Grantee Research Project Results

Here, we summarize the main results and achievements under this grant, organized according to activity/objective.

 

1a. Studies of aerosol phase chemistry from BVOC oxidation. In the first year of this project, we proposed to begin smog chamber experiments with α-pinene, and begin analysis of the aerosol phase composition, using DESI-MS. To establish our methods, and to begin work on a simple system, we first studied the oligomerization of methacrolein under acidic conditions. We studied the nature of the oligomer produced in the condensed phase reaction of methacrolein with itself, in the presence of H₂SO₄, as a simple model system involving a multifunctional atmospheric OVOC. Oligomerization produces a highly complex oligomeric material, whose structure is impacted by oxidation, hydrogenation and hydrolysis reactions involving the presence of sulfuric acid. A DESI-mass spectrum of a methacrolein oligomer is shown in Figure 1, which shows a number of oligomers, for up to 11 methacrolein units in the chain. DESI-MS analysis of the oligomeric material shows evidence for products arising from Diels-Alder cycloaddition reactions (e.g., as shown in Figure 2), which have not been previously reported for atmospheric SOA. In addition, H₂SO₄ acts to initiate oxidation and hydrolysis reactions that influence the composition of the aerosol. Interestingly, there also are H₂ addition reactions (i.e., reduction) simultaneously occurring in the oligomeric mixture. Several structures and mechanisms have been identified/developed.

 

Figure 1. DESI-MS specttrum of MACR oligomer. 

 

 

 

 

Figure 2. Potential diels-Aider chemistry

 

 

 

 

As a result of reviewer comments related to our submitted paper, we have been focusing on the observation of organosulfates in cloud water. This was done using the cloud water collector on our aircraft (see Figure 3), the Airborne Laboratory for Atmospheric Research.

 

 

 

 

 

 

 

 

 

 

Figure 3., Cloud water collector on the Purdue Airborne
Laboratory for Atmospheric Research (ALAR).

 

To investigate the extent to which oligomeric structures might be present in the condensed phase in isoprene-impacted atmospheres, cloud water samples collected from the Missouri Ozarks site were analyzed using ESI-MS (DESI-MS analysis yielded few peaks significantly above those in the blank samples). ESI spectra from a cloud water sample (at 1.8 km) analyzed directly from the solution phase are shown in Figures 4 and 5 below for positive and negative mode, respectively. Peaks marked with red arrows are those not present in the blanks. These spectra provide unambiguous evidence that oligomeric material is present in cloud water, as shown previously for rainwater. Intensities were sufficient to determine if organosulfates were present by neutral loss experiments and those peaks are marked by blue arrows (Figure 5). Organosulfates have been previously observed in rainwater using ESI and high resolution mass spectrometry (Altieri et al., 2008). The parent ion scan of m/z 97 (HSO₄^-) did not yield any significant peaks. These spectra are atypical of soft ionization spectra of smog chamber or ambient SOA samples in that there are several individual species with high abundance, rather than envelopes where a peak is present at every mass. CMAQ modeling is being utilized to further test the hypothesis of the relative contribution of both gas- and aqueous-phase biogenic VOC oxidation, compared to anthropogenic VOC oxidation, leading to SOA formation in the area sampled by ALAR. Preliminary results show that isoprene and monoterpene oxidation products dominate the SOA production at the altitude sampled by ALAR, and current model studies involve comparison of days with and without clouds in the model.

 

 

 

 

 

 

 

 

 

 

Figure 4. (+) EST-MS spectrum of cloud water from 
7,000 ft. Peaks labled with red arrows are absent
from the blank.

 

Figure 5.  (-) ESI-MS spectrum of cloud water from 7,000
ft. Peaks labled with red arrows are absent from the blank. 
Peaks labled with blue arroows are not in the blank but
are present in the 80 Da neutral loss spectrum which
indicates that these species are not organosulfates.

 

1b. Development and Testing of the Proton Transfer Reaction Linear Ion Trap (PTRLIT). An objective of this work was to apply a newly developed technology, the PTRLIT, to studies of terpene oxidation and the nature of oxidation products that may undergo gas-to-particle conversion. While the PTRLIT was developed with methyl vinyl ketone (MVK) and methacrolein (MACR) as test cases, it had not been previously evaluated for terpenes and their oxidation products. We have completed a series of experiments in which we evaluated the ability to trap and selectively detect terpene isomers through collision-induced-dissociation (CID) experiments. The PTRLIT was intercompared with a conventional triple quadrupole mass spectrometer. We found that the CID spectra are very similar in the two cases. However, the PTRLIT enables ion chemistry in the trap for pursuit of improved selectivity. These results are described in our paper, Müller et al., 2009. Work for the future will involve use of reagent gases in the trap, for example, vinyl methyl ether (Colorado et al., 1998), that could be used to improve the selectivity of the PTRLIT for terpenes and sesquiterpenes. During the summer of 2008, we conducted measurements of a variety of VOCs in the atmosphere above the mixed deciduous/coniferous forest at the University of Michigan Biological Station (UMBS) using the PTRLIT. The PTRLIT was shown to enable selective detection of the structural isomers MVK and MACR, and those data have established the PTRLIT as a new method that takes the PTRMS instrument to the next level in selectivity and sensitivity. Among the things we learned was the extent to which there can be interferences in the conventional PTRMS. A good example is isoprene measurements, regarded as very good via the PTRMS. However, we compare in Figure 6 isoprene determined operating in conventional PTRMS ("scan out") mode, and in PTRLIT ("CID") mode, in which we isolate the parent ion (m/z=69), and then conduct CID on that ion, and quantify isoprene through measurement of the fragment at m/z=41. As shown, while often the apparent agreement between the two determinations is very good, on July 19 and 25, the differences can be as large as a factor of two in mid-day. Possible interferences include furan, and we are in the process of evaluating that possibility. This is discussed in Mielke et al. (2010).

