5.1.2. Flask Samples

Overview

The flask sampling program has undergone a few changes in 1996-1997 designed mainly to facilitate processing of samples and data. Over the past decade the number of gases analyzed by this group has risen from 3 to over 25 and the number of weekly sampling sites has more than doubled, from 5 to 11 (Figure 5.1, Table 5.1). Additional flasks are collected as part of special projects, including oceanic expeditions, polar firn sampling, and aircraft missions. Flasks that once were analyzed by one instrument are now analyzed by four instruments, two of which are mass spectrometers (Table 5.2). This increase in the number of flasks sampled and gases measured has resulted in a considerable increase in the laboratory workload. Attempts to offset this have focused on automation of instruments, streamlining of data processing, and improvement of quality control. In 1991 the first of the automated instruments, a three-channel gas chromatograph (GC) electron capture detector (ECD) system capable of measuring seven gases, was put in service. This system replaced the old manual, one-channel GC/ECD that measured only three gases [Thompson et al., 1985]. After a few years of debugging and inter-comparison, the old, manual GC/ECD was retired in 1996. Software for running the automated GC/ECD system was upgraded in 1995. In the past 2 years quality control programs were written and a database developed that now allows evaluation of data from samples immediately following analysis.

TABLE 5.2. Instrumentation for NOAH Flask Analysis

Other efforts at automation included the GC mass spectrometer (MS) mainly responsible for measurement of HCFC’s and methyl halides, and an additional GC electron capture detector (ECD) system, Low Electron Attachment Potential Species (LEAPS), mainly used for measurement of halons and methyl bromide. Programs also have been written to speed up processing of data from these instruments. Peak integration is now fully automated on both GC/MS instruments, although manual integration of peaks is still necessary on the LEAPS GC/ECD system.

Thirty-two new, 3-L sampling flasks from Meriter Corporation (San Jose, California) were purchased for inclusion in our sampling network. These flasks are constructed from no. 316 stainless steel, are highly electropolished, and have minimal internal weld exposure. Stability is excellent for most gases in these flasks, although CCl₄ and, to a lesser extent, CH₃CCl₃ can degrade over the long term as they do in all stainless steel flasks filled with dry air. Recent work with glass flasks in sampling firn air has shown that glass flasks identical to those used in the carbon cycle network can be used for sampling halocarbons over the short term. Further tests of flasks already in the network showed that many of them, depending upon their history, can be contaminated. We continue to purchase and test these in small lots as a possible option for use in special projects and for measuring gases that are less stable in selected stainless steel flasks.

A number of improvements at our flask sampling sites during 1996 and 1997 were made. At SMO the sample inlet was moved from the stack to the pump board Air-Cadet inlet system of the Radiatively Important Trace Species (RITS) GC to avoid cross contamination from other observatory instruments with inlets attached to the stack. Samples are now collected from the continuous flow, pressurized inlet at all sites except CGO, which will be changed over in 1998. Because of concerns about the durability of the Air-Cadet pumps for use with the RITS inlet system and flask sampling, existing pumps were replaced with KNF Neuberger N-05 pumps during visits to the South Pole Observatory, Antarctica (SPO), Mauna Loa Observatory, Hawaii (MLO), Cape Kumukahi, Hawaii (KUM), and SMO. Pumps at the remaining sites will be upgraded in 1998. At KUM before November 1997, a pump system was connected upstream of the continuously flowing, Air Cadet pump to facilitate flask sampling. In November 1997 the Air cadet pump was replaced by a continuously flowing KNF Neuberger pump. Flasks can now be sampled from a simple manifold. BRW and CGO will be visited in 1998 to receive the same improvements. At NWR, AC power was installed at T-van. The battery-powered pump system will be replaced by a new system that allows flasks to be filled to higher pressures.

