Impact/Purpose:

There is currently a critical need for real-time methods of organic aerosol particle analysis for use in ambient monitoring, source apportionment, and health effects studies. In this research program, we plan to address this need by developing and applying a thermodenuder-Aerodyne Aerosol Mass Spectrometer (TD-AMS) technique for ambient organic fine particle analysis. The objectives of the project are to (1) construct the thermodenuder and optimize the TD-AMS, (2) use the TD-AMS in laboratory studies to develop a mass spectra-volatility database for the major atmospheric combustion and secondary organic aerosol sources, and (3) apply the database to a TD-AMS study of organic aerosol in the Los Angeles Air Basin.

Description:

1. Thermodenuder Development:

Two TD systems were designed, constructed, and tested at Aerodyne. In this design, the vaporizer consists of a 50 cm long, 1 inch OD stainless steel tube wrapped with three heating tapes and fiberglass insulation and then mounted in a stainless steel housing. Each of the three heating zones is temperature-controlled by a PID hardware controller whose set-points can be set independently from one another to achieve a uniform temperature profile. The system is operated at a flow of 0.6 l min^-1, which corresponds to an effective plug flow residence time in the vaporizer of ~15 s. The denuder consists of a 50 cm long, 0.75 inch diameter wire mesh tube held in place inside a 4 inch OD aluminum tube by end caps attached to the denuder with Viton O-rings to provide an air-tight seal. The volume outside the mesh is filled with activated carbon charcoal. The main difference between this and previous designs is that this vaporizer was designed with additional temperature control and reduced thermal mass to allow rapid temperature stepping for ambient analysis. Rapid stepping allows one to obtain information about a wide range of volatilities on a time scale comparable to or shorter than most aerosol variations in ambient air. One of these TDs was sent to UC-Riverside (UCR) where it was used in the laboratory to develop a method for measuring the vapor pressures of organic aerosol components and for studies of SOA. The other was sent to the University of Colorado-Boulder (CU) where its performance was further evaluated in the laboratory and its design improved upon for ambient analyses. It was then employed in a series of field studies of combustion-derived aerosol and ambient aerosol.

The final CU design employs two fans to shorten the cool-down period, as well as a valve and 3-zone temperature control system in software (Labview and Visual Basic 5.0). The valve system allows for switching between ambient air that either bypasses or passes through the TD for constant time periods. The temperature control system is typically programmed to either step through several temperatures repeatedly (e.g., 50, 75, 100, 125, 150, 175, and 200°C in 20 min intervals) or to ramp the temperature continuously. During standard ambient campaign operation a full cycle time of 160 minutes allows for measurements at 7 different temperature steps from 50 to 200°C plus ambient.

Particle number losses were characterized experimentally at CU as a function of size and TD temperature using NaCl particles, which do not evaporate at TD temperatures. The measured losses were similar to those measured in previous studies and to those expected from theoretical calculations. The total mass loss measured for typical ambient submicron particle size distributions increased linearly from approximately 5% at 54°C to 20% at 230°C. The increased loss with temperature is mostly due to thermophoresis (but also diffusion) acting between the vaporizer and denuder stages. The measured relationship between mass loss and temperature is used to correct ambient data for these losses.

