Coupling Filter-Based Thermal Desorption Chemical Ionization Mass Spectrometry with Liquid Chromatography/Electrospray Ionization Mass Spectrometry for Molecular Analysis of Secondary Organic Aerosol
2 Filter-based thermal desorption (F-TD) techniques, such as the filter inlet for gases 3 and aerosols (FIGAERO), are widely employed to investigate the molecular compo-4 sition and physicochemical properties of secondary organic aerosol (SOA). Here, we 5 1 introduce an enhanced capability of F-TD through combination of a customized F-6 TD inlet with chemical ionization mass spectrometry (CIMS) and ultra-performance 7 liquid chromatography/ electrospray ionization mass spectrometry
... ometry (UPLC/ESI-MS). 8 The utility of F-TD/CIMS + UPLC/ESI-MS is demonstrated by application to α-9 pinene ozonolysis SOA, for which increased filter aerosol mass loading is shown to slow 10 the evaporation rates of deposited compounds. Evidence for oligomer decomposition 11 producing multi-mode F-TD/CIMS thermograms is provided by measurement of the 12 mass fraction remaining (MFR) of monomeric and dimeric α-pinene oxidation products 13 on the filter via UPLC/ESI-MS. In-situ evaporation of aerosol particles suggests that 14 α-pinene-derived hydroperoxides are thermally labile; thus, analysis of particle-phase 15 (hydro)peroxides via F-TD may not be appropriate. A synthesized pinene-derived 16 dimer ester (C 20 H 32 O 5 ) is found to be thermally stable up to 200 • C, whereas particle-17 phase dimers (C 19 H 30 O 5 ) are observed to form during F-TD analysis via thermally 18 induced condensation of synthesized pinene-derived alcohols and diacids. The comple-19 mentary F-TD/CIMS + UPLC/ESI-MS method offers previously inaccessible insight 20 into the molecular composition and thermal desorption behavior of SOA that both 21 clarifies and expands on analysis via traditional F-TD techniques. 22 Heat-induced evaporation of particles, as a pre-treatment process, has long been em-32 ployed to investigate aerosol volatility, as well as particle-phase composition 3-5 . Two com-33 mon particle-evaporation techniques include: (1) detection of the vapors coming off the 34 particles (e.g., with a mass spectrometer 6-8 ), and (2) removing the vapors with adsorptive 35 materials (e.g., with a thermodenuder 9,10 ) to determine the mass fraction remaining (MFR) 36 in the particle phase after evaporation. These two approaches can provide complementary 37 understanding of SOA chemical properties if one can simultaneously analyze both the vapors 38 from the particles as well as the mass remaining in the particles. 39 To address the minute amount of single-particle mass, particles are routinely collected 40 on filters or impactors 11,12 and analysis is focused on the concentrated material. A recently 41 developed technique, the filter inlet for gases and aerosols (FIGAERO), 13 is now widely 42 employed to investigate SOA composition. The details of FIGAERO operation, which lever-43 ages filter-based evaporation to detect both gas-and particle-phase components in a semi-44 continuous mode, are described extensively elsewhere. 14-17 45 Application of filter-based evaporation of aerosol-phase compounds upon heating has 46 significantly advanced analysis of SOA composition across a wide range of molecular volatil-47 ities. Nevertheless, some intrinsic properties of this method introduce uncertainties into 48 the data analysis. First, the temperature at which the chemical ionization mass spectrom-49 eter (CIMS) detects the maximum ion signal during thermal desorption (so-called T max ) 50 for a pure compound can differ as a function of the analyte matrix (i.e., pure compound 51 vs. SOA sample). 13 This results because 1. (semi-)solid particle phase states determined 52 by aerosol composition may retard the evaporation, 17 2. increasing aerosol mass loading 53 on the filter 18 may limit inter-layer particle-phase diffusion, and 3. the complexity of SOA 54 may induce stronger non-covalent H-bonding between the compound of interest and other 55 SOA constituents as compared to those of the pure compound. 19 Second, temperature ramp-56 ing to separate particle-phase components based on molecular volatility can lead to thermal 57 decomposition of larger molecules or oligomerization in the condensed phase with poorly con-58 3 strained transformation rates. 20,21 Third, owing to the relatively large surface area of filters in 59 F-TD, evaporated molecules can interact with the filter (e.g., adsorption and desorption) be-60 fore entering the CIMS. Schobesberger et al. 5 developed a comprehensive model framework to 61 explain observed thermograms (CIMS signal vs. temperature) for different compounds that 62 incorporates the effect of the filter itself and is based on temperature-dependent reversible 63 oligomerization between monomers and oligomers and irreversible thermal decomposition. 