Volume 41, Issue 3
Free Access

Ion-aerosol-cloud processes in the lower atmosphere

R. G. Harrison

R. G. Harrison

Department of Meteorology, University of Reading, Reading, UK

Search for more papers by this author
K. S. Carslaw

K. S. Carslaw

Institute for Atmospheric Science, School of the Environment, University of Leeds, Leeds, UK

Search for more papers by this author
First published: 16 September 2003
Citations: 278


[1] Natural terrestrial radioactivity and cosmic ray ionization lead to the formation of air ions and charged aerosol particles even away from regions of active charge separation, such as in thunderstorms. The natural electrified state of the atmosphere has been studied for over a century; however, the effect of ionization on the physical properties of aerosols and clouds has rarely been studied in its own right except in thunderstorms. Here we review the status of our understanding of atmospheric charged particles and their influence on aerosol and cloud microphysical processes. We also review mechanisms that have been recently proposed to connect variations in the atmospheric ionization rate with variations in global cloudiness and weather systems. We conclude that a mechanism linking cosmic ray ionization and cloud properties cannot be excluded and that there are established electrical effects on aerosol and cloud microphysics. Necessary further work includes measurements of cloud, droplet, and aerosol charging and ion-aerosol conversion, together with modeling of the electrical aspects of nonthunderstorm cloud microphysics.


[2] Fair weather atmospheric electricity and the solar-terrestrial electrical environment have been studied for many years. The serious study of atmospheric electricity began with the experiments of B. Franklin, J. Canton, G. Becarria, and C. A. de Coulomb more than 200 years ago [Chauveau, 1925; Chalmers, 1967; Herbert, 1997]. Early observations showed that the electric field in the lower atmosphere responded strongly to meteorological changes [Everett, 1868], and Lord Kelvin suggested that future weather forecasting would be possible with an electrometer [Thomson, 1884].

[3] Of central importance to the physics connecting the atmospheric electrical system with clouds and weather is the behavior of cosmic rays, which are modulated by solar activity and are themselves modulators of the atmospheric electrical system. Cosmic rays generate ions deep within the troposphere and control the atmospheric electric field even down to the planetary surface. Compelling reasons to examine the atmospheric effects of cosmic rays have arisen in recent years: Theory now links the ions produced by cosmic rays with aerosol formation and changes in the rates of aerosol coagulation and removal of particles by water droplets in clouds. The myriad processes of cloud physics therefore occur within a weak ionic plasma, susceptible to changes on all spatial and temporal scales.

[4] Cosmic ray ionization is ubiquitous in the Earth's atmosphere, making up ∼20% of the ionization over land surfaces [Hoppel et al., 1986], becoming more significant with altitude and dominating ion production above the planetary boundary layer, typically within 1 km of the surface. Cosmic rays are the principal source of ionization over the oceans. The molecular ion pairs produced have concentrations at the land surface of 500 cm−3 [Chalmers, 1967], rising to a maximum of 1000–3000 cm−3 at 15 km [Ermakov et al., 1997]. In the free troposphere the number concentration of ions is comparable to the number concentration of droplets found in stratiform liquid water clouds [Pruppacher and Klett, 1997]. Earth's vertical tropospheric electric field is known to respond to changes in cosmic rays [Markson, 1981].

[5] Atmospheric electrical variables are highly dependent on a wide range of atmospheric processes, ranging from scales of a few meters on the edges of clouds up to the global scale; see Figure 1. For example, the surface electric field is as sensitive to variations in the turbulent flux of local space charge and aerosol pollution as it is to variations in the global thunderstorm activity [Barlow and Harrison, 1999; Harrison and Aplin, 2002]. In contrast, the electrical state of the atmosphere has been viewed as having an insignificant effect on the physical, chemical, and meteorological processes that drive such variations [Hoppel et al., 1986]. Consequently, atmospheric electricity has been seen as largely irrelevant to both the chemistry of air pollution and global meteorology.

Details are in the caption following the image
Summary of the atmospheric processes in which ions and electrification are implicated, as discussed in this paper. Cosmic rays and the turbulent transport of surface radioactivity lead to the formation of molecular ions, with cosmic rays the dominant source over the oceans and at high altitudes. The molecular ions exchange charge with atmospheric aerosol and cloud particles, by collision, and also contribute to the formation of ultrafine aerosols. In regions of disturbed weather, thunderstorms produce substantial local electric fields, causing positive charge to be carried to the ionosphere and negative charge to be carried to the surface. The ionosphere is conductive, and the positive potential it carries with respect to the surface is distributed around the planet to regions of fair weather, and a small current (marked as the fair-weather drift current) flows vertically. This generates the vertical fair-weather electric field. In regions containing layers of nonthunderstorm cloud, the electric field may be enhanced.

[6] The importance of atmospheric electricity, and cosmic ionization in particular, to cloud and atmospheric processes was realized by C. T. R. Wilson [Galison, 1997]. Variations in ionization were suggested [Ney, 1959] as an explanation for changes to weather and climate [Bering, 1995]. Theoretical considerations [Mason, 1971] prevent direct nucleation of water droplets on ions at atmospheric supersaturations, but it has become apparent that there may be other effects of ionization. Natural and laboratory-generated aerosols may acquire substantial electrical charges, which can influence their dynamical behavior. Coagulation rates [Fuchs, 1964] and aerosol removal rates by droplets (scavenging) [Pruppacher and Klett, 1997] can be greatly affected by electrical charging of aerosols, which can be further enhanced if the particles are radioactive [Clement et al., 1995; Tripathi and Harrison, 2001]. High levels of radioactive ionization [Bricard et al., 1968, 1972; Megaw and Wiffen, 1961] can produce ultrafine aerosol particles in artificial and filtered air, and particle formation has also been reported [Vohra et al., 1984] at atmospheric levels of radioactivity. Although radiolytically produced ultrafine aerosol particles produced by ionization would have to persist and grow substantially to permit water condensation at atmospheric supersaturations, it is not justified to completely neglect possible direct or indirect electrical influences on aerosols and clouds [Harrison, 1997, 2000]. Recent modeling work [Yu and Turco, 2001] indicates that small ions can provide a source of atmospheric cloud condensation nuclei, which indicates a potential effect on clouds and ultimately climate [Carslaw et al., 2002] Many studies have suggested that variations in cosmic rays on the scales of days and years influence global cloudiness [Svensmark and Friis-Christensen, 1997], cloud cover [Todd and Knievton, 2001], mid-latitude cyclones [Tinsley and Deen, 1991] and high-level cloud [Pudovkin and Veretenenko, 1995]. Variations in ionization have also been suggested to account for variations in the planetary albedo of Neptune [Lockwood and Thompson, 1979; Moses et al., 1992].

[7] The electrical properties of aerosols are well recognized in sciences outside geophysics. Aerosol electrification modifies aerosol deposition in the human lung [Hashish et al., 1988], including deposition of therapeutic aerosols [Cohen et al., 1998; Hashish and Bailey, 1991]. Modifications to particle deposition may also occur as the result of corona ions from overhead power lines [Fews et al., 1998]. Domestic ionizers are designed to release large quantities of negative ions, causing unipolar aerosol charging. There are known biological effects of small ions [Kellogg, 1984], and bacteria may be killed or their growth inhibited by positive or negative ions [Krueger and Reed, 1976]. Additional complication in evaluating human health effects include the possible electrically enhanced removal of bacterial aerosol [Harrison, 1995] and the production of hydrogen peroxide from negative ions in some circumstances [Richardson et al., 2003]. Excesses of negative ions have been found to relieve the symptoms of hay fever [Palti et al., 1966], reduce pain and promote healing in patients suffering from burns [Kornblueh, 1968], modify reaction times [Hawkins and Barber, 1978], and reduce the incidence of reports of headaches, nausea, and dizziness in office workers [Hawkins, 1982].

[8] Although electrification is recognized to influence particle microphysical processes to some degree, it is far from clear that variations in the atmospheric electrical state will lead to observable changes in aerosol or cloud properties. The influence of the atmospheric electrical state on global meteorology and planetary cloudiness caused by variations in aerosol and cloud properties [e.g., Hobbs, 1993; Charlson and Heintzenberg, 1994] would, if confirmed by testable physical mechanisms, be important for atmospheric science. However, establishing a causal link between large-scale meteorological processes and small variations in the atmospheric electrical parameters is hampered in several ways. First, our basic understanding of the electrical state of the atmosphere itself is incomplete [Krider and Roble, 1986], as high-resolution global monitoring of clouds and lightning is in its early stages. Second, any influence of electrical variability on clouds and aerosols is likely to be masked by large natural variability. Third, appropriate models and measurements relating the electrical state of the atmosphere and aerosol and cloud processes do not yet exist with which to test possible links. Fourth, even if a link between small variations in electrical parameters and local cloud processes could be established, an extrapolation from the local effects to global climate scales would be complex.

[9] The goal of this paper is to review the “state of the science” regarding atmospheric electrical processes and their effect on aerosol and cloud microphysical processes. We begin by reviewing our understanding of the basic physical processes involved in atmospheric electrification (section 1) and atmospheric ionization (section 2) through to aerosol electrification (section 3) and the effects of such electrification on aerosol physical processes. Such a didactic review is necessary because much of the literature on this subject appears to be largely unknown to practitioners of modern atmospheric aerosol science. Our summary of aerosol electrification studies is the first to be aimed at those interested in aerosol physical properties and aerosol-cloud interactions rather than the electrical properties of the atmosphere. We then summarize the studies that have suggested a link between the electrical state of the atmosphere and weather (section 4) and examine the magnitude of electrical effects in each case. Section 5 considers ion-aerosol-cloud processes on other planets in our solar system. We close by outlining what further studies are needed to advance the field, emphasizing the need to formulate hypotheses amenable either to physical experiments or numerical modeling (section 6.)


[10] The Earth has an atmospheric electrical system. It consists of a few regions of intense electrical activity and charge separation (thunderstorms) and large areas with limited charge separation where only small currents flow, driven through the considerable resistance of air by the surface-ionosphere voltage difference [Chalmers, 1967; Bering,1995; Rycroft et al., 2000]. Electric currents flow vertically in thunderstorms [MacGorman and Rust, 1998] where charge is separated by cloud-particle processes, causing a positive charge to pass to conducting regions in the upper atmosphere (at ∼80 km) and causing a negative charge to pass to the planetary surface by the combined effect of lightning and point discharge. Fossil evidence [Harland and Hacker, 1966] shows that electrical activity has existed on the planet for at least 250 Myrs, and lightning may even be implicated in the formation of amino acids in early atmospheres [Miller, 1953] This review is not directly concerned with thunderstorm electrification processes, although the cumulative effect of thunderstorms on the global electrical system is central to the work discussed.

