Sources of insoluble inclusions in stratospheric sulfate particles




Abstract– Properties of aerosol collected in the stratosphere from altitudes of 20–45 km are reviewed. Removal of the soluble material from predominantly sulfate particles collected at 20 km revealed the presence of insoluble individual particles, or small groups of them, typically 40–50 nm in diameter. The size distribution of components of chain aggregates found above 35 km was almost identical, suggesting that rupture of the chains by condensing sulfuric acid, as they fell into the sulfate layer from above, was the source of the inclusions. Particles collected above 35 km on thin films of metal all showed the presence of a partially volatile liquid. On a copper surface, the liquid was stabilized, and of greater extent than the solid component. Three observations suggest that the upper stratospheric particles and their associated liquid were partly or wholly organic and derived from cometary dust too small to be heated on entering the atmosphere. These are: (1) the presence of a liquid that reacts with copper and the similarity to the behavior of particles collected on copper during a manned space flight, (2) their morphological similarity to published photographs of particles collected in the mesosphere from rockets, (3) the consistency with recent spacecraft observations of the size distribution of components sub-10 μm aggregates in cometary dust and the presence within them of carbon compounds.


Mossop (1965) described the particles of the stratospheric sulfate layer captured at an altitude of 20 km on 2000 km flights of the USAF U-2 aircraft in the Australian region between 29 January and 14 March 1963. Four of the flights ranged from about latitude 16S to 36S along longitude 145E, while the remainder had paths south of Tasmania between latitudes 40S and 45S. Particles were collected directly onto electron microscope grids cemented on a metal strip width of 3 mm placed vertically to the airstream in a housing mounted onto the fuselage of the aircraft. After a suitable exposure time, the strip was retracted into the aircraft housing. Control grids on the rear of the mounting strip were used to assess contamination. The collected particles were shown to be flattened domes with a modal diameter of about 750 nm consisting predominantly of ammonium sulfate. Grids were floated collection side up on pure water to remove the soluble material by dialysis through the nitrocellulose supporting membrane, revealing one or more much smaller insoluble inclusions in at least 95% of particles. The larger of these could be seen to consist of aggregates of particles with diameters of the order of 50 nm. Mossop considered that the inclusions had acted as nuclei for the formation of the sulfate particles, but was unable to obtain any diffraction patterns from them that would indicate their composition. If the inclusions were the particles on which the sulfuric acid condensed, their concentration would have influenced both the concentration and size of the particles that formed. Their possible sources are (1) residues from volcanic eruptions sufficiently large to inject particles into the stratosphere, (2) terrestrial particles, such as portions of carbon chains from forest fires carried aloft in tropospheric-stratospheric air interchanges, (3) products of ablation, vaporization, and recondensation of micrometeorites, or (4) unaltered meteoric material too small to have been significantly heated on entering the atmosphere. The first option seemed unlikely as no major volcanic eruptions had been reported for many years. The second could not be ruled out and there was no evidence to show that the particles had come from above. The aim here is to use the results of subsequent collections of stratospheric aerosol from much higher altitudes to try to find the source, or sources, of Mossop’s dialyzed particles.

Mossop’s Dialyzed Particle Collections

Mossop’s original electron microscope pictures still exist and now that high-resolution scanning is available, it is much easier to examine the particles in detail than it was in 1963. As Mossop’s 1965 paper may now be difficult to access, some of the typical dialyzed particles are shown in Fig. 1a, together with a composite picture of typical inclusions (Fig. 1b). Most of the particles that appeared to be single at low magnifications were in fact aggregates of two or more, the individual components being mainly 20–60 nm in diameter. Many of the particles had angular facets that make them unlike soot particles. Most were partially transparent to electrons. A feature that Mossop illustrated, but did not discuss, was the ring of insoluble material that marked the boundary of the original sulfate particle, illustrated in close-up in Fig. 1c. Close examination shows that few rings contained grains larger than 2 nm in diameter, most being smaller than the resolution limit.

Figure 1.

 a) Dialyzed sulfate particles, the rings marking their original boundaries, showing small inclusions within. b) Composite picture of inclusions within typical dialyzed particles the typical diameters of component particles being 20–60 nm. c) Portion of a typical ring surrounding a dialyzed particle.

