Measurement of Gaseous HAP Emissions from Idling Aircraft as a Function of Engine and Ambient Conditions (2012)

Below is the uncorrected machine-read text of this chapter, intended to provide our own search engines and external engines with highly rich, chapter-representative searchable text of each book. Because it is UNCORRECTED material, please consider the following text as a useful but insufficient proxy for the authoritative book pages.

88 This section describes the results of the quality assurance procedures for the ACRP Project 02-03a that took place in February 2009, October 2009, and March 2010. The key dilution tracer species appropriate to emission index characterization for aircraft engine is CO2. The CO2 emission index can be predicted quite accurately using fuel composition. If the combustor is lit, but operating away from its peak efficiency, CO is among the compounds that indicate inefficiency. Typically the CO2 EI is 3160 g kg-1 fuel (Baniszewski et al. 2010). During these tests, the CO EI did not exceed 105 g kg-1 (or <5% of CO2 in ppmC). This suggests that using CO2 as the exclusive dilution tracer (or more formally the form that all fuel carbon is emitted in) for aircraft engine exhaust can bias the speciated hydrocarbon emission index by <5%. This section begins by discussing analytical approach and calibration data for the principal dilution tracer and combus- tion product, CO2. The subsequent section will describe the methodology and quality assurance procedures for the spe- cies measured using tunable infrared differential absorption (TILDAS). These include, CO, formaldehyde, ethane and for a subset of the work, NO2 and methane. The majority of the HAP oxygenated and aromatic hydrocarbon compounds were measured using two related mass spectrometric methods: pro- ton transfer reaction mass spectrometry (PTR-MS) and NO+ mass spectrometry (NOMS). The validation of these measure- ment is described, both the independent calibration and when available the gas chromatographic comparison when sampling the aircraft exhaust matrix. In the final section the flame ion- ization detector (FID) operation and use is described in detail since this was the time the FID had been deployed on the truck sampling manifold. Quality Assurance Description for Carbon Dioxide This section includes a technical description of the different carbon dioxide measurement instruments deployed in ACRP 02-03a measurement campaigns at MDW 2009, DAL2010, and MDW 2010. Three different measurements of carbon dioxide were made with two employed on the sample line used for gas phase species characterization and one on the sample line used to characterize particulate characteristics. The three different carbon dioxide instruments, a LiCor 820, and two LiCor 6262 were deployed. When the concen- tration range is in excess of 2500 ppmv, we use the Licor model 820 instrument equipped with the short absorption cell. When the concentration range is less than 2500 ppmv, we rely on two Licor model 6262 instruments. Occasionally, Licor model 840 are also used when the Licor 6262 is not available on a particular sampling line. The on-site calibra- tion procedures are similar in principal for all of the units. The Licor 6262 have additional sorbent chemicals that are changed in the laboratory prior to deployment. The LiCor instruments are non-dispersive infrared gas analyzers and have precisions under 1 ppm for the model 820 and under 600 ppb for the model 6262. Zeroes and spans were taken by overblowing the inlet to the instrument with either CO2 free N2 or calibration span gas. The instruments were calibrated to the span values after the readings were taken. All spans, subsequent to the initial setup calibration, were within 1% of the span value, which was gas taken from a tank certified to 1000 ± 20 ppm by the manufacturer (Scott Specialty Gases) and certified to 994 ppm by an absolute CO2 measurement (accurate to 1%). Intercomparisons of this CO2 standard with a second, independent 1% accuracy tank (Scott Marin gases) owned by University of California, Berkeley showed that both standards agreed within 0.7%. The experience with these instruments suggests they typi- cally hold their calibration very well when they are not subjected to caustic pressure changes and when particulate matter is filtered from the sample gas. The sample gas is moved through the instrument in two ways depending on the experiment. In the dedicated engine testing on the 1 meter probe (DAL 2009 and MDW 2010), there is little hope of having the manifold pressure be constant. To prevent the instrument from expe- riencing the pressure fluctuations in the manifold, we use a A p p e n d i x e Quality Assurance Documentation for Analytical Methods

