Fresenius' Joumal of
Fresenius J Anal Chem (1995) 351:27-32
@ Springer-Verlag 1995
A continuous flow instrument with flow regulation and automatic sensitivity range adaptation for the determination of atmospheric HzS W. Jaeschke, H. Schunn, W. Haunoid Zentrum fiir Umweltforschung, Johann Wolfgang Goethe-UniversitS_t, Georg Voigt-Strasse 14, D-60325 Frankfurt a.M., Germany Received: 21 May 1994/Revised: 9 July 1994/Accepted: 15 July 1994
Abstract. A set-up for continuous measurements of atmospheric H2S concentrations in the range between 0.138 lag/m3 and 17.5 lag/m3 is described. All flows are regulated by flow sensors which are connected to the respective pump by a feed back circuit. The instrument is working in two sensitivity ranges. One is provided for low HeS concentrations (lower than 1.75 i_tg/m3) and the other for higher levels (higher than 1.75 btg/m3). During continuous measurements the set-up enables an automatic adaptation of one of the sensitivity ranges to ambient HzS concentrations. The critical step of absorption of gaseous H2S into the liquid phase of the continuous flow system could be stabilised by the development of a new fluid level control system. After numerous test measurements under laboratory conditions the instrument was tested for the first time in June 1993 during field measurements in the tropics. These measurements took place in the environment of natural and industrial sources of H2S in the vicinity of Salvador da Bahia in Brazil. Data are presented which prove the ability of the instrument.
1 Introduction H2S is emitted into the atmosphere by biological and anthropogenic sources. On a global scale biological sources emit 9.14 Tg (Teragram × 10 lz g) (H2S)/year 1,1~. This exceeds by far the anthropogenic emissions of only 1-5 Tg (H2S)/year 1,2]. Considering biological emissions, vegetation and soil are contributing 6.6 Tg which is 72%. One Tg is emitted by anaerobic processes from swamps and tidal flats corresponding to 11% and approximately the same amount is emitted by volcanic activities 1,3]. Only 0.5 Tg (5%) are exhaled from the oceans and a small amount of 0.04 Tg is emitted through biomass burning. Main anthropogenic sources are refineries, cellulose factories and coke productions. Recently some data have Dedicated to Professor D. Dieter Klockow on the occasion of his 60th birthday Correspondence to: W. Jaeschke
been published about the emission of }:'[2S by automobiles due to the application of catalysts in the exhaust pipe I-4, 5]. This anthropogenic source may become very important in the following years. The main sink of atmospheric H 2 S is the reaction with photodynamically produced OH radicals. H2S is first oxidised to SO2 and further on to sulphate. All oxidation products are removed from the atmosphere either by wet or dry deposition. The rate constant of the reaction of H i S with OH radicals is 5×10 12cm3mol ls-1 1,6]. Assuming a mean OH radical concentration of 5 x 105 molecules/cm3 an atmospheric residence time of 4-5 days can be assessed for H2S. During the last decades many attempts have been made to characterise and quantify biogenic fluxes of H2S by the aid of field experiments. Analytical methods applied for this purpose mostly consisted of discontinuously working sampling procedures and subsequent analyses of the samples. It was either done by gas chromatography [7~, by fluorescence quenching method [8], by molybdeneblue-sorption (VDI 2454, 1974) or by methyleneblue-impinger method (VDI 5454, 1976). A common way of sampling atmospheric H2S is the application of filters 1,8~ which are impregnated with AgNO3. When H2S containing air is sucked through such a filter it reacts with AgNO3 to form AgS. The sulphide is removed from the filters by a cyanide containing washing solution which dissolves AgS by complexation. The content of sulphide in the washing solution is determined by a fluorescence quenching effect which is caused by reaction of sulphide with fluoresceine mercuricacetate (FMA). This filter method has the disadvantage that it is working discontinuously. The same holds for gas chromatography and the other methods. For more detailed studies of biogenic H2S emissions it is necessary to observe the behaviour of the sources continuously and with higher time resolution. This should enable the observation of interactions of H2S emissions with meteorogical conditions 1,9,10] such as sunshine, rain, wind direction, temperature, humidity and in special cases tidal movements [11]. Therefore a necessity was seen for the Construction of a continuously working H2S instrument which is based on the continuous flow principle.
