Temporal variations in formaldehyde, acetaldehyde and acetone and ...

7 downloads 0 Views 825KB Size Report
Abstract--Formaldehyde was measured at the TOR station, Schauinsland, (48°N, 6°E, 1250 m asl) in. September 1992 during the TRACT intensive field ...
Pergamon

Atmospheric EnvironmentVol. 30, No. 21, pp. 3667-3676, 1996 Copyright © 1996 ElsevierScience Ltd P I I : S1352-2310(96)00025-8 Printed in Great1352_2310/96Britain. All rights$15.00reserved+ 0.00

T E M P O R A L V A R I A T I O N S IN F O R M A L D E H Y D E , ACETALDEHYDE AND ACETONE AND BUDGET O F F O R M A L D E H Y D E AT A R U R A L SITE IN SOUTHERN GERMANY J. SLEMR and W. JUNKERMANN Fraunhofer Institute IFU, Krenzeckbahnstrafle 19, D-82467 GarmischPartenkirchen, Germany

and A. VOLZ-THOMAS ICG-2, Forschungszentrnm Jiilich, D-52425 Jiilich, Germany (First received 3 November 1994 and in final form 1 January 1996) Abstract--Formaldehyde was measured at the TOR station, Schauinsland, (48°N, 6°E, 1250 m asl) in September 1992 during the TRACT intensive field campaign, using concurrently a continuous enzymatic/fluorometric technique and the DNPH cartridge method, which also allowed the concentrations of CHaCHO and CHaCOCH a to be determined. The DNPH method gave, on an average, 0.84 ppb higher CH20 concentrations than the fluorometric technique, which was most likely due to contamination during sampling and HPLC analysis. The average relative mole fractions of the carbonyl compounds measured by the DNPH me~Ihodwere 37% CH20, 14% CHaCHO, and 49% CHaCOCH a. Mixing ratios of formaldehyde (measured fluorometrically) ranged from 0.2 to 2.8 ppb. They were highest during the daytime when polluted air from the Rhine valley was advected. Lower mixing ratios were usually found during the night or early morning when the site was situated above the nocturnal inversion. Comparison of the diurnal variation of CH20 with the concentrations of O a, NO, NO2, NOy, and PAN and with meteorological data indicated that both transport and photochemical formation contributed to the observed CH20 levels. Budget for CH20 was estimated using covariance of CHzO with NO, and PAN. Copyright © 1996 Elsevier Science Ltd Key word index: Formaldehyde, acetaldehyde, acetone, troposphere, photochemistry, transport, emission.

1. INTRODUCTION

Carbonyl compoun&~ are the most important intermediate species in the oxidation of hydrocarbons (Atkinson, 1990; Altshuller, 1991a, b). Their photolysis is considered to be a m~Ljorsource of free radicals in the moderately and highly polluted atmosphere (Kleinman, 1991). In summertime, formaldehyde, which is photolysed at longer wavelengths than ozone, can dominate the radical production during early morning and late afternoon. Carbonyl compounds can even become the predominant radical source in the polluted planetary bc~undary layer during winter and early spring (Kleinman, 1991). Despite their high reactivity, carbonyl compounds can comprise a significant fraction of N M H C in rural air (Schubert et al., 1988; Fehsenfeld et al., 1992) and hence, can substantially influence the photochemical production of ozone... Since the radical budget in

photochemical models is very sensitive to the assumptions made about the concentrations of formaldehyde and higher aldehydes (Liu et al., 1987; Dodge, 1990), estimation of the production rates of ozone and other oxidants in differently polluted air masses requires a realistic assessment of the concentrations of aldehydes. Carbonyl compounds are amongst the few species that are both emitted directly into the atmosphere and produced in situ, e.g. during the photodegradation of organic compounds (Cleveland et al., 1977; Graedel, 1978; Grosjean et al., 1983; Carlier et al., 1986). Primary emissions of carbonyl compounds are due to anthropogenic activities, such as incomplete combustion processes, and to emissions from natural sources (Kimmerer and Kozlowski, 1882; Isidorov et al., 1985; Carlier et al, 1986; Brunke, 1988). Photochemical oxidation of volatile organic compounds (VOC) by reaction with hydroxyl radicals is

3667

3668

J. SLEMR et al.

considered to be the predominant source of carbonyl compounds in air masses that are loaded with strong anthropogenic emissions (e.g. Grosjean et al., 1983), in addition to reactions with ozone and NO3 at night (Altshuller, 1993). Chemical destruction of C H 2 0 is caused by photolysis and by reactions with OH, HO2, and NO3 (Finlayson-Pitts and Pitts, 1986). Other sinks are dry and wet deposition. In view of its lifetime of several hours during daytime, C H 2 0 can be transported over distances of several tens of km. We have measured carbonyl compounds (formaldehyde, acetaldehyde, and acetone) at the T O R (Tropospheric Ozone Research) field station Schauinsland in September 1992. The measurements were part of an intensive field campaign which was made in cooperation with T R A C T (Transport of Tropospheric Pollutants over Complex Terrain), another subproject of E U R O T R A C . The T R A C T campaign focused on the investigation of transport processes on a regional scale and on the production of photooxidants.