2: Development of TD Method for Measuring Vapor Pressures of OA Components

Laboratory studies were carried out at UCR to develop a method for measuring the vapor pressures and volatility-dependent mass spectra of organic aerosol particles using a TD-AMS. Such data are valuable for modeling organic gas-particle partitioning and for obtaining more detailed composition information from AMS mass spectra. The method for measuring vapor pressures is based on an empirically determined relationship between the TD temperature at which 50% of the organic aerosol mass evaporates (T₅₀) and the organic component vapor pressure at 25°C (P₂₅). The T₅₀ is a convenient and appropriate parameter for characterizing the volatility of OA, since SOA, combustion-derived OA, and ambient OA all have smooth TD profiles with similar shapes that are indicative of the evaporation of mixtures of compounds with a wide range of volatilities. The empirical approach developed here avoids complex modeling of aerosol evaporation, which normally requires information on several particle properties such as composition, heat of vaporization, phase, mixing state, size distribution, and mass loading, most of which are either unknown (and therefore must be estimated or guessed) or time consuming to determine for ambient aerosols. Laboratory measurements of T₅₀ were made on a variety of monodisperse, single-component OA particles with known P₂₅ values and the results were used to create a logP₂₅ vs. T₅₀ calibration curve for use with particles with unknown vapor pressures. A thermal desorption particle beam mass spectrometer (TDPBMS) developed previously in the Ziemann lab was used to analyze particle composition rather than an AMS, but since both instruments employ thermal desorption and electron ionization they give similar mass spectral information. The effects of aerosol mass loading and particle size on the calibration curve were investigated experimentally and theoretically to estimate the contribution of these parameters to measurement uncertainties. Vapor pressure distributions and volatility-dependent mass spectra were then measured for particles composed of simple mixtures of standard organic compounds and for SOA formed from reactions of alkanes, monoterpenes, and aromatics with various oxidants in an environmental chamber, which together are representative of the variety of hydrocarbon-like and oxygenated organic aerosols present in the atmosphere. The evaluation indicates that this empirical method can be used to estimate OA component vapor pressures to within approximately an order of magnitude and that volatility differences among components can be used to improve aerosol mass spectral data analysis. This is the first experimental method that has been developed to estimate vapor pressures of organic components in ambient aerosols.

3: Laboratory Studies of Secondary Organic Aerosol:

The environmental chamber experiments described above provided TD profiles and TDPBMS mass spectra (similar to AMS mass spectra) for SOA formed from reactions of alkanes, monoterpenes, and aromatics with various oxidants. These profiles and also those obtained for many different single-component standards were compared with desorption profiles obtained using temperature-programmed thermal desorption (TPTD). This is a method developed previously by the Ziemann group in which aerosol is sampled into the TDPBMS, collected on a cooled vaporizer, and then desorbed at a ramp rate of 2°C min^-1. Components desorb according to volatility, allowing mass spectra of separated components to be obtained. It was observed here that the temperatures of the maxima in TPTD profiles were consistently ~15°C less than the T₅₀ values in the corresponding TD profiles. Because of the close correspondence of these quantities, it is not necessary to perform chamber experiments to create SOA for all systems of interest. Instead, we are able to simulate TD profiles of SOA by converting TPTD profiles in our large SOA database. These are placed at a newly created web page in the AMS User’s website http://cires.colorado.edu/jimenez-group/TDPBMSsd/ that is maintained by Professor Jimenez at CU, along with real-time TDPBMS mass spectra of SOA, which are similar to those obtained with the AMS since both use thermal desorption and electron ionization. This has been done thus far for a selection of SOA systems. More are being added and this will be a standard procedure for our future SOA studies.

In addition to these studies by the UCR group, SOA was analyzed with the CU TD-AMS equipped with a high-resolution time-of-flight AMS (HR-ToF-AMS) during the SOAR-1 campaign. Five environmental chamber reactions were studied, including α-pinene ozonolysis and the photooxidation of gasoline vapor with NO_x (where SOA precursors are dominated by aromatics). The reactions produce prototypical biogenic and anthropogenic SOA. The SOA from both reactions had T₅₀ values of ≤84°C and SOA mass evaporation rates in the TD of 0.8-0.9% K^-1. These results were very useful for comparing to ambient measurements.

4: Laboratory Studies of Combustion-Derived Organic Aerosol:

The TD-AMS system equipped with a HR-ToF-AMS was used in the FLAME-1 and FLAME-2 (Fire Lab At Missoula Experiment, Phases 1 and 2) studies to investigate the volatility of biomass burning OA (BBOA) at the US Forest Service Fire Sciences Lab in Missoula, MT in June 2006, and May-June 2007. Each of these studies analyzed highly diluted (~x10⁴) smoke from open air burning of ~15-20 different biomass specimens (~200 grams each). The smoke was confined to a large chamber and OA from flaming and smoldering phases of combustion were allowed to mix and stabilize before starting the TD-AMS cycle. The range of T₅₀ values of BBOAs measured in this study was 60-280°C, which also covers the range that has been measured by the TD-AMS for OA under any circumstances. BBOA from lodgepole pine needles and sticks was the most volatile, while that produced by burning sage and rabbitbrush was by far the least volatile OA measured to date with the TD-AMS. Most BBOAs were quite volatile, with more than half having T₅₀ values ≤100°C and all but one being ≤180°C. Qualitatively, it appears that the more volatile BBOA was dominated by smoldering combustion, while the less volatile BBOA was more influenced by flaming combustion. Most BBOA, at least under the conditions of FLAME-1, appears to be of similar or higher volatility than urban OA, although much less volatile BBOA is possible.

5. Ambient Measurements:

The TD-AMS system equipped with the HR-ToF-AMS was used in several field studies of ambient aerosols: ICARTT (International Consortium for Atmospheric Research on Transport and Transformation), SOAR-1 (Study of Organic Aerosols in Riverside), and MILAGRO (Megacity Initiative: Local And Global Research Observations). The ICARTT campaign, which took place in Chebogue Point, Nova Scotia in July-August 2004, essentially provided preliminary volatility data for use in the further development of the TD method. The SOAR-1 campaign was organized by Professors Jimenez and Ziemann to evaluate the performance of the TD-AMS. It was carried out at UCR in July-August 2005. Riverside is located at the eastern end of the Los Angeles basin, which is subject to both local emissions and advected, aged pollution from Los Angeles and thus is heavily impacted by both primary and secondary aerosol. Although it was originally intended that this study would include only the UCR and CU groups, in the end, ~60 scientists from 17 universities and research institutes and companies participated in what is one of the most complete analyses of organic aerosols performed to date. More detailed information on the participants and measurements can be found at http://cires.colorado.edu/jimenez-group/Field/Riverside05/ . The MILAGRO campaign was carried out in March 2006, with the CU group sampling at the T0 Supersite inside Mexico City (north of downtown), which was designed to measure fresh emissions near the city center as well as air aging in the polluted valley.

The volatility of the OA in Mexico City and Riverside was similar. The air exhibited a diurnal profile of the OA mass fraction remaining after passing through the TD. The morning aerosol was dominated by reduced, hydrocarbon-like organic aerosol (HOA) that is due to primary organic aerosol (POA) emissions, while the afternoon aerosol was dominated by oxygenated organic aerosol (OOA) that is overwhelmingly SOA. Although different air masses contained mixtures of different types of OA that combined to produce an average OA volatility, a diurnal cycle with higher volatility in the morning rush hour than in the afternoon was clearly observed, indicating that HOA was more volatile than OOA. The average value of T₅₀ for OA during the MILAGRO campaign was 106°C. Furthermore, the average rate of evaporation of mass of this OA upon heating in the TD near ambient temperature was ~0.7% K^-1. This value provides a constraint on the volatility of OA used in models and is also useful for estimating OA evaporation in aerosol instruments and during aircraft sampling.

Factor analysis of the high-resolution mass spectra from MILAGRO indicates that the OOA in Mexico City can be subdivided into a less oxidized, fresh SOA (OOA-2) that exhibited a broad afternoon peak, and a more aged SOA (OOA-1) that displayed a flatter diurnal profile. BBOA could also be separated; it was more episodic but had a diurnal cycle similar to HOA. Thermograms of total OA analyzed during periods dominated by either BBOA (~60% of the OA mass) or OOA-2 (~65%), or strongly influenced by OOA-1 (~40%) or HOA (~45%) showed that when HOA or BBOA concentrations were higher than OOA concentrations, the average volatility of the total OA was considerably higher than when influenced more heavily by OOA. Thermograms of mass spectral markers of different OA components showed that the overall volatility of the HOA was significantly higher than that of total OA or either OOA-1 or OOA-2, but was less volatile than BBOA. For example, for the major markers of HOA, OOA-1, and OOA-2 the T₅₀ values were 82°C, 141°C, and 94°C, respectively. The T₅₀ value for the major BBOA marker was the same as for HOA, but at higher temperatures the BBOA was more volatile, demonstrating the high volatility of BBOA in Mexico City. Thermograms also indicate that an increase in temperature of 10°C above ambient leads to a net loss of ~9% of HOA mass, but only ~4-7% of OOA mass.