89 sealed diaphragm pump to compress sample and “push” it through the instrument. The flow through the instrument is monitored with a venturi flow meter on the outlet to ensure that the instrument is always sampling continuously. In the mode when sampling from atmosphere, when sam- pling from the front of the truck (exclusively at MDW 2009 and during the airport operations phases of all tests), we use the vacuum pump and a critical orifice to draw sample through the instrument. Generally, in this mode the QC-TILDAS instru- ment, described earlier, controls valving which periodically floods the inlet with zero air. This serves several functions. The licors are zeroed by this process, the time response is evaluated and any sample line temporal offset (1–2 seconds) is measured for all instruments on the manifold. Quality Assurance: TILDAS Instruments Measurements of CO, formaldehyde, ethane, NO2 and methane The instruments documented in this section employ Tun- able Infrared Laser Differential Absorption Spectroscopy (TILDAS) as the fundamental analytical method for quan- tifying trace compounds. Although TILDAS methods using tunable diode lasers have been widely used for a variety of trace gas measurements (Sachse et al. 1987, Zahniser et al. 1995) the requirement for cryogenic cooling of lasers and detectors and the uneven quality of lead salt diode lasers has limited wider application of TILDAS methods. Improve- ments in engineering led to the development of the instru- ment deployed to the aircraft exhaust studies, robust and portable instrumentation that can operate without cryogenic cooling of the laser (Herndon et al. 2006, Jimenez et al. 2005, McManus et al. 2005, Nelson et al. 2002, Nelson et al. 2004) During the ACRP 02-03a deployments the twin dual quan- tum cascade laser based instruments were configured to mea- sure CO and formaldehyde (in the first chassis) and ethane and a second species (in the second chassis). The second species was NO2 in the MDW 2009 and DAL 2010 but methane in the MDW 2010 test. An example spectrum depicting the spectrum and analysis software is depicted in the inset (Figure E-1). The measurement of CO, though not specified by the research objective, is still fundamental to the overall approach of using fuel carbon content to arrive at mass per mass of fuel emission indices following the ICAO convention. CO was quantified using the singlet transition located at 2165.6 cm-1. Formaldehyde (HCHO) was measured using a clus- ter of absorption lines at 1764.9 cm-1. Ethene (C2H4) was measured using a pair of lines at 952.08 cm-1. Nitrogen dioxide (NO2)is also not specified by the program objective, however, we note that it has been extremely useful in prior measurements for helping rule out other potential emissions sources as well as been a keen diagnostic of the idling air- craft engine. It was measured using the absorption lines near 1599.9 cm-1. Although the QC-TILDAS method is fundamentally an absolute Beers-law absorption technique that does not require calibration, Aerodyne has found that the use of cali- bration procedures, either in the field or in the laboratory prior to deployment is a mandatory practice. The analysis Figure E-1. Experimental spectrum with 210 m pathlength, 30 Torr, and CH4 mixing ratio of 1870 ppb with an averaging time of 30 s. The upper panel shows experimental spectrum (points) and fit (solid line) to the data using HITRAN spectral lines for 12CH4 (peak absorbance 0.07) and 13CH4 (peak absorbance 0.0022). The lower panel shows the residuals to the fit with a root-mean-square deviation of 10-5 absorbance units.

90 software records all direct absorption spectra, which can facilitate archival re-analysis that retains substantial value if there are any future discoveries about the spectroscopy. There are some fundamental sources of systematic error though that have to be checked and monitored that can foil this method. Typically the calibration is used to fix the source of the problem. The analytical method is an absolute number-density detection scheme, that is to say it counts the molecules along the beam path in the absorption cell. In order to convert this measurement to a mixing ratio, the cell pressure and temperature need to be accurately measured. The tem- perature measurement is known to be good to +/- 1K which is less than 1%. The pre-measurement checks for pressure are to compare the three manometers in the truck during stopped or very shallow flow to ensure they agree. Periodically the pres- sure heads are calibrated by the factory. They do require as much warm up time as the instrument itself for true readings. The CO measurement was calibrated using a separate gas standard diluted in a dynamic calibration box. The result indi- cates the spectroscopic retrieval is within 4% of the imputed calibration concentration and the methodology has linearity over the range of CO encountered in this test (see Figure E-2). A recent comparison of this instrument chassis and laser at a separate measurement campaign (not ACRP 02-03a related) of our calibration standard to one provided by collaborators from the University of California, Berkeley was excellent. They used a CO standard cylinder from Scott-Marin (a supplier) with a NIST-traceable uncertainty of 1%. QCL measurements (incorporating the correction indicated by calibrations with our CO source) of a sample of their 10.21 ppm calibration gas read 10.35 ppm—a difference of only 1%. This supports the accuracy of our calibrated CO measurements and implies the calibration procedure accounts for systematic biases in the components of the instrument including imperfections in our laser. The pulsed quantum cascade lasers in this instrument have a broader linewidth compared to the cw lasers, and the laser lineshape is not always symmetric. Any multi-mode char- acter in the laser would also contribute to the QCL measure- ment being low. We conclude that the overall spectroscopic certainty in the analytical method of TILDAS applied to the detection of CO is accurate to 4%. A new laser device that operates at a more favorable wave- length for the detection of C2H4 was acquired just prior to MDW 2009. The old laser was calibrated against gas chro- matography during an eight week campaign and found to be within 3% of those measurements. The ethene measurement that will be done under this work has been calibrated against a standard cylinder of ethene, similar to the calibration of CO and NO. The same compressed gas cylinder contained the standards for methane, acetylene, and ethylene, thus cali- brations for all three species were executed simultaneously. Figure E-3 shows the excellent agreement (1%) between the Figure E-2. In field calibration procedure used for CO. Figure E-3. The measured ethene mixing ratios are plotted vs the concentration of the diluted standard. The two sets of points represent two different dilution schemes (grey squares = dual gilibrator dilution; red triangles = automatic dilution calibration system)