2. EXPERIMENTAL

The measurements were performed between 7 and 21 September 1992, at the Schauinsland observatory (48°N, 6°E, 1248m asl). The site is located on a ridge near the Schauinsland summit, approximately 11 km southeast of Freiburg in the Black Forest. The surrounding hills are partially covered by coniferous/deciduous forests, the ground vegetation is dominated by grass. The distribution of local wind directions at the site is strongly influenced by local orography. As has been discussed by Volz-Thomas et al. (1991, 1993) and Geifl and Volz-Thomas (1993), the locally observed wind directions can be segregated into three sectors: (i) Northwesterly winds are mainly observed in summer during daytime under the persistence of high pressure. The wind system is driven by up-slope thermals, which advect freshly polluted air masses from the Rhine valley. (ii) Southwesterly winds are, for most of the time, representative of clean continental air without any recent input of pollutants. (iii) Southeasterly winds usually advect photochemically aged air masses which have been subject to anthropogenic emissions. In addition to anthropogenic sources, the air masses advected to the site are influenced by hydrocarbon emissions from the vegetation in the vicinity of Schauinsland (Kley et a/., 1993; Klemp et al., 1993). The TOR station is equipped with instruments to measure continuously 03 (UV absorption), PAN (GC/chemiluminescence),H202 (enzyme catalysed fluorescence), CO (GC/HgO), NO (chemiluminescence), NO~ (photolytic converter/ chemiluminescence), NOy (gold converter/chemiluminescence), NMHC (GC/FID; PID), and meteorological data. A detailed description of the site and the analytical systems is given by Geil3 and Volz-Thomas (1993) and Volz-Thomas (1994). Continuous determination of formaldehyde

Formaldehyde was measured using the enzymatic/fluorometric technique described by Lazrus et al. (1988). In the first step, formaldehyde is continuously stripped from air into acidified water. In the aqueous phase, CH20 reduces the coenzyme nicotinamide adenine dinuclcotide (NAD +) to

NADH, which is detected fluorometrically. The reaction is catalysed by formaldehyde dehydrogenase. As observed by Lazrus et al. (1988), trace gases like 03, Ct~24 hydrocarbons, NO, NO + 0 3, methylhydroperoxide, H20 2 and CHaCHO do not interfere in the enzymatic/fluorometric method. A possible interference by SO 2 was eliminated by adding H202 to the scrubbing solution (Lazrus et al., 1988). The formaldehyde mixing ratio [CH20] g in air is related to the measured molarity in the liquid [CH20] ~ by:

[CH20]g = 22.4 x [CH20]I

X Vl x Vg i~- 1 .

(1)

Possible errors originate from uncertainties in [CH20]1, in the flow of the stripping solution, V1 [:min-1], in the air sample flow, Vs [standard :min-1], and in the collection efficiency, r/. The collection efficiency was 91% + 1% at 3ppb and 25°C, as determined by using a standard gas mixture from a permeation source. The observed collection efficiency is close to the value of 95%, calculated from Henry's law expression (Lazrus et al., 1988, and a reference therein). The collection efficiency varies only by about 1% in changing from 0.10 ppb CH20 to 29.6 ppb CH20 (Lazrus et al., 1988). The instrument was routinely calibrated with aqueous standards prepared by serial dilution of a stock solution, which was standardised by oxime titration (Organikum, 1981). The calibration curve was found to be linear from 0.1ppb to at least 46ppb. [cn20]l was determined with a precision of about + 5%, including errors in the calibration procedure (correlation coefficients of calibration curves were 0.98-0.99) and the standardisation of the CH20 stock solution. The deviations of V1and Vz were determined to + 1.6% and _+ 1.3%, respectively. An additional systematic error arises from the decreasing activity of the formaldehyde dehydrogenase (FDH) in the time between replenishments. During laboratory tests, the enzyme activity was found to decrease at a rate of approximately 1% per hour. It was accounted for by calibrating the instrument each day prior to and after the measurement period. We estimate the total uncertainty in the enzymatic/fluorometric technique to be about 10%. Under optimum operational conditions, the instrument reached a detection limit of 120ppt; the delay time was 15 rain and the time resolution i min. In the course of the field campaign, precipitation of formaldehyde dehydrogenase occurred, which had not been observed before. For this reason, the flow system gradually plugged up and affected the instrument baseline. Cleaning and additional calibration was necessary, leading to frequent interruptions in the measurements. D N P H method

Additional analyses of CH20 were made by derivatization of DNPH-coated silica gel cartridges followed by HPLC analysis of the hydrazone after elution from the cartridges. This technique also allows the determination of the concentrations of CH3CHO, CH3COCH a and higher homologues to be made. Details of the analytical procedure used have been described previously (Slemr, 1991). As the hydrazones of the carbonyl compounds investigated in this study are stable for at least 2 weeks, the cartridges were shipped to IFU for HPLC analysis. In order to eliminate interference by ozone, potassium iodide filters upstream of the DNPH cartridges were used during sampfing. In order to determine artefacts introduced during storage and shipping, cartridges for blank samples were handled as those for air samples. The technique achieves a detection limit of 0.25 ppb for sampling periods of I h, which is mainly determined by the magnitude of the blank. The mixing ratio of formaldehyde [CH20]s resulting from the HPLC analysis is given by [CH20]g = (cs - Cb)X Ve X M - 1 x 22.4 x V~-1 .