In addition to improvements at these sites, sampling with glass flasks was initiated at Palmer Station, Antarctica (PSA) at the end of 1997 in an attempt to understand the seasonal cycling of some of the more reactive halogens at a southern hemispheric coastal site. The objective is to compare results from this site with those from BRW where seasonal cycles are pronounced and perhaps influenced by the springtime breakup of ice. With assistance from CMDL’s Carbon Cycle Group (CCG), samples are collected in glass flasks two times per month. The glass flasks are filled to about 1.3 kPa with the CCG MAKS sampling apparatus by personnel trained by CCG. Although contamination is observed for some halocarbons, pump tests and preliminary results suggest good data can be obtained for many other compounds with this technique.

CFCs and Chlorocarbons

Measurements from the automated flask GC (OTTO) show that atmospheric mixing ratios of CFC-11 and CFC-113 continued to decline through 1996 and 1997 at rates similar to those previously reported for earlier years [Montzka et al., 1996] , while the growth rate of CFC-12, although still positive, decreased from 5.9 to around 4 ppt yr^-1 (see cover figure). As a result of declining concentrations and growth rates for CFCs, methyl chloroform, and CCl₄, the amount of chlorine, equivalent chlorine (chlorine + bromine weighted by an efficiency factor), and effective equivalent chlorine (equivalent chlorine weighted by destruction rates in the midlatitude stratosphere) contained within long-lived, halogenated gases (CFCs, HCFCs, CH₃CCl₃, CCl₄, and halons), peaked in 1992-1994 and declined through 1995 [Montzka et al., 1996]. Cunnold et al. [1997] found similar results for chlorine-containing compounds measured by the Advanced Global Atmospheric Gases Experiment (AGAGE). The measurements suggest that declines in these quantities continued during 1996-1997 at rates similar to those observed in mid-1995 (Figure 5.2). Amounts observed at the end of 1997 represent a decrease of 2-4% from the peak Cl, ECl, and EECl delivered to the atmosphere from these gases in earlier years. Relative declines in the total atmospheric burden of Cl, ECl, and EECl are smaller because other gases (e.g., CH₃Cl and CH₃Br) also contribute significantly to the atmospheric burden of these quantities.

Fig. 5.2. Trends in chlorine, equivalent chlorine (chlorine + bromine multiplied by an efficiency factor of 50), and effective equivalent chlorine (equivalent chlorine where compound-specific halogen release rates are considered) from CFCs, HCFCs, halons, CCl₄, and CH₃CCl₃. Symbols refer to the northern hemisphere (filled diamonds), southern hemisphere (filled squares), and global tropospheric mean (crosses).

The gas contributing the most to this decline is CH₃CCl₃, which has an atmospheric lifetime of less than 5 years. As this gas is removed from the atmosphere, the overall rate of the decline in total chlorine will become slower. However, distributions and trends of this gas allow understanding of other atmospheric processes. For example, the difference in the atmospheric mixing ratio of CH₃CCl₃ between hemispheres has become dramatically smaller since 1992 as emissions have declined. The global latitudinal distribution of CH₃CCl₃ in 1992 and earlier years reflected the distribution of sources; mixing ratios in the northern hemisphere were higher than in the southern hemisphere because this solvent was emitted pre-dominantly in the northern hemisphere. As emissions become insignificant, the distribution of CH₃CCl₃ will instead reflect the latitudinal distribution of sinks for this compound, which is dominated by the reaction of CH₃CCl₃ with the hydroxyl radical (Figure 5.3). Since the summer of 1996, mixing ratios at SMO have been lower than at CGO, likely as a result of the greater abundance of OH in the tropics. Continued monitoring of methyl chloroform as emissions diminish further should allow for refined estimates of the global lifetime of CH₃CCl₃ and, therefore, of other trace gases that react with OH. It also will be useful in estimating the relative mean OH abundance in the northern and southern hemispheres.

Fig. 5.3. The annual mean abundance of methylchloroform relative to the abundance at Cape Grim, Tasmania, for that year. Similar years are connected with lines to guide the eye. Mixing ratios are determined from GC/MS analysis of paired flask samples.