91 QCL instrument’s internal spectroscopic quantification of ethene and the calibration mixture prepared from the stan- dard cylinder. The accuracy of the HCHO measurement (using QC- TILDAS) is checked by sampling the output of a well- characterized permeation tube at a known dilution rate. The permeation flow stability requirement is high. As a result, these characterizations are typically performed when ample time is available to stabilize the output of the device. Because the on-wing aircraft testing involves being hosted by an airline, we often do not have time in the field to wait for stabilization. Thus the formaldehyde perm source char- acterization for the MDW 2009 test was performed prior to the test at Aerodyne. Similarly, the instrument was checked during and subsequent to the MDW 2010 sampling cam- paign. We have no evidence of any systematic drift in the verification of the formaldehyde instrument. The procedure is described in the following example. The permeation flow was diluted into a total flow of 3.11 LPM as measured with a recently calibrated gilibrator flow meter. The reported concentration at cell pressures of both 26 Torr and 14 Torr was 17.73 ppbv. The calculated mixing ratio, based on a permeation rate of 62 ± 8 ng/min, atmospheric pressure of 735 Torr and ambient temperature of 298 K, is {1.24E15 molecules/min / 3.11 liters/min} × {1 liter/1000 cm3} / 2.38E9 molecules/cm3 = 1.675E-8, or 16.75 ppb. The difference of 0.98 ppbv out of 16.75 ppbv is 5.9%, well within the uncertainty of the HCHO permeation rate (12.9%). The overall accuracy of the HCHO measurement is 7%. A second permeation rate using a different (lower range) HCHO device was also used to verify the spectroscopic cer- tainty in the TILDAS/QCL approach. The result of this inde- pendent test is depicted in Figure E-5. The data in Figure E-6 shows results of two calibrations performed after MDW 2010; the red triangles using the flow meter pair in the dilution box and the orange diamonds using two gilibrator measurements of flow to determine dilution levels. The agreement between both methods is excellent and values of methane measured with the spectrometer have not been adjusted by the factor 1.02 suggested by the calibration procedure. This provides a citable calibration based strictly on the well-researched and documented spectroscopy behind the HITRAN database (Rothman et al. 2008). The overall systematic uncertainty in the methane instrument is esti- mated to be 6%, based in part on the linestrength literature for these methane absorption lines and in part on the agree- ment with the independent calibration performance check documented here. Proton Transfer Reaction and NO+ Mass Spectrometry Method Overview Proton transfer reaction mass spectrometry (PTR-MS) is a chemical ionization mass spectrometry technique that utilizes H3O+ as the principal reagent ion. H3O+ reagent ions are generated in an external hollow cathode ion source through direct ionization of water vapor. These reagent ions are electrostatically injected into a drift tube reaction region where they merge with the gas to be sampled that has been reduced in pressure (~2 mbar). The drift tube reaction region is formed by a series of concentric stainless steel rings com- pressed between Viton or Teflon o-rings, which serve to elec- trically isolate the drift rings and provide a vacuum seal. The drift rings are electrically connected via a series of resistors. An electric potential applied to the top of the drift tube creates 25 20 15 10 5 0 R ep or te d [H CH O] (p pb v) 12:21:00 PM 7/25/2010 12:22:00 PM 12:23:00 PM 12:24:00 PM Figure E-4. The “top-hat” appearance of the time-series data shown in the figure above results from sudden introductions of the permeation tube flow (HCHO standard) into the total instrument flow.

92 a uniform electric field which transports any positive ions through the drift tube. H3O+ reagent ions within the drift tube are pulled through the sample gas by the electric field where they will react upon collision with any molecule hav- ing a proton affinity greater than that of water. It is important to note that the primary components of air: N2, O2, Ar, CO2, and the alkanes all have proton affinities less than water and thus do not react with H3O+. Most other organic substances except for acetylene and ethene react with H3O+ via a proton transfer reaction, reaction 1. H O R RH H Ok3 2+ ++ → + ( )R1 The proton transfer reaction forms the protonated molecule RH+, which is a stable reaction product in many cases. The drift tube reaction region is terminated by a plate that contains a small aperture through which a fraction of the unreacted reagent ions and product ions are extracted, focused into a quadrupole mass spectrometer and detected using a second- ary electron multiplier. The resulting mass spectrum contains quantitative information regarding the composition of the gas sample, providing that the composition of the sample is known or can be deduced. Figure E-5. Independent Calibration of HCHO Spectroscopy. The Figure depicts a spectrum collected during an independent check of the permeation rate of a second calibration device. The raw agreement between the catalytic conversion method and the online spectroscopic method is 6% for this trial. This verifies the quoted accuracy of the overall TILDAS-QCL approach. methane based on gilibrator measured dilution of 470 ppmv standard (ppm) Figure E-6. The results from the QCL measurement of methane are plotted vs. the expected mixing ratio determined by dilution of the methane standard. The dilution levels for the points noted as orange diamonds were generated using two gilibrator based flow measurements. The dilution levels for the points noted by the red triangles were determined using the in-lab dilution and calibration system.

