View
5
Download
0
Category
Preview:
Citation preview
Research Collection
Doctoral Thesis
Thermodynamic Properties and Nucleation Processes of UpperTropospheric and Lower Stratospheric Aerosol Particles
Author(s): Knopf, Daniel A.
Publication Date: 2003
Permanent Link: https://doi.org/10.3929/ethz-a-004555208
Rights / License: In Copyright - Non-Commercial Use Permitted
This page was generated automatically upon download from the ETH Zurich Research Collection. For moreinformation please consult the Terms of use.
ETH Library
Diss. ETH No. 15103
Thermodynamic Properties and Nucleation
Processes of Upper Tropospheric and Lower
Stratospheric Aerosol Particles
A dissertation submitted to the
SWISS FEDERAL INSTITUTE OF TECHNOLOGY
ZURICH
for the degree of
Doctor of Natural Sciences
presented byDANIEL A. KNOPF
Dipl. Phys.born 1. May 1973
citizen of Germany
accepted on the recommendation of
Prof. Thomas Peter, examiner
Prof. Ulrich Schurath, co-examiner
Dr. Thomas Koop, co-examiner
2003
1
Zusammenfassung
Atmosphärische Aerosolteilchen unterhegen thermodynamischen und kinetischen Prozessen
durch die Wechselwirkung mit ihrer Umwelt. Diese Prozesse beinhalten unter an¬
derem Änderungen in ihrer Zusammensetzung durch die Aufnahme von Gasmolekülen und
Phasenübergänge, die durch Kristallbildung hervorgerufen werden. Eine atmosphärische Re¬
gion von besonderer Bedeutung, ist die obere Troposphäre/untere Stratosphäre (OT/US), in
der Aerosolteilchen an der Wolkenbildung und an heterogenen Reaktionen beteiligt sind. Diese
Prozesse sind für die Ozonchemie, die Strahlungswechselwirkung und die Dehydratation von
Luft beim Durchqueren der Tropopause von grosser Bedeutung. In dieser Arbeit wurden La¬
borexperimente mit Hilfe von optischer Mikroskopie in Kombination mit Raman-Spektroskopie
durchgeführt, die Relevanz für Aerosolprozesse in der OT/US besitzen.
Die Dissoziationsreaktion des Hydrogensulfations, HSOJ ?=± SO|~ + H+, wird in wässrigen
H2S04-Lösungen in einem Konzentrationsbereich von 0.54-15.23 mol kg-1 und einem Tempe¬
raturbereich von 180-330 K unter Verwendung der Raman-Spektroskopie untersucht. Alle unter¬
suchten H2S04-Lösungen zeigen einen kontinuierlichen Anstieg des Dissoziationsgrads von HSOJ
mit abnehmender Temperatur. Dies steht im Widerspruch zu Vorhersagen thermodynamischer
Modelle wässriger H2S04-Lösungen. Ein Pitzer Ionen-Wechselwirkungsmodell wird eingesetzt,
um eine thermodynamische Dissoziationskonstante des Hydrogensulfations, Kn(T), abzuleiten,
die thermodynamisch konsistent ist und mit den experimentellen Daten übereinstimmt. Die
neue Parameterisierung von Ku(T) ist gültig von 180 K bis 473 K. Im typischen Temperaturbe¬
reich der OT/US zeigen Berechnungen des Ionen-Wechselwirkungsmodells deutliche Abweichun¬
gen der Aktivitätskoeffizienten, der Wasseraktivitäten, der Wasserdampfdrücke und der HCl
Löslichkeiten zu bestehenden thermodynamischen Modellen von H2SO4/H2O Lösungen.
Die Zirrus-Eiswolkenbildung wird durch Messungen der oberen Grenzen des homogenen Eis-
nukleationsratenkoeffizienten und des pseudo-heterogenen Eisnukleationsratenkoeffizienten in
wässrigen (NRi)2S04 Tropfen untersucht. Für Temperaturen von 215 K ergeben die Experi¬
mente Ratenkoeffizienten von 106 cm-3 s_1 bzw. 3.2-102 cm-2 s~1. Diese Eisnukleationsraten-
koeffizienten sind kleiner als die durch Infrarotspektroskopie an strömenden Aerosolpartikeln
(AFT-IR: Aerosol Flow Tube-Infrared Spectroscopy) bestimmten Werte, sind aber im Ein¬
klang mit in Emulsions- und anderen optischen Mikroskop-Experimenten gewonnen Raten¬
koeffizienten. Es wird vermutet, dass die Gründe für die widersprüchlichen Datensätze in der
Aufarbeitung der experimentellen Daten liegen.
Die Bildung Polarer Stratosphärenwolken wird theoretisch und experimentell erforscht, in¬
dem homogene und pseudo-heterogene Nukleationsratenkoeffizienten von NAD und NAT
in HNO3/H2O und HN03/H2S04/H20-Lösungströpfchen untersucht werden. Die unter
stratosphärischen Bedingungen in flüssigen Tröpfchen gemessenen oberen Grenzen der homo¬
genen Nukleationsratenkoeffizienten von NAD und NAT sind 2-10-5 cm-3 s_1 und 8-10~2
cm-3 s-1. Die oberen Grenzen der pseudo-heterogenen Nukleationsratenkoeffizienten von NAD
und NAT betragen 1.5-10-6 cm-2 s-1 und 9-10~4 cm-2 s_1. Diese experimentell bestimmten
Nukleationsratenkoeffizienten sind bis zu 8 Grössenordnungen kleiner als Ratenkoeffizienten
berechnet durch eine Nukleationsparameterisierung, die in einer kürzlich veröffentlichten De-
nitrifizierungsstudie zur Anwendung kam. Die hier vorgestellten Nukleationsratenkoeffizienten
11
ergeben unter stratosphärischen Bedingungen Bildungsraten der aus Salpetersäure Hydraten
bestehenden Teilchen von maximal 3-10-10 cm-3 (Luft) h-1 und 6-10-6 cm-3 (Luft) h_1 für die
homogene und pseudo-heterogene Nukleation. Diese Produktionsraten sind zu gering, um die
kürzlich beobachteten Teilchenanzahlen grosser salpetersäurehaltiger Teilchen zu erklären und
reichen deshalb nicht aus, die daraus folgende Denitrifizierung des arktischen polaren Wirbels
zu beschreiben.
m
Abstract
Atmospheric aerosol particles are subject to thermodynamic and kinetic processes through inter¬
action with their environment. Such processes involve, among others, composition changes due
to uptake of gas phase molecules and liquid-to-solid phase transitions induced by crystal nucle¬
ation. An atmospheric region of particular interest is the upper troposphere/lower stratosphere
(UT/LS), where aerosol particles are involved in cloud formation and heterogeneous reactions,
which are important for ozone chemistry, radiation, and the dehydration of air crossing the
tropopause. In this work, laboratory experiments, employing optical microscopy combined with
Raman spectroscopy, have been performed to study aerosol processes relevant to the UT/LS.
The dissociation reaction of the bisulfate ion, HSO4 ^ SO2- + H+, is investigated in aqueous
H2SO4 solutions with concentrations of 0.54-15.23 mol kg-1 in the temperature range of 180-326
K using Raman spectroscopy. All investigated H2SO4 solutions show a continuous increase in
the degree of dissociation of HSOJ with decreasing temperature, in contrast to predictions from
thermodynamic models of aqueous H2SO4 solutions. A Pitzer ion interaction model is used to
derive a thermodynamically consistent formulation of the thermodynamic dissociation constant
of the bisulfate ion, Kn(T), that is in agreement with the experimental data. The new formula¬
tion of Ku(T) is valid from 180 K to 473 K. Calculations with the ion interaction model reveal
considerable differences in ion activity coefficients, water activities, water vapor pressure, and
HCl solubilities, when compared to existing thermodynamic models of H2SO4/H2O solutions,
in particular at temperatures typical for the UT/LS.Cirrus ice cloud formation is studied by measuring upper limits of the homogeneous and pseudo-
heterogeneous ice nucleation rate coefficients in aqueous (NHj)2S04 droplets. At temperatures
of about 215 K, the experiments reveal values of these rate coefficients that are smaller than ~106
cm-3 s_1 and ~3.2-102 cm-2 s-1, respectively. These values are smaller than the nucleation
rate coefficients obtained by AFT-IR (Aerosol Flowtube-Infrared) spectroscopy studies, but are
in agreement with emulsion studies and other optical microscope experiments. The reasons for
the discrepancies between the different data sets cannot be resolved but are suspected to have
their origin in the evaluation procedures of the experimental data.
Polar Stratospheric Cloud formation is studied by investigating, both theoretically and experi¬
mentally, the homogeneous nucleation and pseudo-heterogeneous nucleation of NAD and NAT
in HNO3/H2O and HNO3/H2SO4/H2O solution droplets. For polar stratospheric conditions,
the upper limits of the homogeneous nucleation rate coefficients of NAD and NAT in liquid
aerosols derived from the experiments are 210-5 cm-3 s_1 and 810-2 cm-3 s_1, respectively.
The upper limits of the pseudo-heterogeneous nucleation rate coefficients of NAD and NAT are
1.5-10-6 cm-2 s-1 and 9-10-4 cm-2 s_1, respectively. These experimentally derived nucleation
rate coefficients are lower by up to 8 orders of magnitude than values in recently published nu¬
cleation parameterizations used in denitrification studies. From the nucleation rate coefficients
presented here, maximum hourly production rates of nitric acid hydrate particles at strato¬
spheric conditions are calculated which yield about 3-10-10 cm"3 (air) h_1 and 610-6 cm~3
(air) h_1 for homogeneous and pseudo-heterogenous nucleation, respectively. These production
rates are too low to explain the number densities of large nitric acid containing particles recently
observed in the Arctic stratosphere and, therefore, are insufficient to account for the subsequent
IV
stratospheric denitrification of the Arctic polar vortex.
Contents
1 Introduction 1
1.1 Aerosols in the atmosphere 1
1.1.1 Particulate matter 1
1.1.2 The atmosphere 3
1.1.3 Aerosols in the UT/LS 3
1.2 Relevance of atmospheric aerosols 7
1.2.1 Climate forcing 7
1.2.2 Heterogeneous chemistry 9
1.3 Processes in UT/LS aerosols 10
1.3.1 Thermodynamic processes11
1.3.2 Kinetic processes11
1.4 Objectives of this PhD thesis 12
2 Theory 13
2.1 Thermodynamic processes in UT/LS aerosol 13
2.1.1 The Gibbs free energy13
2.1.2 Chemical potential of solutions 14
2.1.3 Henry's law constant 15
2.1.4 Water activity 16
2.1.5 Phase diagram of H2S04/H20 16
v
vi
2.1.6 Phase diagram of (NH4)2S04/H20 17
2.1.7 Phase diagram of HN03/H20 18
2.2 Kinetic processes in UT/LS aerosol 20
2.2.1 Classical nucleation theory 20
2.2.2 Surface nucleation 23
2.3 Raman spectroscopy 24
2.3.1 Classical derivation of the Raman effect 26
3 Experimental 29
3.1 Sample preparation 29
3.2 Sample cell 31
3.3 Hydrophobic coating 31
3.4 Experimental setup 32
3.5 Experimental procedure 34
4 Thermodynamic processes in UT/LS aerosol particles 35
4.1 Abstract 39
4.2 Introduction 39
4.3 Experimental Section 40
4.4 Results and Discussion 42
4.4.1 Analysis of Experimental Data 43
4.4.2 Results of the Pitzer Ion Interaction Model 49
4.5 Atmospheric Implications 53
4.6 Conclusions 56
4.7 Appendix 56
4.7.1 Tables 56
4.7.2 Derivation of the thermodynamic dissociation constant of HSOJ 61
4.7.3 Extended Pitzer Ion Interaction Model 62
Contents vii
4.8 HCl solubility in H2S04/H20 solutions 67
4.9 Analysis of H2SO4/H2O Raman spectra 69
4.10 Analysis of (NH4)2S04/H20 Raman spectra 75
4.11 The ferroelectric phase transition of (NH4)2S04 77
5 Kinetic processes in UT/LS aerosol particles 83
5.1 Abstract 87
5.2 Introduction 87
5.3 Nucleation formulation analysis 88
5.4 Experimental 90
5.5 Results and discussion 94
5.6 Conclusions 98
5.7 Pseudo-heterogeneous nucleation of PSCs 101
5.8 Homogeneous ice nucleation in (NH4)2S04/H20 droplets 107
6 Final remarks 113
6.1 Summary and conclusion 113
6.2 Outlook 116
A Experimental 119
A.l Electrical circuit for the operation of the inkjet-cartridge 119
B Raman spectroscopy 121
B.l Assignments of the normal vibrations of the investigated Raman spectra 121
C Heterogeneous chemistry 125
D Nucleation rate coefficients and production rates 127
D.l Derivation of upper nucleation limits 127
D.2 Derivation of stratospheric production rates of NAD and NAT 128
VlllContents
E Parametrizations of NAD and NAT nucleation mechanisms 131
E.l Homogeneous nucleation parametrization of NAD and NAT 131
E.2 Pseudo-heterogeneous nucleation activation energies of NAD and NAT 132
List of Figures 133
Bibliography 137
Acknowledgements 149
Chapter 1
Introduction
1.1 Aerosols in the atmosphere
1.1.1 Particulate matter
Aerosols are defined as relatively stable suspensions of solid or liquid particles in a gas (Finlayson-
Pitts and Pitts, 2000). Atmospheric aerosols consists of particles and particulate matter with
diameters of ~0.002-100 fim.
The particles arise from natural sources like sea spray and volcanos, but also from anthropogenic
activities such as combustion of fuels. Aerosol particles can be emitted directly (primary aerosol)
into the atmosphere or can be formed in the atmosphere by gas-to-particle conversion processes
(secondary aerosol). The airborne particles can change their size and composition due to several
interaction processes, such as condensation and evaporation of vapor species, coagulation with
other aerosol particles, and chemical reactions. In an environment supersaturated with respect
to water the particles also can become activated into cloud droplets.
Figure 1.1 shows a typical aerosol size distribution. Whitby and Sverdrup (1980) suggested
that three distinct aerosol size modes exist: Particles larger than 2.5 /im are identified as coarse
particles and those with diameters smaller than 2.5 /xm are called fine particles. Most of the
aerosol mass and aerosol number belongs to this fine particle mode. This mode can be further
devided into the accumulation range (0.08 /im to 1-2 /im) and the transient or Aitken range
(0.01-0.08 /im). In recent years the technology for measuring very small particles has advanced,
so a fourth size mode could be identified known as ultra fine particles, which describe aerosol
particles with diameters smaller than 0.01 /im. Figure 1.1 also shows the main atmospheric
aerosol formation mechanisms, leading to the different particle size modes. These formation
mechanisms involve the chemical conversion of gases to low volatile gases, the condensation of
hot vapor to particulate matter occurring during combustion, and non-chemical processes such
as wind blown dust and sea spray. Homogeneous nucleation plays a fundamental role in the
formation processes of aerosol particles. These particles can grow in diameter by coagulation
until they reach sizes corresponding to the Aitken mode and accumulation mode. The two
1
2 CHAPTER 1. INTRODUCTION
Chemical conversion
of gases to low
volatility vapors
Chemical conversion
of gases to low
volatility vapors
Wind blown dust
+
Emissions
Sea Spray+
Volcanos
+
Plant particles
Particle diameter (urn)
„ „
Transient nucleiUltra fine or Aitken nuclei
particles__
| range->+*- -*+*-
Accumulation
range
fine particles -
Mechanicallygeneratedaerosol range
Coarse particles -
Figure 1.1: Sketch of atmospheric aerosol formation mechanisms and corresponding four mode aerosol
size distribution (Finlayson-Pitts and Pitts, 2000). This graph was adapted from Whitby and Sverdrup
(1980) whose original hypothesis consist only of a three mode aerosol size distribution which is indicated
as solid line.
main mechanisms which axe responsible for the removal of atmospheric aerosol particles are
dry deposition and wet deposition. The first applies mainly to coarse particles which sediment
due to gravitational forces. The latter is the main loss process for particles belonging to the
accumulation size mode due to the scavenging by raindrops (washout).Table 1.1 shows average values for mass and composition of typical tropospheric aerosols. A
significant fraction of the aerosol is due to anthropogenic origin. The particles contain sulfate,
ammonium, nitrate, and a large fraction of elemental and organic carbon. Elemental carbon
(soot, graphite) is directly emitted by combustion processes. Organic carbon can be emitted
directly or forms by condensation of low volatile organic gases such as polycychc aromatic
hydrocarbons (PAHs).