64 Owing to these issues associated with thermal-induced evaporation of aerosols, interpre-65 tation of data derived from thermal desorption measurements and direct comparison between 66 studies are difficult. Alternatively, another widely used, comparatively mild, method to pre-67 treat aerosol samples is extraction followed by electrospray ionization (ESI), often coupled 68 with liquid chromatography (LC). It is noted, however, that the extraction efficiency of 69 the condensed-phase compound depends on the polarity of the solvent. 22 Furthermore, ESI 70 efficiencies have been shown to depend strongly on molecular structure. 23 71 Here, we introduce and demonstrate a complementary capability of filter-based thermal 72 desorption (F-TD) by combining a customized inlet system, the key feature of FIGAERO, 73 with CIMS and off-line ultra-performance liquid chromatography/electrospray ionization 74 mass spectrometry (UPLC/ESI-MS). The F-TD inlet processes the filter sample with a 75 conventional procedure (ramp to a temperature setpoint and hold), from which one obtains 76 thermograms of evaporating species via CIMS, while UPLC/ESI-MS provides complemen-77 tary information about the components remaining on the filter after evaporation under a 78 range of temperatures. The combination of F-TD/CIMS + UPLC/ESI-MS enables a more 79 comprehensive analysis of aerosol molecular composition and volatility than that afforded by 80 traditional thermal desorption techniques. With F-TD/CIMS + UPLC/ESI-MS, we seek to: 81 (1) provide direct evidence for the role of thermal decomposition in producing multi-mode 82 thermograms; (2) demonstrate that the amount of aerosol mass deposited on the filter affects 83 the performance of thermal desorption in terms of both evaporation fluxes and the MFR in 84 the particle phase, and (3) explore the thermal stability and reactivity of typical aerosol 85 4 Experimental 87 Filter-Based Thermal Desorption System 88 A custom F-TD inlet system was developed for the Caltech CIMS. 24 The basic operating 89 principles of the F-TD inlet follow those of the FIGAERO. 13 The F-TD inlet scheme and a 90 detailed comparison with the FIGAERO can be found in the Supplementary Material (SI.I 91 and Fig. S1a). Evaluation of the performance of the F-TD system is described in SI.II. 92 Briefly, the F-TD inlet comprises three parts: N 2 flow control system, heating system, and 93 filter holder. The main N 2 flow is set at 2 LPM by a mass flow controller (Horiba, SEC-94 4400MC), while a critical orifice extracts 200 ccm N 2 from a bypass flow open to the room air; 95 together these flows maintain the pressure in the ion-molecule reaction (IMR) chamber at 96 26.6 Torr. The heating tube of 8-inch length and 3/8-inch OD stainless steel is surrounded by 97 ultra-high temperature heating tape (Omega, STH051-020). The heating process involves 98 a ramping controller (Watlow, F4SH-CKA0-01RG) and two K-type thermocouple probes 99 (PerfectPrime) that monitor temperature: Probe 1 (TL0260) is positioned above the filter 100 and provides the feedback to the ramping controller, while Probe 2 (TL0700) is twined with 101 the heating tape. The probes are connected to an Arduino board (UNO Rev3), and real-time 102 data are recorded by a customized program in MATLAB (R2010) through a serial cable. The 103 stainless steel filter holder is customized for 25 mm diameter filters with a hole for Probe 104 1. A stainless steel porous disc pre-treated by FluoroPel (Cytonix, 801A) is used to support 105 the filter. The essential difference between the F-TD inlet and that of FIGAERO is the 106 pressure to which the filter is exposed: 26.6 Torr vs. ambient pressure. The pressure of 107 the F-TD, however, is still much higher than the vapor pressure of typical SOA oxidation 108 products, thus having limited impact on their evaporation behavior compared to that in the 109 FIGAERO (see explanation in SI.I). Vapor wall loss is also minimized during transportation 110 5 Page 5 of 38 ACS Paragon Plus Environment Environmental Science & Technology 157 components on the filters. 28,29 Briefly, filter extracts were separated by a Waters ACQUITY 158 UPLC I-Class system equipped with an ACQUITY BEH C 18 column (1.7 µm, 2.1 mm × 50 159 mm) fitted with an ACQUITY BEH C 18 VanGaurd pre-column (1.7 µm, 2.1 mm × 5 mm) 160 at 30 • C. Samples were injected with a 10 µL volume, and the total flow rate through the 161 7 (35) Yasmeen, F.; Vermeylen, R.; Szmigielski, R.; Iinuma, Y.; Böge, O.; Herrmann, H.; 595 Maenhaut, W.; Claeys, M. Terpenylic acid and related compounds: precursors for 596 dimers in secondary organic aerosol from the ozonolysis of α-and β-pinene. Atmos. 597 Chem. Phys. 2010, 10, 9383-9392. 598 (36) Yasmeen, F.; Vermeylen, R.; Maurin, N.; Perraudin, E.; Doussin, J.-F.; Claeys, M.