28 In a first attempt, a similar instrument was constructed for field measurements in tropical regions like Brazil [121. However, during its application two disadvantages arose. When the instrument was installed for unattended runs during periods of 6 or 7 h (one night), often deviations in the solvent fluxes occurred resulting in incorrect analytical informations. However even in cases when fluxes could be kept constant the atmospheric HaS concentrations sometimes rose to levels above the working range of the method. In this paper an improved instrument is described which is able to overcome these disadvantages.
2 Description of the method A scheme of the entire set-up is shown in Fig. 1. Ambient air containing traces of HzS is sucked through the instrument by the compressor {A}. The air flow is regulated by the gas flow controller {B}. Traces of HzS in the sampled air are absorbed by a washing solution which contains 0.1 mol NaOH/1. By the aid of peristaltic pump {C} this sampling solution is transported from the storage container {D} to the PTFE mixing coil {E} where it meets the air to be analysed. By the time the mixture of air and absorption solution leaves the coil, all HaS should be in the liquid phase as sodium sulphide. Then liquid and gaseous phases are separated in the separator {F}. The gas phase is sucked away from the separator by compressor {A}. In order to protect gas flow controller {B} and compressor {A} against possible overload of the separator {F}, the air stream is passed through a Woulfe flask {G}. The separated liquid phase, containing the absorbed traces of sulphide, is transported by peristalticpump {H} to the level regulator {I}. The level regulator has two control functions. It controls the flow of the washing solution through the mixing
Air Inlet
G
coil and the outflow of the washing solution to the analytical instrument. The working principle is as follows: the height of the liquid column in the level regulator is controlled by pressure sensor {J}, which is communicating in a feed back connection with peristaltic pump {C}. In cases of very dry ambient air, the fluid level in the regulator may fall beyond a nominal value. If that happens, the main control board {K} gives a signal to inlet pump {C} to accelerate its flow until the nominal level of liquid is reached. In contrast, when in the presence of high humidity the level of liquid is increasing above a nominal value, a signal is sent to the pump for flow reduction. With this set-up, disturbances which are usually caused by fluctuations of the relative humidity are suppressed and do not affect the subsequent analytical part of the flow system. Without any disturbances the washing solution is transported through a modified flow cell of the fluorimeter {L} by peristaltic pump {M}. Before the flow reaches the fluorescence instrument a solution of 1.94 x 10-7 mol/1 fluoresceine mercuricacetate (FMA) (Fluka 46980) in 0.1 tool NaOH/I is added via the T-piece {N}. The solution is transported by peristaltic pump {P} from storage container {O} via flow controller {Q} to the T-piece {N}. In order to provide a constant flux, peristaltic pump {P} has a feed back connection with the flow controller {Q}. To make sure that the washing solution and the reagent solution are mixed homogeneously, they are both led through the mixing coil {R}. The chemical reaction which is causing a quenching of the dye is described by Natusch [8]. The fluorescence signal is detected by ftuorimeter {L} (Merck LiChrograph F-1050) and recorded with a laptop. The absorption wavelength of the FMA is 480 nm and the emission wavelength is 530 nm. The analogue signal of the fluorimeter is in the range between 0 and 1.32 V. High amounts of sulphide cause total quenching. In this case the fluorescence signal is nearly 0V. In contrast low
Fig. 1. Scheme of the entire set-up of the continuous flow instrument with regulated fluxes and automatical range adaptation for the determination of atmospheric H2S. A: Compressor; B: Gas flow controller; C: Peristaltic pump for washing solution; D: Storage container for washing solution; E: Coil for mixing sample air and washing solution; F: Gas liquid separator; G: Woulfe flask; H: Peristaltic pump for sulphide containing washing solution; I: Level regulator; J: Pressure sensor; K: Main control board; L: Fluorimeter with a modified flow cell; M: Peristaltic pump for waste; N: T-piece; O: Storage container for dye solution; P: Peristaltic pump for dye solution; Q: Flow controller for dye solution; R: Coil for mixing dye and washing solutions; S: Flow sensor for waste; T: Waste container; U: Syringe for injection of standard solutions or liquid samples
29
,ressure sensor teflon tube flow chamber air fluid level
> H,S C o n c e n t r a t i o
Fluorescencequenching
~ ........... 1O0 o '/~-
l inlet
I I hi
nm ~.SLI, I'~.,W I,,I..,l~
,
outlet
Fig. 2. Detailedschemefor the liquid flow sensor
amounts of sulphide cause only a small quenching effect. In this case the fluorescence signal is close to 1.3 V. After passing the flow cell of the fluorimeter {L}, the reagents are transported by peristaltic pump {M} through flow controller {S} into the waste container {T}. Flow controller {S} is controlling the outlet of the liquid flow by a feed back connection to the peristaltic pump {M}. For liquid calibration, standard solutions can be injected into the flow of the washing solution with syringe {U}. The same injection procedure can be used if liquid samples are to be analysed.
