Environ. Sci. Technol. 2001, 35, 3344-3349
Evaporation Rates and Vapor Pressures of Individual Aerosol Species Formed in the Atmospheric Oxidation of r- and β-Pinene MERETE BILDE† AND S P Y R O S N . P A N D I S * ,‡ Department of Chemistry, University of Copenhagen, DK 2100 Copenhagen, Denmark, and Department of Chemical Engineering, Carnegie Mellon University, Pittsburgh, Pennsylvania 15213
The semivolatile oxidation products (trans-norpinic acid, pinic acid, cis-pinonic acid, etc.) of the biogenic monoterpenes (R-pinene, β-pinene, etc.) contribute to the atmospheric burden of particulate matter. Using the tandem differential mobility analysis (TDMA) technique evaporation rates of glutaric acid, trans-norpinic acid, and pinic acid particles were measured in a laminar flow reactor. The vapor pressure of glutaric acid was found to be log(p0glutaric/Pa) ) - 3510 K/T + 8.647 over the temperature range 290300 K in good agreement with the values previously reported by Tao and McMurry (1989). The measured vapor pressure of trans-norpinic acid over the temperature range 290-312 K is log(p0norpinic/Pa) ) - 2196.9 K/T + 3.522, and the vapor pressure of pinic acid is log(p0pinic/ Pa) ) - 5691.7 K/T + 14.73 over the temperature range 290323 K. The uncertainty on the reported vapor pressures is estimated to be approximately ( 50%. The vapor pressure of cis-pinonic acid is estimated to be of the order of 7 × 10-5 Pa at 296 K.
1. Introduction Starting in the 1960s (1) observations and measurements have demonstrated that biogenic volatile organic compounds (VOCs) contribute to the atmospheric burden of particulate matter. Monoterpenes are an important class of biogenic VOCs (2, 3), and recently organic compounds attributable to monoterpene oxidation have been identified in the particle phase in forest atmospheres (4-7). Aerosols participate in heterogeneous chemical reactions and influence the global radiation budget directly as well as indirectly via their role as cloud condensation nuclei (8). Natural sources of VOCs in the atmosphere are estimated to exceed anthropogenic sources by as much as an order of magnitude (2, 9), and biogenic secondary organic aerosols are likely to influence atmospheric chemistry on both regional and global scales (10). Biogenic organic components for example influence particle hygroscopic properties (11). Even so, the magnitude of the natural contribution to the burden of particulate matter in the atmosphere is not well quantified (12). Upon release to the atmosphere biogenic VOCs such as monoterpenes are oxidized by O3, OH, and NO3. These * Corresponding author phone: (412)268-3531; fax: (412)268-7139; e-mail:
[email protected]. † University of Copenhagen. ‡ Carnegie Mellon University. 3344
9
ENVIRONMENTAL SCIENCE & TECHNOLOGY / VOL. 35, NO. 16, 2001
reactions often lead to semivolatile multifunctional products containing carbonyl, carboxy, and hydroxy groups (13, 14). The products typically have higher molecular weights than their precursors, and for some their vapor pressures are low enough to allow their transfer to the particle phase. Current efforts to model the formation and role of secondary organic aerosols are however limited by lack of information about the partitioning of such semivolatile organic compounds between the gas and particle phase (15, 16). The need for such data is also seen from the work of Pankow (17, 18) who developed the governing equations for the partitioning of semivolatile organic compounds between the gas and the organic aerosol phase and discussed how the vapor pressure of the subcooled liquid can be estimated from the vapor pressure of the solid state. Few experimental values of vapor pressures of secondary organic aerosol components exist. The vapor pressure of five monocarboxylic acids (C14-C18) and two dicarboxylic acids (glutaric and adipic acid) were measured by Tao and McMurry (19) over the temperature range 283-323 K using a tandem differential mobility analyzer (TDMA) system. Vapor pressures of the monocarboxylic acids were in the range 3 × 10-7 to 10-4 Pa at 296 K and vapor pressures of the dicarboxylic acids were 8 × 10-4 and 10-5 Pa for glutaric and adipic acid at 296 K. Heats of formation were in the range 163-191 kJ mol-1 for the monocarboxylic acids