Laboratory Cloud Simulation

Capabilities and Future Directions

Frank Stratmann1, Ottmar Möhler2, Raymond Shaw1'3, and Heike Wex1

1Leibniz Institute for Tropospheric Research, Leipzig, Germany 2Institute for Meteorology and Climate Research, Forschungszentrum Karlsruhe, Karlsruhe, Germany 3Michigan Technological University, Houghton, MI, U.S.A.

Abstract

Since the atmosphere offers only observation possibilities and does not permit the setting of initial and boundary conditions, laboratory studies are important tools to examine cloud processes under well-defined, repeatable conditions and expand our understanding. An overview is provided of the capabilities and limitations of laboratory studies on physical cloud processes. Aerosol particle hygroscopic growth and activation, droplet dynamic growth, ice nucleation, and droplet-turbulence interactions are addressed and laboratory devices suitable for investigating and simulating these cloud physical processes are presented. These devices range from portable instruments for measuring selected particle and cloud droplet properties, to setups allowing process studies under simulated cloud conditions, to experiments allowing a nearly full-scale cloud simulation. The issue of laboratory-generated particles and their importance in laboratory cloud simulation is discussed. Finally, suggestions are presented for possible future research topics and devices in the field of laboratory cloud simulation. Specific suggestions include the initiation and/or continuation of investigations regarding particle hygroscopic growth and activation, the accommodation coefficients of water vapor on liquid water and ice, aerosol effects on primary ice formation in clouds, aerosol-based parameterizations of cloud ice formation, secondary ice formation/ice multiplication, the production and characterization of particles suitable for cloud simulation experiments, and experiments that combine turbulence and microphysics. The latter topic is important, as it offers the only possible way of simulating and quantifying the possible interactions and feedbacks between the microphysical (activation, growth, freezing) and turbulent transport processes within clouds.

Introduction

Atmospheric clouds constitute a complex system of global importance. Clouds influence climate and are the source of precipitation. Atmospheric clouds occur sporadically in locations that are usually difficult to reach. Therefore, the investigation of atmospheric clouds (i.e., the attempt to understand cloud formation and the dynamic processes within clouds in situ) is an ambitious, expensive, and often impossible task. To complicate matters, each cloud is unique; that is, measurements in atmospheric clouds suffer from a lack of re-producibility in terms of their initial and boundary conditions. Together with the complexity of the cloud system itself, this explains why cloud formation and cloud dynamic processes have not yet been quantified and why they are still not well understood.

Experiments are thus needed to allow (a) the isolation, examination, and quantification of individual cloud processes and (b) the simulation of combinations of selected cloud processes. Such investigations must be performed under well-controlled and reproducible initial and boundary conditions. None of these requirements can be fulfilled in situ and thus laboratory studies must address the entire range: from selected single cloud-relevant processes to the simulation of clouds on a laboratory scale. To date, laboratory investigations are not, and most likely will never be, able to reproduce all real-world cloud characteristics. In other words, although single characteristics, such as temperature and pressure ranges, timescales, and lifetimes, can be reproduced in the laboratory for selected cloud types, laboratory experiments are incomplete for characteristics such as cloud size, cloud surface area to volume ratio, the absence of any hard surfaces, and the Reynolds number. Nonetheless, laboratory studies are mandatory if we are to increase our understanding of selected cloud processes as well as clouds in general.

In this chapter we address the laboratory investigation of cloud physical processes, excluding, however, cloud chemical processes from our considerations. First we introduce important processes such as hygroscopic growth and activation of aerosol particles, droplet dynamic growth, ice nucleation and droplet freezing, and droplet-turbulence interactions. Thereafter we describe a selection of up-to-date methods, instruments, and facilities for investigating cloud physical processes. In closing we offer suggestions and discuss important future research topics as well as the need for further ideas and facilities.

Cloud Processes Simulated in the Laboratory Hygroscopic Growth and Activation

The ability of atmospheric aerosol particles to take up water is the primary link to their influence on clouds. For water vapor concentrations below saturation

(i.e., relative humidity, RH < 100%), particles may grow hygroscopically, such that when a threshold supersaturation (RH > 100%) is reached, the particles activate to cloud droplets. Both processes—hygroscopic growth and activation— depend strongly on particle properties such as size, structure, and chemical composition (see Kreidenweis et al., this volume).

It is well accepted that hygroscopic growth and activation can be modeled by means of the Köhler equation (Köhler 1923). However, the correct determination of the parameters and coefficients in this equation (e.g., water activity and surface tension), and consequently the consistent theoretical description of both hygroscopic growth and activation, is still a topic of ongoing research. The key is that although hygroscopic growth of solution droplets below 95% RH is controlled by water activity (Raoult term), activation is strongly influenced by the Kelvin term and is both dependent on droplet size and surface tension. This implies that only combined hygroscopic growth and activation measurements are sufficient to characterize fully the particle behavior upon humidification.

Many inorganic and organic substances of atmospheric importance have been examined in the laboratory with respect to their hygroscopic growth and activation. Summarizing the results of these investigations, one may say that (a) in addition to particle size and composition, surface tension exerts a strong infl uence on the activation process (corresponding sensitivities are given in Wex et al. 2008); (b) the surface tension of an activating droplet might be significantly different from that of water if surface active substances are present, and (c) for most investigated substances, at activation, it seems appropriate to assume ideal behavior of the solution droplet and the surface tension is that of water.

We still lack, however, the means to describe consistently, quantitatively, and efficiently the hygroscopic growth and activation of atmospheric (i.e., internally mixed multicomponent) aerosol particles (internal mixtures of soluble, slightly soluble and insoluble organic and inorganic substances) based on a reasonably small number of physical and chemical particle properties.

Droplet Growth

After activation, cloud droplets grow as a result of further condensation of water molecules on the droplet surface. In the course of this process, water vapor is depleted from and heat is released to the droplets' surrounding. This influences the vapor concentration, the temperature, and consequently the water vapor saturation fields in the vicinity of growing droplets and therefore much of the cloud. Consequently, growth by condensation is one of the key processes that controls cloud properties such as water vapor supersaturation and cloud droplet number and size. One of the main parameters in describing condensational droplet growth is the accommodation coefficient of water vapor on liquid water. This accommodation coefficient has been controversially discussed for a long time. Literature suggests accommodation coefficients between 0.01

and 1.0 (Davidovits et al. 2006); however, we do not understand the reasons for the observed discrepancies.

