Contributions to combustion chemistry research

Transcription

1 Plenary Paper Topic: General 8 th U. S. National Combustion Meeting Organized by the Western States Section of the Combustion Institute and hosted by the University of Utah May 19-22, 2013 Contributions to combustion chemistry research Julia Koppmann, Andreas Brockhinke, Katharina Kohse-Höinghaus, Department of Chemistry, Bielefeld University, Universitätsstraße 25, D Bielefeld, Germany With increasingly tight emission regulations, introduction of alternative fuels or fuel additives potentially derived from biomass, and discussion of combustion regimes at low temperatures and high pressures, combustion chemistry research, already important for conventional fuels and today s combustion characteristics, has gained considerable additional weight. For any of the fuels to be potentially introduced on the market, knowledge of the pollutant spectrum to be expected would be desirable, requiring information on the associated reaction pathways. To establish and validate complete, fully detailed combustion reaction mechanisms, accurate and reliable experimental data are needed from well controlled experiments, so that transfer of the fundamental knowledge into the specific application can be performed, possibly using reduced chemistry. The present plenary lecture and associated paper report several recent examples of combustion chemistry research performed in our group and within collaborations. It highlights some results for flames of C4 hydrocarbons, specifically butane and butene, as well as for flames of selected oxygenated fuels and from combustion experiments at low temperatures. While most of these experimental results discussed here have been recently published elsewhere, new and unpublished results are provided using quantum cascade laser (QCL) absorption spectroscopy, particularly in dimethyl ether (DME) flames that had been characterized with in-situ molecular-beam mass spectrometry (MBMS) before. The potential of this technique for further flame chemistry studies is promising. 1. Introduction Detailed chemical knowledge is a prerequisite for an informed judgment on many combustion processes, especially regarding potential emission hazards. Ignition properties and chemical reaction times at lower temperature require investigation of relevant details in fuel oxidation, and intermediate species such as peroxides may play a role that have lesser influence in high-temperature reaction regimes. Furthermore, with an increasing number of potential alternative fuels entering the discussion, especially oxygenated molecular structures that are derived from biomass, previously unexplored reaction pathways must be considered, and thermochemical, transport, and reaction rate details for submechanisms must be determined or refined. Large efforts have been devoted in the past years to understand these combustion chemistry aspects from a systematic perspective [1-11], and reaction mechanisms have been provided, often with a hierarchical structure, that may be used for a prediction of ignition, combustion, and emission characteristics under laboratory and practically relevant conditions. For the validation of such chemical mechanisms, reliable experiments that provide speciation are needed, with special emphasis on stable and reactive intermediates. It is beyond the scope of this article to provide a complete picture of the numerous studies that address full species characterization in combustion systems, and we therefore only draw attention to several recent overview articles that cover some important aspects [12-17]. Here, some contributions from our own work will be exemplarily addressed that permit the detection of quantitative species profiles. This is done from a limited perspective studying laminar flames and flow reactors. Several techniques have been particularly useful in such situations, including in-situ molecular-beam mass spectrometry (MBMS) and sensitive laser diagnostic techniques such as laser-induced fluorescence (LIF) and cavity ring-down spectroscopy (CRDS). While MBMS is necessarily invasive, using a probe to withdraw the gas sample to be studied which must be introduced into the reaction system [18,19], it has the advantage that a complete species mix can be characterized. Laser techniques such as LIF and CRDS are, in contrast, not intrusive but are typically only available to detect smaller molecules with about 2-4 atoms; they may offer remarkable sensitivity into the ppb range [20] under typical flame conditions and can be used advantageously to complement MBMS studies in chemically complex combustion situations [21]. Here, we will introduce, in addition to these powerful techniques for flame chemistry diagnostics, two recent strategies from our work that complement this measurement arsenal: these are in-situ gas chromatography (GC) from the same volume as MBMS [22] and quantum cascade laser (QCL)-based infrared absorption spectroscopy.

