4586
J. Phys. Chem. A 2010, 114, 4586–4594
Aerosol Volatility and Enthalpy of Sublimation of Carboxylic Acids Kent Salo, Åsa M. Jonsson,† Patrik U. Andersson, and Mattias Hallquist* Department of Chemistry, Atmospheric Science, UniVersity of Gothenburg, SE 412 96 Go¨teborg, Sweden ReceiVed: October 22, 2009; ReVised Manuscript ReceiVed: March 2, 2010
The enthalpy of sublimation has been determined for nine carboxylic acids, two cyclic (pinonic and pinic acid) and seven straight-chain dicarboxylic acids (C4 to C10). The enthalpy of sublimation was determined from volatility measurements of nano aerosol particles using a volatility tandem differential mobility analyzer (VTDMA) set-up. Compared to the previous use of a VTDMA, this novel method gives enthalpy of sublimation determined over an extended temperature range (∆T ∼ 40 K). The determined enthalpy of sublimation for the straight-chain dicarboxylic acids ranged from 96 to 161 kJ mol-1, and the calculated vapor pressures at 298 K are in the range of 10-6-10-3 Pa. These values indicate that dicarboxylic acids can take part in gasto-particle partitioning at ambient conditions and may contribute to atmospheric nucleation, even though homogeneous nucleation is unlikely. To obtain consistent results, some experimental complications in producing nanosized crystalline aerosol particles were addressed. It was demonstrated that pinonic acid “used as received” needed a further purification step before being suspended as a nanoparticle aerosol. Furthermore, it was noted from distinct differences in thermal properties that aerosols generated from pimelic acid solutions gave two types of particles. These two types were attributed to crystalline and amorphous configurations, and based on measured thermal properties, the enthalpy of vaporization was 127 kJ mol-1 and that of sublimation was 161 kJ mol-1. This paper describes a new method that is complementary to other similar methods and provides an extension of existing experimental data on physical properties of atmospherically relevant compounds. Introduction Aerosol particles in the atmosphere have significant environmental effects; for example, influencing the radiation budget and altering precipitation systems.1 Another impact of aerosol particles is adverse health effects, such as affecting the respiratory systems and inducing cardiovascular diseases.2 A significant contribution to atmospheric aerosol particles is the formation of products of low volatility from chemically processed volatile organic compounds (VOCs). VOCs are oxidized and transformed via atmospheric radical and photochemical processes to less volatile compounds that can be subject to gas-to-particle partitioning, contributing to secondary organic aerosol (SOA) formation.3 The actual mechanistic pathways and identities of the low-volatility products originating from oxidation of VOCs are not fully established. However, a typical class of products from atmospheric oxidation processes yielding SOA is the carboxylic acids.3 For example, the important biogenic SOA precursors R-pinene and β-pinene give pinic and pinonic acids.4,5 Carboxylic and dicarboxylic acids are primarily attributed to be a major source of organic aerosol mass by gas-to-particle partitioning or surface reactions.6 Typical concentrations of dicarboxylic acids in ambient air are a few nanograms per cubic meter and thus contribute to a large fraction of the total identifiable resolved organic mass in fine aerosols, and they have been identified and quantified in both rural and urban air.7-10 In addition to being of atmospheric interest, dicarboxylic acids have attracted interest due to physical properties of the bulk phase; for example, the now well-known melting point alternation with the parity of the overall number of carbons.11 * Correspondingauthor.Phone:+46317869019.E-mail:
[email protected]. † IVL Swedish Environmental Research Institute Ltd., Box 5302, SE 400 14 Go¨teborg.