Another important growth mechanism is droplet collision-coalescence. One driving force behind the collision-coalescence process is the balance of gravitational and fluid drag forces, which causes cloud droplets of different size to fall with different velocities. This differential sedimentation results in collisions between larger and smaller droplets and consequently growth of the larger droplets. However, differential sedimentation is not the only process that causes droplets to collide. Another important mechanism is turbulence-induced droplet motion and it is still not well quantified (discussed further below).

Ice Nucleation and Droplet Freezing

After activation of aerosol particles, the resulting cloud droplets can be further lifted up in the atmosphere, where they can be cooled to temperatures well below 0°C. The spontaneous freezing of pure water, however, is hindered by an energy barrier, mainly because of the surface tension between the new ice germ, formed by stochastic processes, and the surrounding supercooled liquid water phase. Only at temperatures below ca. -35°C are ice germs of sufficient size formed, thereby inducing the freezing of the whole cloud droplet. Homogeneous freezing rates of water droplets measured in two laboratories at simulated cloud conditions were consistent with formulations of classical nu-cleation theory (DeMott and Rogers 1990; Benz et al. 2005). Recent laboratory measurements of the freezing rates of single water microdroplets demonstrated that the homogeneous freezing process is volume dependent (Duft and Leisner 2004b). These results conflict with the hypothesis that homogeneous freezing is preferentially initiated at the droplet surface, at least for droplets larger than a few micrometers in diameter.

At temperatures between 0° and -35°C, the freezing of supercooled cloud droplets can be induced by solid aerosol particles with specific surface properties, termed ice nuclei (IN). This heterogeneous ice nucleation process is initiated by a particle immersed in the droplet (immersion mode) or in contact with the surface of a droplet (contact mode). If cloud condensation nuclei (CCN) activation happens at sufficiently low temperatures, the same particle can first act as a CCN and then subsequently as an IN (condensation mode). The direct deposition of water vapor to the surface of a solid particle is called depositionmode ice nucleation. Until now, little has been known about the abundance and nature of heterogeneous IN or the relative importance of the different modes of ice nucleation in mixed-phase clouds. Various ongoing research programs are currently trying to relate the heterogeneous ice nucleation efficiency to specific aerosol properties and to develop aerosol-based formulations for heterogeneous ice nucleation in cloud, weather forecast, and climate models.

At temperatures below -35°C, freezing rates of growing water droplets are sufficiently fast so that the clouds contain only ice particles and no liquid water droplets. These ice clouds (cirrus) are formed either by homogeneous freezing of supercooled water or solution droplets, or by heterogeneous ice nucleation processes (cf. Kärcher and Spichtinger, this volume). The homogeneous freezing threshold of micrometer-sized solution particles increases from a saturation ratio of about 1.4 at -40°C to ca. 1.7 at -80°C. Mineral dust particles are thought to act as very efficient IN at cold temperatures and ice saturation ratios below 1.2 (Möhler, Bunz et al. 2006). Cirrus clouds tend to contain fewer but larger ice crystals if predominantly formed by heterogeneous ice nucleation processes. The number and size of cirrus ice crystals have important consequences for the balance between shortwave cooling and longwave heating of cirrus clouds in the climate system. More experiments and model investigations are needed to assess and quantify the competition between heterogeneous and homogeneous ice formation in cirrus clouds.

Ice Particle Growth and Habits

The ice phase affects the cloud cooling and warming capabilities and, in many cases, initiates precipitation and influences the distribution and intensity of precipitation. The number of primary ice crystals in mixed-phase clouds formed by heterogeneous IN is typically much smaller than the number concentration of droplets. Ice crystals can, however, grow to larger sizes at the expense of the liquid droplets (Bergeron-Findeisen process). Furthermore, the ice phase can be enhanced by secondary processes (e.g., ice splintering, the so-called Hallett-Mossop effect, riming, and ice crystal aggregation) and cause rapid changes to the internal structures of mixed-phase clouds. The quantitative description of these changes requires a better understanding of the basic processes, such as the accommodation of water molecules on ice particles, the rate of ice splinter formation, the collision rates of ice particles with droplets, the aggregation rates of ice crystals, or the influence of turbulence on, for example, condensational growth, aggregation, and riming.

Only a few laboratory experiments have investigated the Hallett-Mossop effect of ice splinter formation upon collisions between supercooled cloud droplets and ice crystals (e.g., Saunders and Hosseini 2001). Most cloud model parameterizations of secondary ice processes in clouds are based on empiric considerations.

Measurements of the mass accommodation coeffi cient of water vapor on ice are relatively scarce and not entirely consistent with each other. The data tend to support the idea that the accommodation coeffi cient is a function of temperature, and possibly of supersaturation and crystal size. Recent measurements in a carefully controlled laboratory experiment, involving small crystals suspended in an electrodynamic balance, resulted in an accommodation coefficient of 0.006 ± 0.0015 at a temperature of -50°C (Magee et al. 2006). A very low mass accommodation coefficient would have important implications for cirrus cloud properties and upper tropospheric humidity. Thus, further experiments are needed.

The growth of pristine ice crystals with distinct habits (e.g., columns, plates, needles, or bullet rosettes, or more compact, polycrystalline shape) is controlled mainly by the temperature and supersaturation present during crystal growth. However, it may also depend on the formation mechanism of the ice crystal (see, e.g., Bailey and Hallett 2002). The relation between preferential crystal habit and growth conditions is well established, but little is known about the influence of the ice formation mode on crystal shape. There is also a lack of experimental data for quantifying the effect of crystal habit on crystal growth rates, collision rates, settling velocities, and radiative properties.

As discussed above, the dynamic change of ice number and mass concentration in clouds is still difficult to understand and quantify. Issues requiring further attention include the accommodation coefficient of water vapor on ice and the influence of turbulence or crystal habits on, for example, condensational growth, aggregation, and riming.

Generation and Characterization of Particles Used in Laboratory Cloud Simulations

The properties of the particles used in l aboratory cloud research are not the main focus of this chapter. However, since these properties may heavily influence the results obtained during the experiments performed, we provide some thoughts on this topic.

Simulation of aerosol-cloud interactions in the laboratory requires the following:

1. Particles must be generated with well-defined physical and chemical properties (sometimes in large amounts).

2. Particle properties (e.g., by condensation of different vapors) must be tailored to the specific needs of the performed experiment.

