Measurements of Atmospheric Dimethylsulfide, Hydrogen Sulfide, and Carbon Disulfide During GTE/CITE 3

Measurements of atmospheric dimethylsulfide (DMS), hydrogen sulfide (H2S), and carbon disulfide (CS2) were made over the North and South Atlantic Ocean as part of the Global Tropospheric Experiment/Chemical Instrumentation Test and Evaluation (GTE/CITE 3) project. DMS and CS2 samples were collected and analyzed using an automated gas chromatography/flame photometric detection system with a sampling frequency of 10 min. H2S samples were collected using silver nitrate impregnated filters and analyzed by fluorescence quenching. The DMS data from both hemispheres have a bimodal distribution. Over the North Atlantic this reflects the difference between marine and continental air masses. Over the South Atlantic it may reflect differences in the sea surface source of DMS, corresponding to different air mass source regions. The median boundary layer H2S and CS2 levels were significantly higher in the northern hemisphere than the southern hemisphere, reflecting the higher frequency of samples influenced by pollutant and/or coastal emissions. Composite vertical profiles of DMS and H2S are similar to each other, and are consistent with a sea surface source. Vertical profiles of CS2 have maxima in the free troposphere, implicating a continental source. The low levels of H2S and CS2 found in the southern hemisphere constrain the role of these compounds in global budgets to significantly less than previously estimated.


INTRODUCTION
The distribution and chemistry of reduced sulfur gases in the atmosphere are of current interest with regard to several important geochemical processes.Much of the acidity in rainfall can be attributed to oxidized sulfur compounds.Quantification of natural background precursors is therefore an essential step in assessing the magnitude of anthropogenic perturbations, and in predicting the effect of pollutant emission controls.It has also been suggested that the atmospheric cycling of biogenic sulfur gases may play a role in the maintenance of the global energy budget by providing the precursors for the formation of submicron sulfate aerosols [Shaw, 1983].Aerosols derived from a marine sulfur source may be the predominant cloud condensation nuclei over the remote oceans, controlling the albedo of marine clouds [Charlson et al., 1987].
The lifetime of biogenic sulfur gases such as dimethylsulfide (DMS), hydrogen sulfide (H2S), and carbon disulfide (CS2) is on the order of a day in the troposphere.Hence, even low concentrations may indicate a substantial flux.Most environments contain these gases at the low parts-pertrillion (pptv) level [Maroulis and Bandy, 1978;Slatt et al., 1978;Andreae and Raemdonck, 1983;Andreae et al., 1985;Ferek et al., 1986;Berresheim, 1987;Kim and Andreae, 1987;Saltzman and Cooper, 1988].Many of the problems in the atmospheric chemistry of biogenic sulfur gases are best studied in remote regions, where the tropospheric oxidant balance is undisturbed by anthropogenic activities.While primarily an intercomparison study, the Global Tropospheric Experiment/Chemical Instrumentation Test and Evaluation (GTE/CITE 3) expedition consisted of sampling periods over both the polluted North Atlantic Ocean and the 1Now at Plymouth Marine Laboratory, Plymouth, England.
Copyright 1993 by the American Geophysical Union.

Paper number 92JD00218.
0148-0227/93/92JD-00218505.00 relatively unpolluted tropical South Atlantic Ocean.This offered an excellent opportunity to study the contrast between these two environments.The comprehensive support measurements made during these flights facilitated analysis of the sources, sinks, and lifetimes of the various sulfur gases in the different regimes studied.
In this report, we describe an automated sampling and analysis system and present the first airborne DMS and CS2 measurements made using such a system.The instrument is discussed in detail here, as it has not previously been described.Some of the analytical principles applied may also be of interest in the analysis of a number of sulfur-containing and other organic compounds.Data are also presented for atmospheric H2S , simultaneously collected using a AgNO3 impregnated filter technique [Natusch et al., 1972;Cooper and Saltzman, 1987;Saltzman and Cooper, 1988].

Sampling
Measurements presented in this report were made from the NASA Wallops Flight Facility Lockheed Electra.Flights over the North Atlantic Ocean were staged from Wallops Island, Virginia; flights over the South Atlantic Ocean were staged from Natal, Brazil.
Samples were drawn into the analytical systems through Teflon PFA tubing (1/4-inch OD for DMS, 3/8-inch OD for H2S) extending through bulkhead unions in the top of the aircraft.A 30-cm rearward facing rigid support ensured that samples were collected outside of the aircraft boundary layer.DMS samples or standards were drawn into the system using a small diaphragm pump while the aircraft was stationary, or a venturi pump while airborne.H2S samples were collected only while airborne, using venturi pumps.The reaction of iodide (I-) with ozone in neutral aqueous solution has been used for more than a century for the quantitative determination of atmospheric oxidants [Brodie, 1872].We have employed this reaction for sulfur gas analysis on the presumption that the product, triiodide ion (I•-), would be inert toward DMS.This indeed appears to be the case, as shown by a series of previously reported standard addition experiments [Saltzman and Cooper, 1989;Cooper and Saltzman, 1991].

