INTRODUCTION. In spite of our recent finding that reaction probabilities of ozone (Kamm et al., 1999) and other important oxidising trace gases (Saathoff et al., ...
TRACE GAS INTERACTIONS WITH SPARK GENERATED AND DIESEL ENGINE SOOT A contribution to subproject CMD
H. Saathoff, S. Kamm, O. Möhler, K.-H. Naumann, and U. Schurath Forschungszentrum Karlsruhe, Institut für Meteorologie und Klimaforschung Postfach 3640, D-76021 Karlsruhe, Germany
INTRODUCTION In spite of our recent finding that reaction probabilities of ozone (Kamm et al., 1999) and other important oxidising trace gases (Saathoff et al., 1999) with spark generated soot are very low in dry synthetic air, and tend to decrease with increasing exposure time and decreasing temperature, speculations are still running high that soot might affect the chemistry even of the lower stratosphere with average soot concentrations of only about 0,64 ng/m3 (Strawa et al., 1999). Therefore we wanted to complement our investigations with experiments under more typical tropospheric conditions. Recently we have complemented our earlier investigations in dry air by studying interactions of N2O5 with spark generator soot at 50% relative humidity. Furthermore we have started a programme to investigate trace gas interactions with Diesel engine soot. EXPERIMENTAL AND MODELLING To investigate the kinetics and mechanisms of the soot - trace gas interactions we use an evacuable aerosol chamber of 84 m³ volume which can be thermostated in the temperature range from 183 to 333 K. An important advantage of our large aerosol chamber is the fact that trace gas – soot interactions can be investigated on time scales of several days, thus approaching typical tropospheric aerosol lifetimes, and over a wide range of temperatures and relative humidities. To obtain reaction probabilities of O3, NO2, HNO3, N2O5, NO3, HO2, and HO2NO2 on soot we measured the trace gas behaviour in our reactor in the absence and in the presence of the soot particles. Model soot particles were generated using two graphite spark generators (GfG 1000, Palas). The initial mass concentrations typically amounted to about 200 µg/m3 corresponding to roughly 2.7 m2 of accessible soot surface area inside the reaction chamber. To investigate Diesel soot particles it was found necessary to separate the soot particles from a VW-TDI Diesel engine (working at 2500 rpm, 17 kW) from water vapour and other gaseous pollutants (NOx, VOCs). This was achieved by passing the 1:10 (VKL 10, Palas) pre-diluted exhaust gas through three large denuders in series. The cleaned aerosol was sent through a 100 m pipe into the AIDA aerosol chamber. Typical initial Diesel soot mass concentrations were about 90 µg/m3. Ozone was generated with silent discharge generators (Witte, Sorbios) using O2 (5.0). NO2 was introduced from a 0.1% mixture in synthetic air (99.5% purity). N2O5 was synthesised by oxidation of NO with O3 and used as a 0.4% mixture in nitrogen. HO2NO2 was produced by reaction of NO2BF4 with H2O2 (>90% purity). Aerosol characterisation included number concentration (CNC, TSI), mass concentration (filter samples and coulometric analysis), size distribution (SMPS, TSI), and particle shape (electron microscopy). Trace gas concentrations were monitored in situ employing two long path absorption systems (112 m/254 m, White type) equipped with FT-IR/FT-VIS spectrometer Proc.: EUROTRAC Symposium 2000, Garmisch-Partenkirchen, March 27-31, 2000
(IFS 66v (IR) and IFS 66/S (VIS), Bruker) as well as ex situ with O3 and NO/NOx monitors. The experimental investigations are supported by model calculations with the physicochemical aerosol simulation code COSIMA. This model does an excellent job in describing time-dependent size distributions of both spark generated and Diesel engine soot in our aerosol chamber on time scales of several days, thus lending credibility to calculations of the accessible aerosol surface area, which is needed to run the chemical code. This program simultaneously accounts for the dynamics of fractal particles, particle sources and sampling losses, gas to surface transport, homogeneous as well as heterogeneous chemistry, and the optical properties of the particles. The following table summarises some of the model parameters used. COSIMA Model Parameter
Spark Soot
Fractal dimension
Diesel Soot
2.0
2.0
Diameter of the primary particles [nm]
6±2
27±3
3
Density of the primary particles [g/cm ]
2.0
1.7
Volume filling factor
1.43
1.43
100-200
50-80
Accessible surface area [m2/g] (size dependent, cf. (Xiong et al., 1992))
RESULTS FOR DIESEL ENGINE SOOT -4
120
9x10
100
-4
7x10
-4
80
-1
6x10
Extinction [m ]
3
Mass Concentration [µg/m ]
-4
8x10
-4
5x10
60
-4
4x10
Carbon Mass (Filter Sample) Model Extinction (473 nm)
40
20
-4
3x10
-4
2x10
-4
1x10 0 0
2
4
6
8
10
12
14
16
18
20
0 22
Time [h]
Fig.1: Evolution of the Diesel soot mass concentration and extinction.. 30000 25000
dN/dln dme
20000 15000
0.1 h 0.5 h 6.1 h 20 h Model
10000 5000 0 0,03
0,1
1
dme [µm]
Fig.2: Evolution of the Diesel soot size distribution.