The possibility of analyzing specific organic compounds in aerosol particles has increased during the past decade, with improvements in both techniques and methods and thereby has led to an increased knowledge of the chemical composition of aerosol particles.3,12 The organic fraction in atmospheric aerosols varies with sampling sites but is considerable in both rural and urban air.13 The complex mixture of compounds that form organic aerosols leads to a need to consider a wide range of volatilities.14 This volatility of the organic aerosols is of importance for gas-to-particle partitioning and stems from vapor pressures and the concentration of the individual compounds and partly to the physical state of the condensed phase.15,16 There is a challenge to measure vapor pressures of compounds of low volatility due to the relatively low equilibrium concentration that can be achieved in the gas phase. Especially, this is the case for low-volatility polar compounds that may interact considerably with any surfaces. However, there are a number of different methods to determine vapor pressures and the corresponding enthalpy of vaporization for organic compounds, by monitoring either the condensed phase or the gas phase. For example, some experimental methods are based on effusion of condensed bulk samples using a Knudsen cell, in which the mass loss of the condensed phase is proportional to the vapor pressure and the dimension of an effusion orifice.17,18 Nowadays, the use of modern analytical techniques can provide means to measure the gas phase concentration directly, also for compounds with low saturation vapor pressures (e.g., below 10-4 Pa). This was recently applied in vapor pressure measurements of substituted and unsubstituted carboxylic acids using a thermal desorption particle beam mass spectrometer.19 In their study, the aerosol samples were collected on a cryogenically cooled surface, and the thermally desorbed vapors were analyzed with a quadrupole mass spectrometer. Additionally, there have evolved several empirical computational methods: for example,
10.1021/jp910105h 2010 American Chemical Society Published on Web 03/17/2010
Sublimation of Carboxylic Acids
J. Phys. Chem. A, Vol. 114, No. 13, 2010 4587
TABLE 1: Structures and Physical Properties of the Organic Acids Used in This Work
a Surface free energy. b Interparticle distance where potential is zero. c Depth of the potential energy well. d Bilde et al.26 e Thalladi and Nusse.11 f Measured within this work. g Bilde and Pandis.25 h Assumed to be the same as for suberic acid. i Calculated according to Bird and Stewart.42 j Acros Organics product data sheet.
based on other physical properties20 or related to functional group contributions to the overall vapor pressure for a specific compound (UNIFAC-p° and SIMPOL.1).21,22 In this work, a volatility tandem differential mobility analyzer (VTDMA) system has been used to quantify the volatility of aerosol particles, that is, vapor pressures at selected temperatures. The basis of this technique has been described by Rader and McMurry,23 and the VTDMA system used in this study has been described in Jonsson et al.24 A similar system, operating between 290 and 314 K, has been used previously to study vapor pressures of, for example, dicarboxylic acids.25,26 The VTDMA technique has also been extensively used for field measurements with the aim to characterize and identify different aerosol sources.27-32 Recently, this technique has been applied in laboratory studies with focus on secondary organic aerosol (SOA) formation.24,33-37 It has previously been demonstrated that the designed VTDMA system applied in the present study can serve as a useful tool to follow small changes in thermal
properties of aerosol particles.24 Furthermore, as is stressed in this study, an additional application for the VTDMA system is to obtain enthalpy of evaporation of low-volatility substances. Specifically, the objective of this study was to classify the volatility (i.e., the rate of evaporation) of straight-chain dicarboxylic acids and infer the enthalpies of sublimation and saturation vapor pressures. In addition, experiments were performed with two cyclic compounds: one dicarboxylic acid (pinic acid) and one keto acid (pinonic acid) that are known products, with substantial yields, from ozonolysis of R-pinene. The results are discussed in relation to the volatility determination of complex mixtures of compounds and SOA volatility in general. Experimental Section A VTDMA system was used to determine the volatility properties of organic aerosol particles from nine selected
4588
J. Phys. Chem. A, Vol. 114, No. 13, 2010
Salo et al.
Figure 1. A schematic of the four main parts in the experimental setup. In clockwise order: 1, Aerosol generation; 2, selection of a narrow aerosol size distribution; 3, evaporation of the aerosol at selected temperatures; and 4, determination of the residual particle size distribution.
