Ice nucleation properties of mineral dust particles: Determination of onset RH i , IN active fraction, nucleation time-lag, and the e ﬀ ect of active sites on contact angles

A newly developed ice nucleation experimental set up was used to investigate the heterogeneous ice nucleation properties of three Saharan and one Spanish dust particle samples. It was observed that the spread in the onset relative humidities with respect to ice (RH i ) for Saharan dust particles varied from 104% to 110%, whereas for the Span- 5 ish dust from 106% to 110%. The elemental composition analysis shows a prominent Ca feature in the Spanish dust sample which could potentially explain the di ﬀ erences in nucleation threshold. Although the spread in the onset RH i for the Saharan dust samples were in agreement, the active fractions and nucleation time-lags calculated at various temperature and RH i conditions were found to di ﬀ er. This could be due to the 10 subtle variation in the elemental composition of the dust samples, and surface irregularities like steps, cracks, cavities etc. A combination of classical nucleation theory and active site theory is used to understand the importance of these surface irregularities on the nucleability parameter contact angle that is widely used in the ice cloud modeling. These calculations show that the surface irregularities can reduce the contact 15 angle by approximately 10 ◦ .


Introduction
Ice clouds constitute the largest source of uncertainty in predicting the Earths' climate behavior according to last Intergovernmental Panel on Climate Change 2007 report (Forster et al., 2007). The uncertainty arises in part because of the lack of understand- 20 ing of the complex ways in which these clouds are formed. The dominant ice formation mechanism, for temperature below −38 • C, is homogeneous nucleation of the supercooled water droplets and aqueous aerosol particles, with the nucleation rate increasing for colder temperatures (Jeffrey and Austin, 1997;Pruppacher and Klett, 1997: henceforth P&K97). At temperatures warmer than −38 • C ice formation take place het- 25 erogeneously. Heterogeneous ice nucleation requires special atmospheric aerosols 11300 servations and theories. While observational and theoretical results for homogeneous nucleation can be compared with good agreement, theoretical treatments of heterogeneous nucleation that involve IN surfaces are very difficult to relate to observations (Cantrell and Heymsfield, 2005). Past studies (DeMott et al., 2003;Cziczo et al., 2004;Richardson et al., 2007) have 15 identified mineral dust particles as the IN inside the ice crystals collected on the ground or from clouds. Various laboratory experiments (Archuleta et al., 2005;Kanji and Abbatt, 2006;Möhler et al., 2006) have also shown that mineral dust can initiate ice formation at low saturations and warmer temperatures than homogeneous freezing, and thus can alter the ice cloud properties to the larger extent than a purely homogeneous 20 nucleation scenario. In the present experimental study, we not only investigate mineral dust as IN, but also quantify this understanding by evaluating nucleability parameters (e.g., contact angle) which can eventually be used in cloud models. Laboratory studies dealing with heterogeneous ice nucleation on dust particles, in particular deposition ice nucleation, have been undertaken since the 1950s. Early 25 studies (Mason and Maybank, 1958;Roberts and Hallett 1967;Schaller and Fukuta, 1979) reported mineral dust particles as efficient IN. Recent studies (Bailey and Hallett, 2002;Archuleta et al., 2005;Salam et al., 2006;Möhler et al., 2006;Dymarska et al., 2006;Knopf and Koop, 2006;Kanji and Abbatt, 2006) have also shown that Introduction

Conclusions
References Tables  Figures   Back  Close Full Screen / Esc

Printer-friendly Version
Interactive Discussion mineral dust is a good IN for various temperatures, supersaturations with respect to ice (SS i ) and sizes of dust particles. In order to better quantify the experimental ice nucleation rates, test nucleation theories, and further develop ice cloud parameterizations schemes it is also necessary to investigate ice nucleation studies of IN as a function of time. 5 Few empirical ice formation formulations for heterogeneous ice nucleation have been developed, and the commonly used ones are either SS i dependent (Meyers et al., 1992), temperature dependent (Fletcher, 1962), or combination of both (Cotton et al., 1986). These formulations rely on data from instruments that measure ice nucleation properties at controlled temperature and saturation ratios independent of IN physical 10 and chemical properties. In one ice chamber study Möhler et al. (2006) show the fractions of ice crystals formed are mainly a function of SS i at a given temperature. However, the relative importance of temperature and SS i on ice nucleation is not yet established. Here we have performed systematic experiments to address these questions. 15 Ice formation formulations in cloud models often use classical ice nucleation theory (CNT) (Comstock et al., 2008) described by P&K97. CNT in its limit, i.e. when independent of IN nature (e.g., surface characteristics), offers a reasonable estimate of ice nucleation observations. CNT fails in describing IN surface characteristics involved in the ice formation process, e.g., distribution of active sites over the IN surface. It is one 20 of the aims of this study to quantify the IN surface characteristics using active site theory (Gorbunov and Kakutkina, 1982) and recalculate the nucleability parameters that are used directly by CNT. In addition, the accuracy of models also relies on incorporating a broad distribution of nucleation onsets observed within single type of IN (Knopf and Koop, 2006). 25 A total of four different dust samples were collected from four different locations (three from the Saharan desert and one from Spain) for the present study. The advantage of using these dust particles is that we can experiment with the natural dust particles which eliminate the variable of atmospheric processing. Introduction