The global mixing ratio of CCl₄ is more difficult to determine from flask samples because CCl₄ (and to a much lesser extent CH₃CCl₃ and CH₃Br) can be degraded in dry air samples stored for extended periods in stainless steel flasks. This is indicated by anomalously low mixing ratios and poor flask pair agreement in many of the samples collected at SPO and less frequently from wintertime samples collected at ALT, BRW, MLO, and NWR. Nevertheless, measurements from reliable flask samples (those for which pair agreement is within acceptable limits) support the trends and abundance determined by the RITS program, showing a continued decrease of 0.8 ppt yr^-1 for atmospheric CCl₄ through 1996 and 1997.

N₂O

The growth rate of atmospheric N₂O over the past two decades has ranged from under 0.5 to over 1.0 ppt yr^-1. The mean, globally-averaged growth rate of this gas from flask measurements during this time was 0.75 ± 0.03 (95% confidence limits (C.L.)) ppt yr^-1, which amounts to about 0.25% yr^-1. These data and growth rates are from 20 years of flask analyses and are corroborated by measurements with RITS in situ instrumentation over the past 11 years (section 5.1.3). The factors that cause this increase and determine the isotopic composition of atmospheric N₂O are currently unexplained [Bouwman et al., 1995; Cicerone, 1989; Kim and Craig, 1993] . However, our measurements of firn air [Battle et al., 1996] and other ice-core records [Khalil and Rasmussen, 1989; Leuenberger and Siegenthaler, 1992; Machida et al., 1995] show clearly that N₂O has been increasing in the atmosphere for at least the past 100 years.

SF₆

SF₆ is a trace gas only recently introduced into the atmosphere. It has a lifetime of ~3200 years [Ravishankara et al., 1993] and a greenhouse warming potential (GWP) of 15,000-35,000 [Schimel et al., 1996] , making it an extraordinarily strong greenhouse gas on a per molecule basis. SF₆ is used mainly as an insulator in electrical transformers and circuit breakers. Once leaked into the atmosphere, it will persist for millennia. Al-though present at low ppt levels in today’s atmosphere and currently of little global consequence, SF₆ has been increasing in abundance since the early 1970s [Geller et al., 1997; Maiss and Levin, 1994] . SF₆ data, which include archived air samples and recent samples from the flask network, show that the growth rate has not changed much over the past decade (Figure 5.4). The growth rate of 0.20 ± 0.03 (95% C.L.) ppt yr^-1 from the flask network samples, which run from 1995-1998, does not differ significantly from the 0.22 ± 0.03 (95% C.L.) ppt yr^-1 determined from archived NWR samples which go back to 1987. Northern hemisphere (NH), southern hemisphere (SH), and global growth rates for this gas are identical, which is to be expected for a gas with constant source strength and long lifetime.

Fig. 5.4. CMDL measurements of atmospheric SF₆. (a) Global and hemispheric averages of the dry mole fraction of SF₆ in flask samples dating back to 1995. (b) Annual growth rates calculated from year-to-year differences in the global mean N₂O. The dark line is a loess smooth of the data.

Halons

Measurements show that the atmospheric burden of halons H-1301 and H-1211 has doubled and that of H-2402 has increased by over half during the past decade (Figure 5.5) [Butler et al., 1998] . Halon mixing ratios continued to increase in recent years despite an international ban on their production and sales in developed nations effective January 1, 1994. The growth rate of H-1301 appears to have slowed recently, but it remains significant (Table 5.3, Figure 5.5a) and, within stated uncertainties, the 1997 growth rate does not differ from that reported for the end of 1996 [Butler et al., 1998] . Atmospheric H-1211 is increasing at a much higher rate than H-1301 and has not shown much sign of slowing over the past decade (Table 5.3, Figure 5.5b). The 1997 tropospheric growth is virtually identical to growth over the past decade. Although the growth rate of H-2402 is substantially slower than that of the other two halons, H-2402 contains two bromine atoms per molecule. Thus the increase of Br due to growth of H-2402 in the atmosphere in 1996 is almost half that of H-1301 and about one-tenth that of H-1211.