carboxylic acids shown in Table 1. The nine acids investigated were the C4-C10 dicarboxylic acids and pinonic and pinic acid. A schematic of the setup used for the aerosol generation and volatility measurements is shown in Figure 1. The setup consists of four major parts that represent generation, size selection, conditioning, and size change analysis of the carboxylic acid aerosol particles. The carboxylic acid aerosol particles were generated from ∼0.2 wt % organic acid aqueous solutions (Milli-Q-plus) using a TSI constant output atomizer (TSI 3076) operated in a nonrecirculating mode. The atomizer was fed with the acid solution using a peristaltic pump (Cole-Palmer, MasterFlex 7520) and atomized using filtered compressed purified air (Balston 75-52 FT-IR purge gas generator). The size distribution of the produced aerosol can be modified by changing the concentration of the aqueous solutions. To reach an adequate aerosol size and number concentration of less soluble acids, heated solutions (313-333 K) were used. The compounds used were commercially available products, and in most cases, were used as received. Succinic acid (g99%), glutaric acid (g99%), adipic acid (g99%), and azelaic acid (98%) were from Merck. Suberic acid (98%) and sebacic acid (98%) were from Acros Organics. Pimelic acid (>98%) was from Alfa-Aesar, and pinic acid was from Sigma Aldrich Library of Rare Chemicals (purity not stated). For the pinonic acid (98%) from Sigma-Aldrich used in these experiments, a purification step was tested in which the acid was recrystallized from water.
After nebulizing the acid solution, the aerosol was dried using two silica diffusion driers in-line. This method generates an aerosol with a relative humidity well below 5%. Between the driers, a dilution volume was used to dilute the aerosol flow with dry air when necessary to control the particle concentration. The generated dry aerosol was analyzed using a VTDMA unit, parts 2-4 in Figure 1, that has been described in detail elsewhere and is only briefly presented here.24 A narrow size range of the generated aerosol was selected using a differential mobility analyzer (TSI 3071) operated in a recirculating mode, as is shown in part 2 of Figure 1. Particles, organic vapors, and water were removed in the recirculation of the sheath air. A typical sample/sheath air ratio of 1:20 was used to ensure the narrow size selection. Typically, the initial mean particle diameters selected were between 80 and 110 nm. The mass concentration of the selection was, depending on the compound, in the range of 0.1-1 µg m-3, with a typical number concentration of ∼50-500 cm-3. The size-selected aerosol was subsequently directed under laminar flow conditions through one of the four heated parallel tubes that makes the conditioning oven unit. The heated part of each of the ovens is a 50 cm stainless steel tube mounted in an aluminum block with a heating element. The temperature was controlled and monitored with sensors and temperature controllers (Pt 100, Hanyoung MX4). The temperature of the four tubes can be set independently from 298 to 573 K. At the exit of the
Sublimation of Carboxylic Acids
J. Phys. Chem. A, Vol. 114, No. 13, 2010 4589 temperature, t is the evaporation time, and γi is the surface free energy. f(Kni, R) is a correction term for particle diameters in the transition regime41 given by
1 + Kni
f(Kni, R) )
1 + 0.3773 × Kni + 1.33 × Kni ×
(1 + Kni) R (2)
where R is the accommodation coefficient that was set to unity in this work. Kn is the Knudsen number Figure 2. Particle number size distributions for pinonic acid at 298 (solid line), 343 (dashed line), and 423 K (dotted line).
Kn )
2λi Dp
(3)
where λi is the mean free path of molecule i in air. The diffusivity of species i in air Di,air is given by
( T3
Di,air ) 5.9542 × 1024
Figure 3. Temperature dependence of volume fraction remaining (VFR) for glutaric acid (open circles), succinic acid (gray circles), and ammonium sulfate (black circles).