1.1. Aerosols in the atmosphere 3
Table 1.1: Moss concentration and composition of tropospheric aerosols (Heintzenberg, 1989). The
composition is given in percentage.
Region Mass [/xg m 3] C (elemental) C (organic) NH+ NO3 SOl"
Remote 4.8 0.3 11 7 3 22
Nonurban continental 15 5 24 11 4 37
Urban 32 9 31 8 6 28
In the free troposphere above about 6 km height the average mass concentration of the particles
is about 1 /ig m-3. At lower altitudes the mass concentration can vary from 0.7 /ig m-3 in
remote continental sites to 150 /ig m-3 in desert regions (Jaenicke, 1993). An average aerosol
number density of about 300 cm-3 is observed in the well-mixed troposphere above 6 km height.
Higher particle number concentrations of up to 3104 cm-3 can be found in remote continental
areas at lower altitudes (Jaenicke, 1993).
1.1.2 The atmosphere
Figure 1.2 shows the pressure and temperature of the standard atmosphere as a function of
altitude. The pressure decreases exponentially with altitude. As a rule of thumb, the pressure
decreases by a factor oftwo over 5 km in height. Temperature decreases with altitude throughout
the troposphere up to the tropopause, where the temperature experiences a minimum. The
height of the tropopause is defined by the lowest altitude level where the temperature decreases
by less than two kelvin per kilometer (WMO, 1992). The temperature of the tropopause can be
as low as 180 K. The height of the tropopause changes with latitude and season, ranging from
8 km at the poles up to 16-17 km in the tropics. While temperatures in the lower stratosphere
are generally warmer than at the tropopause it can be as low as 180 K in the polar regions, a
prerequisite for the formation of Polar Stratospheric Clouds (PSCs). The temperature in the
stratosphere rises with altitude until a temperature maximum is reached at the stratopause.
The gray shaded region in Fig. 1.2 indicates the part of the atmosphere which is called Upper
Troposphere/Lower Stratosphere (UT/LS). It is the region of the atmosphere between ~10-25
km, which is one of the coldest parts of the atmosphere. This thesis focuses on aerosol particles
and their interactions with the surrounding environment in this part of the atmosphere.
1.1.3 Aerosols in the UT/LS
The chemical composition of aerosol particles in the UT/LS can vary significantly. It is
generally agreed that aerosol in the upper troposphere are usually sulfate particles with a
varying degree of neutralization ranging from sulfuric acid to ammonium sulfate (Martin, 2000;
Colberg, 2002). Additional components such as nitrates and organics have been also found in
upper tropospheric aerosols (Murphy et al., 1998).
4 CHAPTER 1. INTRODUCTION
<B133
1C-3 10"2 10'1 1 10 102
00
I 1 1 I 1
Thermosphère /
/
90 /
1
j Mesopause
80 - \
70
60
50
-
X. Mésosphère XI®*.
X Stratopause
\ ;
40 \ /
Stratosphere \ /
30 / \
20
/ x/ xi X.
10
TrofopHtise \ X.\ XN X
01 I I I
Troposphere v^
1 1 I 1 1 1 I I 1 h I I I 1
100 200 300
Temperature (K)
Figure 1.2: The pressure and temperature ofthe standard atmosphere are plotted as afunction of altitude
(adapted from Finlayson-Pitts and Pitts (2000)). The different parts of the atmosphere are indicated. The
gray shaded area is defined as the upper troposphere/lower stratosphere.
Since the 1970's it has been known that stratospheric aerosol particles consist of sulfuric acid and
water (Junge and Manson, 1961). The source for the H2SO4 in the globally distributed strato¬
spheric background aerosol is the oxidation of carbonyl sulfide (OCS). This is oxidized to SO2
which further reacts with OH to finally form H2SO4. H2SO4 nucleates to form aerosol particles.
Another important source for stratospheric H2SO4 axe major volcanic eruptions. Because of its
very low vapor pressure, the stratospheric background aerosol consists predominantly of sulfate
as was recently confirmed by laser ionization mass spectrometry, see Fig. 1.3. During this field
measurements only in a few percent of the particles was sulfate not the dominant compound
(Murphy et al., 1998).Stratospheric background aerosol particles act as precursor for the formation of PSCs. The latter
have been observed since 1870 (Stanford and Davis, 1974). During winter time PSCs form in the
polar regions at an altitude of 15-30 km, when the temperature drops below ~198 K. Usually
PSCs persist longer over the Antarctic region since the polar vortex is dynamically more stable
than that over the Arctic pole and, therefore, the temperatures remain longer cold enough for
PSC formation (Schoeberl and Hartmann, 1991). PSCs consist mainly of ternary mixtures of
1.1. Aerosols in the atmosphere 5
0.3
|o.2m
0.1
0.0
OH"
1»8«$©7#o6?219 km; 31°M
(M37K
MS04"
HSO,
20
SOjT 804"
^-4-
rrso/
40 60 80
Ion mass/charge
100 100
Figure 1.3: Negative ion mass spectra of individual sampled aerosol particles in the lower stratosphere
Murphy et al. (1998).
Table 1.2: PSC properties. Adapted from Turco et al. (1989).
Type la Type lb Type II
T threshold < 198 K > 187 K < 187 K
composition H2SO4/HNO3/H2O STS H2O + traces
diameter > 1 /im 0.5-1 /im 5-100 /im
phase crystalline liquid ice
H2SO4, HNO3, and H2O. Table 1.2 presents typical properties ofPSC particles such as existence
temperature, composition, diameter, and the kind of phase. There are three forms of PSC par¬
ticles. Type I PSCs contain signifant amounts of HNO3, whereas type II PSCs consist mainly of
water ice. PSCs of type I axe further divided into type la and lb. Laboratory experiments sug¬
gested that PSCs of type la consist of crystalline NAT (nitric acid trihydrate) which is the most
stable nitric acid hydrate under stratospheric conditions (Hanson and Mauersberger, 1988a).
Worsnop et al. (1993) showed in laboratory experiments that also crystalline NAD (nitric acid
dihydrate) could exist in PSCs. Type lb PSCs are composed of supercooled ternary solutions
(STS). First suggestions of their existence were based on model calculations. Figure 1.4 repre¬
sents the volume density of PSCs measured in the Airborne Arctic Stratospheric Experiment
1989 (AASE) by Dye et al. (1992) and the thermodynamic equilibrium model calculations of
Carslaw et al. (1994).
Figure 1.4 shows that neither aqueous H2SO4 nor NAT particles can explain the measured
volume densities. The model calculations indicate that in this field measurements STS aerosols
6 CHAPTER 1. INTRODUCTION
10
gm
E
10.1
0,01185 190 195 200 205 210
T/K
Figure 1.4: The dots represent particle volumes measured by Dye et al. (1992). The lines indicate
model calculations of Carslaw et al. (1994). The thick solid line represents model calculations assuming
the growth of liquid STS particles by Hi 0 and HNO3 uptake. The dashed line corresponds to a model
simulation taking into account the NAT deposition on frozen particles. The dotted line shows the growth
of binary liquid H2SO4/H2O by H2O uptake. The model calculations were performed at 55 mbar for 5
ppmv H2O, lOppbv HNO3, and 0.5 ppbv H2SO4. The thin solid lines indicate a sensitivity study assuming
5 and 15 ppbv gas-phase HNO3, respectively. The NAT saturation temperature for 10 ppbv HNO3 (196.2
K) and the frost point (188.9 K) are also displayed.
were sampled. The model calculations in Fig. 1.4 are performed with a Pitzer ion interaction
model (Carslaw et al., 1995a). In the meantime in situ composition measurements of PSC
particles have confirmed that PSC particles can consist of STS or NAT particles (Schreiner
et al., 1999; Voigt et al., 2000).At temperatures below the frost point Type II, PSCs can form. Due to the large amount
of available water vapor these PSC particles can grow to diameters up to 100 /im. Type la
particles are typically larger than type lb aerosols, since only a few particles nucleate NAT
which subsequently deplete the available gaseous HNO3. The liquid type lb particles grow
simultaneously, thereby distributing the gaseous HNO3 amount to a larger number of particles.
Therefore, their size is smaller than that of type la PSCs.
If HNO3 containing particles grow to particle sizes larger than 2 /im in diameter, they can
sediment to lower altitudes due to gravitational forces. This process can lead to a significant
removal of HNO3 from the stratosphere and is known as stratospheric denitrification (Seinfeld
and Pandis, 1998). At lower altitudes and, hence, higher temperatures, the particles evaporate
and HNO3 is released again into the gas phase. The removal of HNO3 is crucial for stratospheric
gas-phase chemistry since NO2 (NO2 is a photolysis product of HNO3) converts the ozone
destroying gas CIO into the unreactive reservoir species CIONO2 (Seinfeld and Pandis, 1998).
A still unresolved question is the observation of large HNO3 containing particles up to 15 /im
in diameter with number densities of about 10-4 cm-3 during the Arctic winter of 1999/2000
(ICE) (NAT)
Ll Li I
1.2. Relevance of atmospheric aerosols 7
(Fahey et al., 2001). Such large HNO3 containing particles can have a major impact on the
denitrification of the stratosphere. Tabazadeh et al. (2001) proposed that the observed number
densities of the large particles can be explained by homogeneous nucleation of NAD and NAT
in STS particles. In contrast, Knopf et al. (2002) derived NAD and NAT production rates from
laboratory experiments showing that homogeneous nucleation rates in STS particles are too
small to explain stratospheric denitrification. Thus, other formation processes of the observed
large particles must be involved, such as heterogeneous nucleation of NAT onto PSC type II
particles (Waibel et al., 1999) or the proposed mother cloud/NAT-rock mechanism (Fueglistaler
et al., 2002). A very recent suggestion of Tabazadeh et al. (2002a) follows the idea of a pseudo-
heterogeneous nucleation mechanism of NAD and NAT in STS particles. These suggestions will
be discussed in more detail later in this thesis.
1.2 Relevance of atmospheric aerosols
1.2.1 Climate forcing
The Earth's climate is controlled by the energy balance between the incoming solar radiation
and the emission of long-wave radiation from the Earth-atmosphere system into space. 30 % of
the incoming solar radiation is scattered back to space due to the Earth's albedo. The latter is
strongly influenced by the presence of aerosols and clouds. A major proportion of the radiation
emitted by the Earth into space is absorbed by greenhouse gases, aerosols, and clouds, thereby
warming the atmosphere.A measure of any perturbation in the radiative energy budget of the Earth's climate system is
given by the term "radiative forcing". The definition for radiative forcing given here is adapted
from the Intergovernmental Panel on Climate Change (Houghton et al., 2001): uThe change in
radiative forcing of the surface-troposphere system due to the perturbation in the amount of or
the newly introduction of an agent in the atmosphere, respectively, is defined as the change in
net (downward minus upward energy flux) irradiance (incoming solar radiation plus outgoing
long-wave radiation in units W m~2) at the tropopause after allowing for stratospheric temper¬
atures to readjust to radiative equilibrium, but with surface and tropospheric temperatures and
state held fixed at the unperturbed values".
The aerosol effect on radiative forcing can be divided into the direct and the indirect aerosol
effect. The direct effect describes the interactions of the aerosol with radiation. The indirect
effect takes into account the ability of aerosol particles to serve as cloud condensation nuclei
(CCN) or ice nuclei (IN) thereby changing the optical properties of clouds. Figure 1.5 sum¬
marizes the anticipated radiative forcing of various substances by IPCC. The scientific level of
understanding for the particular processes is also indicated (Houghton et al., 2001). Here we
will focus on the effects related to aerosols. Sulfate containing aerosols and particles generated
by biomass burning have a negative radiative forcing, i. e. they contribute to a cooling of the
atmosphere. The influence of mineral dust particles is not clear yet. They can contribute to
either cooling or heating. Aerosols containing black carbon emitted from fossil fuel burning axe
expected to heat the atmosphere, since these particles are strong absorbers. Aerosols containing
8 CHAPTER 1. INTRODUCTION
| -2
Hatocarbons
CH«
Aerosols
CO,
Tropoapnericozone
Blackcarbon irom
fossilfuel
burning
MineralDust
-cçrStratasphariç
ozone
^ii «-*-»—nnT [ * ^pl"^ Organic 11 carbon ri*»«,«
Organiccarbon
fuel
burning
Blomass
owning
Aviation-induced
,*
N
Contrails cirrus
Tn.CD
Solar
Aaissaf
indlwctsi set
Land-
(afcedo)only
High Med. Med. Low VeryLow
Verytow
Very VeryLow Low
VeryLow
VaryLow
VeryLow
VsryLow
Leva of Scientific Understanding
Figure 1.5: The effect on radiative forcing is shown for various atmospheric radiatively active agents.
The level of scientific understanding is also indicated for the presented processes (Houghton et al., 2001).
organic carbon are assumed to cool the atmosphere. The scientific level of understanding of
these processes is in general low to very low.
It is assumed that changes in cloud properties induced by aerosol particles contribute negatively
to radiative forcing. However, because of the large uncertainties involved no forcing value was
assigned by IPCC and only a range is given, indicated by the error bars in Fig. 1.5. In the follow¬
ing sections the influence of aerosols on the direct and indirect aerosol effects will be discussed
in further detail.
Direct aerosol effect
The aerosol particles in the atmosphere interact with the incoming solar radiation by scattering
and absorption. Therefore, the particles affect the radiation budget of the Earth-atmosphere
system. The main uncertainties in quantifying this direct aerosol effect are accurate estimates
in the determination of the primary aerosol sources (Houghton et al., 2001). Secondary aerosol
species also have uncertainties both in the sources of the precursor gases and in the atmospheric
processes that convert some of those gases to aerosol particles.
The optical interaction of aerosols with solar radiation depends on the phase and radius of
the particles. Due to the temperature and relative humidity changes in the atmosphere, aerosol
particles are forced to take up or to release water and, hence, change their diameters accordingly.
Therefore, particle radius and phase are crucial for the optical properties of the aerosols, i.
1.2. Relevance of atmospheric aerosols 9
e. wether the particles have a net positive or negative effect on radiative forcing. As long
as the particles are liquid or partially liquid the change in particle volume due to changes
in atmospheric conditions can be calculated using thermodynamic models such as Pitzer ion
interaction models. The optical properties of the liquid aerosol will change significantly, when a
crystalline phase forms in the particles. The nucleation of a new phase is a kinetic process and,
therefore, cannot be predicted by thermodynamic models. Hence, the atmospheric conditions
at which crystalline phases nucleate in aqueous droplets must be determined by laboratory
experiments. The experimentally derived nucleation rates can then be applied to radiation
models to obtain an estimate of the radiative forcing.
