GEOPHYSICALRESEARCH LETTERS, VOL. 14, NO. 3, PAGES206-209, MARCH1987

UPTAKE OF CARBONYL BY SILVER NITRATE IMPREGNATED FILTERS: IMPLICATIONS FOR THE MEASUREMENTOF LOW LEVEL ATMOSPHERIC H2S

David J. Cooper and Eric S. Saltzman

Rosenstiel School of Marine and Atmospheric Science, University of Miami

Abstract. Measurements of atmospheric exposure to elevated SO2 and 03, but concluded sulfide are often made by first collecting H2S on that these losses would not be significant at a paper filter impregnated with silver nitrate. ambient oxidant concentrations. No interference Using this technique on a recent shipboard study, has been found from the reduced compounds we observed anomalous breakthrough characteristics dimethyl sulfide, thiol, and of sulfide, which caused us to suspect the (Jaeschke et al., 1980). To our presence of an artifact. We have investigated the knowledge, no one has tested œor interference from interaction of various sulfur species with these carbonyl sulfide, despite the fact that OCS is the filters as precursors for the artifact sulfide. most abundant sulfur in marine air, with Laboratory experiments indicate that OCS generates concentrations in the marine boundary layer ot 500

sulfide on the •ilters with an efficiency of 1-2% +_ 50 pptv (Khalil and Rasmussen, 1984). under conditions comparable to those used for In this paper we present our anomalous field field sampling in the marine boundary layer. The data and the results of laboratory ext•riments magnitude of the artifact increases with designed to study the interaction of OCS with increasing temperature and is insensitive to silver nitrate impregnated filters. The humidity and insolation. It appears likely that implications of these results for the analysis of OCS is responsible for our shipboard results. trace concentrations of atmospheric H2S are Since much of the data on H2S concentrations in discussed. marine air have been omtained with this technique, they are most likely in error, producing Methods overestimates of background H2S concentrations. Shipboard and laboratory measurements of H2S Introduction were made by a method essentially identical to that of Natusch et al. (1972). An air stream was A balance of the global requires drawn through a silver nitrate impregnated filter knowledge of the contribution of both natural and (Whatman 41, 47 mmdiameter), held in a Teflon PFA anthropogenic fluxes of sulfur compounds to the filter holder. The H2S is fixed on the filter as atmosphere. is a major component , which was extracted from the of biogenic sulœur emissions •rom both coastal and filter using 20 ml of 0.1 M / terrestrial environments, primarily as a result of sodium solution. The sulfide was microbial suifate reduction in anoxic envir. on- determined by the fluorescence quenching of dilute ments. Measurements of H2S in marine air (Slatt fluorescein mercuric acetate (FMA), using 100 ul et ai., 1978; Delmas and Servant, 1982; Herrmann of 20 uM •iA in 0.1 M NaOH in each analysis. A and Jaeschke, 1984) yielded concentrations which Turner Designs model 10-853 fluorometer was used suggest that there are substantial sources of H2S with a blue lamp and 10 nm bandpass filters at 500 in remote oceanic regions, and that these sources nm excitation and 520 nm emission wavelengths. may be important in the sulfur budget oœ the Routine sampling of background marine air at marine boundarylayer. our maximumflow rate of 8 liter min-! required a All the reported measurements of low level sampling time of approximately one hour to give a concentrations of atmospheric hydrogen sulfide detection limit of 3 pptv. For the laboratory have relied on variations of the method of Natusch work, we increased the sampling times to give et al. (1972), which owes its popularity to its better precision at low sulfide concentrations. A high sensitivity and to the purported absence of detection limit of lower than one pptv was any known atmospheric interferents. Results of our achieved by sampling for several hours. field work, however, led us to believe that a low In the laboratory experiments, screw cap test level positive interference was present. tubes (45 ml) sealed with Mininert Teflon valves Axelro• •t al. (1969) had previously• reporte• were filled with an atmosphere of , cont- no interferenceby2 C1-, SO•2-, PO•3-, NO3-, aining approximately 1 ppm OCS from a permeation C2M302-, CHO2-, CO3'- , , ethano.1, tube. For these studies, measurements of carbonyl , , mercaptoethanoi, NH•OH,Mn 2+, sulfide were made gas chromatographically using a IN1.•.2+ , t•O• 2+ , L,U• 2+ , L,(/....2+ , •'•,-2+ D , ]3e• 2+ , •eo 3+ , K+ , a , 5' x 1/8" Teflon Tenax GC column (Alltech Assoc- slight fluoresenceenhancement by I- andslight iates, Deerfield,IL) heldisothermal at 40øC. N2 quenchingby Br-, SO•2- NO2- cysteine and wasused as the carrier gas at a flow rate of 20 cystine at elevated concentrations. Natuschet cmB min -1 . Aliquotsof gas wereinjected onto the al. (1972) •ound no loss of H.S from •ilters column through a septum. A sulfur specific flame exposedto elevatedNO2, and s•ight loss from photometricdetector was used, which has a detection limit of approximately 2 picomoles S. Copyright 1987 by the American Geophysical Union. Calibration was performed by injecting microliter aliquots of a standard gas mixture (Scott Paper number 6L6463. Specialty , Plumsteadville, PA) and using a 0094-8276/87/006L-6463503.00 permeation tube/mass tlow controller system.