Particulates in the Stratosphere from above 35 km

From 1968 to 1974, further sampling of stratospheric particles was carried out (Bigg 1975, 1976) from balloon-borne payloads released at sites at 34°S 142°E, 33°S 148°E, and 23°S 144°E. Particles were collected by impaction onto electron microscope grids mounted in a circle on a disk rotating below the nozzle of an impactor, giving a height profile of particle types. On some flights, a few of the grids were precoated with a thin film of copper, nickel, silver or aluminum to see if the particles caused any reaction. The impactor pump (a magnetically operated piston device) was turned on by a pressure switch acting at an altitude of 27 km on flights extending beyond 35 km and was turned off before the payload returned to the ground by parachute, where it was speedily recovered. All the flights above 35 km occurred at the times of stratospheric wind direction reversals, either in March or September–October. Twelve of the 14 flights originated at 34.2°S, 142.2°E, including the three highest which reached 39, 42, and 45 km. The remaining two originated at 33.1°S, 148.2°E. There were also more than 30 flights operating between 19 and 33 km, at various times of the year originating at 34°S 142°E or 23°S 144°E. For these, the pressure switch starting sampling was set to operate at 14 km.

In contrast to the particles which Mossop found to be mostly moist ammonium sulfate, liquid sulfuric acid particles dominated at 20 km, reacting completely with copper substrates to leave holes surrounded by crystals. It is possible that the eruption of Mt. Agung on Bali in March 1963 removed all the stratospheric ammonia and provided enough stratospheric sulfuric acid gas to keep its concentration low for many years. Concentrations of sulfuric acid or ammonium sulfate particles usually fell dramatically above about 27 km, indicating a common limit to upward diffusion of sulfur-containing gases sufficient to ensure nucleation. The dominant particles above 35 km on most flights consisted of branched chains having a maximum length of about 10 μm, but typical lengths of 1–2 μm. The example in Fig. 2a, captured on a carbon substrate, was unusual in showing an attached liquid (marked with an arrow). More often on a carbon surface, this was detached as a small particle near the chain, or was undetectable. Captured on a silver substrate, patches were always left where the silver grains had been flattened into a smooth surface (Fig. 2b), and these often extended to many times the diameter of the particle. This suggests the presence of a liquid that altered the surface before evaporating. Fig. 2c shows one of the less common compact aggregates captured on a thin film of nickel. The associated liquid is indicated with an arrow. When aggregates were captured on a grid coated with a thin film of copper, extensive liquid pools surrounded them (Fig. 2d). The liquid was stable in the high vacuum of the electron microscope and much larger in relation to the particle than seen on other substrates, suggesting that a reaction had stabilized a substantial volatile component. Electron dense spheres or aggregates of them were present, but were less than 5% of total particles. A few had portions of chain aggregates attached (Fig. 2e), but more often they were unaccompanied (Fig. 2f). Only those with attached chains showed any associated liquid.

Figure 2.

 a) Chain aggregate on a carbon surface. An arrow points to the associated liquid. Areas of dissolved silver grains were always found when a similar aggregate b) was captured on silver. c) A compact aggregate on a nickel surface, with an arrow marking associated liquid. d) On a copper surface, the extent of the liquid surrounding chain aggregates becomes much larger. e) Electron-dense spheres with a small attached chain aggregate on a silver surface, with some indication of an attached liquid. f) Electron-dense spheres on a carbon surface without any obvious liquid.

It is a reasonable guess that the spherical electron-dense particles of Figs 2e, f are in fact melted metallic particles from ablated meteoroids. As such, they would be unlikely to have recondensing liquids associated with them, unless they had subsequently collected an aggregate chain as suggested by Fig. 2e.

Up to 20% of the total particles from above 35 km had single compact centers and were also associated with a liquid that was more conspicuous on a silver or copper substrate than on carbon (Fig. 3).

Figure 3.

 Particles other than chain aggregates above 39 km. The top row on carbon substrates, E, F, and H on silver, G on copper. Note that all except B have been surrounded by a liquid, more extensive on copper or silver substrates.