 

Figure 6.  Comparison of PTRMS and PTRLIT modes
of operation of the PTRLIT, for isoprene at UMBS.

 

1c. Studies of aerosol production from the surface of the Great Lakes. During the summer of 2008, we conducted aerosol measurements from the UMBS tower. The data, in which several new particle events were observed, have inspired us to examine the nature of the BVOCs contributing to SOA at the UMBS site. We then pursued the hypothesis that these events result from BVOC oxidation in the near-canopy environment, through measurements of the aerosol profile via SMPS measurements from the ALAR aircraft (see installation in Figure 7), conducted by graduate student, Nate Slade. We conducted a series of vertical profile measurements of aerosol through the boundary layer above the PROPHET tower at UMBS in the summer of 2009.

Figure 7.  ALAR aircraft (top),
with SMPS installed (bottom).

 

What we observed is a maximum in the total particle number in the middle of the boundary layer, consistent with what would be expected from transport of existing particles, and loss by dry deposition to the canopy surface. However, in the process of obtaining control profiles upwind over Lake Michigan, we made a surprising discovery, that there is new particle formation near the lake surface, and the concentration of these small particles depends exponentially on wave height. In Figure 8, we show a sample vertical profile, and aerosol size spectrum for the lowest altitude over the lake, that shows the small particle mode in the spectrum.

 

Figure 8. Vertical profile of temperature and aeorsol
number concentations of Airken mode particles 
(20-40 mm) over UMBS forest (green) and Lake
Michigan (blue) for JUly 26 flight. 

 

We believe that this mode results from evaporation of water from film droplets generated from breaking waves, leaving only the non-volatile ions and organic matter to produce a particle (analogous to the well-understood mechanism for production of sea-salt aerosol from marine waves). We hypothesize that this may represent a mechanism for transfer of non-volatile POPs into the atmosphere. Current activities involve studies of marker compounds in the lake water that might be used to test this hypothesis further. We published the observation and our interpretation in Slade et al., GRL, 2010.

 

1d. Studies of the flux of gases from an urban environment. In the course of testing the winds and aerosol equipment on board ALAR, and testing an idea for examining the urban impact on regional aerosol, we conducted a series of experiments of the CO₂ and CH₄ outflow from Indianapolis, as a test case. That work led to a paper describing a mass-balance method for measuring the urban area-wide emission flux of greenhouse gases. This was presented in the Mays et al., 2009 paper. An example of the measurement data used in this approach is shown in Figure 9 below.

 

 

 

 

 

 

 

 

 

 

Figure 9, Example of CO₂ data from Indianapolis. 

 

1e. Studies of α-pinene oxidation and gas-particle partitioning of the products. In this project, we conducted studies in which α-pinene oxidation undergoes oxidation in our 5,500 liter all-PFA Teflon photochemical reaction chamber. Thus far, these have involved traditional VOC/NO_x irradiations, during which NO, NO₂, NO_y, individual organic nitrates, and total organic nitrates have been measured using a denuder-based method to separate the gas-phase and particle-phase nitrates. The aim is to measure the individual product yields and from the measurements in the two phases, and the total aerosol mass concentration, the partition coefficient, as described in Equation I (by Pankow). O₃ is measured with traditional UV absorption instruments, and particle size distributions and number density are determined using
K_p,i = C_i,p/C_a * C_i,g) = f*RT/(10⁶ * MW_a*ς*p⁰_i)    I

our Scanning Mobility Particle Spectrometer (SMPS), purchased through this grant. Until recently, we have been focused on determination of the terpene nitrate yields (RONO₂), as many recent publications have reported evidence of nitrate functionalities in ambient aerosol from forest-impacted air masses. We have synthesized and purified four of the “terpene nitrates” (TNs) that originate from OH radical addition across the double bond of α-pinene, and are working on methods to ensure that we can quantitatively measure the concentrations of these isomers. Recently, we conducted measurements of the yields of the nitrates in the presence and absence of seed aerosol. A set of preliminary data for the total (gas + particle) yield of total organic nitrates is shown in Figure 10 (with ammonium nitrate seed aerosol).