Indirect aerosol effect
The indirect aerosol effect links various processes such as the ability of aerosol particles to serve
as CCN and IN with the resulting radiative forcing due to clouds. In the lower troposphere the
indirect aerosol effect combines two individual effects:
First, anthropogenic emissions increase the number of aerosols which can act as CCN. Therefore,
the number of cloud droplets is increased and the cloud droplet size distribution is shifted to
lower diameters. This leads to an increases of the optical depth of the cloud, enhancing the
cloud albedo (Twomey, 1974).The second effect refers to the lower precipitation efficiency and the increased thickness of clouds
due to the reduction in cloud droplet size and the increase of cloud droplet number density, re¬
spectively (Albrecht, 1989; Pincus and Baker, 1994).In the upper troposphere cirrus ice clouds form due the low temperatures in this part of the
atmosphere. Although the role of cirrus ice clouds on climate is not yet quantified, ice formation
by homogeneous nucleation of aerosol particles is believed to have an impact on the global ra¬
diative forcing (Houghton et al., 2001). Several Global Circulation Models (GCM) studies have
supported the expected influence of ice formation from supercooled water on global radiative
forcing (Senior and Mitchell, 1993; Fowler and Randall, 1996). Lohmann and Feichter (1997)
performed a sensitivity modelling study showing a globally averaged cloud forcing of +16.9 W
m-2 obtained by allowing only ice in clouds with a temperature below 273 K compared to clouds
containing only water droplets for temperatures above 238 K. Therefore, even small changes of
the ice content in the clouds can have a significant impact on the global radiative forcing.
Current models still suffer from uncertainties in the parameterization of the microphysical for¬
mation mechanisms of ice particles in high-altitude clouds (Jensen et al., 1994a,b). One way to
significantly improve this situation axe detailed laboratory experiments on ice nucleation pro¬
cesses in aerosols. These experimental data are required to develop microphysical nucleation
models such as the one recently presented by Koop et al. (2000) for homogeneous ice nucleation.
1.2.2 Heterogeneous chemistry
Atmospheric trace gases can react heterogeneously on aerosol particle surfaces and subsequently
homogeneously within the particle volume. The most prominent example of the importance of
10 CHAPTER 1. INTRODUCTION
heterogeneous reactions on aerosols is the Antarctic Ozone hole. The large Ozone losses during
the Antarctic spring time can be explained only by including heterogeneous reactions on the
surface of PSCs (Solomon et al., 1986; Ravishankara and Sheperd, 1999). The heterogeneous
reactions activate chlorine from its reservoir species (HCl and CIONO2) into CI2, which then
is photolyzed into two chlorine atoms in eaxly spring. These chlorine radicals can react with
O3, thereby destroying a large amount of the ozone within a short time period. There axe four
important heterogeneous reactions on PSCs:
HCI + CIONO2 - CI2 + HNO3 (1.1)
HC1 + N205 -» CINO2 + HNO3. (1.2)
CIONO2 + H2O -* HOCI + HNO3 (1.3)
HC1 + HOC1 -» CI2 + H2O (1.4)
Reaction 1.1 was shown to be a two-step process: CIONO2 hydrolyzes on the surface of the par¬
ticle forming HOC1 (see Reaction 1.3), which subsequently reacts homogeneously with dissolved
HCl to CI2 (see reaction 1.4) (Hanson and Ravishankara, 1991, 1993; Abbatt et al., 1992).
Also, cirrus ice clouds have the potential to serve as a surface for heterogeneous reactions in¬
volving chlorine activation (Borrmann et al., 1997b). During AASE II the research aircraft
ER-2 measured a positive correlation between aerosol surface and CIO concentrations within
cirrus clouds (Borrmann et al., 1997a). Model simulations of a 3-D chemical transport model
including typical heterogeneous reactions on ice surfaces also confirm the correlation between
the occurrence of subvisible ice clouds and measured CIO concentrations (Bregman et al., 2002).
Due to the their major atmospheric impact, the heterogeneous reactions of CIONO2 and N2O5
with HCl and H2O were intensively investigated in numerous laboratory experiments (see e. g.
Hanson and Ravishankara, 1992, 1993; Williams and Golden, 1993; Elrod et al., 1995; Hanson,
1998; Zhang et al., 1993a; Robinson et al., 1998). The experimental data show that heteroge¬
neous reactions involving HCl tend to be fast on solid surfaces such as ice, NAT, and Sulfuric
Acid Tetrahydrate (SAT) and also on liquid aerosol surfaces consisting of binary H2SO4/H2O
or ternary H2SO4/HNO3/H2O solutions.
The reaction rates of heterogeneous reactions depend on the uptake ability of gaseous molecules
onto/into the solid and liquid particles. The solubility of trace gases into liquid aerosol parti¬
cles under atmospheric conditions is expressed by the Henry's law constant, which is usually
predicted for systems such as H2S04/H20, H2SO4/HNO3/H2O, and NH3/H2SO4/H2O using
Pitzer ion interaction models.
1.3 Processes in UT/LS aerosols
In the UT/LS the aerosol is exposed to temperatures down to 180 K and relative humidities
with respect to ice of 10-150 % (Gierens et al., 1999; Jensen et al., 1994a,b; Heymsfield et al.,
1998). The influence of the aerosols on the environment involves, among others, radiative
forcing by particles, cloud formation, and heterogeneous chemistry on the particle surface. These
interactions can only be understood and quantified through the knowledge of the physical and
1.3. Processes in UT/LS aerosols 11
chemical aerosol properties. In the UT/LS the aerosol particles are often composed of inorganic
water soluble species such as H2SO4, NH3, and HNO3. These substances dissociate in the
aqueous solutions, thereby forming ions. Hence, in most cases UT/LS aerosols can be considered
to consist of aqueous electrolytic solutions.
1.3.1 Thermodynamic processes
The theory of Debye and Hückel (1923a,b), describing the ionic forces within aqueous elec¬
trolytes, paved the way for the invention of thermodynamic models of electrolytes based on
the ion interaction (Pitzer) equations (Pitzer, 1991). The functionality of these models requires
accurate thermodynamic data sets of the investigated systems over a large temperature and con¬
centration range. These models can be used to predict thermodynamic properties of aqueous
solution aerosol particles under atmospheric conditions. These properties involve the speciation
of the various ions within the solution or the change in water vapor pressure and water activity
of the solution with temperature and concentration. The two most important parameters de¬
termining the concentration of solutes in aqueous atmospheric aerosol particles are temperature
and relative humidity. When an aerosol particle is in equihbrium with its environment, the
gas phase relative humidity is equal to the liquid phase water activity, which in turn can be
calculated using Pitzer models.
Heterogeneous chemistry is also affected by the thermodynamic properties because it depends
strongly on the solubihty of the involved gases. The solubihty of a trace gas into an aqueous
solution changes with temperature and relative humidity, and under equilibrium conditions sol¬
ubilities can be predicted using Pitzer models. Liquid aqueous UT/LS aerosol particles can also
experience a large degree of supercooling, i. e. they are metastable with respect to crystalline
phases. The composition in this supercooled temperature regime and the change in speciation,
water activity, and solubihty with changing environmental conditions, can be obtained by Pitzer
models. However, the available data to constrain Pitzer models at the low temperatures that
can occur in the UT/LS is very limited. Therefore, experimental investigations of thermody¬
namic aerosol properties at very low temperatures will improve the predictions of the particle
characteristics and interactions in the UT/LS region.
1.3.2 Kinetic processes
Due to large fluctuations in atmospheric conditions, the aerosol particles in the UT/LS expe¬
rience phase changes. An increase in relative humidity can drive a solid particle to become
an aqueous aerosol particle (deliquescence). Decreasing relative humidity can lead to solidifica¬
tion of an aqueous particle (efflorescence). Under certain atmospheric conditions ice or certain
hydrates nucleate in an aqueous solution, thereby forming cirrus ice clouds or PSC particles,
respectively. In the UT/LS region two types of clouds are dominant: Cirrus ice clouds in the
upper troposphere and PSCs in the lower stratosphere. Cirrus ice cloud formation is assumed to
occur, at least in part, due to homogeneous ice nucleation of aqueous H2SO4 and (NH4)2S04 so¬
lutions (Houghton et al., 2001; Martin, 2000). PSC formation requires the nucleation of ice and
12 CHAPTER 1. INTRODUCTION
of nitric acid hydrates from supercooled ternary HNO3/H2SO4/H2O solutions (Peter, 1997).
Nucleation of a crystalline phase in an aqueous solution droplet is a process whose kinetics,
though dependent on thermodynamic properties, cannot be described by the laws of thermody¬
namics, but has to be measured in laboratory experiments or in situ in the atmosphere.
1.4 Objectives of this PhD thesis
In this Ph.D. thesis thermodynamic and kinetic processes will be studied in aerosol particles of
the UT/LS region of the atmosphere.
The investigation of thermodynamic processes of UT/LS aerosols, such as the changes in water
activity and in solubility of involved trace gases due to changes in the atmospheric conditions,
requires the knowledge of the thermodynamic properties of the investigated aqueous aerosol par¬
ticles at low temperatures. Experimentally derived thermodynamic data sets of aqueous solution
droplets obtained at low temperatures will be applied to constrain thermodynamic models of the
investigated systems. The improved thermodynamic models will yield more accurate predictions
of the thermodynamic properties of UT/LS aerosols and, therefore, a better understanding of
the interactions of the particles with their environment.
Kinetic processes in UT/LS aerosol particles such as PSC and cirrus ice cloud formation will
be investigated by employing nucleation experiments. In the context of PSC formation the ob¬
servation of large nitric acid containing particles will be addressed by measuring experimentally
homogeneous nucleation rate coefficients of NAD and NAT in aqueous HNO3 and HNO3/H2SO4
droplets under stratospheric conditions. Recently suggested theoretical parameterizations of ho¬
mogeneous and pseudo-heterogeneous nucleation mechanisms of NAD and NAT in STS aerosols
will be verified theoretically and experimentally by applying the nucleation data sets of NAD
and NAT derived in this work.
The influence of high altitude cirrus ice clouds on the globally radiative forcing will be followed up
by the investigation of the ice formation mechanisms in these altitudes. Therefore, homogeneous
ice nucleation rate coefficients will be derived experimentally in aqueous (NH4)2S04 particles.
The experimentally obtained data set will allow the comparison with ice nucleation data de¬
rived by other experimental methods and the verification of a possible pseudo-heterogeneous
nucleation pathway.
Chapter 2
Theory
In section 2.1 thermodynamic principles axe presented which axe necessary to describe thermo¬
dynamic processes such as gas-phase to liquid-phase interactions for varying temperatures and
compositions of the investigated solutions. The phase diagrams of the aqueous systems studied
in this work wiU be shown. In section 2.2 kinetic principles are discussed, i. e. the formation of a
critical nucleus in a supersaturated environment leading to crystalhzation of the solution. A brief
derivation of the Classical Nucleation Theory (CNT) will be given. The pseudo-heterogeneous
nucleation mechanism, i. e. nucleation induced at the surface of a droplet, recently suggested by
Tabazadeh et al. (2002a,b) and Djikaev et al. (2002) will be presented. The last section deals
with the theoretical background of Raman spectroscopy which is one of the main investigating
tool of this thesis.
2.1 Thermodynamic processes in UT/LS aerosol
2.1.1 The Gibbs free energy
From the first and second law of thermodynamics the Gibbs free energy, G, of a system is defined
as (Atkins, 1994):G = U +pV-TS, (2.1)
where U is the internal energy of the system, p and V its pressure and volume, respectively,
and S is the entropy of the system. T is absolute temperature. The study of atmospheric
processes is facilitated by the introduction of the Gibbs free energy since the variables p and T
axe convenient to obtain. Using the Gibbs free energy one can define the chemical potential for
a species i in a solution, /ij, for constant T, p, and constant moles of additional other solution
species, nv as:
*-(£) ,(2-2)
13
14 CHAPTER 2. THEORY
where nt is the number of moles of species i. From this the general equation
k
G = Y,ßtnt, (2.3)î=i
can be derived. The total Gibbs free energy of a system is the sum of all single chemical potentials
weighted by the corresponding number of moles. The second law of thermodynamics states that
the entropy of a system in an adiabatic (dQ = 0) enclosure increases for an irreversible process
and remains constant in a reversible one. This is expressed as dS > 0. This corresponds to
dG < 0 or that a system will tend to decrease its Gibbs free energy for any process to occur
spontaneously.
2.1.2 Chemical potential of solutions
A solution is defined as ideal if the chemical potential of every component is a hnear function
of the logarithm of its aqueous mole fraction, xt, according to the relation (Seinfeld and Pandis,
1998):Ht = l4(T,p) + RTIn x%, (2.4)
where /i* is the chemical potential of the pure species (xt = 1) under the same pressure and
temperature as the solution under discussion. R is the universal gas constant. When the partial
pressures of the components vary linearly with xt, i. e.
x, = h ,(2.5)
the solution is called ideal. pt is the vapor pressure of species i over the solution and p° is the
vapor pressure over the pure component i. Equation 2.5 is a form of Raoult's law, which for
real solutions usually holds only in the dilute concentration range.
Atmospheric aerosols are usually concentrated aqueous solutions that deviate significantly from
ideality. The deviation from ideality is described by introducing the activity coefficient, 7,, and,
thus, the chemical potential is given as (Seinfeld and Pandis, 1998):
Ai, = /i:(T,p) + Ärin(7,*a:t). (2.6)
The activity coefficient is a function of p, T, and x%. For an ideal solution 7* = 1. /i* is defined
as the chemical potential at the hypothetical state for which 7* — 1 and x% — 1. The product
of activity coefficient and mole fraction, 7*x4, is called activity, al.
An often used concentration scale is molahty, mt, defined as moles of solute i per kilogram of
solvent given in units mol kg-1. On the molahty scale the chemical potential, /i,, is defined as
IM = n\{T,p) + RThiat, (2.7)
where at = 7»^-. /4 is the value of the chemical potential asm,-*1 and ^ —* 1, i. e. when m, =
mt = 1 mol kg-1. In this case the hypothetical state of ideality corresponds to a concentration
of 1 mol kg-1.
2.1. Thermodynamic processes in UT/LS aerosol 15
2.1.3 Henry's law constant
The distribution of a species between the gas and aqueous phase is described by the Henry's law
constant, Hx, often given in units mol atm-1. Hx for a particular species X is the equilibrium
constant for the reaction
X(g) ^ X(aq) Hx = [XU/px, (2-8)
where [X]aq is the concentration of species X in the solution and px is its partial pressure. For
a real solution the activity of the species X, ax, is used instead of [X]. The effective Henry'slaw constant, H^, (given in units mol l-1 atm-1) takes into account further reaction of X in
the liquid phase (Seinfeld and Pandis, 1998). The solution of gaseous HCl will be given as an
example. The first step is the hydrolysis of HCl into the aqueous solution, the second step is
the dissociation of HCl:
HCl(g) ^ HCl(aq) (2.9)
HCl(aq) ^ Cr + H+. (2.10)
The equihbrium constant for the above reactions are:
i*HC. =^ (2-11)PHC1
Kdis = ^aiSSi, (2.12)ÛHC1
where i^dis is the thermodynamic dissociation constant of HCl. Since anci = #hciPhci Eq. 2.12
can be written as
tfdis = f^^. (2-13)iiHClPHCl
Therefore, the amount of dissociated HCl is given by
-Kdis-ffHClPHCl foiA\aa- =
. (2.14)
aH+
The total amount of dissolved gaseous HCl, afja, can be obtained by
aHCi = «HCi + aci- (2-15)
= #hci-Phci(i +—) (2-16)
Equation 2.16 shows similarity to the definition of the Henry's law constant. Therefore, one can
define the effective Henry's law constant as
tfSci = HHa(l + —). (2-17)
Equation 2.17 indicates that the solubility of gaseous HCl depends also on the amount of H+ in
the solution.
16 CHAPTER 2. THEORY
2.1A Water activity
In a given air parcel the water content is not affected by transport of water vapor to the condensed
water of the aerosols due to the large water amount in the gas phase compared to the liquid
water in the aerosol phase. Considering the equilibrium
H20(g) ^ H20(aq)
and using the criterion for thermodynamic equilibrium between the gas phase and aqueous phase
(AtH20(g) = AtH20(aq)) the water activity, Ow, can be written as:
pw RH, .
aw =
^=
IÖÖ' (2-18)
where pw is the water vapor pressure of the solution and p% is the saturation vapor pressure of
the pure liquid at the same conditions. RH is the relative humidity given in percent and is equal
to aw. Thus, for each RH the water activity for any aqueous aerosol solution is fixed as long as
equilibrium conditions axe maintained.
