Deliquescence and Efflorescence of Potassium Salts Relevant to Biomass-Burning Aerosol Particles

Abstract

Particles containing potassium salts are common in biomass burning plumes and include compounds such as KCl, KNO ₃ , and K ₂ SO ₄ . We determined the deliquescence and efflorescence relative humidities of single- and multi-component potassium salts using an environmental transmission electron microscope and an environmental scanning electron microscope. KCl particles deliquesced at a relative humidity (RH) of 85% and effloresced at 56%. K ₂ SO ₄ particles deliquesced at 96% and effloresced at 60%. KNO ₃ particles grew continuously with increasing humidity. If KNO ₃ particles were heat-treated at 104°C overnight, they deliquesced at an RH above 90%. Atomized solutions of multi-component potassium compounds (KCl & KNO ₃ and KCl & K ₂ SO ₄ ) formed particles that contained distinct grains of each compound upon drying. Particles formed from the KCl & KNO ₃ mixtures grew continuously with increasing humidity until deliquescence occurred at an RH of 82% and efflorescence occurred at 58%. Particles from the KCl & K ₂ SO ₄ mixture began to take up water at an RH of 84%. They deliquesced at >90% and effloresced at 52%. This work provides new data for mixed aqueous electrolytes that are important for atmospheric science. Efflorescence values for K ₂ SO ₄ , KNO ₃ , and internally mixed particles of K ₂ SO ₄ and KCl are reported here for the first time.

INTRODUCTION

Biomass burning occurs on a global scale and results in the emission of large amounts of soot, organic, and inorganic material into the atmosphere. Such aerosol particles affect the climate in part because of their capacity to absorb or scatter solar radiation and to act as cloud-condensation nuclei (Hobbs et al. 1997; Andreae et al. 2004; IPCC 2007).

Large numbers of organic and inorganic particles occur in biomass plumes. The inorganic particles include compounds such as KCl, KNO₃, and K₂SO₄ (Posfai et al. 2003; Rissler et al. 2006). These potassium compounds are commonly used as markers for biomass burning (Cachier et al. 1995; Rissler et al. 2005). They have also been detected in dust samples, and their importance has been demonstrated in aerosol models (Kim and Seinfeld 1995; Fountoukis and Nenes 2007).

Several studies examined smoke particles from different biomass fires (Cachier et al. 1995; Gaudichet et al. 1995; Li et al. 2003; Posfai et al. 2003). Flaming and young smoke includes large numbers of particles containing crystals of KCl. Aged and reacted biomass smoke contains greater amounts of K₂SO₄ and KNO₃ than KCl as well as high concentrations of particles containing C, N, and S (Li et al. 2003). This change with age suggests that KCl particles are converted to K₂SO₄ and KNO₃ through gas-particle reactions with S- and N-bearing species, presumably from the biomass plume. These potassium compounds are emitted as KCl and then partially or fully react with HNO₃ and H₂SO₄ during transport in the atmosphere. Partially reacted particles become internal mixtures of either KCl & KNO₃ or KCl & K₂SO₄, whereas fully reacted particles consist of only KNO₃ or K₂SO₄, with HCl gas presumably liberated. Similar aging of potassium salts was reported for dust samples (Ro et al. 2005), where K₂CO₃ particles became internally mixed with sulfates and, less commonly, nitrates.

The transformation of KCl to K₂SO₄ and KNO₃ with aging leads to concomitant changes in the hygroscopic properties of smoke particles. Whether particles are solid or aqueous has important implications for their environmental effects. Aqueous particles are larger and scatter solar radiation back to space more efficiently than solid particles. They can also modify atmospheric chemistry through reactions such as N₂O₅ hydrolysis (Jacob 2000; Kane et al. 2001) and secondary organic aerosol production (Lim et al. 2005).