2 2. Methods For the analysis of laminar premixed flames, mostly at reduced pressure, we have used combinations of laser spectroscopy [23] and in-situ MBMS variants [13,15] since a long time, either with instruments based in our own laboratory or with access to special instrumentation through collaborations. Ionization methods for MBMS species quantification have included resonance-enhanced multi-photon ionization (REMPI) [24] as well as combinations of electron-ionization (EI) and photoionization (PI), typically using two independent instruments in each study [25-28]. The MBMS analysis has been recently adapted to low-temperature oxidation under flow reactor [29] and flame conditions [30] at atmospheric pressure. Descriptions of the techniques, instruments, and detection conditions as well as of the strategies for species identification and quantitative concentration determination have been given in detail in the recent literature [22,26,29] and shall not be repeated here. In a given laminar premixed low-pressure flame condition, typically about stable and reactive intermediate species profiles are determined in so-called burner scans, providing the mole fraction of each species as a function of height h above the burner. Such datasets present a reasonable basis for the comparison with one-dimensional combustion models using detailed reaction mechanisms. To minimize fragmentation, especially by EI, nominal ionization energies of typically ev are used, and fragmentation corrections are applied when necessary. Since the precision of absolute concentration values depends on the availability of ionization cross sections for each ionization method as well as on several instrumental factors, a very useful strategy has been the choice of identical flame conditions for isomeric fuels in order to make differences in the species profiles more obvious; such differences may arise as a consequence of the molecular structure of the fuel and its destruction and oxidation pathways [22,25-27]. While absolute mole fractions, especially for radical species for which accurate cross sections are quite often not available, may be determined only within a factor of 2-4, relative measurements of the same species under identical conditions in flames of different isomeric fuels will permit clear distinction of structure-related variations. This has been demonstrated in flames of, e.g., the isomers of butane [26], butene [22], and butanol [27] as well as in those of isomeric small amines [25] and esters [31]. Flames of numerous fuels featuring important structural variations, such as, for example, branched and linear, saturated and unsaturated functions as well as different functional groups have been experimentally investigated under very similar flame conditions with combinations of PI-MBMS and EI-MBMS using independent instruments. Thus, a large coherent database for low-pressure flat premixed flames of hydrocarbon, oxygenated and of some nitrogenated fuels is available as a valuable resource for the development of combustion mechanisms. The important aspect of single-photon PI-MBMS using vacuum ultraviolet radiation for the analysis of combustion chemistry is the possibility of performing energy scans with high energy resolution of typically below 50 mev. This is done in the range of chemically interesting ionization energies, i.e. from about 8 to 20 ev, resulting in photoionization efficiency (PIE) spectra from which different isomers can be identified [12,15-17]. The advent of this technique has considerably enhanced the amount of information obtainable from reacting environments. It has allowed detection of species such as enols and peroxides that had not been identified in combustion systems before. Thus, reaction mechanisms could be updated to include reactions of such species. Also, detection of elusive species had permitted to corroborate steps in the reaction sequence that had not been experimentally amenable before. However, beam time at synchrotron facilities is limited and not every combustion mechanism detail may need such advanced-level instrumentation. Also, available PI-MBMS spectrometers at beamlines in the USA and China have featured lower mass resolution than the EI-MBMS set-up in our own laboratory which, with typically m/ m=4000, has permitted to distinguish species with different heavy atoms in combustion of oxygenated and nitrogenated fuels by their exact mass [27,28]. As a standard analytic technique, GC is being used in flame chemistry diagnostics, often with microprobe sampling. GC is not capable of detecting radical species, but may be advantageous for isomer separation of stable species. As a novel feature for flame chemistry analysis, we have therefore recently reported the coupling of EI-MBMS and GC from the same sampled volume in a flame [22] via a newly developed interface; a schematic of the set-up is given in Fig. 1. In this configuration, a gas sample is transferred from the first