Oxygenated VOCs as significant but varied contributors to VOC emissions from vehicles. Preprint acp-2022-130

Overview

This manuscript characterizes gaseous emissions from a number of vehicles meeting a wide range of Chinese emissions standards, to include gasoline, diesel, and liquified petroleum gas (LPG), as measured using a chassis dynamometer setup. Measurements are primarily presented for those using a PTR-ToF-MS, and included canister sampling with GC-MS/FID analysis and a few species by Iodide CIMS, along with common measurements (CO₂, etc.) using a portable emissions measurement system.  Oxygenated VOC’s (OVOC) are indicated to be molecules with less than 18 carbons.

This work shows the strong influence of OVOC in diesel exhaust (>50% by mass) compared to a much smaller influence in gasoline vehicles (~15%). Clear differences between cold-start and hot-start emissions are also observed, notably they are much more significant for gasoline vehicles than for diesel vehicles, and aromatics and OVOC had similar temporal profiles. Some ratios of emissions (e.g. toluene to larger aromatics) are unique between gasoline and diesel vehicles, and are suggested as potentially useful for emissions attribution.

Overall the work as presented is quite thorough, and the intended goals of the work are clearly made. The insights from the work are a good contribution to the field. There are a few details that should be addressed, however prior to suitability for publication, notably quality control.

General Comments

It is concerning that the agreement between canister with GC-MS/FID and PTR-ToF measurements for toluene are so disparate in more than 20% of the tested vehicles, as shown in Figure S6c. These discrepancies are essentially ignored in the manuscript. How can canister measurements be near-zero while PTR-ToF measurements are 250 mg/km, and vice versa? Perhaps this might cold occur for more exotic species, but for toluene I would expect agreement at least within a factor of 2 in all cases, as it is a high volatility species that is easily ionized in PTR and observed with GC-MS/FID. Perhaps in some cases one or the other measurement was not made and simply reported as zero? This issue should be clarified. Furthermore, agreement between generally accepted canister measurements and PTR-ToF measurements must be reported for a wider variety of species, to include oxygenated species and larger aromatics.

The mileage of the vehicles tested is quite variable, are there any correlations in your data with mileage, are these different for gasoline vs. diesel?

Was any analysis of the fuels done? To make clear sense of the emissions, the compositions of each fuel type, in terms of saturates (linear and cyclic), aromatics (BTEX and others), and oxygenates should be given. This is especially important for the diesel fuel, which can vary significantly in terms of aromatic content. Were the fuels summer or winter blends? The results presented have much narrower significance without clearer information on the fuel composition. Did the gasoline fuel have any ethanol content, as might be expected for gasoline in China after 2017? Ethanol content will have significant effects on small OVOC emissions. The discussion beginning on line 377 is well explained by the difference in aromatic content of the two fuels.

When considering the usefulness of ratios between emitted species as diagnostic for diesel vs. gasoline species, you should also consider their atmospheric lifetimes for oxidation.

Specific Comments

Figure 9. The fits to your data are poorly presented this way, either for predicting the values through the whole range or for giving physical insight. Perhaps you should make the axes linear rather than logarithmic. At the very least, you should explain that the strange curves to these linear fits in log-log space are due to the y-intercept, or perhaps only plot these fits in the region where they appear linear (where the intercept is small compared to the fit value) and note that you plot only in the region of reasonable fit.

Please review again thoroughly for grammar. A few corrections are:

Line 224 “species emitted by vehicles”

Line 377 “Comparing gasoline and diesel vehicles,”

Line 445. “can be determined”

Line 486. “Substantially larger”

Wang et al. present an analysis of VOC emissions measured from vehicle dynamometer testing for vehicles designed under different emission standards (China I - IV). The authors evaluate total and speciated VOC emissions from both gasoline, diesel, and LPG under a variety of conditions (cold start, warm start, speed, etc). The authors detail the different emission factors between each vehicle, and observe a distinct difference between the OVOCs emitted by gasoline and diesel engines. The latter produces significantly higher fraction of OVOCs than by gasoline, which appears to be at least partly associated with the pollution control technology.

I found the paper to be very well-written, well-reasoned, and full of good information. I appreciated the study as a nice piece of work describing fossil fuel emissions from motor vehicles in China.

My only substantive comment is that I don’t have a sense of the fuel composition and how this might contribute to the high OVOCs observed in diesel exhaust. And how do the OVOC emissions compare against diesel exhaust studies reported elsewhere? Gentner et al. (2013) also see elevated OVOC emissions in diesel compared to gasoline. Are these differences comparable to what is observed here, or is there something different between the aftertreatment or fuels that could contribute to any differences?

Comments:

Line 70 - 71 Based on the reference, I presume that the authors are specifically noting the decline of VOCs in urban regions in China? For clarity, I would suggest re-writing this sentence to say “Furthermore, VOC emission significantly decreased in China due to stricter emission standards.”

Line 76: Could the authors provide some context on the China VI emission standard? I recognize that the standard is dependent on power ranges, but a few sentences on VOC emissions at max power output would be useful. This would also be useful in the methods (lines 112 - 122) to give readers context as to what the China I - IV standards represent in terms of VOC emissions.

Line 81: Would suggest modifying “group” to say “class of compounds”

Line 80 - 83: Are the authors primarily discussing VOC measurements from dynamometer studies, or tunnel studies, or ambient studies? I think the distinction matters given that results from laboratory, tunnel, or ambient measurements can be interpreted differently given differences in co-emitted sources that can convolute the measured signal from tailpipe emissions

Lines 268 - 271: Are there also differences in the aftertreatment that might lead to higher OVOC emissions? The authors note the temperature of the device at line 240, and I’m curious if previous work has looked at VOC speciation under different aftertreatment conditions.

Figure 1: It would be useful to see the acronyms (LDDT, MDDT, HDDT, and BUS) defined in the caption as a reminder to the reader.

Title of Section 3.2: The title doesn’t quite reflect the discussion that follows. Might I suggest “Analysis of PTR-ToF-MS mass spectra to evaluate VOC speciation”?

Lines 320 -323: This is a nice result, and partially addresses my question at lines 268-271. Could the authors point to this figure and discussion to demonstrate that the changes to the VOC distribution isn’t significantly different between cold start and normal operation?

Lines 424 - 425 : I like the discussion in this section on using the aromatics to delineate between diesel and gasoline. I agree with the authors that these ratios might be difficult to assess in the ambient owing to additional sources of aromatics (e.g. solvent emissions) and secondary production of formaldehyde and acetaldehyde. Are there any unique masses, with high enough signal in ambient air, that could be used to more definitively separate gasoline vs diesel emissions? I also wonder if ratios to CO or other combustion markers might be insightful.

Figure S6. The intercomparisons are nice for the fast time-resolution systems, but there are significant differences between the GC and PTR for toluene - is this due to differences in sampling techniques (e.g., grab sampling artifacts vs real-time sampling), or something due to fragmentation in the PTR to produce a signal at m/z 93? I believe the other reviewer also commented on this, and I agree that some explanation is warranted here.