Limited work has been done on the compositions of mixed aqueous electrolytes even though many hygroscopic atmospheric particles consist of several salts (Martin 2000). The goals of this study were to determine the deliquescence and efflorescence relative humidity of single- and multi-component potassium salts. We used scanning and transmission electron microscopes (SEM and TEM, respectively), both fitted with environmental cells. With these instruments we have the ability to detect small changes on the surfaces of individual particles as they are exposed to RH conditions that they encounter in the atmosphere.

EXPERIMENTAL

Aerosols containing particles of KCl, KNO₃, K₂SO₄, and their mixtures were generated from 1 M solutions prepared from analytical reagent-grade chemicals (97% purity, Aldrich). Solutions of K₂SO₄ were saturated at this concentration. The aerosols were dried using a silicon diffusion drier (TSI Model 3062) to an RH of ∼20% (TSI, 2003). Single particles were deposited by diffusion, in air, onto TEM grids (Ted Pella inc #1820; lacey carbon Type-A, 300 mesh copper). This was achieved by holding a TEM grid directly in front of the output from the silica gel dryer for approximately 2 min.

Water-uptake experiments were carried out using a 200-kV FEI Tecnai F20 TEM fitted with a differentially pumped environmental cell (ETEM) (Wise et al. 2005; Semeniuk et al. 2007; Wise et al. 2008). The cell was kept at constant temperature through a balance of liquid nitrogen cooling and resistive heating. Liquid nitrogen was contained in a dewar outside the microscope column. The dewar was connected via a conducting rod to the specimen-holder tip. The rod, which was insulated with a vacuum jacket from the environment inside the microscope, contained an electrical heater to control and a silicon diode to measure the sample temperature. The holder was designed so that its tip was kept at a constant temperature. The sample temperature was maintained at 15 ± 0.2°C so that an RH of 90 ± 1% was attained if 15 ± 0.05 torr of water vapor was allowed into the environmental cell.

Water vapor was introduced into the environmental cell of the TEM from a glass bulb containing room-temperature distilled water. No carrier gas was used. The vapor pressure in the environmental cell was regulated using a mechanical leak valve and measured using a capacitance manometer located outside the ETEM. More detailed information on the operation of the ETEM and the calculation of the RH is discussed by Wise et al. (2005).

Prior to each water-uptake experiment; the accuracy of the RH measurements was verified by measuring the deliquescence relative humidity (DRH) of laboratory-generated NaCl particles. Their DRH was consistently 75 ± 1%, in agreement with the known DRH of NaCl (DRH = 75.3%) (Wise et al. 2005; Biskos et al. 2006). The error was calculated from standard deviations of at least ten measurements. There was no obvious difference in the DRH or ERH values if the substrate was made from silicon, a discontinuous carbon layer, or a Formar®/carbon backing (Wise et al. 2005). During these experiments, sequences of images at various increasing and decreasing RH values were recorded for each particle. At an RH between 90% and 95%, water condenses on the TEM stage, making water-vapor pressure difficult to read and control. Therefore, RH readings in the ETEM above 90% no longer agreed with the literature DRH for the potassium compounds. Any DRH or phase changes that occurred at an RH above 90% are labelled “>90%.”

An XL30 scanning electron microscope (SEM FEI, Inc.) fitted with a field-emission gun and an environmental chamber (ESEM) to introduce water vapor (Hoffman et al. 2004) was also used for comparison to results from the ETEM. ESEM experiments were performed at a temperature of 5°C. An accelerating voltage of 20 kV, beam current of 148 pA, and slow scanning (line time 16.7 ms and 484 lines per frame) were used.

Detailed chemical and morphological characterization of all particles was performed in vacuum using a Philips CM200 TEM. Semi-quantitative energy-dispersive X-ray spectrometry (EDS) was used for chemical characterization.

RESULTS AND DISCUSSION

Deliquescence and Efflorescence of Particles Atomized from Simple Solutions

A series of ETEM measurements in which the RH was increased from 0% to >90% were carried out for KCl, K₂SO₄, and KNO₃ aerosol particles (Figures 1, 2, and 3 respectively). The measurements are summarized in Table 1.