2.1.5 Phase diagram of H2S04/H20
Junge (1961) discovered a layer of stratospheric aerosols which consists mainly of liquid
H2SO4/H2O particles. Since the vapor pressure of sulfuric acid of about 10~15-10-20 mbar is
very low it can be dealt as a non-volatile substance in the troposphere and lower stratosphere.The upper bound of the stratospheric aerosol layer is defined by the complete evaporation of the
aerosol particles due to higher temperatures (250 K) and lower water partial pressures (Carslawet al., 1997). In lower regions of the atmosphere other gaseous species exist such as ammonia
which are taken up by the liquid H2SO4/H2O particles to form ternary solutions or solids.
Figure 2.1 shows the phase diagram for aqueous sulfuric acid in the weight percent scale taken
from Gable et al. (1950). This phase diagram was corroborated by various thermodynamicmodels and measurements (Tabazadeh et al., 1994; Carslaw et al., 1995a; Luo et al., 1995;
Clegg and Brimblecombe, 1995; Middlebrook et al., 1993; Zhang et al., 1993b). The followingsolid phases can be identified in the phase diagram:
H2SO4 - H20: The Sulfuric Acid Monohydrate (SAM) is thermodynamically stable in the
temperature range of 220-250 K. Koop et al. (1997b) strongly suggest that aqueous H2SO4
does not nucleate as SAM due to low nucleation rates.
H2S04 • 2 H20 and H2S04 • 3 H20: The Sulfuric Acid Dihydrate (SAD) and the Sulfuric
Acid Trihydrate (SATr) which do not play a significant role in the stratosphere due to very low
nucleation rates (Koop et al., 1997b).
2.1. Thermodynamic processes in UT/LS aerosol 17
ouu
280 /\ )
rture[K]
i\>
ro
o
o
.ICE \
/ SAM W,
\ v-2i
Q.
V /sat vy7 /
jk H2S04
| 22° \ J SATr SAD
200y^SAH
180 -
300
280
260
- 240
220
200
180
0 10 20 30 40 50 60 70 80 90 100
H2S04 [Wt%]
Figure 2.1: The binary phase diagram of aqueous H2SO4 is shown (Gable et al., 1950). The solid lines
correspond to the melting curves of the indicated solids ice, SAH, SAT, SATr, SAD, and SAM.
H2SO4 • 4 H2O: The Sulfuric Acid Tetrahydrate (SAT) is under stratospheric conditions the
most important hydrate. Although it is supersaturated at temperatures below 240 K it forms
rarely, since its nucleation rate is very low (Koop et al., 1995, 1997b).
H2SO4 • 6.5 H2O: The Sulfuric Acid Hemihexahydrate (SAH) has a lower existence temper¬
ature than SAT. Since SAT will form more readily than SAH as temperatures decreases, the
nucleation of SAH will occur very rarely.
2.1.6 Phase diagram of (NH4)2S04/H20
Gaseous ammonia, NH3, is immediately taken up by liquid H2SO4/H2O particles in which it
reacts to form the ammonium ion, NHJ" (Swartz et al., 1999). In the case of aqueous (NÜ4)2S04the vapor pressure of NH3 is very low, similar to the vapor pressure of H2SO4. Therefore, it
can be assumed that NH3 is also a non-volatile substance under most atmospheric conditions.
Figure 2.2 shows the phase diagram of aqueous (NHi)2S04 in the weight percent scale. The
phase diagram indicates that solid (NH4)2S04 could be thermodynamically stable in a wide
temperature and concentration range.
18 CHAPTER 2. THEORY
300
290
5f 280
s"
«j 270
CDQ.
E
H- 260
250
2400 10 20 30 40 50 60 70 80 90 100
(NH4)2S04 [Wt%]
Figure 2.2: The phase diagram of aqueous (NH^SO^ is shown. The solid lines correspond to the
melting curves of the indicated solids ice and (NH^SO^.
2.1.7 Phase diagram of HN03/H20
Due to the importance of HNO3 in PSC formation the binary HN03/H20-system has been
investigated intensively in the recent years (Hanson and Mauersberger, 1988a,b; Worsnop et al.,
1993; Carslaw et al., 1995a; Massucci et al., 1999; Beyer and Hansen, 2002). Since the vapor
pressure of HNO3 is much higher than the one of H2SO4, no binary HNO3/H2O aerosols exist
under stratospheric equilibrium conditions. Below 196 K HNO3 is taken up by H2SO4/H2Oaerosols forming ternary solution particles. Under non-equihbrium conditions, induced by rapid
temperature changes which can occur in lee wave situations, quasi-binary HNO3/H2O aerosols
with only traces of H2SO4 can form due to a large uptake of HNO3 (Meilinger et al., 1995).
Figure 2.3 shows the phase diagram of aqueous HNO3 in weight percent scale. The followingsolid phases can be found in the binary HN03/H20-system:
HNO3 • H20: The Nitric Acid Monohydrate (NAM). Hanson and Mauersberger (1988a,b) and
Worsnop et al. (1993) show by vapor pressure measurements that the vapor pressure ofNAM is
higher than the stratospheric partial pressures of HNO3. Therefore, NAM does not exist under
stratospheric conditions.
t——1——r ouu
290
(NH4)2S04
(solid)
280
270
260
250
'---'--'-' OAfi
2.1. Thermodynamic processes in UT/LS aerosol 19
300
280
gr 260
"fi
g3 240
<DQ.
E<»r- 220
200
1800 10 20 30 40 50 60 70 80 90 100
HN03 [Wt%]
Figure 2.3: The phase diagram of aqueous HNO3 is shown. The solid lines correspond to the melting
curves of the indicated solids ice, NAT, NAD, and NAM.
HNO3 • 2 H20: The Nitric Acid Dihydrate (NAD). Worsnop et al. (1993) present first
thermodynamic data of NAD. The authors also state that NAD has a potentially lower
nucleation barrier than NAT, i. e. NAD nucleation is favored over nitric acid trihydrate (NAT)formation. Therefore, under stratospheric conditions NAD is always metastable with respect
to NAT. Ji and Petit (1993) investigated the binary HN03/H20-system using Differential
Scanning Calorimetry (DSC). These authors give a melting point of NAD of about 232.7 K.
A newer study of the binary HN03/H20-system by Beyer and Hansen (2002) presents a 2 K
higher melting point for NAD. Tsias et al. (1997) showed theoretically, that NAD can form in
highly HNO3 concentrated (~58 wt%) non-equilibrium solution droplets during strong warming
events in mountain wave PSCs.
HNO3 • 3 H20: The Nitric Acid Trihydrate (NAT). Hanson and Mauersberger (1988a) showed
that this hydrate is the stable form of condensed phase HNO3 under stratospheric conditions
above the ice frost point. The saturation temperature is about 196 K at stratospheric conditions,
i. e. at 50 mbar ambient pressure, 5 ppmv H2O, 10 ppbv HNO3, and 0.5 ppbv H2SO4.
ICE '
NAT \
"
NAD
ham\J J ,
HNO3"
ouu
280
260
240
- 220
- 200
20 CHAPTER 2. THEORY
2.2 Kinetic processes in UT/LS aerosol
Aerosol particles experience phase transitions due to changes in atmospheric conditions, i. e.
temperature and RH variation. A pure liquid aerosol particle has a defined melting point. If
the ambient temperature decreases below the melting temperature, the solid phase is thermo¬
dynamically preferred. Since the nucleation process and, therefore, the phase transition, is a
kinetic process, an aerosol particle can remain liquid, although the temperature is lower than its
melting point. Under these conditions the liquid particle is in a supercooled state, i. e. the liquidis supersaturated with respect to its solid, but nucleation is kinetically hindered. In the case of
pure water a supersaturated state exists when the vapor pressure of the liquid is higher than
the vapor pressure of its solid phase at the same conditions. Temperature can be decreased and
supersaturation in the supercooled liquid can be increased until a critical number of molecules
in the liquid phase form a critical cluster, which is the definition of the nucleation process. In
most cases crystallization of the whole liquid follows immediately after the nucleation occurs.
The nucleation process depends crucially on the supersaturation of the liquid phase with respect
to its solid phase. Nucleation theory can be used to obtain the critical size of the cluster, to
calculate supersaturations required for nucleation, and to derive nucleation rate coefficients. The
latter define the expected start of the crystallization process for given atmospheric conditions
and aerosol particle size.
The following sections deal with the derivation of the nucleation processes which occur in super¬
saturated aerosol droplets. The first section presents classical nucleation theory (CNT), in which
the nucleation rate scales with the volume of the investigated particle. The second section deals
with a newly proposed pseudo-heterogeneous nucleation mechanism (Tabazadeh et al., 2002a,b;
Djikaev et al., 2002), which assumes that nucleation is induced at the surface of the particle.
Thus, the derived nucleation rates scale with the surface of the particle.
2.2.1 Classical nucleation theory
Here, the classical nucleation theory is briefly presented. There axe two common approachesto derive CNT: the kinetic approach and the constrained equilibrium approach (Seinfeld and
Pandis, 1998). The first develops CNT by determining the rates of collision of monomers, i. e.
monomers sticking together and forming a critical cluster in the liquid, and the rates of hittingand disengaging monomers within the liquid. From these processes, a nucleation rate coefficient
is obtained. The constrained equihbrium approach treats the formation of a critical cluster by
determining its Gibbs free energy of formation, which also allows to derive a nucleation rate
coefficient. In this section CNT will be described using the constrained equilibrium approach.
Homogeneous nucleation occurs exclusively in a supersaturated environment. In the case of a
pure liquid X, e. g. water, the liquid phase activity of X corresponds to the saturation ratio S:
liq
* = &• (2-19)Px
2.2. Kinetic processes in UT/LS aerosol 21
where p£ is the vapor pressure of the pure liquid and p1 is the vapor pressure of its solid at
the same conditions. When a liquid is saturated with respect to its solid then 5 = 1.
Homogeneous nucleation depends on the transfer of molecules from the liquid to the solid phase,
i. e. the increase of the number of molecules to a critical cluster size. If 5 is sufficiently large, this
critical cluster size will be exceeded and a new phase starts growing. A transfer of i molecules
from the liquid phase forms an i-zner cluster of radius r. The corresponding change in the Gibbs
free energy is
AGi = (ßsoi - imq)i + 4irasoir2 , (2.20)
where /isoj and nuq are the chemical potentials of the solid and liquid phase, respectively. asoi is
the surface tension between the solid and liquid phase and r is the radius of the critical cluster.
The first term in Eq. 2.20 describes the Gibbs free energy of the transfer of a molecule to the
cluster and the second specifies the Gibbs free energy of forming an interface. The number of
molecules in the solid, i, can be obtained by
« =^
= ^, ("I)
where Vsoi is the volume of the critical cluster and vsoi is the volume of one i-mer in the solid.
The difference in the chemical potentials can be expressed with respect to the corresponding
vapor pressures:
mq-»soi = kTmS (2.22)
= *ThÄ, (2.23)Px
where k is the Boltzman's constant. The Gibbs free energy change of i-mer formation can now
be written using Eq. 2.23 and Eq. 2.21 within Eq. 2.20 as
a ^ 24irkTlnS o
,nnA\
AGi = 4ivasolr2 - r3. (2.24)
3 Vs0l
The first term describes the Gibbs free energy increase due to the formation of a surface of the
critical cluster. The second term is the Gibbs free energy decrease due to the transfer of molecules
from the liquid to the solid phase. This equation assumes that the critical cluster has the same
properties as the bulk. Typically this is known as capillarity approximation (Seinfeld and Pandis,
1998; Defay and Prigogine, 1966). Also, the surface tension, asoi, is not known for such a small
critical nucleus. It is only known for some pure substances at the melting temperature of the
bulk crystal, whereas nucleation occurs at lower temperatures in the supersaturated regime.
These two assumptions are still a controversial subject in the formulation of CNT.
Figure 2.4 gives an example for AG as a function of r for different values of 5. In a subsaturated
environment (5 < 1) the Gibbs free energy increases with increasing cluster size r due to the
formation of a surface. For supersaturated conditions (5 > 1) AG increases initially with r due
to the formation of a surface but is compensated at a critical cluster size r* by the decrease of
AG due to the formation of the solid. This critical i-mer radius can be derived by
r* =
2asolVso1(2.25)
22 CHAPTER 2. THEORY
\ /s<i
<
AG/
/S,>1
/ ^^ !^
AO,* ^^\^\\|
1 \ \1 ! \Ste>S,\
r2*
Figure 2.4: The Gibbs free energy difference due to the formation of a critical cluster as function of
saturation ratio S and critical cluster size, r*.
The Gibbs free energy at r* is obtained by substitution of Eq. 2.25 in Eq. 2.24, which yields
*_
16tt vfo-3solAG ~
~MJfcTln5)2-(2-26>
Thus, an increase in 5 decreases the Gibbs free energy barrier, AG*, and the critical i-mer
radius, r* (Eq. 2.25), i. e. less molecules for the formation of the critical cluster are necessary
(see Fig. 2.4). If the nucleus is smaller than r* it will dissociate at once because of the increase
of AG. If the critical cluster reaches the size r* nucleation will start, because of the continuous
decrease of AG with an additional i-mer. Therefore, the nucleation barrier depends on the Gibbs
free energy for the formation of a critical cluster, AG*, and on the necessary energy to transfer
molecules to the cluster. Thus, the activation energy can be written as given by Turnbull and
Fisher (1949):
AGact(T) = AG*(T) + AGdi/(T), (2.27)
where AGdif is the molar Gibbs free energy of activation for diffusion of molecules across the
liquid-solid boundary. From this, the homogeneous nucleation rate coefficient in units cm-3 s-1
can be derived:
AGactÇr)-kTJhorn = nuq—exp
RT(2.28)
where n^ is the molecular number density of the species and h is the Plank's constant. Although
osoi and AGdif are not known, AGact can be obtained by
AGact(T) _ JnJhomnliqh
„jiKJ-
(2.29)
2.2. Kinetic processes in UT/LS aerosol 23
using experimentally derived //«„„-values.
CNT has the advantage that it is convenient to apply to experimental data due to the simple
underlying mathematical expressions. The disadvantage of CNT is the assumption that the
critical cluster behaves like the bulk. Therefore, CNT uses o~soi and AGdif which are obtained
from macroscopic samples. These quantities are often not available (MacKenzie, 1997).
2.2.2 Surface nucleation
Djikaev et al. (2002) and Tabazadeh et al. (2002a,b) propose a new idea of the nucleation
mechanism of aerosol particles. The authors claim to have found evidence that nucleation
starts at the droplet surface, i. e. at the air-solution interface (Djikaev et al., 2002; Tabazadeh
et al., 2002a,b). Here, the arguments for this pseudo-heterogeneous phase transformation or
surface-induced nucleation, suggested by the authors are presented and the derivation of the
pseudo-heterogeneous nucleation rate coefficients are shown.
fp liq ^svap
a- a"sol
Figure 2.5: Sketch of a liquid (liq) that rests on its solid (sol) and is surrounded by its vapor (vap).
The three interfaces are vapor-solid (vs), liquid-solid (Is), vapor-liquid (vl). The corresponding surface
tensions are avs, au, and avi, respectively.
Figure 2.5 shows a liquid droplet sitting on its solid for an one-component system. aV3, ats, ovi
correspond to the surface tensions of the vapor-solid, liquid-solid, and vapor-liquid interfaces,
respectively. The contact angle, 6, is defined as
=, (2.30)
&vl
which is also known as Young's relation (Pruppacher and Klett, 1997). For 6 > 0° one follows
from Eq. 2.30, that:
Ovs < °~vl + o~ls • (2-31)
Equation 2.31 indicates the condition of partial wetting of a solid by its liquid.