3 Principle of the flow sensor A prerequisite for the construction of the analytical set-up described in section 2 was the availability of extremely sensitive flow sensors. Reproducible values for sensitivity, baseline of the signal and detection limit can be observed only under constant flux conditions. In practice solvent flows established by peristaltic pumps may deviate due to the corrosion of the tube or untight connections caused by shocks during transport especially during field campaigns. These deviations of the solvent flow can only be prevented by liable flow sensors connected in a feed back circuit to the pump responsible for the respective solvent flow. Since such flow sensors have been designed and constructed in our working group [13], they should be described here in more detail. The principle of the flow sensor (Fig. 2) is based on the pressure difference which occurs when a medium is led through an orifice. Generally, this principle is applied to control the flow of gases. For the work with a liquid, the orifice is replaced by a so-called throttle tube. Its resistance is determined by its length and inside diameter. Both parameters can be chosen according to the desired flow rate. The resistance of the throttle tube causes a difference in pressure between its inlet and outlet. The difference which is proportional to the flow rate is measured by a piezoresistive pressure sensor (Fa. Honeywell Type 142 SColD). In order to provide sufficient buffer volume, the
I #
t
50 %
,,
I I #
I
•
t t
t: It
t
t t
11~.~,Output-Signal
t t
I
I
I
I
':/!
| I Iii
I II U | |
I
II
0%
Range A
Range B
Range A Time
Fig. 3. Measuring signal during automatic changing
senslnvity
range
Teflon tube which connects the flow line with the pressure sensor is mounted as a loop. With this set-up it is possible to send an electronic signal of the flow rate to the peristaltic pump in a feed back connection. By the aid of an operational amplifier the liquid flow is controlled and automatically stabilised at a chosen nominal value. These flows are determined by the desired ratios of the reactants in the solvents. They can be varied in order to obtain different ranges of analytical sensitivity. The proportionality between the liquid flow and the pressure difference given in-Voltage must be established. As an example the linear regression of the outlet flow is: Y(voltage) = 2.4804 X(ml/min)+ 1.0072.
4 Principle of automatic sensitivity variation The principle of the automatic sensitivity range variation can be explained by Fig. 3. In the presence of low H2S concentrations the instrument runs with low flow-rates of dye solution (0.37 ml/min). This means, that in the most sensitive range A, small traces of HaS in the range of 0.21 pg/m 3 are still causing significant signals of fluorescence quenching in the order of 10%. If now the concentration of atmospheric H2S is rising, the quenching signal is allowed to rise up to a threshold value of 80%. When this level is reached, the peristaltic pump, which provides
30 the flow of the dye solution gets an electronic signal to accelerate the flow to a higher value (1.13 ml/min). This means that the ratio of reactants is shifted and higher amounts of dye are available. Therefore the quenching signal is decreasing to only 10%. The instrument is now working in the less sensitive range B. In this range it can determine H2S concentrations up to 16 lag/m 3 corresponding to a quenching signal of 90%. Above 90% the relation between the sulphide concentration and the quenching signal is no longer linear. As soon as ambient H2S concentrations are decreasing again and in range B a quenching signal of 10% is reached, the flow of the dye solution is reduced. This reduction changes the 10% signal into an 80% signal. The instrument continues to work in the most sensitive range A as long as the quenching signal remains below 80%. Before starting measurements the two sensitivity ranges have to be calibrated separately.