FIG. 1 KCl particles deposited onto a lacey-carbon TEM grid. All images are at the same magnification. The RH was increased from (a) 0% to (d) 85% and then decreased to 56%. Closed black arrows indicate whether the image was obtained as the RH was being increased (↑) or decreased (↓). The open arrows in (c) indicate water uptake prior to deliquescence. Deliquescence occurred at an RH of (e) 85% and efflorescence occurred at an RH of (f) 56%.

FIG. 2 K₂SO₄ particles deposited onto a lacey-carbon TEM grid, (a), (b), and (c) are of a different field of view than (d), (e), and (f). Relative humidity was increased from (a) 0% to (d) >90% and then decreased back to 60%. The open black arrow in (c) indicates water uptake prior to deliquescence. Deliquescence occurred at an RH (d) >90% and efflorescence occurred at an RH of (f) 60%. ∗Measurements above 90% are not accurate.

FIG. 3 KNO₃ particles deposited onto a lacey-carbon TEM grid. All images are at the same magnification. No abrupt deliquescent phase transition was recorded as the RH increased from (a) 0% to (f) >90%. Each image is of a different field of view.

TABLE 1 Summary of deliquescence and efflorescence measurements made using both the ETEM and ESEM together with corresponding literature values

	DRH		ERH	Lit. values
	ETEM	ESEM	ETEM	DRH	ERH
Single salts
KCl	85 ± 1%		56 ± 1%	84% (Cohen et al 1986)	56%
K2SO4	>90%	96 ± 1%	60 ± 2%	97% (Goldberg 1981)	N/A
KNO3	Cont. growth ^1 & >90% ^2	93%	—	92.3% (Adams and Merz 1929)	N/A
Multi-component salts∗	DRH		ERH	Lit. values
	Initial ^3	Final ^4		DRH	ERH
KCl & KNO3	Cont. growth ^2	82%	58%	78.6% ^4 (Adams and Merz 1929) @ 30°C	N/A
KCl & HNO3	Cont. growth ^2		—	N/A	N/A
KCl & K2SO4	84%	>90%	52 ± 2%	83.2% (Adams and Merz 1929) @ 30°C	N/A
KCl & H2SO4	86%	>90%	60 ± 2%	86% -DRH of KHSO4 (Pilinis et al., 1989) ^5	N/A
N/A—not available.

Atomized KCl particles were deposited onto a TEM grid and exposed to an increasing RH in the ETEM (Figure 1). KCl particles deliquesced at an RH of 85 ± 1%, consistent with the literature value of 84% (Cohen et al. 1987a; Hamza et al. 2004; Semeniuk et al. 2007). Figure 1 shows no visible morphological changes until the RH was increased to 83%, at which point each particle took up an observable quantity of water before deliquescing at 85%. This water uptake prior to deliquescence was reversible, meaning that when the RH was dropped by 1 or 2% the water disappeared. A similar effect was reported and discussed for NaCl particles by Wise et al. (2008). When the RH was decreased to 56 ± 1%, the KCl particles recrystallized (Figure 1f), consistent with efflorescence RH (ERH) measurements of Cohen et al. (1987a) and Hamza et al. (2004).

For K₂SO₄ particles, deliquescence did not occur in the ETEM until the RH was above 90% (Table 1). Similar to the other salts studied, reversible water uptake prior to deliquescence occurred (Figure 2c). When the RH was decreased, K₂SO₄ particles effloresced at an RH of 60 ± 2% (Figure 2f). To our knowledge, this is the first report of the ERH of K₂SO₄.

The DRH of K₂SO₄ was verified using an ESEM, where K₂SO₄ particles deliquesced at an RH of 96%, consistent with measurements of Adams and Merz (1929) and Goldberg (1981). Repeated experiments on KCl and K₂SO₄ showed that the electron beam of the TEM did not affect their DRH or ERH values.

