Environ. Sci. Technol. 2001, 35, 4495-4501
The Hygroscopic Properties of Dicarboxylic and Multifunctional Acids: Measurements and UNIFAC Predictions CHANGGENG PENG, MAN NIN CHAN, AND CHAK K. CHAN* Department of Chemical Engineering, Hong Kong University of Science and Technology, Clear Water Bay, Kowloon, Hong Kong
The role of water-soluble organic compounds on the hygroscopic properties of atmospheric aerosols has recently been the subject of many studies. In particular, low molecular weight dicarboxylic acids and some multifunctional organic acids have been found or are expected to exist in atmospheric aerosols in urban, semiurban, rural, and remote sites. Unlike for their inorganic counterparts, the hygroscopic properties of organic acids have not been well characterized. In this study, the hygroscopic properties of selected water-soluble dicarboxylic acids (oxalic acid, malonic acid, succinic acid, and glutaric acid) and multifunctional acids (citric acid, DLmalic acid, and L-(+)-tartaric acid) were studied using single droplets levitated in an electrodynamic balance at 25 °C. The water activities of bulk samples of dilute solutions were also measured. Solute evaporation was observed in the dicarboxylic acids but not in the multifunctional acids. Oxalic acid, succinic acid, and glutaric acid droplets crystallize upon evaporation of water, but, except for glutaric acid droplets, do not deliquesce even at 90% relative humidity (RH). Mass transfer limitation of the deliquescence process was observed in glutaric acid. Neither crystallization nor deliquescence was observed in malonic acid, citric acid, DL-malic acid, or L-(+)-tartaric acid. Malonic acid and these three hydroxy-carboxylic acids absorb water even at RH much lower than their respective deliquescence RH. The growth factor (Gf), defined as the ratio of the particle diameter at RH ) 10% to that at RH ) 90%, of oxalic acid and succinic acid was close to unity, indicating no hygroscopicity in this range. The remaining acids (malonic acid, glutaric acid, citric acid, malic acid, and tartaric acid) showed roughly similar hygroscopicity of a Gf of 1.301.53, which is similar to that of “more hygroscopic” aerosols in field measurements reported in the literature. A generalized equation for these four acids, Gf ) (1-aw)-0.163, was developed to represent the hygroscopicity of these acids. Water activity predictions from calculations using the UNIFAC model were found to agree with the measured water activity data to within 40% for most of the acids but the deviations were as large as about 100% for malic acid and tartaric acid. We modified the functional group interaction parameters of the COOH-H2O, OH-H2O, and OHCOOH pairs by fitting the UNIFAC model with the measured * Corresponding author phone: (852)2358-7124; fax: (852)23580054; e-mail:
[email protected]. 10.1021/es0107531 CCC: $20.00 Published on Web 10/18/2001
2001 American Chemical Society
data. The modified UNIFAC model improves the agreement of predictions and measurements to within 38% for all the acids studied.
