Data Systems Published by Copernicus Publications on behalf of the European Geosciences Discussions Union.

Abstract. As the relative humidity varies from high to low values in the atmosphere, particles containing organic species and inorganic salts may undergo liquid–liquid phase separation. The majority of the laboratory work on this subject has used ammonium sulfate as the inorganic salt. In the following we studied liquid–liquid phase separation in particles containing organics mixed with the following salts: ammonium sulfate, ammonium bisulfate, ammonium nitrate and sodium chloride. In each experiment one organic was mixed with one inorganic salt and the liquid–liquid phase separation relative humidity (SRH) was determined. Since we studied 23 different organics mixed with four different salts, a total of 92 different particle types were investigated. Out of the 92 types, 49 underwent liquid–liquid phase separation. For all the inorganic salts, liquid–liquid phase separation was never observed when the oxygen-to-carbon elemental ratio (O : C) g 0.8 and was always observed for O : C


Introduction
A large fraction of atmospheric particles contain both organic material and inorganic salts (Murphy et al., 1998(Murphy et al., , 2006;;Zhang et al., 2007;Pratt and Prather, 2010).The number of possible inorganic salts is relatively small with ammonium sulfate, ammonium bisulfate, ammonium nitrate and sodium chloride thought to be important (Finlayson-Pitts and Pitts, 2000;Seinfeld and Pandis, 2006).In contrast, the number of organic species can be in the 100s to 1000s (Hamilton et al., 2004;Goldstein and Galbally, 2007) while only around 10 % of these organic species have been identified at the molecular level (Hallquist et al., 2009).
Many studies on phase transitions of mixed organicinorganic salt particles of atmospheric relevance have focused on deliquescence and efflorescence (see for example Brooks et al., 2002Brooks et al., , 2003;;Choi and Chan, 2002;Chan and Chan, 2003;Wise et al., 2003;Braban and Abbatt, 2004;Pant et al., 2004;Parsons et al., 2004aParsons et al., , 2006;;Badger et al., 2006;Marcolli and Krieger, 2006;Salcedo, 2006;Ling and Chan, 2008;Treuel et al., 2009;Bertram et al., 2011;Smith et al., 2012).These studies have shown that when liquidliquid phase separation does not occur, the addition of organics to inorganic salts lowers the deliquescence and efflorescence relative humidities of the salts.As a result, some atmospheric particles may remain in a single homogeneous liquid state for the full range of relative humidities found in the atmosphere.
In the following we studied liquid-liquid phase separation in particles containing organics mixed with the following salts: ammonium sulfate, ammonium bisulfate, ammonium nitrate and sodium chloride.In each experiment one organic was mixed with one inorganic salt and the liquid-liquid phase separation relative humidity (SRH) was determined.Since we studied 23 different organics mixed with four different salts, a total of 92 different particle types were investigated.These studies provide insight into the effect of salt type on liquid-liquid phase separation in atmospheric particles.
All organic compounds studied herein were purchased from Sigma-Aldrich with purities ≥ 98 %, with the exception of suberic acid monomethyl ester and 1,2,6-hexanetriol, which were purchased from Sigma-Aldrich with a purity of 97 %, and glycerol, which was obtained from Thermo Fisher Scientific with a purity of 99.9 %.All organics were used without further purification.
The organic-to-inorganic mass ratio (OIR) in the particle was fixed at 2.0 ± 0.1 for most of the experiments.This value is in the range of OIR values observed in many field studies (Zhang et al., 2007;Jimenez et al., 2009).In addition, previous research using solutions or particles containing organics mixed with ammonium sulfate suggest that SRH often is not dependent on the OIR for a wide range of OIR values (Ciobanu et al., 2009;Bertram et al., 2011;Song et al., 2012a, b).As an example, Bertram et al. (2011) investigated the effect of OIR on SRH in eleven different types of particles containing organics and ammonium sulfate.For eight out of the eleven systems investigated, the SRH varied by less than 6 % for OIR values ranging from 0.1 to 10.As another example, Song et al. (2012b) measured SRH in particles containing ammonium sulfate and up to ten organics with OIR ranging from 0.17 to 2. The results of that study showed that for nine out of the fourteen systems that underwent liquidliquid phase separation, the SRH varied by less than 15 % as the OIR varied from 0.17 to 2.
Particles were generated by nebulizing (Meinhard) a solution of one organic compound and one salt, prepared in high-purity water (Millipore, 18.2 M cm) or in a mixture of water and methanol if the water solubility of the organic compound was less than 1 weight/weight %.The particle stream from the nebulizer was directed at a hydrophobic slide surface.As the droplets impacted on the slide surface, they coagulated into supermicron droplets.The water or the water/methanol mixture was then evaporated to generate organic-inorganic salt particles with lateral dimensions ranging from 10 to 35 µm.
The only two organic species that had a solubility of less than 1 % (weight/weight) in water were liquid diethyl sebacate and liquid suberic acid monomethyl ester.Based on this solubility information, liquid-liquid phase separation is expected for these species even without the presence of a salt at roughly ≥ 99 % RH.
The glass slide was mounted to a temperature and relative humidity controlled flow cell, which was coupled to an optical reflectance microscope (Zeiss Axiotech; 50 × objective) (Koop et al., 2000;Parsons et al., 2004b;Pant et al., 2006;Bodsworth et al., 2010).The temperature of the cell was held constant at 290 ± 1 K in all the experiments described here.To control the relative humidity in the flow cell, a combination of dry and humidified nitrogen gas was continuously passed through the cell.The total flow rate was approximately 1.5 L min −1 .The relative humidity of the gas was determined using a chilled mirror hygrometer (General Eastern), which was calibrated by measuring the deliquescence relative humidity of ammonium sulfate particles.
At the beginning of an experiment, the RH in the flow cell was first set to nearly 95 % and particles were allowed to equilibrate for about 15 min.The RH was then ramped down at a rate of 0.4-0.6 % min −1 , and images of the particles were captured approximately every 10 s until one of the following conditions occurred: liquid-liquid phase separation was observed, the particles effloresced, or an RH of ≤ 0.5 % was reached.For each different type of organic-inorganic salt particle, experiments were repeated at least three times.Roughly 5 particles were monitored in each experiment.For particles containing diethyl sebacate or suberic acid monomethyl ester, experiments started at RH = 100 ± 2.5 % since the RH of liquid-liquid phase separation was greater than 95 %.
While the RH was decreased, liquid-liquid phase separation could be identified from the recorded images.To illustrate this point we have included images and movies of particles containing 2-methylglutaric acid mixed with different inorganic salts as the relative humidity was decreased (Fig. 1 a-d and Movies S1-S4).The images and movies show that all particle types containing 2-methylglutaric acid at an OIR of 2.0 ± 0.1 underwent liquid-liquid phase separation.

