Experimental investigation of ice nucleation by different...

Meteorologische Zeitschrift Vol 14 No 4 485-497 (August 2005)ccopy by Gebruder Borntraeger 2005 Article

Experimental investigation of ice nucleation by differenttypes of aerosols in the aerosol chamber AIDAimplications to microphysics of cirrus clouds

ALEXANDER MANGOLDlowast1 ROBERT WAGNER2 HARALD SAATHOFF2 ULRICH SCHURATH2 CARSTEN GIESEMANN3 VOLKER EBERT3 MARTINA KRAMER1 and OTTMAR MOHLER2

1Institut fur Chemie und Dynamik der Geosphare I (ICG-I) Stratosphare Forschungszentrum Julich (FZJ)Germany2Institut fur Meteorologie und Klimaforschung IMK-AAF Forschungszentrum Karlsruhe (FZK) Germany3Physikalisch-Chemisches Institut Universitat Heidelberg Germany

(Manuscript received January 22 2004 in revised form September 3 2004 accepted October 8 2004)

AbstractThe aerosol chamber AIDA was used as a moderate expansion cloud chamber with cooling rates at the onsetof ice nucleation between ndash13 and ndash30 K minminus1 to investigate the nucleation and growth of ice crystalsin sulphuric acid ammonium sulphate and mineral dust aerosols at temperatures between 196 and 224 KSupercooled sulphuric acid droplets with mean diameters of about 02 to 03 microm nucleated ice by homo-geneous freezing at RHice increasing from 144 to 166 with temperatures from 220 and 196 K This isin good agreement both with previous results of AIDA experiments and literature data In contrast ammo-nium sulphate particles of similar size nucleated ice at the significantly lower RHice of 120 to 127 in thesame temperature range Fourier-Transform infrared (FTIR) extinction spectra of the aerosol revealed that theammonium sulphate particles mainly consisted of the liquid phase The number concentration of ice crys-tals formed during the homogeneous freezing experiments agree well with model results from the literatureHigher ice crystal number concentrations formed during the ammonium sulphate compared to the sulphuricacid experiments can be explained by the also somewhat higher cooling rates at ice nucleation Depositionice nucleation on mineral dust particles turned out to be the most efficient ice nucleation mechanism both withrespect to RHice at the onset of ice nucleation (102 to 105 in the temperature range 209 to 224 K) and theice crystal number concentration Almost all mineral dust particles nucleated ice at the lower temperatures

ZusammenfassungAn der Aerosolkammer AIDA wurde die Nukleation und das Wachstum von Eiskristallen in Schwefel-saure- Ammoniumsulfat- und Mineralstaub-Aerosolen im Temperaturbereich von 196 bis 224 K undKuhlraten von ndash13 bis ndash30 K minminus1 bei der Eisnukleation untersucht In der Aerosolkammer wurden hierfurdynamische Wolkenprozesse durch eine kontrollierte Druckabsenkung des Kammervolumens simuliert BeiExperimenten mit unterkuhlten Schwefelsaure-Tropfchen eines mittleren Durchmessers von etwa 02 bis 03microm entstanden Eiskristalle durch homogenes Gefrieren bei einer Eisfeuchte die im Temperaturbereich von220 bis 196 K von 144 auf 166 anstieg Dies stimmt sowohl mit Ergebnissen vorheriger AIDA Experi-mente als auch mit Literaturdaten gut uberein Im Gegensatz dazu entstanden bei Experimenten innerhalb desgleichen Temperaturbereichs mit Ammoniumsulfat-Partikeln ahnlicher Groszlige Eiskristalle bei den deutlichniedrigeren Eisfeuchten von 120 bis 127 Mit Hilfe von Fourier-Transform-Infrarot-(FTIR) Extinktions-spektren konnte gezeigt werden dass hauptsachlich flussige Ammoniumsulfat-Partikel vorhanden waren DieAnzahlkonzentrationen der Eiskristalle die bei den homogenen Gefrierexperimenten entstanden stimmengut mit Modellergebnissen aus der Literatur uberein Die hoheren Eiskristallanzahlkonzentrationen bei denAmmoniumsulfat-Experimenten im Vergleich zu den Schwefelsaure-Experimenten konnen durch die eben-falls leicht hoheren Kuhlraten bei der Eisnukleation erklart werden Mineralstaub-Partikel zeigten als Keimefur Depositionsgefrieren die hochste Wirksamkeit sowohl in Bezug auf die Gefrierfeuchten (102 bis 105 im Temperaturbereich von 209 bis 224 K) als auch auf die Anzahlkonzentrationen der Eiskristalle Bei dentieferen Temperaturen bildeten fast alle Mineralstaub-Partikel Eiskristalle

1 Introduction

Cirrus clouds may either form via homogeneous or het-erogeneous freezing of aerosol particles or via a combi-

lowastCorresponding author Alexander Mangold Forschungszen-trum Julich GmbH ICG-I 52425 Julich Germany e-mailamangoldfz-juelichde

nation of both processes (GIERENS 2003 HAAG et al2003b) The homogeneous freezing process of pure so-lution droplets occurs at temperatures between 235 and185 K at relative humidities with respect to ice (RHice)between 140 and 170 Homogeneous freezing was in-vestigated in various laboratory and field studies (eg

DOI 1011270941-294820050053

0941-294820050053 $ 585

ccopy Gebruder Borntraeger Berlin Stuttgart 2005

486 A Mangold et al Experimental investigation of ice nucleation Meteorol Z 14 2005

HEYMSFIELD and MILOSHEVICH 1993 JENSEN et al1998 MARTIN 2000 KOOP et al 2000 FIELD et al2001 MOHLER et al 2003) Based on laboratory ex-periments KOOP et al (2000) provided a water activitybased parameterisation for the homogeneous freezing ofsupercooled solution droplets which was confirmed bydetailed measurements in ice clouds formed by homoge-neous freezing of sulphuric acid solution droplets at theaerosol chamber AIDA of Forschungszentrum Karls-ruhe (MOHLER et al 2003 HAAG et al 2003a) Us-ing the freezing nucleation rates of KOOP et al (2000)KARCHER and LOHMANN (2002ab) developed a mi-crophysical parameterisation of the homogeneous freez-ing of supercooled solution droplets showing that theice crystal number density in cirrus clouds can mainly bedescribed as function of updraft velocity (ie the coolingrate) and temperature They also stated that enhancednumber concentrations of sulphate aerosols occurringeg after volcanic eruptions would not influence sig-nificantly the process of cirrus formation and thereforethe relationship between aerosol and ice particle numberconcentrations in homogeneously formed cirrus cloudswould be much weaker than in liquid water clouds

Heterogeneous freezing of particles containing an in-soluble inclusion occurs at ice supersaturations signif-icantly below the pure homogeneous freezing thresh-old HAAG et al (2003b) showed based on in situmeasurements of RHice in and outside of cirrus clouds(OVARLEZ et al 2002) that in parts of the pollutednorthern hemisphere cirrus clouds are formed via theheterogeneous pathway but that homogeneous ice nu-clei are also involved There are several laboratory andfield studies investigating the heterogeneous freezing ofwater on soot coated with sulphuric acid (DEMOTT etal 1999) mineral dust (DEMOTT et al 2003) mineraldust immersed in aqueous ammonium sulphate particles(ZUBERI et al 2002) crystallised ammonium sulphate(ZUBERI et al 2001) as well as the ice nucleation onflame soot aerosol of different organic carbon content(MOHLER et al 2005 this issue) These studies showthe temperature and RHice dependence of the freezingprocess for the various types of aerosol particles HUNG

et al (2003) found that the heterogeneous ice nucleationrates of ammonium sulphate solution droplets with amineral dust core depend beside temperature and ice su-persaturation on the diameter of the immersed mineralparticle KARCHER and LOHMANN (2003) provided amicrophysical parameterisation of the heterogeneous icenucleation in the immersion mode For intermediate up-draft velocities they found a strong aerosol effect oncirrus properties probably also affecting the radiativeproperties of cirrus clouds and thereby indirectly affect-ing the climate The presence of two types of ice nucleiduring ice formation and growth differing distinctly bytheir freezing thresholds would lead to a lower ice crys-

tal concentration This is due to competition of the twotypes of ice nuclei for the available water vapour duringthe growth phase

Up to now there are only few field studies exam-ining the factors controlling the ice nucleation pro-cesses and the resulting frequency of cirrus cloud oc-currence and their optical properties From the INCAfield campaign in the northern and southern hemi-sphere in 2000 a first insight into the ice formation pro-cesses is gained (OVARLEZ et al 2002 HAAG et al2003b) The CRYSTAL-FACE field campaign in 2002provided a comprehensive dataset for the characterisa-tion and description of the cirrus cloud formation (seehttpcloud1arcnasagovcrystalfaceindexhtml) Butadditional investigations of the relationship betweenaerosols and cirrus clouds especially with respect to icenucleation and crystal properties are of high interest

Here we present results of detailed laboratory mea-surements of homogeneous and heterogeneous ice nu-cleation performed at the aerosol chamber AIDA un-der simulated atmospheric cirrus conditions The ex-periments were conducted at temperatures between 224and 196 K using FTIR extinction spectroscopy to deter-mine microphysical parameters of the ice clouds Sul-phuric acid and ammonium sulphate solution dropletswere used during homogeneous ice nucleation experi-ments while mineral dust particles (Arizona Test Dust)served as nuclei for heterogeneous ice formation Inthis paper we first focus on describing our experimentalmethods Thereafter we show the influence of tempera-ture and cooling rate on characteristic parameters of theformed ice clouds especially the ice water content theice crystal number concentration and mean crystal sizefor the different types of aerosols Differences betweenthe freezing relative humidities of the different aerosoltypes are also discussed