Introduction Organic compounds constitute a significant portion of atmospheric fine particles (1-3). Atmospheric aerosols change their sizes and chemical compositions by absorbing water, which affects their deposition characteristics, size distributions, radiative properties, and chemical reactivity (4). Recent studies indicate that organic compounds or surfactants may form organic films on atmospheric particles. Such films impede the hygroscopic growth of inorganic compounds such as NaCl (5, 6) and CaCl (7). Other studies have shown that organic compounds both positively and negatively alter the hygroscopic properties of inorganic aerosols (8-11). Organic compounds can be conceptualized as consisting of two fractions in terms of their interactions with water. One is a hydrophobic fraction and the other is a hydrophilic fraction (9). The latter fraction, including low molecular weight carboxylic acids, dicarboxylic acids, alcohols, aldehydes, ketones, nitrates, and multifunctional compounds, is condensable and water-soluble. Water-soluble organic compounds (WSOC) have been identified as important components of the organic fraction in atmospheric aerosols (12, 13). In field measurements of WSOC, most effort has been applied to carboxylic acids such as formic, acetic, succinic, glutaric, oxalic, pyruvic, and malonic acids (12, 1417). By examining the atmospheric abundance, the solubility in water, and the condensability of a list of organic compounds, Saxena and Hildemann (18) proposed that, in addition to carboxylic acids, C2-C7 polyols, amino acids, and other oxygenated multifunctional compounds may also exist as atmospheric WSOC. These WSOC are likely to absorb water and thus add to the amount of water absorbed by inorganic compounds. Recently, Ansari and Pandis (19) investigated water absorption by secondary organic aerosols (SOA) and its effects on inorganic aerosol behavior. They found, on average, that SOA accounts for approximately 7% of total aerosol water, based on the UNIFAC (UNIQUAC Functional Group Activity Coefficients) model predictions. Because there is more water present in the system due to the SOA contribution, they found that 10% more nitrate was transferred from the gas to the aerosol phase compared with when the SOA contribution to total aerosol water was excluded. They pointed out that additional measurements of the water activities of organic solutions are necessary for a full evaluation of the utility of the UNIFAC model in predicting the hygroscopic properties of atmospherically relevant organic compounds. Saxena et al. (9) have also discussed the importance of controlled laboratory studies to understand the hygroscopic properties of organic species in atmospheric aerosols. So far, however, water activity data of WSOC for modeling purposes are scarce (20). Only a few controlled hygroscopic measurements from laboratory studies of WSOC are reported in the literature (21, 22). Seinfeld et al. (23) found that both the amounts of condensed organic mass and water in the SOA phase as the RH increases. The single particle levitation approach has been proven to be a valuable tool for investigation of the hygroscopic properties of aerosols (24-27). Peng and Chan (22) presented the first systematic study of the water cycles of selected watersoluble organic sodium and ammonium salts of atmospheric VOL. 35, NO. 22, 2001 / ENVIRONMENTAL SCIENCE & TECHNOLOGY
9
4495
TABLE 1. Selected Properties of the Water-Soluble Organic Acids Studied formula weight
densitya (g/cm3) (at 25 °C)
succinic acid glutaric acid
HO2CCH2CO2He HO2CCO2H HO2CCO2H‚2H2Oc HO2CCH2CH2CO2H HO2CCH2CH2CH2CO2H
104.06 90.04 126.07 118.09 132.12
1.63 1.9017 1.65319 1.552 1.42920
citric acid DL-malic acid L-(+)-tartaric acid
HO2CCH2C(OH)(CO2H)CH2CO2H HO2CCH(OH)CH2CO2H HO2CCH(OH)CH(OH)CO2H
192.12 134.09 150.09
1.665 1.609 1.7598
species malonic acid oxalic acid
molecular formula
solubilityb (g/100 g H2O at 25 °C) 161 12 8.8 64a 116b 160.8d 136.9d 143.6d
vapor pressure (mmHg at 25 °C)b
manufacturer
purity (%)
Aldrich
99
7.9E-6 7.78E-6
AnalaR AnalaR Riedel-deHaen
99.5 99.5 99
1.7E-11 3.3E-8 2.4E-12
Sigma Aldrich Sigma
99.5 99 99
8.26E-5
a From ref 32. The superscript means the temperature at which the density and solubility are measured. b Reference 18. c The dihydrate from the manufacturer was used in this study and the water content measured by TGA is 28.5%. d Reference 33. e There is no data available for the vapor pressure of malonic acid.
