Emission of Nonchlorinated and Chlorinated Aromatics in the Flue

Publication Date (Web): February 10, 2001 ... PCDD/PCDF Ratio in the Precursor Formation Model over CuO Surface .... in MSWI Crude Gas and Ashes durin...
0 downloads 0 Views 506KB Size
Download PDF
Environ. Sci. Technol. 2001, 35, 1019-1030

Emission of Nonchlorinated and Chlorinated Aromatics in the Flue Gas of Incineration Plants during and after Transient Disturbances of Combustion Conditions: Delayed Emission Effects R . Z I M M E R M A N N , * ,†,§,⊥ M. BLUMENSTOCK,† H. J. HEGER,† K . - W . S C H R A M M , † A N D A . K E T T R U P †,‡ GSF-Forschungszentrum fu ¨ r Umwelt und Gesundheit, Institut fu ¨r O ¨ kologische Chemie, Ingolsta¨dter Landstrasse 1, D-85764 Neuherberg, Germany, Lehrstuhl fu ¨r O ¨ kologische Chemie und Umweltanalytik, Technische Universita¨t Mu ¨ nchen, D-85350 Freising/Weihenstephan, Germany, Lehrstuhl fu ¨ r Festko¨rperchemie, Universita¨t Augsburg, D-86159 Augsburg, Germany, and BIfA-Bayerisches Institut fu ¨ r Abfallforschung GmbH, Abteilung Chemie, D-86167 Augsburg, Germany

The profiles of different products of incomplete combustion (PIC) in the flue gas of a 1 MW pilot combustion facility were investigated under normal steady-state and disturbed combustion conditions. The behavior of emission profiles after disturbed combustion conditions was investigated in order to obtain a better understanding of emission memory effects. Highly time-resolved, quantitative on-line measurements of several aromatic species down to low ppbv or higher pptv concentrations were performed by a mobile resonance-enhanced multiphoton ionization time-offlight mass spectrometer. Conventional analytical methods (gas chromatography-mass spectrometry and highperformance liquid chromatography) were also applied for measurement of polycyclic aromatic hydrocarbons (PAH) and polychlorinated dibenzo-p-dioxins and -furans (PCDD/ F). The sampling point was located in the high-temperature region of the plant at the outlet of the post-combustion chamber at temperatures between 650 and 880 °C, prior to any emission reduction devices. The investigation pointed out that after a short phase of disturbed combustion conditions, e.g., due to process changes, transient puffs, or malfunctions, the composition of combustion byproducts in the flue gas can be changed drastically for a very long time (“memory emission” effect). It is suggested that carbonaceous layers, deposited on the inner walls in the high-temperature zone of the plant, might be responsible for the observed memory emission of some PAH species. Drastic changes in the profiles of the PCDD/F homologues were also observed during memory emission conditions. The PAH memory most likely is due to pyrolytic degradation of the carbonaceous layers, while the altered PCDD/F homologue pattern may be mediated by the high catalytic activity of the freshly formed deposit layers. Finally, it should be emphasized that a rich pattern of aromatic species, including PCDD/F, was found in a temperature 10.1021/es000143l CCC: $20.00 Published on Web 02/10/2001

 2001 American Chemical Society

regime well above the typical temperature window (∼300 °C) for de novo PCDD/F formation.

Introduction It is known that transient, non-steady-state conditions in the incinerator systems can induce dramatic concentration increases of products of incomplete combustion (PIC) (13). Transient fluctuations of combustion conditions thus may contribute significantly to the emission of air toxics by industrial combustion processes (4). The transiently increased formation of PIC compounds such as polycyclic aromatic hydrocarbons (PAH) is associated with soot formation and takes place in the high-temperature region of the combustion system (5, 6). Furthermore, it was observed that after disturbed combustion conditions the emissions of PIC compounds sometimes are increased for a longer period of time (7, 8). However, the investigation of highly transient emission effects requires the application of novel analytical on-line methods allowing a real-time detection of transient phenomena for the diagnosis of industrial combustion processes. Resonance-enhanced multiphoton ionization time-of-flight mass spectrometry (REMPI-TOFMS) represents such a novel technique for on-line detection of traces of molecular species (PIC compounds) in combustion flue gas. In 1996, REMPI-TOFMS laser mass spectrometry was used for the first time for on-line analysis of combustion byproducts in flue gases of an incineration plant (9, 10). Other applications of REMPI-TOFMS for monitoring purposes at different incinerators followed (11, 12). Upon the application of the REMPI-TOFMS technique for on-line monitoring of aromatic compounds in the flues gas, highly dynamic changes in the emission concentrations of different aromatic species were detected due to disturbances of the combustion process (13). It was further concluded that the observed pattern of PAH may give valuable information about the combustion process (e.g., temperature, malfunction, type of feeding, etc.) (14). The development of REMPI-TOFMS combustion process monitoring systems in our group was supported by a parallel research project that focuses on the correlation between indicator substances, which are accessible for REMPI-TOFMS on-line monitoring, and other regulated byproducts in flue gases that can be measured by conventional analytical techniques such as GC-MS (8, 15). In particular, monochlorobenzene was identified as a reliable, on-line measurable surrogate for the emission of toxic polychlorinated dibenzop-dioxins and -furans (PCDD/F) (i.e., for the international toxicity equivalent, I-TEQ) from industrial incinerators (13, 15). The work presented here focuses on the effects of transient changes in the combustion conditions on the formation of PIC compounds. In this context, REMPI-TOFMS laser mass spectrometry was applied for highly time-resolved measurement of the fluctuations in the concentrations of some PAH species and monocyclic aromatics. Furthermore, the behavior of chlorinated aromatic species such as the PCDD/F or chlorinated benzenes during and after disturbed combustion conditions is of interest. Literature results show that the formation of PCDD/F in incineration processes not only takes place in the de novo temperature region (T ∼ 300 °C) but * Corresponding author: e-mail: [email protected]; telephone: ++49 89 3187-4544; fax: ++49 89 3187-3371. † GSF-Forschungszentrum fu ¨ r Umwelt und Gesundheit. ‡ Technische Universita ¨ t Mu ¨ nchen. § Universita ¨ t Augsburg. ⊥ BIfA GmbH. VOL. 35, NO. 6, 2001 / ENVIRONMENTAL SCIENCE & TECHNOLOGY