80
o~ 60
,_== ®
4o
Range B • liquid • gas
20
b
80
Z 6O
5 Calibration
._==
The analytical method can either be calibrated in the gaseous or the liquid mode. ]?or the gas phase calibration small amounts of H2S emitted from a permeation tube are diluted by pure nitrogen with a flow of 100ml/min. The permeation tube is kept in a heated chamber with a proportional temperature control providing a constant temperature of 30°C. The temperature drift is less than _+ 0.1°C. The gas mixture leaving the thermostat chamber is further diluted with clean air by a two step gas dilution system. The clean air is generated by pressing ambient air through silicagel and activated charcoal. With this system it is possible to provide a test gas with H2S concentrations down to the low ppt range. In order to use this source of calibration gas under field conditions a mobile version has been designed which is described elsewhere [14]. In the liquid phase the instrument is calibrated with standard solutions. These are prepared by dissolving a crystal of N a z S . 9 H 2 0 in 0.1mol NaOH/1. The content of the stock solution is determined photometrically by the reaction of sulphide with leucomethyleneblue (Dr. Lange Test LCW 053). For calibration purposes diluted standard solutions can be injected into the flow of the washing solution at the T-piece {N} (Fig. 1). To convert the sulphide concentration of the liquid standard into that of an air stream, the following equation can be used: C~a = Cw~ x Fw~/Fsa x F ....
1 O0
(1)
C~, = Concentration in sampling air [gg/m 3] Cws = Concentration in standard solution [btg/ml] F,v~ = Flow of the washing solution [ml/min] F~a = Flow of the sampling air [m3/min] F .... = Correction factor [%] In general in the described flow system with integrated gas absorption a liquid phase and a gas phase calibration cannot yield identical calibration curves, because H2S is adsorbed only up to an equilibrium concentration. Therefore a correction factor F .... has to be included in Eq. (1).
0
4O
Range A liquid • gas
20
0
~ 0
2
4
6
8
10
12
14
16
18
p g H2S/m 3
Fig. 4a, b. Comparison of calibration curves obtained either by liquid standards or test gases, a Most sensitive measuring range A. b Less sensitive measuring range B
This factor can be determined by the comparison of liquid phase and gas phase calibration. Such comparisons performed in the two sensitivity ranges are shown in Fig. 4. It can be seen from Fig. 4a that in sensitivity range A the difference between the two kinds of calibration is so small that it can be neglected. This is due to the fact that the absolute amounts of H2S which remains as equilibrium concentrations in the gas phase are so small, that they are in the order of the standard deviation of the blank value. However, a comparison of the calibration curves obtained in range B yields a noticeable difference (Fig. 4b). The knowledge of this difference enables the determination of a correction factor which makes it possible to apply liquid phase calibration for gas phase measurements according to Eq. (1). The detection limit of the method has been determined in range A by the mean of the blank value plus three times its standard deviation. In the case of a fluorescence quenching method the blank is represented by the maximum fluorescence signal. Its value plus three times the standard deviation corresponds to a sulphide concentration in the washing solution of 0.089 ng/ml - or according to Eq.(1) - to an atmospheric H2S concentration of 138 ng/m 3.