3. Particles must be characterized according to their properties of interest with sufficient accuracy.

Some important properties to include in this context are particle diameter, shape, internal structure, volume, mass, surface area, surface properties such as number of active sites, and the chemical particle composition. Here we do not discuss the production or design of particles for a specific experiment; instead we address briefly the characterization of some selected properties.

Characterizing particle properties of interest for a certain cloud experiment represents a true challenge and is sometimes even impossible to achieve. Let us consider the process of heterogeneous ice nucleation in a solution droplet with a solid mineral dust or soot core, to which we add a thin (e.g., 5 nm) organic coating. To characterize the droplet as being frozen, we need to know the volume of the mineral dust core, the amount of organic substance on the core particles, the surface properties of the core particles (e.g., number of active sites), the size of the droplet, and the location of the core within the droplet.

Determination of any of these properties is nontrivial and requires high-end instruments such as differential mobility analyzer (DMA), low pressure impac-tor, mass spectrometer, and optical particle/droplet spectrometers. The surface properties of the core particle, which is most likely crucial for understanding the ice nucleation process, are currently almost impossible to quantify.

Reliable generation and sufficient characterization of the particles used in laboratory cloud simulations are important issues, and efforts to further our capabilities are needed.

Droplet-Turbulence Interactions

As we attempt to understand and quantify clouds, it is necessary to consider the role of turbulence, including the specific interactions between particles and the turbulence as well as multiscale turbulent transport processes. Droplet growth by condensation and collision-coalescence or aggregation are fundamental Lagrangian processes and depend on the history of individual cloud droplets or ice crystals throughout the cloud. Because cloud particles have a high density compared to air, their trajectories are not the same as that of fluid elements. The thermodynamic fields that they sample and the angles and speeds at which they collide are influenced by the surrounding turbulence. On larger scales, turbulent transport processes cause mixing within a cloud as well as mixing between the cloud and its surroundings. Turbulent transport processes affect strongly the temperature, the water vapor concentration, and saturation distributions within and at the edges of clouds. Consequently, turbulent transport processes influence cloud microphysical processes (e.g., cloud droplet activation and growth) and therefore have effects on cloud microphysical properties (e.g., cloud droplet number, cloud droplet size distribution, and even the small-scale spatial distribution of cloud droplets). To make things even more complex, this is not necessarily a one-way process. Cloud microphysical processes such as condensation and evaporation may have an effect on local turbulence through latent heat effects, resulting in a feedback between cloud mi-crophysics and turbulence. Examples for droplet-turbulence interactions that have been investigated and simulated in laboratory studies are entrainment, inertial particle dynamics, and turbulence-influenced condensational growth and collision-coalescence.

Entrainment, as mentioned above, is one of the main interaction processes between the cloud and its surroundings. It describes the transport of "dry" air, gases, and aerosol particles from the cloud's environment into the cloud by turbulent mixing processes. The opposite process—the transport of cloudy air outside the cloud—is called detrainment. Entrainment processes are thought to be responsible for the broad droplet size distributions observed in real clouds, which is in contrast to results obtained from theoretical calculations. However, the essential physics of the entrainment process are not fully understood: questions remain, for example, about the competing roles of homogeneous and inhomogeneous mixing.

The influence of turbulence on droplet collision-coalescence is one of the key processes that needs to be understood in terms of the growth of droplets within clouds. In a quiescent flow, droplets would collide solely through this differential sedimentation mechanism. However, in a turbulent flow, significant velocity and acceleration components perpendicular to the vertical will enhance the collision probability and thus droplet growth.

These cloud droplet-turbulence interaction processes have been investigated primarily through numerical simulation and field measurements over the last ten years, but controlled laboratory experiments, where specific mechanisms can be isolated and verified, are still needed.

Methods for Simulating Cloud Processes in the Laboratory

Having introduced the main topics of laboratory cloud investigations, let us now turn to some of the important experimental approaches used in these studies. The list of devices discussed below is not meant to be inclusive, but rather is intended to illustrate major techniques currently in use or under development. The laboratory devices presented here range from portable instruments applicable in both the laboratory and the fi eld, to larger-scale chambers for investigating individual cloud processes, and finally to large cloud simulators. We note, however, that currently, and most likely also in the near future, no device is capable of simulating a "real" cloud with all its relevant processes and complexity. As the cloud physical processes described above are investigated with different devices at various scales, they will be classified into the following groups:

1. Devices for measuring selected particle and cloud droplet properties.

2. Devices allowing process studies under simulated cloud conditions.

3. Devices allowing a large-scale cloud simulation.

Devices for Measuring Selected Particle and Cloud Droplet Properties

Here we present instruments that can be used in both laboratory and field studies to determine particle and droplet properties. Specifically, we discuss the humidity tandem differential mobility analyzer (H-TDMA), different CCN and IN counters, and the electrodynamic balance (EDB).

Humidity Tandem Differential Mobility Analyzer

H-TDMA is a tool used widely to simulate and quantify size-segregated hygroscopic growth of airborne particles in the RH range up to 98% (McMurry and Stolzenburg 1989). All H-TDMAs feature the same principle of operation: from an existing dry particle size distribution, a well-defined size fraction is extracted by means of a first DMA. Subsequent to size selection, the aerosol particles are humidified, and the resulting changes in particle size are determined by means of a second DMA. Results are usually given in terms of growth factors (i.e., the ratio of particle/droplet sizes) prior to and after humidification. The main differences between different H-TDMA systems are (a) the humidification method and (b) the manner in which RH and temperature are controlled.

H-TDMAs are excellent tools for simulating and quantifying hygroscopic particle growth in the RH range up to 98%. They have been widely used in both laboratory and field studies. However, they are operated in the RH range where particle growth is controlled mainly by water activity (and not by the Kelvin effect).

Cloud Condensation Nucleus Counters

Particle activation to cloud droplets can be simulated and quantified by means of CCN counters. Activation is achieved by subjecting the aerosol particles to an environment with supersaturated water vapor. Different techniques are available for creating such environments. Adiabatic expansion is one such method, but it usually leads to supersaturations (order of percent to hundreds of percent) larger than those observed in real clouds. Another method involves wetted parallel plates with different temperatures (Radke and Hobbs 1969). Here, the heat and mass transfer processes between the two plates together with the nonlinearity of saturation vapor pressure cause the air-particle-water vapor mixture in the gap between the two plates to become supersaturated, with the supersaturation peaking approximately halfway between the surfaces. Recently, a continuous-flow streamwise thermal-gradient CCN counter has become commercially available (Roberts and Nenes 2005). Here, an aerosol beam is sent through the center of a cylindrical column with wetted walls, such that temperature increases continuously along the column. Supersaturation is achieved because water vapor diffuses faster than heat from the walls to the center of the column.