9

1019

also can be quite significant in the high-temperature regime (T > 600 °C) at combustion processes (16, 17). Therefore, it was also tried to follow the changes of the profiles of PCDD/F (homologues) and other chlorinated aromatics at high temperatures (T > 600 °C) during and after disturbed combustion conditions at the highest possible time resolution (here: sampling time of 1 h) with conventional analytical techniques like GC-MS.

Experimental Section The presented measurements of chemical species in combustion flue gas have been performed at a waste incineration pilot plant. During the measurements, waste wood was used as feeding material. The thermal power of the plant is about 1 MW. The plant consists of a primary combustion chamber (grate system) combined with a spatially separated postcombustion chamber. After the boiler section, several flue gas cleaning devices are installed. The sampling site for the measurements of PIC was close to the outlet of the afterburner chamber with flue gas temperatures higher than 850 °C. On-line Measurements of PAH and Monocyclic Aromatics by REMPI-TOFMS Laser Mass Spectrometry. REMPITOFMS laser mass spectrometry was used for on-line measurement of organic species in the flue gas. REMPITOFMS represents a highly selective and sensitive twodimensional analytical technique, combining UV (laser) spectroscopy and mass spectrometry (18-22). The technical setup of the applied REMPI-TOFMS spectrometer and details on the sampling and calibration procedure have been published previously (9, 10). Briefly, flue gas from the combustion chamber is drawn through a quartz glass probe of 8 mm i.d. at a rate of ≈0.1 m3/h by a sampling pump. At the investigated sampling point, dust concentrations in the range of 20-150 mg/m3 were observed. A quartz microfiber plan filter (size cutoff at about 50 nm particle diameter) in a special housing heated to 250 °C was used for dust removal. At this temperature, no adsorption of PAH up to 300 m/z on the filter material is observed. According to literature results, the onset for the new formation of aromatic compounds such as PAH is well above 300 °C (6). Downstream of the filter unit for dust removal, a small fraction of the flue gas is sampled by the capillary sampling system of the timeof-flight mass spectrometer (TOFMS), employing a 2 m deactivated 320 µm i.d. fused silica capillary. The capillary passes through a flexible heating tube (heated to 250-300 °C) and directly acts as pressure restrictor between ambient pressure (sampling system) and the vacuum of the ion source of the TOFMS. The tip of the capillary leads into the center of the ion source of the TOFMS, running within a heated conical metal needle. A directed effusive molecular beam containing the sample molecules is generated from the orifice at the end of the transfer capillary (23). This molecular beam is irradiated directly underneath the orifice with UV laser pulses with a 5-10 Hz repetition rate. The power density of the pulses is in the 106 W/cm2 range, and the pulse length is 10 ns. Intermediate UV-active electronic states of compounds present in the molecular beam can be selectively excited by laser photon absorption if the photon energy (i.e., the laser wavelength) is in resonance with the respective UV transition. This resonance condition introduces the additional selectivity to the ionization process, as only previously excited molecules can subsequently be ionized by absorption of an additional laser photon. Figure 1 shows the scheme of the one-color, two-photon resonanceenhanced ionization (REMPI) process for comprehensive ionization of PAH species and nonchlorinated monocyclic aromatics. In the middle of Figure 1, the electronic term scheme of a PAH molecule is sketched (S0, ground state; S1, first excited singlet state; Sn, higher excited singlet states; IP, ionization potential). On the left, the corresponding UV 1020