pumping stage of the MBMS to the GC (Thermo Fisher Scientific). The first pumping stage may be separated by a gate valve, and after blocking of the molecular beam, the chambers may be connected by the bypass shown in Fig. 1. With a valve system (VICI ED48UWE), a gas volume to be analyzed may be sampled at low pressure into a well-defined volume (sampling loop) at the sampling position and then transferred to the GC column in the measurement position. The vacuum bypass is then blocked and the sample is compressed by the helium carrier gas to match the 1.5 bar inlet conditions of the GC. The analysis is thus done under the same sampling conditions as for the MBMS detection. For in-situ analysis of a combustion environment featuring mainly small hydrocarbons, a PLOT column (Alumina BOND, Na 2 SO 4 deactivation, Restek) is employed using a suitable temperature-step program from 50 C to 200 C with a heating rate of 20 C min -1 for elution. 2

3 Fig. 1. Experimental setup of the MBMS apparatus (top), showing flame chamber, first pumping stage, ionization chamber, and reflectron TOF-MS. Detailed sampling configuration (bottom) including the low-pressure GC sampling interface. Valve positions are indicated: solid lines: sampling, broken lines: measurement. Fig. 1 in Ref. [22], reprinted with permission of The Combustion Institute. As stated above, MBMS techniques have the general drawback of being invasive. This is of influence in the comparison of species profiles taken with optical versus MBMS methods, since the sampling probe, depending on its geometrical configuration, will perturb temperature and species mole fractions in the flame [18,19,32]. Examples of shifts that have been applied to experimental temperature profiles for a comparison with MBMS measurements are widely reported in the literature, and a history of probe perturbation analyses has been given in the very recent article by Skovorodko et al. [32], together with an experimental and numerical characterization of such effects. In our recent MBMS studies, we have used a perturbed temperature profile which relies on the pressure in the first pumping stage in the MBMS measurements itself, calibrated by LIF or CRDS in the burnt-gas region [19,22,26,27], rather than on an unperturbed temperature profile from laser spectroscopic experiments. This has been done assuming that a gas sample that has arrived in the first pumping stage of the mass spectrometer set-up (as a reference for the temperature) has experienced the same sampling effects as the species that are detected by MBMS. This approach has made shifts in the temperature profile as input to flat premixed flame model calculations unnecessary [22,26,27,33] in most cases. In flame investigations that report species profiles from both laser and mass spectrometric techniques [21], two different temperature profiles would then have to be used for simulations, an unperturbed one for comparison of the model calculations with those species that have been detected by optical methods versus a perturbed one for comparison of the model with those detected by MBMS. While this can be done in principle, influences of the sampling probe on species mole fractions have been noted [18,19,32] in addition to the temperature effects. It is therefore advisable to measure the same species under identical flame conditions with MBMS and by non-invasive methods, as has been done by some authors in the past [see references in 19,32]. Struckmeier et al. [19] provide an example of quantitative methyl and formaldehyde detection using CRDS, EI-MBMS, and PI-MBMS. Within the joint error limits, these profiles are in agreement, but they are noted to exhibit some differences in shape and magnitude. The discussion in [19] also shows that only a limited number of species will be suitable for such direct comparisons, due to limitations of both types of diagnostics. 3

4 CaF 2 plate HeNe-laser vacuum pump As a relatively recent tool, we have started to apply mid-infrared (IR) laser absorption techniques using QCL sources for detailed flame chemistry investigations. Mid-IR radiation is extremely valuable in combustion-related diagnostics since it can probe fundamental vibrations. Absorption in this wavelength regime has been employed under various atmospheric and combustion-related conditions including applications in shock tubes, gas cells, combustion exhaust gases, coal-fired power plants, and heavy-duty diesel engines [34-39]. Access to these wavelengths has been established with distributed feedback (DFB) laser radiation, difference frequency generation (DFG) approaches as well as QCLs. For example, Pyun et al. [34] have detected CO in a shock tube with QCL absorption at a wavelength of 4.56 µm and Vanderover and Oehlschlaeger [36] have demonstrated CO and temperature measurements in shock-heated gases using QCL absorption. NO detection with a QCL at 5.2 µm has been reported recently by Chao et al. [37] in particle-laden