Diesel soot has a lifetime of about 10 days in the AIDA aerosol chamber. Figures 1 and 2 show the observed aerosol dynamics and the COSIMA model results. After introducing the cleaned Diesel exhaust (remaining NOx content about 20 ppb) into the AIDA chamber 120 ppb ozone were added (Fig.3). A relative humidity of 0.7% in AIDA was calculated from the measured dew point. The evolution of the soot and ozone concentrations was measured for two days. The significant differences in the ozone concentration profiles in presence and absence of Diesel soot are due to reaction of the remaining NOx with ozone. No significant heterogeneous reaction of O3 could be observed and consequently we can calculate an upper limit for the reaction probability of ozone on Diesel soot assuming a passivation time of 12 h
like for spark generated soot: O3 -(Diesel soot)→ products;
N2O5 -(Diesel soot)→ 2 HNO3;
120 110 100
Concentrations [ppb]
After introducing Diesel exhaust 1:10 diluted without further purification into the AIDA chamber (Fig.4) about 240 ppb of NO and 60 ppb of NO2 were measured. A relative humidity of 0.5% in AIDA was calculated from the water content of the diluted exhaust gas. 6 hours later about 800 ppb of ozone were added to this mixture which led to the transformation of NOx into N2O5 and eventually HNO3. With the assumption of unchanged wall losses for the reactive trace gases compared to earlier experiments we were able to estimate an upper limit for the reaction probability for the hydrolysis of N2O5 on Diesel soot:
γ(296K, 0.7% r.h.) ≤ 5*10-6
90 80 70 60 50
O3 only
40
O3 + Soot 106 µg/m O3 only NOx (Diesel Exhaust) Model γ = 0
30 20 10
Model γ = 5*10
-6
Model γ = 1*10
-5
3
T = (296±2) K [H2O] = (200±30) ppm
0 0
10
20
30
40
50
Time [h]
Fig.3: Evolution of the trace gas concentrations in the presence and absence of Diesel soot. The Lines show the COSIMA model result for different reaction probabilities.