and 102 and 117 kJ mol-1, for glutaric and adipic acid, respectively. Hallquist et al. (20), using a Knudsen effusion apparatus, measured the vapor pressures of two oxidation products of R-pinene and ∆3carene, namely pinonaldehyde and 2,2-dimethyl-3-(2-oxypropyl)cyclopropaneacetaldehyde (in the paper referred to as caronaldehyde), respectively. Vapor pressures were measured over the temperature range 255-276 K, and the values at 296 K were 4.2 Pa for pinonaldehyde and 2.3 Pa 2,2dimethyl-3-(2-oxypropyl)cyclopropaneacetaldehyde. The corresponding heats of formation were 75.5 and 77.4 kJ mol-1, respectively. The current study was undertaken to extend the existing database on vapor pressures of biogenic semivolatile organic compounds in the atmosphere. Pinic acid, pinonic acid, and norpinic acid were chosen as representatives for compounds believed to be involved in secondary aerosol formation. The chemical structures of the compounds discussed in this work are shown in Figure 1. Pinic acid was identified in aerosol samples by Christoffersen et al. (21) as a product of the reaction of R-pinene with ozone. It has later been detected in the particle phase in other smog chamber studies of the same reaction (22-24) and tentatively by Jang and Kamens (25). Yu et al. (22) also identified pinic acid from the reaction of R-pinene with ozone, and Glacius et al. (26) have confirmed this finding. Pinonic acid has been reported in the aerosol phase in a number of smog chamber studies of the reaction of R-pinene with ozone (12, 21, 25-28) and recently also in the reaction of R-pinene with ozone (22). Norpinic acid has been identified in the aerosol phase following the reactions of β-pinene with ozone in smog chambers (12, 22, 26) and β-pinene (22, 26). Recently, pinic acid, pinonic acid, and norpinonic acid were identified in the particulate phase in forest atmospheres in Canada, California (4), and Portugal (5, 6).
2. Experimental Section The experimental technique used is the tandem differential mobility analyzer (TDMA) System (29, 30). Experiments were performed using the TDMA and laminar flow reactor system described previously by Dassios and Pandis (31). A schematic 10.1021/es001946b CCC: $20.00
2001 American Chemical Society Published on Web 07/19/2001
FIGURE 1. Chemical structures of the compounds discussed in this work.
FIGURE 2. TDMA experimental system. drawing of the system is shown in Figure 2. The apparatus consists of three parts: monodisperse aerosol generation system, laminar flow reactor, and aerosol classifier. Aerosols were generated by atomizing aqueous solutions of organic compounds in a TSI 3076 constant output atomizer operated in nonrecirculation mode. A constant flow of liquid was provided by a syringe pump (KD Scientific model 200) using 60 mL syringes from Becton Dickinson equipped with an inorganic membrane filter (Anotop 25) from Whatman (>0.2 mm). Measurements performed using syringe flowrates of 0.2 mL/min or less proved problematic, and thus syringe flow rates of 0.4 mL/min or higher were used in all experiments described herein. After passage through two silica gel diffusion dryers (each of length 84 cm, ID: 11 cm) the particles were exposed to a Kr-85 bipolar ion source (TSI, 3077) to ensure a Boltzmann distribution of charges. The particles were introduced into the first differential mobility analyzer (DMA-1) to select an almost monodisperse part of the aerosol distribution. Next, the monodisperse particles entered a 4 m long stainless steel laminar flow reactor (inner diameter 2.206 cm) and were allowed to evaporate. The reactor has four ports along its length for sampling or measurement of temperature and relative humidity. The temperature of the reactor was controlled by flowing water from a circulating bath (Fisher Scientific 9610) through a jacket surrounding the reactor. In the majority of the experiments the aerosol was premixed with preheated particle-free air before entering the reactor. The temperature in the reactor as monitored using a digital thermometer (Fisher Scientific) and Vaisala humidity and temperature transmitters at four points along its length was found to be uniform radially and along the length of the evaporation zone during the time period of an experiment to within (1 K.