Dimethylsulfide
The oxidant scrubbers used in this study were 250-mL glass bubblers with glass frits 20 mm in diameter.These contained 60 mL of neutral potassium iodide solution (2% In normal use, the solution is routinely changed on a weekly basis, although no deterioration in performance has ever been noted.The theoretical capacity of this scrubber exceeds 1000 m 3 of air containing 100 ppbv of oxidants.By using an oxidant scrubber with such a large capacity, automated sampling and analysis of DMS and other reactive gases become possible.We have developed a system using Tenax GC (Alltech Associates, Inc., Chicago, Illinois) as a preconcentration medium for DMS, because of the relatively low temperatures required for desorption.The use of Tenax GC and other polymers for the determination of trace organics in air has previously been questioned due to the breakdown of the substrate and the release of various byproducts [Walling et al., 1986].In the absence of an oxidant scrubber, we have observed the conversion of DMS to dimethylsulfoxide and dimethylsulfone on Tenax columns, in addition to the appearance of elevated levels of organic components.The removal of oxidants from the sampled airstream was found to eliminate these interferences.
System design and operation.DMS and CS 2 are preconcentrated on Tenax GC cooled to approximately -20øC using a thermoelectric heat pump.These samples are desorbed thermally by reversing the voltage polarity to the thermoelectric units.Analysis is via GC using flame photometric detection (FPD).
A schematic diagram of the system is shown in Figure 1, and the timing cycle in Figure 2. The airstream is drawn first through the oxidant scrubber, then through a thermoelectrically cooled zone (TEC 1 or TEC 2), which contains a drier tube and Tenax trap.When sampling is complete, the sampling valve is switched, and the polarity of the 12 V dc supply to TEC 1 is reversed.This diverts the carrier gas through the Tenax trap and heats the block, desorbing the sample onto a cooled preconcentration column (TEC 3).The airstream continues to flow through the now heated drier tube to waste, thus removing moisture which accumulated during sampling.After a brief transfer period, the polarity of the thermoelectric module cooling the preconcentration column is reversed, heating the unit and initiating the chromatographic run.Simultaneously with the injection of the sample collected on one channel, collection of the next sample on the second Tenax trap is started.After sample desorption, TEC 1 or TEC 2 is again cooled to begin the next sampling period.The total cycle time is 20 min per channel.Thus, with two channels operating alternately, one measurement is made every 10 min.
The system was calibrated manually during this study by switching of the airstream to sulfur-free air and injecting liquid standards (5-200 /xL) onto a plug of silanized glass wool, Teflon wool, or directly into the KI bubblers.The preparation of liquid standards for CS2 and DMS has been described previously [Saltzman and Cooper, 1988;Cooper and Saltzman, 1991].A typical calibration curve covers the range of 2-80 pmol, which is equivalent to 4-180 pptv in a 10-L sample.
The thermoelectric units were clamped between an aluminum heat sink, mounted on the inside of an insulated cooler (0øC), and an aluminum block.For the two sampling channels, a 3/8-inch-thick block was drilled to permit the entry of the drier tube and the Tenax sampling trap.Connecticut).The thermoelectric units were operated at full load during cooling with no temperature control.
Airflow rates through the sampling channels were controlled using mass flow controllers (0-2 standard L/min, MKS Instruments, Andover, Massachusetts), and the air volumes were integrated using a voltage-to-frequency converter and counter.The system was typically operated at airflow rates of 0.5-1.0 standard L/rain with a sampling time of 10 min.
Chromatographic separation was achieved using a 1/8 x 8 inch Chromosil 330 column (Supelco Inc., State College, Pennsylvania) held isothermal at 50øC.The carrier gas for the chromatographic separation was N2 at 40 mL/min.These conditions provide baseline resolution between CS2 and DMS.It should be noted that CS2 emanates from many polymeric components in flow systems, and particular care is needed to ensure that the carrier gas stream is free from contamination.All tank gases used in the system were scrubbed with silica gel, molecular sieve, and charcoal prior to use.For the carrier gas and sulfur-free air (for calibration) an additional Pd-coated molecular sieve trap (Science Glass, Miami, Florida) was used immediately prior to the switching valve.
The sulfur gases were detected using an FPD (Tracor Instruments, Houston, Texas) and a combined high-voltage supply and electrometer (Pacific Instruments, Concord, California).In order to minimize the effect of changes in cabin pressure on the detector response, the flame chamber was maintained at a constant pressure of 15 psia (1.03 bar).The flame exhaust flowed through an absolute back pressure regulator (Moore Products, Allentown, Pennsylvania) to waste through a bulkhead in the airframe.This regulator was heated to 85øC to prevent condensation of water.The detection limit of the FPD is approximately 1 pmol of sulfur, which corresponds to 3 pptv of DMS or 1.5 pptv of CS2 in a 10-L sample.In practice, the system has a small CS2 background, for which the samples are corrected.Because of the nonlinear nature of the FPD, this has the effect of lowering the detection limit of CS2 to less than 0.5 pptv.
The data acquisition and control functions were carried out using a PC-based chromatography software and analog- The temperature dependence of adsorption, and hence breakthrough, can be described using the Arrhenius relationship [Namiesnik, 1988].Results of the breakthrough experiments using DMS are shown in Figure 4 in the form of an Arrhenius plot.The relatively large scatter in the plot results from the fact that breakthrough volumes in this experiment

