon on August 1, 2000. A linear fit regression line was calculated through all data using the sampler volumes as independent variables and the CFCdetermined volumes as dependent variables. The linear regression for all data provides a slope of 0.985. The level of precision is demonstrated by the correlation coefficient, r 2 ) 0.977, and is best for sampling volumes below 6 STP L. Most of the data obtained at higher flow rates are based on balloon profiles where collection in Tedlar bags was used as the means of determining the physical volumes, since the pump flow measurements were highly scattered (Figure 1). The volume determination using the Tedlar bags is estimated to have a ∼5% precision and accuracy, hence much of the variance in Figure 2a may be due to the imprecision in the Tedlar bag method. Figure 2b shows the correlation between methylchloroform-determined volumes and measured sample volumes for the same sample set. Although the regression line for the composite data has a slope (1.02) very close to unity, the correlation coefficient is poor, with r 2 ) 0.836. This poor precision largely results from lack of agreement among data obtained from the three different experiments. The groundbased samples overpredict volumes, especially at high flow rates, while the balloon-based samples underpredict volumes. As with CFC-113 (Figure 2a), the Tedlar bag balloon data show greater scatter than for the other two sampling platforms where mass flow controllers were used. The underprediction of methylchloroform internal standard volumes in the balloon samples is likely a result of some breakthrough of this compound under the humid sampling conditions encountered at this site (see above). In Figure 2c, the same analysis is used for volumes based on CCl4 measurements. Linear regression on the composite data provides a slope close to unity (0.991). The correlation coefficient, r 2 ) 0.965, also is quite good, especially conVOL. 37, NO. 5, 2003 / ENVIRONMENTAL SCIENCE & TECHNOLOGY

9

1005

TABLE 3. Data Analysis of the Sample Volume Ratio (CFC Determined Volumes: Measured Volumes) for Subsets of Ambient Air Samples Collected in the Free Troposphere, Boundary Layer, and Ground Level Aira sampler: n: CCl4

CFC-113

CH3CCl3

average std dev %RSD average std dev %RSD average std dev %RSD

10-Port 27 BL & FT

Balloon 13 BL

10-Port 26 GL

66 total

1.00 0.12 11.8 0.90 0.04 4.2 0.71 0.41 5.8

0.97 0.08 7.8 0.96 0.07 7.5 0.91 0.09 9.4

0.88 0.09 10.7 0.92 0.07 8.0 1.13 0.17 15.4

0.95 0.12 12.2 0.92 0.06 6.9 0.91 0.22 24.4

a Ratios of “1” are expected for perfect agreement. Precision and accuracy were evaluated for each group as well as the combined results. Data shown are the mean ratios, standard deviation, and relative standard deviation (RSD).

These data show that CCl4 overall gives the best agreement between the two methods in all sample sets. However, lower overall standard deviation was found for CFC-113. It is somewhat difficult to deconvolute and assess these errors because the different contributing factors to these errors including (a) procedure of reference volume measurement, (b) CFC atmospheric variability, (c) chemical analysis procedure, and (d) reference standard and used correction. Conclusively, it can be seen that overall accuracy and precision on the order of ∼