combustion exhaust, and she and her collaborators have been able to apply two dual-channel sensors for NO/NH 3 and CO/O 2 in exhaust from a coal-fired power plant, including QCL-based NO absorption at 5.2 µm [38]. Furthermore, Kawahara et al. [39] have described a fiber probe for residual gas analysis inside a heavy-duty diesel engine, using 4.3 µm QCL radiation. These and other investigations have demonstrated the wide applicability of quantum cascade lasers to gas analysis for different combustion conditions. Here, we focus on the use of QCLs as a source for direct laser absorption for combustion chemistry research, complementing LIF, CRDS, and in particular MBMS analysis of laminar, flat premixed low-pressure flames as one example of research with these laser sources in our laboratory. The set-up used in this work is shown schematically in Fig. 2. 5,9 μm 7,8 μm 4,5 μm QCLcontroller valve data processing lid coolant off-axisparabolic mirror digitizer CaF 2 lens burner coolant fuel / oxidizer MCT-detector Fig. 2. Set-up for quantum cascade laser absorption spectroscopy in flat premixed low-pressure flames. The laser beam is focused at the center of the moveable burner. A co-flow around the burner matrix is used to prevent build-up of absorbing species in the chamber and to permit Abel inversion. The three QCL units (Cascade Technologies) operating at 4.5 µm, 5.9 µm, and 7.8 µm are controlled individually with a controller unit. Typical operation conditions are pulse lengths of 500 ns and 20 khz repetition rate. The laser beams are combined by several sets of mirrors and are focused by a CaF 2 lens into the burner chamber. A joint path with a He-Ne laser serves to align the beams. The beam profiles are examined and shaped by suitable pinholes. After transmission through the burner chamber, the beam is focused with an off-axis parabolic mirror onto an IR detector (Vigo Systems PVI-2TE). Signals are then digitized and processed. To minimize interference by combustion reactant and product gases that could accumulate around the burner, a flat sintered burner of 66 mm diameter with an additional shroud of 20 mm width for inert gas co-flow is used. To reduce reflections, the aluminum burner chamber is completely anodized. Furthermore, windows are mounted in Brewster s angle to permit complete transmission of radiation with polarization parallel to the plane of incidence. The window surfaces are not parallel but placed at an angle to each other to avoid etaloning effects. Inside the cylindric window supports, the laser beams are guided through N 2 -purged tubes of a few mm diameter into close proximity of the flame, to minimize accumulation of combustion gases in these regions. Wavelength calibration is performed using an N 2 O calibration gas cell and a Ge-etalon. Edge effects are removed using a tomographic reconstruction procedure based on the work of Dasch [40] and described in detail in [41]. 4

5 3. Results and Discussion Several examples of combustion chemistry diagnostics using the described techniques are presented in the following. As one target, flames of C4 fuels were investigated, burning the two butane isomers, the three butene isomers and the four butanol isomers under comparable fuel-rich conditions. Laminar premixed C 4 H x O y /O 2 /Ar flames (with x=8,10;y=0 and x=10;y=1) were stabilized on bronze-matrix burners of typically mm diameter at a pressure of 40 mbar and an equivalence ratio of ϕ=1.7. With the current interest in alternative fuels, and with butanol as a potentially promising renewable fuel, a comprehensive model for C1-C4 combustion chemistry is highly warranted, and suitable experimental validation data with extensive quantitative species information under comparable conditions is scarce. We have provided species profiles for about 30 intermediates including radicals in flames of all these C4 fuels [22,26,27]. For the C4 hydrocarbon fuels butane and butene, modeling of our experimental conditions [22, 26] was performed with a general hydrocarbon oxidation mechanism starting from work by Hoyermann et al. [42] with addition of n-heptane and toluene chemistry [43,44]. In general, the trends observed in flames of both butane isomers are well represented by the model, with interesting reaction paths differences evident from a reaction flux analysis [26]. While the combustion of n- butane proceeds mainly through 1-C 4 H 9 and 2-C 4 H 9 radicals towards C 2 H 4, C 2 H 2 and HCCO, i-butane consumption proceeds through t-c 4 H 9 and i-c 4 H 9 towards C 3 H 6 and then via further C2 and C3 species to C 2 H 2 and HCCO. The mole fraction profiles for selected unsaturated species including C n H 2 (with n=2,4,6) and C m H 4 (with m=4,5,6) are given in Fig. 3. Such compounds are considered to be involved in carbon growth. 40 x C 2 H x 0.45 C 4 H x 0.52 C 4 H C 5 H x 0.12 C 6 H C 6 H n-butane EI n-butane PI i-butane EI i-butane PI n-butane model i-butane model Fig. 3. Mole fraction profiles of selected unsaturated species in flames of butane isomers. Symbols are experimental data (left axis), and dashed lines are modeling results (right axis). To facilitate comparison, a scaling factor (model axis/experiment axis) is indicated in the upper left corner. Thin solid lines are drawn to guide the eye for C 5 H 4 and C 6 H 4 which are not included in the model. Fig. 5 in [26], reprinted with permission of Oldenbourg Verlag. 5