93 The PTR-MS can be adapted to employ NO+ as the reagent ion by switching the hollow cathode source gas from water vapor to dry air. NO+ reacts via a charge transfer reaction, reaction 2, with compounds having ionization energies lower than 9.26 eV: NO R R NOk+ ++ → + ( )R2 NO+ provides greater measurement specificity as only the dienes and the aromatic hydrocarbon exhaust components fit this criterion. This technique was developed specifically for the measurement of 1,3-butadiene. Quantification Quantification of the PTR-MS ion signals is possible directly from first principles, but is most reliably done via calibration with certified gas standards. In this test the concentrations reported were evaluated from calibrated response factors, except for naphthalene for which a calibration standard was not available. Naphthalene was quantified using a sensitivity factor based on a surrogate (discussed below). The standard equation for quantifying a target compound, designated generically as (R) is shown in equation PTR-1. R I I X I S RH H O R H O H O R [ ] = +       + + +( )3 3 2 106 T P300 2 2 2    ( )PTR-1 The term [R] represents the concentration of R in ppbv. SR is the sensitivity factor expressed as normalized counts per second (ncps) per ppbv T and P represent the drift tube temperature and pressure respectively that are referenced to a standard condition. The term I I X I RH H O R H O H O + + ++    ( ) •106 3 3 2 represents the product ion response expressed in ncps, which is the mass spectral intensity of RH+ measured in cps per 1 million reagent ions. This normalization step accounts for any variation in the product ion intensity resulting from changes in the reagent ion intensity. The intensity of H3O+ is too large to measure directly and its intensity is determined by measurement of the O-18 iso- tope of this ion detected at m/z 21, which is then multiplied by 500 to correct for the isotopic dilution. Measurement of the intensity of H3O+(H2O) is measured directly at m/z 37. Some components react with both H3O+ and H3O+(H2O) while others do not. The XR term is a factor between 0-1 that accounts for the reactivity difference between H3O+(H2O) and H3O+ towards R. Quantification with the NO-MS is accomplished in an analogous fashion as shown in equation NOMS-2. R I I S T P R NO R [ ] =        + + 10 300 2 16 2 .  2 ( )NOMS-2 The term [R] represents the concentration of R in ppbv. SR is the sensitivity factor expressed as normalized counts per second (ncps) per ppbv T and P represent the drift tube tem- perature and pressure respectively that are referenced to a standard condition. Figure E-7. Schematic of the PTR-MS.

94 Calibration Calibrations were performed by dynamically diluting cer- tified gas standards with dry compressed air or VOC free air generated by passing ambient air through a heated Pt cata- lyst. Several gas standards were employed. The 1,3-butadiene calibrations were performed using single component stan- dard, Scott Specialty Gases. Sensitivity factors for all other components were evaluated using a multi-component stan- dard (Apel-Reimer) owned by MSU. The stated accuracy of standards is +/- 5%. Table E-1 provides composition infor- mation on the gas standards used. Flows of the gas standards and the dilution gas were con- trolled using mass flow controllers. The outflow from the mass flow controllers was mixed together and delivered to the sample inlets of the two PTR-MS instruments and GC/FID. In all cases, the PTR-MS instruments and the GC/FID were all calibrated using the same gas mixtures. Calibration checks were performed daily. Day to day varia- tions were within +/-10% and attributed to statistical vari- ability of the method and an average sensitivity factor was employed to compute the reported concentrations. The calibration factors employed in this study are summarized in Table E-2. Gas Standard Identifier Components Concentration Scott Specialty Gas 1,3-butadiene in N 10.1 ppmv MSU-multicomponent methanol in N 520 ppbv acetonitrile in N 520 ppbv propene in N 480 ppbv acetaldehyde in N 490 ppbv acetone in N 500 ppbv isoprene in N 440 ppbv methacrolein in N 410 ppbv benzene in N 510 ppbv toluene in N 500 ppbv styrene in N 480 ppbv p-xylene in N 480 ppbv 1,2,4-trimethylbenzene in N 480 ppbv alpha pinene in N 2 2 2 2 2 2 2 2 2 2 2 2 2 2 410 ppbv Table E-1. Gas standards employed in the calibration of the PTR-MS and GC/FID during ACRP 02-03a. MDW 2009 DAL 2009 MDW 2010 Compound NO-MS PTR-MS PTR-MS NO-MS PTR-MS SR SR XR SR XR SR SR XR acetaldehyde --- 24.2 1 27.7 0.34 10.3 1 1,3-butadiene 14.2 --- --- 12.0 --- benzene 15.2 14.7 1 22.2 -0.21 13.0 6.3 1 toluene 20.6 15.7 1 29.2 0.045 17.6 6.2 1 styrene 25.2 --- 27.4 0.35 21.0 6.1 1 p-xylene 27.4 --- 31.1 0.33 22.3 5.5 1 naphthalene 25.5 --- 32.0 0.37 20.7 4.5 1 Table E-2. Compound response factors for the PTR-MS and NO-MS instruments during ACRP 02-03a. The PNL PTR-MS was deployed at MDW 2009 and MDW 2010. The MSU PTR-MS instrument was deployed as the NO-MS at MDW 2009 and MDW 2010 and as a PTR-MS at DAL 2009.