Measurements using CCN counters have been used to determine the concentration and fraction of activated atmospheric aerosol particles at a given supersaturation, independent of particle size (i.e., the number of aerosol particles activated was compared to the total number concentration). Recently, a DMA has been sometimes used upstream of the CCN counter to obtain information on the size-dependent activation of the examined particles. This allows the retrieval of pairs of values for critical supersaturation and critical diameter for activation.

CCN counters have been widely used in laboratory and field. They are very usefUl devices for measuring the activation of aerosol particles to droplets. However, they do so under instrument-specific conditions. Timescales and supersaturation profiles inside the instruments differ from those that prevail in the atmosphere, and thus a one-to-one translation from data measured with CCN counters to conditions in atmospheric clouds might not always be feasible.

Ice Nuclei Counters

IN counters are instruments that determine the number concentration of IN in a given particle population at a particular ice supersaturation. They can be viewed as devices to simulate a subset of freezing processes within a cloud. Examples of currently available IN counters are (a) the continuous flow diffusion chamber (CFDC) (Rogers 1988; Rogers et al. 2001), (b) the Zürich ice nucleation chamber (ZINC) (Stetzer et al. 2008), and (c) the fast ice nucleus chamber (FINCH) (Bundke et al. 2007). Both CFDC and ZINC are based on a similar principle but feature different geometries (cylindrical and parallel plate, respectively). In both devices, particles are subjected to an environment supersaturated with respect to ice by means of two "parallel" surfaces that are both coated with ice but feature different temperatures. As a result of heat and mass transfer between the two surfaces and the nonlinearity of the temperature dependence of saturation vapor pressure, the air-particle-water vapor mixture in the gap between the two surfaces is supersaturated with respect to ice, and the supersaturation peaks approximately half way between the surfaces. Depending on the temperature difference between the two surfaces, different supersaturations can be achieved. Since the surface temperatures can be varied, IN number and or ice nucleation probabilities can be determined as functions of ice supersaturation. The number of ice particles is determined by means of an optical particle counter at the outlet of the instrument.

FINCH operates according to a different principle. Here supersaturation is achieved by turbulently mixing warm humid air with cool dry air in a flow tube. Again, ice particles are detected by means of an optical particle counter.

All three instruments have been successfully applied for investigating both atmospheric and laboratory-generated aerosol particles. However, similar to CCN counters, IN counters measure IN numbers under instrument-specific conditions. Timescales and supersaturation profiles inside the instruments differ from those that exist in clouds. Furthermore, IN counters currently lack the possibility to determine ice particle sizes/masses and thus ice particle growth rates. Most of the instruments operate in deposition or condensation nucleation modes, and therefore immersion and contact modes are not measurable.

Electrodynamic Balance

Single aerosol particles and droplets can be levitated and stored in EDBs (for a review, see Davis 1997). Storage time is principally infinite and limited only for practical reasons. EDBs have been used in many fields of aerosol research, including heterogeneous chemistry, deliquescence and efflorescence of salt particles, and optical properties of particles typically larger than a few micrometer in diameter. More recently, EDB devices in temperature-controlled environments have also been used to investigate the freezing of microparti-cles such as sulfuric acid (Imre et al. 1997) or pure water droplets (Duft and Leisner 2004b).

Several configurations of EDBs have been suggested and used for various applications (Davis 1997). A more sophisticated version of an EDB consists of a central rotationally symmetric torus electrode, with the symmetry axis in vertical orientation and two end cap electrodes on the same symmetry axis. The dimension of an EDB for particle levitation is in the range of 5-10 cm. Optimal storage and levitation results of microdroplets at atmospheric pressure are obtained if the caps and torus are of hyperboloidal shape. An AC voltage with an amplitude on the order of 1 kV and a frequency of a few hundred Hz is applied between the torus electrode and the end caps; this focuses the charged particles in the trap to a small area centered around the symmetry axis. A superimposed DC voltage between the two end caps is adjusted to balance the gravitational force on the particle in the trap.

Mass changes, for example, of evaporating water droplets in the EDB are sensitively monitored with the balance voltage. For spherical droplets, the index of refraction can also be obtained from Mie-scattering features observed optically (e.g., Duft and Leisner 2004a). If the index of refraction is known, the same setup can be used to measure the droplet size accurately. Freezing of droplets is detected by depolarization-sensitive light-scattering techniques. Whereas droplet processes can be studied in EDBs over a wide temperature range, the EDB setup is limited in achieving and controlling high humidities or even supersaturations with respect to ice and water. In addition, possible effects of droplet charge and the applied electrical field should be kept in mind.

Devices Allowing Process Studies under Simulated Cloud Conditions

Let us now turn to those devices that allow the investigation of cloud physical processes at simulated cloud conditions. Again, our discussion is not intended to be complete but features examples for important state-of-the-art-cloud simulation techniques and devices. Specifically, we discuss (in alphabetical order), the aerosol interactions and dynamics in the atmosphere (AIDA) chamber, the Leipzig aerosol cloud interaction simulator (LACIS), the Meteorological Research Institute (MRI) chamber, and the University of Mainz wind tunnel.

Also, we address devices (wind tunnels and chambers) used to study cloud droplet-turbulence interactions.

Aerosol Interaction and Dynamics in the Atmosphere Chamber

Aerosol and cloud processes can be investigated in the AIDA facility of the Forschungszentrum Karlsruhe under a wide range of simulated atmospheric conditions, including temperature, pressure, humidity, as well as trace gas and aerosol constituents (Benz et al. 2005; Möhler, Bunz et al. 2006). Experiments conducted in the large AIDA chamber are complemented by experiments with single levitated microdroplets (see above) as well as the development and application of process models with detailed aerosol and cloud microphysical formulations (Möhler, Field et al. 2006).

At the core of the AIDA facility is a cylindrical aluminum vessel with a volume of 84 m3. The vessel can be evacuated to pressures below 1 hPa and is located in a thermally insulated container that can be cooled to temperatures as low as -90°C. The cooling system maintains homogeneous temperature conditions for hours to days, with temporal and spatial fluctuations throughout the container below ±0.3°C. Experiments begin at well-defined pressure, temperature, and RH conditions and with a well-specified mix of trace constituents. With respect to ice and/or liquid water, supersaturation is achieved by pumping expansion and corresponding adiabatic cooling of the vessel volume. Thereby, conditions similar to lee wave or convective clouds can be simulated with cooling rates between about 0.1 and 5 K min-1. During a typical cloud expansion simulation run, the pressure in the AIDA vessel is lowered from 1000 to 800 hPa within about 5 to 10 minutes.