Hydrogen Sulfide Analysis
Hydrogen sulfide samples were collected and analyzed as described previously [Cooper, 1986;Saltzman and Cooper, 1988]   South Atlantic tropical maritime air masses were sampled during flights 13-19.Flight paths over the South Atlantic were selected primarily for the purpose of diurnal studies, with three pairs of duplicated missions.This paper focuses mainly on the spatial distribution of DMS, H2S, and CS2 found during these flights, while the diurnal variation is discussed in more detail by [Saltzman et al., 1993].
The boundary layer measurements of all compounds are summarized in Table 1,  The South Atlantic data are mapped in Figure 9. Unlike the North Atlantic data, there are no clear geographic gradients in the DMS levels, even though significant variability is evident in the boundary layer data.An interesting feature evident in the maps of the South Atlantic DMS data is the occurrence of relatively high DMS levels in several of the samples taken at 5000 ft, which were considered to be free tropospheric air.This suggests that a portion of the sample was enriched in air from the boundary layer.Two explanations for such an enrichment are (1) that boundary layer air had been transported vertically through cloud processes, as proposed by Chatfield and Crutzen [1984] and noted previously by Ferek et al. [1986] or (2) that the height of the boundary layer depth was close to 5000 ft, so that boundary layer air was collected for at least part of the sample.The simultaneous dew point and ozone profiles also show abrupt changes at about 6000-ft altitude.
In general, there does not appear to be any significant  9a) shows that considerable variation was occurring between flights over identical paths.The ratio of nighttime maximum to daytime minimum found between flights 16 and 17, a factor of approximately 3, is considerably higher than found previously in remote marine air [Andreae et al., 1985;Berresheim, 1987;Saltzman and Cooper, 1988].A nighttime/daytime ratio of approximately 1.5 was found during flights 14 and 15, similar to the previous studies.Flights 18 and 19 were conducted sequentially on the same day, with sunrise occurring in the middle of flight 18.In this case the DMS declines steadily in a manner that is consistent with fairly rapid daytime photochemical removal.
While the large differences in DMS levels between flights 16 and 17 appear to be due to a slightly different air mass trajectory, the air mass trajectories and the tracers shown in Figure 8 suggest that the conditions during flights 18 and 19 were similar.The rapid decrease in DMS during this flight pair therefore suggests that either (1) oxidant levels may have been higher than previously encountered in remote marine air or (2) vertical mixing processes may have intensified during the flights.These differences have been discussed in detail by Saltzman et al. [1993].
Diurnal variation by a factor of approximately 2 in H2S levels is also consistent with the preceding discussion.The diurnal variation of CS2, though smaller, appears to support a change in source between flights 16 and 17, with the highest levels occurring during the daytime flights.This suggests that variation in the sources of these compounds produced the observed temporal variability in the first two flight pairs, and not oxidation processes.

Vertical Profiles of DMS, CS2, and H2S
The free tropospheric measurements of DMS, H2S, and CS2 are summarized in Table 2.It is clear that DMS and H2S concentrations were significantly lower in the free troposphere than the boundary layer (Table 1), which is consistent with a sea surface source for these compounds.In contrast, levels of CS2 in the free troposphere were not significantly different to those in the boundary layer.This suggests that horizontal advection may be an important source of CS2 to  nental origin of free tropospheric H2S in addition to an oceanic source.