Figure 2.6 shows a crystal forming at the surface of its liquid. If we use Young's relation on the
facet of the crystal which is in contact with the vapor phase we obtain the same inequality given
by Eq. 2.31. Therefore, surface nucleation would be favored if at least one of the facets of the
24 CHAPTER 2. THEORY
vap
Figure 2.6: Sketch of a crystal nucleus (sol) that forms at the surface of its liquid (liq). The facet in
contact with the vapor (vap) is indicated by a dashed line. The solid lines represent the contact of the
crystal with its liquid. The corresponding surface tensions are avs, ois, av\ similar to Fig. 2.5. This
figure is adapted from Djikaev et al. (2002) and Tabazadeh et al. (2002a).
crystal is only partially wettable by its own melt. The authors (Djikaev et al., 2002; Tabazadeh
et al., 2002a,b) state that Eq. 2.31 is the pre-requisite for the occurrence of surface-induced
nucleation.
Equation 2.28 gives an expression for the derivation of the homogeneous rate coefficient within
the volume of a liquid. Tabazadeh et al. (2002a) present a similar relation for the pseudo-
heterogeneous nucleation rate coefficient of a nucleus on the droplet surface:
hornKT
kTNs— exp
h
&Gsact(T)RT
(2.32)
where Ns is the total number of molecules per unit surface of the liquid. The authors relate the
surface-based to volume-based homogeneous nucleation rate coefficients in the following way:
Jhom — Wt/St) Jfiom > (2.33)
where Vt is the total aerosol volume and St is the total surface area of an observed aerosol
ensemble. A monodisperse distribution of droplets yields the following relation:
Tshorn (r/3)JL». (2.34)
where r is the droplet radius.
From these presented equations one can easily convert the experimentally derived volume-based
Jftom-values into surface-based J^om-values.
2.3 Raman spectroscopy
The Raman effect in liquids was discovered in 1928 by Raman and Krishnan (1928). Almost
simultaneously Landsberg and Mandelstam (1928) observed the Raman effect in crystals. The¬
oretically, the Raman effect was already 1923 predicted by Smekal (1923). Placzek (1934)
2.3. Raman spectroscopy 25
developed a semi-quantum mechanical theory of the Raman effect. It abandons the quantum
mechanical treatment of the light, but regards the molecule energies in a quantum mechanical
way. This theory still holds for most of the common Raman spectroscopic applications.
The Raman effect is an inelastic scattering process of two photons. If a photon with energy, Erj
e| s,
/?v0 hvi
i
hv0 hvi
Figure 2.7: Term diagram of the inelastic scattering process. On the left hand side the Stokes scattering
process is shown and on the right hand side the Anti-Stokes scattering process is shown (Schrader, 1995).
= Iivq, hits a molecule the elastic scattering process, i. e. Rayleigh-scattering, which emits the
same energy quantum, Erj, as the incoming photon, has the highest probability to occur. The
inelastic scattering process, i. e. the Raman-scattering, where an exchange of vibrational energy
takes place, has a much lower probability. The Raman-scattering process emits the energy quan¬
tum of hvQ =f hvs- Figure 2.7 shows the principle of Raman scattering. At ambient temperature
most molecules occupy their vibrational ground state at No. By absorbing the energy quanta
hvQ, the molecules reach an excited state. These molecules emit energy quanta hu^, which is
lower by hi>s than the incoming energy quanta, leaving the molecules in an excited state (Ng)little higher than their vibrational ground state (No). If the excited molecules have emitted the
energy quanta of hi>o the molecules would have reached again their ground vibrational state.
This process would correspond to Rayleigh-scattering. The emittance of the energy quantum
hVft is called Stokes-Raman scattering, since Stokes postulated in 1852 that the light produced
by fluorescence has always a longer wavelength than the excitation wavelength. Therefore, the
energy of the Stokes-Raman intensity line can be written as
Er — huR — hvo — hvs (2.35)
According to Boltzmann's law, even at ambient temperature there is still a small number of
molecules in the vibrationally excited state, Ns. If an energy quantum of hvo hits the already
excited molecules they reach a higher excited state and can emit the energy quanta hv^ to reach
their vibrational ground state. This transition is called Anti-Stokes-Raman-scattering, because
it exhibits a larger frequency and, hence, a higher amount of energy than the corresponding
Stokes-Raman-scattering and Rayleigh-scattering, respectively. The energy quanta of the Anti-
26 CHAPTER 2. THEORY
Stokes-Raman intensity fine is
Eft = huR — hvQ + hus (2.36)
Figure 2.8 shows the intensity distribution of the Rayleigh-scattering process (at z^o), the Stokes-
line (at v^), and the Anti-Stokes-line (at u^). Since the highest probability is obtained for the
elastic scattering process, the Rayleigh-line is the most pronounced signal. The Anti-Stokes-
line is the lowest signal, since at ambient temperature, the number of molecules in the excited
state are much lower than the number of molecules in the vibrational ground state. Since the
occupation of the vibrational ground and excited states depend on the temperature-dependentBoltzmann factor, the temperature can be obtained from the ratio of the Anti-Stokes-line and
Stokes-line.
Stokes
— vs —
Anti-Stokes* vs
Vr v0 Vr
Figure 2.8: The intensity distribution in a Raman spectrum as function of frequency and scattering
process. From left to right: Stokes-scattering, Rayleigh-scattering, and Anti-Stokes-scattering (Schrader,
1995).
A change in the excitation wavelength will change the resulting Raman spectrum. Also, impuri¬
ties inside the sample can cause fluorescence which can strongly perturb the Raman spectrum.
The reason for this is the relatively small Raman-scattering cross section compared to the elastic
Rayleigh scattering cross section. Therefore, highly monochromatic and powerful light sources
axe needed for Raman spectroscopy. Optical niters of high quality are also necessary to cut
off the Rayleigh scattered light, whose intensity can be up to a factor of 106 higher than the
scattered Raman light.
2.3.1 Classical derivation of the Raman effect
The following derivation of the Raman effect is given by Kiefer et al. (1995) and Schrader (1995)which is based on the semi-quantum mechanical theory of Placzek (1934).The classical approach of light scattering is based on the idea that a dipole moment in a liquid is
induced by the electromagnetic field of fight. This dipole moment oscillates with the frequency of
the incoming electromagnetic field. Therefore, the oscillating dipole moment can be considered
as a source for the emittance of electromagnetic radiation. This secondary emitted radiation is
2.3. Raman spectroscopy 27
distributed over an angle of 4n. The incoming electromagnetic field consists of a magnetic part
and an electric part. The latter one will be considered in this derivation. The incoming electric
field, E, oscillates with the frequency, vq, and can be expressed as:
Ê = Éo- cos(27ri/0t), (2.37)
where Eq is the amplitude of the electric field. E induces a dipole moment, ß, of
ß = a-E, (2.38)
where a is the polarizability of the molecule. From Eq. 2.37 and Eq. 2.38 one follows
ß = a-EQ- cos(27ri/0t). (2.39)
The polarizability, a, is a tensor which projects one component of the vector E to produce
the corresponding component of ß. a depends on the molecule symmetry and also on inter¬
nal molecule vibrations. The flexibility of electrons and nuclei in a molecule depends on their
mutual distance. If this distance is small, an external field has only a small influence on the
positions of the electrons and nuclei, and a large influence, if their distance is large. Therefore,
the polarizability of a molecule can be modulated by the vibration of the incoming light. The
emitted radiation depends also on the polarizability and, hence, on the vibration. If the excita¬
tion wavelength deviates significantly from the resonance wavelength of the molecule, a can be
expanded in a Taylor series with respect to the normal coordinates qk at its equilibrium position
qk = 0 and aborted after the hnear element. (The assumption of an expansion of a in a Taylor
series is the main point of Placzek's theory (Placzek, 1934), since the Schrödinger equation, i.
e. the quantum mechanical treatment, cannot be solved for all wave functions of this scattering
problem.) The expansion in a Taylor series yields:
3Q-/ ,~ v
Oij = «auk = 0) + £ ^ qk, (2.40)fc=1
\ °Qk J qk=Q
where aiij(qk = 0) is the polarizability in the equilibrium position. Q is the number of the atoms,
3Q — / is the number of normal vibrations. / depends on the molecular geometry, e. g. / = 5 for
a linear molecule and / = 6 for a non-linear molecule. In the case of small molecule vibrations
the normal coordinates qk can be approximated by a harmonic oscillation:
qk = qkO- cos(2irukt), (2.41)
where g^o is the amplitude of the fcth oscillation and vk is the frequency of the fcth oscillation.
From equations 2.39-2.41 one obtains
M
k v%A,=o'<x(qk = 0) + Y] ( -£— 1 qko • cos(27Ti/fci) Ê0cos(2nvot). (2.42)
28 CHAPTER 2. THEORY
Applying trigonometrical transformation yields
ß = a(qk = 0) • Êq cos(27ti/o*) +%
v'
Rayleigh — scattering
qk0Êo cos(27r(z/fco - vo)t) +9fe=0
Stokes — Raman — scattering
qk0Ê0 cos(27r(i/fe0 + ^o)*) • (2.43)9fc=0
Anti — Stokes — Raman — scattering
The fraction of the emitted light can be identified. Light with the frequency of vq corresponds to
Rayleigh-scattering and fight with the frequency of Vko — vq and ukQ + vq corresponds to Stokes-
and Anti-Stokes-Raman-scattering, respectively. Equation 2.43 indicates that Raman scattering
only occurs when the polarizability a changes during an oscillation through the equilibrium
position, i. e.
(£)*" <
Therefore, a molecular vibration can be observed in the Raman spectrum only if there is a
modulation of the molecular polarizability by the vibration. From this, one can conclude which
molecules are Raman-active and which molecules axe Raman-inactive.
l^p/dcA
2-MaJ
Chapter 3
Experimental
All experiments performed in this work have made use of droplet samples. The following sections
describe all necessary steps to generate a droplet sample which is used in the Raman spectro¬
scopic measurements and nucleation experiments. This includes the preparation of the aqueous
solutions, the droplet production, the sample holder (i. e. the droplet cell), and the experimental
setup. In the last section of this chapter the typical experimental procedures are illustrated.
3.1 Sample preparation
The solutions investigated in this work are mixtures of H2SO4/H2O, (NH4)2S04/H20,
HNO3/H2O, and HNO3/H2SO4/H2O. The H2S04 and HNO3 containing solutions were
prepared from stock solutions which were titrated against 1 M NaOH. The (NHi)2S04 solutions
were prepared from solid (NH4)2S04 and Milhpore water (Resistivity > 18.2 Mfi-cm). In
addition, the solutions were filtered through a 0.2 /im pore size membrane to exclude insoluble
impurities.These solution were used to produce two types of droplet samples. On the one hand relatively
large droplets with diameters of about 0.4-1.5 mm, and on the other hand small droplets
with diameters of about 10-50 /im. The whole droplet production was performed inside a
laminar flow clean bench to minimize the contamination with dust particles. The large droplets
were produced using a micropipette. The small droplets were generated by an atomizer or an
inkjet-cartridge:
Figure 3.1 shows a homemade atomizer made of glass, since acidic solutions (H2SO4/H2O, or
(NHi)2S04/H20 solutions with typical concentrations of about 20 wt%) are used within the
droplet production. Gaseous nitrogen with a pressure of about 0.4 bar flows over a nozzle,
thereby, drawing the aqueous solution through the capillary and creating numerous droplets
with a mean diameter of about 0.7 /im (Knopf et al., 2001). The second vessel serves as a
precipitation reservoir for larger drops. The particles will be deposited onto a hydrophobicallycoated glass plate. This plate, 13 mm in diameter, is immersed for a few seconds in the aerosol
29
30 CHAPTER 3. EXPERIMENTAL
flow. The resulting droplet diameters can be changed by varying the nitrogen pressure, the
concentration of the aqueous solution, and the contact time of aerosol and plate. The higher the
amount of the non-volatile substances in the solution the larger the diameter of the droplets for
a given relative humidity. This method generates a high number density of deposited particles
with average diameters between 5-200 /im.
œrosolsproy
reservoir \X
nozzle
=4-aerosol drain
quartz
plate
aerosol
generator
Figure 3.1: Sketch of the atomizer.
A lower number density of deposited droplets - or even single droplets - are obtained by the
operation of a modified Hewlett-Packard inkjet-cartridge. The setup is shown in Fig. 3.2. The
plate is placed on a micrometer stage. A few millimeters above this stage the inkjet-cartridgeis fixed. By adjusting the micrometer stage and by operating the pulse generator in single
pulse mode defined droplet arrays can be generated. Furthermore, the inkjet-cartridge can be
run in a 100 Hz mode, hence, producing many more droplets. Preferentially, very dilute acidic
inkjet-cartridge
aerosol
quartz plate
micrometer stage
Figure 3.2: Sketch of the single droplet generator using an inkjet-cartridge.
3.2. Sample cell 31
solutions (up to a maximum of 5 wt%) are used within the inkjet-cartridge, due to the danger of
solubilization of the inkjet-cartridge surfaces. The inkjet-cartridge produces droplets of about 58
/im in diameter (Diiwel, 2003), whose diameters can be changed by varying the concentration of
the solution for a given relative humidity. The electric circuit to operate such an inkjet-cartridge
is given in appendix A.l.
3.2 Sample cell
An o-ring or a Teflon washer (for the large droplets) or an aluminum foil (for the small droplets)treated with high-vacuum-grease served as a spacer for a second quartz plate, which sealed
the droplets against ambient air. Model calculations have been performed which show that
the amount of gas-phase water inside the small cell is negligible compared to the liquid-phase
water of the droplets. Therefore, no concentration changes inside the droplets occur due to
evaporation and condensation of gas-phase water from the droplet cell to the droplets. Hence,
the composition of the droplets in the presented experiments remains fixed. This kind of
sample, consisting of two glass plates separated by a spacer will be called aerosol cell or droplet
cell in the remaining parts of the thesis.
The investigation of the ferroelectric phase transition of solid (NrLi)2S04 requires a different
sample preparation. A very diluted aqueous (NH4)2S04 solution was spread out on the the
quartz plate. The plate was placed on the temperature stage and was warmed to about 330 K,
thus, evaporating the liquid water. A thin layer of solid polycrystalline (NH4)2S04 remained
on the plate. Afterwards, the cell was sealed by an aluminum foil and a second quartz plate to
avoid water uptake from the ambient air.
3.3 Hydrophobic coating
The droplets were deposited on a hydrophobically coated glass or quartz (Herasil) plate. Herasil
quartz plates were used for Raman spectroscopic measurements due to the large transmission
efficiency of about 92 % in a wavelength range of 300-1000 nm, which coincides with the exci¬
tation wavelength used in the presented experiments. The effect of the surface on the droplets
can be minimized by coating the glass surface with a hydrophobic substrate. In this work three
different substrates were tested for the production of a hydrophobic coating:
1. Silanization Solution I: 5 % Dimethlydichlorosilane in heptanefrom Fluka BioChemika
2. Aqua Sil: Organosilane concentrate
from Hampton Research
3. OTS: Octadecyltrichlorosilanefrom Aldrich
The glass plates were purified by immersion into Caro's acid (prepared from a 2.5/1 mole ratio
32 CHAPTER 3. EXPERIMENTAL
of H2SO4/H2O2 using 93 wt% H2S04 and 35 wt% H202) for a period of one day.
In the case of Silanization Solution I the formation of the substrate occurred via gas-phase
reaction. The dry plates rest in a closed container for two days which is flooded by gaseous
Dimethlydichlorosilane. Afterwards, they were flushed with water and dried.
The preparation with Aquqa Sil was as follows: 1 ml of Aqua Sil was dissolved in 100 ml water
(Milhpore water, Resistivity > 18.2 Mfi-cm). The plates were dipped for a few seconds into the
solution and then flushed with water.