206 Cooper and Saltzman' Uptake of Carbonyl Sulfide P0?

25 is higher for samples collected during the daytime, compared to nighttime samples. '• 20- A possible explanation for these results is the uptake of carbonyl sulfide (OCS) by the filters. OCS is known to interact with metal surfaces -- 15- (Ammons, 1980; Graedei et al., 1981) and to o ._ undergo hydrolysis in aqueous solution (Johnson, • 10- 1981; and references therein). In both cases ß sulfide is the most likely product. Consequently, ß ß we carried out a series of experiments to o 5- ß ß ß ß ß ß ß determine whether OCS could be responsible for the ß o artifact observed in our field studies.

0 .'; , -. , , ß , ' , ß e, ee_ee-, O0 04 J8 1• • 20 O0 Laboratory Studies Time[hour] Thefirst experimentwas designed to determine Fig. 1. Diurnal variation of the apparant H2b whether OCS has an influence on the collected measured on back-up (second stage) silver nitrate sulfide and the apparent breakthrough. Two filters in remote marine air. identical samples were collected in parallel, each with five filter stages, at the same flow rates used for the field studies (7.8 liter min-1). An Study Sites OCS permeation tube was connected in line with one set of filters, adding approximately 1.3 ppbv to The field data discussed in this study were the ambient OCS. While we did not determine the obtained on cruise CI8601 of the R/V Columbus ambient OCS, previous studies have shown it to be

Iselin from Miami to the Cariaco Trench, off the ubiquitous in the atmosphere, at 500 +-- 50 pptv coast of Venezuela, during February and March, (Khalil and Rasmussen, 1984). Figure 2a shows the 198b. The H2S data discussed in this paper were results of this experiment. The unspiked set of obtained from air masses representative of back- filters exhibited a concentration of 48 pptv ground conditions over the subtropical Atlantic sulfide on the front filter. The back-up •ilters Ocean (E. Saltzman and D. Cooper, manuscript in exhibited sulfide signals equivalent to 1.8-3.3 preparation). The laboratory studies were pptv H2S, similar in magnitude to the artifact conducted from a fourth floor rooftop laboratory observed at sea. The spiked set o• filters at the Rosenstiel School of Marine and Atmospheric exhibited higher sulfide signals on all filter Science on Virginia Key, FL. Winds were predomin- stages, with the difference between the two sample antly onshore during these studies. The site has trains ranging from Z! pptv at the front •ilter to been described in detail by Slatt et al. (1978).

Field Observations 80

The measurements of H2S were made in the field ß o using the normal (front) silver nitrate filter 60 with a second (back-up) filter in series as a routine precaution. Previous laboratory and field tests have shown the collection efficiency of a 40 single filter to be greater than •8% (Cooper, 1986). However, on this cruise, which required longer sampling times than previous terrestrial 20- H2 S measurements, a sulfide signal was consis- tently found on the back-up filter. This signal averaged approximately 30% of that on the front filter, with a calculated mean suicide (apparent H2S) concentration of 4.9 pptv (standard deviation = 4.7, n = 44). The concentration on the back-up b filter did not correlate with the sulfide 30- concentration on the front filter. In addition, - when a third stage was added to the filter train, a signal was measured that was equivalent to that 20- on the second stage filter. These observations suggested the presence of a previously undiscovered interference to the lO method. Furthermore, the peculiar "breakthrough" characteristics suggested that the iqte r•ering species was not quantitatively collected on the o silver nitrate filters, but was present at a concentration significantly higher than B2S. An interesting feature of the shipboard Filter Stage measurements is a diurnal cycle in the sulfide Fig. 2. Sulfide concentrations measured in content of the second stage filters, shown in ambient air on Virginia Key, using silver nitrate Figure 1. While there is considerable scatter in filters (a) in the presence (0) and absence (•) the data (a composite from 12 days of sampling), of added carbonyl sulfide, and (b) in •olders it is evident that the second stage concentration covered by aluminum foil (•) and uncovered (0). P08 Cooper and Saltzman: bptake o• Carbonyl Sulfide