2 Experimental

21 AIDA Aerosol and cloud chamberfacility

The AIDA aerosol chamber is a large vessel of 84 m3

volume which can homogeneously be cooled down to183 K The pressure range covers 01 to 1000 hPa Un-der constant wall and gas temperature conditions icesaturation is maintained by a thin ice layer on the cham-ber walls The ice supersaturation necessary for the ho-mogeneous or heterogeneous nucleation of ice crystalsis achieved by lsquovolume expansionrsquo due to controlledpumping usually from 1000 to 800 hPa During pump-ing the relative humidity with respect to ice (RHice) in-creases by up to 50 minminus1

The onset of ice formation is precisely detected bymeasuring the increasing intensity and depolarisation of

Meteorol Z 14 2005 A Mangold et al Experimental investigation of ice nucleation 487

Figure 1 FTIR spectra monitoring the in situ neutralisation of supercooled H2SO4H2O solution droplets by the addition of NH3(g) at a

temperature of 212 K (indicated by the arrow between the two lower spectra) left complete spectrum CO2(g) absorption regime omitted

right zoom from 2000ndash900 cmminus1 The spectra are offset for clarity A reference spectrum of crystalline ammonium sulphate particles

(uppermost spectrum) added to the AIDA chamber at 224 K is shown for comparison Additional arrows (in combination with labels) are

used to assign the characteristic extinction bands in the aerosol spectra

laser light back-scattered by the growing ice particleswith high sensitivity and a time resolution of 1 Hz (fora detailed description see MOHLER et al 2003) TheLyman-α-fluorescence hygrometer FISH (ZOGER et al1999) is used to measure the total water concentration(gas phase + condensed phase) at a time resolution ofabout 1 Hz Its overall accuracy is about 6 with a wa-ter vapour detection limit of about 002 Pa at 1000 hPatotal pressure Simultaneously the water vapour concen-tration in the AIDA vessel is measured in situ at 1370plusmn2nm by a tunable diode laser (TDL) absorption spectrom-eter The laser which is located outside the thermostatedhousing is coupled by an optical fibre to a White multi-path cell with 82 m optical path length permanently in-stalled inside the AIDA vessel This system provides atime resolution of about 1 Hz an accuracy of 5ndash10 and a resolution in the H2O(g) mixing ratio of up to 15ppb It will be described in detail elsewhere (EBERT etal 2004) The data evaluation procedures used are sim-ilar to earlier in situ TDL spectrometers which were de-veloped eg for a sampling-free detection of gaseous O2in multi-phase water sprays (SCHLOSSER et al 2002)CO in power plants (TEICHERT et al 2003) or simul-taneous in situ measurements of stratospheric CH4 andH2O (GURLIT et al 2005)

The water vapour saturation pressures and RHice arecalculated from the water vapour and temperature mea-surements using ice saturation vapour pressures ac-cording to MARTI and MAUERSBERGER (1993) Mainsources of the RHice uncertainty result from temperature

inhomogeneities mainly due to internal heat sources(eg the heated sampling tube for the total water mea-surements) and incomplete mixing during strong pump-ing These uncertainties are plusmn 01 K at constant p and Tconditions and less thanplusmn 03 K during expansions Theoverall error for RHice is estimated to range between 6and 10 of the actual RH value

Ice particle number concentrations and mean sizesare retrieved from FTIR extinction spectra In section24 a detailed description of the FTIR measurementsis given Number concentrations and optical diametersof growing ice crystals are also determined with twooptical particle spectrometers (PCS2000 and WELASPalas) The total aerosol number concentration is mea-sured with a condensation particle counter (CNC3010TSI) slightly modified for operation at reduced sam-pling pressures (SEIFERT et al 2004) The estimatederror for the PCS2000 is 30 and for the CNC301020 For a more detailed description of the instrumen-tation error estimation and the methods of AIDA icenucleation experiments see MOHLER et al (2003)

22 Aerosol generation

The sulphuric acid (SA) aerosol particles are generatedby saturating a synthetic air flow (7 l minminus1) at 120Cwith sulphuric acid Binary droplets of sulphuric acidand water nucleate upon cooling of the airsulphuric acidmixture while passing it through a stainless steel con-nection tube between the saturator and the AIDA cham-ber The size distribution of the added aerosol closely


fitted log-normal distributions with a count median di-ameter between 02 and 03 microm Aerosol number con-centrations were about 12000 cmminus3 after addition to theAIDA vessel was completed

2120

2155

2190

2225

2260

T [

K]

800

850

900

950

1000

p [h

Pa]

T_gas

T_wall

pressure

012345

H2O

[P

a]

8090100110120130140150

RH

ice

[]

Onset of freezingltminusminusminus

120000 120500 121000 1215000

10

20

30

40

50

024681012

dic

e [micro

m]

0

20

40

60

N_i

ce_a

bs [c

mminus

3 ]

time CEST

H2O

[ppm

v]

total water gas phase water sat_ice_gas sat_ice_wall RHice ice water content (FISHminusTDL) ice water content (FTIR) ice particle median diameter ice particle number conc

Figure 2 Time series of an AIDA ice nucleation experiment with

sulphuric acid aerosol at 224 K upper panel total pressure gas and

wall temperature in the AIDA chamber middle panel water vapour

pressure of gas phase water ice saturation pressure with respect to

gas and wall temperature as well as the relative ice saturation lower

panel water vapour mixing ratio of gas phase water ice saturation

with respect to wall temperature as well as the ice water content re-

trieved from FISH-TDL data and from FTIR spectra and ice crystal

number concentration and median diameter the black vertical line is

indicating the onset of freezing

To form deliquesced ammonium sulphate (AS) par-ticles the supercooled SA solution droplets are neu-tralised in situ by the addition of gaseous ammoniaAfter complete neutralisation to ammonium sulphateits gas phase absorption features becomes visible inthe FTIR spectra (Fig 1 right) As pointed out byHUNG et al (2002) and ZUBERI et al (2001) the exactphase of the ammonium sulphate particles ie entirelydeliquesced particles or an externally mixed aerosolcontaining small amounts of crystalline (NH4)2SO4 is

a crucial parameter when analysing ice freezing ex-periments Recently COLBERG et al (2003) provideda model to predict the physical state of atmosphericH2SO4NH3H2O aerosol particles The efflorescencerelative humidity (ERH relative humidity at which thecrystallisation process starts inside the deliquesced liq-uid particles) for deliquesced AS particles ranges be-tween 35 and 40 in the respective temperature rangeof our experiments During the generation of the ASaerosol in the AIDA chamber the relative humidity wasconstantly well above the ERH value From the infraredspectrum of a representative AS aerosol sample shownin Figure 1 (a neutralisation at T = 212 K) we canclearly identify the presence of condensed water due tothe appearance of the spectral shoulder at about 3400cmminus1 (Fig 1 left) attributable to the O-H stretchingregime of liquid water (CZICZO and ABBATT 1999)Figure 1 right shows an expanded view of the 2000ndash900 cmminus1 region As a result of the low signal-to-noiseratio it is futile to explore spectral details on the shapeof the sulphate extinction band at 1100 cmminus1 or the1420 cmminus1 ammonium peak as demonstrated by Hunget al (2002) to identify small amounts of crystalline(NH4)2SO4 Hence there is evidence that the AS aerosolmay have been deliquesced during the ice nucleation ex-periments but an uncertainty regarding the presence anexternally mixed aerosol containing small amounts ofcrystalline (NH4)2SO4 is left

The mineral dust aerosol particles are generated bydispersing a sample of Arizona Test Dust (Powder Tech-nology Inc USA) in a brush generator piping it througha dispersion nozzle and adding it to the chamber Thesize distribution of the added dust aerosol covered therange from 01 to 15 microm with a count median diameterof about 05 microm The number concentrations were about500 to 700 cmminus3 This dust aerosol is composed of amixture of different minerals mainly silicates calciteand clay minerals

In successive ice nucleation experiments at the sametemperature level Nptcl decreases because no furtheraerosol is added to the chamber The decrease is mainlycaused by the evacuations of the chamber during the icenucleation experiments and to a lesser extent due tocontinuous sedimentation of larger aerosol particles

23 Typical ice nucleation experiment

Figure 2 shows the most important parameters charac-terising a typical ice nucleation experiment in the AIDAchamber At the beginning of an experiment the pres-sure p and the gas temperature T are constant (sim1000hPa and sim225 K sim211 K sim200 K for the respectiveexperiments at different temperature levels) The partialpressure of water vapour egas (Fig 2 middle panel bluecurve) is controlled by the saturation vapour pressureover ice of the slightly colder ice coated walls esat ice wall


(Fig 2 middle panel black curve) With the start ofpumping (see Fig 2 upper panel for the time evolutionof pressure gas and wall temperature) egas starts to de-crease almost linearly with the decreasing total pressureDue to the expansion cooling the ice saturation pressureesat ice gas (Fig 2 middle panel green curve) steeply de-creases with decreasing gas temperature and thereforethe relative humidity with respect to ice RHice increases(Fig 2 middle panel orange curve) As soon as RHiceexceeds the critical ice saturation RHice nuc ice particlesbegin to form