A comprehensive set of commercial as well as specially designed and homemade instruments is available at the AIDA chamber for generating various aerosols as well as for measuring physical and chemical particle properties, water vapor and trace gas concentrations, optical particle properties, and ice cloud characteristics. This set includes standard aerosol instrumentation (e.g., condensation nuclei counters and electrical mobility, aerodynamic and optical sizing techniques, Fourier transform infrared, and tunable diode laser spectroscopy) and specially designed scattering, depolarization, and imaging techniques for the detection and characterization of ice particles.

One of the strengths of the AIDA facility is that cloud processes can be investigated under realistic conditions with internal variability and fluctuation of temperature and humidity. A full AIDA cloud simulation cycle includes preparation of aerosol, hygroscopic growth, CCN activation, and, if temperatures are low enough, ice nucleation. Effects of aerosol aging and cloud processing can be investigated in subsequent expansion cycles with the same aerosol. However, the humidity control in the chamber is limited to an accuracy of only a few percent as a result of temperature fluctuations and wall effects.

Leipzig Aerosol Cloud Interaction Simulator

LACIS (Stratmann et al. 2004) is a device to investigate complex phase transition processes, such as particle/droplet hygroscopic growth, activation, ice heterogeneous nucleation, and ice particle growth. LACIS has been designed to simulate accurately cloud processes that take place on short timescales (i.e., up to a minute). Its main focus is on particle size-resolved investigations; it features particle/droplet generation and detection devices capable of producing, detecting, and characterizing monodisperse particle and droplet size distributions. Thermodynamic parameters such as temperature, pressure, RH, critical supersaturation, composition, and concentration of particles/droplets and of chemical composition in the carrier gas can be varied in ranges similar to the lower troposphere.

LACIS consists of a 15 mm diameter laminar flow tube with temperature-controlled walls. The length of the flow tube can vary from 0.5 to 10 m. Residence times up to 60 s and temperatures down to -50°C are feasible. Using high-precision instruments, the thermodynamic conditions in the LACIS flow tube are controlled with extremely high accuracy and reproducibility (temperature about 0.05°C, RH about 0.5%, and critical supersaturation about 0.05%). Particle/droplet size distributions produced in LACIS are determined by means of specially designed optical particle counters mounted along the axis of LACIS. These optical particle counters are capable of counting, sizing, and distinguishing the phase state (e.g., liquid, frozen) of individual droplets inside LACIS.

LACIS may be operated under both sub- and supersaturated conditions with respect to water and ice. Thus it combines the capabilities of simulating hygroscopic growth, dynamic post-activation growth, droplet freezing, and subsequent ice particle growth into one instrument.

LACIS has been used mainly to investigate the hygroscopic growth and activation behavior of different types of aerosol particles. In one recent study, it was applied to connect successfully high RH hygroscopic growth and activation behavior for selected inorganic and organic substances (e.g., Ziese et al. 2008). Recently, initial investigations have been performed on the immersion and deposition freezing of monodisperse dust particles.

Meteorological Research Institute Cloud Chamber

A new dynamic cloud chamber for simulating and investigating IN and CCN processes has been recently developed at the Meteorological Research Institute (MRI) in Tsukuba, Japan (T. Tajiri, pers. comm.). The chamber has a similar design to the Colorado State University cloud chamber (DeMott and Rogers 1990). Adiabatic cloud expansion processes are simulated by synchronously controlling air pressure and wall temperature in the ranges 1013 to 30 hPa and

30 to -100°C, respectively. Adiabatic expansion can be simulated equivalent to updraft speeds of up to 30 m s-1.

The chamber consists of two stainless steel vessels, an outer pressure-controlled chamber, and an inner temperature-controlled chamber. The inner cylindrical vessel has a height of 1.8 m, a diameter of 1 m, and a volume of 1.4 m3. The walls of the inner vessel are temperature-controlled by a circulating coolant. The particle injection (air supply) port is located at the top of the chamber. A small fan stirs the injected air to achieve homogeneity inside the volume. The chamber is equipped with various instruments for aerosol characterization, CCN measurements, as well as droplet and ice crystal detection.

During cloud formation experiments, the simulated ascents follow dry adia-batic expansion until the air temperature corresponds to the lifting condensation level (LCL). Thereafter, moist adiabatic expansion is simulated. The initial condition of pressure, temperature, dewpoint temperature, and ascent rate must be known to calculate cooling/evacuation rate and LCL. Accurate temperature and pressure controls are made by a combination of feed-forward and three-term PID control methods. The actual temperature and pressure are typically held to within 0.5°C and 0.3 hPa of the command profile.

Among the advantages of the MRI chamber are the wide temperature range (as low as -100°C) and the fact that adiabatic cloud expansion cycles can be simulated with active control of both gas and dew or frost point temperature. The chamber will be used to study cloud formation processes such as CCN activation, ice nucleation, and aspects of artificial cloud seeding. Because of the smaller volume, only a low number of sampling instruments can be operated at the chamber, especially at low expansion rates.

University of Mainz Wind Tunnel

A vertical wind tunnel at the University of Mainz, Germany, is used to investigate and simulate cloud processes such as uptake of trace gases by raindrops, the influences of turbulence on the collisional growth of cloud droplets, and the impaction scavenging of aerosol particles (Pruppacher 1988; Vohl 1989). The wind tunnel allows free suspension of water drops or other hydrometeors at their terminal velocity in its experimental section. The flow inside the wind tunnel is created by two vacuum pumps in the upper horizontal part of the tunnel. The velocity of updraft in the experimental section is controlled by a variable flow sonic nozzle. In the lower horizontal part of the tunnel, there are particle filters and air conditioning units which make it possible to adjust the air humidity and temperature in the experimental section. Trace gases, aerosol particles, or cloud droplets can be introduced into the wind tunnel upstream of the experimental section. Downstream of the experimental section it is possible to take air samples to determine the air temperature and dew point or to measure trace gas concentrations.

Inside the experimental section, the suspended hydrometeors can be observed visually, and after defined residence times, the droplets can be collected and removed from the wind tunnel, fixed in sample bottles, and analyzed by means of ion chromatography to determine their composition.