(2)

Temporal variations in formaldehyde, acetaldehyde and acetone

3

4

3669

Diurnal profiles of CH20, CH3CHO and CH3COCH3, measured with the D N P H method, are presented in Fig. 2. They were the most abundant carbonyl compounds measured at Schauinsland. Higher homologues were observed at much lower concentrations (Fig. 1). The most abundant carbonyls were speciated as 37% CH20, 14% CH3CHO, and 49% CH3COCHa. All three carbonyl compounds exhibited similar diurnal variations. The good correlation between the species was confirmed by linear regression analysis, which gave the following results: [CH3CHO] = 0.42 [CH20] + 0.01 (R = 0.83; n = 29)

Fig. 1. HPLC chromatogram of a sample collected at Schauinsland on 18 September 1992. 1: FormaldehydeDNPH; 2: acetaldehyde-DNPH, 3: acetone-DNPH, 4: methyl butyraldehyde-DNPH (internal standard). The samplinlg time was 40 min. Uncertainties in ICH20]0 result from errors in the determination of the CH20-DNPH concentrations in the solutions eluted from the sample and blank cartridges (c~ and cb, [gml-1]), from errors in the measurement of the eluted volume lie [ml] and of tlhe sample volume Vg [standard 1]. (M is the molecular wei~:ht of CH20-DNPH.) The concentrations, cs and cb, were determined using HPLC. A typical chromatogram is shown :in Fig. 1. The reproducibility of the HPLC analysis is about _+ 3%. Calibration was performed by adding of an internal standard to each sample. The reproducibility of this p:rocedure was found to be + 3%. Linear regression lines used as calibration curves showed correlation coefficients of > 0.98. Vs was measured by an automated sequential sampler with an accuracy of + 5%. Under laboratory conditions, the estimated uncertainty of the entire technique is better than 15%. Under field conditions, deviations between measurements made simultaneously by the DNPH method and tunable diode laser spectroscopy (TDLAS) were found to average approximately 30% (Shepson et al., 1991).

RESULTS

The temporal variations of C H 2 0 and other carbonyl compounds, along with those of 03, PAN, NO, NOy, global radiation, wind speed and wind direction observed during the c~Lmpaign, are depicted in Fig. 2. The measurement period (11-21 September 1992) was characterised by warm weather and clear skies, with daytime maximum temperatures of 13-22°C. Precipitation was not obserw;d during the entire campaign. Mixing ratios of formaldehyde, determined by the fluorometric method, ranged from 0.2 to 2.8 ppb. The highest levels of C H 2 0 were generally observed during daytime and the lowest during night. The lower concentrations found at Schauinsland were in the range of C H 2 0 concentrations observed in marine air over the Atlantic by Lowe and Schmidt (1983; 0.1-0.3 ppb) and by Schubert et al. (1988; 0.15- 0.61 ppb). This is in line with earlier findin~;s that Schauinsland experiences very clean air under conditions of southwesterly winds (Volz-Thomas et al., 1991, 1993).

[CH3COCH3] = 1.08 [CH20] + 0.59 (R = 0.91; n = 29) [CHaCOCH3] = 2.15 [CH3CHO] + 0.88 (R = 0.91; n = 29) The C H 2 0 concentrations derived from the fluorometric technique yielded even higher correlation coefficients when they were used for regression analysis. The intercepts for CH3COCH3 are significantly different from zero, e.g. between 0.6 and 0.9ppb, which is consistent with a higher background concentration of CH3COCH3 because of its longer lifetime (Chatfield et al., 1987). Formaldehyde concentrations measured by the fluorometric method reflect, in general, the same temporal variations as those recorded by the D N P H method (Fig. 2); however, they are systematically lower. The discrepancy found in the measurements made at Schauinsland is larger than that to be expected from the uncertainties in both methods. A linear regression (Fig. 3(A)) revealed a slope of 0.97 [CH20"]DNI'rl/[CH20"]fl .... which is not significantly different from unity, and an offset of 0.84 ppb in the D N P H measurements. One possible reason for the discrepancy could be that PTFE inlet lines had been used for the D N P H samples. PTFE is more porous and permeable than the PFA tubes used for the fluorometric analysis. Another possibility was contamination in the laboratory during elution of hydrazons from cartridges and during HPLC analyses. Follow-up intercomparison measurements, carried out with an improved D N P H method and the enzymatic/fluorometric method showed good agreement for two standard gas mixtures obtained from a permeation source (DNPH: 10.0 and 5.0ppb; fluorometry: 10.1 and 5.5ppb) and for ambient air samples (Fig. 3(B)). These intercomparison measurements suggest that the D N P H measurements made at Schauinsland in 1992 were indeed subject to contamination. This applies to C H 2 0 and CHaCOCH3, since both species are present in laboratory air in concentrations of up to 10 ppb. On the other hand, acetaldehyde concentrations in laboratory air were found to be below 1 ppb which is reflected in substantially lower blanks.

3670

J. SLEMR et al.

2

CI-I20

-

e~ l0 4

-

CH3COCH3

2-

/

? CH20

~,

cmcHo

0

i !:

15 NO2

1

10

NO

/i

0 80

A,,N

60

OZONE

40

20

PAN * 10

0 80

eq 60

GLOBAL RADIATION

~ 4o ~ 2o 400 WIND DIRECTION

O O

o~ 300 -~ 200 o

100

WIND SPEED * 10

0 10.9.