9

ENVIRONMENTAL SCIENCE & TECHNOLOGY / VOL. 35, NO. 6, 2001

FIGURE 1. Schematic representation of the one-color, two-photon resonance-enhanced ionization process that was applied for comprehensive ionization of PAH species for on-line analysis of flue gas. For explanation, see text. absorption spectrum is depicted while on the right the REMPI ionization scheme is shown. With the laser wavelength of 266 or 248 nm, the ionization occurs via an intermediate state of the broad rovibronic continuum of the Sn electronic systems. Thus, most PAH are detectable while most other flue gas compounds are suppressed at this wavelengths. The symbols σ1 and σ2 denote the cross sections of the absorption and ionization step, respectively. The yield of the one-color, two-photon REMPI process is proportional to σ1σ2[A]I 2, if deactivation processes such as inter-system crossing (24) can be neglected. In the formula, [A] stands for the concentration of the analyte molecules in the laser focus, I is the laser intensity, and the product σ1σ2 represents the ideal cross section of the two-photon REMPI process (σREMPI,ideal). The ions formed in the REMPI process are accelerated by electrostatic fields into the field free drift region of the TOF mass analyzer. The high optical discrimination of potential interferents allows direct on-line analysis of target compounds or compound classes from complex samples (9). Quantification of the REMPI results is performed via the external standardization method, using a calibration gas generator (permeation and diffusion standards) that is integrated in the REMPI/TOFMS system and supplies ppbv concentrations of toluene, naphthalene, monochlorobenzene, and other compounds in air (10). For PAH, naphthalene is used as a reference compound (standard concentration: 58 ppbv ( 10%). For phenanthrene and pyrene, the REMPI cross sections (σREMPI,experimental) relative to the REMPI cross section of naphthalene are known for the wavelengths 266 and 248 nm (from GC-REMPI-TOFMS experiments; 10, 25) and are used for calibration. An estimate of the concentrations of the monomethylated naphthalenes, phenanthrenes, and pyrenes was performed based on UV spectroscopic data (26, 27). Details on the calibration procedure of the monomethylated species are given elsewhere (42). For comprehensive registration of the pattern of aromatic compounds, the fourth harmonic frequency of a Nd:YAG laser (266 nm) or the KrF excimer laser line (248 nm) was used. Most of the PAH with three or more condensed rings exhibit broad absorption band systems (S2 or Sn π* r π transitions) in this wavelength region (26, 27). Thus, the difference of the laser wavelengths is more or less irrelevant for the structure of the observed REMPI PAH pattern. However, for a quantification of the individual PAH

TABLE 1. Overview of the Presented Measurementsa description REMPI-TOF measurements conventional measurements of micropollutants

case 1

case 2

case 3

process control measures: switching from combustion to gasification on grating REMPI@266-TOFMS (Figure 2) none

process control measures: switching from combustion to gasification on grating REMPI@248-TOFMS (Figure 3) none

malfunction in PCCb due to largely oxygen-deficient combustion conditions REMPI@248-TOFMS (Figure 4) PAH (Figure 5a; Table 3), PCDD/F (Figure 5b; Table 3), PCBz (Figure 6a; Table 3), PCPh (Figure 6b; Table 3)

a Three incidents, two process switching procedures, and one malfunction have been investigated by on-line REMPI-TOFMS laser mass spectrometry and partly by conventional off-line analysis. b Post-combustion chamber.

species, specific calibration factors are required (10). In summary, the optical REMPI selectivity together with mass selectivity is sufficient to generate laser-induced mass spectra that contain chemical fingerprint information about the combustion process. Off-line Measurements of PCDD/F and PAH by GC-MS or HPLC. Isokinetic sampling was performed close to the outlet of the post-combustion chamber (PCC), using a sampling system of Stro¨hlein GmbH, Germany, with a cooled probe according to the German VDI DIN 3499 guideline (28). An impinger system with R-ethoxyethanol was used for trapping the semi-volatile PIC compounds (PCDD/F, PAH, and PCB). The sampling flow rate was 1 m3/h, and the sampling time was 1 h. Laboratory experiments suggest that a considerable formation of PCDD/F starts after a reaction time of some seconds (29) in the formation temperature window of approximately 300 °C. Thus, to minimize the formation of chlorinated aromatics in the cooled sampling probe, it is necessary to account for the residence time of a gas plug before it is cooled below 200 °C. This temperature is considered as the lower limit of de novo formation processes in the flue gas (30). The residence time of the flue gas in the sampling line at temperatures above 200 °C is well below 1 s, suggesting that sampling artifacts can be neglected. For each individual sampling, the glass inlet of the probe and all the other glassware were exchanged by new, clean pieces in order to avoid sample carryover in the sampling train. An internal 13C12-labeled standard cocktail of PCDD/F and PCB with at least one labeled congener for each degree of chlorination (with exception of the Cl1DD/DF homologues) was added to the R-ethoxyethanol before it was stirred for 24 h with a mixture of water/toluene (9/1, v/v). The cleanup procedure is described elsewhere (31). The more volatile PCBz were sampled simultaneously from a small bypass sampling stream of the sampling train. Glass cartridges (6 mm i.d.) containing 2 g of activated charcoal with 13C6-labeled internal standard (one isomer for each degree of chlorination) were used for trapping. A second “back-up” cartridge filled with 1.5 g of charcoal was used to indicate breakthrough of analytes. Sampling rates were about 80 L/h, and the temperature of the whole sampling line was held at 70 °C to avoid condensation of water in the tubes. The chlorinated benzenes were eluted with 2 mL of CS2 according to NIOSH Guideline 1501 (32). A cleanup for the PCBz was not necessary. These cartridges were also spiked with 13C6-labeled benzene, 2H8-labeled toluene, and 2H10labeled ethylbenzene. 13C6-labeled naphthalene was used as recovery standard for the PCBz and BETX analysis and added just before the measurement. The PCDD/F were analyzed using a GC-HRMS (GC: HP 5890 series II, high-resolution MS: MAT 95). The PCB, PCBz, BETX and naphthalene were analyzed by a GC-MS (GC: HP 5890 series II, low resolution MS: SSQ 7000 or GC: Varian 3400, low resolution MS: ITS 40 tracker). Details of chemical analysis have been published elsewhere (15). Recoveries