Because the wall temperature remains almost con-stant during the expansion there is an increasing differ-ence between esat ice wall and egas causing a continuousflux of water vapour from the ice layer on the wall intothe gas phase After ice particles have formed they startto take up the excess water and grow as long as RHice ge100 (Fig 2 lower panel green stars) This addition-ally lowers egas (Fig 2 middle panel blue curve) andtherefore increases the water vapour flux from the wallice layers resulting in a marked increase of the total wa-ter signal (Fig 2 lower panel red curve) Therefore thefinal ice crystal size and the ice water content also de-pend on the amount of water evaporating from the icecoated chamber walls When pumping is stopped at 800hPa the gas temperature starts to increase RHice dropsbelow 100 and the ice crystals start to evaporate Thegas phase water increases due to the evaporating icecrystals and because of the still existing (but weakening)gradient from esat ice wall to egas The total water remainsnearly constant because the increase of water due to thewater vapour flux from the wall is partly compensatedby sedimentation of ice particles

The ice water content (IWC) is directly obtained bysubtracting the gas phase water vapour concentrationmeasured by the TDL absorption from the total waterconcentration measured by the FISH instrument (Fig2 lower panel black diamonds) The IWC is also re-trieved from the FTIR extinction spectra (Fig 2 lowerpanel red circles) For an explanation of this methodsee section 24 The IWC peaks around the end of thepumping period and decreases before RHice drops be-low 100 As the ice crystals are expected to growcontinuously while RHice is ge100 this inappropri-ate decrease of the IWC must be due to sampling lossesof larger ice crystals This effect is probably only signif-icant at higher temperatures when the ice crystals growto larger diameters due to more abundant water vapourThe ice particle sampling efficiency of the heated totalwater inlet is estimated to be 100 for diameters upto 7 microm and decreases for larger particles The result-ing overall accuracy for the IWC derived from the FISHand TDL measurements is about 10ndash15 In the caseof the homogeneous freezing experiments the total wa-ter measurements also include the liquid water content

of the solution aerosol droplets growing by water uptakedue to increasing relative humidity We estimate this liq-uid water fraction to range between 006 ppmv for theexperiments around 225 K and 004 ppmv around 200K which is well below our detection limit

The ice crystal number concentration reaches itsmaximum shortly after ice nucleation (Fig 2 lowerpanel orange triangles) Subsequently the number con-centration decreases continuously due to the ongoingpumping and ice particle sedimentation The accuracyof ice crystal number concentration and size retrievedfrom the FTIR spectra is discussed in section 24

All homogeneous ice nucleation experiments are per-formed at pumping rates (dpdt) of about -45 hPa minminus1During the heterogeneous ice nucleation experimentsdpdt was about ndash30 hPa minminus1 Note that during allAIDA experiments the pumping rate ndash and not the cool-ing rate ndash is controlled During the pumping period aheat flux from the warmer walls to the cooler gas phaseexists which decreases the cooling rate with time ofpumping Therefore the cooling rate at ice nucleation(dTdt)nuc may be different for each experiment

24 FTIR measurements

FTIR extinction spectra of the formed ice particles aremeasured in situ with a White-type multiple reflectioncell yielding a horizontal optical path up to 2543 m35 m above the bottom of the 7 m high AIDA vesselSpectra are recorded with a Bruker IFS 66v FTIR spec-trometer in the wave number range from v = 800 to 6000cmminus1 at a resolution of 4 cmminus1 and at a rate of 3 spectraper minute Figure 3 (left) demonstrates the suitabilityof FTIR extinction measurements to study the forma-tion and growth of ice crystals during AIDA expansioncooling experiments considering as an example an ac-tivation with mineral dust particles as ice nuclei (initialgas temperature at the beginning of this experiment was225 K) The series of FTIR spectra clearly reveals thesuccessive increase of the distinctive extinction bands ofice crystals (CLAPP et al 1995) Note that the character-istic infrared signatures of the provided ice nuclei couldbe neglected

By analysing the entire sequence of FTIR spectrathe temporal evolution of the number concentration andmean size of the ice crystals as well as the total IWC canbe retrieved Following the notation given by ARNOTT

et al (1997) the optical depth τ(vj) at a specific wavenumber v j is calculated using

τ(ν j) = LN

sumi=1

n(Di)σ(Di ν j) j = 1 middot middot middotM (21)

where L denotes the optical path length n(Dj) the num-ber concentration of ice crystals in a particular size binn(Dj) N the total number of ice crystal size bins and


Figure 3 Left Ice formation and growth monitored by in situ FTIR extinction spectroscopy during a typical AIDA expansion cooling

experiment Mineral dust aerosol particles served as ice nuclei The H2O(g) absorption bands were subtracted Right Comparison between

a measured extinction spectrum of ice crystals and retrieval results assuming different ice crystal shapes spheres (Mie fit) and cylinders

(T-matrix fit)

M the total number of wave numbers The size bin-averaged extinction cross section σ(Di ν j)

σ(Di ν j) =1

∆D

Di+ ∆D2

int

Diminus ∆D2

σ(D ν j)dD (22)

is calculated at M = 107 wave numbers between 6000and 800 cmminus1 for N = 135 individual size bins rangingfrom n(Dj) = 01 to 20 microm

In this section we want to briefly compare the re-trieval results based on two different approaches tocalculate the extinction cross sections σ(Di ν j) Firstas done in several recent laboratory studies (CLAPPet al 1995 HUNG and MARTIN 2002) we assumeMie theory to be valid to calculate the infrared extinc-tion cross sections of ice crystals Second we inves-tigate the influence of particle asphericity by applyingthe T-matrix code for randomly orientated ice cylinders(MISHCHENKO and TRAVIS 1998) adopted as surro-gates for hexagonal columns (LEE et al 2003) In thisapproach we choose an aspect ratio DL (D diameter Llength of the cylinder) of 07 (MITCHELL and ARNOTT1994) The low-temperature optical constants of waterice from RAJARAM et al (2001) in the near-infrared(6000ndash3700 cmminus1) and CLAPP et al (1995) in the mid-infrared region (3700ndash800 cmminus1) are employed in thecalculations The size distribution n(Di) of the ice crys-

tals is constrained to log-normally distributed particlesizes Using the downhill simplex method (PRESS et al1992) the ice crystal number density N as well as thecount median diameter CMD and the mode width σgof the log-normal size distribution are retrieved by min-imising the summed squared residuals between experi-mental and calculated spectra

Figure 3 (right) compares a measured extinctionspectrum of ice crystals selected from the series ofspectra shown on the left side with the infrared spec-tra calculated from the retrieved size distributions forice spheres and ice cylinders As already observed byLIU et al (1999) in similar calculations only small de-viations between the individual retrieval results occurboth calculated spectra agree nicely with the measuredextinction spectrum However there exist subtle differ-ences in the retrieved size distribution parameters Theretrieval based on Mie theory predicts N = 190 cmminus3 σg= 122 and CMD = 525 m whereas N = 149 cmminus3 σg= 110 and CMD (ie diameter of the volume equiv-alent sphere) = 597 microm are obtained when applyingthe T-matrix code The temporal evolution of retrievedice crystal size and number concentration for the com-plete AIDA expansion experiment is shown on the leftside of Figure 4 Generally the T-matrix approach pre-dicts ice crystal number densities which are about 20 lower than those retrieved by applying Mie theorywhereas the retrieved particle diameters are approxi-


Figure 4 FTIR-retrieved ice particle size distribution parameters for different assumptions on the ice crystal shape Left ice particle

number concentration and mean size Right ice water content

mately 10 larger compared to the calculations assum-ing ice spheres

The FTIR retrievals of ice crystal number concen-trations are nicely validated by simultaneous measure-ments with the optical particle spectrometers PCS2000and WELAS The ice crystal number concentrationsmeasured by these instruments differ by only 10 to 20 from the FTIR values Those of the PCS2000 re-veal a tendency towards an overestimation at higher tem-peratures and an underestimation at lower temperaturesThe values for the ice crystal number concentration andmean size presented in this paper refer to the results ofthe T-matrix calculation

Finally from the individual fit parameters N g andCMD as well as the density of ice (PRUPPACHER andKLETT 1997 Eq 3-2) the total ice water volume mix-ing ratio is calculated and displayed for the selected icenucleation experiment on the right side of Figure 4 Ob-viously in the size regime of ice crystals covered by ourstudy the effect of particle a-sphericity on the retrievedIWC is negligible (deviations below 1 ) Thereforethe analysis of the FTIR spectra should yield an accu-rate value for this quantity (relying on the accuracy ofthe published optical constants for water ice) which canbe directly compared to the IWC derived from the FISHand the TDL measurements (see Fig 2 lower panel)

3 Results and discussion

In the following two sections we present the results ofthe AIDA ice nucleation experiments Sulphuric acid