TABLE 5.3. Atmospheric Halons

Tropospheric mole fractions for H-1301 and H-1211 are for the end of 1997. The tropospheric mole fraction for H-2402 is for the end of 1996. Global growth rates for H-1301 and H-1211 are given as the observed change in the latitudinally weighted, global, mean mixing ratios for 1995-1997 for H-1211 and for 1995-1996 for H-2402.

Fig. 5.5. Hemispheric and global bimonthly averages of tropospheric mole fractions of (a) H-1301, (b) H-1211, and (c) H-2402. Data are taken from the CMDL flask network (squares), research cruises (triangles), and cylinders of archived air (diamonds). Northern hemispheric results are shown as shaded symbols, southern hemispheric results as open symbols, and global means as solid symbols. Bimonthly, hemispheric averages are calculated by weighting measurements by the cosine of the sampling latitude within each hemisphere. Global averages are computed as means of the hemispheric averages.

Few measurements of halons that allow for accurate comparisons to the results presented here have been reported over the past decade. Usually such reports are associated with field missions that are limited in geographic distribution, period of sampling, or both. Some are part of stratospheric investigations, so contribute only a few values for the troposphere. Even if these differences in sampling are taken into account, it is still clear that measurements from these studies in the past have not agreed well (Figure 5.6). Such widespread disagreement among laboratories underscores the need for extensive intercalibration among investigators making these measurements. Small offsets in calibration can lead to large errors in estimates of potential ozone depletion because of the possibility of the halons offsetting gains in stratospheric ozone protection resulting from reductions in chlorocarbon emissions [e.g., Montzka et al., 1996]. Small errors in estimating the atmospheric burden of halons can lead to significant errors in estimates of the atmospheric burden or trend of equivalent chlorine in the atmosphere [Daniel et al. 1996].

Fig. 5.6. Measurements of tropospheric halons over the past decade. Solid lines are CMDL global averages for (a) H-1211 and (b) H-1301. Symbols signify measurements by other laboratories and research groups. Filled squares represent work by Khalil and Rasmussen [Ehhalt et al. 1988, Khalil and Rasmussen, 1992], open triangles are from a study by Singh et al. [1988], open squares represent measurements by Oertel [1992], filled circles are measurements by C.J.-L. Wang, D.R. Blake, N. Blake, and F.S. Rowland, as given in Lorenzen-Schmidt [1994], filled diamonds are values from Lorenzen-Schmidt [1994], open circles are from Schauffler et al. [1993], and filled triangles are from Fabian et al. [1994].

Evaluations of the growth rates and the amounts of “banked” halon available for use suggest that H-1301 emissions could continue at the present rate for another 40 years before depleting the bank of H-1301 [Butler et al., 1998] . This would leave an atmospheric mixing ratio of 3.6 ppt, or 57% higher than observed today. Under the same scenario, reserves of H-1211 would be depleted in 8 to 12 years leaving an atmospheric mixing ratio of 4.6-5.0 ppt, or 31-43% higher than observed today. However, there is a significant discrepancy between H-1211 emissions calculated from production and use and emissions deduced from atmospheric measurements. The discrepancy can be reconciled by lowering estimated emissions by ~25%, reducing the atmospheric lifetime of H-1211 from the 20 years given in Kaye et al., [1994] to 11 years or some combination of the two [Butler et al., 1998] . These uncertainties cause considerable doubt in modeled predictions of the future burden and fate of H-1211 in the atmosphere. It is not certain whether the continued rapid increase in H-1211 results from a depletion of known reserves, inordinately high fugitive emissions during its production in third world countries, or unreported production of halon.

Chlorofluorocarbon Alternatives Measurement Program (CAMP)

Measurements of chlorofluorocarbon alternatives continued on two instruments during 1996 and 1997. On average, one to three flask pairs per month from eight remote and three regional sampling locations were analyzed on the older GC-MS instrument. In addition, about one sample flask pair per month was analyzed from the remote locations on the instrument dedicated to making measurements of HFC-134a. The main changes made in this program involved automated data manipulation on both instruments as of January 1997 and automated analysis of larger flasks on the older GC-MS instrument as of September 1997.