In the case of OTS a 1 mmol solution of OTS in chloroform was prepared. Then, the plates
were brought for a few seconds into contact with the solution and then flushed with water.
The quality of the hydrophobic coating was determined by placing a droplet with known volume
on the coated glass plate and measuring the droplet diameter. The more the diameter of the
placed droplet approached the diameter of a sphere with similar volume, the higher the quality
of the hydrophobic coating. Another quality test of the hydrophobic coating was performed by
investigating the homogeneous nucleation of pure water droplets. The quality of the hydropho¬
bic coating was considered to be good when pure water droplets reached the expected degree of
supercooling with respect to their size (Pruppacher and Klett, 1997), hence, indicating that no
heterogeneous nucleation due to the contact with the substrate occurred.
The highest quality of the hydrophobic coating was achieved by using the Silanization Solu¬
tion I, followed by OTS and Aqua Sil. Therefore, all glass/quartz plates used in the presented
experiments were treated by the Silanization Solution I.
3.4 Experimental setup
Figure 3.3 shows the experimental setup. It consists of two main parts: a Confocal Raman
Microscope (Jobin Yvon, model: Labram) and a homemade temperature stage.
The Confocal Raman Microscope allows the visual observation of phase transitions of the
droplets. The instrument is equipped with several objectives with a magnification of 10, 50, and
100. Raman spectra of the droplets can also be recorded. The Raman microscope is equipped
with a Nd:YAG-laser which is operated at a wavelength of 532 nm and has a maximum power
of 100 mW for illumination. The laser power can be changed by inserting filters with different
optical densities. The light backscattered from the sample is passed onto a grating of 1800
mm-1 and focused on the CCD detector of the spectrograph. This yields a spectral resolution
of about 2-4 cm-1 within the observed range of 500-4000 cm-1.
Figure 3.4 shows a sketch of the temperature stage. Cooling of the temperature stage is
achieved through the evaporation of liquid nitrogen and the counterheating is generated by a
heating foil. Liquid nitrogen enters the small chamber below a heating foil and a copper plate.
The liquid N2 evaporates and, hence, cools the whole temperature stage. Gaseous nitrogen
leaves the temperature stage through a valve and is pumped away by a membrane pump. By
adjusting the orifice of the valve the amount of pumped gaseous N2 is controlled and, thus,
the cooling capacity. The droplet cell is placed inside the depression of the copper plate. A
resistance temperature sensor (Pt 100) is fixed onto the copper plate close to the droplet cell.
The whole temperature stage is surrounded by a box which is flushed with gaseous nitrogen
3.4. Experimental setup 33
laser
video
analysis
spectrographCCD detector
^532 nm
microscope
temperature stagear
grating:1800 g/mm
droplet cell
Figure 3.3: Sketch of the experimental setup.
to prevent ice condensation on the droplet cell and the apparatus. The heating foil and
the temperature sensor axe connected with a low temperature controller (LTC-11, Neocera).The LTC-11 allows to maintain a constant temperature or to perform definite temperature
ramps of the cooling stage. The LTC-11 is controlled by a Windows-based computer using
Hewlett-Packard Visual Engineering Environment (HP VEE) which is a graphical programming
language optimized for instrument control (Helsel, 1995). The complete experimental run
including a serial of heating and cooling ramp can be set. Experiment time and actual droplet
temperature are transferred to the computer and saved on harddisk every 0.1 seconds. This
data is also displayed on the monitor during the experiment using a Videotext overlay module
(Engineering, 2001). The temperature stage is able to vary the temperature of the sample
in a range of 160-350 K. The temperature of the sample is measured by the Pt 100 sensor
whose linear resistance/temperature response was confirmed by measuring the melting points,
Tm, of heptane (TTO=182.55 K), octane (Tm=216.35 K), decane (Tm=243.45 K), dodecane
(Tm=263.5 K), and water (Tm=273.15 K). These substances were put into the droplet cell
and the melting points were determined using a heating rate of 1 K/min. The statistical mean
of 10 melting point measurements were used for the calibration. As expected the calibration
data reveals a hnear relationship between the measured resistance and the temperature. The
calibration error was obtained by error propagation law analysis of the data and the fit function.
The temperature accuracy of the homemade temperature stage is better than ±0.1 K. A
dynamic calibration was also performed, i. e. the change of nucleation temperature of the above
mentioned substances (heptane, octane, decane, dodecane, water) was investigated by variation
of the cooling rate. The freezing temperatures show no deviation for cooling rates in the range
of 1-20 K/min. Most experiments were performed with a cooling rate of maximum 10 K/min.
34 CHAPTER 3. EXPERIMENTAL
objective
heating foilPt-100
shielding box
evaporation chamberniiài
copper plate
light source
Figure 3.4: Sketch of the temperature stage.
3.5 Experimental procedure
Two different types of experiments were performed in this study: the record of temperature-
dependent Raman spectra and nucleation experiments.The first kind of experiment was performed in the following way: At a fixed temperature a
Raman spectrum was recorded. Afterwards, the sample was either cooled or heated with a
rate of 10 K/min to a new temperature 5-10 K lower or higher, respectively, than the previous
temperature. As the temperature was reached a new Raman spectrum was recorded.
The experimental procedure for the nucleation experiments was performed in the following way:
A cooling ramp with a maximum cooling rate of 10 K/min was conducted until the nucleation of
all droplets had occurred. The whole experimental run is recorded on a video tape. The actual
temperature of the droplets and the experimental time is displayed on the monitor and on the
video tape. The tapes were analyzed ex post, i. e. the droplet diameters and, hence, the volume
and the surfaces of the droplets, the nucleation temperature, and the cooling rate were noticed
for further analysis.
Chapter 4
Thermodynamic processes in UT/LSaerosol particles
This chapter presents an investigation of the ionic speciation in aqueous H2SO4 under thermo¬
dynamic equilibrium conditions and relevant atmospheric temperatures. The dissociation of the
bisulfate ion, HSOJ, is analyzed using Raman spectroscopy. The dissociation data obtained at
low temperature is implemented into a Pitzer model to derive a more consistent thermodynamic
model of the H2S04/H20-system. This leads to the derivation of a new thermodynamic disso¬
ciation constant of the bisulfate ion for a temperature range of 180 K to 473 K. Sections 4.1
to 4.7 axe identical to the publication "Thermodynamic Dissociation Constant of the Bisulfate
Ion from Raman and Ion Interaction Modeling Studies of Aqueous Sulfuric Acid at Low Tem¬
peratures" reproduced with permission from the Journal of Physical Chemistry Part A, 107,
4322-4332. Unpublished work copyright 2003 American Chemical Society.Section 4.8 describes in further detail the changes of HCl solubility in aqueous H2SO4 aerosol
particles due to implementation of the new thermodynamic dissociation constant of the bisulfate
ion.
The last three sections of this chapter present further analysis of H2SO4/H2O and (NH4)2S04Raman spectra. This involves the effects of temperature and water uptake on the composition
and phase of the investigated solutions.
35
36 CHAPTER 4. THERMODYNAMIC PROCESSES
Seite Leer /
Blank leaf
37
Thermodynamic Dissociation Constant of the Bisulfate Ion from Raman
and Ion Interaction Modeling Studies of Aqueous Sulfuric Acid at Low
Temperatures
D. A. Knopf*, B. P. Luo, U. K. Krieger, and Thomas Koop
Institute for Atmospheric and Chmate Science, Swiss Federal Institute of Technology, Hongger-
berg HPP, 8093 Zurich, Switzerland
* To whom correspondence should be addressed. Email: Daniel.Knopf@iac.umnw.ethz.ch.
Reproduced with permission from the Journal of Physical Chemistry Part
A, 2003. Unpublished work copyright 2003 American Chemical Society.
38 CHAPTER 4. THERMODYNAMIC PROCESSES
Seite Leer /Blank leaf
4.1. Abstract 39
4.1 Abstract
The dissociation reaction of the bisulfate ion, HSOJ ^ SO4- + H+, is investigated in aqueous
H2SO4 solutions with concentrations of 0.54-15.23 mol kg-1 in the temperature range of 180-
326 K using Raman spectroscopy. All investigated H2SO4 solutions show a continuous increase
in the degree of dissociation of HSOJ with decreasing temperature, in contrast to predictionsfrom thermodynamic models of aqueous H2SO4 solutions. A Pitzer ion interaction model is
used to derive a thermodynamically consistent formulation of the thermodynamic dissociation
constant of the bisulfate ion, Ku(T), that is in agreement with the experimental data. The
new formulation of Ku(T) is valid from 180 K to 473 K. All ion interaction parameters and the
corresponding parametrizations of the Pitzer ion interaction model are presented. Calculations
with this model reveal significant differences in ion activity coefficients, water activities, water
vapor pressure, and HCl solubilities, when compared to existing thermodynamic models of
H2SO4/H2O solutions, in particular at lower temperatures.
4.2 Introduction
Aqueous sulfuric acid (H2SO4) is one of the most important mineral acids in chemical industries
(Donovan and Salamone, 1983). Because of this, its thermodynamic properties such as partial
pressures as well as osmotic and activity coefficients were intensively studied over the past
decades (Rard et al., 1976; Staples, 1981; Bolsaitis and Elliott, 1990; Zeleznik, 1991). In the
atmosphere, sulfuric acid affects many properties of ambient aerosols. Stratospheric backgroundaerosols consists of highly concentrated aqueous sulfuric acid droplets (Junge and Manson, 1961;
Hamill and Toon, 1991). Tropospheric aerosols can contain mixtures of various inorganic and
organic species but H2SO4 is often a major component (Murphy et al., 1998). Furthermore, the
solubihty of volatile gases such as HCl and NH3 in liquid aerosols depends on the concentration
of dissolved H+-ions (Luo et al., 1994; Swartz et al., 1999) which, in turn, depends on the degree
of dissociation of H2SO4. The dissociation of H2SO4 is a two-step process:
H2S04 ^ HS04+H+ (I)
HSO4 ^ SO|- + H+ (II)
It has been shown that the dissociation of H2SO4 is essentially complete for concentrations
up to 40 mol kg-1 at temperatures between 273 K and 323 K (Young et al., 1959). These
measurements further suggest that full dissociation occurs also at lower temperatures at these
concentrations. On the other hand, the dissociation of the bisulfate ion, HSO4 , depends strongly
on temperature (Young et al., 1959; Chen and Irish, 1971; Dawson et al., 1986; Dickson et al.,
1990; Tomikawa and Kanno, 1998). The thermodynamic dissociation constant of the HSOj-ion,
K\\(T), is defined by the activities of the particular ions (see Appendix 4.7.2 for a full derivation
oftfii(r)):
aH+(T)ao02-(T)*n(T) =
n S%, (4.1)
40 CHAPTER 4. THERMODYNAMIC PROCESSES
mn+ (T)mS02- (T) 7h+ (r)7s02- (T)
»»hsoj-O' 7hso4-C0Q(T) • 7(D, (4.3)
where aî7 m*, and ji denote the activity, molahty, and activity coefficient of ion i (i = H+,
SO^-, HSOJ) in equilibrium, m) is by definition 1 mol kg-1, and Q(T) and -){T) are the molal
dissociation quotient and activity coefficient product, respectively.
Knowledge of Kn(T) is a prerequisite for the description of the thermodynamic properties of
a multicomponent system containing H2SO4, such as the NH3/H2SO4/H2O system at atmo¬
spheric temperatures («180-300 K). However, there are only few experimental studies of the
properties of aqueous H2SO4 at low temperatures Zhang et al. (1993b); Massucci et al. (1996);Das et al. (1997); Tomikawa and Kanno (1998). Therefore, thermodynamic solution models axe
employed to predict ion activity coefficients and ion concentrations in aqueous solutions at low
temperatures in a consistent way. One widely used model of this kind is the aerosol inorganics
model (AIM) (Clegg et al., 1998) which is based on the Pitzer ion interaction approach (Pitzer,
1991). Since low-temperature data on the dissociation of sulfuric acid are not available, the
formulation of the thermodynamic dissociation constant Ku(T) that has been implemented in
the AIM model (Clegg et al., 1994, 1998), is the one taken from Dickson et al. (1990), who
derived Kii(T) from measurements in the temperature range of 298-523 K.
One way to experimentally investigate the dissociation of the bisulfate ion is Raman spec¬
troscopy. Raman data of aqueous H2SO4 solutions in the temperature range of 278-328 K
Young et al. (1959) axe available but the only existing Raman study at temperatures below
273 K focused on H2SO4/H2O solutions in the glassy state and investigated the dissociation
of the bisulfate ion in more detail only for a solution 4.37 mol kg-1 in concentration to 233 K
Tomikawa and Kanno (1998).In this paper we present new experimental data on the degree of dissociation of the bisulfate
ion in sulfuric acid solutions derived from Raman spectroscopic measurements at concentrations
of 0.54-15.23 mol kg-1 and temperatures of 180-326 K. We use a Pitzer ion interaction model
(Pitzer, 1991) to derive a thermodynamically consistent formulation of K\\(T) which is in agree¬
ment with the experimental data. The ion activity coefficients, 7H+, and 7so2-, 7^+ •
7So2- >
and water activity, aw, for selected H2SO4 solutions are calculated using the new formulation of
Kji(T). These results axe compared to values derived by the AIM model of (Clegg et al., 1998).
4.3 Experimental Section
Figure 4.1 shows the experimental setup. Raman spectra of aqueous droplets axe obtained using
a confocal Raman microscope (Jobin Yvon, model: Labram) operated with a Nd:YAG-laser at
a wavelength of 532 nm and a power of 25-100 mW for illumination. The backscattered light is
passed onto a grating (1800 mm-1) and focused on the CCD detector of the spectrograph. The
resulting spectral resolution is about 2-4 cm-1 within the observed range of 500-4000 cm-1.
A homemade temperature stage is attached to the microscope table. The temperature of the
stage can be varied between 180 and 326 K. The temperature was measured using a resistance
4.3. Experimental Section 41
0
video
analysis
laserspectrographCCD detector
\ I
532 nm
microscope
x_ -7
grating:1800 g/mm
droplet cell
I-JJMtemperafure stage
Figure 4.1: Sketch of the experimental setup.
temperature sensor (Pt 100) whose linear resistance/temperature response was confirmed by
measuring the melting points of heptane, octane, decane, dodecane, and water. Phase changes
(i.e. freezing or melting) are observed visually with the microscope part of the setup.
Table 4.1 shows the composition of the investigated solutions of H2SO4/H2O and
(NH4)2S04/H20. The H2SO4 solutions were prepared from stock solutions which were
titrated against 1 M NaOH. The (NH4)2S04 solutions were prepared from solid
(NH4)2S04 and Milhpore water (Resistivity > 18.2 Mfi-cm). In addition, the solutions were
filtered through a 0.2 /im pore size membrane. The volume of the droplets varied between 0.5
and 10 /xL (diameters of about 0.1-0.26 cm). The droplets were deposited with a micropipet on
a silanized (hydrophobic) quartz plate inside a laminar flow clean bench. Either an O-ring or
a Teflon washer treated with high-vacuum grease served as a spacer for a second quartz plate,
which sealed the droplets against ambient air. Afterward the droplet cell was placed on the
temperature stage.A typical experiment started by taking a Raman spectrum at room temperature. Subsequently,
the droplet was either cooled or heated in temperature steps of 5-10 K (at a rate of 10 K min-1)and a new Raman spectrum was taken at each temperature.