The volatility of OA during periods strongly influenced by urban HOA was similar to that of SOA formed in chamber studies, for which T₅₀ ≤84°C. Conversely, average urban total OA and OA that was heavily influenced by urban OOA was much less volatile, with T₅₀ values of 106°C and 120-125°C. These observations are consistent with results obtained using a new analysis technique developed by the CU group that allows the direct estimate of the oxygen-to-carbon (O/C) ratio of OA using HR-ToF-AMS data. In general, the estimated average O/C ratio for the MILAGRO campaign and for chamber-generated SOA and BBOA increased with increasing TD temperature. This demonstrates that reduced compounds are more volatile and therefore evaporate at lower temperatures, thus enhancing the proportion of more oxygenated compounds in OA at higher temperatures and thereby increasing the O/C ratio.

6. Potential TD-AMS Problems, Solutions, and Improvements:

One common concern when using a TD is the possibility of artifacts due to re-condensation of organic vapor onto particles downstream of the vaporizer, which will lead to an underestimate of particle volatility. This effect is significantly reduced by the use of a charcoal denuder as in the TD developed here. Furthermore, we have shown that re-condensation can be post-diagnosed (and partially corrected for) by analyzing the thermograms vs. aerosol surface area concentration after the TD, since re-condensation will be larger when the aerosol surface area is higher. Plotting the fraction of particle mass remaining after the TD as a function of the total ambient particle mass concentration (used as a surrogate for total particle surface area concentration) yields lines with slopes close to zero in the absence of re-condensation and positive slopes when re-condensation occurs. This analysis was carried out for the relatively high ambient aerosol mass concentrations in Mexico City, and the results indicate that re-condensation was typically not a significant problem. Another potential concern is that evaporation of particle components in the TD will lead to variations in the collection efficiency of the AMS vaporizer, but this has not been observed to play a significant role in ambient measurements to date.

For the future, design improvements may be useful to flatten the longitudinal temperature profile at the entrance and exit of the TD vaporizer and minimize thermophoretic losses, especially at the transition between the vaporizer and denuder. Increasing the number of heaters in the TD vaporizer (for example from the three used here to five, with each being shorter and having higher power capacity) would help achieve this objective and provide a longer residence time in the vaporizer near the temperature set-point. It also might be beneficial to develop ways to change the TD temperature more rapidly, which would provide higher measurement time resolution, but this would probably require a more complex control system and/or active cooling to allow the temperature to stabilize before taking the next data point.

7. Conclusions:

The work described here has successfully met the original goals of the project. The results demonstrate that the TD-AMS is a powerful new instrument for characterizing the chemical composition and volatility of atmospheric organic PM_2.5. It has been shown that the TD can provide a significant degree of separation of components based on volatility, which improves mass spectral identification and yields information on component vapor pressures, making it possible to use TD profiles to estimate the distribution of vapor pressures of organic aerosol components. The information obtained by the AMS has previously come from the dependence of aerosol mass spectra on particle size and time. The addition of desorption temperature as a new correlation parameter for aerosol characterization and as a means for estimating component vapor pressures significantly enhances the power of the AMS method. Through its application the TD-AMS method will provide improved data for understanding organic aerosol formation mechanisms, aerosol properties, and source apportionment. Such data can be used to develop more accurate air pollution models, thereby allowing for more efficient targeting of aerosol sources in pollution control strategies and aiding in the evaluation of the effects of PM_2.5 on human health and the environment.