(
)
dDp 4Di,airMi 0 4γiMi f(Kni, R) )p exp dt FiDpRT i DpFiRT
)
(4)
2 pσi,air Ωi,air
where p is the total pressure. Ωi,air is calculated as described by Bird and Stewart42 using the Lennard-Jones parameters for the molecule i in air; that is, σi,air and εi,air, where σ is the interparticle distance where the potential is zero and ε is the depth of the potential energy well,
σi,i + σair 2
(5)
εi,air ) √εi,i × εair
(6)
σi,air ) heated part, the evaporated gas was adsorbed by activated charcoal diffusion scrubbers to prevent recondensation. The resulting aerosol was classified using an SMPS system (TSI 3096). No further TDMA algorithms were used to fit the achieved particle counts. As an example, Figure 2 displays the change in the number particle size distribution due to increased evaporation of the aerosol particles at higher temperatures. The change in the particle mode diameter was monitored for each temperature setting and normalized to the diameter determined at the reference temperature (298 K), resulting in a normalized mode particle diameter (NMDp) as a function of temperature. From the NMDp, a volume fraction remaining (VFR) can be calculated (VFR ) (NMDp)3), assuming spherical particles. In Figure 3, the temperature dependence of VFR, a thermogram, is exemplified for ammonium sulfate, succinic acid, and glutaric acid aerosol particles. A typical thermogram consists of 10-20 measured VFRs at selected temperatures ranging from 298 to 523 K. For particles in the transition region (between 0.02 and 3 µm at atmospheric pressure), the evaporation of the particles can be used to calculate the saturation vapor pressure at each temperature.23,25,26,38-40 In brief, the evaporation of a single component aerosol particle can be described by
1 1 + Mi Mair
and
where subscript i, i denotes the pure substance and air denotes pure air. εair/kB is 97.0 K, and σair is 3.617 Å.42 By integrating eq 1, one obtains the saturation vapor pressure, p0, as
p0 ) -
FiRT 4Di,air∆tMi
D
(
Dp,f p exp ∫Dp,i f(Kni, R)
)
-4γiMi dDp DpFiRT
(7)
where Dp,i is the initial and Dp,f is the final particle diameter, and ∆t is the residence time in the oven calculated by assuming a plugflow. Furthermore, by applying the Clausius-Clapeyron equation,
d ln p0 ∆H0 )1 R d T
()
(8)
(1)
where Dp is the particle diameter, Di,air is the diffusivity of molecule i in air, Mi is the molar mass, Fi is the density, p0i is the saturation vapor pressure, R is the gas constant, T is the
and assuming that the enthalpy (∆H0) is constant over the studied temperature range, ∆H0 and the saturation vapor pressure 0 ) is obtained. Due to the low humidity applied, at 298 K (p298K the generated aerosols are dry and supposed to be crystalline, and the obtained enthalpy is therefore assumed to represent
4590
J. Phys. Chem. A, Vol. 114, No. 13, 2010
Salo et al.
TABLE 2: Enthalpy of Sublimationa and Calculated Vapor Pressuresb this study name
P0 (298 K) (10-5 Pa)
literature data
∆Hsub (kJ mol-1)
P0 (298 K) (10-5 Pa)
∆Hsub (kJ mol-1) 138 ( 11e 120i 118 ( 3j 123.1l 67 ( 7d 91 ( 7e 102f 132i 154 ( 6e 118f 140g 135 ( 13h 146i 129.3 ( 2.5j 121.9k 147 ( 11e 178g 149 ( 10h 124i 184 ( 12e 148g 148i 143.1 ( 2.8k 153 ( 24e 138i 147i 160.7 ( 2.5j
succinic acid
+2.0 6.4-1.8
112 ( 12 (303-328 K)
4.46 ( 2.2e 13.7i
glutaric acid
+31 85-22
101 ( 20 (298-318 K)
adipic acid
+1.8 5.8-1.4
97 ( 8 (303-333 K)
74 ( 37d 67 ( 34e 102f 40i 1.4 ( 0.7e 1.44f 1.7g 3.4 ( 1.2h 3.02i
pimelic acid
+8.0 17-5.0 +6.0 1.8-1.3
suberic acid
+0.6 1.4-0.4
azelaic acid
+0.8 4.7-0.7
sebacic acid
+0.5 0.9-0.4
pinonic acid
+0.2 0.42-0.1
pinic acid
+1.5 10-1.8
127 ( 20c (301-323 K) 161 ( 50 (310-327 K) 101 ( 10 (303-357 K)
9.7 ( 5e 1.2h 7.2 ( 1.7h 2.47i 0.2 ( 0.1e 0. 34h 0.33i
96 ( 5 (303-343 100 ( 12 (308-363 90 ( 7 (318-358 83 ( 5c (303-326
0.9 ( 0.5e 0.75i 0.14i
K) K) K)
7 ( 2d 4.27 ( 2.1d
Figure 4. Normalized modal particle diameter (NMDp) of suberic acid aerosols for different residence times in the heated part of the oven. Residence times at 298 K calculated by assuming a plug flow, 0.7 s (black circles), 1.1 s (gray circles), 2.3 s (open circles), and 4.5 s (crosses). For the vapor pressure calculations, these residence times were corrected for the actual temperature.