[11] Although it is not straightforward to eliminate all possible sources of charge separation contributing additional electric fields (e.g., sea spray and charged dust production), a convention of regarding the nonthunderstorm regions of the atmosphere as fair-weather regions has arisen. In fair-weather conditions it is assumed that no local processes of charge separation arise from any cause and that the atmospheric electrical quantities are steady, modulated only by the global circuit. Under fair-weather conditions a voltage gradient can be measured just above the Earth's surface using a very high impedance voltmeter [e.g., Harrison, 2002a]. The voltage gradient is caused by the large potential difference between the lowest layer of the ionosphere, the electrosphere, and the surface. The resulting surface potential gradient in fair weather is ∼150 V m−1 but can be as high as about 105 V m−1 in thunderstorms before a lightning discharge [Krider and Roble, 1986]. (There is sometimes confusion between the atmospheric electrical convention of calculating the vertical potential gradient as +dV/dz and electrostatics, in which electric field E = −dV/dz. The term “potential gradient” (PG) is in widespread use and will be adopted here. The PG is positive in fair weather.)

[12] The majority of free charge in the fair-weather part of the atmosphere arises from natural ionization, principally from radon, cosmic rays, and terrestrial gamma radiation [Chalmers, 1967]. Partitioning between the sources varies vertically, with cosmic ray ionization dominating away from the continental surface. Over the open oceans, cosmic rays are the principal source of ionization even down to the surface. A steady state space charge density is generated from the balance between the ionization rate and the recombination of positive and negative ions formed. (No free electrons can be sustained, because of the presence of atmospheric water vapor.) Collisions between ions and aerosols lead to electrification of the aerosol and ion removal. The space charge density ρ is
equation image
where e is the magnitude of the electronic charge, n+ and n are the air ion concentrations and there are dNj(a) aerosol particles with radii from a to a + da per unit volume carrying j electronic charges. A mean value for the surface layer space charge under fair-weather conditions is about 10 pC m−3 (60e cm−3) [Bent and Hutchinson, 1966]. Variations in vertical electric field E and space charge are coupled by Poisson's equation
equation image
where z is altitude and ε0 is the permittivity of free space.

[13] Variations in the space charge density and, in particular, the generation of regions of unipolar space charge, are likely to be an important factor in aerosol processes, as we show in section 4. There are many sources of variability in the space charge density. Adkins [1959] found a difference between the polarity of space charge in rain and snow. Melting snow has been observed to release negative charge, giving −5500e cm−3 between 0 and 1m above the snow surface, when the space charge measured between 1 m and 21 m was +770e cm−3 [Bent and Hutchinson, 1966]. Particulate atmospheric space charge is transported like any other atmospheric constituent, through turbulent diffusion and advection [Whitlock and Chalmers, 1956]. Under quiescent micrometeorological conditions, such as those around sunrise [Law, 1963], an electrode effect leads to enhanced positive ion concentrations within centimeters of the negatively charged planetary surface.

[14] The electrospheric (or ionospheric) potential VI and, in fair-weather conditions, the surface potential gradient vary with the electrical activity of global thunderstorms. In fair weather, with no local effects, the diurnal cycles in surface potential gradient and VI are similar [Markson, 1985], resulting from the diurnal cycle in global heating and thunderstorm activity [Whipple, 1929; Price, 1993]; see Figure 2. It is therefore assumed that thunderstorm activity modulates the global atmospheric electric circuit, and recent evidence exists of electrical discharges passing from the upper part of a thundercloud to the lower ionosphere [Pasko et al., 2002]. The diurnal variation was identified in oceanic potential gradient measurements made during voyages of the wooden geomagnetic research vessel Carnegie in the early part of the twentieth century: It is a characteristic global signature present in many atmospheric electrical measurements in unpolluted air. Averaging is, however, usually necessary to remove local effects, but there is evidence to suggest the global diurnal variation has remained very similar for the majority of the twentieth century [Harrison, 2003a]. Multidecade decreases in the potential gradient at continental sites [Harrison, 2002b; Märcez and Harrison, 2003] may have resulted from aerosol changes [Williams, 2003] or geophysical changes reducing cosmic rays [Harrison, 2003b].

Details are in the caption following the image
Diurnal variation in (a) the surface Potential Gradient (relative units, left-hand axis) and ionospheric potential (right-hand axis) and (b) active thunderstorm areas of each of the continents (redrawn from Price [1993]). The universal form of the diurnal variation is generally known as the Carnegie curve, and the similarity between the thunderstorm area and potential gradient variations was found by Whipple and Scrase [1936].
[15] The surface-electrosphere potential difference VI causes an ionic leakage current to flow vertically. Currents of order 2000 A flow in the global circuit. Applying Ohm's law with VI ∼ 300 kV, gives a global atmospheric electrical resistance RT = 230 Ω. Variations in RT arise from changes in ion concentrations: RT has its principal contribution from the planetary boundary layer, because of ion removal by aerosol. The concentric sphere system formed by the electrosphere and the planet has a finite capacitance C, with a time constant RTC of ∼10 min [Chalmers, 1967]. The continued existence of an atmospheric electric field indicates that charge generation processes are continuously active. Table 1 gives fundamental properties of the atmospheric electric system. The atmospheric electric circuit's total current of ∼2000 A can be expressed as a vertical conduction current density Jz over the Earth's surface, which is ∼2 pA m−2. It is determined by VI and the resistance Rc of a unit column between the surface and the electrosphere, which is given by
equation image
where σ(z) is the local air conductivity at altitude z. (The upper limit of the integral represents the high-conductivity region of the electrosphere.) The vertical potential gradient at z is related to Jz and σ by
equation image
For bipolar ion number concentrations n+ and n the total conductivity is given by
equation image
where μ+ and μ are the average ion mobilities and σ+ and σ the associated polar air conductivities. Rapid direct measurement of the conduction current Jz is difficult, so it is usually derived from simultaneous conductivity and electric field measurements. By such an approach the conduction current is found to be constant vertically in stable, fair-weather conditions [Gringel et al., 1978].

[16] The local potential gradient can deviate considerably from its average fair-weather value, typically up to ±2000 V m−1 in situations of disturbed weather. Small-scale potential gradient variations are caused by convection and advection of regions of local space charge as well as by processes that can locally alter the space charge density or ion mobility, in particular, fog, clouds, precipitation, or pollution with enhanced aerosol loading [Sagalyn and Faucher, 1955]. Precipitation from unstable air such as that in a cold front causes rapid fluctuations in both sign and magnitude. Thus, although the Carnegie cycle in potential gradient can be detected in a long-term mean of surface potential gradient measurements, lower tropospheric potential gradients are dominated by high temporal and spatial variability, especially in aerosol-laden air.


3.1. Atmospheric Ionization

[17] Air ions or “small ions” are produced by the natural radioactivity of the air and by cosmic radiation. The air's natural radioactivity consists of airborne alpha emitters such as radon and direct gamma radiation from the soil. Cosmic radiation consists of extremely high-energy (>GeV) particles, mostly protons and helium nuclei [Hillas, 1972]. Both forms of radioactivity separate an electron from a molecule of nitrogen or oxygen, which is subsequently captured by neutral molecules on a very short timescale. At the continental surface the small ion production rate is ∼10 ion pairs cm−3 s−1, which varies with the air density. In continental regions the relative contributions from air, soil, and cosmic origins are about 4.5 ion pairs cm−3 s−1, 3.5 ion pairs cm−3 s−1, and 2 ion pairs cm−3 s−1, respectively, although the relative contributions vary with local geology [Bricard, 1965].

[18] In addition to radioactive ionization, which produces positive and negative ions in equal numbers, a net charge can be introduced into the atmosphere by processes such as combustion, rainfall, and breaking ocean waves. Ionization by ultraviolet radiation is unimportant in the lower atmosphere, although this is the dominant source of ions above 60 km altitude. Ions are also formed by corona discharge from points or branches of trees having a high potential difference with their surroundings. The amount of atmospheric point discharge is a sensitive function of wind speed and electric field [Whipple and Scrase, 1936]. A classical experiment [Schonland, 1953] measured point discharge currents of the order of microamperes from an uprooted, electrically isolated tree.

[19] Two geophysical factors modulate atmospheric cosmic ray ionization: geomagnetic latitude and solar variability. The combination of geomagnetism and the cosmic ray energy spectrum leads to a distribution in the cosmic rays entering the atmosphere, only a small fraction of which penetrates to the Earth's surface. Very high energy cosmic radiation is necessary to penetrate the geomagnetic field and enter the atmosphere at low geomagnetic latitudes, whereas much lower energy cosmic radiation enters the atmosphere at higher geomagnetic latitudes. The principal deviations of geomagnetic from geographic latitude [Wolfendale, 1963] are a southward shift up to 15° over South America and a 10° northward shift over Africa through South Asia.

[20] Solar activity modulates the heliospheric magnetic field, which acts as a shield to low-energy cosmic radiation. At periods of high solar activity the cosmic ray flux reaching Earth is reduced, although the modulation principally affects the lower-energy cosmic radiation [Hillas, 1972]. Figure 3a shows the variation of cosmic ray ionization and ion concentrations with height, including the limits expected from the solar modulation over one solar cycle. Long-term surface measurements of cosmic radiation also show the effect of the solar cycle modulation (Figure 3b) at the neutron monitor stations operated by the University of Chicago. The low-latitude station at Huancayo shows both a smaller mean count level and a smaller modulation. Hensen and van der Hage [1994] give a simple parameterization for the variation in surface cosmic ray ionization rate with latitude and time. Many significant properties of cosmic rays were identified by S. E. Forbush, who served on the geomagnetic survey ship Carnegie and survived its final fire in 1929 [Van Allen, 1998]. Shorter-term cosmic ray variations associated with sporadic magnetic storms (of 2–20% and lasting 2–3 days) are known as Forbush decreases.