The hygroscopic behavior of KNO₃ is more complex than the compounds discussed above. It is more sensitive to electron radiation than the other alkali salts studied, and therefore consecutive exposures of the same area of the TEM grid caused damage to the particles, making them difficult to study using electron microscopy. The sensitivity of KNO₃ and other alkali metal nitrates to radiation damage is discussed by Kulkarni and Garg (1991). In order to examine these particles with minimal damage, each area was exposed to the electron beam for approximately 5 s. The electron beam was then blanked, and the grid position was changed so that each photograph was from a new area of the grid.

Atomized KNO₃ particles were exposed to an RH that was increased from 0% to above 90% in the ETEM (Figure 3). Since the particles lacked well-developed crystal faces, it was difficult to detect morphological changes as the humidity increased. However, the particles spread out on the TEM grid at higher RH values, suggesting that they had absorbed water (Figure 3f). In the ESEM, gradual growth was observed as the humidity was increased. When the RH was decreased to 0%, these particles gradually decreased in size but did not readily crystallize. A range of studies using other approaches suggests that particulate nitrates such as NaNO₃, NH₄NO₃, and Ca(NO₃)₂ occurred as amorphous solids when the RH was decreased to below 10% after having been above the DRH (Tang and Munkelwitz 1994; McInnes et al. 1996; Lee and Hsu 2000; Gysel et al. 2002; Hoffman et al. 2004; Laskin et al. 2005; Schlenker and Martin 2005; Gibson et al. 2006; Liu et al. 2008).

An optical microscope with polarized light illumination was used to view TEM grids containing atomized KNO₃ particles. Interference colors (Figure 4b) indicate that most such atomized KNO₃ particles are crystalline. This result differs from IR and X-ray measurements that indicated that nitrate particles deposited onto a surface occur as amorphous solids (Hoffman et al. 2004; Liu et al. 2008).

FIG. 4 Polarized-light optical microscopy images of KNO₃ particles deposited onto a lacey-carbon TEM grid. Image (a) is viewed in plane-polarized transmitted-light. Image (b) shows the same area viewed in cross-polarized transmitted light. Only crystalline particles give interference patterns when viewed with cross-polarized light.

In a second set of experiments, the atomized KNO₃ particles were heated overnight at 104°C to remove surface and structurally absorbed water prior to being exposed to an RH increasing from 0% to above 90% (Figure 5). This temperature was not sufficient for the thermal decomposition of KNO₃, which occurs at 400°C. Morphological changes occurred only above 90%, when particles became rounded, suggesting that they deliquesced (Figure 5f). In the ESEM, a similar result (DRH = 93%) was obtained when larger, mechanically formed KNO₃ particles were studied (Figure 6). These measurements are compatible with literature DRH values of 92.3% (Adams and Merz 1929). The difference in hygroscopic properties between atomized and oven-treated KNO₃ (and mechanically produced KNO₃) possibly arises from residual water that remains in atomized particles but was removed upon heating.

FIG. 5 KNO₃ particles deposited onto a lacey-carbon TEM grid and heated overnight at 104°C. Images (a) through (e) are at the same magnification. Each image shows a different field of view. Relative humidity was increased from (a) 0% to (f) >90%. Deliquescence occurred at (f) >90%.

FIG. 6 ESEM images of KNO₃ particles prepared from mechanically grinding the reagent powder. Relative humidity was increased from (a) 0% to (d) 93%. Deliquescence occurred at an RH of (d) 93%.

Deliquescence and Efflorescence of Particles Atomized from Complex Solutions

As a model for atmospherically processed KCl particles, we studied the hydration behavior of particles atomized from 1 M solutions of KCl & KNO₃ (1:1), KCl & K₂SO₄(1:1), KCl & HNO₃(1:1), and KCl & H₂SO₄(1:1). Particles atomized from 1M solutions of KCl & KNO₃ or KCl & K₂SO₄ contained distinct grains of both compounds (Figures 7a and 8a). Using the Philips CMZOD TEM for imaging and EDS, we identified the shapes and compositions of single grains deposited onto the TEM grids before transferring the grids to the ETEM. As KCl grains are generally more euhedral than KNO₃ grains, it was possible to visually distinguish between them in mixed particles. For particles containing both KCl and K₂SO₄, KCl particles consisted of single grains, whereas the K₂SO₄ particles consisted of multiple grains of smaller elongated particles. The coexistence of these compounds as separate phases has been reported for ambient particles (Posfai et al. 1994, 1995), but differs from reports by Ge et al. (1996) and Hoffman et al. (2004).