Although the grids were prepared and loaded in a clean room, stray particles from various sources can occur on them. Hemispherical particles of ammonium sulfate that partially evaporate in the electron beam dominated in the outside air, but were not found in the collections above 39 km. A surrounding liquid was a good indication that the particles were collected at altitude, and its absence on particle B renders its source suspect.

Comparison of the size Distribution of Dialyzed Inclusions and Components of Chain Aggregates

The individual components of all the inclusions that Mossop (1965) photographed at ×10,000 magnification have been examined and the diameters of those whose outlines were not obscured by adjoining particles recorded. There were 357 cases. This is compared in Fig. 4 with the size distribution of 235 of the most easily sized particles in chain aggregates collected at heights above 39 km. The agreement is remarkably close.

Figure 4.

 Comparison of size distributions of components of insoluble nuclei in sulfate particles at 20 km altitude (solid line) and components of chain aggregates collected at heights above 39 km (dashed line).

Contamination Problems

Contamination from the balloon and associated payload has been suggested as a source of many of the particles appearing on grids sampled above 35 km. There are three reasons for believing this to be unlikely. (1) The balloon and payload had penetrated the turbulent troposphere and more stable lower stratosphere before sampling began at 27 km, during which time any detachable particles would have been lost. (2) Anyone who has attempted to micromanipulate micrometer-sized particles on a surface knows how strongly they resist detachment. The surfaces probably become electrically charged during flight, which would increase the adhesion. (3) The particles found in the balloon-borne collections have very close counterparts in those obtained from rockets in the mesosphere by Farlow et al. (1970), (see the Mesospheric Particles and Cometary Dust section). It is unlikely that the contaminants in their experiments would so closely resemble those in the balloon samples. However, chain aggregates pose a particular problem because they are formed when high concentrations of small particles of any sort coagulate. Soot chains are ubiquitous in continental atmospheres, morphologically not very different from the long chains of Fig. 2, and could therefore be collected during specimen handling. Even over remote oceans chain aggregates are found. They are thought to have been derived from coagulation of high concentrations of viruses, organic debris, or minerals in the ocean, and injected into the atmosphere by bursting bubbles (Leck and Bigg 2005). The feature that distinguishes the high-altitude aggregates from others is their reaction with copper. Although a copper substrate was only used on a few of the grids in each flight, all chains reacted with it, suggesting that contaminants were rare.

There was only one occasion in all the flights (13 March 1969) in which particles clearly of tropospheric origin were found at a height above 35 km. Brochosomes, very distinctive particles originating from leaf hoppers (Bigg 2003), were present, heavily coated with sulfate. Free acid was not observed, all particles being coated with ammonium sulfate, consistent with uplift of tropospheric air with a high ammonia concentration. Photographs of the horizon were taken at 600 m intervals during the entire ascent. A strong aerosol layer was present at 17 km, the altocumulus cloud horizon being clearly marked. Beyond about 25 km, aerosol scattering obscured both this layer and the horizon, which was unusual. The next flight above 30 km was on 26 March 1969 (32.3 km) and showed again a relatively large concentration of ammonium sulfate particles all the way to the top level and no sulfuric acid. On 21 May 1969 (32.1 km), strong aerosol layering was present at several levels, the horizon was clear at 32 km, and sulfuric acid particles once more dominated to 27 km.

An explosive volcanic eruption seems to be the most likely explanation for these anomalous observations. The most probable source was an eruption of Iya, Indonesia on 27–30 January 1969 at latitude 8.95°S according to the Global Volcanoes Program of the Smithsonian Institution. Although clearly a rare event, it implies that, on occasions, particles at that altitude are not necessarily of extraterrestrial origin. However, such particles differ considerably from the much rarer particles that dominated in most collections.

The higher the collection altitude of a particle other than inadvertent contaminants, the less likely that it is of terrestrial origin. Kornblum (1969) calculated the fall velocity of particles of various sizes in the mesosphere, showing that 250 nm particles would fall at about 8 km per day at 50 km. Particles like those of Fig. 3 would therefore be removed rapidly from heights above 40 km. As the chain aggregates have a much larger aerodynamic diameter than a sphere of equivalent mass, they would fall less rapidly than that, but should still fall out from 40 km in several weeks. The >30 degree temperature inversion that usually exists between 25 and 45 km is certain to inhibit all except explosive upward motions. For this reason, most weight has been given to particles obtained on the three highest flights at 45, 42, and 39 km, even though the particles were usually typical of most of those above 35 km.