γ(296K, 0.5% r.h.) ≤ 10-3
RESULTS FOR SPARK GENERATED SOOT:
600
Adding Diesel Exhaust
Concentrations [ppb]
500 To obtain reaction probabilities of O3, NO2, HNO3, N2O5, NO3, HO2, 400 and HO2NO2 on spark generated soot O (FTIR) we measured the trace gas behaviour 300 NO (FTIR) in our reactor in the absence and in N O (FTIR) HNO (FTIR) 200 the presence of the soot particles. The O (Monitor) NO (Monitor) initial trace gas concentrations, the NO (Monitor) 100 Model temperature, and the relative humidity were varied. For the ozone 0 -6 -4 -2 0 2 4 6 8 10 reaction a complex mechanism Time [h] including passivation led to a time dependent effective reaction Fig.4: Evolution of the trace gas concentrations after filling aerosol from a Diesel engine into the AIDA. At a time probability. For details see reference exhaust equal to zero 800 ppb ozone were added. (Kamm et al., 1999). In order to distinguish between the reactions of the radical precursors N2O5, and HO2NO2 and the radicals NO3, and HO2 experiments were started with different initial NO2 concentrations to change the equilibrium radical concentrations. Although the walls of the aerosol chamber act as efficient sink for the reactive trace gases especially at higher humidities, a significant reaction of N2O5 on soot was still detectable. For HO2NO2/HO2 the wall loss was too fast an therefore only upper limits of the respective reaction probabilities were obtained. The following table lists the reaction probabilities obtained by fitting the simulated trace gas behaviour to the observations with the COSIMA model. Reaction Probabilities at 296 K Reactions 3
2
2
5
3
3
2
Literature O3 --(Soot)→ NO2 --(Soot)→ HNO3 --(Soot)→ N2O5 N2O5 N2O5 NO3 NO3 HO2 HNO4
products NO NO, NO2
This work (Palas)
-3
γ = 3⋅10 (Fendel et al., 1995) γeff ≤ 2⋅10-6 -8 γ = 3⋅10-4 (Kalberer et al., 1996) γ ≤ 4⋅10 γ = 3.8⋅10-2 (Rogarski et al., 1997) γ ≤ 5⋅10-6
--(Soot)→ 2 HNO3 --(Soot /50%r.H.)→ 2 HNO3 --(Soot)→ NO + NO2 γ = 3⋅10-3 --(Soot)→ NO / NO2 --(Soot /50%r.H.)→ products --(Soot)→ products γ≤1 --(Soot)→ products
(Brouwer et al., 1986)
(Ivanov et al., 1997)
(Kamm et al.)
γ = (4 ± 2)⋅10-5 γ = (2 ± 1)⋅10-4 γ = (4 ± 2)⋅10-6 γ ≤ 3⋅10-4 γ ≤ 1⋅10-3 γ ≤ 10-2 (278 K) γ ≤ 10-5 (278 K)
CONCLUSIONS The reactivity of Diesel soot towards ozone is comparable or smaller than that of spark generated soot. Its reactivity towards N2O5 is not dramatically larger than that of spark generated soot but further investigations are necessary to substantiate this preliminary result. The reaction probabilities we measured for spark generated and for Diesel soot at atmospheric time scales, temperatures, and humidities are much smaller than most published values measured on shorter time scales. The high initial reactivities do not persist on relevant time scales. Consequently box model simulations by Vogel et al. (1999) using our data showed only very little influence of tropospheric soot aerosol on the oxidation capacity even for high aerosol concentrations. For the rather small carbonaceous aerosol concentrations in the tropopause region a possible impact of heterogeneous reactions on atmospheric chemistry is extremly unlikely. Complementary to short time laboratory experiments studies on airborne material conducted over several days are indispensable to realistically assess the possible impact of soot particles on atmospheric chemistry. Due to the existence of saturation and passivation processes the extrapolation of results from investigations covering only short time scales to atmospheric conditions can be severely misleading. REFERENCES Kamm S., O. Möhler, K.-H. Naumann, H. Saathoff, U. Schurath; Atmos. Environ. 33, (1999) 4651-4661. Saathoff H., S. Kamm, O. Möhler, K.-H. Naumann, A. Nink, U. Schurath; Reactions of N2O5, NO3, and HO2 with Soot Aerosol; Proceedings of the EC / EUROTRAC 2 Joint Workshop: “Chemical Mechanism Development”, Aachen.September 1999, pp. 215-218. Strawa et al.; J. Geophys. Res. 104 (D2), (1999) 26753-26766. Fendel W., D. Matter, H. Burtscher, A. Schmidt-Ott; Atmos. Environ. 29 (1995) 967-973. Kalberer M., K. Tabor, M. Ammann, Y. Parrat, E. Weingartner, D. Piguet, E. Rossler, D.T. Jost, A. Turler, H.W. Gaggeler, U. Baltensperger; J. Phys. Chem. 100(38) (1996) 15487-15493. Rogaski C.A., D.M. Golden, L.R. Williams; Geophys. Res. Lett. 24 (1997) 381-384. Brouwer L., M.J. Rossi, D.M. Golden; J. Phys. Chem. 90 (1986) 4599-4603. Ivanov et al.; 4th International Conference on Chemical Kinetics, Gaithersburg 1997. Vogel et al.; Presentation at the European Geophysical Society, The Hague 1999. Xiong Y., S.E. Pratsinis, A.W. Weimer; AIChE Journal 38 (1992) 1685-1692.