FIG. 7 Particles from an equimolar solution of KCl & KNO₃ deposited onto a lacey-carbon TEM grid. Images (b) through (f) are at the same magnification. Particle identities are indicated with the black arrows in (a). Relative humidity was increased from (a) 0% to (d) 82% and then decreased to (f) 58%. Images (a), (d), (e), and (f) are of a different field of view. Solid phases, indicated by the open black arrows, remain visible at an RH of (c) 78%.

FIG. 8 Particles from an equimolar solution of KCl & K₂SO₄ deposited onto a lacey-carbon TEM grid. Images (c) through (f) are at the same magnification. Particle compositions are indicated with the black arrows in (a). Relative humidity was increased from (a) 0% to (d) >90% and then decreased to (f) 52%. Solid phases, indicated by the open black arrows, remain visible from an RH of (c) 84% to (d) >90% and are surrounded by a layer of water.

Hygroscopic measurements of particles atomized from the KCl & KNO₃ solution showed that the presumed KNO₃ fraction of these particles gradually took up water without a distinct DRH. It is difficult to assess the eutonic effect for this mixture since KNO₃ does not exhibit a distinct DRH. Solid particles, presumably KCl, were visible inside droplets up to an RH of 78% but were no longer present at 82% (Figure 7c and d). Tilting of the sample within the ETEM allowed us to see whether solid grains were on the interior or exterior of the droplets. Solid grains inside droplets were viewed by changing the height of the focal plane. The compositions of those solids were assumed to be those of the least soluble species of the internally mixed particle. Adams and Merz (1929) measured the eutonic RH of a solution containing KCl & KNO₃ to be 78.6% at 30°C. This value is compatible with ours of 82%, within experimental uncertainty. When the humidity was decreased, particles effloresced at an RH of 58%.

Particles atomized from a 1 M solution of KCl & HNO₃ displayed gradual hygroscopic growth. Similar to atomized KNO₃ particles, heating overnight at 104°C removed adsorbed water, and deliquescence did not occur until the RH reached values >90%. EDS spectra showed that the particles consisted of K, O, and N, indicating that KNO₃ particles formed from the mixture and HCl gas was presumably liberated.

For particles atomized from solutions of KCl & K₂SO₄, the presumed KCl fraction of the particles began to take up water at an RH of 84% (Figure 8). This value is only 1% lower than the measured DRH for pure KCl (85%), which is within experimental uncertainty. The eutonic effect can vary depending on the composition of a mixed particle. For example, Tang and Munkelwitz (1994) observed that particles containing both NaCl and Na₂SO₄ took up water initially at 74.2%, an RH close to the RH for NaCl (75.3%). However, particles containing NaCl and NaNO₃ took up water at an RH of 68%.

Solid cores, presumably K₂SO₄, were visible within single droplets from RH values of 84% to >90% (Figure 8c). Full deliquescence occurred at an RH >90% (Figure 8d). The RH at which a mixture of KCl & K₂SO₄ fully deliquesced was reported to be 83.2% at 30°C (Adams and Merz 1929). The difference between our results and those of Adams and Merz (1929) could result from the temperature differences between these studies. Our experiments were performed at 15°C.

Particles from the solution of KCl & K₂SO₄ effloresced at an RH of 52%. This value is lower than that of either KCl (ERH 56%) or K₂SO₄(ERH 60%). To our knowledge this is the first report of the mutual ERH for particles from a KCl & K₂SO₄ solution. In a similar study (Cohen et al. 1987b), solutions of KCl & NaCl (1:1) or NaCl & (NH₄)₂SO₄ (1:1) also effloresced at values lower than the ERH of either of the single salts. However, if soluble inorganic salts are mixed with insoluble dust particles, the ERH of the soluble component increases because of heterogeneous nucleation (Han and Martin 1999).