Effect of H : C (hydrogen-to-carbon elemental ratio), O : C, and inorganic salt type on liquid-liquid phase separation
Listed in Table 2 are the measured SRH values for the different particle types investigated.Out of 92 particle types, 49 underwent liquid-liquid phase separation between 100 % and ≤ 0.5 % RH.These results are summarized in Fig. 2, which shows the dependence of observed liquid-liquid phase separation on the O : C and H : C (hydrogen-to-carbon elemental ratios) of the organic components of the particles.No trend with H : C is apparent for any of the salts studied.However, a trend with O : C is apparent: for all salts, liquid-liquid phase separation was never observed for O : C ≥ 0.8 and was always observed for O : C < 0.5.For 0.5 ≤ O : C < 0.8, the results depended on the salt type.For ammonium sulfate, phase separation was always observed for 0.5 ≤ O : C < 0.57 and phase separation was frequently observed for 0.57 ≤ O : C < 0.8.For the other three salts, phase separation was frequently observed for 0.5 ≤ O : C < 0.8.To further investigate the effect of O : C on liquid-liquid phase separation, the measured SRH values are plotted as a function of O : C in Fig. 3.The solid curves in the figure are fits to all the data using a Sigmoidal-Boltzmann function.Following Song et al. (2012b), the Sigmoidal-Boltzmann function was chosen to avoid physically unrealistic values at both low and high O : C values.Many of the systems in which phase separation was observed (i.e.SRH-values > 0 %) lie above the fit line since the curve was fit to both zero and non-zero SRH values.Alternatively we could have fit only the non-zero SRH values, but this would give extra weight to cases where phase separation was observed.The results of the fits are given in Table 3.The fit to the ammonium sulfate SRH-data is qualitatively consistent with fits previously reported in the literature for ammonium sulfate SRHdata (Bertram et al., 2011;Song et al., 2012b).
Although the results for each inorganic salt do not fall perfectly on the fit curves in Fig. 3, a correlation between O : C and SRH is observed.This suggests that O : C is a useful parameter for estimating, to a first approximation, the relative humidity for liquid-liquid phase separation, as shown previously for particles containing organics with ammonium sulfate (Bertram et al., 2011;Song et al., 2012a, b).For highaccuracy predictions, additional information such as the organic functional groups is required (Song et al., 2012b).Marcolli and Krieger recently measured SRH in bulk solutions containing one organic mixed with ammonium nitrate or sodium chloride (Marcolli and Krieger, 2006).Organics studied were 1,2-hexanediol (O : C = 0.33), 1,4-butanediol (O : C = 0.5), polyethylene glycol (MW = 400 and O : C = 0.56) and glycerol (O : C = 1).Only one of these organics, glycerol, was investigated in the current study (see Table 1).The SRH data from Marcolli and Krieger (2006) is included in Fig. 3 as a function of O : C (red triangles), together with the results of this study.The overlap between the current data and the data from Marcolli and Krieger (2006) suggest good agreement between the particle studies and the bulk studies.
SRH results shown in Fig. 3 do not vary drastically with the types of inorganic salt.However, out of the 23 organics investigated, the SRH of 20 organics followed the SRH trend: (NH 4 ) 2 SO 4 ≥ NH 4 HSO 4 ≥ NaCl ≥ NH 4 NO 3 (see Fig. 4a).In other words, the SRH-value measured with (NH 4 ) 2 SO 4 was greater than or equal to the SRH-value measured with NH 4 HSO 4 , and so on.Early in the last century, Randall and Failey showed the following trends for the salting out efficiencies of ions relevant to our work: Randall and Failey, 1927).In addition, the Hofmeister series, which consists of a ranking of cations and anions in terms of their ability to salt out proteins follows the same trend (i.e.Na + > NH + 4 and SO 2− 4 > Cl − > NO − 3 ) (Hofmeister, 1887(Hofmeister, , 1888;;Kunz et al., 2004).These trends allow one to compare the salting out efficiency (or SRH) of two salts if they have a common cation or anion.Based on these early salting out studies, we would expect (NH 4 ) 2 SO 4 to have a greater salting out efficiency (or higher SRH) compared to NH 4 HSO 4 since the salting out efficiency follows the trend NH + 4 > H + .In addition, we would expect (NH 4 ) 2 SO 4 to have a greater salting out efficiency (or higher SRH) than NH 4 NO 3 , since the salting out efficiency follows the trend SO 2− 4 > NO − 3 .These expectations are consistent with the trends observed for 20 out of the 23 organics investigated (Fig. 4a).
Three organics (2,5-dihydroxybenzonic acid, α, 4-dihydroxy-3-methoxybenzeneaceticacid, and 2,2dimethylsuccinic acid) were inconsistent with the SRH trend (NH 4 ) 2 SO 4 ≥ NH 4 HSO 4 ≥ NaCl ≥ NH 4 NO 3 when using an OIR of 2.0 ± 0.1.The results from these organics are illustrated in Fig. 4b.Interestingly, Bertram et al. (2011) also measured SRH values for these organics mixed with ammonium sulfate and found that the SRH values for these organics varied by more than 6 % with the OIR of the particles.In fact, the SRH of these three organics were the only ones observed to vary by more than 6 % with OIR out of the 13 organics studied by Bertram et al. (2011).To investigate this further we have measured SRH for these three organics mixed with the different salts using OIR values lower than 2.0 ± 0.1.The results from these measurements as well as the results from Bertram et al. (2011) are shown in Fig. 5 and are summarized in Table 4. Figure 5a shows that for OIR ≤ 0.5, SRH of 2,2dimethylsuccinic acid followed the trend of (NH 4 ) 2 SO 4 ≥ NH 4 HSO 4 ≥ NaCl ≥ NH 4 NO 3 , consistent with the 20 organics shown in Fig. 4a.Hence the anomaly at OIR = 2 for 2,2-dimethylsuccinic acid is absent at an OIR of ≤ 0.5.In other words, the trend for 2,2-dimethylsuccinic acid is consistent with the other 20 organics as long as the OIR is in a range where particles undergo liquid-liquid phase separation.
Figure 5b and 5c show that α,4-dihydroxy-3methoxybenzeneacetic acid and 2,5-dihydroxybenzonic acid do not follow the trend of (NH 4 ) 2 SO 4 ≥ NH 4 HSO 4 ≥ NaCl ≥ NH 4 NO 3 regardless of the OIR studied.Note out of all organics studied, these two organics are the only two organics which contained aromatic functional groups.One possible reason for the differences in trends of SRH observed for these organics may be due to strong cation-π interactions (Kumpf and Dougherty, 1993;Ma and Dougherty, 1997;Song et al., 2012b).Previous work by Song et al. has suggested that cation-π interactions may decrease the salting out effect of the ammonium cations, thereby influencing SRH values in particles containing ammonium salts and aromatic compounds (Song et al., 2012b).
SRH was only investigated as a function of OIR for three organics (2,5-dihydroxybenzonic acid, α,4-dihydroxy-3-methoxybenzeneacetic acid, and 2,2-dimethylsuccinic acid).For the organics that were only investigated at OIR=2.0 ± 0.1, we don't expect that SRH will be a strong function of OIR in most cases, based on previous studies with organics and ammonium sulfate (see Experimental Section).However, experiments are needed to confirm this expectation.
Recent work has shown that chloride anions may react with organic acids in the particle phase to form organic salts (Laskin et al., 2012).This type of reaction could potentially occur in our studies.However, if such reactions are occurring in our studies, they don't appear to drastically affect the occurrence of liquid-liquid phase separation.