95 Estimation of Uncertainty The PTR-MS technique does not have any official adopted protocol for evaluation of measurement uncertainty. Mea- surement uncertainty arises from two main sources: 1) the purity of the ion signal (i.e., does all of the signal arise from a singular compound or from multiple compounds) and 2) uncertainties in the measured ion intensities and the cali- bration response factors SR and XR. With respect to 1, it must be recognized that the interpretation of the mass spectrum of the PTR-MS or the NO-MS is critically dependent on the sample matrix. The ions used to monitor the compounds reported in this study have been evaluated extensively for the PTR-MS (Knighton et al. 2007; Wey et al. 2006) and to a lesser extent for NO-MS (Knighton et al. 2009), which is a newer technique. Except for naphthalene, a GC coupled to the PTR-MS (Timko et al. 2011) or the NO-MS (Anderson et al. 2010) has been used to verify the ion assignments in the jet engine exhaust matrix. The sample introduction system for our GC does not efficiently transfer semivolatile compounds (i.e., naphthalene) to the GC column. The cumulative body of our work on jet engine exhaust supports that acetaldehyde, 1,3-butadiene, benzene, toluene, and styrene can be reliably measured within the aircraft exhaust matrix and no addi- tional discussion of these compounds is warranted. The determination of the C2-benzenes (sum of ethylben- zene and xylene isomers) and naphthalene requires additional discussion. The PTR-MS measurement at m/z 107 actually reflects the sum of the C2-benzenes plus benzaldehyde (C7H6O), because the quadrupole mass spectrometer lacks the resolution to separate these compounds. This result is illustrated in the accompanying Figure E-8, which shows the GC/PTR-MS trace of m/z 107 taken from a jet aircraft engine burning JP8 fuel. The peaks have been annotated to identify the C2-benzene isomers and benzaldehyde. The distribution (relative peaks areas) of these species has been consistently observed in other GC/PTR-MS measurements and agrees well with the exhaust profile reported in SPECIATE. This result in concert with our observation of near-idle VOC emis- sions scaling has been used to adjust the PTR-MS measure- ments made in DAL 2009 to reflect only the C2-benzenes. The PTR-MS measurements in DAL 2009 were scaled based on the SPECIATE distribution that the C2-benzenes represent 57% of the measured m/z 107 signal. This has been done so that the PTR-MS results can be compared directly to the mea- surements made with the NO-MS. NO+ does not react with benzaldehyde and the NO-MS instrument provides a direct measure of only the C2-benzenes. The attribution of naphthalene to the m/z 129 (PTR-MS) and m/z 128 (NO-MS) measurements comes with greater uncertainty, as these results have not been confirmed by independent GC evaluation. Furthermore, the lack of calibration standard for naphthalene increases the uncer- tainty in the absolute quantification. There are no known or anticipated interferences to the determination of naph- thalene with the PTR-MS (Knighton et al. 2007). The NO+ determination is less certain as it is known that this ion can react with 1-heptene by an association reaction and with the larger alkenes (>C7) via an insertion mechanism to pro- duce ions at m/z 128 (Diskin et al. 2002). The electric field employed in PTR-MS and NO-MS techniques strongly discriminates against association reactions so the inter- ference from 1-heptene is expected to be minimal. Less is known about how the electric field will affect the insertion reaction channels. Quantification of naphthalene is based off the calibrated response factor of 1,2,4-trimethylbenzene. 1,2,4-trimethylbenzene is chosen as surrogate for naphtha- lene because it is similar in molecular weight, has comparable Figure E-8. GC/PTR-MS trace of m/z 107 for a jet engine burning JP8 fuel.

96 reaction rate constants and is present in calibration gas. It is recognized that the naphthalene determination is heavily caveated and that these measurements should not be over interpreted. We believe the naphthalene measurement is sufficiently valuable to justify its inclusion in this data set. Naphthalene is the only semivolatile compound present in jet engine exhaust at sufficient levels to be assessable by either the PTR-MS and NO-MS. As a semivolatile it will be the first to condense onto the sample lines or existing particles and thus serves as a valuable marker for evaluating the integrity of the extraction probes and the associated sample lines. The precision of the sensitivity factors derived from the individual calibration experiments generally falls within the +/-10% level. Evaluation of the uncertainty in the ion intensity measurements requires defining a specific time base. A single 1-second measurement will have a greater uncertainty than a time averaged series or ensemble of data points. Rather than specifically deriving an uncertainty value it is more appropriate to provide an estimate. Since Poisson statistics governs the variability in the ion intensity measurements the noise in the ion signal scales in propor- tion to the magnitude of the response. This means the relative uncertainly remains essentially constant and independent of sample concentration. Assuming a 10% uncertainty to both the ion intensity measurements and the calibration fac- tors leads to an overall uncertainty of approximately 15%. The uncertainty will be greater for the determination of the C2-benzenes since this determination is for a collection of isomers. The p-xylene calibration factor has been applied for the quantification of the C2-benzenes. Naphthalene has the greatest uncertainty since it employs a surrogate sensi- tivity factor. The uncertainty is naphthalene measurement is estimated to be 30%. Comparison of the PTR-MS and NO-MS measurements at MDW 2010 Four compounds—benzene, toluene, styrene, and naphthalene—were monitored and quantified by both the PTR-MS and NO-MS techniques during the MDW 2010 dedicated engine tests. Figure E-9 depicts the results of side-by-side comparison and illustrates that both instru- ments provided reproducible results. All of the measure- ments appear to be well correlated, but show poorer quanti- tative agreement than was anticipated. The PTR-MS data are ~30% higher than the corresponding NO-MS measurements Figure E-9. Comparison of PTR-MS and NO-MS measurements at DAL 2010.