Conclusions
References Tables  Figures   Back  Close Full Screen / Esc

Printer-friendly Version
Interactive Discussion In the present study, using a newly developed experimental set-up (Kulkarni et al., 2009), we have investigated the ice nucleation efficiency of natural mineral dust particles as a function of temperature, saturation ratio and nucleation time. In Sect. 2, an experimental procedure to determine various ice nucleation properties including a procedure for the ice nucleation experimental set up are presented. In Sect. 3 we present 5 the different ice nucleation characteristics, e.g., nucleation onset as a function of temperature and saturation ratio, demonstrate the importance of active sites by measuring the nucleation time-lag, formulate a new ice cloud parameterization, and calculate the nucleability parameters used in the CNT. The conclusions are presented in Sect. 4.

Preparation of the experimental set up
Four natural mineral dust samples were used in this study. The samples were collected at ground level from three different locations in West Africa: Nigeria (2 • E, 15 • N), Dakar (outside the city), Dakar, during the African Monsoon Multidisciplinary Analyses (AMMA) campaign, and one location from Spain (South East coast). The Dakar sam-15 ple collected outside the city is termed as Dakar-1 sample henceforth. The collected samples were stored and transported in inert, leak proof plastic bottles to avoid contamination. The Spanish sample was selected because it has different mineralogical elemental composition compared to West African dust samples. The collected dust samples were sieved to obtain dust particles less than 38 µm, which were then used 20 during the experiments. Figure 1 shows a schematic diagram of the Thermal Gradient Diffusion Chamber (TGDC) experimental apparatus. The chamber is made up of two parallel horizontal plates with the inside of both upper and lower plates coated with ice. It consists of an optical microscope mounted in such a way that the objective lens of the microscope 25 can be raised and lowered into the optical port of the top plate. A clear glass window Interactive Discussion 2 mm thick seals the water vapor from escaping from the chamber. The experimental apparatus is housed inside a walk-in cold room, which is controlled to cool the plates of the apparatus. The thickness of ice layers is maintained constant by freezing a known quantity of deionized water (18 MΩ). A detailed procedure of preparing the ice layers is outlined in Kulkarni et al. (2009). An average ice layer thickness of 4 mm was used 5 to produce a 12 mm gap between the top and bottom ice layers. A heating mat warms the top plate and this produces a thermal and vapor diffusion gradient between the plates. Using Flatau et al. (1992), the saturation vapor pressure over the ice (e si ) at respective temperatures are calculated. These calculations are further used to give the relative humidity with respect to ice (RH i ) at different heights (d ) for the lower half of 10 the chamber using Eqs.
(1-4) as follows, first observed on any dust particle under the observation area of the microscope. The onset RH i of the dust particles from each location was investigated in the overall temperature range −10 to −34 • C. In these experiments, approximately the same numbers of particles (500 per mm 2 ) were deposited on the Teflon substrate. The substrate holding the sample holder is raised to 1 mm distance above the bottom plate 10 ice layer and the microscope is adjusted to view the dust samples. If nucleation is observed then the experiment is terminated; otherwise the sample holder is raised another 1 mm. During the experiment, optical images of the particles are recorded. The total number of dust particles counted in the field of view varies between 5 and 15, and we define the nucleation threshold as formation of ice on any 1 dust particle. 15 The active IN fraction was calculated from the fraction of ice crystal numbers observed at a given temperature and SS i to the total number of dust particles in the field of view of the microscope. Experiments to calculate the active fraction were performed at two different temperatures (−20 • C and −30 • C) and two different RH i values (110% and 116%). In this experiment the dust particles were deposited on the Teflon sub-20 strate and the sampling rod was raised directly to the point of the desired RH i . Images were taken from time=0 s until no more new dust particles nucleated ice (time=tmax). It was observed the dust particles do not nucleate immediately at time t=0; there was a characteristic time-lag before nucleation was observed. Once this time-lag was exceeded, the dust particles began nucleating and continued until time=tmax, when no ACPD 9,2009 Ice nucleation properties of mineral dust particles G. Kulkarni