109 ( 21d
K)
a Calculated for the given temperature intervals. b Extrapolated to 298 K. c Enthalpy of vaporization. d Bilde and Pandis.25 e Bilde et al.26 f Tao and McMurry.39 g Chattopadyay et al.44 h Sahle et al.47 i Chattopadyay and Ziemann.19 j Cox and Pilche.45 k Albyn.43 l DeWit et al.46
enthalpy of sublimation; that is, solid-to-gas transformation. This assumption is further discussed in the Results and Discussion section. To apply this method for the VTDMA measurements, some assumptions are needed. First, one assumes that all the particles are spherical with isotropic surface free energy. Second, the vapor pressure from the evaporated species and the latent heat effects are assumed to be negligible. From the experiments in this study, it was found that the partial pressure from the evaporated acids were well below their saturation vapor pressures at the end of the heated part of the oven for all temperatures used. The compound-specific input data needed for the calculations of the vapor pressures are found in Table 1. Results and Discussion Vapor Pressures and Enthalpies of Sublimation. NMDp and VFR were determined as a function of temperature for all nine carboxylic acids, and corresponding vapor pressures were calculated at the respective temperatures. Table 2 gives a summary of enthalpies of sublimation and vapor pressures extrapolated to 298 K for all acids. The enthalpies of sublimation determined in this work ranged from 83 to 161 kJ mol-1, and the calculated vapor pressures at 298 K were in the range of 10-6-10-3 Pa. The obtained values are in the same order as previous studies.19,25,26,38,43-47 It should be noted that our values of the enthalpy of sublimation were determined at temperatures somewhat higher than 298 K but generally over a larger temperature interval (∆T ∼ 40 K). This is due to differences in
Figure 5. Calculated saturation vapor pressures of suberic acid as a function of temperature from data in Figure 4 for the different sets of residence times: 0.7 s (black circles), 1.1 s (gray circles), 2.3 s (open circles) and 4.5 s (crosses) at 298 K.
vapor pressure and enthalpy of sublimation for the compounds, and a specific temperature range had to be applied for each acid; that is, the range where most evaporation occurred for the specific residence time (see Figure 3). In addition, for one of the acids, suberic acid, experiments were performed at four different residence times in the oven. The resulting NMDp’s are shown in Figure 4, and as can be seen, there is a clear dependence of the amount evaporated with residence time. However, as is shown in Figure 5, the calculated vapor pressure did not significantly change with residence time. This is indirectly a confirmation of the assumption that the particles are evaporating during the time in the oven and that no recondensation is occurring. From the calculated vapor pressure at each temperature, the data was evaluated assuming a Clausius-Clapeyron relationship; that is, a linear dependence of log p0 versus 1/T. As an example, Figure 6 shows data for succinic acid and glutaric acid plotted as log p0 versus 1/T. For comparison, reported room temperature vapor pressures for these two acids are also shown, and there is a good agreement with our extrapolated values. In previous publications, thorough uncertainty analyses have been performed when applying the theoretical framework and using the physical constants summarized in Table 1.26,38 These reported uncertainties are assumed to be valid and are also used within this work. In addition, any experimental uncertainties related to flow rates and evaporative temperatures were con-
Sublimation of Carboxylic Acids
Figure 6. The temperature dependence of saturation vapor pressures: succinic acid (circles) and glutaric acid (triangles). This study (filled black symbols), Tao and McMurry39 (gray circle), Bilde and Pandis25 (open circle), Chattopadyay and Ziemann19 (open triangle), and Bilde et al.26 (gray triangle).
Figure 7. Enthalpy of sublimation (∆Hsub) for straight-chain dicarboxylic acids: this work (filled circles), Bilde and Pandis25 (open circles), Bilde et al.26 (crosses), Tao and McMurry39 (triangles), Chattopadyay et al.44 (stars), Chattopadyay and Ziemann19 (plus signs), Sahle et al.47 (squares), and Cox and Pilcher45 (minus signs).
sidered. The sample flow and, thereby, the evaporation time is believed to be controlled within 0.05 LPM. A change within this range will change the evaporation time by 1 s, giving a 20% change in p0. A perturbation of the evaporation temperature with 1 K results in