Mesospheric Particles and Cometary Dust

Murphy et al. (1998) and Cziczo et al. (2001) found that about half the sulfate particles at heights up to 19 km contained meteoric material so that at least some of the particles observed above 35 km must have been of meteoric origin. Hemenway et al. (1964) and Witt et al. (1964) illustrated typical particle types obtained from rockets above 60 km that have closely similar counterparts in Figs. 2 and 3. They found liquid surrounding the particles and, because the collection passed through a noctilucent cloud, believed it to be water. The liquid attached to captured high-altitude stratospheric particles could not be water because it failed to evaporate in the electron microscope and reacted with copper. The reaction with copper of particles in space found by Brownlee et al. (1968) suggests that it may have been an organic liquid having a low vapor pressure. Farlow et al. (1970) showed electron microscope photographs of six particle types collected in the mesosphere from a rocket, of which all but Type 2 (compact aggregates) and type 5 (chain aggregates) were considered to be probable contaminants. However, Type 6 particles were also present in the upper stratosphere collections, suggesting that they may have been incorrectly described as contaminants. Although Farlow et al. stated that no liquid surrounded the particles, it is often difficult to detect on carbon coated grids.

The composition, method of formation, and morphology of most mesospheric particles remain relatively unexplored. Rosinski and Snow (1961) and Hunten et al. (1980) modeled the recombination of material ablated from incoming meteoroids, showing that a smoke of nanometer-sized particles would result. Havnes et al. (2009) also provided evidence of the presence of nanometer-sized particles loosely bound with ice that would be consistent with the “smoke” derived from the recombination of ablated meteoroids. Could these have formed the chain aggregates? Ablation products would be spread in a sloping line in a region where diffusion would very rapidly dilute the evaporated material. Recombination would result in a very wide size dispersion, the particles becoming ever smaller as the concentration of available material for nucleation and deposition diminished, down to the smokes of a few nanometer envisaged by Hunten et al. The result would be quite unlike the consistent size distribution of the components of upper stratospheric chain aggregates. In addition, liquid material would have evaporated at a higher altitude than the ablation occurred and, if not decomposed, would have condensed on pre-existing particles at that altitude. The fact that Murphy et al. (1998) found such a high proportion of sulfate particles at 19 km containing meteoric material could be explained by deposition of vaporized material on chain aggregates, which subsequently broke into smaller components that formed the nuclei for formation of sulfate particles.

Direct identification of mesospheric particles has been largely inconclusive. Witt et al. (1964) unsuccessfully attempted analyses using neutron activation. Electron microprobe analyses showed a few particles containing iron and nickel, and others with silicon or calcium, but mostly revealed no elements. Farlow et al. (1970) also could obtain no X-ray signatures of any element in their collected particles. Providing that their detectors were sufficiently sensitive, this could imply that the particles were largely composed of carbon or organic compounds. For this to be possible, their source would have to be dust particles sufficiently small to have survived entry into the atmosphere without substantial heating. Hunten et al. (1980) discussed the sizes of meteoroids that do not reach temperatures high enough to ablate. It is of the order of 100 μm diameter, far larger than the size of observed aggregates in stratospheric and mesospheric collections. It therefore seems unlikely that the chains are recombination products, and recent evidence suggests that such particles might exist in cometary dust.

The remarkable collection of cometary dust made by the Stardust spacecraft in the tail of comet 18P/Wild 2 revealed that the peak in the grain size distribution of aggregate particles was <100 nm diameter (Price et al. 2010). This is not in conflict with the grain size of the chain aggregates seen in Fig. 4. The Stardust collections also provided evidence of the presence of complex aromatic hydrocarbons (Clemett et al. 2010), the amino acid glycine (Elsila et al. 2009), and volatile organics (Bajt et al. 2009) in spite of the heating caused by rapid deceleration of the particles in the collecting medium. It is therefore possible that organic compounds might form at least part of aggregates entering the atmosphere that have been part of comets. It is worth noting that copper forms complexes with many organic compounds, including the glycine that has been detected in comet Wild 2.