for PCDD/F, PCB, and PCBz were in the range of 60-100%. No breakthrough of the PCBz and BETX was observed upon the analysis of the backup cartridges. PAH were analyzed by HPLC (HP 1100) using a fluorescence detection system without cleanup. All flue gas concentrations are given according to normal conditions: dry air, 273 K, 1013 hPa, and 11 vol % O2.

Results and Discussion This paper deals with the dynamic behavior of micropollutants (PIC compounds) in the flue gas of an incineration plant after more or less strong deviations from optimal, steady-state combustion conditions. These deviations are due to intended process control measures or malfunctions at the investigated 1 MW incineration pilot plant. Shredded waste wood, partly contaminated with plastics and paints, was used as feed material for the pilot incineration plant during the measurements. In total three cases are considered, two of them (cases 1 and 2) being due to process control measures and one (case 3) due to a malfunction, which was observed by chance. The considered three cases and measurements are summarized in Table 1. Disturbances of Combustion Conditions Due to Process Control Measures. The first two cases (REMPI results in Figures 2 and 3) show the effects of process control measures that are causing unstable combustion conditions for a short time. In detail, the air supply to the grate (primary chamber) was reduced in order to achieve a change from a combustion process to a gasification process in the primary chamber. However, it was tried to increase the air supply in the PCC (secondary chamber) simultaneously in order to keep the overall λ-value (ratio of supplied oxygen to amount of oxygen required for stoichiometric combustion of the fuel) above 1 (over-stoichiometric). In Figure 2, eight concentration-to-time traces (naphthalene, methylnaphthalene, phenanthrene, methylphenanthrene, pyrene/fluoranthene, methylpyrene/methylfluoranthene, the four-ring PAH at 228 m/z, and the five-ring PAH at 252 m/z) recorded with a REMPI laser wavelength of 266 nm during a process control measure (case 1) are shown. The beginning of the process manipulation procedure at t ∼ 200 s is marked by an arrow in Figure 2. During the control measure, combustion conditions are fluctuating, leading to transient peaking of different PAH species. In the beginning (t ) 200-450 s) four emission peaks each with a full width at half-maximum (fwhm) below 60 s are observed. This behavior is most pronounced for naphthalene. The monocyclic aromatics (benzene, etc. not shown in Figure 2) however exhibit a similar but even more drastic change with respect to naphthalene. It is remarkable that, even for semivolatile five-ring PAH with mass 252 m/z, the four emission peaks do not show time broadening or retention behavior with respect to benzene or naphthalene. This fact also VOL. 35, NO. 6, 2001 / ENVIRONMENTAL SCIENCE & TECHNOLOGY

9

1021

FIGURE 2. REMPI-TOFMS at 266 nm measurement of PAH (time-to-concentration profiles) during a process switch from combustion to gasificaton conditions on the grate (case 1). Phases of “direct emission” and “memory emission” of aromatics are indicated. The vertical arrow indicates the beginning of the control measures. confirms that both the sampling probe system and the sample inlet system of the laser mass spectrometer do not cause significant internal memories due to adsorption/desorption processes. At later times, some species show broader emission features (starting from approximately t ∼ 350 s in Figure 2), while others do not. This is most obvious when comparing the phenanthrene trace with the trace of pyrene in Figure 2. While pyrene exhibits a second sequence of sharp, transient emission peaks (t ) 650-950 s on time axis, marked as “2. direct emission”), phenanthrene shows a totally different behavior. The first sequence of emission peaks merges with a subsequent intense and broad emission maximum. Comparable but weaker emission features are observed for the methylphenanthrenes and the five-ring PAH. Figure 3 shows a similar process control incident (case 2) as presented in Figure 2 (case 1), observed by REMPI with 1022

9

ENVIRONMENTAL SCIENCE & TECHNOLOGY / VOL. 35, NO. 6, 2001

a laser wavelength of 248 nm. However, the deviation from optimal steady-state combustion conditions in case 2 was larger. An additional fossil fuel burner was fired to avoid a temperature decrease in the PCC. The naphthalene, methylnaphthalene, phenanthrene, methylphenanthrene, and pyrene/fluoranthene traces are depicted. The naphthalene trace exhibits some transient emission peaks of low intensity at the beginning (see arrow), which are due to the initiation of the process control measures. Shortly afterward, a steep increase in the naphthalene concentration occurs. Similar to the first case, the changes in the benzene and toluene traces (not shown here) track the changes in the naphthalene trace but at a larger magnitude. The phenanthrene concentration increases with a delay of some hundred seconds, the signal of the methylphenanthrenes and of pyrene/fluoranthrene start to increase after an even longer delay.