Downstream of the reactor the particle stream was introduced into a Scanning Mobility Particle Sizer system (SMPS, TSI Model 3934) consisting of a second DMA (DMA2) and a particle counter (Condensation Nuclei Counter, TSI Model 3010). DMA-2 scans the entire possible voltage range and thus the final size distribution was measured. Experiments were performed for initial particle diameters in the range 150-230 nm and for temperatures in the range 290323 K. The relative humidity was less than 5% throughout the system. To account for evaporation of particles inside the DMAs and connecting tubing and discrepancies between the two DMAs the change in particle diameters reported in the following was obtained by measuring the diameter using DMA-2 before and after evaporation in the laminar flow reactor. In this way the reported change in diameter corresponds to evaporation inside the laminar flow reactor only. All particles passing through the reactor were sampled, and initial and final diameters were obtained as the maximum of the parabola that best fit the top data point and its nearest neighbors. Laminar flow conditions were maintained through the reactor, and volumetric flow rates were measured using the build in TSI flow meters and an external Gilibrator flow cell from Gilian Instrument Corp. Using volumetric flow rates between 1.8 and 8.9 L/min and letting the particles evaporate in the reactor over a distance of typically 4 m (in some cases shorter, for example 3 m) gave evaporation times in the range 4-26 s. Experiments were performed in a carrier gas of air purified using five inline filters: silica gel desiccant from J. T. Baker, packed beds of citric acid and NaHCO3, a particulate filter capsule (HEPA, 99.97% retention of 0.3 µm DOP aerosol) from Pall Corporation, and a Carbon filter (Carbon Cap 150) supplied by Whatman. Aqueous solutions used were as follows: glutaric acid, 0.3-45 mM; pinic acid, 0.2-5 mM; cis-pinonic acid, 3-11 mM; and trans-norpinic acid, 0.2-2 mM. Purified water was obtained by distillation of deionized water or by filtration of deionized water using a Millipore system (Millipore Co.). Chemicals were obtained at the highest purity available from commercial sources and were used as received: glutaric acid was supplied by Aldrich Chemical Co. (>99%) and by Sigma Chemical Co. (>98%). Pinic acid and trans-norpinic acid were obtained from SigmaAldrich Rare Chemical Library, and cis-pinonic (98%) acid was supplied by Aldrich Chemical Co. All chemicals were solids at room temperature. To avoid chemical changes of the organic acids with time all chemicals were stored in a refrigerator and new samples were ordered regularly. Experiments indicated that impurities might be a problem for initial particle diameters less that 150 nm, and to minimize the potential importance of impurities we therefore restricted the conditions to initial particle diameters larger than 150 nm.
3. Theory The reduction in diameter with time of an evaporating particle of compound i in the transition regime in an environment of zero gas-phase concentration can be described by eq 1 (32, 33)
(
)
dDp 4σiMi 4Di,airMi 0 )p exp F(Kni,Ri) dt FiDpRT i DpFiRT
(1)
where Mi is the molar mass of species i, Fi is the density, R is the gas constant, T is the temperature, pi0 is the vapor pressure of species i over a flat surface, σi is the surface free energy, and Dp is the particle diameter. Table 1 gives the VOL. 35, NO. 16, 2001 / ENVIRONMENTAL SCIENCE & TECHNOLOGY
9
3345
TABLE 1: Properties of the Investigated Carboxylic Acids Mi
glutaric acid cis-pinonic acid pinic acid trans-norpinic acid
Vc
Tm a
10-3 kg mol-1 132.12 184.24 186.21 172.18
cm3 mol-1
K 368 375 343 410
365 567.5 532.5 477.5
b
Fi 103 kg m-3 1.424 1 1 1
c
σii d Å 6.01 6.96 6.82 6.57
ii /kB e
K 706.85 720.29 658.85 787.49
a Melting points measured as part of this work. b Critical volumes V are estimated using the group contribution method by Lyderson (41). c Density of glutaric acid from Sigma Aldrich. Densities of pinonic, pinic, and norpinic acid are estimated. d σii is calculated as 0.841 Vc1/3 (34, p e 22). ii/kB is calculated as 1.92 Tm (34, p 22). c
necessary physical data and Lennard Jones parameters to calculate the diffusivity Di,air of species i in air, as described in ref 34. Equation 1 relies on four assumptions: (1) the particle is spherical; (2) surface free energy is isotropic; (3) latent heat effects can be neglected; and (4) the partial pressure of species i far from the particle surface is negligible. The validity of assumption (4) will be discussed in section 4.2. For any given values of particle diameters before (Di) and after (Df) evaporation, evaporation time (∆t), and surface free energy it is possible to calculate the vapor pressure from eq 1. Several expressions have been suggested for the correction term F(Kni,ai). We will apply two of them. First, the expression suggested by Fuchs and Sutugin (35)
F(Kni,ai) )
1 + Kni 1 + 0.773Kni + 1.33Kni(1 + Kni)/R
(2)
where Kni ) 2λi/Dp is the Knudsen number, and the mean free path of a gas molecule of species i in air given as λi ) 3Di,air/ci and ci ) x8RT/πMi. For the purpose of this study we assume that the accommodation coefficient Ri ) 1 for all i. Second, the expression is suggested by Bademosi and Liu (36, 37)
[
F(Kni,ai) ) 1 +
Jc Jk
1-
( ) 0.13Jc βKniJk
]
-1
1 -1
+
6.08βJk Jc
(3)
where Jc/Jk ) 8Dij/EDpci ) 4βKni/E. We assume that the evaporation constant E is one. The definitions of λ and β implied in eq 3 are λ ) Dij/ciβ and β ) 1.132 π/8x8[Mi + Mair/Mair]0.5.