Automated data manipulation allows results to be calculated and compiled more efficiently. This is achieved through the use of macros that determine chromatographic peak areas and calculate and compile results with a commercially-available spreadsheet software package. Automation allows for unattended analysis of up to eight flasks or four high-pressure cylinders. Flasks and secondary air standards are connected via a 16-port stream selection valve to the instrument inlet. Flows from flasks are regulated with different lengths of small diameter stainless steel tubing that are matched to the initial flask pressures. With automated analysis, agreement between replicate injections and between simultaneously filled flasks is similar to or improved over manual analysis. Results from a subset of flasks analyzed by both methods agreed for nearly all compounds. Some small offsets were observed for HCFC-142b and HCFC-141b and appear to result from problems associated with manual analysis of flasks. Protocols for routine checking of sample integrity as it passes through the multiple sampling ports are being implemented.

Mixing ratios of the most abundant HCFCs (HCFC-22, -141b, and -142b) continue to increase throughout the troposphere (Figure 5.7, Table 5.4). In mid-1997 these three gases accounted for about 5% or 150 ppt of the atmospheric burden of chlorine contained within long-lived, anthropogenic halocarbons. This amount was increasing by about 10 ppt per year in 1997 or similar to that reported for earlier years [Elkins et al., 1996a; Montzka et al., 1996].

Fig. 5.7. Atmospheric dry mole fractions (ppt) of the most abundant HCFCs and HFC-134a. Each point represents the mean of two simultaneously filled flasks from one of eight stations: ALT, filled circle; BRW, filled triangle; NWR, filled diamond; KUM, crosses; MLO, open square; SMO, open triangle; CGO, filled diamond; SPO, filled circle. Also plotted are results from analysis of archived air samples (open circles) filled at NWR and in a past cruise from both hemispheres (mid-1987).

TABLE 5.4. Global Mid-Year Burden and Rate of Change for HCFCs, and HFC-134a

Quantities estimated from latitudinally weighted measurements at seven remote sampling locations.

*Growth rate estimated from 1992-1997.

Continued increases were also observed for HFC-134a, a gas for which restrictions on future use are being considered as part of the Kyoto Protocol (Figure 5.7, Table 5.4). Global mixing ratios of this CFC replacement are currently below 10 ppt. Because of laboratory air contamination and other issues, the number of good measurements made in 1997 was limited, but improvements are being implemented to avoid these problems in the future.

Short-Lived Gases

Measurements of CH₂Cl₂, CHCl₃, and C₂Cl₄ by GC-MS techniques continued from remote flask sampling locations during 1996-1997 (Figure 5.8) and as part of CAMP. Beginning in 1995 the use of a new type of flask built at Max Planck Institute for Chemistry (MPI), Mainz, Germany, allowed for more reliable measurements of CH₂Cl₂ and CHCl₃, and reliable measurements of CH₃Cl and CH₃Br (Figure 5.8). These MPI flasks are larger (2.4 L versus 0.8 L), made out of a higher grade stainless steel, and do not contain any seals that require Teflon tape. Results from simultaneously filled pairs of these flasks generally agree to within the instrument measurement capabilities, suggesting that mixing ratios of gases contained within the flasks do not change during storage and transport. This was not true for some gases, particularly CH₃Cl and CH₃Br in the older, 0.8-L flasks.

Fig. 5.8. Atmospheric dry mole fractions (ppt) determined for selected chlorinated trace gases and CH₃Br. Symbols are identical to those described in Figure 5.7. Results shown for CH₃Cl, CH₃Br, and CHCl₃ are from 2.4-L flasks only. Results plotted for CH₂Cl₂ and C₂Cl₄ are from both 2.4-L and 0.8-L flask samples. All results are based on preliminary calibration scales.

[BACK] [CONTENTS] [NEXT]