Atmospheric implications
A large fraction of submicron particles in the atmosphere contain organics mixed with some combination of ammonium (NH + 4 ), protons (H + ), sulfate (SO 2− 4 ) and nitrate (NO − 3 ) (Adams et al., 1999;Lee et al., 2003;Martin et al., 2004;Tolocka et al., 2005;Murphy et al., 2006;Seinfeld and Pandis, 2006;Zhang et al., 2007;Pratt and Prather, 2010).As a result the inorganic salts ammonium sulfate, ammonium bisulfate and ammonium nitrate are thought to be important in atmospheric particles.As mentioned in the Introduction, most of the previous laboratory work on liquid-liquid phase transitions of atmospheric importance have used (NH 4 ) 2 SO 4 as the inorganic salt, even though sulfate is not always fully neutralized in atmospheric particles and NO − 3 can make up a large fraction of the inorganic anions under certain conditions (Dibb et al., 1996;Huebert et al., 1998;Tolocka et al., 2005;Murphy et al., 2006;Zhang et al., 2007;Pratt and Prather, 2010).To address this disconnect, we have carried out liquid-liquid phase separation experiments with particles containing organics mixed with (NH 4 ) 2 SO 4 , NH 4 HSO 4 and NH 4 NO 3 .The results from these studies show that in all cases, liquid-liquid phase separation is a common occurrence when O : C < 0.8 and always observed when O : C < 0.5.These ranges of O : C values are frequently observed in the atmosphere, suggesting that liquid-liquid phase separation is a common process in atmospheric particles, regardless of the identity of the salt.
In the marine boundary layer, supermicron particles containing NaCl make up a large fraction of the particulate mass (Quinn and Bates, 2005;Seinfeld and Pandis, 2006).These particles, which are produced from a bubble bursting mechanism (Woodcock et al., 1953;Blanchard and Woodcock, 1957), can often contain relatively low O : C organics such as sterols, fatty acids and fatty alcohols (Schneider and Gagosian, 1985;Peltzer and Gagosian, 1987;Sicre et al., 1990;Kawamura et al., 2003).Based on the liquid-liquid phase separation results for NaCl containing particles presented here and the O : C of sterols, fatty acids and fatty alcohols (O : C less than approximately 0.5) thought to be present in the marine boundary layer, liquid-liquid phase separation is also expected to be a common occurrence in marine environments.