(SA) and ammonium sulphate (AS) solution dropletswere used during homogeneous freezing experimentswhereas Arizona Test Dust particles served as ice nucleiduring heterogeneous freezing experiments To com-pare the results of the experiments with different aerosoltypes we reduce the time dependent course of each ex-periment to characteristic data sets at certain points oftime At the onset time of ice nucleation (tnuc) we derivethe critical ice saturation and cooling rate (RHice nuc and(dTdt)nuc respectively) When the ice crystal numberconcentration has reached its maximum value (Nice abs)also the corresponding mean size (dice) is taken for com-parison The ice water content (IWC) values refer tothat time interval when the IWC has reached its max-imum value The IWC results are discussed in section31 The other microphysical parameters are shown insection 32

31 Ice water content (IWC)

The IWC derived from the difference between measure-ments of total and gas phase water (FISH-TDL) as wellas from Fourier transform infrared (FTIR) extinctionspectroscopy is shown in Table 1 and Figure 5 upperpanel In general there is good agreement between thetwo methods for measuring the IWC not only concern-ing the maximum values but also during the dynami-cal growth and evaporation of the ice crystals (Fig 2lower panel black diamonds and red circles) Howeverespecially at higher temperatures the values retrievedfrom the FISH-TDL measurements have a tendency toslightly underestimate the IWC compared to the FTIR


Table 1 Maximum ice water content (IWC) and corresponding temperature and pressure during AIDA ice nucleation experiments with

different aerosol types (SA = sulphuric acid AS = ammonium sulphate and ATD = Arizona Test Dust) values for IWCFISHminusTDL arearithmetic averages over 20 s and their respective standard deviations

_____________________________________________________________________

Type of IWC by IWC by T p

Aerosol FISHndashTDL FTIR

Experiment [ppmv] [ppmv] [K] [hPa]

_____________________________________________________________________

SA_1 156 012 135 1950 8093

SA_2 727 025 722 2053 8060

SA_3 2788 052 2674 2162 8013

AS_1 139 014 138 1943 8057

AS_2 759 049 884 2046 8028

AS_3 2774 078 3206 2161 8030

AS_4 3024 085 3180 2157 8013

ATD_1 354 017 430 2053 7990

ATD_2 2910 039 3496 2181 8002

ATD_3 1909 047 2458 2200 8014

_____________________________________________________________________

values As explained in section 23 this is probably dueto sampling losses of larger ice crystals

The maximum IWC increases with increasing tem-perature since there is more water vapour available forcrystal growth at higher temperature This increase wasnicely reproduced by both methods (see Fig 5 and Ta-ble 1) The differences in the IWC at comparable tem-peratures may be due to different ice particle total num-ber and surface area concentrations or habits of the icecrystals The larger the number of ice nuclei and thehigher the ice particle surface area concentration (iethe higher the ice nucleation efficiency) the faster isthe water vapour depletion of the gas phase and the ear-lier RHice decreases below 100 terminating the wa-ter vapour flux to the ice phase This mechanism couldhave additionally lowered the IWC for the experimentswith mineral dust aerosol (see also next section) Pro-cess models may be applied in future studies to furtherinvestigate the ice crystal growth during AIDA experi-ments

32 Microphysics

The parameters important to characterise the formationand life cycle of ice clouds namely the freezing onsetrelative humidity with respect to ice RHice nuc the cool-ing rate (dTdt)nuc the temperature Tnuc the pressurepnuc the total initial aerosol number concentration Nptclthe fraction of particles acting as ice nuclei Nice rel aswell as the maximum ice crystal number concentration

Nice abs and respective mean size dice of the ice crystalsare listed in Table 2 for all experiments

321 Freezing onset relative humidity (RHice nuc)

Analysing the values of RHice nuc for the homogeneousfreezing experiments with SA and AS aerosol it is ob-vious that the AS particles froze at a lower RHice nuc(120ndash127 ) than the supercooled SA solution droplets(144ndash166 ) The values of RHice nuc for SA aerosolare in very good agreement with the parameterisationof the homogeneous nucleation rate given by KOOP etal (2000) which was also found recently from anotherset of SA AIDA experiments (MOHLER et al 2003)

The lower values of RHice nuc for AS aerosol obvi-ously do not agree with the parameterisation of KOOP etal (2000) who stated that the nucleation rates of solu-tion droplets at the same temperature and water activityof the solute should be independent of the nature of thesolute That means SA and AS particles should havethe same RHice nuc as long as the particles are in thermo-dynamic equilibrium and approximately have the samesize (which is the case in our experiments) CZICZO andABBATT (1999) also measured homogeneous freezingthresholds of AS aerosol lower than predicted for thehomogeneous ice nucleation For the experiments dis-cussed here the analysis of FTIR spectra reveals that theAS particles consisted mainly of the liquid phase How-ever as also pointed out in detail in section 22 we cannot completely exclude from the FTIR spectra the ex-istence of an externally mixed aerosol containing small


Table 2 Parameters of AIDA ice nucleation experiments with different aerosol types (SA = sulphuric acid AS = ammonium sulphate ATD

= Arizona Test Dust) partly at the moment of ice nucleation tnuc Tnuc = gas temperature pnuc = total pressure dTdtnuc = cooling rate

wnuc = corresponding updraft velocities RHice nuc = ice saturation Nice rel = fraction of particles acting as ice nuclei Nice abs = maximum

ice crystal number concentration with corresponding count median diameter (dice) Nptcl = total initial aerosol number concentration

_____________________________________________________________________________________

Aerosol Tnuc pnuc dTdtnuc wnuc RHice_nuc

Experiment [K] [hPa] [K min-1

] [m s-1

] []

_______________________________________________________________________________ ______

SA_1 1955 9176 ndash210 36 1658

SA_2 2066 9020 ndash157 27 1511

SA_3 2196 9215 ndash240 41 1435

AS_1 1972 9564 ndash241 41 1265

AS_2 2090 9505 ndash252 43 1222

AS_3 2214 9641 ndash298 51 1199

AS_4 2209 9548 ndash263 45 1259

ATD_1 2087 9708 ndash131 22 1015

ATD_2 2223 9840 ndash178 30 1046

ATD_3 2240 9757 ndash174 30 1016

_____________________________________________________________________________________

Nice_abs dice Nptcl Nice_rel

[cm-3

] [microm] [cm-3

] []

_____________________________________________________________________________________

SA_1 230 167 10074 23

SA_2 83 346 11758 07

SA_3 51 470 13838 04

AS_1 520 113 2900 173

AS_2 236 322 8040 29

AS_3 78 712 8941 09

AS_4 70 717 6492 11

ATD_1 170 226 175 971

ATD_2 170 396 255 669

ATD_3 185 265 352 525

_____________________________________________________________________________________

amounts of crystalline (NH4)2SO4 Therefore the lowervalues of RHice nuc may also be explained by heteroge-neous effects Additionally one ice nucleation experi-ment was performed with crystalline AS added to theAIDA chamber at 224 K (corresponding FTIR-spectrumshown in Fig 1) First results indicate that ice crystalsalready occurred at a RHice slightly above 100 So ifthere were some externally mixed crystalline AS par-ticles present in the experiments discussed above we

should have observed the first ice crystals shortly afterRHice exceeded 100 Additionally the formation ofAS particles by in situ neutralisation of supercooled SAdroplets with ammonia clearly above the efflorescencehumidity supports the assumption that the AS particleshave been fully deliquesced during our experiments

The Arizona Test Dust particles were found to beeven more efficient ice nuclei than the AS particlesdiscussed above They froze heterogeneously at very


1

10

100IW

C [p

pmv]

FISHminusTDL FTIRSAASATD______________

0

100200300400500600

Nic

e_ab

s [c

mminus

3 ]

02

4

6

8

dic

e [micro

m]

195 200 205 210 215 220 225

temperature [K]

01

10

100

1000

Nic

e_re

l [

]

100

1000

10000

Np

tcl [

cmminus

3 ]

SA AS ATD__________

____

____

__

Figure 5 Microphysical parameters of AIDA ice clouds for differ-

ent aerosol types (SA = sulphuric acid (diamonds as symbol) AS

= ammonium sulphate (circles) ATD = Arizona Test Dust (trian-

gles) upper panel Maximum ice water content (IWC) filled sym-

bols denote IWC derived from FISH-TDL data empty symbols de-

note IWC retrieved from FTIR spectra middle panel maximum ice

crystal number concentrations (Nice abs red symbols) and the corre-

sponding median diameters (dice blue symbols) lower panel frac-

tions of particles acting as ice nuclei (Nice rel red symbols) and total

initial aerosol concentrations (Nptcl blue symbols) The dashed and

dotted lines are only to guide the eye

low values of RHice nuc (102ndash105 ) This is signifi-cantly below the homogeneous freezing thresholds andalso lower than RHice nuc measured for soot particles(MOHLER et al 2005 this issue) The ice crystals wereformed by deposition nucleation on the surface of thedry mineral dust particles

322 Number concentration and mean size of icecrystals (Nice abs dice)

The maximum number concentration and correspondingmean size of ice crystals measured during the homo-geneous and heterogeneous AIDA freezing experimentsare shown in Figure 5 middle panel red and blue sym-bols respectively