Elemental analysis of dust particles
Using an Environmental Scanning Electron Microscope -Energy Dispersive X-ray (ESEM-EDX) set-up, dust particles were investigated for their morphology and elemental chemical composition. Figure 2 shows the detailed surface features of an example 5 dust particle from Dakar-1 source. The irregularities of the dust surface may be due to erosion and weathering processes at the source region. Such irregularities over the surface might be responsible for the change in the surface free energy and reactivity with different chemical compounds (Kärcher and Lohmann, 2003). Foreign gases such as SO 2 , NH 3 might occupy the active sites as opposed to H 2 0 and this can reduce 10 the nucleability of IN (P&K97). It is also possible that irregular features, cracks, steps, or pores, might fill up with water from the vapor phase and enhance the activation of the dust particle. An interesting effect, termed to as "memory effect" or "preactivation" has been observed previously (Mason and Maybank, 1958;Roberts and Hallett, 1968;Knopf and Koop, 2006). These authors surmise that ice deposited inside the cracks 15 might survive ice sub-saturated vapor pressure conditions due to a decrease in the equilibrium vapor pressure as a result of concave curvature (also known as negative Kelvin effect). While we have not investigated these effects experimentally we have mathematically quantified the significance of these irregular features using active site theory described in the Sect. 3.5.

20
Ice embryo formation on the dust surface can be a selective process; a small area of the total surface can be hydrophilic and might favor the embryo formation. To minimize the uncertainties due to chemical inhomogeneity and for a greater understanding of the role of the dust surface in the ice nucleation, we determined the elemental composition of individual and bulk dust particles. 25 In the first examination of the dust, the elemental composition of a complete individual dust particle (larger than 10 µm) was analyzed, and later compared to the composition obtained from various small views (approximately 2×2 µm) over the same 11306 Introduction

Tables Figures
Back Close

Full Screen / Esc
Printer-friendly Version Interactive Discussion particle. In total 40 dust particles were analyzed, and a similar elemental composition distribution with ±5% variation of weight percent between the two analyses was observed. Thus it is inferred that the chemical composition over the entire dust surface is relatively homogeneous at this scale. The second study involved bulk elemental composition analysis of 25-30 dust par-5 ticles. Table 1 shows the different elements observed with their respective average weight percent distribution of Si, Al, Mg, Ca, Na and Fe. Trace amounts (<1%) of other elements, including P, K, Ti, Na and Cl, were also observed in the analysis but are not presented. The data from Table 1 show that variations exist in the amount of elements from each source region. Nigerian dust has more Si but less Ca compared with the Dakar-1 and Dakar dust samples. The Dakar-1 and Dakar dust particles were also associated with a higher Fe content. Coastal dust particles have a larger percentage of Na compared with other source locations, which might be due to sea salt mixing with the dust particles. The Spanish dust had the highest percentage of Ca (60%) compared with other source locations, and this might have been derived from carbonate 15 minerals. Variations of elemental composition of dust particles across the four source regions could be significant for ice nucleation if they are transported throughout the atmosphere. For example, from Table 1 Spanish dust is observed to contain a comparatively larger amount of calcium and dust particles containing high levels of calcium are found to be reactive with respect to nitric acid (Dentener et al., 1996). Nitrates 20 are hygroscopic in nature and deliquesce at low relative humidity with respect to water (Tang and Fung, 1997;Al-Abadleh et al., 2003). Therefore dust particles with an abundance of calcium, once transported for long periods might be neutralized with nitric acid. These particles would then have the ability to serve as effective IN in the appropriate atmospheric conditions. In short, dust particles provide a reactive site for many Introduction