97 for benzene, toluene and styrene. The naphthalene measure- ment shows a different result where the PTR-MS measure- ment is ~40% lower than that of NO-MS. This latter result is suggestive that the NO-MS determination may suffer from interferences due to the presence of the 1-heptene and longer chain alkenes as described above. The consistent offset between the PTR-MS and the NO-MS for benzene, toluene and styrene is suggestive of a calibration problem rather than a measurement difference. While a calibration difference would be an easy explanation, this seems unlikely as both instru- ments were calibrated in concert using the same calibration gas and delivery system. There is no clear objective resolution to why the methods differ or to which data is the more accu- rate. In many venues agreement at this level is considered to be excellent. Comparison of the PTR-MS and the GC/FID at DAL 2009 A SRI model 8610C gas chromatograph (GC) using a flame ionization detector (FID) was deployed for the analysis of hydrocarbon exhaust gas emissions. Exhaust gas samples are trapped and pre-concentrated on a solid adsorbent (TENAX). A vacuum pump is used to pull the gas sample through the trap (~100 ml/min) for 2 minutes. At the conclusion of the trapping period, the adsorbent trap is heated and the contents of the trap are injected onto the head of the chromatographic column using a 10-way multiport valve. The chromatographic column, Restex MXT-1, utilizes a temperature program where the temperature is held at 50°C for 4 minutes, ramped at 10°C per minute to 200°C and then held for 4 minutes. Calibration of the detector response, adsorbent trapping characteristic and column retention characteristics of the GC/FID were evaluated by analyzing a diluted mixture of the certified gas standard used to calibrate the PTR-MS. Fig- ure E-10 shows a chromatogram obtained for the analysis of the PTR-MS calibration gas mixture. The components present in PTR-MS calibration gas mixture are tabulated in Table E-3. The peaks that are labeled in Figure E-10 are identified in Table E-3. Separation on the MST-1 column is Figure E-10. Chromatogram of PTR-MS calibration gas mixture. Compound Boiling point (°C) Concentration (ppbV) Peak number propene -48 480 - acetaldehyde 21 490 1 isoprene 34 440 2 acetone 56 500 3 methanol 65 520 - methacrolein 69 410 4 benzene 80 510 5 acetonitrile 82 520 - toluene 111 500 6 p-xylene 138 480 7 styrene 146 480 8 alpha-pinene 157 440 9 1,2,4-trimethylbenzene 169 480 10 Table E-3. Calibration standard component list. Boiling point and concentration in the calibration gas standard.

98 essentially determined by compound boiling point and the elu- tion order was determined on this basis. Note that not all of the compounds present in the calibration mixture are observed in the chromatogram. The FID response to compounds contain- ing heteroatoms is diminished relative to their carbon num- bers and is essentially absent for 1-carbon compounds such as methanol and acetonitrile. Propene does not appear to be retained on TENAX and its peak could not be identified. Standard curves were produced for each of the five aromatic compounds by plotting peak area response versus sample concentration. The standard curve for benzene is shown in Figure E-11. The slope of these plots provides response per unit concentration. The magnitude of the slopes should vary in proportion the number of carbons in the compounds. Table E-4 is a summary of the compound responses for the aromatic compounds. Benzene and toluene show a constant per carbon response as expected. The higher molecular weight compounds exhibit smaller per carbon responses. This result is not expected and may be the result of retention of these less volatile compounds in the sample lines. Quantification of exhaust samples assumes that the peak area/ppbC response factor of 0.73 expressed by benzene and toluene is correct. Two engine exhaust gas samples were analyzed by the GC/FID. One sample was taken during the initial engine warm-up period while the second sample was obtained dur- ing test point 16 (ground idle, zero bleed). The resulting chro- matograms from these samples are shown in Figure E-12. The two chromatograms are very similar and show essentially Figure E-11. FID response curve for benzene. compound peak area/ppbV peak area/ppbC benzene 4.38 0.73 toluene 5.12 0.73 p-xylene 5.30 0.66 styrene 4.78 0.60 1,2,4-trimethylbenzene 5.74 0.64 Table E-4. FID response factors. Figure E-12. GC analysis of diluted engine exhaust.