Details are in the caption following the image
(a) Variation of ionization rate (drawn for solar maximum and solar minimum) and typical ion concentration variations with height [from Gringel et al., 1986]. (b) Surface measurements of neutrons at climax (latitude 39.37°N, left-hand axis) and Huancayo (latitude 12.03°S, right-hand axis), showing the effect of geomagnetic (spatial) and solar (temporal) modulation. (The Huancayo data have been corrected for geomagnetic drift.)
[21] The vertical distribution of ionization is determined by the relative contribution of cosmic ray ionization and radioactive gases transported from the continental surface, such as radon. Immediately adjacent to the land surface, gamma radiation also contributes. The rate of ionization by atmospheric radioactivity is determined by vertical eddy transport of the gases away from the surface. For material with a radioactive decay constant λ the ion production rate at a height z is approximately [Junge, 1963]
equation image
where q10 is the ion production rate at z = 0 and K is the eddy diffusion coefficient. For radon 222, λ = 2.1 × 10−6 s−1. Above the boundary layer and over the oceans, atmospheric ionization is dominated by cosmic ray ionization. The cosmic ray ionization rate is ∼2 ion pairs cm−3 s−1 at the surface and increases by an order of magnitude at 5 km. The height dependence arises from the stopping power of the atmosphere, which increases with the density and the energy of the particles.

3.2. Ion Composition

[22] A convention in atmospheric electricity has been to categorize molecular ions according to their electrical mobility μ. Mobility is readily observed by ion counters and provides a method of ion spectrometry [Aplin and Harrison, 2001]. The ion mobility enters into equation (2) for the conduction current and therefore links the physical properties of the charge carriers (molecules and aerosol) with the behavior of the atmospheric electrical circuit. Ion mobility also controls the rate of uptake of air ions by the aerosol. The ion mobility μ is related to the ion diffusion coefficient D by Einstein's [1956] relation
equation image
and therefore scales with temperature T, pressure, and charge carrier mass in the same way. Composition differences between the positive and negative air ions cause negative ions to have a greater mobility than positive ions. The magnitude of the difference is typically 20% but varies with humidity in response to changes in the ion hydration as well as core ion composition, which may be different in different regions of the atmosphere. This asymmetry is central to the atmospheric aerosol electrification (section 4.2).

[23] In forming atmospheric small ions, the ionized molecules N2+ and O2+ undergo a series of chemical and ion exchange reactions that are every bit as complex as neutral chemical reaction sequences in the atmosphere [Beig and Brasseur, 1999, 2000]. The composition of the ions and ionic clusters is important because it influences their mobility and hence the electrical conductivity of the lower atmosphere and the rate of charge transfer to aerosols. The recombination of ions could play a role in ultrafine aerosol formation in the lower atmosphere [Mohnen, 1977; Yu and Turco, 2001]. The ion mobility depends on the ion mass and temperature and pressure through the diffusion coefficient D, while the nucleation of new aerosols is dependent also on the chemical properties of the ions involved.

[24] The reaction of the primary ions N2+ and O2+ with common atmospheric compounds results in a series of increasingly stable terminal ions, such as NO3 and HSO4, which accumulate polar molecules such as NH3 and H2O. Early measurements suggested that the majority of lower tropospheric negative ions would be composed of O2(H2O)n, CO3(H2O)n, or NO3(H2O)n clusters as well as HSO4 core ions and clusters such as H3O+(H2O)n, H+(H2O)n, NO+(H2O)n, NO2+(H2O)n, and NH4+(H2O)n for the positive ions. Improved measurement techniques have led to the discovery of much larger mass ions composed of heterocyclic nitrogen-containing organic molecules such as pyridine (C5H5N) and lutidine (C5H3(CH3)2N) as well as the neutral species [Eisele, 1988].

[25] There is evidence that the level of hydration varies with humidity; studies have shown that n = 3 is an especially stable configuration, and this ion is found to be common at humidities of ∼20%. There is also some experimental evidence of infrared absorption arising from such cluster ions [Carlon and Harden, 1980; Carlon, 1982], which may be apparent in atmospheric measurements [Aplin, 2003]. A hydronium ion is formed when a water molecule attaches itself to one of the hydrogen atoms in an existing oxonium ion. Hydration by three water molecules forms the stable H3O+ (H2O)3 ion. At extremely low humidities, ions such as NO+ and NO2+ occur, although in atmospheric air these ions will always become hydrated.

[26] Beig and Brasseur [2000] have developed a model of tropospheric ion composition based on the limited available rate coefficients and containing 20 positive ions and 29 negative ions. Their calculations suggest that in most atmospheric environments pyridinated clustered ions (e.g., H+(NH3)x(pyridine) · (H2O)y) are the dominant positive ions, with acetone dominating above 6 km. In contrast to previous studies their results suggest that ammonia and methyl cyanide are not the major parent species in tropospheric positive ions. However, over the oceans, where pyridine concentrations are low, ammoniated ions can still dominate. The most abundant negative ions calculated by the model have NO3 or HSO4 as the core, with HNO3 and H2SO4 as ligands. Aerosol in the lowest few kilometers of the atmosphere represented almost half of all negative ions. The net negative charge carried by the aerosol is due to the asymmetry in positive and negative ion mobilities (see section 4.2).

[27] The relative rates of ion-ion recombination and ion-neutral reaction are important in terms of determining the growth of the ion clusters in competition with their neutralization. We return to this issue in section 4.4 on aerosol nucleation.

3.3. Atmospheric Ion Concentrations and Polarities

3.3.1. Bipolar Ion Concentrations and the Effect of Aerosol

[28] Straightforward relationships exist between ion production and removal rates, depending on the amount of aerosol present. Neglecting the effect of sign, the time variation in ion number concentration n in aerosol-free air is given by
equation image
where q is the formation rate of ion pairs per unit volume and α is the recombination coefficient [Schonland, 1953]. The solution is
equation image
with a steady state concentration
equation image
Calculated recombination rates between charged molecular ions are typically 10−4 greater than the rate between neutral molecular clusters of comparable size [Landau and Lifshitz, 1980; Harrison, 2000]. Inserting typical atmospheric values of q ∼ 10 ion pairs cm−3 s−1 and α ∼ 1.6 × 10−6 cm3 s−1 [Schonland, 1953] gives n ∼ 2500 ion pairs cm−3. Typical values of small ion concentrations observed in clear mountain air are rather lower, with ∼500 ions cm−3 of each sign [Chalmers, 1967]. This difference is evidence that other loss processes, principally ion-aerosol attachment are significant in determining the ion concentrations in the lower troposphere. The small ion recombination timescale varies as 1/αn and is typically 250 s for n = 2500 ion pairs cm−3, increasing to 1250 s for 500 ion pairs cm−3, more representative of lower tropospheric ion concentrations. Steady state ion concentrations, which also determine when the aerosol charging becomes steady, therefore occur on timescales of the order of minutes.
[29] A similar expression to equation (8) can be written for the ion concentration in the presence of aerosol. Representing the aerosol as a monodisperse (i.e., all with the same radius) collection of particles with number concentration Z, the ion balance equation becomes
equation image
with solution
equation image
Here β is the ion-aerosol attachment coefficient, which varies with aerosol radius and charge [Gunn, 1954; Fuchs, 1963; Bricard, 1962; Hoppel and Frick, 1986]. Typical values of the attachment coefficient are given in Figure 4. Natural variations in aerosol concentrations, for example, associated with the production and dispersion of pollution in the planetary boundary layer, result in changes in ion concentrations. In the limiting case when ion-aerosol attachment dominates over ion-ion recombination as the principal mechanism of ion loss, the steady state ion concentration becomes
equation image
A related consequence is that the aerosol may then carry the large proportion of the space charge (final term in equation (1)). In this case, as aerosol is more strongly affected by inertial rather than electrical forces, the frequency spectrum of space charge variability is found to have properties typical of the inertial motions [Anderson, 1977].
Details are in the caption following the image
Ion-aerosol attachment coefficient β, calculated as a function of aerosol radius a and the number of elementary charges carried j, using the theory of Gunn [1954]. (Solid lines show attractive electrical forces, and the neutral value of β is found from the limiting case of the attachment coefficient expression, 4πkTμa/e). The Gunn theory considers Coulomb forces but neglects image forces, leading to repulsive forces for small radii particles. Experimental points are for the neutral aerosol case [from Pui et al., 1988].

3.3.2. Regions of Unipolar Space Charge Around Cloud and Aerosol Layers

[30] Although ionization leads to the formation of equal numbers of positive and negative ions in the atmosphere, local variations in the electric field can give rise to regions of net unipolar space charge. Such regions are potentially very important for aerosol microphysical processes, as detailed in section 4. Regions in which ions of one polarity dominate are created wherever there exists a gradient in the air conductivity in the presence of nonzero current density, which, through equations (2) and (4), leads to the presence of space charge. The conductivity is reduced in the atmosphere wherever ions are scavenged by aerosol (or cloud droplets), so regions of net unipolar space charge will often exist on the upper and lower surfaces of aerosol or cloud layers.

[31] Figure 5 shows a typical vertical profile of electric field through and around a layer of cloud. Above and below the cloud the electric field has its fair-weather value, but the field is enhanced within the cloud interior. The upper and lower parts of the cloud become charged by the vertical drift current of ions. Regions of negative and positive space charge as large as 5e cm−3 (1e cm−3 = 0.16 pC m−3) have been observed in nonthunderstorm shower clouds [Imyanitov and Chubarina, 1967]. Reiter [1986] has compiled an extensive data set of vertical electric field profiles in the Bavarian Alps, using a cable car–mounted sensor frequently passing through cloud layers. The density of space charge depends on many factors, including the stability of the aerosol or cloud layer (turbulent mixing can reduce the conductivity gradient), the ionization rate q, and the ionospheric potential VI and the columnar resistance Rc which both affect Jz. Since both the ionization rate and VI are modulated by the cosmic ray flux [Markson, 1981], the aerosol charge will also be modulated. Sections 4 and 5 describe how such regions of the atmosphere may be important for modulation of aerosol microphysical processes. A parameter which is modulated at the cloud boundaries is the ion asymmetry ratio x, which is the ratio of the positive and negative conductivities σ+, closely related to the number concentration ratio of the bipolar ion species. A nonunit value of x is necessary for the aerosol to carry a net charge (section 4.2). Table 2 summarizes measurements of the ion asymmetry ratio x, in the presence of cloud.