FIGURE 3. REMPI-TOFMS at 248 nm measurement of PAH (case 2). during a similar process switch from combustion to gasificaton conditions on the grating as shown in Figure 2. Phases of increased ”direct emission” and “memory emission” of aromatic compounds are indicated. In Figure 4, two REMPI at 266 nm mass spectra from case 1 are shown (see also Figure 2). The upper mass spectrum is recorded during disturbed combustion conditions (at approximately 300 s in Figure 2), and the lower one was recorded some time after the disturbances (at approximately 600 s in Figure 2). The “disturbed” spectrum is characterized by large signals of benzene (78 m/z) and toluene (92 m/z) as well as by the occurrence of aromatic substances with localized multiple bonds as phenylacetylene (102 m/z), indene (116 m/z), and acenaphthylene (152 m/z). The latter compounds are typical intermediates formed upon the building up of PAH structures in the flame from aliphatic C2 and C3 species (6, 33) and have also been detected in different flame sampling experiments (34, 35). The lower spectrum in Figure 4 was recorded after the disturbed combustion conditions. The spectrum reflects the sustained emission of larger PAH as observed in Figures 2 and 3 (and also in case 3, see below). In conclusion, the emission of PAH and other aromatics can be classified into two regimes: (i) direct emission phase of aromatics and PAH (during disturbed combustion conditions) (ii) PAH memory emission phase (after disturbed combustion conditions). The transient, direct emission phase of PAH is directly associated with disturbed combustion conditions, increased

CO levels, and occurrence of intermediates from the molecular growth process of aliphatic molecular species to PAH, like phenylacetylene (see Figure 4). During the memory emission phase, some PAH species show increased concentrations, while no increased signals of the molecular growth intermediates or BTX compounds are observed. However, cases 1 and 2 represent relatively minor incidents, induced by process control measures, and the time scale of the memory emission observed by REMPI-TOFMS is in the order of some 10 min. During operation of industrial incineration facilities, far more drastic disturbances occur, e.g., during barrel uptake at rotary kiln hazardous waste incinerators (12) or due to malfunctions and control problems. In the following, the effect of a drastic disturbance due to a malfunction in the PCC (case 3) is considered. Drastic Disturbance of Combustion Conditions Due to a Malfunction. Case 3 (Tables 1-3) was due to a malfunction in the post-combustion system (observed by chance) and represents an example for a very pronounced disturbance of the combustion conditions. This malfunction was due to largely substoichiometric combustion conditions (oxygen deficiency) in the PCC. In Table 2, flue gas properties (temperatures; concentrations of O2, CO2, CO, and NO2) are given for normal combustion conditions, for the malfunction, and for 1 h after the malfunction. For this case in addition to REMPI-TOFMS results (Figure 5), conventional analytical VOL. 35, NO. 6, 2001 / ENVIRONMENTAL SCIENCE & TECHNOLOGY

9

1023

FIGURE 4. Two on-line recorded REMPI-TOFMS at 266 nm mass spectra of aromatic compounds from flue gas during a process control phase (switching from combustion to gasification conditions on the grate, case 1). The REMPI mass spectra were recorded during substoichiometric combustion conditions. (t ∼ 300 s. in Figure 2), showing a “direct emission” pattern of aromatics (top) and after the disturbances (t ∼ 600 s in Figure 2), showing the ”memory emission” pattern of aromatics (bottom).

TABLE 2. Measured Flue Gas Parametersa for Case 3, Which Represents a Malfunction in the Post-Combustion Region Due to Largely Oxygen-Deficient Conditionsb flue gas properties

no. 1

normal operation no. 2 no. 3 average

Ttop PCC (°C) 1060 1100 1050 Tsampling point (°C) 675 670 900 O2 (%) 10 8 4 CO2 (%) 11 12 16 CO (ppm) 2 0 27 NO2 (mg/m3) 409 323 357

1070 748 7 13 10 363

malfunction after 1 h 1050 775 8 10 227 262

720 740 5 11 30 297

a Temperature at the top of the PCC, temperature at the tip of the sampling probe, oxygen concentration, carbon dioxide concentration, carbon monoxide concentration, and nitrogen dioxide equivalent concentration. b In columns 2-4, values for normal combustion conditions (steady-state) are given (nos. 1-3 and 1-h average, respectively). Column 5 gives the averages of the values from columns 2-4 (normal combustion conditions). The last two columns show the values obtained during and 1 h after the malfunction.