Homogeneous freezing Nice abs (Fig 5 middlepanel red circles and diamonds) increases with decreas-ing temperature for both SA and AS aerosol The icecrystal sizes are much smaller at lower temperatures(Fig 5 middle panel blue circles and diamonds) which

is due to the larger ice crystal number concentrationand the lower IWC (see above) The ice crystal num-ber concentrations agree well with the parameterisa-tion of cirrus cloud formation by homogeneous freez-ing developed by KARCHER and LOHMANN (2002a)based on the homogeneous freezing parameterisation ofKOOP et al (2000) Accordingly the number concen-tration of ice crystals is rather insensitive to the aerosolsize distribution but increases with decreasing temper-ature and increasing updraft velocity (ie higher cool-ing rates) which is confirmed by detailed process mod-elling studies Based on these simulations KARCHERand LOHMANN (2002ab) propose only a weak indirectaerosol effect on cirrus cloud properties The AIDA icecrystal number concentrations support this finding espe-cially with regard to the high and varying initial aerosolconcentrations at the AIDA experiments However anincreased occurrence of cirrus clouds due to the lowerfreezing thresholds may be observed when increasingthe fraction of AS particles in the atmosphere

In our SA experiments the cooling rates at ice nucle-ation (dTdt)nuc ranged from ndash16 to ndash24 K minminus1 (seeTable 2) which correspond to adiabatic cooling rates atupdraft velocities of about 27 to 41 m sminus1 At an up-draft velocity of 4 m sminus1 the cirrus parameterisationpredicts ice crystal number concentrations of about 400and 60 cmminus3 at freezing temperatures of 1964 and 216K respectively (see Fig 3 in KARCHER and LOHMANN2002a) The measured ice crystal number concentrationsare 230 and 51 cmminus3 at freezing temperatures of 1955and 2196 K respectively (see Table 2)

During the AS experiments the ice nucleation oc-curred earlier and therefore at somewhat higher coolingrates (dTdt)nuc between ndash24 and ndash30 K minminus1 Thesevalues correspond to updraft velocities between 41 and51 m sminus1 At an updraft velocity of 5 m sminus1 the cirrusparameterisation predicts ice crystal number concentra-tions of about 600 and 80 cmminus3 at freezing temperaturesof 1964 and 216 K respectively Therefore the higherice crystal number concentrations of 520 and 70 cmminus3measured for AS compared to SA aerosol may mainly beexplained by the higher cooling rates (dTdt)nuc As Nptclis lower for the AS than for the SA aerosol at comparabletemperatures the relative ice crystal number Nice rel ishigher for AS than for SA aerosol The good agreementof the AIDA results compared to the parameterisationof KARCHER and LOHMANN (2002a) again gives evi-dence that ice was nucleated by homogeneous freezingof our AS particles rather than by heterogeneous ice nu-cleation As will be discussed below heterogeneous icenucleation may produce at least much higher fractionsof ice crystals with respect to the total aerosol concen-tration

It should be mentioned that the experiments dis-cussed here have been made at relatively high cooling


rates (ie high corresponding vertical velocities) wherethe detailed process modelling gives somewhat higherice crystal number concentrations than the cirrus param-eterisation of KARCHER and LOHMANN (2002a) andwhere the ice crystal number concentration gets moredependent on the aerosol size distribution Thereforeadditional process studies using the measured AS andSA size distributions would be helpful to further provethe reliability of the cirrus parameterisation at high up-draft velocities

Heterogeneous freezing Arizona Test Dust (ATD)particles were not only more efficient ice nuclei withrespect to RHice nuc compared to SA and AS solutiondroplets (see section 321) but also with respect tothe fraction of Nice rel of aerosol particles nucleatingice Nice rel (Fig 5 lower panel red triangles) wasabout one order of magnitude higher even at lower val-ues of (dTdt)nuc compared to the AS and SA exper-iments In contrast to homogeneous freezing there isno clear dependency of Nice abs on temperature (Fig5 middle panel red triangles) Note however that atthe lower temperature Nice abs was limited by the totalaerosol number concentration Nptcl At the higher tem-peratures Nice abs was markedly higher than for SA andAS aerosol The higher ice crystal number concentra-tions indicate that there is at least for ATD particles amuch stronger dependency of the deposition nucleationrate on the relative humidity compared to the homoge-neous freezing mechanism In other words certain min-erals may nucleate ice in a very narrow band of relativehumidity This would imply that in the atmosphere thenumber of ice crystals formed on mineral dust particlesis almost independent of the temperature or cooling rateand mainly limited by the number concentration of themineral particles

Altogether mineral dust particles seem to be very ef-ficient ice nuclei and therefore may have a significant ef-fect on the number concentration size and habit of icecrystals Thus the upper tropospheric aerosol may indi-rectly affect the climate by changing the radiative prop-erties of cirrus clouds A strong indirect aerosol effect isalready found by KARCHER and LOHMANN (2003) atintermediate updraft velocities (ie cooling rates) whenadding immersion freezing of heterogeneous ice nucleito their microphysical model of homogeneous freezing(KARCHER and LOHMANN 2002ab)

4 Summary

During dynamic expansion ice nucleation experimentsstarted at temperatures between 224 and 196 K in theaerosol chamber AIDA ice clouds were formed by icenucleation processes of sulphuric acid (SA) ammoniumsulphate (AS) and mineral dust (Arizona Test DustATD) aerosol The formation and properties of the ice

clouds were comprehensively analysed with respect tothe ice water content (IWC) the freezing onset relativehumidity with respect to ice (RHice nuc) as well as themaximum number concentration of ice crystals (Nice abs)and their corresponding mean size (dice) The numberconcentration and mean size of the ice crystals as well asthe IWC were retrieved from Fourier transform infrared(FTIR) extinction spectroscopy The IWC was also di-rectly obtained from the difference of independent totaland gas phase water measurements

(i) Ice water content There is very good agreementbetween the IWC data derived from the difference be-tween measurements of total water and interstitial wa-ter vapour and those retrieved from the FTIR spectraLower IWC at lower temperatures mainly reflects thedecreasing ice saturation pressure with decreasing tem-perature Slight differences between the different aerosoltypes at the same temperature could qualitatively beexplained by different ice surface area concentrationsor different amounts of water evaporating from the icecoated chamber walls during the experiments

(ii) Homogeneous freezing experiments For SAaerosol the RHice nuc values between 144 and 166 measured at temperatures between 220 and 196 K agreevery well with previous AIDA results (MOHLER et al2003) For AS particles significantly lower values ofRHice nuc between 120 and 127 were measured in thesame temperature range thus confirming the results ofCZICZO and ABBATT (1999) Because the FTIR analy-sis reveals that the AS particles consisted mainly of theliquid phase this seems to contradict the activity basedparameterisation for homogeneous freezing of solutions(KOOP et al 2000) The formation of AS particles byin situ neutralisation of supercooled SA droplets withammonia clearly above the efflorescence relative humid-ity supports the assumption that the AS particles havebeen fully deliquesced during our experiments How-ever we can not completely exclude from the FTIR anal-ysis the existence of a minor volume fraction of solidcrystals inside the AS particles The number concentra-tion of ice crystals formed during the SA and AS experi-ments agree well with the parameterisation developed byKARCHER and LOHMANN (2002ab) for the formationof ice crystals in cirrus clouds by homogeneous freez-ing The higher ice crystal number concentrations at theAS experiments can be explained by the higher coolingrates at the onset of freezing

(iii) Heterogeneous freezing experiments ArizonaTest Dust mineral particles nucleated ice by deposi-tion freezing at relative humidities only slightly aboveice saturation clearly below the freezing thresholds forthe homogeneous freezing mechanism and the heteroge-neous ice nucleation of soot particles (MOHLER et al2005 this issue) The mineral dust particles have alsobeen most efficient with respect to the ice crystal num-


ber concentration and the fraction of particles nucleatingice at comparable temperatures This gives evidence thatthere may be a much stronger dependency of the nucle-ation rate on the relative humidity for the deposition icenucleation on mineral particles compared to the homo-geneous freezing mechanism This could have importantimplications for the parameterisation of heterogeneousice nucleation processes in atmospheric models

Further process modelling and AIDA ice nucleationstudies are planned to elucidate the relationship betweenthe formation life cycle and climatologically relevantoptical properties of cirrus clouds with basic aerosolproperties and microphysical processes

Acknowledgements

We gratefully acknowledge the continuous support andtechnical assistance by all staff members during theAIDA ice nucleation experiments especially C LINKES BUTTNER O STETZER and M SCHNAITER Wehighly appreciate the collaboration of H TEICHERTwith the TDL instrumental setup and data retrieval Thiswork contributes to the HGF project ldquoParticles and Cir-rus Clouds (PAZI)rdquo

References

ARNOTT W P C SCHMITT Y LIU J HALLETT 1997Droplet size spectra and water-vapor concentration of labo-ratory water clouds inversion of Fourier transform infrared(500-5000 cmminus1) optical-depth measurement ndash Appl Opt36 5205ndash5216

CLAPP M L R E MILLER D R WORSNOP 1995Frequency-dependent optical constants of water ice ob-tained directly from aerosol extinction spectra ndash J PhysChem 99 6317ndash6326

COLBERG CA BP LUO H WERNLI T KOOP TH PE-TER 2003 A novel model to predict the physical state ofatmospheric H2SO4NH3H2O aerosol particles ndash AtmosChem Phys 3 909ndash924

CZICZO DJ JPD ABBATT 1999 Deliquescence efflo-rescence and supercooling of ammonium sulfate aerosolsat low temperatures Implications for cirrus cloud formationand aerosol phase in the atmosphere ndash J Geophys Res At-mos 104 13781ndash13790