99 the same peaks in the same relative proportions. The benzene peaks are the largest single peaks in both chromatograms. Other identifiable peaks in these chromatograms have been labeled. The integrated peak areas in Figure E-12 can be used to assess the average concentration present in the exhaust stream during the sample period. These results can then be compared to the PTR-MS measurements that were being made simultaneously. The PTR-MS benzene measurements are shown in Figure E-13 and indicate the beginning and end times for the GC sample acquisition. The warm-up sample was taken at approximately 2:35 while test point 16 corre- sponds to the sample taken at approximately 3:25. Table E-5 is a summary of the inter-comparison of the GC/FID and PTR-MS measurements. The PTR-MS measurements reflect the average of the concentrations recorded while the GC sam- ple was being taken. The C2-benzenes represent the sum of ethyl benzene and the three xylene isomers. For the GC/FID measurements the individual peak areas of these compounds have been summed together. The PTR-MS cannot provide a direct measure of the C2-benzenes because benzaldehyde, a prominent component in aviation exhaust is also detected at the same mass-to-charge ratio. The SPECIATE profile indicates that 57% of measured intensity originates from the C2-benzenes with the remaining 43% coming from benzaldehyde. The concentration reported in the PTR-MS column has been adjusted accordingly. Inspection of the results in Table E-5 indicates that for this limited set of com- pounds that the two methods are in excellent agreement. Summary of PTR-MS and NO-MS Measurements The ACRP Project 02-03a involved the deployment of two PTR-MS instruments, MSU PTR-MS and the EMSL PTR- MS, at MDW 2009 and MDW 2010 and one PTR-MS (MSU) at DAL 2009. At MDW the MSU PTR-MS was operated in alternate reagent NO+ mode. These instruments were chal- lenged to operate under a wide range of environmental con- ditions over three campaigns spanning a period 13 months. Overall the level of agreement between the PTR-MS and NO-MS is considered to be excellent. The comparison of the PTR-MS and GC/FID demonstrated excellent agreement. Any study that challenges different instruments risks the pos- sibility that the instruments will deliver different results. Many of these inter-comparison studies reveal that while instruments appear identical certain instrument/operator pairs often suc- ceed where others don’t. It is not without some bias, but we favor the MSU measurements, except for naphthalene, for inclusion in the final data archive. This decision is based in part on intimate knowledge of the MSU instrument and its perfor- mance as well as a demonstrated consistency over a period of many year and campaigns. Table E-6 summarizes the com- pounds measured and the instrumental method used for the data submitted to the final archive. Figure E-13. Time series of PTR-MS benzene and GC-FID benzene. Green continuous trace is the benzene from the PTR-MS and the red bars are the GC-FID quantified benzene where height is indicative of mixing ratio. compound Warm-up Test point 16 PTR-MS FID PTR-MS FID benzene 79.4 79.9 47.6 48.6 toluene 28.9 24.9 17.0 13.5 Table E-5. Inter-comparison of GC/FID and PTR-MS measurements. Compound MDW 2009 DAL 2009 MDW 2010 acetaldehyde PTR-MS PTR-MS PTR-MS 1,3-butadiene NO-MS ---- NO-MS benzene NO-MS PTR-MS NO-MS toluene NO-MS PTR-MS NO-MS C2-benzenes NO-MS PTR-MS NO-MS naphthalene NO-MS PTR-MS PTR-MS Table E-6. Summary of compounds measured and instrument technique employed.

100 Figure E-14. Segment of time series data collected on the second day with the FID zero and span procedure engaged (not shown). Note that the CO (black) and the HCHO (green) are highly correlated. The FID data is also correlated (orange/yellow) but is noisier and is offset due to atmospheric methane and the sum of other urban area hydrocarbons. The time response of the FID is also slightly longer (slower) than the HCHO and CO instruments. Flame Ionization Detector Deployment at MDW 2009 Some special remarks are required on the flame ionization detector (FID) instrument. The FID instrument was operated during the first day in its rack using manual calibration proto- cols recommended in the manual. This instrument operated at its highest gain setting and experienced frequent drifts in the zero signal level, sometimes beyond the scale of the digi- tization range. We found that the switching from sample to calibrate did not always reproduce pressures in the detector. This prompted us to devise a new method that would perform zeroes and calibration without using the internal valves. For the second day of testing, we developed a new zero and cali- bration mode for the instrument. We operated it in sampling mode only with the potentiometers set at middle scale. We over blew the inlet with zero air every five minutes during the test on an automatic timer. Following each zero we over blew the inlet with the propane standard (2.4 ppmC). The frequent “zero” data was used to correct the data in the intervening sampling periods. Each subsequent span was used to scale the intervening sampling period. A time series resulting from this procedure is depicted in Figure E-14. The FID data for the second day of testing is adequate for analysis. For future tests we will improve the instrument in the laboratory and provide more temperature and vibration isola- tion when installed in the truck. Qualitatively, a portion of the noise of this instrument was due to people walking in the truck, which suggests it can be fixed with internal improvements. Despite its widespread use, there are several limitation of the FID instrument measurement approach when character- izing a complex mixture of hydrocarbons. For example, the FID instrument does not employ any chromatographic separation of compounds. The instrument has unknown response factors to compounds containing heteroatoms and no response to formaldehyde. Essentially, for each oxygen in the molecule, the instrumental response to one carbon is removed. This means the instrument does not really count all of the carbon in a par- tially oxidized hydrocarbon. Furthermore, the signal level can only be calibrated in units of carbon atoms. In order to express the output from this instrument as mass, frequently the mass of methane is used; or it reports methane equivalent emissions. This campaign represented the first opportunity to com- pare the FID response to that of the spectroscopic HCHO measurement when the sample was diluted into ambient air. The correlation of HCHO with FID response has been used to relate the results of several tests to the ICAO databank mea- surements. For example in Herndon et al. (2006), the range of 0.21–0.26 g HCHO per FID g methane was used to estimate the relationship between taxiway operational measurements and the ICAO databank. An alternative method using the