interest using an electrodynamic balance (EDB). They found that most organic salts have hygroscopicities similar to that of typical inorganic atmospheric salts such as (NH4)2SO4 and NaCl. Furthermore, they found that sodium malonate and sodium maleate show “nondeliquescent” properties in the water cycle, although they crystallize in bulk sample studies. More data on the water cycles of atmospheric WSOC are needed to extend currently used aerosol thermodynamic models to predict the hygroscopic growth of atmospheric aerosols containing both inorganic and organic species. In this study, the water cycles of a few low molecular weight dicarboxylic acids (malonic acid, oxalic acid, succinic acid, and glutaric acid) and a few multifunctional acids (citric acid, malic acid, and tartaric acid) of atmospheric interest were studied using an EDB at 25 °C. These low molecular weight dicarboxylic acids are commonly found in field measurements (15-17, 28). Malic acid has been identified in atmospheric aerosols (12, 15). Citric acid and tartaric acid are also expected to exist in the atmosphere (18). Monocarboxylic acids (e.g., formic acid and acetic acid), ketocarboxylic acids (e.g., pyruvic acid and glyoxylic acid), and methanesulfonic acid are too volatile for water cycle measurements using the methods of this study. The performance of the UNIFAC model in prediction of water activities of these aqueous organic solutions will also be evaluated.
Experimental Section A temperature-controlled EDB, as described by Peng and Chan (22), was used in this study. Briefly, a combination of AC and DC fields was used to trap and levitate a charged particle in the EDB. The relative mass of a particle equilibrated at different relative humidities was determined by measuring the balancing DC voltage. When a droplet is equilibrated with its ambient environment, its water activity (aw) is related to the RH by aw ) RH/100, ignoring the Kelvin effect because of the large size of the particles. The size of each particle studied was not measured but was estimated to be about 10-20 microns in diameter from visual inspection using a microscope. In this paper, aw and RH are used interchangeably. The RH of the air fed into the EDB was changed by varying the mixing ratio of a stream of dry air and a saturated stream from a water bubbler. The airflow that controlled the RH was momentarily stopped when the balancing voltage was measured. The RH was determined by a dew-point hygrometer (DewPrime I, EdgeTech) with a precision of 0.2 °C. The ambient temperature was measured with a digital thermocouple with a precision of 0.1 °C. The equilibrium mass fraction of the solute (mfs), defined as the ratio of the dry solute mass to the solution mass, was determined at different RH. In general, each measurement took about 40 min to enable the EDB to attain the preset RH. The experimental details were described by Chan et al. (29) and 4496
9
ENVIRONMENTAL SCIENCE & TECHNOLOGY / VOL. 35, NO. 22, 2001
FIGURE 1. The water cycle of malonic acid. Clegg et al. (30). All measurements were made at 25 °C. The range of RH studied was between 5% and 93%. In the single particle experiments, three to four different particles were studied for each chemical at 25 °C. The crystallization RH (CRH) and deliquescence RH (DRH) were determined in the course of the water cycle, if applicable. Since RH was changed in discrete steps in these experiments, only the range of RH values within which phase transformation occurred were reported. It is necessary to identify a reference state, such as RH ) 85%, to calculate the mfs at other RH in the single particle experiments (25, 27). The bulk data to be correlated with the single particle data are usually available at high RHs. However, for low solubility chemicals such as oxalic acid and succinic acid, whose bulk data do not overlap with the single particle data, an assumed water-free state at RH ) 5% was used as the reference state, as described by Chan et al. (25) and Peng and Chan (22). This assumption was validated. The vapor pressure data (Table 1) shows that the three multifunctional acids are nonvolatile but the four dicarboxylic acids may evaporate in the course of the measurement of the water activities of their droplets (11, 21). Nonreproducible erroneous measurements, sometimes with mfs exceeding unity (for example “particle a” in Figure 1), resulted from solute evaporation. To reduce the evaporation of the solute, measurement at a single RH setting was made between each successive reequilibration of the droplet to the standard state. This procedure was repeated a few times to obtain a set of mfs - aw data. The evaporation of the solute was reduced to 2-3%, which is acceptable in EDB measurements, considering the lack of data and the difficulty of the measurements. This reequilibration method was used for the dicarboxylic acids in the water cycle measurement. The water activities of the bulk solutions were measured using an AquaLab water activity meter (Model 3TE, Decagon
FIGURE 2. The water cycle of oxalic acid.