DEMOTT PJ Y CHEN SM KREIDENWEIS DCROGERS DE SHERMAN 1999 Ice formation by blackcarbon particles ndash Geophys Res Lett 26 2429ndash2432

DEMOTT PJ K SASSEN MR POELLOT D BAUM-GARDNER DC ROGERS S BROOKS AJ PRENNISM KREIDENWEIS 2003 African dust aerosols as atmo-spheric ice nuclei ndash Geophys Res Lett 30(14) 1732 DOI1010292003GL017410

EBERT V H TEICHERT C GIESEMANN U HSAATHOFF SCHURATH 2004 Fibre-coupled in situ laserabsorption spectrometer for the selective detection of watervapour traces down to the ppb-level accepted for publica-tion ndash In Proceedings of 4th Conference on Applicationsand Trends in Optical Analysis Technology 7ndash8 Oct 2004Dusseldorf (in german)

FIELD PR RJ COTTON K NOONE P GLANTZPH KAYE E HIRST RS GREENAWAY C JOST RGABRIEL T REINER M ANDREAE CPR SAUNDERSA ARCHER T CHOULARTON M SMITH B BROOKSC HOELL B BANDY D JOHNSON A HEYMSFIELD2001 Ice nucleation in orographic wave clouds Measure-ments made during INTACC ndash Quart J Roy Meteor Soc127 1493ndash1512

GIERENS K 2003 On the transition between hetereoge-neous and homogeneous freezing ndash Atmos Chem Phys3 437ndash446

GURLIT W JP BURROWS R ZIMMERMANN U PLATTC GIESEMANN J WOLFRUM V EBERT 2005 Light-weight diode laser spectrometer ldquoCHILDrdquo for balloon-borne measurements of water vapor and methane ndash Ap-plied Optics 44(1) 91ndash102

HAAG W B KARCHER S SCHAEFERS O STETZER OMOHLER U SCHURATH M KRAMER C SCHILLER2003a Numerical simulations of homogeneous freezingprocesses in the aerosol chamber AIDA ndash Atmos ChemPhys 3 195ndash210

HAAG W B KARCHER J STROM A MINIKIN ULOHMANN J OVARLEZ A STOHL 2003b Freezingthresholds and cirrus cloud formation mechanisms inferredfrom in situ measurements of relative humidity ndash AtmosChem Phys 3 1791ndash1806

HEYMSFIELD AJ LM MILOSHEVICH 1993 Homoge-neous ice nucleation and supercooled liquid water in oro-graphic wave clouds ndash J Atmos Sci 50 2335ndash2353

HUNG H-M S T MARTIN 2002 Infrared spectroscopicevidence for the ice formationmechanisms active in aerosolflow tubes ndash Appl Spectrosc 56 1067ndash1081

HUNG H-M A MALINOWSKI ST MARTIN 2002 Icenucleation kinetics of aerosols containing aqueous and solidammonium sulfate particles J Phys Chem A 106 293ndash306

mdash mdash mdash 2003 Kinetics of heterogeneous ice nucleationon the surfaces of mineral dust cores inserted into aqueousammonium sulfate particles J Phys Chem A 107 1296ndash1306

JENSEN EJ OB TOON A TABAZADEH GWSACHSE BE ANDERSON KR CHAN CW TWOHYB GANDRUD SM AULENBACH A HEYMSFIELD JHALLETT B GARY 1998 Ice nucleation processes in up-per tropospheric wave-clouds observed during SUCCESSndash Geophys Res Lett 25 1363ndash1366

KARCHER B U LOHMANN 2002a A parameterizationof cirrus cloud formation Homogeneous freezing of su-percooled aerosols ndash J Geophys Res 107(D2) 4010doi1010292001JD000470

mdash mdash 2002b A parameterization of cirrus cloud for-mation Homogeneous freezing including effects ofaerosol size ndash J Geophys Res 107(D23) 4698DOI1010292001JD001429

mdash mdash 2003 A parameterization of cirrus cloud forma-tion Heterogeneous freezing ndash J Geophys Res 108(D14)4402 DOI1010292002JD003220

KOOP T B LUO A TSIAS T PETER 2000 Water aci-tivity as the determinant for homogeneous ice nucleation inaqueous solutions ndash Nature 406 611ndash614

LEE Y-K P YANG MI MISHCHENKO BA BAUMYX HU H-L HUANG WJ WISCOMBE AJ BARAN


2003 Use of circular cylinders as surrogates for hexago-nal pristine ice crystals in scattering calculations at infraredwavelengths ndash Appl Opt 42 2653ndash2664

LIU Y WP ARNOTT J HALLETT 1999 Particle size dis-tribution retrieval from multispectral optical depth Influ-ences of particle nonsphericity and refractive index ndash JGeophys Res Atmos 104 31753ndash31762

MARTI J K MAUERSBERGER 1993 A survey and newmeasurements of ice vapour pressure at temperatures be-tween 170 K and 250 K ndash Geophys Res Lett 20 363ndash366

MARTIN ST 2000 Phase transitions of aqueous atmo-spheric particles ndash Chem Rev 100 3403ndash3453

MISHCHENKO MI LD TRAVIS 1998 Capabilities andlimitations of a current Fortran implementation of the T-Matrix method for randomly oriented rotationally symmet-ric scatterers ndash J Quant Spectroscop Radiat Transfer 60309ndash324

MITCHELL DL WP ARNOTT 1994 A model predictingthe evolution of ice particle size spectra and radiative prop-erties of cirrus clouds Part II Dependence of absorptionand extinction on ice crystal morphology ndash J Atmos Sci51 817ndash832

MOHLER O O STETZER S SCHAEFERS C LINKEM SCHNAITER R TIEDE H SAATHOFF M KRAMERA MANGOLD P BUDZ P ZINK J SCHREINER KMAUERSBERGER W HAAG B KARCHER U SCHU-RATH 2003 Experimental investigation of homogeneousfreezing of sulphuric acid particles in the aerosol chamberAIDA ndash Atmos Chem Phys 3 211ndash223

MOHLER O C LINKE H SAATHOFF M SCHNAITERR WAGNER A MANGOLD M KRAMER U SCHU-RATH 2005 Ice nucleation on flame soot aerosol of dif-ferent organic carbon content ndash Meteorol Z 14 477ndash484

OVARLEZ J J-F GAYET K GIERENS J STROM HOVARLEZ F AURIOL R BUSEN U SCHUMANN 2002Water vapour measurements inside cirrus clouds in North-ern and Southern hemispheres during INCA ndash GeophysRes Lett 29(16) 1813 Doi1010292001GL014440

PRESS WH SA TEUKOLSKY WT VETTERLING BPFLANNERY 1992 Numerical recipes in C The art of scien-tific computing ndash Cambridge University Press CambridgeNew York Port Chester Melbourne Sidney 994 pp

PRUPPACHER HR JD KLETT 1997 Microphysics ofclouds and precipitation ndash Kluwer Acad Pub Dordrecht980 pp

RAJARAM B DL GLANDORF DB CURTIS MATOLBERT OB TOON N OCKMAN 2001 Temperature-dependent optical constants of water ice in the near in-frared new results and critical review of the available mea-surements ndash Appl Opt 40 4449ndash4462

SCHLOSSER HE J WOLFRUM BA WILLIAMS RSSHEINSON JW FLEMING V EBERT 2002 In situ deter-mination of molecular oxygen concentrations in full-scalefire suppression tests using TDLAS ndash Proc Comb Inst29 353ndash360

SEIFERT M R TIEDE M SCHNAITER C LINKE OMOHLER U SCHURATH J STROM 2004 Operation andperformance of a differential mobility particle sizer anda TSI 3010 condensation particle counter at stratospherictemperatures and pressures ndash J Aerosol Sci 35 981ndash993

TEICHERT H T FERNHOLZ V EBERT 2003 In situ mea-surement of CO H2O and gas temperature in a lignite-firedpower-plant ndash Appl Opt 42 2043ndash2051

ZOGER M A AFCHINE N EICKE M-T GERHARDS EKLEIN DS MCKENNA U MORSCHEL U SCHMIDTV TAN F TUITJER T WOYKE C SCHILLER 1999Fast in situ stratospheric hygrometers A new family ofballoon-borne and airborne Lyman- photofragment fluo-rescence hygrometers ndash J Geophys Res 104(D1) 1807ndash1816

ZUBERI B AK BERTRAM T KOOP LT MOLINAMJ MOLINA 2001 Heterogeneous freezing of aqueousparticles induced by crystallized (NH4)2SO4-H2O ice andletovicite ndash J Phys Chem A 105 6458ndash6464

ZUBERI B AK BERTRAM CA CASSA LT MOLINAMJ MOLINA 2002 Heterogeneous nucleation of ice in(NH4)2SO4-H2O particles with mineral dust immersions ndashGeophys Res Lett 29(10) 1010292001GL014289