To exclude any possible bias in the temperature and composition of the droplet due to the en¬
ergy transfer from the laser light, a sensitivity study was performed. At a fixed temperature of
298 K, Raman spectra were recorded, in which a droplet was exposed to different illumination
times and laser intensities. Figure 4.2 shows 12 Raman spectra of a H2SO4/H2O droplet 10
fiL in volume with a concentration of 4.37 mol kg-1. Six of the 12 Raman spectra were taken
with a laser power of 100 mW and varying laser excitation times between 1 s and 1 h. The
other 6 Raman spectra were recorded with a laser power of 25 mW and varying illumination
times from 1 s to 2 h. The Raman spectra are indistinguishable from each other, showing that
the energy transfer from the laser into the droplets has no significant influence on the droplet
42 CHAPTER 4. THERMODYNAMIC PROCESSES
1.2
1.0
0.8
0.6
0.4
0.2
0.0)0
Raman shift [cm"1]
Figure 4.2: Raman spectra of an aqueous H2SO4, droplet 10 ßL in volume with a concentration of 4-37
mol kg-1 at room temperature. Twelve Raman spectra are shown for which the laser excitation time
varies between 1 s and 2 h. Six of the 12 Raman spectra were taken with a laser power of 25 m W. The
other six Raman spectra were taken with a laser power of 100 mW. The Raman spectra are normalized
to the ^(SO^-) vibration band.
composition or temperature, which would have been seen as a change in the vibration band ratio
ui(S02~)/ui(H.SO^) (see below). The influence of the laser light on droplet temperature was
also checked by measuring melting temperatures of aqueous nitric acid droplets while they were
excited by the laser Knopf et al. (2002). The experimentally obtained melting temperatures
were in agreement with literature data (Carslaw et al., 1995a) within 1 K, which also indicates
that no significant change in composition occurred.
4.4 Results and Discussion
To determine the ion activity product, 7(T), in an aqueous H2SO4 solution measured values
of Q(T) and data on Ku(T) are required. In the following, we present the analysis of the
experimental data and derive a new formulation of K\\(T) using a Pitzer ion interaction model
Pitzer (1991).
1.2
0.0
T 1 " •—1——1—•—1——r
•«i(HS041
"1<S04*)
- l__i l_ _< I L. _• L.
500 600 700 800 900 1000 1100 1200 1300 14(
4.4. Results and Discussion 43
4.4.1 Analysis of Experimental Data
Raman spectroscopy can be used for a quantitative analysis of ion speciation if the vibration
bands of the individual species can be identified. The assignment of the various S04_, HSOJ,and H30+ vibration bands according to Querry et al. (1974) and Cox et al. (1981) are indicated
in Fig. 4.2. We chose the integrated line intensities of the vibration bands i/i(S04_) at 980
cm-1 and ^(HSOJ) at 1040 cm-1 to obtain the corresponding molal ratio of rag02-/mHSO-
(Dawson et al., 1986; Tomikawa and Kanno, 1998), where m denotes the molahty of the particular
ion. The integrated line intensities, Iu, were obtained by simultaneously fitting a Lorentzian
function to each peak in the 800-1300 cm-1 interval. The line intensities are proportional to
the concentration of the respective ion i:
F(i)=m(i).r(i), (4.4)
where m(i) is the molal concentration and Jv(i) is the molal scattering coefficient of ion i. Jv(i)
depends on the Raman scattering cross section of the ion, cru(i), and on instrumental properties,
-^instr:
r(i) = o-l'(i)-AiBStT. (4.5)
The molal ratio of SO|_ and HSOJ can be derived from the measured integrated fine intensities
in the following way:
m(SOl-)mQHSOj)
Therefore, the conversion of an intensity ratio into a molal ratio depends only on the ratio of
the Raman scattering cross sections of the particular ions. Equations 4.6 and 4.8 show that for
one particular experiment the ratio of <jv can be substituted by the ratio of the corresponding
Jv, because Amstr cancels out. This requires that av(i) or likewise Jv{i) are constant in the
investigated concentration and temperature range.
Dawson et al. (1986) investigated the temperature and concentration dependence of J980(SO4_)and J1040(USO^) using sodium sulfate and ammonium bisulfate solutions. They found both
molal scattering coefficients to be constant within 1 o error in the temperature range of
298.15-523.15 K and for concentrations of 0.514-2.25 mol kg-1. Hayes et al. (1984) also found
that J980(SO4-) is temperature and concentration independent in the temperature range of
298.15-358.15 K in (NH4)2S04/H20 solutions with concentrations of 0.53-3.14 mol kg-1.Here, we investigate the temperature and concentration dependence of J980(SO|~) at lower
temperatures and higher concentrations than those of the studies mentioned above. The
line intensities of the vibration bands vi(SOl~), ^(SO^-), and ^(SO^-) obtained from
jl040(HgOr) 7980(302-)
J980(S02-)'
/1040(HSO4)
<71040(HSO4-)Ainstr J«»(SO?-)<7980(SOl-)Ainstr
'
/1040(HSO4-)
a1040(HSQr) 7980(302-)
CT980(SO|-) 71040(HSO4 )'
44 CHAPTER 4. THERMODYNAMIC PROCESSES
0.9
X
0.3
c3
r» 0.25
m
Pm 0.2
£•COcCD
0.15
0.05
0.0
I I I I • I I
(a)
"i(S04^
KrfstV)^so/-)
500 600 700 800 900 1000 1100 1200
Raman shift [cm*1]
0.25
0.15
0.05
500 600 700 800 900 1000 1100 1200
Raman shift [cm"1]
Figure 4.3: Raman spectra of (NH^foSO^/^O droplets. Panel (a) shows 10 Raman spectra of a droplet
0.5 fiL in volume and with a concentration of 0.99 mol kg"1 for temperatures between 245 K and 285 K
every 5-10 K. Panel (b) shows 12 Raman spectra of a droplet 1 fiL in volume and with a concentration
of 5.35 mol kg~r for temperatures between 220 K and 296 K every 5-10 K. The Raman spectra are
normalized to the ui(SOl~) vibration band.
Raman spectra of (NH4)2S04/H20 droplets with concentrations of 0.99-5.35 mol kg-1 and at
temperatures of 220-296 K were analyzed. Figure 4.3 displays 10 Raman spectra of a 0.99 mol
kg-1 (NH4)2S04/H20 solution at temperatures between 245 and 285 K and 14 spectra of a 5.35
mol kg-1 (NH4)2S04/H20 solution at temperatures between 220 and 296 K. The ^(SO?-)and 1/4 (SO4-) bands show no change in intensity when compared to the normalized i/i(S04 )vibration band over the investigated temperature and concentration ranges. We conclude
that J980(SO*-) and, thus, also <t980(SO|~) are constant at temperatures of 220-296 K for
concentrations up to 5.35 mol kg-1. Assuming this also to be the case for J1040(HSO4~), we can
use the measured molal scattering coefficients by Dawson et al. (1986) to obtain the ratio of the
Raman scattering cross sections: a^,/ct98^_ = J980(SO|")/J1040hso: so^
(HSO4) = 1.035±0.024.
The small relative difference between the Raman scattering cross sections of 0.035 indicates that
the excitation of the SO|~ and HSO4 stretching vibrations are very similar. This gives further
support for the above assumption that J1040(HSO4~) is independent of temperature and concen¬
tration, just as in the case of the investigated J980(SO|~) band discussed above and in Figure 4.3.
Figure 4.4 shows the phase diagram of the H2SO4/H2O system (Gable et al., 1950). The tem¬
perature and concentration ranges of solutions that were investigated by Raman spectroscopy in
4.4. Results and Discussion 45
6 8 10 12 14 16 18 20
mHaso4 [mol kg"1]
Figure 4.4: Phase diagram of H2SO4/H2O (Gable et al, 1950). Solid lines represent the melting
curves of several crystalline solids of H2SO4/H2 0. SAH: sulfuric acid hemihexahydrate; SAT: sulfuric
acid tetrahydrate; SATr: sulfuric acid trihydrate. The dashed lines indicate the temperatures and concen¬
trations where Raman experiments have been performed.
this study are indicated by the dashed fines. The experimental data axe limited to the tempera¬
ture range where the droplets remained liquid. As seen in Figure 4.44, the measurements could
be extended well into the supercooled regime. Figure 4.5 shows Raman spectra of a H2SO4/H2O
droplet with a concentration of 2.55 mol kg-1 at different temperatures. The spectra are nor¬
malized to the ^(SO^-) vibration band and reveal a strong decrease in the intensity of the
^(HSOJ) vibration band at low temperatures. We conclude that the concentration of HSO4decreases with decreasing temperature. The dashed line corresponds to a Raman spectrum of a
frozen droplet. The peak at about 3100 cm-1 indicates the presence of ice. We also investigated
the ratio, R, of the integrated line intensities of the SO|~~ and HSO4 vibration bands in the
range of 800-1300 cm-1 (with a negligible intensity stemming from the U2(H.sO+) vibration
band) and the integrated line intensities of the ï/i(H20) and ^3(^0) vibration bands in the
46 CHAPTER 4. THERMODYNAMIC PROCESSES
CO
c
cd
!o
COc
CD
290 K
n(HS04)i/,(S04a)
500 1000 1500 2000 2500 3000 3500 4000
Raman shift [cm"1]
Figure 4.5: Raman spectra of a H2SO4/H2O droplet 0.5 ßL in volume and with a concentration of
2.55 mol kg~l. Spectra are shown from 290 K in 10 K steps until freezing occurs (228 K). The dashed
line corresponds to a Raman spectrum of the frozen H2SO4/H2O droplet. Individual spectra are shifted
vertically for better visibility. The Raman spectra are normalized to the i>i(S04~) vibration band.
range of 2500-4000 cm-1, that is
1300
£ /"(SOI") + /"(HSOJ) + J"(H30+)R =
i/=800
4000
E J*(H20)i/=2500
(4.9)
We found R to be constant for a particular solution concentration over the investigated temper¬
ature range within an experimental uncertainty of about 10%, indicating that no concentration
changes due to water evaporation or condensation occurred during the cooling of the droplets.
Figure 4.6 shows the mS02- /mBSO- ratio obtained from our Raman measurements for 0.54-
15.23 mol kg-1 H2SO4 solutions at temperatures of 180-326 K. The experimental data (open
squares) show a continuous increase of the "^so2_/mHSO~ rati° w^ith decreasing temperature
for all investigated concentrations. Tomikawa and Kanno (1998) present results which show
4.4. Results and Discussion 47
2.0
1.5
1.0
0.5
0.0
OCOI
E 12
-~» 9
<% 6
O o
CO d
E o
10
8
6
4
2
0
4
3
2
1
0
1.5
1.0
0.5
0.0
180 200 220 240 260 280 300 320
—I 1 —i 1 1 —
• Œ m
0.54 mol kg'1
• S S
*1.13 mol kg"1
H 1.5
1.0
0.5
0.0
• «S g I g
3
2
-|1
T T T• • 2.55 mol kg'1
"ra„••i g
9Bet t T
4.37 mol kg
• • • •
H—i—i
+
• • S* O* a A.
,—I—, | ? f f» f» Mip»iP»p
6.79 mol kg"1
r * r r r ? y m « s i t » t
9.84 mol kg
f • r ' f ? r f r t ? 9 1 »ff15.23 mol kg'1J
js_^ e_£ e t e
Q] Q,m"m œ nm m m m rj
-C t S « — «* •»
2.0
180 200 220 240 260 280 300 320
Temperature [K]
o
5
4
3
2
1
0
12
9
6
3
0
10
8
6
4
2
0
4
3
2
1
0
1.5
1.0
0.5
0.0
Figure 4.6: Ratios o/mS02-/mHSO- m H2SO4/H2O solutions with concentrations of 0.54-15.23 mol
kg-1 as function of temperature. Open squares with error bars indicate the data obtained from our Raman
spectra. Circles represent values derived using the AIM model (Clegg et al, 1998). Diamonds show data
of a Raman study by Tomikawa and Kanno (1998). Note the different scales for each concentration.
full dissociation at even lower temperatures for aqueous H2SO4 solutions with concentrations of
4.37-15.23 mol kg-1 in the glassy state (about 143-158 K). Our Raman data for the 4.37 mol
kg-1 H2SO4 solution axe in very good agreement with the data by Tomikawa and Kanno (1998)for the same concentration (diamonds in Figure 4.6). In contrast, the "^go2-/rnHSO_ ra*i°s
predicted by the AIM model (Clegg et al., 1998) are much smaller than our measurements and
48 CHAPTER 4. THERMODYNAMIC PROCESSES
exhibit a maximum for all concentrations at about 180-240 K. In addition, for concentrations
greater than 6.79 mol kg-1, the model predictions deviate significantly from the experimental
data even at room temperature.
From the wS02-/mHS0- ratios the degree of dissociation of the HSOJ ion, c*HSO-, can be
calculated:
aHso:
ms02
m"
ms02
m.
1 +mS02
-l
m-r
(4.10)
where m^ is the total HS04 molahty before dissociation (Note 1). Figure 4.7 shows a
HSO.hso;
4 6 8 10 12
mH2so4 [mol kg"1]
16
Figure 4.7: Degree of dissociation of the HSO^ ion versus the solution H2SO4 molality. Diamonds,
squares, and circles with corresponding error bars represent experimentally derived data at 190, 230, and
290 K, respectively. The dotted lines indicate the values predicted by the AIM model (Clegg et al, 1998).
The solid lines show values calculated by the Pitzer ion interaction model using K\\(T) derived in this
study.
derived from our experimental data (symbols) as a function of mH2so4 for temperatures of 190,
230, and 290 K. The dotted lines correspond to values predicted by the AIM model (Clegget al., 1998). The solid lines represent aHSO- values calculated by our Pitzer model (see below).
There is a large discrepancy between the experimentally derived data and the AIM values, in
particular at low temperatures and high concentrations.
4.4. Results and Discussion 49
One possible explanation for the observed discrepancy between the AIM model and the experi¬
mental data could be an inaccurate parametrization of the thermodynamic dissociation constant
of HSOJ, Kn(T), in the model of Clegg et al. (1998) at lower temperatures. The experimental
data used to derive the formulation of Kn(T) were limited to a temperature range of 298-523 K
(Dickson et al., 1990). Nevertheless, this formulation had been adopted in the AIM model for
temperatures down to 180 K. For these reasons we will investigate the temperature dependence
of K\\(T) in more detail in the following section.
4.4.2 Results of the Pitzer Ion Interaction Model
We have used an extended Pitzer ion interaction model (Pitzer, 1991) to calculate ion activity
coefficients and Kn(T) for the H2SO4/H2O system at low temperatures. It is based on the
molality concentration scale and is valid up to concentrations of 40 mol kg-1. A detailed de¬
scription of the working equations is given in the Appendix.For a consistent calculation of the activity coefficients of the various ions (7h+ , 7so2_ > Thso- ) a*
low temperatures, the dissociation of HSOJ has to be considered. Thus, the temperature depen¬
dent second thermodynamic dissociation constant, Ku(T), must be known for the investigated
temperature range. In Figure 4.8, In K\\{T) is plotted as a function of inverse temperature. The
dotted fine shows the formulation of K\\{T) given by Clegg et al. (1994), which is also imple¬
mented in the AIM model (Clegg et al., 1998; Note 2). This formulation was originally derived
by Dickson et al. (1990) for the temperature range of 298-523 K. Some of the high temperature
data of K\\ of Marshall and Jones (1966) and Dickson et al. (1990) are also plotted as diamonds
in Figure 4.8. The dissociation of HSO4 at room and higher temperature is an exothermic
reaction. Extrapolation of the formulation of Ku(T) of Dickson et al. (1990) to low temper¬
atures in the AIM model (Clegg et al., 1998) suggests that the exothermic reaction changes
to an endothermic reaction at around 233 K. From K\\(T) the corresponding standard Gibbs
free energy, AGn(T), enthalpy, AH^T), and entropy, AS^T), of the dissociation reaction of
HSO4 (HSO4 -s. SO^_ + H+) can be derived:
AG^(T) = -RTlnKn(T) (4.11)
A/4(T) = -R*h*°Çl (4.12)d^
AS,(T) _ A^(T)-Ao;,(T)| (413)
where T is temperature and R is the universal gas constant. The dotted lines in Figure 4.9
show AGjj(T), Ai?jj(T), and A5n(T) for the dissociation reaction of the bisulfate ion using the
formulation of -Kn(T) of Clegg et al. (1994). AG\X has a minimum at about 220 K with increasing
values at lower temperatures. Also, A.ffn and ASjj increase with decreasing temperature over
the entire temperature range. ASjj at 298.15 K is about -110 J K_1mol-1 (Lide, 1998), which
is in agreement with the formulation of Kn(T) of Clegg et al. (1994). However, the increase in
A5jj at very low temperature contradicts the Nernst heat theorem (Nernst, 1906; Berry et al.,
50 CHAPTER 4. THERMODYNAMIC PROCESSES
T[K]400 333.3 285.7 250 222.2 200 181.8
-2.0
-2.5
-3.0
-3.5
-4.0
^-4.5ç
-5.0
-5.5
-6.0
-6.5
-7.00.0025 0.003 0.0035 0.004 0.0045 0.005 0.0055
T"1 [K"1]
Figure 4.8: In K\\ plotted as a function of inverse temperature. The dotted line corresponds to the
formulation given by Clegg et al. (1994). The solid line represents K\\ derived in this study. Dashed lines
represent the results of a sensitivity study (see text for details). Open squares show Kn(T)-values using
the measured <*HSO- and the activity coefficients derived by our Pitzer model. Diamonds show some of
the high-temperature data of Ku from Marshall and Jones (1966) and Dickson et al. (1990).