Thus far, the collisional growth of cloud droplets, impaction scavenging of aerosol particles, and gas uptake by single water drops and/or ice crystals have been studied under laminar and/or turbulent conditions. Droplet immersion and contact freezing processes have also been investigated.

Other Wind Tunnels

We would like to mention two wind tunnels that can be used for measuring interactions between water droplets and turbulence: the wind tunnel at Cornell University Ithaca, New York, U.S.A. (Ayyalasomayajula et al. 2006; Saw et al. 2008, submitted) and the tunnel at the Max Planck Institute for Dynamics and Self Organization in Gottingen, Germany. A key aspect of both tunnels is the ability to reach large turbulence Reynolds numbers, which is necessary when considering turbulence in geophysical flows such as clouds. The Cornell wind tunnel has a cross section of 1 m x 0.9 m, and a length of 20 m, with turbulence driven by an active grid (triangular agitator wings attached to randomly rotating grid bars). It is capable of generating turbulence with a Reynolds number ranging from roughly 104 to 106. A broad droplet size distribution, with mean diameter approximately 20 p.m, is generated by an array of four spray nozzles. The Gottingen wind tunnel, which is currently being assembled in the newly constructed experimental hall and which is expected to be operational by mid-2008, is designed to achieve a Reynolds number of approximately 108, the highest Reynolds number ever to be achieved in a standard (i.e., non-superfluid) laboratory flow. The tunnel is 12 m long and the pipe is 1.8 m in diameter. The gas is pressurized to have a density 150 times greater than ambient air. Both tunnels are designed to investigate particle/droplet-turbulence interactions and allow Lagrangian measurements (i.e. tracking of particles/droplets by a high speed camera moving along the side of the tunnel at the mean flow speed).

Warsaw University Cloud Chamber

To investigate interactions between turbulence and cloud droplets at Warsaw University, a glass box of 1.0 m x 1.0 m x 1.8 m (Malinowski et al. 1998; Korczyk et al. 2006) was built. Droplets generated by a commercial ultrasonic humidifier are visualized in a vertical or horizontal cross section through the chamber interior by means of a laser beam formed into the shape of a narrow (~1.2 mm thickness) plane. The light scattered by the illuminated droplets at 90° is detected by photo and video or, more recently, by digital CCD cameras and used to image droplet patterns. Conditions within the chamber are not fully controlled, but are instead monitored. The cloud droplet size spectrum is measured as follows: Droplets are collected on a glass plate (covered with a thin film of paraffin oil) and subsequently imaged by microscope. Images are then processed to determine droplet size and number (i.e., the droplet size distribution). The liquid water content of the cloudy plume is measured with a cotton filter: the increase in mass of the filter after pumping a given volume of cloudy air through it gives the measure of liquid water content. Temperature and humidity profiles within the chamber are monitored using a set of thermocouples and capacitance sensors.

Experiments conducted with this chamber addressed the geometric patterns created by turbulent mixing of the cloud with its environment and the preferential concentrations of cloud droplets in weak turbulence. Currently, a particle imaging velocimetry (PIV) technique is used to retrieve motion of the cloud droplets investigating fine-scale details regarding the turbulent mixing of the cloud with its environment.

Michigan Technological University Cloud Chamber

For studying Lagrangian properties of cloud droplets in turbulence, a laboratory system has been developed at Michigan Technological University in Houghton, MI, U.S.A. (Fugal et al. 2007). The chamber is a cube with speaker-driven jets positioned at each vertex. The jets are randomly forced and interact in the center of the cube, so as to produce approximately homogeneous, isotropic turbulence. Water droplets of a predetermined size are allowed to settle into the chamber from above, after having achieved their terminal fall velocity inside a seeding cylinder. The water droplet positions within the central volume are determined by digitally reconstructing in-line holograms recorded by fast CMOS cameras. The holograms can be recorded at rates up to 6000 frames per second, allowing particles to be tracked in real time. Initial results show clearly the transition between gravitationally and turbulence-dominated Lagrangian velocity and acceleration statistics. For large cloud droplets and small drizzle droplets, velocities that are several times greater than the terminal velocity and accelerations up to ten times the gravitational acceleration are observed as the turbulence intensity is increased.

Devices Allowing a Nearly Full-Scale Cloud Simulation

The last group of devices that we address concerns simulations conducted close to real cloud scale. Two devices—one existing and one currently under discussion—will be presented, both of which feature former mine shafts as cloud simulators. Before discussing the details of such devices, let us consider the rationale behind such huge cloud simulators. Shafts allow the simulation of cloud processes, including the adiabatic expansion of rising air parcels, for reasonably realistic length scales and residence times with hydrometeors being at their respective terminal velocity. Sampling conditions are also relatively easy. Control of the thermodynamic parameters inside the shaft, however, is lower than in usual laboratory experiments. Clouds generated in vertical shafts are thought to permit investigation of central cloud research topics, such as the roles of aerosol particles, turbulence, electrical effects and temperature on cloud history and the formation of precipitation. They also allow experiments on the behavior of complex aerosol (e.g., biomass burning, combustion aerosol) in clouds, the simulation of ice particle interactions in supercooled cloud environments (thought to cause charge separation and therefore lightning), and the investigation of cloud chemical processes under conditions close to those in a real cloud.

Furthermore, vertical shaft clouds exhibit higher Reynolds numbers than can be achieved in laboratory experiments (but still not as high as in the open atmosphere), thus permitting better experimental data to address the question of how microscale turbulence (~ mm to cm length scales) may affect cloud processes such as condensation and collision-coalescence or aggregation. Other possible research topics may include cloud radiative transfer and remote sensing.

Two devices for the generation of shaft clouds are (a) the artificial cloud experimental system (ACES) (e.g., Yamagata et al. 1998, 2004), operated by Hokkaido University in Sapporo, Japan, and (b) the Cloud Physics Facility in South Dakota (Homestake DUSEL), suggested by J. Helsdon.1 According to our understanding, ACES suffers currently from a lack of financial support and the Cloud Physics Facility at DUSEL has been postponed.

The facility operated by Hokkaido University features an area of 2.5 m x 5 m and a vertical length of 430 m. Updraft velocities can be varied in the range between 0.5 and 2 m s-1, and temperatures range from 13.5 at the bottom to 10.5 at the top of the shaft.

The facility suggested by J. Helsdon would have an area of 3-5 m x 3-5 m, a vertical length of approximately 1 km, an active grid turbulence generator, and in-situ measurements of below-, mid-, and cloud-top properties.