6 Experimental trends, including EI-MBMS and PI-MBMS results are in excellent agreement with each other and, for all compounds shown in Fig. 3, are well represented by the model. It should be noted, however, that the mole fractions are scaled by the factors given in the upper left corner so that profile shapes can be easily compared, and maxima deviate by about a factor of two for C 2 H 2, C 4 H 4, and C 4 H 2. Deviations are larger for C 6 H 2 ; C 6 H 4 and C 5 H 4 are not included in the model. From the energy scans performed with PI-MBMS, C 4 H 2 is diacetylene, C 4 H 4 is vinylacetylene, and C 5 H 4 is 1,3- pentadiyne. Several contributions are possible for C 6 H 4 which could be the linear isomers hexa-1,5-diyne-3-ene or hexa- 1,3-diyne-5-ene; it could also include the cyclic isomer benzyne. Since the ionization energies are rather similar, the separation was not unambiguously possible here within the experimental signal-to-noise ratio. C 6 H 2 was identified as 1,3,5-hexatriyne. For both isomeric butane fuels, the measured concentrations of these species are quite similar, C 2 H 2 being a major intermediate along the reaction pathways of both fuels, and dehydrogenation and build-up reactions contributing in potentially similar fashion for both fuels. Structure-related differences in the combustion reactions lead to different mole fractions observed for further C2 as well as C3 intermediates, including C 2 H 4, C 2 H 5, C 3 H 6 and C 3 H 5, with experimental trends being again in good agreement with the model [26]. Isomer identification in the butane flames relied on energy scans from PI-MBMS. In the butene flames in [22], some key isomers were separated instead by using the in-situ combination of EI-MBMS and GC. An exemplary result is given in Fig. 4. Fig. 4. Identification of C 5 H 8 isomers in butene flames: GC analysis for m/z= at h=3 mm, and cold-gas calibration (lower right) for cyclopentene, isoprene, 1,4-pentadiene, and trans-1,3-pentadiene. Fig. 12 in [22], reprinted with permission of The Combustion Institute. The isomer composition for C 5 H 8 given in Fig. 4, detected in the reaction zones at h=3 mm of the three different butene flames, shows different contributions. Detected compounds, identified in accord with the calibration spectra (bottom right in Fig. 4) include cyclopentene, isoprene, 1,4-pentadiene and trans-1,3-pentadiene. Two further peaks seen in the chromatograms could not be assigned unambiguously; most probably, peak 2 represents cis-1,3-pentadiene. Significant contributions in all three flames are seen from isoprene and trans-1,3-pentadiene, with an isoprene/pentadiene ratio that is similar for the linear 1-butene and 2-butene fuels, and a much more dominant contribution of isoprene to C 5 H 8 for the flame of the branched i-butene fuel. Both experiment and model result in the highest mole fraction of C 5 H 8 for the i-butene flame, followed by those for 2-butene and 1-butene flames, respectively. Similarly, GC analysis was performed in all three flames regarding C 3 H 4 isomers at m/z= (allene and propyne), C 4 H 6 at m/z= (1,3- butadiene, 1,2-butadiene, 1-butyne, and 2-butyne), and C 5 H 10 at m/z= (trans-2-pentene, cis-2-pentene, 1-pentene, 2-methyl-1-butene. and 3-methyl-1-butene) [22]. Not only are such species identifications important for the comparison with flame model simulations, but they also permit the evaluation of a detected signal at a given mass using the cross section of the dominant isomer; valuable information that would not be available from the EI-MBMS experiment alone. 6

7 mole fraction [10-5 ] counts T h [K] While the previous examples reported investigations in flames, similar diagnostics is valuable in other reactive environments that permit characterization of the detailed fuel breakdown and oxidation chemistry. Mass spectrometric analysis in a laminar flow reactor at atmospheric pressure was recently used in our group to study details of the lowtemperature oxidation of dimethyl ether (DME) [29]. Different stoichiometries were investigated under highly diluted conditions (97% Ar) in the range of K. Major species and important stable intermediates such as CH 2 O, CH 4, and the C2 species C 2 H 6, C 2 H 4, and C 2 H 2 were detected as a function of temperature and equivalence ratio, with CO, H 2 and CH 2 O observed in the negative temperature coefficient (NTC) region near 600 K. As shown in Fig. 5, formic acid was also detected exclusively in the NTC region a m/z 6 d 4 2 C 2 H 4 O 2 = 0.8 = 