31 Table 1. Quenching effects by various reduced sulphur gases [15]
Compound
Concentration [pg/m 3]
CH3SH
6.77 i2.8 38.6 42.1 125 486 1.26 2.38 7.16 1.41 2.69 8.07 481 1350
COS
CS2
(CH3)2S SOz
1,0
1,0
0,8
0,8
06
o,6
QuenchiNg [%] 0.68 1.59 4.54 0.23 1.36 7.48 0 0 - 0.35 0.12 0.46 0.81 0.12 1.92
Z 03
0,4
O,4
o,2
02
,
00 13:00
16:00
~9:00
22:00
01:00
04:00
,
0,0
07:00
Time
Fig. 5. Diurnal variation of H2S concentration measured at the campus of the Federal University of Bahia between June 20 and June 21 1993
Another aspect, which has to be discussed in the case of a new analytical development is the possibility of interferences. Therefore the influence of other reduced sulphur gases have been intensively studied. Table 1 shows quenching effects which are caused by different concentrations of various reduced sulphus gases. They have been introduced into the air inlet of the system using a gas dilution system [14]. As can be seen from Table 1 none of the applied gases is causing significant quenching effects, at least not in those concentration ranges in which they are present in the atmosphere.
10
~6
I
14
14
12
12
8
0
6
3
6
4
6 Application of the method during field measurements
As indicated in the introduction the flow method for continuous measurements of atmospheric H2S was designed to permit more intensive studies regarding the contribution of natural and anthropogenic H2S sources to the atmospheric sulphur content on regional as well as on global scale. In particular the knowledge about exhalations of H2S from natural ecosystems and from anthropogenic activities in the tropics of the southern hemisphere is very weak. Therefore the Reconcaro Baiano, which is the area surrounding and comprising the bay of Salvador in Brazil (Bahia de Todos Santos) was selected for the intended investigations. It provides several estuaries and extensive mangrove areas as natural sources. On the other hand, since this area has experienced one of the fastest industrial growths of Brazil, a tot of anthropogenic emissions are present. For instance the largest petrochemical complex of South America with 50 plants is situated at this bay. The data presented here are the results of preliminary measurements which have been performed in June 1993 in co-operation with the group of Prof. Tania Tavares from the Federal University of Bahia. In Fig. 5 diurnal variation of the H2S concentration is shown which was measured at the campus of the University of Bahia from June 20 to June 21 1993. The instrument was left unattended and worked properly for 24 h. The behaviour of the natural background could be measured. H2S concentrations reaches maximum values in the
2
0
.
18:00
.
.
20:00
.
22:00
00:00
i
02 O0
i
04 O0
'
06;00
Time
Fig. 6. H2S concentrations measured during the night between June 22 and June 23 I993 in the area of an oil refinery at the coast of the Bay of Bahia. Maximum values are due to test emissions of exhausts of the plant
morning hours which are comparable to those at the respective time of the previous day. However, most of the time the H2S concentrations were below the detection limit of 0.138 pg/m 3. As another example, the results of the unattended measurements during the night of June 22 to June 23 1993 are shown in Fig. 6. In that night the instrument was placed in the middle of the oil refinery at the coast of the All Saints Bay. In the area of this industrial source, ambient H2S concentrations are about ten times higher than at the university campus and no variation determined by sink reaction can be seen. The maximum values which go up to 9 btg/m 3 in the time between one and two o'clock of the night are due to test emissions which have been performed by technicians of the plant, in order to test the sensitivity of our method. The influence of natural sources was studied in the night of June 23 to June 24 1993 in the estuaries
32 I t
'
'
I I
'
i
,
18
'
16
14 12
14
12
10 ~ 8 % z
z~ 6
8 4
4
2
8
2 I
I
I
I
I
I
18:00
21:00
00:00
03:80
06:00
09:00
data could be always evaluated and discussed together with the Brazilian partners during the field compaign. The instrument became the basis of a common project for the study of the exhalation of reduced sulphur gases from natural ecosystems and from anthropogenic activities in Bahia, which is founded by the VW-Stiftung. While this paper is written more detailed measurements are being performed in Bahia with the new continuous flow system. All these scientific contacts with the Brazilian partners have been intensively stimulated by Prof. Dieter Klockow. Therefore we have dedicated this paper in honour of his 60th birthday.