Conclusions
Out of 92 types of particles studied, 49 underwent liquidliquid phase separation.For all the inorganic salts, liquidliquid phase separation was never observed when O : C ≥ 0.8 and was always observed for O : C < 0.5.For 0.5 ≤ O : C < 0.8, the results depended on the salt type.In addition, a correlation between the separation relative humidity (SRH) and O : C was observed for all inorganic salts, suggesting that O : C is a useful parameter for estimating, to a first approximation, the relative humidity for liquid-liquid phase separation, although additional information will be required for predictions with high accuracy.
Out of the 23 organics investigated, the SRH of 20 organics had the following trend: (NH 4 ) 2 SO 4 ≥ NH 4 HSO 4 ≥ NaCl ≥ NH 4 NO 3 .The trend is consistent with previous salting out studies and the Hofmeister series.Based on the range of O : C values found in the atmosphere and the current results, liquid-liquid phase separation is likely a common occurrence in marine and non-marine environments.
Table 4. Summary of SRH results as a function of the organic-to-inorganic (OIR) mass ratio for following organics: 2,2-dimethylsuccinic acid; α,4-dihydroxy-3-methoxybenzeneacetic acid; and 2,5-dihydroxybenzonic acid.Included are results from both the current studies and results from Bertram et al. (2011)  a Not observed means liquid-liquid phase separation was not observed for the range of relative humidities probed.In some cases SRH was not probed below 20-40 % RH since at RH values less than these values efflorescence of the inorganic salt occurred.b Data from previous study of Bertram et al. (2011).