4. Results and Discussion The vapor pressure and surface free energy of glutaric acid have been reported previously (19), and as a test of our system we first measured the vapor pressure of glutaric acid before proceeding to the pinene oxidation products. 4.1. Glutaric Acid. Glutaric acid experiments were performed at 290, 296, and 300 K. The data analysis was performed as described in detail for trans-norpinic acid, section 4.2. Figure 3 shows the vapor pressures obtained using an estimated surface free energy of 0.2 J m-2 at all temperatures versus the inverse temperature. We estimate that the relative uncertainty on the quoted vapor pressures is (50%. Linear least-squares analysis gives
log(pglutaric0/Pa) ) -(3510 ( 368.8)K/T + (8.647 ( 1.248) Uncertainties are two standard errors. Assuming a ClausiusClapeyron relationship between vapor pressure and temperature gives an enthalpy of vaporization of (67(7) kJ mol-1. Also shown in Figure 3 are the results reported by Tao and 3346
9
ENVIRONMENTAL SCIENCE & TECHNOLOGY / VOL. 35, NO. 16, 2001
FIGURE 3. Vapor pressures/approximate mixing ratios (296 K, 1 atm) of glutaric acid versus inverse temperature/temperature: (b) this work and (4) Tao and McMurry. McMurry (triangles), log(p0glutaric/Pa) ) - 5347K/T + 14.95. Given the large uncertainty in the determination of the surface free energy in this type of experiments (as discussed in the next section) the agreement between the two sets of data is satisfying. Even if the experimental techniques used in these two studies were similar, the approaches used for the quantification of the experimental uncertainty were different, resulting in different uncertainty estimates. 4.2. trans-Norpinic Acid. Evaporation rates of transnorpinic aerosols were measured over the temperature range 290-312 K. Evaporation times were in the range 7-22 s. In principle it is possible to determine surface free energy and vapor pressure simultaneously by measuring at least two sets of initial and final diameters and solve the resulting differential equations (1) for σi and pi0 as done by Tao and McMurry (19). We find however, that there is a large uncertainty associated with the experimental determination of the surface free energy. For our experimental data the simultaneous determination of surface free energy and vapor pressure was in many cases inconclusive with respect to surface free energy, and we choose to find the range of vapor pressures that best fits the experimental data for estimated values of surface free energy. Unfortunately, there is no good theoretical method to estimate surface free energies of solids (38, 39), but we can obtain a range of reasonable values by looking at available experimental literature data. Surface tensions for pure saturated organic liquids are typically in the range 0.02-0.04 J m-2 over the temperature range 280320 K (40). Surface free energies of organic solids are expected to be higher. Tao and McMurry (19) report values in the range 0.1-0.45 J m-2 for glutaric and adipic acid over the temperature range 285-313 K. To the best of our knowledge
FIGURE 4. Calculated versus observed final diameters for transnorpinic acid at 290 K for an estimated surface free energy of 0.2 J m-2: (b) p0 ) 6 × 10-5 Pa, (4) p0 ) 3 × 10-5 Pa, ()) p0 ) 9 × 10-5 Pa. The dotted lines are linear least-squares fit to the low- and high-pressure data and the solid line is the one-to-one correspondence line.
TABLE 2: Uncertainty Analysis for an Evaporating trans-Norpinic Aerosol at 290 K with Initial Diameter 163.6 nm and Final Diameter 155.7 nma base value
perturbed by
p after perturbation
relative change in p
σ
0.2 J m-2
T
290 K
∆t
22 s
R F
1 1 g cm-3
+0.3 Jm-2 -0.2 Jm-2 +7 K -7 K +11 s -8 s -0.5 +0.4 g cm-3 -0.4 g cm-3 +0.2 +0.4 +3 nm -3 nm +3 nm -3 nm
3.2 × 10-5 Pa 7.7 × 10-5 Pa 5.4 × 10-5 Pa 5.4 × 10-5 Pa 3.6 × 10-5 Pa 8.5 × 10-5 Pa 5.4 × 10-5 Pa 8.4 × 10-5 Pa 2.6 × 10-5 Pa 6.3 × 10-5 Pa 7.5 × 10-5 Pa 7.9 × 10-5 Pa 3.0 × 10-5 Pa 3.0 × 10-5 Pa 7.7 × 10-5 Pa
-41% 43%