HEYMSFIELD and MILOSHEVICH 1993 JENSEN et al1998 MARTIN 2000 KOOP et al 2000 FIELD et al2001 MOHLER et al 2003) Based on laboratory ex-periments KOOP et al (2000) provided a water activitybased parameterisation for the homogeneous freezing ofsupercooled solution droplets which was confirmed bydetailed measurements in ice clouds formed by homoge-neous freezing of sulphuric acid solution droplets at theaerosol chamber AIDA of Forschungszentrum Karls-ruhe (MOHLER et al 2003 HAAG et al 2003a) Us-ing the freezing nucleation rates of KOOP et al (2000)KARCHER and LOHMANN (2002ab) developed a mi-crophysical parameterisation of the homogeneous freez-ing of supercooled solution droplets showing that theice crystal number density in cirrus clouds can mainly bedescribed as function of updraft velocity (ie the coolingrate) and temperature They also stated that enhancednumber concentrations of sulphate aerosols occurringeg after volcanic eruptions would not influence sig-nificantly the process of cirrus formation and thereforethe relationship between aerosol and ice particle numberconcentrations in homogeneously formed cirrus cloudswould be much weaker than in liquid water clouds

Heterogeneous freezing of particles containing an in-soluble inclusion occurs at ice supersaturations signif-icantly below the pure homogeneous freezing thresh-old HAAG et al (2003b) showed based on in situmeasurements of RHice in and outside of cirrus clouds(OVARLEZ et al 2002) that in parts of the pollutednorthern hemisphere cirrus clouds are formed via theheterogeneous pathway but that homogeneous ice nu-clei are also involved There are several laboratory andfield studies investigating the heterogeneous freezing ofwater on soot coated with sulphuric acid (DEMOTT etal 1999) mineral dust (DEMOTT et al 2003) mineraldust immersed in aqueous ammonium sulphate particles(ZUBERI et al 2002) crystallised ammonium sulphate(ZUBERI et al 2001) as well as the ice nucleation onflame soot aerosol of different organic carbon content(MOHLER et al 2005 this issue) These studies showthe temperature and RHice dependence of the freezingprocess for the various types of aerosol particles HUNG

et al (2003) found that the heterogeneous ice nucleationrates of ammonium sulphate solution droplets with amineral dust core depend beside temperature and ice su-persaturation on the diameter of the immersed mineralparticle KARCHER and LOHMANN (2003) provided amicrophysical parameterisation of the heterogeneous icenucleation in the immersion mode For intermediate up-draft velocities they found a strong aerosol effect oncirrus properties probably also affecting the radiativeproperties of cirrus clouds and thereby indirectly affect-ing the climate The presence of two types of ice nucleiduring ice formation and growth differing distinctly bytheir freezing thresholds would lead to a lower ice crys-

tal concentration This is due to competition of the twotypes of ice nuclei for the available water vapour duringthe growth phase

Up to now there are only few field studies exam-ining the factors controlling the ice nucleation pro-cesses and the resulting frequency of cirrus cloud oc-currence and their optical properties From the INCAfield campaign in the northern and southern hemi-sphere in 2000 a first insight into the ice formation pro-cesses is gained (OVARLEZ et al 2002 HAAG et al2003b) The CRYSTAL-FACE field campaign in 2002provided a comprehensive dataset for the characterisa-tion and description of the cirrus cloud formation (seehttpcloud1arcnasagovcrystalfaceindexhtml) Butadditional investigations of the relationship betweenaerosols and cirrus clouds especially with respect to icenucleation and crystal properties are of high interest

Here we present results of detailed laboratory mea-surements of homogeneous and heterogeneous ice nu-cleation performed at the aerosol chamber AIDA un-der simulated atmospheric cirrus conditions The ex-periments were conducted at temperatures between 224and 196 K using FTIR extinction spectroscopy to deter-mine microphysical parameters of the ice clouds Sul-phuric acid and ammonium sulphate solution dropletswere used during homogeneous ice nucleation experi-ments while mineral dust particles (Arizona Test Dust)served as nuclei for heterogeneous ice formation Inthis paper we first focus on describing our experimentalmethods Thereafter we show the influence of tempera-ture and cooling rate on characteristic parameters of theformed ice clouds especially the ice water content theice crystal number concentration and mean crystal sizefor the different types of aerosols Differences betweenthe freezing relative humidities of the different aerosoltypes are also discussed

2 Experimental

21 AIDA Aerosol and cloud chamberfacility

The AIDA aerosol chamber is a large vessel of 84 m3

volume which can homogeneously be cooled down to183 K The pressure range covers 01 to 1000 hPa Un-der constant wall and gas temperature conditions icesaturation is maintained by a thin ice layer on the cham-ber walls The ice supersaturation necessary for the ho-mogeneous or heterogeneous nucleation of ice crystalsis achieved by lsquovolume expansionrsquo due to controlledpumping usually from 1000 to 800 hPa During pump-ing the relative humidity with respect to ice (RHice) in-creases by up to 50 minminus1

The onset of ice formation is precisely detected bymeasuring the increasing intensity and depolarisation of















2120

2155

2190

2225

2260

T [

K]

800

850

900

950

1000

p [h

Pa]

T_gas

T_wall

pressure

012345

H2O

[P

a]

8090100110120130140150

RH

ice

[]


120000 120500 121000 1215000

10

20

30

40

50

024681012

dic

e [micro

m]

0

20

40

60

N_i

ce_a

bs [c

mminus

3 ]

time CEST

H2O

[ppm

v]





























τ(ν j) = LN

sumi=1







(T-matrix fit)


σ(Di ν j) =1

∆D

Di+ ∆D2

int

Diminus ∆D2

σ(D ν j)dD (22)



















_____________________________________________________________________




_____________________________________________________________________

SA_1 156 012 135 1950 8093

SA_2 727 025 722 2053 8060

SA_3 2788 052 2674 2162 8013

AS_1 139 014 138 1943 8057

AS_2 759 049 884 2046 8028

AS_3 2774 078 3206 2161 8030

AS_4 3024 085 3180 2157 8013

ATD_1 354 017 430 2053 7990

ATD_2 2910 039 3496 2181 8002

ATD_3 1909 047 2458 2200 8014

_____________________________________________________________________



32 Microphysics











_____________________________________________________________________________________



] [m s-1

] []

_______________________________________________________________________________ ______

SA_1 1955 9176 ndash210 36 1658

SA_2 2066 9020 ndash157 27 1511

SA_3 2196 9215 ndash240 41 1435

AS_1 1972 9564 ndash241 41 1265

AS_2 2090 9505 ndash252 43 1222

AS_3 2214 9641 ndash298 51 1199

AS_4 2209 9548 ndash263 45 1259

ATD_1 2087 9708 ndash131 22 1015

ATD_2 2223 9840 ndash178 30 1046

ATD_3 2240 9757 ndash174 30 1016

_____________________________________________________________________________________


[cm-3

] [microm] [cm-3

] []

_____________________________________________________________________________________

SA_1 230 167 10074 23

SA_2 83 346 11758 07

SA_3 51 470 13838 04

AS_1 520 113 2900 173

AS_2 236 322 8040 29

AS_3 78 712 8941 09

AS_4 70 717 6492 11

ATD_1 170 226 175 971

ATD_2 170 396 255 669

ATD_3 185 265 352 525

_____________________________________________________________________________________





1

10

100IW

C [p

pmv]


0

100200300400500600

Nic

e_ab

s [c

mminus

3 ]

02

4

6

8

dic

e [micro

m]

195 200 205 210 215 220 225

temperature [K]

01

10

100

1000

Nic

e_re

l [

]

100

1000

10000

Np

tcl [

cmminus

3 ]

SA AS ATD__________

____

____

__
























4 Summary









Acknowledgements


References
























































2120

2155

2190

2225

2260

T [

K]

800

850

900

950

1000

p [h

Pa]

T_gas

T_wall

pressure

012345

H2O

[P

a]

8090100110120130140150

RH

ice

[]


120000 120500 121000 1215000

10

20

30

40

50

024681012

dic

e [micro

m]

0

20

40

60

N_i

ce_a

bs [c

mminus

3 ]

time CEST

H2O

[ppm

v]





























τ(ν j) = LN

sumi=1







(T-matrix fit)


σ(Di ν j) =1

∆D

Di+ ∆D2

int

Diminus ∆D2

σ(D ν j)dD (22)



















_____________________________________________________________________




_____________________________________________________________________

SA_1 156 012 135 1950 8093

SA_2 727 025 722 2053 8060

SA_3 2788 052 2674 2162 8013

AS_1 139 014 138 1943 8057

AS_2 759 049 884 2046 8028

AS_3 2774 078 3206 2161 8030

AS_4 3024 085 3180 2157 8013

ATD_1 354 017 430 2053 7990

ATD_2 2910 039 3496 2181 8002

ATD_3 1909 047 2458 2200 8014

_____________________________________________________________________



32 Microphysics











_____________________________________________________________________________________



] [m s-1

] []

_______________________________________________________________________________ ______

SA_1 1955 9176 ndash210 36 1658

SA_2 2066 9020 ndash157 27 1511

SA_3 2196 9215 ndash240 41 1435

AS_1 1972 9564 ndash241 41 1265

AS_2 2090 9505 ndash252 43 1222

AS_3 2214 9641 ndash298 51 1199

AS_4 2209 9548 ndash263 45 1259

ATD_1 2087 9708 ndash131 22 1015

ATD_2 2223 9840 ndash178 30 1046

ATD_3 2240 9757 ndash174 30 1016

_____________________________________________________________________________________


[cm-3

] [microm] [cm-3

] []

_____________________________________________________________________________________

SA_1 230 167 10074 23

SA_2 83 346 11758 07

SA_3 51 470 13838 04

AS_1 520 113 2900 173

AS_2 236 322 8040 29

AS_3 78 712 8941 09

AS_4 70 717 6492 11

ATD_1 170 226 175 971

ATD_2 170 396 255 669

ATD_3 185 265 352 525

_____________________________________________________________________________________





1

10

100IW

C [p

pmv]