1.0 = T h [K] 1050 b m/z T h [K] signal intensity [10-5 a. u.]8 c Fig. 5. Ion signal at m/z=46. a: Formic acid (46.01) and DME (46.042) at T h =583 K, individual signal contributions are indicated and b: grayscale-coded signal for the entire temperature range and =1.0. c: Integrated signals as a function of T h. d: Methyl formate mole fraction profile. Fig. 5 in [29], reprinted with permission of The Combustion Institute. The identification of formic acid was possible with the high mass resolution of the instrument using numerical integration to separate its contribution from that of the highly dominant fuel (DME) signal. An additional signal was noted in the NTC regime of formula C 2 H 4 O 2, which could be assigned from PI-MBMS measurements to be methyl formate. Identification of these and other intermediates will be a valuable step in refining detailed models for the lowtemperature oxidation chemistry of hydrocarbon and oxygenated fuels. As a complement to mass spectrometry, which has the unique capability to detect the entire species mix in a flame without prior knowledge of the chemical composition, mid-infrared absorption with QCLs is emerging as a powerful technique for flame chemistry diagnostics since it permits to detect a number of stable intermediates without perturbing the flame. As a line-of-sight technique, absorption is suited for applications in the laminar flat premixed flames described above. In the wavelength region covered by the three lasers in the set-up given in Fig. 2, the detection of important intermediates and products including CO, CO 2, CH 4, C 2 H 2, and CH 2 O can be achieved. Also, temperature is accessible using the QCL at 4.5 µm in the range of cm -1 (operating at 12 C) where a large number of CO 2 features and two CO lines are available. Water, acetylene, and methane can be detected with the QCL at 7.8 µm (at an operating temperature of 10 C) in the cm -1 regime. With the QCL at 5.9 µm (at 30 C), H 2 O and CH 2 O signals were detected in the range of cm -1. Mid-IR absorption measurements were performed in a number of flames with methane, propene, furan, and DME as the fuels, at pressures of mbar and for stoichiometric to fuel-rich conditions. Spectra were simulated referring to the HITRAN/HITEMP databases [45,46]. The signals were fit using a spectral function containing a variable combination of Gaussian and Lorentzian contributions; typically, the Gaussian part is dominant in the low-pressure flames because of Doppler broadening. Mole fractions were determined from the absolute signal areas, taking the measured temperature and the pressure into account. Figure 6 shows a spectrum taken in a ϕ=1.2 methane/oxygen/argon flame at 50 mbar and h=25 mm. The upper half displays the measured spectrum, together with the simulated spectrum in the lower half as a mirror image. The contributions from CO and CO 2 are identified in excellent agreement with the HITEMP [46] simulation. The depicted spectrum shows the region that was evaluated for the CO and CO 2 mole fractions as well as for the temperature measurements. 7

8 mole fraction T [K] absorbance [cm -1 ] measurement CO HITEMP simulation CO 2 HITEMP simulation sum wave number [cm -1 ] Fig. 6. Spectra of CO and CO 2 in a ϕ=1.2 methane flame at 50 mbar and h=25 mm; top: experiment, bottom: simulation using the HITEMP database [46]. An example for temperature, CO, and CO 2 measurements using this method is given in Fig. 7 for a furan/ oxygen/argon flame that was recently investigated by EI-MBMS in our group [47]. Results from both methods agree well within the experimental uncertainty CO CO 2 EI-MBMS T h [mm] Fig. 7. CO and CO 2 mole fractions and temperature T, determined by QCL absorption (filled symbols) as a function of height h in a stoichiometric furan flame at 20 mbar. The temperature profile and the EI-MBMS results (open symbols) have been reported in [47]. A more detailed comparison of CO and CO 2 results measured by QCL absorption and in-situ mass spectrometry is given in Fig. 8. Here, the DME/oxygen/argon flames at 33 mbar investigated by Wang et al. [48] with a combination of PI-MBMS and EI-MBMS were taken as a reference. For the five stoichiometries reported in [48], the agreement of absolute CO and CO 2 mole fraction profiles with the previous PI-MBMS results is, in general, quite good, especially in the regions downstream of the flame front. Upon closer inspection, however, some differences are seen, especially in the rising portions of the profiles near the burner surface. The experimental data were compared with numerical simulations using the CANTERA software [49] developed by Goodwin [50] and the combustion model by Zhao et al. [51]. As shown in Fig. 8, the agreement for both CO and CO 2 is encouraging. Trends are very well represented, including the maxima of the CO profiles in all but the richest flame. For all CO profiles, the simulation results in slightly larger values than the QCL experiment, which may hint at a systematic error. Differences are less than 10%, however, and within the experimental uncertainty. The overall situation is similar for the CO 2 results, with slightly larger differences between the two sets of measurements and between experiment and simulation. We are not aware of any previous systematic comparison of MBMS and QCL absorption for these species under premixed low-pressure flame conditions, and remaining uncertainties will have to be investigated in more detail in the future. 8