0
Time
Fig. 7. HzS concentrationsmeasured during the night between June 23 and June 24 1993in the environment of a fisherman'svillage at the coast of Itaparica Island in the Bay of Bahia
downwind of Itaparica Island in a mangrove area. The instrument was installed at the end of a fishermen's village in a small shed. At the beginning of the measurements the instrument showed signals in the range of 6 btg/m 3. However, as can be seen from Fig. 7, at 22 : 00 the H2S concentration became so high that they even exceeded the maximum of the less sensitive measuring range of 17.5 btg/m 3. These high concentrations are obviously due to the fact that during that night the festival of St. John was celebrated, which in Brazil is connected with a lot of campfires and fireworks.
7 Conclusions
The advantage of the development of the continuous flow instrument described above can best be demonstrated by a comparison with the classical sampling methods with subsequent analysis either by gas chromatography or discontinuous fluorescence quenching. Assuming that such a discontinuous sampling method would have been applied, the measuring data shown in section 6 correspond to more than 120 h mean values. The time needed for the chemical analysis of one of the corresponding samples is in the order of 20 min. This means that the total analysis of all samples would have lasted for more than one working week. Moreover, in contrast to the accumulation, method, the data obtained with the continuous flow system are directly available during measurement activities in the field. Therefore, consequences for measuring strategies can be drawn right at the measuring site. This is very important especially when the measurements are conducted in far distant countries as it is described in section 6. In that case the measurement
Acknowledgements. This project is part of a cooperation between the Zentrum fiir Umweltforschung (ZUF) Johann Wolfgang GoetheUniversifftt, Frankfurt a.M. and Companhia de Econologia de Saneamento Ambiental (CETESB, Sao Paulo) under the governmental agreement and cooperationin the field of scientific research and technological developmentbetween Germanyand Brazil. This research has been supported by the Bundesministerium f/ir Forschung und Technologic (Project ENV 3, BMFT project No. 948485 and code 07 INT 059). We thank the members of the Federal University of Bahia Luis Sergio Nunes, Vania Rocha and Tania Tavares for the cooperation and the friendly hospitality during the H2S measurements in Bahia. We also thank the reviewer for their helpful comments.
References
1. Andreae T, Jaeschke W (1992) In: Howarth R, Stewart J, Ivanow M (eds) Sulphur cycling on the continents, Wiley, Chichester, pp 27-61 2. Cullis C, Hirschler M (1980) Atmos Environ 14:1263-78 3. Berresheim H, Jaescke W (1983) J Geophys Res 88 : 3732-40 4. Carlowitz F, Henk M, Gagneret P (1990) Use of a mass spectrometer to continuously monitor H2S and SO2 in automotive exhaust. SAE Technical Paper Series 900272 5. Lies K (1988) Nicht limitierte Automobil-Abgaskomponente. Volkswagen AG 6. Cox R, Sheppard D (1980) Nature 284:330-331 7. Haunold W, Georgii H-W, Ockelmann (1992) LC.GC: 28-35 8. Natusch DFS, Klonis HB et al. (1972) Anal Chem 44 : 2067-70 9. Renneberg H (1991) In: Saltzmann E, Cooper W (eds) ACS Symp Ser 393 : 44-57 10. Lamb B, Westberg H et al. (1987) Atmos Chem 5:469-491 11. Jaeschke W, Claude H, Hermann J (1980) J Geophys Res 85 : 5639-5644 12. Jaeschke W, Dippell J, Sitals R, Haunold W (1992) In: Hendry M J, Krouse HR (eds) Workshop sulphur transformation in soil ecosystems. Saskatoon, Canada, 5-7 November, pp 89-113 13. Haunold W (1993) Patent application: Vorrichtung zur Bestimmung der Durchflul3rate eines fluiden Mediums nach dem Differenzdruckprinzip, P4308313, 7-52 14. Jaeschke W, Dippell J, Sitals R, Haunold W (1994) Inter J Environ Anal Chem 54:315-337 15. Kleifges K-H: (1988) Diplomthesis, Johann Wolfgang Goethe-University Frankfurt