11.9.

12.9.

16.9.

17.9.

18.9.

19.9.

20.9.

21.9.

date Fig. 2. Temporal variations of CH20 (fluorometric method, 5 rain averages); CH20, CH3CHO, and CH3COCH3 (DNPH method); NO and NO2; O3 and PAN; global radiation; wind speed and direction.

Temporal variations in formaldehyde, acetaldehyde and acetone

3671

4. DISCUSSION

4

==3 13. Z

~2 JO O.

O 1 ¢N -1o 0 0

1 2 3 C H 2 0 [ppb], fluorometric method

4

4 t-

a.

04

-IO

~ Z m i |

0 0

1

J

i

2

3

4

C H 2 0 [pplal, fluorometrie method

Fig. 3. Correlation bel:ween CH20 concentrations obtained by the fluorornetric and DNPH techniques. (A) Measurements during the TRACT campaign, (B) intercomparison measurements made after the campaign at IFU in ambient air.

The results for carbonyl compounds obtained at Schauinsland in September 1992 are summarized in Table 1 together with results obtained in other studies. Despite the possible systematic errors in acetone concentrations, the concentrations and the partitioning of the carbonyl compounds measured by us compare quite well with the results from an earlier study at Schauinsland and with studies in other rural areas (Table 1). Similarities in the temporal behaviour of C H 2 0 and CH3CHO concentrations were also observed by other authors. Strongly correlated temporal variations of C H 2 0 and CH3CHO (R = 0.99) were recorded at a remote alpine site at 1200 m asl (Brand, Vorarlberg, Austria) in December 1984 and January 1985 (Seiler, personal communication). Because of the low photochemical activity during winter, most of the variations in aldehyde concentrations observed in this study were most likely caused by meteorological processes. However, very similar trends in C H 2 0 and CHaCHO concentrations were also observed during time periods with strong photochemical activity, for example in urban regions of California during August (Fung and Wright, 1990) and September (Grosjean, 1988). There, photochemical formation and destruction processes did not seem to influence the C H 2 0 / C H a C H O ratio substantially during the measurement periods. The competing influence of physical and chemical processes on the C H 2 0 levels at Schauinsland was most evident on 17 September 1992, when polluted air masses from the Rhine valley reached the site after break-down of the nocturnal inversion. Our discussion of the concentrations of formaldehyde is

Table l. Summary of carbonyl concentrations at Schauinsland and other rural sites Compound

CH20 [ppb]

CH3CHO [ppb]

CH3COCH3 [ppb]

1.0a 2.3a 0.4" 22a

0.7 1.8 0.1 35

2.6 4.8 0.2 37

1.7

0.5

-

1.8

0.6

1.6

1.3

0.6

1.5

£chauinsland, September 1992

Mean concentration Maximum concentration Minimum concentration Number of measurements Schauinsland, June 1984b

Mean concentration Ontario, July/August 1988c

Mean concentration Wank, October 1991 d

Mean concentration

"Averages of fluorometric values over continuous measurements during r)NPH sampling intervals. b Schubert et al. (1988). c Shepson et al. (1991). d Slemr et al. (1993).

3672

J. SLEMR et al.

confined to this episode, because 17 September was a day with well defined transport processes during a thermal upslope flow (Kramp et al., 1995). Analyses of the remaining days were not feasible mainly due to the complexity caused by frequent changes of air masses. Temporal variations of the trace gases and meteorological parameters observed on 17 September are plotted in detail in Fig. 4. At night, the site was shielded from the pollution sources in the Rhine valley by an inversion layer. After the break-up of the nocturnal inversion at about 11 a.m., polluted air with a high content of nitrogen oxides was advected from the northwest. The temporal variation of C H 2 0 during this period was very similar to that of NOx and NO r Around 1 p.m., the PBL was fully developed (Kramp et al., 1995). Thereafter, NO, NO2 and NOy levels decreased because of chemical destruction and dry deposition during transport. The C H 2 0 concentration, however, continued to increase in parallel with those of PAN and Oa and reached a maximum of 2.8 ppb at about 5 p.m. The similarities in the temporal increase of 03, PAN and C H 2 0 suggest that photochemical production in the air masses advected to the site was the predominating source of formaldehyde during that time span. Precursors of C H 2 0 and PAN may have been both of anthropogenic and biogenic origin. As was observed by Kramp et al. (1994, 1995), the anthropogenic hydrocarbons exhibited similar diurnal variations to that of NOx and NOy on 17 September, whereas isoprene and 1-hexene concentrations remained almost constant or increased slightly during the afternoon, indicating the potential influence of natural VOC emissions in the vicinity of the station. The decrease in concentrations of formaldehyde and the other primary and secondary pollutants after 5 p.m. was associated with the termination of up-slope flow, as is evident from the change in wind direction. The diurnal variation of C H 2 0 observed on 17 September resembles a combination of the diurnal variations of primary pollutants, such as NO/NO2, and secondary pollutants, such as PAN and 03. From this, it is evident, that both transport of C H 2 0 emitted in the polluted Rhine valley and photochemical formation of C H 2 0 from precursors in the air mass during transport, make significant contributions to the concentrations observed at Schauinsland. NOx may be used as a tracer for transport and PAN as a tracer for photochemical production. The individual contributions can be quantified empirically from the covariance of C H 2 0 with NO~ and PAN according to equation (3),