9 CH 2 O mole fraction CH 2 O mole fraction CO mole fraction CO 2 mole fraction DME 0.93 DME 1.63 DME 1.16 DME 1.86 DME 1.40 MS (Wang 2009) model Zhao T free h [mm] h [mm] Fig. 8. CO (left) and CO 2 (right) mole fraction profiles determined by QCL absorption in the five DME flames at 33 mbar studied previously by Wang et al. [48] with in-situ MBMS. Simulations with the CANTERA code [49] using the temperature calculated by the model (T free ) and the mechanism of Zhao et al. [51] are included as solid lines. It is intriguing to study the flame front region in more detail using invasive probe-sampling and non-perturbing laser techniques. Formaldehyde exhibits a sizable mole fraction in DME flames and can thus be assessed with good sensitivity by QCL absorption. Figure 9 shows a comparison of the mid-ir absorption results with the EI-MBMS measurements given by Wang et al [48] for two different stoichiometries. Again, numerical simulations with the mechanism by Zhao et al. [51] are included. Temperature profiles were determined by QCL absorption so that the initial temperature rise was well captured. Note that the height scale in Fig. 9 (0-5 mm) represents only a small region of that depicted in Fig. 8 (0-27 mm), so that details of the intermediate profile are much more pronounced QCL EI-MBMS model Zhao T fix DME = 0.93 DME = h [mm] h [mm] 0 Fig. 8. Formaldehyde mole fraction profiles determined by QCL absorption in two DME flames studied previously by Wang et al. [48] with in-situ MBMS. Simulations with the CANTERA code [49] using a temperature profile determined by QCL absorption (T fix ) and the mechanism of Zhao et al. [51] are included as solid lines. The MBMS maxima are shifted downstream by about 2 mm in both cases relative to the QCL absorption measurements; also, the maximum mole fraction determined by MBMS is about 30% lower in the fuel-rich flame. With the non-perturbed temperature profile determined from the QCL absorption measurements as input, the model predictions agree excellently with the QCL concentration measurements. The MBMS results would have to be simulated using a perturbed temperature profile, which was, however, not determined in the experiments by Wang et al. [48]. With more intermediate species quantitatively accessible by QCL diagnostics in flames, and more extensive data available to assess potential limitations, some of the ambiguities associated with probe sampling influences on in-situ MBMS measurements may be alleviated, and the present arsenal of techniques for combustion chemistry diagnostics may be favorably extended. 9

10 4. Conclusions Combustion chemistry research is vital to understand the detailed fuel consumption and oxidation reactions and to assess potential pollutant emissions. Alternative fuels feature new combustion chemistry details, including for example the presence of more elements, such as oxygen and nitrogen, in the fuel molecule and its decomposition and oxidation products. Diagnostics for combustion chemistry must be able to perform structure-selective quantitative analysis. In the present contribution, just a few examples from gaseous combustion environments have been presented. New combustion regimes, at higher pressures or lower temperatures, under high dilution, or in the presence of aerosols and surfaces need techniques capable of detecting chemical influences under such conditions. While the number of advanced techniques for combustion chemistry research is not small, the time resolution needed for diagnostics in practical systems precludes the detailed chemical speciation that is accessible in the laminar flame and flow reactor experiments discussed here. With state-of-the-art in-situ mass spectrometry, using high mass resolution and isomer separation from GC or from energy scans, details for model validation are available in dedicated laboratory systems, whereas mid-infrared absorption spectroscopy using quantum cascade lasers may find more wide-spread use in practical systems where they may provide access to a relevant number of major and intermediate species that characterize the chemical reaction process. Acknowledgements This research was funded by in part by DFG in SFB 686 TP B3 and TP C5. The authors thank Marina Schenk, Kai Moshammer, both Bielefeld, and Dr. Patrick Oßwald, DLR Stuttgart, for a critical reading of the manuscript. Many researchers have contributed to the work presented here, they are named as authors