FIGURE 3. The water cycle of succinic acid.
devices, U.S.A.), which measures the dew point of the vapor phase in equilibrium with a bulk solution of known concentration in a sealed chamber. It measures aw with an accuracy of ( 0.003. Chemicals from manufacturers were used as received without any further purification. Since the acids are hygroscopic, they may absorb water under ambient conditions. The amount of dry solute for making the solutions must be accurately determined for the bulk measurements. The acids were dried at 60 °C in a vacuum oven for 10 h to remove any free water (31). For oxalic acid dihydrate, thermogravimetric analysis (TGA) (DSC 2910, TA Instrument, Inc., U.S.A.) was performed to determine the exact water content including bound and unbound water. All bulk measurements were made at 25 °C. Some properties of these acids relevant to this study are listed in Table 1.
crystallization. This inconsistency between the solid state in bulk studies and single particle studies has often been observed (22, 24) when the water content of the supersaturated droplet at the onset of crystallization is not sufficient to form the stable hydrate (26). Figure 2 shows that the molarwater-to-solute ratio of the oxalic acid particles prior to crystallization was only 1.05, which is not sufficient to support the formation of a dihydrate after crystallization. Deliquescence was not observed even at RH ) 94% because the theoretical DRH is 97.3% based on the bulk measurements. However, slight water absorption was observed in oxalic acid particles at RH larger than 80%. Water absorption prior to the theoretical DRH has been observed in NaCl and has been suggested to be due to capillary effects of small cracks on the solid particle surface (38-41).
Results and Discussion In the following discussions, both bulk data and EDB data will be presented and compared with data from the literature, if available. The water cycle is the compositional change of a solution (e.g., solute-to-solution mass ratio) as a function of aw (or RH) when the ambient RH increases and decreases. Hysteresis between particle growth and evaporation due to the transfer of water between the gas and particle phases is a common characteristic of many deliquescent salts (2). However, as will be shown below, many acids with finite solubility behave like nondeliquescent species and do not crystallize and therefore do not exhibit hysteresis in levitated particle studies. In the associated figures, the aw data measured during particle evaporation and growth are denoted by open and closed symbols, respectively. Dicarboxylic Acids: Malonic Acid, Oxalic Acid, Succinic Acid, and Glutaric Acid. Malonic Acid. As shown in Figure 1, the mfs-aw relationship of malonic acid is depicted by a smooth curve without any sudden jump, which would be an indicative of phase transformation. In our experiments, the malonic acid particles did not crystallize nor deliquesce and therefore did not exhibit hysteresis. Such reversible and continuous water sorption has been observed for some pharmaceutical aerosols (29, 34), some water-soluble organic salts (22), and inorganic salts (24). The mfs reaches unity when RH < 10%, suggesting that the particle is in the anhydrous state. NH4NO3 particles have also been reported to remain in a droplet state at very low RH by a few researchers (35-37). The anhydrous malonic acid particles produced at low RH may be in a supercooled liquid state. This can only be achieved in suspended particles due to the lack of foreign nucleation surfaces. Oxalic Acid. Oxalic acid has a low solubility of only 1.364 mol kg-1 (18). The oxalic acid particles crystallized at RH ) 51.8-56.7% (Figure 2). An anhydrous solid, but not the most stable dihydrate form found in bulk studies, was formed after