6

3t

18 is

Noy

~2 ~ 9 6 3 0 ~"

GLOBALRADIATION

t

M 20 0 3e0 ~ 3oo 4"_ ~0

~ is0 120 ~, 60 0 00:00

04:00

0~1:00

12:00

16:00

20:00

Fig. 4. Temporal variation of 03, CH20, PAN, NO. NO2, NO, global radiation, wind direction, and wind speed on 17 September 1992 at Schauinsland.

obtained from regression analysis. The results of the fit in Fig. 5 show that the simple model describes the observed variation of the C H 2 0 concentration quite well, including the small scale features at the onset of the up-slope flow regime. The good correlation provides strong experimental evidence for the two different sources that are responsible for the observed variation of C H 2 0 at Sehauinsland during up-slope flow. The empirical equation (3) can be compared to a simple conceptual model in which the concentration [CH20]t observed at Sehaninsland is described as the composite of three terms, a background [CH20]b, a contribution from advection [CH20]a, and a contribution from photochemical production [CH20]p, according to equation (4):

[CH20]t = [-CH20]b + a x [NO.It + b x [PAN]t (3) where [ c n 2 0 ] t , [NOx]t and [PAN]t are the observed concentrations. The constants [CH20]b, a and b are

[ C H 2 0 ] t = [CH20"Ib + [ C H 2 0 ] a + [ C H 2 0 ] p .

(4)

Adopting a few simplifying assumptions, the constants a and b can be interpreted, by comparison of equation (3) with (4), as the ratio at which C H 2 0 and

Temporal variations in formaldehyde, acetaldehyde and acetone NO× were emitted fi:om the sources in the Rhine valley and the ratio at which C H 2 0 and P A N were produced photochemically in the air masses during their transport. With destruction being a first order process with respect to formaldehyde, the total losses of C H 2 0 during transport can be separated into those of advected and photochemical CH20. The terms on the right-hand side of eqt,ation (4) thus represent the net amount of C H 2 0 observed at Schauinsland from the different sources after transport. By comparing equations (3) and (4), some information may be obtained on the emission ratio between NO× and C H 2 0 (factor a) and the ratio of C H 2 0 and

2.5

1.5

t 4~ 0.5 04:00

08:00

12:00

16:00

20:00

Fig. 5. Temporal variation of CH20 on 17 September 1992. Solid line: Fluorometric measurement (5 min averages). Symbols: CH20 values calculated from the empirical relation [ ' C H 2 0 ] t = [ ' C H 2 0 ] b q- 0.049 X [NO×It + 1.35 x [PAN]t.

P A N production (factor b). The comparison can be done if lifetimes of C H 2 0 , NO×, and PAN are similar and comparable to a travelling time of the air mass from the Rhine Valley to Schauinsland. Besides chemical destruction, the effective lifetimes are influenced by mixing of advected air masses with cleaner air from aloft during transport and by dry deposition. Chemical lifetimes of CH20, NO× and P A N estimated for conditions encountered on 17 September at 13:00 as a representative point are given in Table 2. The lifetimes were estimated for three concentrations of O H radicals. Average OH concentration in the air masses transported to the site during the time period of interest were estimated by Kramp et al. (1994, 1995) to range between 5.5-8.5 x 106 molecules cm -3. These estimates were derived from the decay of hydrocarbons as a function of their reactivity and are in reasonable agreement with measurements carried out by Platt et al. (1988) in June 1984 at Schauinsland, who found OH concentrations varying between < 1.5 × 106 molecules cm-3 and 8.7 x 106 molecules cm-3. Besides the OH concentration of 6 x 106 molecules cm -3 estimated by Kramp et al. (1994, 1995) we considered also lower concentrations of 1.5 x 106 and 3 x 106 molecules c m - a to estimate chemical lifetimes for C H 2 0 and NO× (Table 2). Losses of hydrocarbons due to dilution during the growth of the mixed layer were also estimated by Kramp et al. (1994, 1995) from the decay of hydrocarbons. The authors determined an effective first order rate coefficient for losses by dilution of the polluted air in the up-slope regime with the cleaner air from aloft during growing of the planetary boundary layer. The concentrations of hydrocarbons (and CH20, NO×, PAN; see Fig. 4) above the inversion layer, measured at night and early morning at

Table 2. Estimated lifetimes of CH20, NO2, and PAN for 17 September 1992 [OH] x 106 [molecules cm- 31 Compound

Time

NO2/NO

NO2/NOx

1.5

3.0

6.0

Chemical lifetimes [h] 13:00

2.7

0.73 6.1 23.2 4.8

CH20 ]NO2 PAN

4.6 3.1 11.6 5.8 4.8 4.8

Dilution lifetimes [h] 4.6 4.6 4.6

Afternoon

Effective lifetimes [hi 13:00 CH20 NO2 PAN

2.7

3673

0.73 2.6 3.8 2.3

2.3 3.3 2.3

1.9 2.6 2.3

Notes: For estimates of the chemical lifetimes, following reactions were considered: NO×: NOz + OH:~HNO3; PAN: PANc*,CH3C(O)OO + NO2, CHaC(O)OO + N O :=>CH 3 -I- CO 2 -k NO2; C H 2 0 : C H 2 0 -I- O H ~ C H O -b H20; C H 2 0 => H 2 -I- CO, C H 2 0 => H q- CHO. Effective lifetimes resulting from dilution and chemical destruction (zcff) were calculated for each component according to: 1/'Ceff = 1/'t'diI + 1/'Cchem.