Figure 10.

 

A plot of C_i,p/C_i,g vs. C_a, for these experiments, is shown in Figure 11. As shown in the Figure, while this should be a linear function of M, it is not, but is instead rather scattered. This may be consistent with recent observations implying that SOA is not liquid-like and that an equilibrium partitioning model may not apply (cf. Perraud et al., PNAS, 2012). Our recent measurements indicate that the majority (62%) of the organic nitrates are in the aerosol phase, even in the absence of seed aerosol. We are in the process of using LC/FTICR-MS to study the extent to which aerosol phase chemistry may be influencing the apparent lack of equilibrium partitioning for the organic nitrates. A distinct possibility is that after uptake, the organic nitrates undergo further processing, rendering them less volatile.

 

1f. 1D model study of the relative importance of various BVOCs to organic nitrate production. The UMBS site represents a unique opportunity to probe our understanding of atmospheric BVOC chemistry, in part because previous measurements of fluxes for isoprene, 2 aromatics, 3 alkanes, 20 monoterpenes (MTs), 8 oxygenated species, and 23 sesquiterpenes, were conducted for this site, by the Detlev Helmig group. So, given literature data for the product yields, for example, for organic nitrates and aerosol, we are able to pursue the question of which BVOCs contribute the most to organic nitrate production, and to aerosol production in this forested environment. This is a particularly intriguing environment, as it is in a successional transition from largely isoprene emitting species to a mix with a substantial increase in terpene and sesquiterpene emitters (white pine). Thus, postdoctoral researcher Kerri Pratt has constructed a 1D model, using eddy diffusivities derived from turbulence data, by Allison Steiner’s group. The model includes 26 vertical layers, from the surface to 4372m above the surface. BVOC concentrations were calculated according to transport and reaction kinetics following equation II, which describes the change in the concentration of VOC i at altitude z and time t (c_i(z,t)): 

 

 

 

 

Figure 12. Comparison of 1D model isoprene
with aircraft measurement data. 

 

VOC terms include emission (E) into the first canopy bin, chemical production (P) and loss (L) based on chemical kinetics, upward and downward vertical fluxes (F_↑ and F_↓, respectively) divided by bin height (h), dry deposition (D), and horizontal advection (H). The (quite good) quality of the vertical mixing approach (and chemistry of BVOC consumption) was evaluated using aircraft vertical profiles for isoprene; a comparison of the aircraft and model-simulated isoprene vertical profile is shown in Figure 12 above. We have completed studies of how the BVOCs are converted to organic nitrates, as one set of important SOA precursors for the BVOCs. A sample set of results is shown in Figure 13, which shows the fractional contribution of organic nitrate production from various processes, as a function of time of day. Perhaps most significant and interesting is that NO₃ chemistry (with terpenes and isoprene) is important during both night and daytime. This result will help motivate appropriate measurements during the upcoming SOAS study in the southeastern United States. We are continuing work with this model now to produce estimates of the relative importance of each BVOC in producing SOA in this environment.

 

Figure 13.  Fraction of the total organic nitrate production rate as 
a function of time of day.

 

1g. Studies of the morning NO_x maximum. For decades, the atmospheric chemistry community has been puzzled about the morning maximum in surface level NO_x concentrations at rural and forest sites. This has been discussed in the literature as being caused by downward mixing upon breakup of the nocturnal boundary layer, followed by downward mixing of polluted air from aloft. We have shown, from measurements at UMBS, that this is not the case; rather, it results from surface emission, for example, from soils, or local to regional scale combustion sources, followed by transport in the stably stratified surface layer. This is very important to BVOC chemistry and production of SOA precursors, as it enhances early morning chemistry, either by enhancing surface-based production of HONO, and thus OH, or by enhancing production of NO₃. Again, this is likely to be important to SOA production in the U.S. Southeast. A paper has been submitted to ACP describing these results.

 

1h. Studies of cloud water microbiology. An important potential processor of cloud water species is microbes. In the summer of 2008, we collected cloud water samples at UMBS, for analysis and speciation of summer time microbes in cumulus clouds, as discussed in our 2007 JGR paper (Hill et al., 2007). We have recently completed all of the analysis. The cloud water was analyzed for the diversity of bacterial phylotypes by denaturing gradient gel electrophoresis (DGGE) and sequencing of 16S rRNA gene amplicons, using a nested PCR approach. DGGE analyses of bacterial communities detected 17-21 bands per sample. Sequencing confirmed the presence of a diverse bacterial community; sequences from seven bacterial phyla were retrieved. Cloud water bacterial communities appeared to be dominated by members of the cyanobacteria, proteobacteria, actinobacteria and firmicutes, as summarized in Figure 14, below.

 

Figure 14. Clasification of sequences from
cloud water samples acquired in 2008.