Succinic Acid. Succinic acid particles showed a trend similar to that of the bulk data in the water cycle measurement (Figure 3), indicating that the assumption of a water-free particle as a reference state at RH e 5% is valid. Crystallization was observed at RH ) 55-59%, but deliquescence was not observed at RH < 94% since the DRH is 98.8%, as inferred from bulk measurements (also shown in Figure 3). Na et al. (21) also measured the water activities of succinic acid solutions using an evacuated electrodynamic balance system. As shown in Figure 3, their data are not in agreement with our measurements. About 2-3% of the solute mass evaporated with the reequilibration method in our experiments but Na et al. (21) reported 30% evaporation loss in their experiments. The discrepancy between their results and the measurements reported in this paper is attributed to the much larger extent of solute evaporation in their experiments. Glutaric Acid. The solubility of glutaric acid solution of 64 g/100 g water reported by Dean (32) is not consistent with the 116 g/100 g water reported by Saxena and Hildemann (18). Our bulk measurements of the saturated solution (aw ) 0.884) in this study was consistent with the latter. The glutaric acid particles crystallized at RH ) 29-33% (Figure 4a). In deliquescence measurements, we found that a glutaric acid particle (particle 5) started to deliquesce at RH ) 83% and kept absorbing water after several hours and completely deliquesced after 12 h, which was much longer than the 40 min used for equilibration in this study. Similar observation was made in the deliquescence of particle 5 at RH ) 85%. This suggests the presence of a mass transfer limitation not found in the other acids studied, which delayed the complete deliquescence at RH ) 83%. Saxena and Hildemann (20) estimated a DRH for glutaric acid in the range 89-99% RH. Cruz and Pandis (11) reported that glutaric acid deliquesces at RH ) 85 ( 5% using a TDMA system, which is generally consistent with our findings. The phenomenon of the mass transfer limitation in the growth process was also observed in sodium pyruvate by Peng and Chan (22). The potential VOL. 35, NO. 22, 2001 / ENVIRONMENTAL SCIENCE & TECHNOLOGY
9
4497
FIGURE 6. The water cycle of malic acid.
FIGURE 4. (a) The water cycle of glutaric acid and (b) glutaric acid growth factors at 25 °C.
FIGURE 7. The water cycle of tartaric acid. As shown in Figures 5-7, neither crystallization nor deliquescence was observed in citric acid, malic acid, or tartaric acid. Each absorbs and desorbs water continuously and reversibly. Unlike the oxalic acid, succinic acid, and glutaric acid in this study, the citric acid, malic acid, and tartaric acid still contain 5, 5, and 10 wt % of residual water, respectively, after evaporation at RH ) 5%. This may be explained by the strong interactions of the polar functional groups with water molecules in these acids at high supersaturations. The presence of residual water in nondeliquescent droplets at low RH has also been observed in other chemicals such as CaCl2 by Cohen et al. (24), disodium fluorescein (DF) by Chan et al. (29), atropine sulfate by Peng et al. (34), and sodium malonate by Peng and Chan (22).
FIGURE 5. The water cycle of citric acid. implication of this mass transfer limitation will be discussed in the section of growth factors below. Multifunctional Acids: Citric Acid, DL-Malic Acid, and L-(+)-Tartaric Acid. These three chemicals contain both hydroxyl and carboxylic functional groups and are very soluble in water. The evaporation of the solute during the water cycle measurements can be ignored because of their low vapor pressure at room temperature (Table 1). The water cycle of citric acid, which serves as a model additive in pharmaceutical applications, has recently been reported (42). Velezmoro and Meirelles (45) reported the water activities of aqueous malic acid and tartaric acid solutions as a function of molarity (mol/L solution). For comparison with their data, density measurements were made by the mass-volume method of the two chemicals at the ambient temperature of 25 °C. Figures 6 and 7 show that the bulk data reported here are consistent with the data reported by Velezmoro and Meirelles (45). 4498
9
ENVIRONMENTAL SCIENCE & TECHNOLOGY / VOL. 35, NO. 22, 2001
Although the water-soluble organic acids studied here all have finite water solubility in bulk (Table 1), malonic acid, citric acid, malic acid, and tartaric acid can maintain a high supersaturation at very low RH in levitated aqueous particles. A similar phenomenon has been observed in sodium malonate and sodium maleate by Peng and Chan (22), in some drugs by Peng et al. (34), in some pharmaceutical additives (glucose and sorbitol) by Peng et al. (42), and in ammonium nitrate by Lightstone et al. (37). The presence of these acids can be expected to alter the hygroscopic properties of atmospheric aerosols, especially at low RH. In particular, the residual water present at low RH (