Murphy et al. (2007) showed that carbon compounds were present in stratospheric sulfate particles at 19 km, but formed only a small part of the total mass. The concentration of particles at ∼40 km was less than 0.1% of those at 19 km, so that even allowing for increased concentration as they fall, their carbon content (if any) would not conflict with that observation.

Brownlee et al. (1968) used a copper surface for one of the collectors on the Gemini-12 space flight and unlike their gold collector, it had “an enormous amount of surface contamination” in the form of small particles surrounded by reaction areas 3–20 times the diameter of the actual particles. It seems likely from the balloon experience that the reactions were not due to contaminants, but to the liquid associated with collected particles. The role of the copper could have been to render far more particles visible by stabilizing their volatile component. If so, it implies that the copper-reactive material is also present in cosmic dust and it is therefore appropriate to consider the possibility of an extraterrestrial source for the particles found above 35 km and the liquid associated with them. The particles observed would have been small enough to enter the atmosphere without heating, providing an explanation for the similar reactions of upper stratospheric particles. The Stardust collections certainly suggest some support for the reality of organic-containing fine aggregates in cometary dust. Obviously additional evidence of composition and organic content of the fine aggregates in comets and in particles of the mesosphere and upper stratosphere is required before it can definitely be asserted that they are the same. However, it appears to be a plausible hypothesis that would explain the source of inclusions in sulfate particles at 20 km. If the source of particles at 40 km altitude is indeed cometary dust, collection from balloons at that altitude might be an easier way of assessing its nature than collecting it from rockets or spacecraft.

Nuclei for the Deposition of Sulfuric Acid

The remarkable similarity in the size distributions of components of chain aggregates above 35 km and the components of nuclei within sulfate particles (Fig. 1a, b) would be explained if the aggregates tended to separate into smaller units after encountering deposition of sulfuric acid. In support of this, chain aggregates above 35 km tended to be considerably larger than those at lower altitudes; within the sulfate layer, those that were found were usually <300 nm in length and formed only a very small proportion of the total particle population. If this is so, the individual components of the chains, or small groups of them, would form ideal nuclei for further deposition of sulfuric acid. Since Mossop found that almost all sulfate particles contained such inclusions, the implications are that concentrations of the lower stratospheric sulfate particles are controlled by the incoming micrometeorites. Size distributions, however, would probably be more strongly influenced by the concentrations of sulfur-containing gases.

The rings of diffuse material surrounding the dialyzed particles of Fig. 1a and shown in Fig. 1c have two possible sources. Either they were composed of the nanometer-sized recondensed smoke particles from ablation of meteors referred to in the Mesospheric Particles and Cometary Dust section, or they were composed of an insoluble liquid pushed to the outside of the sulfate particle by water entering during dialysis. It is not clear from Fig. 1c which is the more likely, but if it is the former, the grains were exceedingly small. In any case, as mentioned above, the size distribution of chain aggregates does not favor their origin as recondensed ablation products.


This work has shown the similarity between the typical particles found in the upper stratosphere and those found in the mesosphere. A new result is that upper stratospheric particles were associated with a liquid that reacted strongly with a thin film of copper and left impressions on thin films of other metals. Collection on copper surfaces has not been reported in mesospheric collections, but was used on a manned space flight where extensive reactions resulted. The size distribution of components of particles collected from comet Wild 2 by spacecraft and their associated chemistry are consistent with the properties of both the mesospheric and upper stratospheric particles. The maximum lengths of the upper stratospheric particles were about 10 μm. If extraterrestrial in origin, they would have entered the atmosphere without appreciable heating. The chain aggregates tended to disintegrate after an initial deposition of sulfuric acid as they fell below 30 km, and it appears from the similarity in size distribution to the nuclei of sulfate particles that the components then acted as centers for the subsequent formation of sulfate particles. This led to the supposition that the concentration of sulfate particles in the lower stratosphere is determined by the input of micrometeorites, although their size distribution would be more controlled by the input of sulfur-containing gases.

Editorial Handling–– Dr. Donald Brownlee