Implications for the Global Sulfur Cycle
Although the highest DMS levels were found in the North Atlantic tropical maritime air, the ratio of DMS to H2 S and/or CS2 in this air mass was similar to the ratio evident in both the South Atlantic marine air and the previous study of Saltzman and Cooper [1988].Table 3 shows  respectively) are similar to those reported by Saltzman and Cooper [1988] and Kim and Andreae [1987], respectively.The average levels measured in the southern hemisphere (median 2.5 and 0.7 pptv) are significantly lower than in the previous studies.The H2S levels are approximately a factor of 3 lower than found in the North Atlantic tropical maritime air masses, and the CS2 levels almost an order of magnitude smaller.These low levels further constrain the importance of biogenic emissions of these compounds in global budgets.
The evidence for advection of air from the continents to the remote marine free troposphere suggests that oceanic sources of these compounds may also be smaller than previously assumed from atmospheric concentrations.

Fig. 1 .
Fig. 1.Schematic of the automated DMS/CS2 analytical system.Reversing temperature zones are indicated by the shaded boxes.Gas flow is indicated by solid lines, and electrical connections are indicated by dashed lines.
wt/vol KI, 0.05 M KH2PO 4, and 0.05 M Na2HPO4).Samples were drawn into the scrubbers through 1/4-inch OD Teflon PFA tubing, connected by Teflon TFE fittings (Cole Parmer Instrument Co., Chicago, Illinois).The scrubbers were mounted inside an insulated cooler in an ice/water mixture, lowering the water vapor content of the sample stream.

Fig. 2 .
Fig. 2. Schematic of the sampling cycle used during CITE 3. Shaded areas correspond to active periods; blank areas correspond to inactive periods.

Fig. 4 .Fig. 5 .
Fig. 4. Results of experiments to determine the breakthrough volume of DMS through Tenax GC as a function of temperature.
. Briefly, air was drawn through silver nitrate impregnated filters (47 mm, Whatman 41) [after Natusch et al., 1972].Flow rates up to 16 L/min were used for sampling periods up to 110 min.Under these conditions the sensitivity of the method is less than 1 pptv.Flow rates were controlled manually using Teflon PFA needle valves (Cole Parmer, Chicago, Illinois), and monitored using mass flowmeters (MKS Instruments, Inc).Sample volumes were obtained from totalizing the output signal of the mass flowmeters.Samples were mostly collected as duplicate pairs of two filters in series.The sulfide signal on the back-up filter was used to correct for the interference of OCS on the front filter, in accordance with Cooper and Saltzman [1987].The value of this correction varied up to approximately 0.1 nmol per sample, equivalent to approximately 5 pptv sulfide in a typical sample.The detection limit of the method and its precision at low concentrations is determined largely from the variability of the filter blanks.With only one exception (flight 9), the blank variability within a given batch of filters was less than 20 pmol, equivalent to approximately 1 pptv on a single filter.This yields a precision of approximately 2 pptv in a 50-min sample collected at 10 L/min.These conditions represent the majority of the airborne intercomparison periods.A more practical estimate of the precision may be the relative standard deviation from the mean of duplicate samples, which averaged 24% in the 45 valid sample pairs.

Fig. 6a .
Fig. 6a.Maps of DMS data over the North Atlantic Ocean.The east coast of the United States is shown from Georgia to Massachusetts.The upper plot shows the free tropospheric data; the lower plot shows the boundary layer data.Note the different scales on the vertical axes.

Fig. 6b .
Fig. 6b.Maps of CS 2 data over the North Atlantic Ocean.The east coast of the United States is shown from Georgia to Massachusetts.The upper plot shows the free tropospheric data; the lower plot shows the boundary layer data.

Fig. 8 .
Fig. 7. Maps of H2S data over the North Atlantic Ocean.The upper plot shows the free tropospheric data; the lower plot shows the boundary layer data.The east coast of the United States is shown from South Carolina to New Jersey.Boxes are plotted to show the geographic range of each sample.Note the different scales on the vertical axes.

Fig. 9a .
Fig. 9a.Maps of DMS data over the South Atlantic Ocean.The upper plot shows the free tropospheric data; the lower plot shows the boundary layer data.Note the different scales on the vertical axes.

Fig. 9b .
Fig. 9b.Maps of CS 2 data over the South Atlantic Ocean.The upper plot shows the free tropospheric data; the lower plot shows the boundary layer data.

Fig. 10 .Fig. 11 .
Fig. 10.Vertical profiles of DMS measured over (a) the North Atlantic and (b) the South Atlantic.All data from constant altitude legs are included.The symbols represent the means of the data at various altitudes, with the standard deviation shown as horizontal bars.
Fig. 12. Vertical profiles of H2S measured over (a) the North Atlantic and (b) the South Atlantic.All data from constant altitude legs in all flights are included.