Particles dried from a solution of KCl & H₂SO₄ took up water at an RH of 86% (Figure 9d). These particles did not deliquescence until the RH reached >90% (Figure 9e). When the RH was decreased, the particles effloresced at 60% (Figure 9f). These values for deliquescence and efflorescence match those of K₂SO₄. However, water uptake at an RH of 86% is also compatible with the DRH of KHSO₄ (Pilinis et al. 1989 ¹ ). At 86%, solid cores remained visible within droplets. If particles did not deliquesce and the RH was decreased below the DRH from 86% to 75%, the particles recrystallized.

FIG. 9 Particles from an equimolar solution of KCl & H₂SO₄ deposited onto a lacey-carbon TEM grid. All images are at the same magnification. Relative humidity was increased from (a) 0% to (e) >90% and then decreased to (f) 60%. Solid phases remain visible from an RH of (d) 86% to (e) >90% and are surrounded by a layer of water indicated by the open black arrows.

The difference in efflorescence that results from decreasing the RH from >90% to 60% or from 86% to 75% was caused by the solid cores that remained in the latter case. A similar result was observed by Lightstone et al. (2000) for mixed ammonium nitrate and succinic acid particles. Our results indicate that the solubility of slightly soluble components and the RH history of a particle are essential in order to successfully model mixed composition aerosol particles in the atmosphere. Our work corroborates recent results by Wang et al. (2008), who reported that the ERH can shift depending on the internal mixture of single particles.

SUMMARY AND DISCUSSION

We describe the deliquescence and efflorescence properties of KCl, K₂SO₄, and KNO₃ compounds that are common in atmospheric aerosols. ERH values for KNO₃, K₂SO₄, and mixtures of potassium compounds (e.g., KCl & K₂SO₄ and KCl & KNO₃) are reported here for the first time.

Atomized KNO₃ particles deposited onto a TEM grid show optical interference patterns when viewed with cross-polarized light, indicating that they were crystalline. This result differs from measurements where nitrate particles occurred as amorphous solids after they were atomized and deposited onto a surface (Hoffman et al. 2004; Liu et al. 2008). Efflorescence measurements provide information on the hygroscopic properties that could enhance gas-particle atmospheric equilibrium models, some of which assume that the water content for KNO₃ and K₂SO₄ below their deliquescence point is zero (Kim et al. 1993a, b).

Organic particles, soot, and tar balls dominate the aerosol produced through biomass burning and appear to be hydrophobic (Semeniuk et al. 2007). The hygroscopic properties of biomass aerosol largely depend on the fraction of inorganic particles. Semeniuk et al. (2007) showed that attached or included inorganic grains (e.g., KCl) in organic particles greatly enhance the water uptake ability of carbonaceous particles. Understanding the hygroscopic properties of the inorganic component of biomass-burning particles and their mixtures will allow a better estimate of the behavior of these particles under ambient RH conditions.

This material is based upon work supported by the National Science Foundation under Grant No. 0304213 from the Division of Atmospheric Chemistry. Any opinions, findings, and conclusions or recommendations expressed in this material are those of the authors and do not necessarily reflect the views of the National Science Foundation. We gratefully acknowledge the use of the facilities at the John M. Cowley Center for High Resolution Electron Microscopy within the Center for Solid State Science at Arizona State University. We also thank Karl Weiss, Weijun Li, and Dr. Kouji Adachi for useful discussions.

Notes

¹Atomized without heating.

²Atomized with heating.

³The RH at which the mixed particles began to take up water.

⁴The RH at which all components of the mixed particle were in the liquid phase.

⁵Calculated from data measured by Mullin (1961).

∗ETEM only.

¹Calculated from data measured by Mullin (1961).