Details are in the caption following the image
Schematic of the charge structure around an isolated layer of cloud, exposed to the vertical conduction current density Jz. An atmospheric electric field sounding through such a cloud [Reiter, 1986] is also shown, from which the variations in cloud charge have been inferred. Bipolar ion concentrations exist in the atmosphere away from the cloud, but the upper and lower parts of the cloud become charged. Within the interior of the cloud, the ion concentration is reduced and the air conductivity falls. To maintain a constant Jz through the cloud, the electric field in the low conductivity region increases. Well away from the cloud layer, the electric field returns to its fair-weather value.
Table 1. Parameters of the Fair-Weather Atmospheric Electric Systema
Parameter/Property Value
Potential gradient, Ez 120 V m−1 (mean at sea level)
Surface charge flux density, Jz (3 ± 1) × 10−12 A m−2
Air conductivity, σ 1.33 × 10−14 Ω−1 m−1 (mean at sea level)
Ion properties
 Small ion concentration, n 500 cm−3
 Small ion mobility, μ 1.2 × 10−4 V−1 m−1 s−2
 Space charge at sea level, ρ 4 × 10−12 C m−3
Circuit properties
 Potential of high atmosphere 2.4 × 105 V
 Total resistance of the atmosphere, RT 230 Ω
 Total air-Earth current 1800 A
 Columnar resistance, Rcol 1.18 × 1017 Ω m2
Derived properties
 Surface charge density 10−9 C m−2
 Total surface charge on Earth −5 × 105 C
 Total charge in atmosphere 677 kC
Table 2. Values of Ion Asymmetry Factor x Reported by Reiter [1977] in Different Conditions
Cloud Weather Condition Altitude, km Ion Asymmetry Factor x
Mountain clouds fair weather 0.74 1.75–2.5
Nimbostratus homogeneous steady rain 1–1.8 2.5
1.8–3 10–20
steady snow fall 1.–1.6 3.2
1.6–3 4–5
Altostratus weak precipitation 1–1.4 4–30
1.4–3 2–10

3.3.3. Experimental Studies

[32] There are few experimental studies that have made electrical atmospheric soundings alongside meteorological observations in conditions other than thunderstorms, although some exist. Many of these observations came from pioneering studies using early experimental technologies, although high-resolution instruments are becoming available [Harrison, 2001]. Sagalyn and Faucher [1955] made a series of horizontal transects using aircraft to determine atmospheric particle concentrations between the surface and 5000 m during the summer of 1954. In almost all their transects, positive and negatively charged particles were detected, with peak concentrations of charged particles up to 2000 cm−3. Significant concentrations (∼100 cm−3) of charged particles were also found in the upper regions well above the freezing layer, although it was not possible to determine if the particles were multiply charged. Imyanitov and Chubarina [1967] made many in-cloud atmospheric electrical measurements during 1958–1959 above Leningrad, Tashkent, and Kiev, finding sharp potential gradient changes at cloud boundaries [see MacGorman and Rust, 1998].

[33] Venkiteshwaran [1958] and Hatakeyama et al. [1958] measured electric fields and air conductivities using modified radiosondes and determined electrical properties in nonelectrified cloud. They observed that the ion concentration increased with height and found that the measuring instrument itself was susceptible to triboelectrification as it passed through cirrus. Gringel [1978] reported the bipolar ion conductivity asymmetry ratio x using temporal averaging, but not simultaneous bipolar observations. Here x remained within ∼20% of 1.0, but variations were found.


[34] The vast majority of aerosol electrification studies have sought to understand the effects of aerosols on the electrical environment of Earth's atmosphere [Krider and Roble, 1986]. Relatively few studies have examined the effects of aerosol electrification on the aerosol properties themselves. In this section we address the issue of whether charge-mediated processes in noncloudy air can significantly affect aerosol properties. By significant we mean to the extent that the altered aerosol properties could affect the properties of cloud or the radiative properties of the aerosols themselves.

[35] Figure 6 shows a schematic of the processes controlling aerosol evolution in the atmosphere from nucleation of condensable vapors to form condensation nuclei (CN) and the growth of the CN to cloud condensation nucleus (CCN) sizes, typically >80 nm. The processes that are influenced by charges carried by the particles are indicated. Processes occurring in or near clouds are discussed separately in section 5. CCN number concentrations directly determine the concentration of cloud droplets and hence important cloud properties such as reflectivity (albedo) and precipitation efficiency [Hobbs, 1993]. The study of processes controlling CCN concentrations is therefore a major focus of current atmospheric aerosol research.

Details are in the caption following the image
Physical effects on cloud microphysics arising from atmospheric cluster ions. Molecular ions are formed by cosmic rays or point discharge in regions of enhanced electric fields and are transported within the global electric circuit. The presence of molecular ions leads to the charging of atmospheric aerosols and the formation, in suitable conditions, of ultrafine aerosols. Electrification of aerosol modifies the aerosol coagulation rates and the rate of aerosol collection by cloud water droplets. Electrical changes to aerosol coagulation and collection have been suggested to influence water and ice clouds, respectively.

[36] We begin by examining the possible magnitude of charge-mediated effects on these processes under atmospheric conditions. We then assess the potential contribution of such processes to observable changes in the properties of the atmospheric aerosol and clouds based on the limited available modeling studies.

4.1. Aerosol Nucleation on Ions

[37] Most descriptions of ion behavior in the atmosphere include a production rate from ionization and two loss rates, one due to ion-ion recombination and another due to attachment of the ions to existing aerosol particles (see section 4.2). However, the process by which air ions act as nucleation sites for the formation of new particles has not, until recently, been considered in aerosol models. The neglect of this process may be justified on the grounds that it is not important for the calculation of ion mobility and hence electric field strength, which has been a principal focus of ion-aerosol electrical studies. However, production of new particles in the atmosphere may be of importance to the microphysical properties of aerosol populations, insofar as they affect clouds.

[38] Early experimental studies suggested that artificial [Burke and Scott, 1973] and natural [Vohra et al., 1967] ionization could lead to the formation of condensation nuclei. For example, Vohra et al. [1984] showed that ionization rates typical of natural levels of radon gas caused the nucleation of condensation nuclei in natural filtered air. They suggested that aerosols might form from the clustering of vapor molecules around ions or from homogeneous nucleation of H2SO4 or HNO3 generated by radiolytic reactions involving sulphur and nitrogen oxides. Similar laboratory experiments were performed by Mäkelä [1992], who was able to isolate ion-induced nucleation in air containing H2SO4 and H2O. At low humidities with alpha particles, Yoon et al. [1992] found that the reduction of available OH led to a lower mean mobility of Po+ ions, suggesting that the nucleation was OH-limited. Kim et al. [1997] were able to define conditions where ion-induced nucleation dominated over homogeneous nucleation in their aerosol chamber experiments involving SO2 and water vapor. Aerosol nucleation was found to occur preferentially on negative ions at low SO2 concentrations. Vohra et al. [1984] observed aerosol nucleation at a rate equivalent to ∼10−5 new CN per ion pair, although this production rate is likely to depend very strongly on the composition of the air. Vohra et al. [1984] also suggested that the ion balance equation (equation (11)) would need to be modified to account for a change in the ambient aerosol number density if ions were directly involved in particle formation. These developments did not occur, and it was subsequently suggested again by Harrison [1998]. Available experimental data are summarized by Aplin [2000] and Harrison [2002c].

[39] Raes et al. [1986] proposed that ion-induced nucleation could produce new aerosol directly, above the sea surface where homogeneous nucleation of H2SO4 and H2O would be negligible (see Figure 7). The ion-induced nucleation rate is limited by the production rate of air ions by cosmic rays (2 cm−3 s−1 in the marine boundary layer away from sources of radon). New particle formation becomes significant at H2SO4 concentrations about a factor of 10 lower than are needed for significant homogeneous nucleation. More recent estimates of the homogeneous nucleation rate suggest that the H2SO4 concentration may even need to be a factor of 5 higher still. Only recently, as we discuss below, has the process of ion-induced nucleation been invoked to explain observed anomalous aerosol production.

Details are in the caption following the image
Ion-induced particle nucleation rates from sulphuric acid in marine air, as a function of sulphuric acid vapor concentration. The ion production rate (2 cm−3 s−1) is typical of that produced by cosmic ray ionization immediately above the ocean surface. At higher concentrations of sulphuric acid vapor, homogeneous nucleation of particles occurs: For ion-induced particle nucleation to dominate particle production, a lower sulphuric acid vapor is required. Reprinted from Raes et al. [1986] with permission from Elsevier.

[40] Observations of high particle concentrations have been made that cannot be explained by classical homogeneous nucleation theory, assuming binary nucleation of H2SO4 and H2O. Such observations exist in the clean marine boundary layer [Covert et al., 1996] and at clean continental sites [Weber et al., 1997]. There are two possible explanations for these observations. The first is that homogeneous nucleation could be accelerated if three species were involved in the formation of the initial molecular clusters (ternary homogeneous nucleation [Coffman and Hegg, 1995; Kulmala et al., 2000]). The second possibility is that nucleation occurs on ions or charged molecular clusters. Such a possibility has some observational support. Hõrrak et al. [1998] observed bursts of intermediate ions (with mobility less than common small ions) in urban air, which they attributed to ion-induced nucleation. These observations have been interpreted as consistent with ion growth by Yu and Turco [2000], using a detailed kinetic model. Harrison and Aplin [2001] made colocated surface measurements of ions and aerosol in atmospheric air and associated increases in condensation nuclei with increased rate of ionization events. Yu and Turco [2001] have examined a burst of ultrafine condensation nuclei observed in the marine boundary layer during the Pacific Exploratory Mission (PEM) Tropics A campaign [Clarke et al., 1998, 1999]. They showed that the observed high concentrations of ultrafine particles (∼104 cm−3) were in good agreement with their model calculations assuming ion-mediated nucleation of H2SO4-H2O aerosol over several hours. Eichkorn et al. [2002] measured very small charged aerosols at 9–10 km altitude that they attributed to ion-mediated nucleation, likely to have been initiated by cosmic rays. Further results of the Yu and Turco [2001] ion-aerosol model and the links to cosmic ray variations are described in section 4.4.

[41] Corona ionization has also been reported to produce ultrafine particles [Allen et al., 1981]. Using a point electrode operated at 4 kV, Nolan [1958] found uncharged nuclei were produced, with their radii determined 10 s after formation to be of the order of 1 nm. Corona yields detectable (>2 nm) particles, with number concentrations proportional to the concentration of trace (parts per million) aromatic organic vapors such as benzene. The effect is inhibited by the presence of alcohols [Ichitsubo et al., 1996]. Significant discharge currents in the atmosphere from electrified points, such as tree tips, are commonly observed [Schonland, 1953].

4.2. Natural Aerosol Charge Distributions

[42] In addition to direct nucleation of new aerosols on ions, air ions are also scavenged by the ambient aerosol particles, leading to removal of ions from the air and electrification of the particles. We now describe the factors that control the magnitude and polarity of the aerosol charge.