3674

J. SLEMR et al.

Schauinsland, were much smaller than those encountered in the up-slope flow. For this reason, dilution is, in good approximation, proportional to the concentrations within the air mass and can thus be treated as a first order loss process. NO~ and PAN should thus account for the losses of the associated formaldehyde fraction. Hence, the factor, a, can be used as a measure of the emission ratio between C H 2 0 and NO~ and the factor, b, as a measure of the relative amounts of CH20 and PAN produced photochemieally. From the results of Kramp et al. (1994, 1995), the dilution coefficient decreased from values around 1 x 10-4 s- 1 in the morning to about 0.6 x 10 -4 s -1 at noon and remained fairly constant thereafter. The dilution lifetime resulting from the latter dilution coefficient (4.6 h) was used to calculate the effective lifetimes given in Table 2. They would be slightly reduced by dry deposition, which is not considered in this simple model. The effective lifetimes of CH20, NO~, and PAN are comparable to the travelling time of 2.5 h (Kramp et al., 1994, 1995) and are only slightly dependent on OH concentrations. Similarities of the effective lifetimes and the travelling time indicate that the application of equations (3) and (4) is not subject to large errors. A further assumption concerns emission sources of C H 2 0 and NO~. During the transport, significant emissions of C H 2 0 and NOx should not be encountered. Biogenic emissions of CH20, in contrast to those of higher earbonyl compounds, have not been reported so far (Isidorov et al., 1985; Fehsenfeld et al., 1992). Our records for C H 2 0 on 11 and 21 September, days with aged air masses advected from the southeast, showed low mixing ratios. Mean values of 0.7-0.8 ppb correspond to a background continental concentration (see below). Therefore, primary biogenie sources do not seem to contribute significantly to the levels of C H 2 0 observed on 17 September. Likewise, the influence of rural NOx is rather small compared to the anthropogenic emissions from the Rhine valley. This is born out by the low NO~, mixing ratios observed on days with southeasterly winds. For instance, on 21 September the mixing ratio of NO~ stayed mostly below 1 ppb, at levels relatively low in comparison with NO~ mixing ratios up to 17 ppb reached on 17 September. The simple empirical model provides estimates of the constants [CH20]b, a, and b. This model, describing temporal variations of C H 2 0 as the result of meteorological effects and photochemical formation during transport, is constrained to conditions found on Schauinsland on 17 September and does not fully account for all mechanisms involved. Nevertheless, the background concentration, [CH20]b = 0.7 _+ 0.2 ppb, is in good agreement with the lower concentrations measured during the campaign at night and during southerly winds and is only a little higher than the concentrations found in clean air in northern Germany (Platt and Perner, 1980) and on the west coast of Ireland (Lowe et al., 1981; Lowe and Schmidt, 1983). Measurements made at Schauins-

land in June 1984 also showed minimum mixing ratios of 0.6 ppb (Schubert et al., 1988). The constant a, which should reflect the ratio between C H 2 0 and NO~ emissions, mainly automobile exhaust, is 0.05 + 0.004. This value exceeds the ratio CH20/NOx ~ 0.02, as measured by Gregori et al. (1989)in a road tunnel (Tauerntunnel, Austria). The discrepancy can in part be explained by operating conditions in the tunnel (cruise), which do not represent the traffic generally, and in part by possible additional sources of formaldehyde in the Rhine valley (e.g. other combustion processes). Considering these large uncertainties usual for emission inventories, the results of the tunnel study and our estimation agree reasonably well. Recent estimates of emission ratios in eastern North America (Li et al., 1994) also led to a value CHzOflqo v --0.05. (In fresh emissions NO v, defined as NO v = NOx + HNO3 + PAN, is very close to NOx, as HNO3 and PAN are negligibly small.) The result obtained for b (1.35 + 0.07) suggests that about 1.4 molecules of C H 2 0 are formed for each PAN molecule in the hydrocarbon mix advected to Schauinsland from the Rhine valley. This ratio is not a universal one but depends on the mix of anthropogenie and biogenic hydrocarbons as well as the NOx to hydrocarbon ratio. However, since NOx and hydrocarbon measurements were made during the campaign, both in the valley and at Schauinsland (Kramp et al., 1994, 1995), the results obtained from our analysis can provide a critical test for the evaluation of photochemical models simulating this episode.

4. CONCLUSIONS

The mixing ratios of CH20, CH3CHO, and CHaCOCHa that were measured in September 1992 at the TOR station, Schauinsland, are similar to those observed at other rural sites. These compounds were the most abundant earbonyls and were on average apportioned as 37% CH2O, 14% CH3CHO, and 49% CH3COCH3. From the covariance of the temporal variations of CH2O, with those of NOx and PAN, the two major sources for formaldehyde at Schauinsland are derived: transport from nearby sources in the Rhine valley and in situ photochemical production during transport. An empirical model of the CHzO budget gives a good estimate of the continental background concentration of formaldehyde of 0.7 ppb. The molar ratio C H 2 0 / NOx (0.05) is higher than the ratio observed in automobile exhaust. The photochemical production was related to that of PAN and shows that in the observed air masses from the Rhine valley, approximately 1.4molecules of C H 2 0 are produced for each molecule of PAN. This relationship should provide a useful test for photochemical models simulating this episode.