One outstanding question is whether shaft clouds permit an accurate amount of adjustment and control of crucial parameters (e.g., temperature, saturation, flow velocity, turbulent intensities, and dissipation rates) to allow the envisioned gain in knowledge and understanding. Furthermore, funding to create and operate such huge devices may be an important constraint.

Future Research and Devices

In the future, we envision that laboratory cloud research will address processes that take place in warm, mixed-phase, and cirrus clouds. These include:

1 http://neutrino.lbl.gov/Homestake/FebWS/presentations/08_Helsdon%20DUSEL%20Home-stake%20L0I.pdf

• hygroscopic growth and activation,

• accommodation coefficients of water vapor on liquid water and ice,

• aerosol effects on primary ice formation in clouds,

• aerosol-based parameterizations of cloud ice formation,

• secondary ice formation or ice multiplication,

• generation and characterization of particles used in laboratory cloud simulations,

• specific cloud droplet-turbulence interactions, and

• combining turbulence and microphysics over multiple scales.

Hygroscopic Growth and Activation

There is still a need for understanding and quantifying the effects of slightly soluble substances, droplet solution non-ideality, and partitioning of surface active substances between particle bulk and particle surface, on high RH hygroscopic growth and activation. This holds for both particles that comprise selected single organic components and particles that consist of internal mixtures of soluble, slightly soluble, and insoluble inorganic and organic substances. Closure studies regarding particle hygroscopic growth and activation behavior and derivation of parameters for a consistent description of particle hygroscopic growth and activation are important topics here.

Possible influences of organic (surface-active) substances on the kinetics of hygroscopic growth and activation are another unresolved, and perhaps critical, issue requiring future attention. This is because application of the Köhler theory for modeling or parameterizing hygroscopic growth and activation implies that the droplet is in equilibrium with its surroundings (i.e., kinetic effects influencing droplet growth are neglected). In case this assumption is not justified, usage of the Köhler theory could lead to erroneous predictions in terms of particle/droplet size and critical supersaturation.

The effects of soluble gases (i.e., their uptake into the particle or droplet and effects on particle hygroscopic growth and activation) must be borne in mind. Influences of particle aging and (cloud) processing on hygroscopic growth and activation behavior are also of great interest.

Of special value in this context are measurements that are performed at RHs > 95% and supersaturations up to ca. 0.5%. In this context, instruments such as the H-TDMA, LACIS, EDBs, and CCN counters have been and will continue to be useful tools.

Accommodation Coefficients of Water Vapor on Liquid Water and Ice

In terms of the accommodation coefficient of water vapor on liquid water, new experimental ideas are needed. The accommodation coefficient may be a function of thermodynamic conditions (temperature, pressure, supersaturation), droplet size, or water vapor flux to the droplet surface. Thus, experiments under conditions resembling those of real clouds (including real cloud time-scales) are required.

Our knowledge is also insufficient regarding the accommodation coefficient of water molecules on ice particles, in particular under realistic atmospheric conditions. This is primarily the result of the challenging conditions under which such experiments must be performed. When equipped with suitable detectors to measure ice particle size or (even better) mass, devices such as LACIS, AIDA, CFDC, ZINC, and FINCH could be utilized to determine ice particle growth rates and derive accommodation coefficients. Such measurements, however, require the use of numerical models to evaluate and interpret the resulting data.

Aerosol Effects on Primary Ice Formation in Clouds

At present, little is known about the influences that specific particle properties (e.g., size, chemical composition or surface structure) have on ice heterogeneous nucleation. We lack both a fundamental understanding of the process and factors involved as well as experimental validation of the theoretical tools used to quantify heterogeneous ice formation processes. In particular, experiments that concern the freezing of internally mixed particles consisting of insoluble cores and inorganic and/or organic coatings are needed to identify and quantify the controlling processes and parameters. Laboratory investigations should utilize available or new cloud simulation facilities as well as ice nucle-ation instruments with sophisticated methods to generate and characterize relevant aerosol particles. In particular, the roles of particle size, surface area, and surface structure and composition (e.g., active surface sites) need to be investigated. Here, AIDA, LACIS, CFDC, ZINC, and FINCH have been and will continue to be useful. In designing experiments, conditions should resemble those of mixed-phase clouds so that insight can be gained, for example, on the relative importance of the different freezing mechanisms.

Aerosol-based Parameterizations of Cloud Ice Formation

A detailed understanding of aerosol-induced ice nucleation processes and their numerical implementation is prerequisite for reliable forecast of clouds, precipitation, and climate change. Homogeneous freezing rates of aerosol particles can be parameterized in numerical models as a function of the temperature, cooling rate, and aerosol parameters. By contrast, aerosol-related parameterizations for heterogeneous ice nucleation processes are more difficult to assess. Most models still describe the abundance of heterogeneous IN only as a function of temperature and humidity. New concepts have only recently been suggested to consider specifi c aerosol properties for the parameterization of heterogeneous ice nucleation in models. These need to be evaluated for application under variable cloud conditions and with relevant tropospheric aerosol systems. Comparison of results from laboratory cloud simulation experiments is needed to test and improve existing parameterizations, or to develop new ones, because it is difficult to constrain formulations of cloud microphysics in models to field measurements.

Secondary Ice Formation or Ice Multiplication (Splintering)

From fi eld experiments and cloud modeling studies, we know that secondary ice multiplication, such as the splintering of existing ice crystals or the so-called Hallett-Mossop effect, are important factors in cloud development and for the initiation and intensity distribution of precipitation. To date, only a few laboratory investigations have been conducted on secondary ice formation, and thus further experiments are needed to investigate and quantify this, if possible for relevant ice crystal shapes and sizes and under simulated cloud conditions. Such experiments could be conducted with single droplets and ice crystals in EDB setups or in cloud simulation chambers.

The University of Manchester has built a new ice cloud chamber facility suitable for such experiments. The device consists of three large cold rooms arranged vertically above each other on three floors of the building. These are joined by a fall tube (diameter of 1 m and height of 10 m) in which ice cloud properties and, in particular, crystal growth can be studied over timescales much longer than has been possible with smaller chambers. Temperatures in the cold rooms can be controlled to -50°C, and each of the three chambers can be controlled separately. These new facilities will allow previous work on ice particle nucleation, riming, charge transfer, and interaction of radiation with ice particles to be continued and expanded to include new areas such as secondary ice formation and crystal growth.