[43] The charges carried by an aerosol population in steady state have been determined theoretically by considering the kinetics of charge acquisition [Mead, 1978; Adachi, 1985; Clement and Harrison, 1992]. An extensive theory for radioactive aerosols has been developed [Clement and Harrison, 2000] and verified [Gensdarmes et al., 2001], but the results are also applicable to nonradioactive aerosols. The attachment of air ions to aerosols occurs through a process of molecular diffusion and migration under an electrostatic force induced by the charges already present on the particle. The collision rate of ions with a single particle depends on the number concentration of air ions of each sign (n±) and is normally expressed in terms of the attachment coefficient β (section 3.3.1) as
equation image
where N± is the number of ions of either sign carried by the particle. The attachment coefficient depends on the magnitude j and sign of the charge already carried by the particle. This uptake rate can be compared with the rate of diffusion of neutral molecules to a neutral spherical particle, which is given by
equation image
where n is the molecular number concentration, a is the particle radius, and N is the number of molecules absorbed by the particle. Expressions for the attachment coefficient have been derived in each of three Knudsen regimes: free molecular [Natanson, 1960; Keefe et al., 1968; Gentry and Brock, 1967], transitional [Bricard, 1962; Fuchs, 1963; Marlow and Brock, 1975], and continuum limit [Bricard, 1949; Gunn, 1954].
[44] The electric force between an ion and a charged, conducting aerosol particle is primarily due to the Coulomb force at large separations, which is sign-dependent, but at small separations the electric image force dominates (see also section 5.2). By Coulomb force we mean the force arising from the total charge on the two particles: The image force is the electric force arising between induced image charges, which is generally a shorter-range force. The sophistication with which the total electrical force is considered varies. Gunn [1954] neglected the image effects, using only the Coulomb force, which leads to an attachment coefficient between an ion of sign ±1 and a spherical particle carrying j charges,
equation image
where λ = e2/8πε0akT and μ+ and μ are the mobilities of positive and negative ions, respectively. The exponential terms in equation (16) are approximately 5.6 × 10−8/a for a singly charged particle. For particles with radii much greater than 0.056 μm, the Gunn attachment coefficients tend to 4πDa so that the uptake rate of ions onto weakly charged large particles approaches the uptake rate of neutral molecules by a neutral particle. Equations (13)(15) imply that a singly charged particle of radius ∼40 nm can capture ions (with opposite charge to that of the particle) at double the rate at which it can capture neutral molecules. On the other hand, a 1 μm particle would need to carry 20e in order to make the uptake rate of ions twice as fast as that of neutral molecules. Ultrafine aerosols of nanometer radii are sensitive to image effects [Marlow and Brock, 1975], as collections of a few hundred molecules are highly polarizable. More sophisticated expressions for β, accounting for image effects, were also derived by Fuchs [1963]. (Calculations of attachment coefficients computed from the Gunn [1954] expressions are compared with experimental data from Pui et al. [1988] in Figure 4.)
[45] The distribution of charges carried by an aerosol population has been derived from probabilistic master equations coupling the charge on one particle to the particles with one more (or one less) charge and thence to the charge state of all the others [Mead, 1978; Boisdron and Brock, 1970]. For a solitary spherical particle the rate of change of number concentration of particles carrying j charges Nj is
equation image
A similar equation can be written for negatively charged particles carrying a charge −j. A set of such equations, solved numerically for all values of j, completely describes a monodisperse aerosol's state of charge as a function of time. Using the attachment coefficients of Gunn [1954], the steady state charge distribution can be shown to be [Clement and Harrison, 1991],
equation image
where μ± are the positive and negative small ion mobilities, n± are their number concentrations, T is the temperature, e is the modulus of the electronic charge, and k is Boltzmann's constant. The mean aerosol charge J [Gunn, 1955] is
equation image
where x (the ratio of the polar ionic conductivities) is the ion asymmetry factor. The ion asymmetry factor becomes unity if the ion concentrations and mobilities are equal, when the modified Boltzmann distribution of charges reduces to the simple Boltzmann distribution [Keefe et al., 1959].

[46] There are few experiments from which the modified Boltzmann distribution can be experimentally validated, notably Hussin et al. [1983] for nanometre particles and Gunn and Woessner [1956] for micron-sized water droplets. Gunn and Woessner [1956] measured the droplet charge distribution with a ratio of positive to negative conductivity of x = 0.82; see Figure 8. The charge distribution described by equation (18) is valid if the total electric force between the ion and the particle is entirely that from a Coulomb potential, neglecting the image force. Breakdown of the Boltzmann charge distribution at small radii has long been recognized. It is clear, however, that the formulation of attachment coefficient used determines the final distribution. The exponential form only arises because of the Coulomb potential in the simplified attachment coefficient [e.g., Gunn, 1954], without image force effects. E. Schrödinger [Pollak and Metnieks, 1963] suspected the Boltzmann form broke down because of nonspherical particles at small radii, but the increasing significance of the image force is the more likely explanation.

Details are in the caption following the image
Comparison of theoretical prediction of aerosol charge distributions [Clement and Harrison, 1991] and experimental values for water droplets [Gunn and Woessner, 1956]. The number concentration of droplets carrying j elementary charges, as a fraction of the number concentration of neutral droplets, is plotted for a range of j for three different droplet radii. The experiment points are for droplets of radii 3.32 μm, charged in air in which the ratio of positive to negative conductivity from bipolar ions was 0.82. The excess of negative ions over positive ions causes the mean of the charge distribution to be negative.

[47] Figure 8 also emphasizes an important point when considering aerosol dynamics in the atmosphere, where there is no such thing as an absolutely neutral aerosol population. Although the mean aerosol charge may be zero (in the case of equal ion mobilities), there will always exist a distribution of aerosol charges, with some small fraction of each aerosol size class containing particles with substantial charges. It is also interesting to note that neither the steady state charge distribution nor the steady mean aerosol charge depend on the ionization rate of the surrounding air, which controls only the rate at which the steady state distribution is approached. Thus a modulation of the ionization rate due to, for example, cosmic rays does not, by itself, lead to a modulation of the steady state aerosol charge. There are, however, specific atmospheric conditions where space charge of one polarity can accumulate, notably in the air above and below clouds and with steep vertical gradients in aerosol loading, with a magnitude proportional to Jz. Under these conditions the charge state of aerosols can indeed be modulated by the ionization rate. It is in such regions of the atmosphere that variations of cosmic ray flux may directly impact aerosol processes (see sections 3.3.2 and 5).

[48] Aerosols can acquire charges that differ from those established at steady state with the ambient space charge density. Particular processes of relevance in the atmosphere are aerosols released by evaporation of cloud droplets and aerosols charged directly by combustion processes. Charge is lost from aerosol at a rate βn±per particle, where n± is the ambient ion number concentration. The ion-aerosol attachment coefficient β depends on the mobility of the ions and the charge carried by the particle (equation (16)), such that it loses charge more rapidly initially. In the troposphere, where n+n ∼ 500 cm−3, a 0.1μm particle carrying 20e will lose charge at a rate of 0.02 s−1, while the same particle carrying 2e will lose charge at a rate of 0.003 s−1. These discharge rates will be slower in air containing only unipolar ions and in clouds or high-humidity regions where ion concentrations and mobilities are reduced. Nevertheless, highly charged particles are unlikely to persist in the lower atmosphere for much longer than a few minutes, with the longest durations likely in fogs and clouds. The short timescale of aerosol charging processes implies that atmospheric aerosol electrification will usually be established by local processes rather than by transport of charged aerosol through the atmosphere.

4.3. Effect of Charge on Aerosol Coagulation

[49] Coagulation or agglomeration of particles in the atmosphere is important because it leads to a shift in the aerosol size distribution to larger particles and reduces the buildup of extremely high concentrations of ultrafine aerosols produced by gas-to-particle conversion. The coagulation is normally expressed in terms of a coagulation kernel (rate coefficient) K, such that the rate of formation of a particle of mass (m1 + m2) is equal to Kn1n2, where n is number concentration. One limit of coagulation, that between two charged molecules, was introduced in the context of ion-ion recombination in section 3.3.1. The charged molecular recombination coefficient α is a factor 104 greater than that for neutral molecules [Landau and Lifshitz, 1980; Harrison, 2000].

[50] Coagulation rates are influenced by charge because of the electrostatic force contribution to the relative motion of the particles. For ultrafine particles, there can be dramatic increases in the coagulation kernel if the particles are charged. In calculating particle formation from ions, Yu and Turco [2001] determined an enhancement in K of 103 for oppositely charged particles of radii 1 nm compared with neutral particles; see Figure 9a. Yu and Turco have shown that such enhancements in the coagulation rate lead to more rapid growth of charged ultrafine particles in the atmosphere, with important consequences for the aerosol size distribution (see section 4.4). The coagulation formulation of Yu and Turco [2000] covers particles from molecular sizes through to particles of several microns and therefore allows the seamless calculation of coagulation through to condensation.

Details are in the caption following the image
Electrical effects on aerosol coagulation. (a) Effect of electrification on coagulation as a function of radius [from Yu and Turco, 2001] for the different cases of (1) oppositely charged particles coagulating, (2) an ion coagulating with a charged particle, (3) a neutral molecule coagulating with a charged particle, and (4) a neutral particle coagulating with a charged particle. (b) Reduction in same-size particle coagulation rates, as a function of the particle charge distribution. The particle charge distribution is determined by the ion asymmetry ratio x (the ratio of positive to negative air conductivity) and the particle radii.
[51] Considering larger particles, with radii <100 nm, the rate of coagulation between particles with like charges is lower than that for neutral particles, while the rate is enhanced for particles with unlike charges. The enhancement f factor [Fuchs, 1964] for Brownian coagulation between two particles of radius a1 and a2 with unlike charge is
equation image
equation image
Experiments [Gillespie, 1953; Rosinski et al., 1962] show changes in coagulation when the particles carry bipolar charge distributions. The effect of a charge distribution leads to more complicated behavior [Clement et al., 1995; Shahub, 1989], as the tails of the charge distribution also contribute asymmetrically to the change in coagulation rate, favoring enhancement over reduction.

[52] In the atmosphere we are interested in the magnitude of f for a wide range of aerosol charge distributions between net neutral and net unipolar. There is a slight increase in the coagulation rate (f = 1.02 for 1 μm mondisperse particles) in the net neutral case, arising from the attraction between the tails of the charge distribution. Figure 9b shows the effect on the coagulation rate of varying the ion asymmetry parameter x, when the coagulation kernel is averaged across all particle charges in the monodisperse case. For ion asymmetries typical of the cloud-free atmosphere (x ∼ 1.2 or less), the enhancement is rather small. On the other hand, strong unipolar charge distributions (corresponding to |x| ≫ 1) can substantially reduce the coagulation rate (Figure 9b). In atmospheric regions of unipolar charging the effect will be to slow the production of larger aerosol particles from smaller particles. The upper and lower boundaries of clouds are regions of the atmosphere containing unipolar charges, as described in section 3.3.2. This results again in regions where |x| ≫ 1, with a similar effect expected on the coagulation.