Temporal variations in formaldehyde, acetaldehyde and acetone Acknowledgement--The authors wish to thank Franz Slemr for helpful discussions, Hans-Werner P~itz for the preparation of the routine data from the TOR station, Schauinsland, and Marta Kern for the careful analysis of carbonyl compounds by the DNPH technique. The development of the fluorometric formaldehyde measurement technique and the operation of the TOR station, Schauinsland, were supported as a part of the EUREKA environmental project EUROTRAC by the German Ministry for Research and Technology under grants No. 01 VQ 8928 and No. 07 EU 723 A, respectively. This article is a contribution to TOR and TRACT, both subprojects of EUROTRAC.

EEFERENCES Altshuller A. P. (1991a) Chemical reactions and transport of alkanes and their products in the troposphere. J. atmos. Chem. 12, 19~1. AltshuUer A. P. (199 lb) Estimating product yields of carboncontaining products from the atmospheric photooxidation of ambient air alkenes. J. atmos. Chem. 13, 131-154. Altshuller A. P. (1993) Production of aldehydes as primary emissions and from secondary atmospheric reactions of alkenes and alkanes during the night and early morning hours. Atmospheric Environment 27A, 21-32. Atkinson R. (1990) G-'Ls-phase tropospheric chemistry of organic compounds: a review. Atmospheric Environment 24A, 1-41. Brunke E.-G. (1988) Influence of guano beds on the tropospheric formaldehyde level in marine air at Cape Point. South African J. Sci. 84, 118-121. Carlier P., Hannachi H. and Mouvier G. (1986) The chemistry of carbonyl compounds in the atmosphere---a review. Atmospheric Environment 20, 2079-2099. Chatfield R. B., GardvLer E. P. and Calvert J. G. (1987) Sources and sinks of acetone in the troposphere: behaviour of reactive hydrocarbons and a stable product. J. geophys. Res. 92, 4201~-4216. Cleveland W. S., Graedel T. E. and Kleiner B. (1977) Urban formaldehyde: observed correlation with source emissions and photochemistry. Atmospheric Environment 11, 357-360. Dodge M. (1990) Form~ddehyde production in photochemical smog as predicted by three state-of-the-science chemical oxidant mechanisms. J. geophys. Res. 95, 3635-3648. Fehsenfeld F., Calvert J., Fall R., Goldan P., Guenther A. B., Hewitt C. N., Lamb B., Liu S., Trainer M., Westberg H. and Zimmerman P. (1992) Emissions of volatile organic compounds from vegetation and the implications for atmospheric chemistry. Global Biogeochem. Cycles 6, 389-430. Finlayson-Pitts B. J. and Pitts J. N. (1986) Atmospheric Chemistry: Fundamentals and Experimental Techniques, p. 506 ft. Wiley-Interscience, New York. Fung K. and Wright 13. (1990) Measurement of formaldehyde and acetaldehyde using 2,4-dinitrophenylhydrazineimpregnated cartridges during the carbonaceous species methods comparison study. Aerosol Sci. Technol. 12, 44-48. Geifl H. and Volz-Thomas A. (eds.) (1993) Lokale und regionale Ozonproduktion: Chemic und Transport. In Berichte des Forschunoszentrums Jiilich; No. 2764, Forschungszentrum Jiilich GmbH, Zentralbibliothek, D-52425 Jiilich. Graedel T.-E. (1978) Carbonyl compounds. In Chemical Compounds in the Atmosphere, pp. 158-209. Academic Press, New York. Gregori M., Lanzerstort~r Ch., Obeflinninger H., Puxbaum H., Biebl P., Gl~iserO. and Villinger J. (1989) Tauerntunnel--