0

100200300400500600

Nic

e_ab

s [c

mminus

3 ]

02

4

6

8

dic

e [micro

m]

195 200 205 210 215 220 225

temperature [K]

01

10

100

1000

Nic

e_re

l [

]

100

1000

10000

Np

tcl [

cmminus

3 ]

SA AS ATD__________

____

____

__
























4 Summary









Acknowledgements


References





















































τ(ν j) = LN

sumi=1







(T-matrix fit)


σ(Di ν j) =1

∆D

Di+ ∆D2

int

Diminus ∆D2

σ(D ν j)dD (22)



















_____________________________________________________________________




_____________________________________________________________________

SA_1 156 012 135 1950 8093

SA_2 727 025 722 2053 8060

SA_3 2788 052 2674 2162 8013

AS_1 139 014 138 1943 8057

AS_2 759 049 884 2046 8028

AS_3 2774 078 3206 2161 8030

AS_4 3024 085 3180 2157 8013

ATD_1 354 017 430 2053 7990

ATD_2 2910 039 3496 2181 8002

ATD_3 1909 047 2458 2200 8014

_____________________________________________________________________



32 Microphysics











_____________________________________________________________________________________



] [m s-1

] []

_______________________________________________________________________________ ______

SA_1 1955 9176 ndash210 36 1658

SA_2 2066 9020 ndash157 27 1511

SA_3 2196 9215 ndash240 41 1435

AS_1 1972 9564 ndash241 41 1265

AS_2 2090 9505 ndash252 43 1222

AS_3 2214 9641 ndash298 51 1199

AS_4 2209 9548 ndash263 45 1259

ATD_1 2087 9708 ndash131 22 1015

ATD_2 2223 9840 ndash178 30 1046

ATD_3 2240 9757 ndash174 30 1016

_____________________________________________________________________________________


[cm-3

] [microm] [cm-3

] []

_____________________________________________________________________________________

SA_1 230 167 10074 23

SA_2 83 346 11758 07

SA_3 51 470 13838 04

AS_1 520 113 2900 173

AS_2 236 322 8040 29

AS_3 78 712 8941 09

AS_4 70 717 6492 11

ATD_1 170 226 175 971

ATD_2 170 396 255 669

ATD_3 185 265 352 525

_____________________________________________________________________________________





1

10

100IW

C [p

pmv]


0

100200300400500600

Nic

e_ab

s [c

mminus

3 ]

02

4

6

8

dic

e [micro

m]

195 200 205 210 215 220 225

temperature [K]

01

10

100

1000

Nic

e_re

l [

]

100

1000

10000

Np

tcl [

cmminus

3 ]

SA AS ATD__________

____

____

__
























4 Summary









Acknowledgements


References














































(T-matrix fit)


σ(Di ν j) =1

∆D

Di+ ∆D2

int

Diminus ∆D2

σ(D ν j)dD (22)



















_____________________________________________________________________




_____________________________________________________________________

SA_1 156 012 135 1950 8093

SA_2 727 025 722 2053 8060

SA_3 2788 052 2674 2162 8013

AS_1 139 014 138 1943 8057

AS_2 759 049 884 2046 8028

AS_3 2774 078 3206 2161 8030

AS_4 3024 085 3180 2157 8013

ATD_1 354 017 430 2053 7990

ATD_2 2910 039 3496 2181 8002

ATD_3 1909 047 2458 2200 8014

_____________________________________________________________________



32 Microphysics











_____________________________________________________________________________________



] [m s-1

] []

_______________________________________________________________________________ ______

SA_1 1955 9176 ndash210 36 1658

SA_2 2066 9020 ndash157 27 1511

SA_3 2196 9215 ndash240 41 1435

AS_1 1972 9564 ndash241 41 1265

AS_2 2090 9505 ndash252 43 1222

AS_3 2214 9641 ndash298 51 1199

AS_4 2209 9548 ndash263 45 1259

ATD_1 2087 9708 ndash131 22 1015

ATD_2 2223 9840 ndash178 30 1046

ATD_3 2240 9757 ndash174 30 1016

_____________________________________________________________________________________


[cm-3

] [microm] [cm-3

] []

_____________________________________________________________________________________

SA_1 230 167 10074 23

SA_2 83 346 11758 07

SA_3 51 470 13838 04

AS_1 520 113 2900 173

AS_2 236 322 8040 29

AS_3 78 712 8941 09

AS_4 70 717 6492 11

ATD_1 170 226 175 971

ATD_2 170 396 255 669

ATD_3 185 265 352 525

_____________________________________________________________________________________





1

10

100IW

C [p

pmv]


0

100200300400500600

Nic

e_ab

s [c

mminus

3 ]

02

4

6

8

dic

e [micro

m]

195 200 205 210 215 220 225

temperature [K]

01

10

100

1000

Nic

e_re

l [

]

100

1000

10000

Np

tcl [

cmminus

3 ]

SA AS ATD__________

____

____

__
























4 Summary









Acknowledgements


References
























































_____________________________________________________________________




_____________________________________________________________________

SA_1 156 012 135 1950 8093

SA_2 727 025 722 2053 8060

SA_3 2788 052 2674 2162 8013

AS_1 139 014 138 1943 8057

AS_2 759 049 884 2046 8028

AS_3 2774 078 3206 2161 8030

AS_4 3024 085 3180 2157 8013

ATD_1 354 017 430 2053 7990

ATD_2 2910 039 3496 2181 8002

ATD_3 1909 047 2458 2200 8014

_____________________________________________________________________



32 Microphysics











_____________________________________________________________________________________



] [m s-1

] []

_______________________________________________________________________________ ______

SA_1 1955 9176 ndash210 36 1658

SA_2 2066 9020 ndash157 27 1511

SA_3 2196 9215 ndash240 41 1435

AS_1 1972 9564 ndash241 41 1265

AS_2 2090 9505 ndash252 43 1222

AS_3 2214 9641 ndash298 51 1199

AS_4 2209 9548 ndash263 45 1259

ATD_1 2087 9708 ndash131 22 1015

ATD_2 2223 9840 ndash178 30 1046

ATD_3 2240 9757 ndash174 30 1016

_____________________________________________________________________________________


[cm-3

] [microm] [cm-3

] []

_____________________________________________________________________________________

SA_1 230 167 10074 23

SA_2 83 346 11758 07

SA_3 51 470 13838 04

AS_1 520 113 2900 173

AS_2 236 322 8040 29

AS_3 78 712 8941 09

AS_4 70 717 6492 11

ATD_1 170 226 175 971

ATD_2 170 396 255 669

ATD_3 185 265 352 525

_____________________________________________________________________________________





1

10

100IW

C [p

pmv]


0

100200300400500600

Nic

e_ab

s [c

mminus

3 ]

02

4

6

8

dic

e [micro

m]

195 200 205 210 215 220 225

temperature [K]

01

10

100

1000

Nic

e_re

l [

]

100

1000

10000

Np

tcl [

cmminus

3 ]

SA AS ATD__________

____

____

__
























4 Summary









Acknowledgements


References















































_____________________________________________________________________________________



] [m s-1

] []

_______________________________________________________________________________ ______

SA_1 1955 9176 ndash210 36 1658

SA_2 2066 9020 ndash157 27 1511

SA_3 2196 9215 ndash240 41 1435

AS_1 1972 9564 ndash241 41 1265

AS_2 2090 9505 ndash252 43 1222

AS_3 2214 9641 ndash298 51 1199

AS_4 2209 9548 ndash263 45 1259

ATD_1 2087 9708 ndash131 22 1015

ATD_2 2223 9840 ndash178 30 1046

ATD_3 2240 9757 ndash174 30 1016

_____________________________________________________________________________________


[cm-3

] [microm] [cm-3

] []

_____________________________________________________________________________________

SA_1 230 167 10074 23

SA_2 83 346 11758 07

SA_3 51 470 13838 04

AS_1 520 113 2900 173

AS_2 236 322 8040 29

AS_3 78 712 8941 09

AS_4 70 717 6492 11

ATD_1 170 226 175 971

ATD_2 170 396 255 669

ATD_3 185 265 352 525

_____________________________________________________________________________________





1

10

100IW

C [p

pmv]


0

100200300400500600

Nic

e_ab

s [c

mminus

3 ]

02

4

6

8

dic

e [micro

m]

195 200 205 210 215 220 225

temperature [K]

01

10

100

1000

Nic

e_re

l [

]

100

1000

10000

Np

tcl [

cmminus

3 ]

SA AS ATD__________

____

____

__
























4 Summary









Acknowledgements


References











































1

10

100IW

C [p

pmv]


0

100200300400500600

Nic

e_ab

s [c

mminus

3 ]

02

4

6

8

dic

e [micro

m]

195 200 205 210 215 220 225

temperature [K]

01

10

100

1000

Nic

e_re

l [

]

100

1000

10000

Np

tcl [

cmminus

3 ]

SA AS ATD__________

____

____

__
























4 Summary









Acknowledgements


References














































4 Summary









Acknowledgements


References













































Acknowledgements


References










































Experimental investigation of ice nucleation by different...

Documents

Transcript of Experimental investigation of ice nucleation by different...