[53] It is worth reemphasizing at this point that we are not particularly interested in establishing whether aerosol charge can influence coagulation rates but whether the coagulation rate is likely to change in response to a varying atmospheric ionization rate q. Because the steady state aerosol charge distribution is not affected by changes in q throughout most of the atmosphere (section 4.2), the coagulation rates in the clear homogeneous atmosphere will also not vary with q. However, as mentioned in section 2, there are specific atmospheric regions where the aerosol charge state can indeed be modulated by variations in q (section 5).

4.4. Modulation of Cloud Condensation Nucleus Concentrations by Cosmic Rays

[54] A description of the processes controlling CCN concentrations in the atmosphere is central to our understanding of how aerosols influence cloud properties and hence also climate change [Charlson and Heintzenberg, 1994; Hobbs, 1993]. Microphysical or “dynamic” models that simulate the chemical and physical properties of atmospheric aerosols almost universally include only neutral processes. Such numerical models describe the formation of new aerosols because of various nucleation mechanisms, coagulation, chemical reactions, growth by condensation of water and soluble gases, activation of aerosols into cloud droplets, etc. Many of these processes, as described in sections 4.2 and 4.3, may be affected by the presence of ions. Until recently, there has been no attempt to incorporate ion-aerosol processes in aerosol microphysical models of the background atmosphere.

[55] Mohnen [1990] suggested a scenario in which modulation of the ionization rate (and hence ion density) in the upper troposphere would lead to changes in the production rate of aerosol embryos that could grow into CCN. However, Mohnen [1990] also recognized that higher ionization rates would not necessarily lead to higher CCN concentrations, particularly if condensable vapor concentrations were limiting. The first detailed ion-aerosol microphysical modeling study was motivated by an interest in aerosol processes in aircraft exhaust [Yu et al., 1998]. It was found that the principal factor controlling the population of ultrafine plume particles was the number of charged ionic species (or chemi-ions) emitted by the aircraft exhaust. In subsequent studies of the background atmosphere [Yu and Turco, 2000] it was found that charged molecular clusters can grow significantly faster than neutral clusters. This mechanism was used in a box model by Yu and Turco [2000] to explain the diurnal variation in the ion mobility spectrum reported by Hõrrak et al. [1998], in which intermediate-size ions (charged aerosol) were found after the presence of small ions had been observed hours earlier.

[56] The box model of Yu and Turco [2001] simulates the evolution of ion clusters through to submicron particles, including the processes of condensational growth and coagulation among neutral and charged particles. Particular care was taken in describing the rate of coagulation, using coagulation kernels appropriate for charged and neutral particles from molecular cluster–scale to micron sizes. The enhanced growth of ion clusters by uptake of condensable vapors was parameterized in terms of an increased accommodation coefficient, recognizing that the thermodynamic data needed for calculating vapor condensation are not available. The simulations were restricted to uptake of H2SO4 and H2O vapors, with the H2SO4 concentration prescribed to vary over a diurnal cycle, with a peak at midday. The model results emphasize that charged clusters have a growth advantage over neutral clusters and can therefore grow more rapidly into nanometer-sized particles with sizes large enough to become stable particles. The ultrafine aerosol concentration is controlled by a competition between new particle formation and scavenging by the larger preexisting background aerosol.

[57] Figure 10 shows the time evolution of the aerosol size distribution in the Yu and Turco [2001] model, including the contribution from charged particles and neutrals. After 30 min of simulation some of the nucleated particles have reached stable nanometer sizes, most of these being derived from charged clusters. In addition, the mode of neutral particles that appears at 30 min results from the neutralization of charged nanoparticles rather than from neutral clusters, which mostly do not reach sizes large enough for stable growth. After 4 hours the new particles have grown as large as 20 nm. Most of these particles are those derived from original charged clusters rather than from neutral particles that acquired charge from ion uptake.

Details are in the caption following the image
Time evolution of aerosol size distributions (expressed as particle concentrations per logarithmic step in diameter) for (a) charged, (b) neutral particles, and (c) total number of particles [from Yu and Turco, 2001].

[58] Particularly important is the growth of ion-induced particles to CCN sizes of typically >80 nm. The simulations of Yu and Turco [2001] were restricted to less than 1 day in duration, which is too short to enable new particles to grow to CCN sizes. It was recognized by Yu and Turco that any realistic simulation over several days would need to include a more complete representation of aerosol processes, including competitive sources of CN and CCN. Nevertheless, their calculations suggest that ion-mediated nucleation can contribute to the CN population and that a proportion of these are likely to be able to grow to CCN sizes, given the right conditions. This fraction will depend on many factors such as competitive sources of CCN, vapor supply, and other sources of new aerosol, and each of these will depend on location, altitude, and season. Interestingly, the simulations suggest that ion-mediated CN formation is likely to be most important in the lower atmosphere, where the supply of condensable vapors such as H2SO4 is not limited.

[59] It is also important to establish the effect on the aerosol number concentration of variations in the ionization rate. The simulations reveal that a 25% increase in the ionization rate leads to a 7–9% increase in concentrations of 3 and 10 nm particles 8 hours after nucleation. In general, Yu and Turco [2001] estimate that a 25% increase in ionization rate could lead to at least a 4% increase in the stable CN number concentration. Further modeling studies are needed in which the growth of these additional CN is tracked through to CCN sizes. Further simulations under other atmospheric conditions, such as in the vicinity of clouds where the ion asymmetry is large, are also needed to understand whether a unipolar charge environment influences the aerosol processes. Yu [2002] indicates that there may be an altitude effect on cosmic ray–induced aerosol particle formation, with greater particle formation in the lower atmosphere and less in the upper troposphere.


[60] In this section we consider how the perturbed electrical environment around nonthunderstorm clouds (section 3.3.3) may impact ion-aerosol and ion-aerosol-cloud processes. Studies of such processes have focused on changes in the formation rate of ice in supercooled clouds and on charge-enhanced scavenging processes leading to modifications to the aerosol physical properties in the vicinity of clouds. Ice formation in nonthunderstorm clouds has several important effects, including the initiation of rainfall and changes in the cloud reflectivity [MacGorman and Rust, 1998], while changes in the aerosol properties near clouds could affect the CCN populations.

5.1. Charged Cloud Condensation Nuclei

[61] The most obvious question is whether charges carried by particles can directly affect their ability to act as CCN. The vapor pressure over a plane surface of water has a maximum value at any given temperature, the saturation vapor pressure. The saturation ratio S (the ratio by which saturation is exceeded) determines whether a droplet of a given size will grow or shrink and is therefore central to the cloud physics of establishing which particles within the atmospheric aerosol are able to act as CCN. If a droplet of radius a and dielectric constant εr condenses on a charged particle of radius a0 carrying a charge q in an environment of saturation ratio S, the change in free energy of the system is [Mason, 1971]
equation image
where NA is Avogadro's number and Mr, ρL, σLV are the relative molecular mass, density, and surface tension of water, respectively. The saturation ratio for a particle of radius a is found from the turning point of equation (22) by
equation image
where A is a “dissolved salt” term to account for solution effects and B is a curvature term. Taken together, it can be seen from these equations that the saturation ratio required for condensation varies inversely with particle size. For condensation to occur on a small particle, such as an ion, for example, in a Wilson cloud chamber, a very large saturation ratio is required, ∼4. A polarity effect is observed in such circumstances, with negative condensation centers being more active than positive ones [Rusanov and Kuni, 1984].

[62] Such a high level of supersaturation (400%) certainly does not occur in the atmosphere, preventing direct condensation on ions. In addition, the electrostatic terms in equation (22) become negligible for large particles and do not appear in equation (23). At a more typical tropospheric supersaturation of 0.6% (S = 1.006) a droplet in equilibrium with water vapor can grow if it has a radius greater than ∼0.13 μm. However, the presence of charge on such particles will have a negligible effect on their growth.

5.2. Scavenging

[63] Aerosol populations passing though a cloud have their particle size distribution modified by the loss of particles to cloud droplets. The removal of aerosol particles by cloud droplets is known as scavenging. The effect of multiple cloud encounters by tropospheric air parcels is thought to be a key process in defining the number of CCN in the lower atmosphere [Hobbs, 1993].

[64] Scavenging is influenced by inertial and electrical forces, and both the Coulomb and electrical image forces combine to give the total electrical force. The image force is always attractive when a charged aerosol particle is close to the droplet. Relatively complicated calculations are required to determine the collection efficiency of the particle by droplets, some of which have been simplified by only considering the Coulomb force [Grover et al., 1977]. Recent calculations include the electrical image force [Tinsley et al., 2000; Tripathi and Harrison, 2002], which is an important extension as the collection of the particle is independent of the relative signs of the droplet and particle charges. Figure 11a shows the partitioning of the water drop's charge with a charged particle nearby. The drop and aerosol particle carry net charges Qd and Qa, respectively, with a distance s between the aerosol and drop centers. It can be shown that the particle charge causes the charge on the conducting sphere to redistribute itself, equivalent to two point charges [Jackson, 1975], providing that the charge on the particle can be approximated by a point charge. Charge conservation requires that, for the two charges Qd = I + D, where I is the image charge and D is the nonimage charge. In magnitude, I = −(A/s)Qa, where s is the separation distance between the aerosol and drop centers. (The image charge is located at a distance c from the center of the drop.) The net electrical force acting between the particles' centers is therefore
equation image
where a positive Fe is repulsive.
Details are in the caption following the image
(a) Schematic of the charged aerosol and the image charge I construction, within a waterdrop of radius A. The aerosol and falling drop carry charges Qd and Qa, respectively, with Qd = I + D, where D is the nonimage charge, considered at the center of the drop. (b) Collision efficiency E for aerosol collection by a droplet of radius 52 μm, plotted as a function of particle radius, for a particle of density 2 g cm−3. A range of different aerosol charges (in elementary charges) is shown. Reprinted from Tripathi and Harrison [2002] with permission from Elsevier.
[65] Collection efficiency calculations integrate the equation of motion of the particle around the droplet. For aerosol particles with radii a > 0.4 μm the collection efficiency E (A,a) for a drop of radius A and aerosol particle of radius a is obtained by integrating
equation image
where m and V are the mass and velocity of the aerosol particle, respectively [McGann and Jennings, 1991] and η is the kinematic viscosity of air. The denominator (1 + αNKn) is the Stokes-Cunningham slip correction factor, g* = gp−ρa)/ρp, where g is the acceleration due to gravity, ρp is the density of the aerosol particle, ρa is the density of the air, U is the velocity of the air around collector drop, and Fe is the total electric force between waterdrop and aerosol. This equation considers the combined influence of gravity, air drag, and electrical forces. The collection efficiency is computed from the trajectory of the aerosol particles moving past the waterdrop, using suitable flow fields generated outside the drop [LeClair et al., 1970]. Using the Fe formulation of equation (24), the collision efficiency as function of particle size is shown in Figure 11b for a water droplet of radius 52 μm. For relatively small particle charges, of <50e, the collection of particles of ∼2 μm is greatly increased over that for neutral particles. We conclude that there are particle sizes for which electrical interactions appreciably enhance their collection by water droplets. So far, the effect of this electrical enhancement has not been included in microphysical models of clouds, so the impact on cloud development remains unknown.