results (GC-MS, HPLC) for nonchlorinated aromatic compounds [benzene, toluene, and xylenes (BTX)] and polycyclic aromatic hydrocarbons (PAH) as well as for polychlorinated PCDD/F, polychlorinated benzenes, and polychlorinated phenols (PCBz and PCPh) are also given (Figures 6 and 7; Table 3). The on-line REMPI-TOFMS measurement sequence shown in Figure 5 was started directly after the end of the 1024

9

ENVIRONMENTAL SCIENCE & TECHNOLOGY / VOL. 35, NO. 6, 2001

active phase of the malfunction. The REMPI measurement, however, shows the dynamics of the sustained emission (PAH memory phase) solely. The signals of the monocyclic aromatics (benzene and toluene) were dominant during the malfunction and declined quickly after the end of the malfunction. In contrast, the naphthalene concentration reduces slower. The concentration of the larger PAH starts to increase after the end of the malfunction, with phenanthrene exhibiting the steepest rising flank. Phenanthrene is followed by the methylphenanthrene isomers, pyrene/ fluoranthene, and methylpyrene/-fluoranthene. In conclusion, the REMPI intensity-to-time curves of the postmalfunction phase fit well to the observations and conclusions drawn from the first two cases. In case 3, five sequential conventional measurements with a sampling time of 1 h each were performed (with exception of the measurement during the malfunction, which lasted 45 min). The time for changing the sampling glassware between the measurements was about 30 min. The first three measurements were done under normal steady-state combustion conditions (prior to the malfunction). The whole active phase of the malfunction happened in the last third of the sampling time (45 min) of the of the forth sampling period. The fifth measurement, which started about 30 min after the end of the active phase of the malfunction, shows the recovery phase after the malfunction. The results for benzene, toluene, different PAH, and PCDD/F are given in Figure 6 as bar graph plot with a logarithmic concentration scale. For each compound, three bars are given, showing (from left to right) the concentration of the analytes prior, during, and after the malfunction period. The respective first bars represent the average concentration over the three measurements performed at normal combustion conditions (standard deviation is indicated). Benzene, toluene, and naphthalene, which are the most prominent aromatic species measured in the flue gas, shows the most dramatic increase during the malfunction. In the case of benzene, an increase of more than 3 orders of magnitude was observed. The concentrations of many larger PAH like anthracene, fluoranthene, or chrysene also increase by more than 1 order of magnitude during the malfunction. One hour after the malfunction, all depicted PAH with exception of acenaphthylene and benz[a]anthracene still exhibit significantly increased concentrations as compared to the normal combustion conditions prior to the malfunction while the concentrations of phenanthrene and fluoranthrene have increased even further. In summary, the conventional measurements confirm the PAH memory effect that was observed firstly by the on-line REMPI-TOFMS measurements. The analytical results on the PCDD/F are shown in Figure 6b. The PCDD/F measurements are in accordance with recent results on high-temperature formation of PCDD/F in combustion systems (i.e., PCDD/F formation above the commonly described formation temperature window of about 250-450 °C) (36). PCDD/F formation at temperatures of 600-1000 °C have been reported (37-39). The mono- to octachlorinated PCDD/F homologues and the concentrations of the nonchlorinated dibenzodioxin and dibenzofuran molecules were analyzed. The observed PCDD/F homologue profiles recorded under normal combustion conditions correspond quite well with literature results (40). The total amount of PCDD/F in the flue gas is relatively low ( 1000 K), they are supposed to be of high chemical activity. Elemental carbon has been proven to be highly

catalytically active for a variety of different reactions (48) in a wide temperature range. Furthermore, it was observed in laboratory experiments that PAH themselves, e.g., adsorbed on graphite surfaces, can be precursors for PCDF formation (46). For catalysis of chlorination/dechlorination reactions, metal species that are also incorporated into the carbonaceous layers may be important. At such high temperatures, metal chlorides are expected to be quite volatile and thus can readily be included in forming deposits. Experiments at test furnaces showed that PCDD/F memory emissions are observable from copper-containing soot surfaces at temperatures of 600 K (7). The observed dramatic changes within the PCDD/F homologue pattern can be explained by surface-catalyzed chlorination/dechlorination reactions. While the increase of the low molecular weight PCDD/F homologues (Cl1-3DD/F) in this scenario is due to chlorination processes of the nonchlorinated dibenzodioxin and dibenzofuran molecules, which are formed in high concentrations during and after the malfunction, the depletion of the high-chlorinated homologues (Cl5-8DD/F) is due to dechlorination processes. Within the investigated chloroaromatic ensembles (polychlorinated benzenes, polychlorinated phenols, polychlorinated dibenzodioxins, and polychlorinated dibenzofurans), the observed homologue pattern changes due to chlorination/dechlorination effects up the malfunction is by far most intense for the PCDD/F. In the substance class of the polychlorinated phenols, only monochlorinated phenols are significantly increased affter the malfunction. The PCBz patterns also do not show significant influence of chlorination/dechlorination processes, except those of slightly increased MCBz and 1,2-DCBz concentrations after the malfunction. At normal conditions, the concentrations of the PCDD/F homologue groups are in the 0.1-1 ng/m3 region, while the concentrations of the single congeners of the chlorinated benzenes or chlorinated phenols range from 10 to 100 ng/m3. So the observed changes in the PCDD/F homologue pattern are real trace effects, the compound classes of higher abundance being not or only slightly affected. An important issue for further investigations is the longVOL. 35, NO. 6, 2001 / ENVIRONMENTAL SCIENCE & TECHNOLOGY