3). Temperature, on the other hand, had a significant effect. Figure 4 shows the loss of OCS to silver nitrate filters in test tubes maintainedat 0, 25 and 50øC. The magnitude of the temperature effect alone is sufficient to explain the differences observed between the foil covered and the uncovered filters in the earlier experiment, and to explain the diurnal variations in the artifact observed in the o0 1 shipboard data. o The b•chanism of Artifact Formation

o • , ' ' While the experiments conducted in this study o 2•O 4•O 6•O 1400 havedemonstrated thatOCS interacts with silver Ti me [ minutesJ nitrate impregnated filters to increase the apparent hydrogen sulfide concentration, the Fig. 3. Effect of light intensity and moisture on mechanism of the low efficiency process is not the loss of OCS to silver nitrate filters. known. The •irst step must be the adsorption of Irradiated/dry ([]); irradiated/humidified (O); gaseous OCS onto the filter surface. The dark/dry (A); dark/humidified (V)- subsequenthydrolysis step could occur in either of two ways. In the first case, OCS may remain stable on the filter until extraction of the 10 pptv at the last stage. These differences filter for analysis. Any OCS on the filter would suggest that the efficiency of the process hydrolyse within seconds in the strongly basic generating the artifact is of the order of 1-2% of NaOH/NaCN solution used to extract the sulfide. the added OCS. œhe experimental evidence argues against this Initially we suspected that the diurnal changes mechanism; if adsorption was the rate controlling in the magnitude of the artifact were related to step, the artifact formation would most likely the effect of sunlight irradiation on the •ilters. exhibit a negative temperature dependence, in It was noted that filters collected during the contrast to the positive dependence observed. daylight hours acquired a grey coloration, indica- The other, more probable, mechanism is that OCS ting the formation of metallic silver via the undergoes both adsorption and hydrolysis on the photoreduction of silver ion. While the reaction filter, with the latter being the rate determining between OCS and silver has not been studied, the step. We speculate that the process involves the interaction of OCS with other metals such as gold, following steps: copper and iron is well known (Ammons, 1980; + + Graedel et al., 1981; and references therein). Ag + OCS <.... > [Ag..S=C=O] An experiment was therefore designed to study the effect of insolation on artifact generation. As in the previous experiment, two parallel sets [Ag..S=C=O]+ + H20 --> --> AgS-+ CO2 + 2H+ - + of samples were collected; this time one set was covered with aluminum foil, preventing light from AgS + Ag .... > Ag2S penetrating the translucent filter holders. The is readily available to form an results, presented in Figure 2b, show a signifi- intermediate hydrated adduct, required in the cant difference between the two sets of filters, second step, as it constitutes a significant with the uncovered filters exhibiting a substan- proportion of the paper filter matrix (13% at •0% tially greater artifact. However,this experiment relative humidity;Natusch et al., 1974). This does not conclusively show that the irradiation of spe cula t ive mechanism is analogous to those the silver nitrate filter is the primary lactor governing artifact gene ration. Absorption of sunlight oy the Teflon PFA filter holders also has the effect of increasing the temperature of the 3 filters during sample collection. A series of laboratory experiments was conduc- ted in an attempt to distinguish more specifically between the environmental factors which may affect 2 the uptake of OCS by the AgNO3 filters. The filters were placed in test tubes containing an atmosphere of approximately 1 ppm OCS in nitrogen, and exposed to various light intensities, humidities and temperatures. The concentration of ß OCS in the test tube was measured by periodically sampling small aliquots of the headspace which were analysed by gas chromatography. 0 • Control experiments, using a moist unimpreg- o nated filter paper, showed negligible loss of OCS from the gas phase, indicating that losses of OCS Time {minutesJ to the glass walls or filter material were not Fig. 4. Effect of temperature on the loss of OCS important. Both light intensity and moisture to silver nitrate filters. 0øC ([]); 25øC(0); content of the air were shownto have a negligible 50øC(A); control with no silver nitrate at effect on the loss of OCSto the filter (Figure 50øC(V). Cooper and Saltzman: Uptake of Carbonyl Sulfide 209