[66] In summary, there are competing electrical factors in the unipolar regions near to a cloud causing (1) a reduction in aerosol coagulation and (2) an increase or decrease in aerosol removal through scavenging. Tinsley et al. [2000, 2001] have suggested that enhanced aerosol scavenging in the cloud edge region could lead either to enhanced ice nucleation rates or changes in the local CCN population. We now discuss these processes and the possible implications in turn.

5.3. Ice Nucleation

[67] The nature of ice nuclei has received significant attention, but many questions remain unanswered [Mason, 1971; Pruppacher and Klett, 1997]. Only a small fraction of the atmospheric aerosol may act as ice nuclei, explained in terms of a chemical or shape effect, depending on the precise nucleation mechanism considered. Much of the early work on ice nucleation [Fletcher, 1966] suggests that typical ice nuclei concentrations are between 10−3 and 10−6 cm−3 below −20°C, rising by several orders of magnitude at temperatures between −20°C and −35°C. Although there are grossly inadequate ice nucleus concentrations to account for the ice observed, ice multiplication mechanisms (where splintering occurs following ice formation leading to additional ice nuclei) contribute to ice formation at higher temperatures (−3°C to −8°C) [Mossop, 1985]. The concentration of ice nuclei show diurnal variations [Rosinski et al., 1995], some of which have their origin as cloud condensation nuclei [Rosinski, 1995].

[68] Ice nucleation occurs either by deposition (vapor to solid) or freezing (liquid to solid) transitions. Included within these general categories are condensation freezing (condensation from the vapor of liquid in which freezing follows), immersion freezing (cooling of a droplet increasing the likelihood of a supercritical embryo's occurrence), and contact-freezing (freezing following contact between a supercooled droplet and an ice nucleus). Heterogeneous nucleation, in which an ice nucleus is required, is important for much of the lower troposphere ice formation. (At temperatures below −40°C, atmospheric supercooled water forms ice by homogeneous nucleation.)

[69] There is currently an incomplete knowledge of ice nucleation processes, as, away from thunderclouds, there are few atmospheric measurements of the electrical interactions of ions, aerosol, and cloud particles. The various electrically influenced microphysical processes that may be relevant for inducing freezing in supercooled tropospheric clouds are therefore likely to be very difficult to observe directly in real clouds.

5.4. Electrofreezing

[70] By electrofreezing we mean any electrical enhancement of the phase transition between supercooled water and ice, although Dawson and Cardell [1973] emphasize that the term is used inconsistently. It is therefore important to distinguish between the many different possible microphysical mechanisms. For freezing to occur, an ice-like alignment of water molecules from the liquid phase is necessary, which then propagates through the supercooled drop as freezing. There remains the possibility that this alignment could be supplied by direct electrical interactions, although it has been found that ionizing radiation itself does not directly cause freezing transitions [Seeley et al., 2001].

[71] In field-induced electrofreezing an electric field causes the freezing of a supercooled waterdrop. Such electric fields are usually large [Abbas and Latham, 1969; Braslavsky and Lipson, 1998], more typical of thunderstorm rather than fair-weather atmospheric electric fields. A series of experiments is described by Smith et al. [1971], in which the observed freezing effect was due to mechanical disruption of the drop by the large electric fields. The disruption was the direct cause of the freezing rather than a direct electrical interaction with the water molecules that compose the drop. Molecular dynamics simulations by Borzsák and Cummings [1997], however, indicate that shear motions alone will not induce crystallization but that they would if an electric field were also present. Charge-enhanced contact nucleation has also been described as electrofreezing, as Pruppacher [1973] describes experiments in which μm-diameter sulphur particles (which usually acted as poor ice nuclei, hardly inducing freezing at all above −20°C) were able to cause ice formation at −8°C, because of inferred negative electrification of the sulphur particles. Electrically enhanced nuclei scavenging or electroscavenging [Tinsley et al., 2000; Tripathi and Harrison, 2002] may also be a possible electrofreezing mechanism (see section 5.2) by increasing the capture rate of the rare aerosol particles that are suitable ice nuclei.

[72] Numerical calculations [Tinsley et al., 2000] of electric image charge effects for particle radii 0.1–1.0 μm colliding with droplets of 18.6–106 μm showed that the collision efficiency increased thirtyfold for aerosols carrying (relatively large) 50 charges. Tripathi and Harrison [2002] report similar calculations but showed that, for particles with radii >0.4 μm, inclusion of the inertial force was necessary, which was neglected by Tinsley et al. [2000, 2001]. For a drop of radius 26 μm, Tripathi and Harrison [2002] estimated that the increase in the aerosol scavenging rate associated with a change in particle charge of ∼10e was approximately comparable with a change in the ice nucleus concentration expected from a temperature decrease of 1°C

[73] We conclude that a direct field-induced ice nucleation effect is most unlikely in nonthunderstorm clouds and that electroscavenging is much more likely to be prevalent instead. Although electrofreezing could occur in any supercooled cloud, in “warmer” liquid water clouds (−5°C to −15°C) ice multiplication processes operate. In clouds colder than −35°C, homogeneous processes are more active. Electrofreezing might therefore be expected to be significant in clouds with temperatures warmer than −35°C and in some cases may provide the primary event from which ice multiplication follows. Only limited laboratory evidence exists to support the idea that electrification alters the direct (nonelectroscavenging) ice nucleation efficiency of aerosol.

5.5. Proposals for Direct Modulation of Cloud Properties by Cosmic Rays

[74] There are currently two principal proposals for a possible modulation of cloud properties by cosmic rays: The first concerns the change in ice nucleation efficiency of aerosol particles because of electrical changes, and the second concerns the formation of cloud condensation nuclei from ions. We have also shown in section 4.3 that the unipolar charge in the upper and lower parts of the cloud layer will also act to inhibit aerosol coagulation. The consequences of this effect for the background aerosol, or for aerosol-cloud interactions, also remain to be investigated.

5.5.1. Effects of Changes in Ice Nucleation Rates

[75] Motivated by observed correlations between sunspot number and various indices of cyclone intensity [Roberts and Olson, 1973], an electrical effect on cloud microphysics leading to modification of weather systems has been proposed in a series of papers by B. Tinsley and coworkers [Tinsley, 1996a, 1996b, 1991, 1997, 2000; Tinsley and Beard, 1997; Tinsley and Deen, 1991; Tinsley et al., 1994, 1989, 2000, 2001; Kirkland et al., 1996] . In summary, the mechanism depends on cosmic ray modulation of aerosol electrification causing modifications to the ice nucleation efficiency of the aerosol. The central hypothesis is that the release of latent heat accompanying droplet freezing might be modified by aerosol electrification, which, in turn, is modulated by variations in cosmic rays. There is therefore the potential for substantial energy amplification between cosmic rays and midlatitude storm systems. The proposed links between ionization and storms are summarized in Table 3, with comments added on each stage.

[76] Tinsley [2000] identified elevated space charge densities around clouds as likely regions in which enhanced aerosol electrification, and hence scavenging, could occur. As described in section 3.3.3, there is both observational evidence and theoretical support for such an enhancement of space charge, although the observed space charge densities are typically up to ∼100e cm−3 [Clark, 1958] and are therefore much lower than the 103–104 cm−3 estimated by Tinsley [2000]. The very high charge densities estimated by Tinsley are due to his assumption of a very thin region (1 m depth) above the cloud over which accumulation occurs, while turbulence is likely to increase this depth in most conditions. Nevertheless, Tinsley is correct in identifying the near-cloud environment as a region in which the charge carried by the aerosol can be modified by cosmic rays (section 3.3.2).

[77] Confirmation of the Tinsley hypothesis depends first on electrofreezing being experimentally established as an atmospheric process and, second, demonstration that the release of sufficient latent heat by electrofreezing can influence weather systems. The second point is amenable to existing numerical models, although the connection between microphysical cloud processes and cyclone intensification is complicated. The first point requires further discussion. As discussed in section 5.4, there is some experimental and theoretical evidence to suggest that electrofreezing could occur, for example, through electrically enhanced scavenging of charged ice nuclei by supercooled droplets. Tinsley [2000] suggests that evaporative residue aerosols are likely to carry large charges sufficient to cause enhanced contact nucleation upon reentrainment into the same or a different cloud. Sustaining significant aerosol charges on residue aerosol for appreciable times will be difficult, however, as highly charged particles attract ions, which act to rapidly neutralize the particle charge, especially so in clear air. However, in clouds, measurements and models indicate that ion concentrations and air conductivity may be considerably lower [Pruppacher and Klett, 1997]. Tinsley et al. [2001] argue that the ice nucleation rate from electroscavenging could, in some maritime stratocumulus, be comparable with the rate of deposition ice nucleation. Although there are many uncertainties associated with several stages of the Tinsley proposal linking cosmic rays to mid latitude storms, it is clear that the central microphysical mechanisms involved are indeed susceptible to electrical modification. The processes are, in principle, amenable to model simulation, laboratory experiments, and in situ measurements. Observations of precipitation changes may provide one source of such measurements, especially at high geomagnetic latitudes [Kniveton and Todd, 2001].