Ice onset RH i determination
Experiments were carried out to examine the onset nucleation properties of mineral dust particles in the deposition ice nucleation mode. Figures 3 to 6 show the plots of onset RH i as a function of temperature for the dust samples. The data points shown as square boxes represent where onset nucleation was observed and the probability 5 of onset nucleation event increases with the increase in area of the square box. It was observed that onset nucleation occurred as low as 104%, and the spread in the onset RH i was found to vary from 104% to 110% for all the dust samples. The results are in general agreement with other past findings, although the experimental set-up, sample preparation and types of dust particles were different. For example Mangold 10 et al. (2005) and Knopf and Koop (2006) studied the ice nucleation abilities of Arizona test dust, and observed onset RH i as low as 105%. In particular, our results agree with both different and similar types of dust particles including Saharan dust studied by the Kanji and Abbatt (2006). They observed that all dust samples initiated ice formation between 102 and 108% RH i , with temperature range between −10 • C to −55 • C.
15 Figure 6 shows the spread in the onset RH i of Spanish dust which was found to have less of a RH i spread than the other location dust particles. Here the range varied from 106% to 110% and this might be due to the variation in the surface chemical composition compared with the other dust particles (see Sect. 3.1 more details) and/or the variation of surface physical features such as cracks, steps, cavities. 20 To test the possible effect of soluble compounds and/or bacteria experiments were performed where we tempered the dust particles at 350 • C for one hour. We did not find any difference between the onset RH i determined before and after tempering. Thus we conclude that any soluble compounds and soil bacteria associated with the dust particles did not play a role during the ice nucleation experiments. 25 The general spread in the onset RH i may be due to the variation in the surface characteristic features such as elemental composition inhomogeneity and distribution of active sites. But in Sect. 3.1 it is shown that dust particles have uniform elemental ACPD Introduction

Tables Figures
Back Close

Full Screen / Esc
Printer-friendly Version Interactive Discussion composition across their surface, and therefore we think this observation might not play an important role in the onset. Thus it can be inferred that the observed spread in the onset RH i might be probably due to only a distribution of active sites and efficiency of active sites to initiate ice nucleation varies from one dust particle to another in the same sample. Additional support to this premise comes from results shown in Sect. 3.3, the 5 variation of nucleation time-lag.

Active fraction and nucleation time-lag
The fraction of dust particles that activate at various temperature and RH i values are shown in of RH i . This might be due to the larger equilibrium ice vapor pressure at warmer temperature compared to colder temperatures. Such that the water vapor molecules at warmer temperature have high mobility, which increases the probability of water vapor molecules attaching to the dust surface. It should be noted that the experimental RH i values at these temperatures are well 20 above the threshold RH i , and therefore their probability of nucleation is greater than zero. The dust from Nigeria and Dakar-1 showed a similar behavior in active IN fraction, but the percentage of active IN is higher for the sample from Nigeria compared to Dakar-1. The differences might be due to subtle variation of elemental composition 25 across the dust surface shown in Sect. 3.1. Other possible reason might be the distribution of active sites, but we do not have any experimental observations to support this idea. Salam et al. (2006)

Printer-friendly Version
Interactive Discussion rillonite types of dust particles. They calculated active IN fraction from the ratio of the ice crystal number concentration active at a given temperature over the total number of ice crystals at 100% relative humidity with respect to water at −40 • C, and observed increasing active IN fraction with increase in RH i . We also observed the similar trend. Their results showed that montmorillonite dust type has higher active IN fraction at −20 5 and −30 • C compared to kaolinite.
The importance of active sites in the deposition ice nucleation mode can be further investigated by understanding the "nucleation time-lag". An appreciable time delay was observed for the appearance of an first ice cluster on any individual dust particle in the field of view (125 µm×94 µm), once the dust particles has been exposed to different 10 temperature and RH i conditions. The time delay at −30 • C is longer compared to −20 • C at two different RH i conditions for Dakar-1 location dust particles.
To date, only one study (Anderson and Hallett, 1976) has been conducted to understand the time-lag of ice nucleation on AgI and CuS surfaces as a function of temperature and RH i . The present work is the first of its type to study the nucleation time-lag 15 of Saharan mineral dust particles at various constant temperature and RH i values in a deposition mode of ice nucleation. Anderson and Hallett (1976) observed the time-lag is longest at low RH i . This is in agreement with the present experiments, where it is observed that the lag is longer at 110% compared with 116% RH i by approximately 40 s. We observe at 110 RH i the time-lag, at −30 • C and −20 • C, is between 80 to 170 s 20 and 85 to 135 s, respectively; whereas, at 116 RH i it is between 85 to 120 s and 65 to 100 s, respectively. It is observed that the nucleation experiments performed at low temperature and RH i values requires more time for the ice nucleation.
The time delay suggests variation of active sites within the dust particles. According to P&K97, each active site on any individual dust surface requires a critical RH i , at a 25 given temperature, which must be applied for a critical length of time for an ice embryo to form and to grow to an ice cluster, and an eventual ice crystal. The time delay for the ice embryo formation can also be associated with the time required to overcome the free energy barrier for nucleation. It should be noted that nucleation time-lag experi-Introduction

