2011_IJMS_Robinson_MAB_FinalThermal desorption metastable atom bombardment ionization aerosol mass spectrometerIntroductionExperimental methodsResults and discussionMetastable source optimizationStability of the dischargeFragmentation of organic compoundsSensitivity of MAB–AMSFLAME-3 field campaignConclusionsAcknowledgementsSupplementary dataSupplementary data2011_IJMS_Robinson_MAB_SI_FinalInternational Journal of Mass Spectrometry 303 (2011) 164–172Contents lists available at ScienceDirectInternational Journal of Mass Spectrometryjournal homepage: www.elsevier.com/locate/ijmsThermal desorption metastable atom bombardment ionization aerosol massspectrometerCarly B. Robinsona,b, Joel R. Kimmelb,c,d,∗, Donald E. Davida,b, John T. Jaynec, Achim Trimbornc,Douglas R. Worsnopc, Jose L. Jimeneza,baDepartment of Chemistry and Biochemistry, University of Colorado, Boulder, CO, United StatesbCooperative Institute for Research in Environmental Sciences (CIRES), University of Colorado, Boulder, CO, United StatescAerodyne Research Inc., Billerica, MA, United StatesdTofwerk AG, Thun, Switzerlandarticle infoArticle history:Received 13 December 2010Received in revised form 25 January 2011Accepted 29 January 2011Available online 15 February 2011Keywords:Metastable atom beam sourceMetastable atom bombardment (MAB)ionizationAerosol mass spectrometryabstractA metastable atom bombardment (MAB) ionization source has been coupled to an existing thermal des-orption aerosol mass spectrometer. The design allows real-time alternation between MAB and electronionization (EI). A jet of metastable species produced in a DC discharge is directed at the ionization volumeof the mass spectrometer, where Penning ionization is thought to be the dominant mechanism. Perfor-mance is characterized in experiments with oleic acid particles. By changing discharge gases between N2,Kr, and Ar, the excited state energy of the metastable species can be adjusted in the range 8.5–11.7 eV.For vaporization at 180◦C, all gases yield significantly less fragmentation than EI, which could improveresults of factor analysis. Fragmentation increases with vaporization temperature, but generated frag-ments have higher average mass than those produced by EI. Analyte signal levels are 0.1% and 0.006% ofequivalent analysis with EI when using Ar*and Kr*, respectively. These sensitivities are not practical forambient studies, but are sufficient for source measurements, as demonstrated with direct measurementsof biomass burning emissions. The measured Ar*flux of 3.6 × 1013sr−1s−1is ∼30 times lower than thebest literature values for similar metastable beam sources, suggesting that sensitivity can be increasedby source design improvements.© 2011 Elsevier B.V. All rights reserved.1. IntroductionAerosols have important effects on regional and global climate,visibility, and human health. At present aerosols are consideredby the Intergovernmental Panel on Climate Change as the mostuncertain component in the radiative forcing of climate [1]. Atmo-spheric aerosols are mixtures of organic and inorganic matter. Theinorganic fraction is better understood, due to the smaller numberof species, fewer sources, and simpler chemistry. Organic aerosols(OA), on the other hand, are complex mixtures of a wide varietyof species with both natural and anthropogenic sources. Thoroughchemical characterization of OA remains a significant challenge,and the sources and processing of OA are poorly known. This leadsto inaccuracies in predictions of future climate forcing and requiresnew approaches for the analysis of OA [2].A number of techniques can quantify and characterize OA,but inevitably each technique has limitations. Thermal–optical∗Corresponding author at: Aerodyne Research Inc., 45 Manning Road, Billerica,MA 01821-3976, United States.E-mail address: [email protected] (J.R. Kimmel).instruments can quantify total organic carbon [3], while totalwater-soluble organic carbon can be quantified via capture into aliquid and online analysis [4]. However both techniques are limitedby their lack of chemical resolution, as they cannot identify sub-types of OA. Ideally OA could be characterized at a molecular level,but due to the extreme range of physical and chemical propertiesof OA species, only a small fraction of the mass of ambient OA hasbeen compositionally resolved [2,5]. Several approaches attemptto characterize the composition of the total bulk OA, althoughat the expense of molecular information. These include FTIR [6],NMR [7], and online aerosol mass spectrometers, such as the Aero-dyne aerosol mass spectrometer (AMS) used in this work [8]. TheAMS flash vaporizes particles and ionizes the gaseous plume withelectron ionization (EI). Recorded MS signals are quantitativelyapportioned to total OA and non-refractory inorganic species. Fac-tor analysis of AMS data allows the identification of several OAcomponents which provide useful information about OA sourcesand processing [8]. But, the high degree of molecular fragmenta-tion generated by EI limits the information that can be extractedabout molecular composition as well as source identification.Better resolution of different sources of OA and classes ofcompounds comprising OA is highly desirable. Softer ionization1387-3806/$ – see front matter © 2011 Elsevier B.V. All rights reserved.doi:10.1016/j.ijms.2011.01.027C.B. Robinson et al. / International Journal of Mass Spectrometry 303 (2011) 164–172 165techniques reduce or eliminate molecular fragmentation byimparting much less energy than EI during the ionization process.Vacuum ultraviolet (VUV) photoionization [9], chemical ionization[10–13], and low energy electron capture ionization [14] haveall been recently applied to OA analysis in the laboratory. Evenfor a softer ionization source the mass spectra of ambient OA areextremely complex. Thus, for a direct analysis instrument like theAMS the main advantage of the more distinct mass spectra wouldbe identification of additional chemical classes of atmosphericOA via factor analysis methods [15]. The AMS represents a well-developed platform for atmospheric aerosol analysis [8], and thusthe implementation of a soft-ionization source is simplified. Aprevious effort coupled a VUV lamp to the thermal desorption AMSplatform [16]. The VUV–AMS showed much reduced fragmentationcompared to EI, but the sensitivity was 0.02% of EI. Currently asignificantly more intense VUV lamp is not