9

1029

term development of the emission patterns of the PCDD/F after malfunctions. In conclusion, the presented combined REMPI-TOFMS and conventional analytical study show that memory effects after disturbed combustion conditions are occurring even in the high-temperature zone of incineration plants. In particular, the PAH and the PCDD/F are affected (Figures 2 and 6). The results encourage further investigation in the field of the emission dynamics of pollutants from combustion processes including a risk assessment of emissions from industrial combustion processes caused by transient emission phenomena like transient puffs or emission memories. The REMPI-TOFMS technique has been proven as a versatile tool for on-line detection of both (i) transient, gasphase based direct emission of aromatic species (Figure 4), occurring during disturbed combustion conditions, as well as (ii) surface-assisted memory emission effects of PAH with slower changing rate (Figures 2, 3, and 5). It thus may be used for development of more adequate operational plans for already running plants in order to reduce the emissions. In this context, it should be investigated whether the time course of the memory emission effects can be influenced by control measures. To further investigate the memory emission effect and elucidate the nature of the surface deposits as well as possible emission control measures, a laboratory-scale experiment consisting of a furnace and an electrical heated flue gas channel is under construction. A future aim would be the use of an on-line recorded REMPI-TOFMS signal as input for an intelligent feedback process control system, which may allow fast corrective action upon emission transients caused by disturbed combustion conditions or the subsequent memory effects. For example, an enforced removal of the deposits by increased oxygen supply and temperature or by special knocking devices may reduce the overall emission of air toxics during memory emission phases.

Acknowledgments The authors thank B. Henkelman and C. Klimm for help with the conventional analysis. Financial support from the Deutsche Umweltstiftung Umwelt, Osnabru ¨ ck, Germany (Projects 04778 and 12447) is gratefully acknowledged. We thank Max Buchner Stiftung and DECHEMA eV, Frankfurt, for personal scholarships (R.Z. and H.J.H.). We thank Dr. U. Boesl for continuous interest and stimulating discussions.

Literature Cited (1) De Fre, R.; Reyman, T. Chemosphere 1989, 19, 331-336. (2) Halonen, I.; Tuppurainen, K.; Ruuskanen, J. Chemosphere 1997, 34, 2649-2662. (3) Hart, J. R. Chemosphere 2001, 42, 559-569. (4) Wendt, J. O. L. In 25th Symposium (International) on Combustion; The Combustion Institute: Pittsburgh, 1994; pp 277-298. (5) Ballschmiter, K.; Bacher, R. Dioxine: Chemie, Analytik, Vorkommen, Umweltverhalten und Toxikologie der halogenierten Dibenzo-p-dioxine und Dibenzofurane; VCH Verlagsgesellschaft mbH: Weinheim, 1996. (6) Soot Formation in Combustion; Bockhorn, H., Ed.; SpringerVerlag: Berlin, 1994; Vol. 59. (7) Lee, C. W.; Kilgroe, J. D.; Raghunathan, K. Environ. Eng. Sci. 1998, 15, 71-84. (8) Blumenstock, M.; Zimmermann, R.; Schramm, K.-W.; Kettrup, A. Chemosphere 2000, 40, 987-993. (9) Zimmermann, R.; Heger, H. J.; Kettrup, A.; Boesl, U. Rapid Commun. Mass Spectrom. 1997, 11, 1095-1102. (10) Heger, H. J.; Zimmermann, R.; Dorfner, R.; Beckmann, M.; Griebel, H.; Kettrup, A.; Boesl, U. Anal. Chem. 1999, 71, 46-57. (11) Thanner, R.; Oser, H.; Grotheer, H.-H. Eur. Mass Spectrom. 1998, 4, 215-222. (12) Zimmermann, R.; Heger, H. J.; Kettrup, A. Fresenius J. Anal. Chem. 1999, 363, 720-730. (13) Zimmermann, R.; Heger, H. J.; Blumenstock, M.; Dorfner, R.; Schramm, K.-W.; Boesl, U.; Kettrup, A. Rapid Commun. Mass Spectrom. 1999, 13, 307-314. 1030