Conclusions
References Tables  Figures   Back  Close Full Screen / Esc

Printer-friendly Version
Interactive Discussion ments were conducted at RH i higher than the threshold RH i (onset RH i ) at respective temperatures. The maximum energy barrier value can be given by (Mason, 1971) Eq. (5), Where σ is the surface tension, and r * is the critical radius of the ice embryo given as 5 Eq. (6), where M is the molecular weight of water vapor, ρ is the ice density, R is universal gas constant, T is the temperature, e is the equilibrium vapor pressure, and e s is the saturation vapor pressure. 10 Assuming r * =0.01 µm (P&K97) we calculated using Maxwell diffusion growth theory the time required for the ice embryo to grow a size (=1.1 µm) detectable by the microscope. It was found that the time required for an embryo to grow to this size is 36 s at −20 • C and 110% RH i . This suggests that the if the total time observed is 85 s then time required for the establishment of ice embryo had taken approximately 49 s, 15 which is the time needed to overcome the energy barrier imposed by the dust surface. Developing ice nucleation parameterizations as a function of free energy barrier and nucleation time-lag can be one of the solutions to represent the complex nature of heterogeneous ice nucleation in the global climate models. Future work should be carried out to understand more about the type of active sites that may increase or decrease 20 the energy barrier. This might help in understanding why sometimes the aerosol particles do not participate in the ice nucleation process under the favorable atmospheric conditions.  Figure 7 shows the fraction of aerosols activated plotted as a ratio of number of ice crystals (Ni) observed to the total number of IN (Nf), as a function of time (t) for the two different geographical locations, Dakar-1 and Nigeria. An approximate behavior for the ice crystal formation can be represented as Eq. (7),

Rate of ice crystal formation as a function of time
where F i =Ni/Nf, A=1.0128, and B=111.33. It is observed that Fig. 7 exhibits no systematic variation with geographic location, except after 300 s of time. This suggests that the dust particles from at least these two sources have the unique ability to initiate ice formation in the early time period of ice 10 cloud development. The wide scatter in the data shown in the Fig. 7 can be attributed to the following two reasons. One reason could be subtle variation in the dust surface elemental composition. Recently Eastwood et al. (2008) showed in their ice nucleation experiments that dust minerals with different elemental composition have a wide range of onset nucleation RH i . Minerals like Quartz (SiO 2 ) and Calcite (CaCO 3 ) were observed to be nucleating at higher RH i compared to Kaolinite (Al 4 Si 4 O 10 (OH) 8 ) and Montmorillonite (Na, Ca) 0.3 (Al, Mg) 6 (Si 4 O 10 ) 3 (OH) 6 -nH 2 O). The montmorillonite has more -OH groups compared to Kaolinite, and the variation might enhance the nucleation as -OH groups in former mineral can attract more water vapor molecules (Salam et al., 2006). Shown in the Table 1 the dust particles from Dakar-1 and Nigeria have 20 subtle variations in the elemental composition, and this might affect the total number of IN at any particular temperature and RH i condition.
The second reason would be the variation of surface irregularities or roughness across the dust particles from both the locations. These surface features can be viewed as the wide distribution of active sites having different free energies of interaction with 25 water molecules. This energetically non uniform distribution of active sites, gives each site a different probability of becoming an active ice-nucleating site (Rosinski, 1980 Nigeria, and all dust particles at each location having nearly similar surface features. The mineralogy specific to a location as well as the local weathering and erosion characteristics might lead to this unique surface pattern. The ESEM images of these dust particles were analyzed to identify these patterns, but just visual inspection of the images did not reveal any patterns. More in-depth image analysis at very high resolution 5 of nanometer scale, including accurately measuring the particle surface area, needs to be carried out to understand more details about the surface patterns. Further Eq. (7) is modified to obtain rate of ice crystal formation (D nice /d t) as a function of total number of aerosol particles (n p ) and time (t) given as Eq. (8), 10 where K =112.755 and B=111.33. The above developed parameterization scheme can be used in the process level ice nucleation studies, which deals with the individual aerosol-cloud interactions. To better understand the effect of dust particles on large-scale cloud properties these process level studies are very useful. Recently, Baker and Peter (2008) highlighted the 15 importance of such studies to predict the development of individual clouds and cloud systems. Future research should be carried out extending these studies to different types of dust particles at various temperature and RH i conditions and characterizing the source locations and aging and transport. 20 In addition to developing the new ice cloud parameterizations schemes, the results are adapted to improve existing schemes and to suggest the new values for the various variables/parameters involved in the schemes. One common scheme used in the ice cloud modeling studies (Comstock et al., 2008) is the classical nucleation theory. The nucleation rate calculated by this theory is as Eq. (9), where J is the nucleation rate (cm −2 s −1 ), A o is the pre-exponential factor (cm −2 s −1 ), ∆G o is the activation energy barrier for homogeneous nucleation, k is the Boltzmann constant, T is the temperature, A is the surface area having active sites or defect area, f (m) is (2+m)(1-m) 2 /4, and m is defined as cos(θ) with θ being the contact angle.