Fig. 1 .
Fig. 1.Optical images and illustrations of particles containing 2methylglutaric acid and one of the inorganic salts with OIR = 2.0 ± 0.1.Shown in the images and illustrations are the relative humidities at which the images were recorded.The inorganic salts studied were ammonium sulfate in row (a), ammonium bisulfate in row (b), sodium cholride in row (c), and ammonium nitrate in row (d).The diameter of the particles shown ranged from 28 to 34 µm.

Fig. 2 .
Fig. 2. Van Krevelen Diagram for the different mixed organicinorganic salt particles (OIR = 2.0 ± 0.1) studied: (a) organicammonium sulfate particles, (b) organic-ammonium bisulfate particles, (c) organic-sodium chloride particles and (d) organicammonium nitrate particles.Open circles indicate that liquid-liquid phase separation was observed, while stars indicate that liquidliquid phase separation was not observed.The vertical hatched regions correspond to the H : C and O : C conditions when liquidliquid phase separation was always observed and the horizontal hatched regions correspond to the H : C and O : C conditions when liquid-liquid phase separation was never observed.

Table 3 .
Parameterizations of SRH results as a function of the oxygen-to-carbon elemental ratio (O : C) of the organic material.

Fig. 3 .
Fig. 3. Summary of SRH results (OIR = 2.0 ±0.1) as a function of oxygen-to-carbon elemental ratio (O : C): (a) organic-ammonium sulfate particles, (b) organic-ammonium bisulfate particles, (c) organic-sodium chloride particles and (d) organic-ammonium nitrate particles.Circles represent the relative humidity at which separation occurred.Error bars associated with the circles represent 2σ of multiple SRH measurements and the uncertainty from the calibration.Stars indicate that liquid-liquid phase separation was not observed.The error bars corresponding to the stars indicate that liquid-liquid phase separation could potentially occur within the range indicated by the error bars, but could not be detected due to the occurrence of efflorescence in the particles.The curves in the panels are Sigmoidal-Boltzmann fits to the data.Red triangles represent the results of liquid-liquid phase separation in bulk solution from Marcolli and Krieger 2006).

Fig. 5 .
Fig. 5. Summary of SRH results as a function of OIR for the following types of particles: (a) 2,2-dimethylsuccinic acid and inorganic salts, (b) α,4-dihydroxy-3-methoxybenzeneacetic acid and inorganic salts, and (c) 2,5-dihydroxybenzonic acid and inorganic salts.Closed symbols represent results from the current study, while open squares represent results from previous studies by Bertram et al. (2011).Uncertainties in the SRH measurements have been left off for clarity.

You et al.: Liquid-liquid phase separation in particles containing organics 2002;
al., Y.

Table 1 .
Summary of different organics used in the liquid-liquid phase separation experiments.

Table 2 .
Summary of SRH results for an organic-to-inorganic mass ratio (OIR) of 2.0 ± 0.1.Uncertainties represent 2σ of multiple SRH measurements and the uncertainty from the calibration.
. Uncertainties represent 2σ of multiple SRH measurements and the uncertainty from the calibration.