3675

Luftschadstoffuntcrsuchung 1988. Bericht 4/89, Abteilung fiir Umweltanalytik, Institut fiir Analytische Chemic, Technische Universitiit Wien. Grosjean D., Swanson R. D. and Ellis C. (1983) Carbonyls in Los Angeles air: contributions of direct emissions and photochemistry. Sci. Total Envir. 29, 65-85. Grosjean D. (1988) Aldehydes, carboxylic acids and inorganic nitrate during NSMCS. Atmospheric Environment 22, 1637-1648. Isidorov V. A., Zcnkevich I. G. and Ioffe B. V. (1985) Volatile organic compounds in the atmosphere of forests. Atmospheric Environment 19, 1-8. Kimmerer T. W. and Kozlowski T. T. (1982) Ethylene, ethane, acetaldehyde, and ethanol production by plants under stress. Plant. Physiol. 69, 840-847. Klcinman L. I. (1991) Seasonal dependence of boundary layer peroxide concentration: the low and high NO x regimes. J. geophys. Res. 96, 20,721-20,733. Klemp D., Flocke F., Kramp F., P~itz W., Volz-Thomas A. and Kley D. (1993) Indications for biogenic sources of light olefins in the vicinity of Schauinsland/Black Forest (TOR station No. 11), Joint Workshop CEC/BIATEX of EUROTRAC, Aveiro, Portugal. In Air Pollution Research Report 47 (edited by Slanina J., Angeletti G. and Beilke S.), ISBN 2-87263-095-3, pp. 271-281. Kley D., Geiss H., Klemp D., Kramp F., Su Y. and VolzThomas A. (1993) The importance of hydrocarbon measurements. In Proc. EUROTRAC Syrup. 92 (edited by Borrell P. M. et al.), pp. 70-79. SPB Academic Publishing bv, The Hague, The Netherlands. Kramp F., Buers H. J., Flocke F., Klemp D., Kley D., P~itz H. W., Schmitz T. and Volz-Thomas A. (1994) Determination of OH concentrations from the decay of Cs-Cs hydrocarbons between Freiburg and Schauinsland: Implications for the budgets of olefins. In Proc. EUROTRAC Symp. 94 (edited by Borrell P. et al.), pp. 373-378. SPB Academic Publishing by, The Hague, The Netherlands. Kramp F., Kley D. and Volz-Thomas A. (1995) Die Rolle reaktiver Kohlenwasserstoffe bei der Photooxidantienbildung in l~indlichen Gebieten: Ein Beitrag zur Bilanzierung der photochemischen Ozonproduktion, Berichte des Forschungszentrum Jiilich, Jiilich-3050. Lazrus A. L., Fong K. L. and Lind J. A. (1988) Automated fluorometric determination of formaldehyde in air. Anal. Chem. 60, 1074-1078. Li S.-M., Anlauf K. G., Wiebe H. A. and Bottenheim J. W. (1994) Estimating primary and secondary production of HCHO in eastern North America based on gas phase measurements and principal component analysis. Geophys. Res. Lett. 21, 669-672. Liu S. C., Trainer M., Fehsenfeld F. C., Parrish D. D., Williams E. J., Fahey D. W., Hiibler G. and Murphy P. C. (1987) Ozone production in the rural troposphere and the implications for regional and global ozone distributions. J. geophys. Res. 92, 4191-4207. Lowe D. C. and Schmidt U. (1983) Formaldehyde (HCHO) measurements in the nonurban atmosphere. J. geophys. Res. 88, 10,844-10,858. Lowe D. C., Schmidt U., Ehhalt D. H., Frischkorn C. G. B. and Niirnberg H. W. (1981) Determination of formaldehyde in clean air. Envir. Sci. Technol. 15, 819-820. Organikum (1981) Organisch-chemisches Grundpraktikum, pp. 486-487. VEB Deutscher Verlag tier Wissenschaften. Platt U. and Perner D. (1980) Direct measurements of atmospheric CH20, HNO2, Oa, NO2, and SO 2 by differential optical absorption in the near UV. J. geophys. Res. 85, 7453-7458. Platt U., Rateike M., Junkermann W., Rudolph J. and Ehhalt D. H. (1988) New tropospheric OH measurements. J. geophys. Res. 93, 5159-5166. Sbepson P. B., Hastie D. R., Schiff H. I., Polizzi M., Bottenheim J. W., Anlauf K., Mackay G. I. and Karecki D. R.

3676

J. SLEMR et al.

(1991) Atmospheric concentrations and temporal variations of C1-C3 carbonyl compounds at two rural sites in Central Ontario. Atmospheric Environment 25A, 2001-2015. Schubert B., Schmidt U. and Ehhalt D. H. (1988) Untersuchungen zum Nachweis und zur Chemie yon Formaldehyd und Acetaldehyd in der unteren Troposphiire, Berichte der

Forschungsanlage Jiilich; Nr. 2257, Forschungsanlage Jiilich GmbH, Zentralbibliothek, D-52425 Jiilich. Slemr J. (1991) Determination of volatile carbonyl compounds in clean air. Fresenius J. Anal. Chem. 340, 672-677. Slemr J., Schulz S., Seemann S. and Kern M. (1993) Development of techniques for the determination of major carbonyl compounds in clean air. In Proc. E U R O T R A C Syrap. 92 (edited by Borrell P. et al.), pp. 170-172. SPB Academic Publishing by, The Hague, The Netherlands.

Volz-Thomas A. (1994) TOR Station Schauinsland. In The T O R Network, EUROTRAC special publication, ISS (edited by Cvitas T. and Kiey D.), pp. 81-94. GarmischPartenkirchen. Volz-Thomas A., Kley D., Buers H. J., Flocke F., Garthe H. J., GeiB H., Gilge S., Heft T., Houben N., Klemp D., Kramp F., Loup H., Mihelcic D., Miisgen P., P~itz H. W., Smit H. G. J. and Su Y. (1991) Local and regional ozone production: chemistry and transport. In EUROTRAC Annual Report, Part 9, TOR. Volz-Tliomas A., Flocke F., Garthe H. J., GeiB H., Gilge S., Heft T., Kley D., Klemp D., Kramp F., Mihelcic D., P~itz H. W., Schulz M. and Su Y. (1993) Photo-oxidants and precursors at Schauinsland, Black Forest. In Proc. of E U R O T R A C Symp. 92 (edited by Borrell P. et al.), pp. 98-103. SPB Academic Publishing by, The Hague, The Netherlands.