discussed by Johnson (1981) for the hydrolysis of wetlands, where the high concentrations make the OCS in aqueous nmdia, with the silver ion acting magnitude of these corrections insignificant. as a catalyst in the reaction. At higher pH, the rate of the reaction can be expected to increase Acknowledgements. This work was supported by as the mechanism shifts to the nucleophilic NSF grant ATM 84-05921. We are grateful for the addition of OH-in place of water. cooperation of both the crew and the scientific party of the R/V oolumbus iselin during CId601. We Implications for the Analysis of H2Sin Air thank Drs. J.M. Prospero and D.L. Savoie for helpful discussion. The low efficiency trapping of OCS on silver nitrate filters observed in our experiments Re fe re nce s generates a sulfide concentration of the correct magnitude to explain the artifact found in Ammons, Jo Mo, Preconcentration Methods for the hydrogen sulfide measurements made both at sea and Determination of Gaseous Sulfur Compoundsin at Virginia Key. The temperature dependance found Air, Ph.D. Dissertation, University of South in the laboratory experiments is the probable Florida, 1980. cause of the diurnal variations found in the Axelrod, H.D., J.H. Cary, J.E. Bonelli, and J.P. magnitude of the artifact. We next considered the Lodge, Jr., Fluorescence Determination of possibility of correcting the field data for this Sub-Parts per Billion Hydrogen Sulfide in the interference in an effort to obtain a real Atmosphere, Anal. Chem., 41, 1856, 1969. background concentration measurement of H2S. Cooper, D. J., Variability in Biogenic Hydrogen The consistency of the artifact sulfide on Sulfide Emissions from Selected Florida sequential back-up filters suggests that it is Ecosystems, M. S. Thesis, University of Miami, reasonable to assume that the efficiency of OCS 1986. conversion is the same at all stages, including Delmas, R. and J. Servant, The Origins of Sulfur the front filter. If this is the case, then a Compoundsin the Atmosphere of a Zone of High simple subtraction of the sulfide concentration on Productivity (Gulf of Guinea), J. Geophys. the back-up filter from that on the front filter Res., 87, 11019, 1982. should give the true ambient •{2S concentration. Graedel, T.E., G.Wo Kammlott, and J.P. Franey, As examples, the data presented •n Figures 2a and Carbonyl Sulfide: Potential Agent of Atmos- 2b can be used to calculate two pairs of replicate pheric Sulfur Corrosion, Science, 212,463, 1981. analyses. As an extreme case, the spiked sample Herrmann, J. and W. Jaeschke, Measurements of H2S in Figure 2a had a signal of 69 pptv on the front and SO2 over the Atlantic Ocean, J. Atmos. filter and 20 pptv on the back-up, yielding a Chem., 1, 112, 1984. calculated concentration of 50 pptv, slightly Jaeschke, W., H. Claude, and J. Herrmann, Sources higher than the 46 pptv calculated • rom the and Sinks of Atmosphetic H 2S , J. Geophys .. unspiked pair. Contaminant H2S from the OCS Rms., 85, 5639, 1980. permeation tube could explain this difference. A Johnson, J.E., The lifetime of carbonyl sulfide in more realistic case, the two samples collected the troposphere, Geophys. Res. Lett., 8, 938, with and w•thout œoil covering (Figure 3b) give 1981.' two identical concentrations of 25 pptv (32-7 and Khalil, M.A.K. and R.A. Rasmussen, Global sources, 27-2 pptv). lifetimes and mass balances of carbonyl sulfide The magnitude of this correct•on •s clearly of (OCS) and (CS2) in the •arth's significance in remote marine air, where H2S atmosphere, Atmos. Environ., 18, 1805, 1984. levels are low. The magnitude of carbonyi sulfide Natusch, D.F.S., R.B. Klonis, H.D. Axelrod, R.J. interference in previous studies is impossible to Teck, and J.P. Lodge, Jr., Sensitive Method for assess, but applying this correction to our own the M•asurement of Atmospheric Hydrogen shipboard data resulted in a substantial lowering Sulfide, Anal. Chem...,44, 2067, 1972. of the H2S ñeveis. Hence, the importance of H2S Natusch, D.F.S., J.R. Sewell, and R.L. Tanner, as a flux of sulfur to the marine atmosphere is Determination of Hydrogen Sulfide in Air- An considerably reduced. Assessment of Impregnated Paper Tape Methods, The artifact may also explain the apparent lack Anal. Chem., 46, 410, 1974. of diurnal variation in H2S noted by Herrmann and Siatt, B.J., D.F.S. Natusch, J.M. Prospero, and Jaeschke (1984) or the reversed cycle noted by D.L. Savoie, Hydrogen sulfide in the atmosphere Delmas and Servant (19•2). A species as reactive of the northern equatorial Atlantic Ocean and as H2S should be removed during the daytime by its relation to the global sulfur cycle, Atmos. photochemically generated OH radicals, but should Environ., 12, 981, 1978. accumulate at night. This behaviour was noted in our own measurements of H2S in remote marine air D. J. Cooper and E. S. Sal•zman, Rosens•iel (E. Salt zman and D. Cooper, manuscript in School of Marine and •mospheric Science, preparation) only after correcting the H2S data University of Miami, •600 Rickenbacker Cswy., tor OCS interference. Miami, F1 While this correction is necessary for H2S data from remote regions, many of the previous studies of atmospheric H2S have been conducted in polluted (Received December•, •986; air, over exposed mudflats, or in coastal accepted January 2•, •987.)