Contact angle and active site theory
The ice forming ability of IN can be expressed in terms of θ between the dust surface 5 and an ice embryo using Young's relation defined as Eq. (10), and shown in Fig. 8, Where m is the wettability parameter, σi j are surface free energies, and subscripts C, V and S refers to catalyzing substrate (dust surface in the present study), vapor (water vapor) and solid (ice), respectively.
The contact line of an ice-embryo growing upon a smooth solid substrate surface is a dashed line shown in Fig. 8. In reality the substrate has a degree of roughness and this decreases the θ to a new contact angle specified as θ r . The rough substrate can be assumed to possess the active sites in the form of steps, cavities etc. To understand the effectiveness of active sites on the ice forming ability of IN, we initially calculate the 15 contact angle without including the active sites (by substituting A=0.0 in Eq. 9). The Table 3 tabulates the contact angle determined for two dust source locations, and it can be observed that the contact angle varies approximately between 15 to 20 • .
To understand the influence of active sites on the contact angle, we first estimate the fraction (F ) of the total surface area which consist of active sites, and then using 20 dust particle surface area (A s ) we calculated the defect area (A) (A=F ×A s ). Then substituting A into Eq. (9) and using the same experimental nucleation rate, we recalculate the contact angles. The following Eqs. (11) and (12) are used to calculate F (Fletcher, 1969;Gorbunov and Kakutkina, 1982;Han et al., 2002 where A o is the minimum area of one active site (=2e-15 cm 2 ), S is the surface area of one active site (=2×A o ), γ is the width of active site distribution (=1), A s is the surface area of aerosol particle (=1 µm in radius). Assuming half probability (P =0.5) 5 distribution (Heneghan and Haymet, 2002) and solving the Eqs. (11) and (12) gives F =9e-07.
The calculated A is substituted into Eq. (9) to re-calculate the revised contact angle. Table 4 tabulates the change in contact angle after including the effect of active surface in the calculation. 10 The tabulated values in Table 4 illustrate the effect of inclusion of active surface on the contact angle. The revised contact angle value is approximately 10 • less than the values when active surface areas are not considered. It is known that the lower the contact angle of the aerosol particle the higher the nucleation efficiency. Generally in ice cloud modeling studies the nucleation rates are calculated by assuming the con-15 stant contact angle over the smooth aerosol particle. Use of revised contact angles may impact these modeling results.

Conclusions
Ice nucleation properties (i.e. the onset RH i , active fraction, and nucleation time-lag) of mineral dust particles collected from Sahara and Spain were determined using a 20 TGDC experimental set-up. Using these measurements two parameterization studies were undertaken: the rate of ice crystal formation as a function of time and a calculation of nucleation parameters (contact angles) using active site theory.
The To understand the effect of surface roughness on the ice forming ability of a dust surface we make use of CNT. By using experimental heterogeneous nucleation rates in conjunction with CNT we have determined the contact angles between an ice embryo and the dust surface. The range of contact angles is found to vary from 15 to 20 • . These results are used in conjunction with a theoretical model to evaluate the significance This study has demonstrated that laboratory measurement data can be successfully utilized to develop ice cloud parameterizations and estimate the range of nucleability parameters involved in existing parameterizations schemes. These measurements may help to develop and evaluate models dealing with aerosol cloud interactions or process level studies, which are necessary for understanding the impact of ice on climate. For a complete understanding of ice nucleation processes, more laboratory studies of this kind are urgently required to address the need to better resolve the impacts of ice 15 on climate. Nigeria, -20 deg C, 110% Dakar-1, -20 deg C, 116% Nigeria, -20 deg C, 116% Curve-fit θ and θ r are the contact angles when the substrate is smooth and rough, respectively. θ r is less than θ because of the surface irregularities of the substrate.