101 hydrocarbon profile of Spicer et al. (1992, 1994) and a syn- thetic estimate of the FID response to various hydrocarbons suggests a value of 0.22 g HCHO per FID gram as methane. The comparison derived from the measurements in this test suggests this ratio is 0.2 g HCHO per FID gram as methane, which agrees very well with the previous estimates. References Anderson, B. E., et al. 2010. Alternative Aviation Fuel Experiment (AAFEX) Rep. Baniszewski, D., D. Martin, and J. DeLeon. 2010. Petroleum Quality Information System 2009 Annual Report Rep. Diskin, A. M., T. S. Wang, D. Smith, and P. Spanel. 2002. A selected ion flow tube (SIFT), study of the reactions of H3O+, NO+ and O2+ ions with a series of Alkenes; in support of SIFT-MS. International Journal of Mass Spectrometry 218: 87–101. Herndon, S. C., M. S. Zahniser, D. D. Nelson, Jr., J. H. Shorter, J. B. McManus, R. Jimenez, C. Warneke, and J. A. de Gouw. 2006. Air- borne measurements of HCHO and HCOOH using a pulsed quan- tum cascade laser system. Accepted J. Geophys. Res. Jimenez, R., S. Herndon, J. H. Shorter, D. D. Nelson, J. B. McManus, and M. S. Zahniser. 2005. Atmospheric trace gas measurements using a dual-quantum cascade laser mid-infrared absorption spec- trometer. SPIE Proceedings 5738(37): 318–331. Knighton, W. B., E. C. Fortner, S. C. Herndon, E. C. Wood, and R. C. Miake-Lye. 2009. Adaptation of a proton transfer reaction mass spectrometer instrument to employ NO+ as reagent ion for the detection of 1, 3-butadiene in the ambient atmosphere. Rapid Comm. in Mass Spec. 23(20): 3301–3308. Knighton, W. B., T. Rogers, C. C. Wey, B. E. Anderson, S. C. Herndon, P. E. Yelvington, and R. C. Miake-Lye. 2007. Quantification of Air- craft Engine Hydrocarbon Emissions Using Proton Transfer Reac- tion Mass Spectrometry. Journal of Propulsion and Power 23(5): 949–958. McManus, J. B., D. D. Nelson, J. H. Shorter, R. Jimenez, S. Herndon, S. Saleska, and M. Zahniser. 2005. A high precision pulsed quantum cascade laser spectrometer for measurements of stable isotopes of carbon dioxide. J. Modern Optics 52: 2309–2321. Nelson, D. D., J. S. Shorter, J. B. McManus, and M. S. Zahniser. 2002. Sub-part-per-billion detection of nitric oxide in air using a thermo- electrically cooled mid-infrared quantum cascade laser spectrometer. Applied Physics B 75: 343–350. Nelson, D. D., J. B. McManus, S. Urbanski, S. Herndon, and M. S. Zahniser. 2004. High precision measurements of atmospheric nitrous oxide and methane using thermoelectrically cooled mid- infrared quantum cascade lasers and detectors. Spectrochimica Acta A 60: 3325–3335. Sachse, G. W., G. F. Hill, L. O. Wade, and N. Y. Chou. 1987. Airborne tunable diode laser system for rapid CO and CH4 measurement (A). J. Opt. Soc. Am. A 4: 54. Timko, M. T., S. Herndon, E. de la Rosa Blanco, E. Wood, Z. Yu, R. C. Miake-Lye, W. B. Knighton, L. Shafer, M. DeWitt, and E. Corporan. 2011. Combustion Products of Petroleum Jet Fuel, a Fischer Tropsch Synthetic Fuel, and a Biomass Fatty Acid Methyl Ester Fuel for a Gas Turbine Engine. Combust. Sci. Technol. in press. Wey, C. C., et al. 2006. Aircraft Particle Emissions Experiment (APEX). NASA/TM-2006-214382, ARL-TR-3903. Zahniser, M. S., D. D. Nelson, Jr., J. B. McManus, and P. L. Kebabian (1995), Measurement of trace gas fluxes using tunable diode laser spectroscopy, Phil. Trans. R. Soc. Lond. A, 351, 357–369. Figure E-15.

Abbreviations and acronyms used without definitions in TRB publications: AAAE American Association of Airport Executives AASHO American Association of State Highway Officials AASHTO American Association of State Highway and Transportation Officials ACI–NA Airports Council International–North America ACRP Airport Cooperative Research Program ADA Americans with Disabilities Act APTA American Public Transportation Association ASCE American Society of Civil Engineers ASME American Society of Mechanical Engineers ASTM American Society for Testing and Materials ATA American Trucking Associations CTAA Community Transportation Association of America CTBSSP Commercial Truck and Bus Safety Synthesis Program DHS Department of Homeland Security DOE Department of Energy EPA Environmental Protection Agency FAA Federal Aviation Administration FHWA Federal Highway Administration FMCSA Federal Motor Carrier Safety Administration FRA Federal Railroad Administration FTA Federal Transit Administration HMCRP Hazardous Materials Cooperative Research Program IEEE Institute of Electrical and Electronics Engineers ISTEA Intermodal Surface Transportation Efficiency Act of 1991 ITE Institute of Transportation Engineers NASA National Aeronautics and Space Administration NASAO National Association of State Aviation Officials NCFRP National Cooperative Freight Research Program NCHRP National Cooperative Highway Research Program NHTSA National Highway Traffic Safety Administration NTSB National Transportation Safety Board PHMSA Pipeline and Hazardous Materials Safety Administration RITA Research and Innovative Technology Administration SAE Society of Automotive Engineers SAFETEA-LU Safe, Accountable, Flexible, Efficient Transportation Equity Act: A Legacy for Users (2005) TCRP Transit Cooperative Research Program TEA-21 Transportation Equity Act for the 21st Century (1998) TRB Transportation Research Board TSA Transportation Security Administration U.S.DOT United States Department of Transportation