5.5.2. Effects of Changes in Ultrafine Particle Production

[78] Close correlations observed between solar cycle length and surface temperatures [Friis-Christensen and Lassen, 1991] have led to suggestions that mechanisms, other than the straightforward changes in solar irradiance, could act to amplify the climatic signal of solar change. Svensmark and Friis-Christensen [1997] presented satellite evidence that changes in cloudiness are positively correlated with the flux of galactic cosmic rays, as modulated inversely by the solar cycle. Svensmark and Friis-Christensen [1997] used the monthly mean (C2) data set from the International Satellite Cloud Climatology Project (ISCCP) to provide variations in cloudiness over the period 1983–1990. Using only data from different geostationary satellites, the analysis was restricted to oceans between ∼60°S and 60°N. This work was developed further by Marsh and Svensmark [2000], using the ISCCP (D2) data set July 1983 to September 1994, derived principally from satellite cloud observations made at infrared wavelengths. They found that the cosmic ray–cloud correlation arose in the low cloud (<3.2 km altitude, as identified by the D2 processing) and hypothesized a mechanism in which CCN were produced by cosmic ray ionization, modulating the radiative properties of the cloud droplets. Table 4 provides comments on aspects of the mechanism suggested.

Table 3. Comments on the Different Stages of Tinsley's Mechanism Linking Solar Modulation of Ionization With Midlatitude Cyclone Intensification
Aspect Assessment Comment
Cosmic rays lead to ionization of atmospheric air throughout the troposphere. well established Cosmic rays do lead to ionization, and above the boundary layer they are the principal source of tropospheric small ions. The cosmic ionization rate depends on latitude (increasing toward the poles) and increases with altitude.
Variations in the solar wind modulate the cosmic ray flux in antiphase with the sunspot number. not contentious Neutron fluxes, ionization rates, and air conductivity have been observed to vary inversely with the solar cycle.
Cosmic ray ionization variations lead to variations in aerosol and water droplet electrification. probable Atmospheric aerosol undergoes charge exchange through ion-aerosol collisions, and charged atmospheric aerosol particles are observed. Variations in the magnitude of aerosol charge on layer boundaries through which the conduction current Jz passes are expected from theory and have been observed.
Electrification of aerosol increases its effectiveness as ice nuclei. possible Some laboratory evidence exists to suggest that charging alters the ice nucleation efficiency of aerosol, but there is no atmospheric experimental evidence. Theory, however, shows electrically enhanced aerosol scavenging increases the acquisition of contact ice nuclei by supercooled droplets.
Solar-induced changes in ionization will influence freezing in all clouds. unlikely The variability in atmospheric clouds and aerosol is such that it cannot be predicted which clouds will be affected by electrofreezing. It seems likely that only a subset will be affected.
Solar influences on electrofreezing intensify cyclone development and modify storm tracks through the changes in cloud temperature and latent heat release. contentious Modification of cyclones is highly selective and depends on where and at what stage in the cyclone's development the latent heat is released. The magnitude, and even sign, of any effect must be regarded as very uncertain.
Table 4. Comments on Cloud Condensation Nucleus Production From Cosmic Ray Ionization, Linking Solar Changes With Clouds
Aspect Assessment Comment
Cosmic ray ionization is ubiquitous throughought the atmosphere and provides a source of ions, modulated by cosmic rays. well established See Table 3.
Ions provide a source of atmospheric condensation nuclei (CN). not contentious Laboratory and atmospheric measurements show ultrafine particles (CN) can be formed from ions. Variations in the trace condensable species control the efficiency of the conversion.
Ions provide a source of cloud condensation nuclei (CCN). probable Detailed microphysical models show that particles large enough to act as cloud condensation nuclei can grow from ion-induced ultrafine aerosol production on the timescale of hours. There will usually, however, be competing effects, so the fraction of new aerosol that ultimately grow to CCN sizes is uncertain but is likely to be small.
CCN changes modify cloud properties, cloud lifetimes, and precipitation. not contentious Reducing uncertainties in understanding the effects of aerosol changes on clouds (the aerosol indirect effect) is a major area of active climate research.
Solar modulation of global clouds, through production of CCN from cosmic ray ionization, has been observed. contentious A significant positive correlation between cosmic rays and cloud cover has been obtained from satellite cloud data over a limited period. Different suggestions have been made to explain spatial pattern of the correlation, some of which do not require CCN production from cosmic rays.

[79] There is uncertainty as to whether there is a long-term correlation between cosmic rays and clouds [Sun and Bradley, 2002] because, although there is a significant correlation between cosmic rays and total cloud over the Atlantic Ocean between 1983 and 1991, there is no correlation with North Atlantic visible cloud cover data between 1953 and 1995. Of course, it is also possible that the correlation may not be due to the hypothesized mechanism but due instead to the synchronous, or even coincidental, response of clouds and cosmic rays to another quantity. The mechanisms discussed in section 4.4 do, however, indicate that there are theoretical reasons to expect the formation of CN and CCN from cosmic ray ionization, which would lead, in turn, to changes in liquid water clouds, and, ultimately, climate [Carslaw et al., 2002]. Yu [2002] has shown that the greatest cosmic ray–induced aerosol production should occur in the lower troposphere, which is consistent with the observations of low-cloud variations strongly correlated with cosmic rays [Marsh and Svensmark, 2000].


[80] Observations over several years show that the albedo of other planets varies in phase with the solar cycle [Lockwood et al., 1986]. The first such reported observations by Lockwood and Thompson [1979] showed that the brightness of Neptune and Saturn's moon, Titan, anticorrelated with solar activity, as defined by sunspot number, between 1972 and 1978. This period included solar minimum, and they were able to detect an approximate 3% increase in the brightness of Neptune and an 8% increase for Titan, followed by a decrease by a similar magnitude. In contrast, they were not able to find changes in brightness that corresponded to shorter-term variations in solar activity, such as solar flares, although the magnitude of solar activity changes in such events is similar to that over a solar cycle.

[81] Lockwood and Thompson [1979] speculated that the variations in brightness might be caused by changes in the cosmic ray flux or the large changes in high-energy ultraviolet radiation, which can be several orders of magnitude at wavelengths <2 nm, both of which could cause changes in the chemistry of the planetary atmospheres [Strobel, 1975]. However, the short period of observation, together with the poor understanding of the complex dynamical and chemical processes on these planetary bodies hampered any further analysis of a possible link.

[82] There are two plausible explanations for the observed albedo variations. The first, the “dark haze” theory, proposes the formation of a dark hydrocarbon polymer smog, driven by solar ultraviolet radiation. The second theory, explored in detail by Moses et al. [1992], proposes that ionization by cosmic radiation drives ion-induced aerosol formation. Moses et al. [1992] developed a model of Neptune's atmosphere including the formation of methane ice aerosol formed around ionized CH5+ ions (see Capone et al. [1977] for a discussion of Neptune's chemistry). Their model includes ion-induced nucleation of new cluster ions followed by growth of the clusters by condensation of methane, but it ignores coagulation. By assuming that all condensation nuclei were generated by cosmic ray ionization, they showed that changes in aerosol loading were sufficient to account for the observed changes in planetary albedo.

[83] The observed variations of planetary albedo and the proposed explanation in terms of ion-induced nucleation are interesting. However, there are several important neglected processes that prevent a complete understanding. These include neglect of convective transport of methane and ions, the assumption that ionic clusters are the only available condensation nuclei, and neglect of the chemical effects of hydrocarbon polymers. There is also uncertainty regarding the rate-controlling mechanism in aerosol formation, either supply of condensable vapor through vertical mixing or the ionization rate. In short, the uncertainties in such a model of Neptune's atmosphere are rather similar to those present in a model of Earth's atmosphere.


[84] A complete understanding of the influence of Earth's electrical environment on aerosols and clouds begins with processes occurring at the scale of individual particles. We have described several microphysical processes that can be influenced by the presence of ions in the atmosphere. For example, aerosols are known to nucleate preferentially on atmospheric cluster ions. These aerosols have a growth advantage over neutral particles and can therefore contribute to increased aerosol number concentrations in the atmosphere. In addition, the electrical environment around clouds is such that high space charge densities can exist. Aerosol and cloud hydrometeor processes on the edges of clouds are therefore susceptible to modification by charge-mediated processes. Increased ice nucleation and increased scavenging rates of aerosol in the turbulent regions around clouds seem likely. More importantly, model calculations and simple physical arguments suggest that a modulation of the atmospheric ionization rate will lead to a modulation in the rates of these charge-mediated processes.

[85] Although several processes have been identified that are susceptible to modulation by a varying atmospheric ionization rate, our ability to estimate the magnitude of any effect is currently limited. This difficulty is related partly to our incomplete understanding of aerosol processes in the neutral atmosphere and, in particular, the difficulty of tracking the influence of one particular modification through a long chain of complex interacting processes. Indeed, some proposed connections, such as the response of global weather systems to a varying ionization rate, will be extremely difficult to establish physically, even if the basic microphysical processes at the scale of individual clouds were understood.

[86] In terms of future work some advances can be made immediately by continuing and expanding existing studies of ion-aerosol-cloud processes. Two particular avenues of exploration appear fruitful. The first of these concerns investigating the effect of ionization on aerosol formation in the background atmosphere. Detailed model simulations indicate that aerosols formed around ions can alter the number of available cloud condensation nuclei. New laboratory and field measurements and comparison with theoretical approaches are needed to confirm these predictions. In addition, an improved understanding of the effect of ion composition on aerosol nucleation and measurements of basic thermodynamic quantities related to ion clusters will be needed for input into the models. Establishing a connection between ionization, aerosols, and clouds will also require further clarification of the complex aerosol-cloud interactions. The second area of fruitful investigation is the aerosol-cloud interactions occurring on the edges of clouds. The unipolar aerosol charge established close to the cloud upper and lower boundaries is modulated by the atmospheric ionization rate and ionospheric potential, both of which vary with the cosmic ray flux. In contrast, away from the perturbed electrical environment around clouds the electrical state of the aerosol is, to first order, independent of the atmospheric ionization rate. Further theoretical and experimental studies are needed to understand whether modulation of the aerosol charge around clouds could lead to observable changes in cloud properties. Large areas of relatively stable lower tropospheric stratus cloud sheets offer an excellent natural laboratory for such investigations. New high spatial resolution atmospheric electrical measurements of both positive and negative ions combined with aerosol properties and micrometeorological parameters are needed in order to understand the potentially highly complex and variable aerosol electrical state around clouds.

[87] In summary, the proposed connections between variations in cosmic rays and Earth's clouds and weather systems offer atmospheric scientists an extremely challenging problem incorporating well-established geophysical disciplines as diverse as aerosol microphysics, atmospheric electricity and the heliosphere. Focused studies of the processes occurring at the level of individual clouds may offer some insight into the driving mechanisms behind the existing cloud-ionization observations, although we are currently far from establishing cause and effect.


[88] This work was supported in part by an award from the Leverhulme Trust in the UK. The neutron data in Figure 3b were obtained by the University of Chicago under National Science Foundation grant ATM-9912341.

[89] Peter Cargill is the Editor responsible for this paper. He thanks two anonymous technical reviewers.