9

ENVIRONMENTAL SCIENCE & TECHNOLOGY / VOL. 35, NO. 6, 2001

(14) Zimmermann, R.; Heger, H. J.; Kettrup, A.; Nikolai, U. Fresenius J. Anal. Chem. 2000, 366, 368-374. (15) Blumenstock, M.; Zimmermann, R.; Schramm, K.-W.; Kaune, A.; Nikolai, U.; Lenoir, D.; Kettrup, A. J. Anal. Appl. Pyrolysis 1999, 49, 179-190. (16) Ballschmiter, K.; Kirschmer, P.; Zoller, W. Chemosphere 1986, 15, 1369-1372. (17) Wikstro¨m, E.; Tysklind, M.; Marklund, S. Environ. Sci. Technol. 2000, 34, 604-609. (18) Boesl, U.; Neusser, H. J.; Schlag, E. W. Z. Naturforsch. 1978, 33A, 1546-1548. (19) Lasers and Mass Spectrometry; Lubman, D. M., Ed.; Oxford University Press: New York, 1990. (20) Tembreull, R.; Lubman, D. M. Anal. Chem. 1984, 56, 1962-1967. (21) Tanada, T. N.; Velazquez, J.; Hemmi, N.; Cool, T. A. Combust. Sci. Technol. 1994, 101, 333-348. (22) Boesl, U. J. Phys. Chem. 1991, 95, 2949-2962. (23) Fricke, J. Phys. Unserer Zeit 1973, 1, 21-27. (24) Zimmermann, R.; Lenoir, D.; Kettrup, A.; Nagel, H.; Boesl, U. In 26th Symposium (International) on Combustion; The Combustion Institute: Pittsburgh, 1996; pp 2859-2868. (25) Heger, H. J. Ph.D. Thesis, Technische Universita¨t Mu ¨ nchen, Mu ¨ nchen, Germany, 1999. (26) Perkampus, H.-H. UV-Vis Atlas of Organic Compounds; VCHVerlagsgesellschaft GmbH: Weinheim, 1992. (27) Friedel, R. A.; Orchin, M. Ultraviolet Spectra of Aromatic Compounds; John Wiley & Sons: New York, 1951. (28) German VDI Guideline 3499 (1-3), 1990. (29) Altwicker, E. R.; Konduri, R. K. N. V.; Lin, C.; Milligan, M. S. Chemosphere 1992, 25, 1935-1944. (30) Hunsinger, H.; Kreisz, S.; Vogg, H. Chemosphere 1996, 32, 109118. (31) Schramm, K.-W.; Merck, M.; Henkelmann, B.; Kettrup, A. Chemosphere 1995, 30, 2249-2257. (32) NIOSH. In Niosh Manual of Analytical Methods; Eller, P. M., Ed.; 1984; Vol. 2, pp 1501-1-1501-7. (33) Frenklach, M. In 22nd Symposium (International) on Combustion; The Combustion Institute: Pittsburgh, 1988; pp 10751082. (34) Melton, T. R.; Vincitore, A. M.; Senkan, S. M. In 27th Symposium (International) on Combustion; The Combustion Institute: 1998; pp 1631-1637. (35) Molecular Precursors of Soot and Quantification of the Associated Health Risk; Siegmann, K., Siegmann, H. C., Eds.; Plenum Press: New York, 1998; pp 143-160. (36) Addink, R.; Olie, K. Environ. Sci. Technol. 1995, 29, 1425-1435. (37) Ghorishi, S. B.; Altwicker, E. R. Hazard. Waste Hazard. Mater. 1996, 13, 11-22. (38) Wikstro¨m, E.; Tyksland, M.; Marklund, S. Environ. Sci. Technol. 1999, 33, 4263-4269. (39) Wikstro¨m, E. Ph.D. Thesis, Umea University, Umea, Sweden, 1999. (40) Wehrmeier, A.; Lenoir, D.; Schramm, K.-W.; Zimmermann, R.; Hahn, K.; Henkelmann, B.; Kettrup, A. Chemosphere 1998, 36, 2775-2801. (41) Strecker, M. In Moderne Feuerungstechnik zur energetischen Verwertung von Holz und Holzabfa¨llen: Emissionsminderung, Konzepte und ausgefu ¨ hrte Anlagen; Marutzky, R., Ed.; Springer VDI Verlag: Heidelberg, 1997; pp 332-350. (42) Zimmermann, R. Habilitation Thesis, Technische Universita¨t Mu ¨ nchen, Germany, 2000. (43) Blumenstock, M. Ph.D. Thesis, Technische Universita¨t Mu ¨nchen, Germany, 2001, in preparation. (44) Zimmermann, R.; Blumenstock, M.; Schramm, K.-W.; Kettrup, A. Organohalogen Compd. 2000, 46, 78-81. (45) Bru ¨ ggert, M.; Hu, Z.; Hu ¨ ttinger, K. J. Carbon 1999, 37, 20212030. (46) Iino, F.; Imagawa, T.; Takeuchi, M.; Sadakata, M. Environ. Sci. Technol. 1999, 33, 1038-1043. (47) Stieglitz, L.; Zwick, G.; Beck, J.; Roth, W.; Vogg, L. Chemosphere 1989, 18, 1219-1226. (48) Sanchez, E.; Yang, Y.; Find, J.; Braun, T.; Schoonmaker, R.; Belz, T.; Sauer, H.; Spillecke, O.; Uchida, Y.; Scjlo¨gl, R. In Science and Technology in Catalysis; Otsuka, K., Hattori, H., Eds.; Kodansha Ltd.: 1998; pp 317-326.

Received for review June 26, 2000. Revised manuscript received November 14, 2000. Accepted December 11, 2000. ES000143L