For example, the development of an automatic, fast temperature-stepping thermodenuder, coupled to a HR-ToF-AMS has enabled the first direct investigations of BBOA volatility and OA volatility in urban air, as well as studies of SOA volatility. A very important result of this work is that urban HOA (a surrogate for POA) and most BBOAs are much more volatile than expected, and are the most volatile organic components in urban particulate matter. They also have similar volatility to the chamber-generated SOA investigated here. Urban OOA (a surrogate for SOA) is much less volatile than urban POA and also the chamber-generated SOA, indicating that some currently unknown factor(s) that are important in atmospheric SOA formation are not reproduced in the laboratory. These results are in direct contradiction to the widespread representations of OA in most atmospheric models, in which POA is treated as non-volatile, with no allowance for evaporation upon heating or dilution, or condensation upon cooling, while SOA is treated as semi-volatile. It is clear from this that the representation of organic aerosol volatility in current atmospheric models is highly inaccurate and needs to be revised.

It is also worth noting an important practical application of this project. One of the original reasons the TD technique was selected as the means for gaining aerosol volatility information with an AMS was that a TD is easily interfaced to an AMS. It was expected that once the power of the TD-AMS technique was demonstrated, the TD would become a desired addition to current and future AMS instruments. Because of the widespread use of the AMS (~65 currently in use), the impact of the TD technique (and this project) will therefore be greatly amplified. The TD technique has already generated considerable interest in the AMS community and it is expected to achieve widespread use. Users of other instruments for aerosol chemical analysis (e.g., ATOFMS, PILS) are starting to use thermodenuders in the way described here, in collaboration with us (e.g., Denkenberger et al., ES&T, 2007) or directly inspired by our work (e.g. Hannigan et al., AAAR Conference, 2007).

URLs/Downloads:

2005 Progress Report

Final Progress Report

2006 Progress Report

URL

2004 Progress Report

Record Details:

Record Type:PROJECT( ABSTRACT )

Start Date:10/01/2003

Completion Date:08/14/2006

Record ID: 59301

Show Additional Record Data

Related Organizations:

Role :OWNER

Organization Name :AERODYNE RESEARCH INC.

Mailing Address :45 Manning Road

Citation :Billerica

State :MA

Zip Code :1821



Role :OWNER

Organization Name :UNIVERSITY OF COLORADO - BOULDER

Mailing Address :Campus Box 17

Citation :Boulder

State :CO

Zip Code :80309



Role :OWNER

Organization Name :UNIVERSITY OF CALIFORNIA - RIVERSIDE

Mailing Address :900 University Ave

Citation :Riverside

State :CA

Zip Code :92521

Project Information:

Approach :

A thermodenuder will be constructed, tested, and coupled to an AMS and then evaluated and optimized using standard particles. The TD-AMS will then be used to develop a database of mass spectra-volatility signatures for interpreting ambient organic aerosol TD-AMS data. TD-AMS measurements will be made on laboratory-generated organic aerosols from the major primary sources (combustion of gasoline, diesel fuel, and wood, and meat cooking) and the major secondary sources (oxidation of aromatics, alkanes and alkenes, and biogenic compounds by OH and NO3 radicals and O3) of fine particle organics. A field study will then be carried out in the Los Angeles Air Basin under conditions when both primary and secondary organic aerosols should be present, and the TD-AMS data analyzed using the new database.

Cost :$409,922.00

Research Component :Particulate Matter

Approach :

A thermodenuder will be constructed, tested, and coupled to an AMS and then evaluated and optimized using standard particles. The TD-AMS will then be used to develop a database of mass spectra-volatility signatures for interpreting ambient organic aerosol TD-AMS data. TD-AMS measurements will be made on laboratory-generated organic aerosols from the major primary sources (combustion of gasoline, diesel fuel, and wood, and meat cooking) and the major secondary sources (oxidation of aromatics, alkanes and alkenes, and biogenic compounds by OH and NO3 radicals and O3) of fine particle organics. A field study will then be carried out in the Los Angeles Air Basin under conditions when both primary and secondary organic aerosols should be present, and the TD-AMS data analyzed using the new database.

Cost :$409,922.00

Research Component :Air Quality and Air Toxics

Project IDs:

ID Code :R831080

Project type :EPA Grant