in the original literature cited in this article, from which these examples were drawn. References [1] F. Battin-Leclerc, Prog. Energy Combust. Sci. 34 (2008) [2] H. Wang, Proc. Combust. Inst. 33 (2011) [3] E. Ranzi, A. Frassoldati, R. Grana, A. Cuoci, T. Faravelli, A. P. Kelley, C. K. Law, Prog. Energy Combust. Sci. 38 (2012) [4] C. K. Westbrook, W. J. Pitz, O. Herbinet, H. J. Curran, E. J. Silke, Combust. Flame 156 (2009) [5] A. Mzé-Ahmed, P. Dagaut, K. Hadj-Ali, G. Dayma, T. Kick, J. Herbst, T. Kathrotia, M. Braun-Unkhoff, J. Herzler, C. Naumann, U. Riedel, Energy Fuels 26 (2012) [6] C. Xu, A. A. Konnov, Energy 43 (2012) [7] J. Zador, C. A. Taatjes, R. X. Fernandes, Prog. Energy Combust. Sci. 37 (2011) [8] K. Kohse-Höinghaus, P. Oßwald, T. A. Cool, T. Kasper, N. Hansen, F. Qi, C. K. Westbrook, P. R. Westmoreland, Angew. Chem. Int. Ed. 49 (2010) [9] C. K. Westbrook, C. V. Naik, O. Herbinet, W. J. Pitz, M. Mehl, S. M. Sarathy, H. J. Curran, Combust. Flame 158 (2011) [10] M. R. Harper, K. M. van Geem, S. P. Pyl, G. B. Marin, W. H. Green, Combust. Flame 158 (2011) [11] C. Bahrini, O. Herbinet, P.-A. Glaude, C. Schoemaecker, C. Fittschen, F. Battin-Leclerc, J. Am. Chem. Soc. 134 (2012) [12] T. A. Cool, K. Nakajima, T. A. Mostefaoui, F. Qi, A. McIlroy, P. R. Westmoreland, M. E. Law, L. Poisson, D. S. Peterka, M. Ahmed, J. Chem. Phys. 119 (2003) [13] C. S. McEnally, L. D. Pfefferle, B. Atakan, K. Kohse-Höinghaus, Prog. Energy Combust. Sci. 32 (2006) [14] C. A. Taatjes, N. Hansen, D. L. Osborn, K. Kohse-Höinghaus, T. A. Cool, P. R. Westmoreland, Phys. Chem. Chem. Phys. 10 (2008) [15] N. Hansen, T. A. Cool, P. R. Westmoreland, K. Kohse-Höinghaus, Prog. Energy Combust. Sci. 35 (2009) [16] Y. Li, F. Qi, Acc. Chem. Res. 43 (2010) [17] F. Qi, Proc. Combust. Inst. 34 (2013) [18] A. T. Hartlieb, B. Atakan, K. Kohse-Höinghaus, Combust. Flame 121 (2000) [19] U. Struckmeier, P. Oßwald, T. Kasper, L. Böhling, M. Heusing, M. Köhler, A. Brockhinke, K. Kohse-Höinghaus, Z. Phys. Chem. 223 (2009) [20] M. Köhler, A. Brockhinke, M. Braun-Unkhoff, K. Kohse-Höinghaus, J. Phys. Chem. A114 (2010) [21] P. Nau, A. Seipel, A. Lucassen, A. Brockhinke, K. Kohse-Höinghaus, Exp. Fluids 49 (2010) [22] M. Schenk, L. Leon, K. Moshammer, P. Oßwald, T. Zeuch, L. Seidel, F. Mauss, K. Kohse-Höinghaus, Combust. Flame 160 (2013) [23] K. Kohse-Höinghaus, R. S. Barlow, M. Aldén, J. Wolfrum, Proc. Combust. Inst. 30 (2005) [24] M. Kamphus, M. Braun-Unkhoff, K. Kohse-Höinghaus, Combust. Flame 152 (2008)

11 [25] A. Lucassen, K. Zhang, J. Warkentin, K. Moshammer, P. Glarborg, P. Marshall, K. Kohse-Höinghaus, Combust. Flame 159 (2012) [26] P. Oßwald, K. Kohse-Höinghaus, U. Struckmeier, T. Zeuch, L. Seidel, L. Leon, F. Mauss, Z. Phys. Chem. 225 (2011) [27] P. Oßwald, H. Güldenberg, K. Kohse-Höinghaus, B. Yang, T. Yuan, F. Qi, Combust. Flame 158 (2011) [28] A. Lucassen, P. Oßwald, U. Struckmeier, K. Kohse-Höinghaus, T. Kasper, N. Hansen, T. A. Cool, P. R. Westmoreland, K. Kohse-Höinghaus, Proc. Combust. Inst. 32 (2009) [29] F. Herrmann, P. Oßwald, K. Kohse-Höinghaus, Proc. Combust. Inst. 34 (2013) [30] K. Zhang, K. Moshammer, P. Oßwald, K. Kohse-Höinghaus, Proc. Combust. Inst. 34 (2013) [31] B. Yang, C. K. Westbrook, T. A. Cool, N. Hansen, K. Kohse-Höinghaus, Proc. Combust. Inst. 34 (2013) [32] P. A. Skovrodko, A. G. Tereshchenko, O. P. Korobeinichev, D. A. Knyazkov, A. G. Shmakov, Combust. Theor. Model. 17 (2013) [33] S. M. Sarathy, S. Vranckx, K. Yasunaga, M. Mehl, P. Oßwald, W. K. Metcalfe, C. K. Westbrook, W. J. Pitz, K. Kohse-Höinghaus, R. X. Fernandes, H. J. Curran, Combust. Flame 159 (2012) [34] S. H. Pyun, W. Ren, K.-Y. Lam, D. F. Davidson, R. K. Hanson, Combust. Flame 160 (2013) [35] L. Tao, M. A. Khan, D. J. Miller, M. A. Zondlo, Opt. Express 20 (2012) [36] J. Vanderover, M. A. Oehlschlaeger, Appl. Phys. B99 (2010) [37] X. Chao, J. B. Jeffries, R. K. Hanson, Appl. Phys. B106 (2012) [38] X. Chao, J. B. Jeffries, R. K. Hanson, Proc. Combust. Inst. 34 (2013) [39] N. Kawahara, E. Tomita, A. Ohtsui, Y. Aoyagi, Proc. Combust. Inst. 33 (2011) [40] C. J. Dasch, Appl. Opt. 31 (1992) [41] R. Villarreal, P. L. Varghese, Appl. Opt. 44 (2005) [42] K. Hoyermann, F. Mauß, T. Zeuch, Phys. Chem. Chem. Phys. 6 (2004) [43] S. S. Ahmed, F. Mauß, G. Moréac, T. Zeuch, Phys. Chem. Chem. Phys. 9 (2007) [44] S. S. Ahmed, F. Mauß, T. Zeuch, Z. Phys. Chem. 223 (2009) [45] HITRAN, [46] HITEMP, [47] D. Liu, C. Togbé, L.-S. Tran, D. Felsmann, P. Oßwald, P. Nau, J. Koppmann, A. Lackner, P.-A. Glaude, B. Sirjean, R. Fournet, F. Battin-Leclerc, K. Kohse-Höinghaus, Combust. Flame (2013) submitted for publication. [48] J. Wang, M. Chaos, B. Yang, T. A. Cool, F. L. Dryer, T. Kasper, N. Hansen, P. Oßwald, K. Kohse-Höinghaus, P. R. Westmoreland, Phys. Chem. Chem. Phys. 11 (2009) [49] D. G. Goodwin, CANTERA C++ User s Guide, [50] D. G. Goodwin, Chem.Vap. Depos. XVI and EuroCVD 14 (2003) 08. [51] Z. Zhao, M. Chaos, A. Kazakov, F. L. Dryer, Int. J. Chem. Kinet. 40 (2008)