Generation and Characterization of Particles Used in Laboratory Cloud Simulations

Reliable generation and sufficient characterization of the particles used in laboratory cloud simulations are already important issues in laboratory cloud research. In addition to the identification of particles aimed at increasing our understanding of particle hygroscopic growth and activation behavior, we must ensure that particles are found which are suitable to the investigation and quantification of the different mechanisms involved in ice formation. Both types of investigations imply generation and characterization of multiphase (e.g., insoluble core and liquid mantle), multicomponent particles, well defined with respect to size and chemical composition. In addition, ice formation investigations require well-defined and characterized particle surfaces.

Another interesting topic involves particle/droplet charge. Electrical charges in particles/droplets may influence microphysical processes such as activation, ice nucleation, and droplet and ice particle growth. However, the generation and characterization of particles/droplets larger than one micrometer with well-defined charge levels, as needed when investigation charge influences microphysical processes, still pose a serious problem.

The above requirements can only be fulfilled if both particle generation and characterization techniques are significantly improved. The greatest need lies in the fields of generating and characterizing particle surface properties.

Specific Cloud Droplet-Turbulence Interactions

Investigations of cloud droplet-turbulence interactions (e.g., entrainment, turbulence-influenced condensational and collisional growth, and the spatial and Lagrangian properties of particles in turbulence) are far from complete, and there is a strong need for controlled laboratory experiments to isolate and quantify the mechanisms. Several challenges exist, including the development of experimental facilities that can match all of the relevant dimensionless parameters governing the processes under consideration. For example, when considering the Lagrangian properties of inertial particles in turbulence, it is necessary to match the particle inertia (Stokes number) as well as the particle settling (settling parameter and Froude number). Many laboratory systems tend to produce turbulence with much higher energy dissipation rates than exist in typical clouds, which makes overlap with the correct parameter ranges challenging. Further complicating matters is the difficulty in achieving large, geophysical Reynolds numbers in laboratory systems: any process sensitive to intermittency will be subject to this constraint, and only very large systems, such as the Gottingen wind tunnel or the large mine shafts described earlier, may be able to approach the Reynolds numbers that are expected to exist, for example, in a cumulus cloud.

Combining Turbulence and Microphysics over Multiple Scales

The topics and experiments discussed above address the investigation, simulation, and quantification of single cloud microphysical and turbulent processes. Although vital to our overall understanding of cloud processes, they can only be considered as an initial step in this process. One key topic that has been intensely debated concerns the issue of what really controls cloud properties such as droplet number and size, lifetime and precipitation behavior. An aspect of this problem is the question of homogeneous versus inhomogeneous mixing. Since the early laboratory experiments that initially led to the development of these ideas in the 1980s, essentially no laboratory studies have been dedicated to this important issue. This illustrates a challenge that we face on many issues involving large-scale turbulence: because entrainment and mixing are inherently multiscale processes, it is difficult to achieve such large ranges of spatial and temporal scales in typical laboratory-sized chambers.

To answer or even address these questions, none of the devices and techniques described above appears to be fully suitable. Therefore we suggest that experiments be designed and performed that allow the controlled and well-defined adjustment of both microphysical and turbulence parameters. Only this way can the possible interactions and feedbacks between the microphysical (activation, growth, freezing) and turbulent transport processes within clouds be simulated and quantified. Devices such as small-scale expansion chambers and wind tunnels, or maybe a combination thereof, might be applicable.

Conclusions

In this chapter, we discussed the simulation of clouds in the laboratory and focused on cloud-related topics and effects that have been or could possibly be investigated in laboratory studies. Topics such as aerosol particle hygroscopic growth and activation, droplet dynamic growth, ice nucleation and droplet freezing, and droplet-turbulence interactions were presented, and a number of devices used for laboratory investigation and simulation of relevant cloud processes were discussed. Since cloud physical processes and their simulation involve different scales, we classified cloud investigation/simulation devices into (a) devices that measure selected particle and cloud droplet properties, (b) devices that allow process studies under simulated cloud conditions, and (c) devices that permit a nearly full-scale cloud simulation. However, we emphasize that at the present time, and most likely in the near foreseeable future, no device exists that is capable of simulating a "real" cloud, with all of its relevant processes and complexity.

For the future, we suggest that investigations be continued and/or initiated to address (a) particle hygroscopic growth and activation, (b) the accommodation coefficients of water vapor on liquid water and ice, (c) aerosol effects on primary ice formation in clouds, (d) aerosol-based parameterizations of cloud ice formation, secondary ice formation or ice multiplication, (e) the production and characterization of particles suitable for cloud simulation experiments, and (f) the combination of turbulence and microphysics. We consider the latter to be of particular importance in the simulation and quantification of possible interactions and feedbacks between the microphysical (activation, growth, freezing) and turbulent transport processes within clouds.

Acknowledgments

We acknowledge, in alphabetic order, the very helpful contributions of Karoline Diehl (University of Mainz, Germany), Yasushi Fujiyoshi (Institute for Low Temperature Science, Hokkaido University, Sapporo, Japan), John Helsdon (South Dakota School of Mines & Technology, Rapid City, South Dakota, U.S.A.), Szymon Malinowski (Warsaw University, Poland), Masataka Murakami (Meteorological Research Institute, Tsukuba, Japan), and Zellman Warhaft (Cornell University, Ithaca, New York, U.S.A.).

References

Ayyalasomayajula, S., A. Gylfason, L. R. Collins, E. Bodenschatz, and Z. Warhaft. 2006. Lagrangian measurements of inertial particle accelerations in grid generated wind tunnel turbulence. Phys. Rev. Lett. 97:144507.

Bailey, M. and J. Hallett. 2002. Nucleation effects on the habit of vapour grown ice crystals from -18° to -42°C. Q. J. Roy. Meteor. Soc. 128:1461-1483.

Benz, S., K. Megahed, O. Möhler et al. 2005. T-dependent rate measurements of homogeneous ice nucleation in cloud droplets using a large atmospheric simulation chamber. J. Photochem. Photobio. A. Chem. 176:208-217.

Bundke, U., H. Bingemer, B. Nillius, R. Jaenicke, and T. Wetter. 2008. The FINCH (Fast Ice Nucleus Chamber) counter. Proc. 17th Intl. Conf. on Nucleation and Atmospheric Aerosols. Atmos. Res., in press.

Davidovits, P., C. E. Kolb, L. R. Williams, J. T

0 0

Post a comment