2000), which says that AS for any reaction vanishes as temperature approaches 0 K; that is
linvr—o AS = 0.
In the following, we use a Pitzer ion interaction model (Pitzer, 1991) to determine a formulation
of K\\(T) which is consistent with the experimental data and with thermodynamics. Several
data sets were implemented in the Pitzer ion interaction model to obtain the new formulation
of Ku(T): the data of ojhso- obtained in this study; data of electromotive force experiments
(Harned and Hamer, 1935); enthalpies and heat capacities of H2SO4/H2O (Giauque et al., 1960);
and dissociation constants in the temperature range of 323-473 K (Marshall and Jones, 1966;
Dickson et al., 1990). Because each of these data sets consists of a different number of data
points, the data were weighted such that each data set had equal weight in the overall fit.
The enthalpy of the dissociation reaction of HSO4 , A.ffn(T), was fitted within the Pitzer ion
4.4. Results and Discussion 51
Ö 50
E
X
<-50
-100
600 k
O
E 400
-5 200 k
<°
-200
-—h
(b)
H——r- H——I-
(c)
50
-50
-100
600
400
200
0
-200
100 150 200 250 300 350 400 450 500
Temperature [K]
Figure 4.9: (a) Gibbs free energy for the dissociation reaction of the bisulfate, AG'n, as a function of
temperature, (b, and c) Corresponding reaction enthalpy, AHU, and entropy, ASjj, respectively. The
dotted lines show results using the formulation ofKn(T) by Clegg et al. (1994). The solid lines represent
results using the formulation of Ku(T) derived in this study. The dashed lines indicate the temperature
range where experimental data are available.
interaction model by11 dco dcp,
Afl£(D = AflS + 4(T - T0) + ^(T2 + T02) - ^T0T, (4.14)
1c° must be changed to Ac£. Due to reasons of copyright the correction will not be implemented.
52 CHAPTER 4. THERMODYNAMIC PROCESSES
where To is 298.15 K, Aüjj is the enthalpy of the dissociation reaction at To, c0, is the heat
capacity of the solution at To, and dcp/dT describes its temperature dependence. We performed
several model runs to derive a new formulation of Kn(T) at low temperatures. To achieve
thermodynamic consistence, we were forced to treat the dissociation as an exothermic reaction in
the temperature range of 165-473 K in our fitting procedure (i. e. Aüjj < 0 for this temperature
range). The final results (the parameters for the Pitzer equations and Ku(T)) axe both consistent
with all the experimental data and the Nernst heat theorem (Nernst, 1906; Berry et al., 2000).The formulation for Ku(T) derived in this study is shown as the solid line in Figure 4.8. The
dashed lines in Figure 4.8 indicate the results of a sensitivity study. They represent fits which
were constrained to Aiïjj < 0 down to 120 K and to 180 K, respectively. The two curves
(dashed lines) differ only slightly from the best fit (solid line), indicating that the formulation
of K\i(T) is rather insensitive to the choice of temperature range where Aiïjj is assumed to
be negative. The open squares in Figure 4.8 were derived, using the measured Q;Hso- an(i *ne
activity coefficients of the Pitzer model used in the present study. The obtained Ku(T) values
match the fit, showing that our Pitzer model works in a consistent way. The solid lines in Figure
4.9 show AGjj, Aüj'j, and ASjj calculated using the formulation of Kn(T) derived in this study.
AGjj decreases with decreasing temperature, and A/ïjj and A5jj approach zero with decreasing
temperature (in the range of available data), in agreement with the Nernst heat theorem (Nernst,
1906; Berry et al., 2000). It should be noted that, when our new dissociation data are included,
the Pitzer model can be tuned to reproduce all experimental data even with Ki\(T) fixed to the
formulation of Clegg et al. (1994). This is achieved by adjusting the ion activity coefficients in
the aqueous solutions. However, although the resulting model parametrization is consistent with
all experimental data, it contradicts the Nernst heat theorem (Nernst, 1906; Berry et al., 2000).
Furthermore, we expect that also the AIM will reproduce all experimental data in agreement
with the Nernst heat theorem when our new dissociation data and the newly derived K\\{T) are
implemented. Values for our newly derived Ku(T) (solid fine in Figure 4.8) can be calculated
from the following equation using the parameters given in table 4.2. Ku was calculated by
integrating eq 4.12 and using eq 4.14 for A/ïjj:2
lnÄ-n(T) = lnirj0i(r0)-/TAi4(r)4 (4.15)
- in*?,- [(Atf8-<$r0 + I^)(I- i)
-K-^inL-l^T-n)} (4.16)
The above formulation can be used in the temperature range of 180-473 K.
With the newly derived formulation for Ku(T) our Pitzer ion interaction model can be used to
predict the degree of dissociation, a°<?_. Comparisons between the modeled a„°^_ and the
experimentally obtained data are given in Table 4.3 and Figure 4.7. Also shown in Table 4.3
are values for the activity coefficient product, 7(T), which were calculated from eq 4.3 using the
Äll(T)-values from Eq. 4.16 together with Q(T)-values obtained directly from the experimental
2In Eq. 4.15 and Eq. 4.16 ^ is missing in front of the integral and the big bracket. Due to reasons of copyright
the correction is not implemented.
4.5. Atmospheric Implications 53
data.
Figure 4.10 and 4.11 show calculated ion activity coefficients, 7h+, 7so2_> a3X^- 7h+'
^SO2-' an(^
water activities, aw, for a 1.13 mol kg-1 and a 9.84 mol kg-1 H2SO4/H2O solution, respectively.
The solid lines correspond to values calculated with our Pitzer model using the new Ku(T)-
formulation; the dotted lines are calculated using the AIM model (Clegg et al., 1998). Significant
differences between the two models exist for all parameters, in particular at low temperatures.
Note that the difference in <zw for the 9.84 mol kg-1 solution is about 10 % at 180 K. This
may be due to the larger HSO4 dissociation and, thus, larger ionic strength in our model at
low temperatures. Also note, that the activity coefficients between the two models also differ
strongly, for example in the case of 7SQ2- in Figure 4.11b by up to 2 orders of magnitude.
4.5 Atmospheric Implications
Relative humidity and temperature can vary over a large range in the atmosphere; for exam¬
ple temperatures can be as low as 180 K in the polar stratosphere and tropical tropopause.
Stratospheric aerosols can consist of highly concentrated H2SO4/H2O droplets (Junge and
Manson, 1961; Hamill and Toon, 1991) at dry conditions. Under these conditions the exper¬
imental data obtained in this work show a significantly higher dissociation for the reaction
HSOJ ^ SO|_ + H+ than was assumed in previous model calculations. This also has im¬
plications for other thermodynamic properties of aqueous H2SO4 solutions. The water vapor
pressure, Ph20> of a H2SO4/H2O solution depends on the water activity of the solution, <zw:
PH2o(T) = av,(T)p0H2O(T), (4.17)
where Ph2q is the water vapor pressure over pure water at the same temperature. Because
we calculate lower water activities at low temperatures than the AIM does, our results imply
slightly lower water vapor pressures of sulfuric acid aerosols under stratospheric conditions.
In addition, the new binary interaction parameters for H2SO4/H2O obtained by our Pitzer
model can serve as input parameters for the derivation of ternary interaction parameters in
aqueous solutions such as the NH3/H2SO4/H2O and HCI/H2SO4/H2O systems, which axe com¬
mon in aerosols of the troposphere and stratosphere. Furthermore, the solubility of trace gases
in H2SO4/H2O solutions such as HCl and NH3 is affected (Luo et al., 1994; Swartz et al., 1999).The larger dissociation constant leads to higher H+ concentrations in H2SO4/H2O solutions
and, therefore, to lower HCl solubilities/Henry's law constants when compared to AIM results
(Carslaw et al., 1995a). Over the concentration range of 4.3-15.23 mol kg-1 and at tempera¬
tures between 180 and 300 K, the maximum difference between our model and the AIM (Carslawet al., 1995a) is about a factor of 3. We note that Carslaw et al. (1995a) show in their Figure
13 that data from vapor pressure measurements (Hanson and Ravishankara, 1993; Zhang et al.,
1993a) fall below their model predictions, while data from uptake experiments (Hanson and
Ravishankara, 1993; Williams and Golden, 1993; Elrod et al., 1995) are closer to predictions.
Later Hanson (1998) reexamined their earlier data which are still below but closer to the predic¬
tions than before. Unfortunately, the scatter between the different data sets (and experimental
54 CHAPTER 4. THERMODYNAMIC PROCESSES
333.3
?»
10"4
3
?101
O* 10","2 -
?
10-°
s
0.975
0.97
3 0.965(0
0.96
0.955
0.95
285.7
T[K]250 222.2 200
-r
181.8
_i_
(a)
(b)"-
(c)
(d)
10"'4
3
10-8
7
6
5
10'
5
10-°
5
0.975
0.97
0.965
0.96
0.955
0.95
0.003 0.0035 0.004 0.0045
r1 [K-1]
0.005 0.0055
Figure 4.10: Activity coefficients 7H+, 7so2-> 7h+"
7so2~ an^ flw °f a ^-^ mo' ^_1 H2SO4/H2O
solution plotted as a function of inverse temperature. The solid lines are calculated by the Pitzer model
using Ku(T) derived in this study. The dotted lines are predictions from the AIM model by Clegg et al.
(1998).
methods) is large and, thus, does not allow us to conclude which of the model predictions is
closer to real solubilities at this stage.
4.5. Atmospheric Implications 55
285.7
T[K]250 222.2 181.8
0.003 0.0035 0.004 0.0045
r1 [K-1]
Figure 4.11: Activity coefficients 7h+, 7go2-> 7h+'
7so2~ an^ flw °f a 9-&4 mo^ ^g_1 H2SO4/H2O
solution plotted as a function of inverse temperature. The solid lines are calculated by the Pitzer model
using Ku(T) derived in this study. The dotted lines are predictions from the AIM model by Clegg et al
(1998).
56 CHAPTER 4. THERMODYNAMIC PROCESSES
4.6 Conclusions
The dissociation of the bisulfate ion (HSOJ ^ SO2- + H+) has been studied in a temperature
range of 180-326 K in H2SO4 solutions with concentrations of 0.54-15.23 mol kg-1 using Raman
spectroscopy. The experimental results show a continuous increase in the degree of dissociation
of HSOJ with decreasing temperatures within the investigated concentration and temperature
range. Our results disagree with predictions from the thermodynamic model AIM (Clegg et al.,
1998), which underestimates the degree of dissociation of HSOJ for high H2SO4 concentrations
and low temperatures by up to a factor of 5. This is most likely due to the implementation of a
thermodynamic dissociation constant in the AIM, that is at odds with the Nernst heat theorem.
Therefore, we have employed a Pitzer ion interaction model to obtain a new thermodynamically
consistent formulation of the thermodynamic dissociation constant, Kjj(T), that is in agreement
with experimental data. The new formulation of -Kn(T) is valid from 180 to 473 K. In the
model, consistency with thermodynamics can be achieved only by assuming that the dissociation
reaction is exothermic over the entire temperature range. Results from our model indicate that
ion activity coefficients can differ by up to 2 orders of magnitude, water activities and vapor
pressures by up to 10%, and HCl solubilities by up to a factor 3 when compared to results from
the AIM (Clegg et al., 1998; Carslaw et al., 1995a).We recommend that future thermodynamic investigations of multicomponent aqueous solutions
containing H2SO4 use the new formulation of K\i(T) for a correct description of the dissociation
reaction of the bisulfate ion.
4.7 Appendix
4.7.1 Tables
Table 4.3: Experimental Data and Modeling Results for the
Investigated Aqueous H2SO4 solutions"
T H2S04 «hso4- «SS- lnQ ln7
[K] [mol kg"1]321.0 0.54 0.22 0.20 -1.667 -3.507
316.0 0.54 0.22 0.22 -1.657 -3.379
311.0 0.54 0.19 0.23 -1.864 -3.033
306.0 0.54 0.23 0.25 -1.620 -3.140
297.1 0.54 0.30 0.29 -1.182 -3.344
290.4 0.54 0.34 0.32 -1.014 -3.337
280.5 0.54 0.38 0.37 -0.787 -3.318
270.7 0.54 0.46 0.42 -0.424 -3.444
260.8 0.54 0.46 0.48 -0.384 -3.258
251.0 0.54 0.53 0.54 -0.079 -3.350
241.0 0.54 0.59 0.60 0.185 -3.413
Table 4.3: (continued)
T H2S04 aHSo4- agg_ InQ In 7
[K] [mol kg"1]*
289.4 1.13 0.37 0.37 -0.077 -4.251
279.4 1.13 0.42 0.42 0.135 -4.212
269.6 1.13 0.48 0.48 0.418 -4.261
259.8 1.13 0.54 0.54 0.696 -4.315
254.9 1.13 0.59 0.57 0.937 -4.448
249.9 1.13 0.60 0.60 1.019 -4.427
240.1 1.13 0.68 0.66 1.401 -4.611
290.8 2.55 0.42 0.41 0.962 -5.325
280.1 2.55 0.46 0.47 1.144 -5.239
270.1 2.55 0.51 0.52 1.398 -5.253
260.8 2.55 0.57 0.58 1.690 -5.332
251.2 2.55 0.64 0.64 1.989 -5.423
246.1 2.55 0.67 0.68 2.145 -5.474
240.1 2.55 0.70 0.71 2.304 -5.513
236.1 2.55 0.73 0.74 2.457 -5.589
233.0 2.55 0.75 0.75 2.565 -5.643
231.0 2.55 0.75 0.77 2.615 -5.656
325.9 4.37 0.34 0.33 1.093 -6.407
320.8 4.37 0.34 0.34 1.112 -6.283
315.9 4.37 0.35 0.35 1.156 -6.189
310.9 4.37 0.36 0.36 1.208 -6.103
306.0 4.37 0.37 0.38 1.276 -6.037
301.0 4.37 0.39 0.39 1.344 -5.973
297.4 4.37 0.40 0.40 1.397 -5.931
289.7 4.37 0.40 0.43 1.398 -5.733
279.9 4.37 0.47 0.47 1.748 -5.837
269.8 4.37 0.52 0.52 1.982 -5.830
259.8 4.37 0.58 0.58 2.250 -5.869
249.7 4.37 0.64 0.64 2.522 -5.926
239.9 4.37 0.70 0.70 2.844 -6.050
230.0 4.37 0.76 0.75 3.195 -6.218
219.7 4.37 0.83 0.81 3.665 -6.515
209.5 4.37 0.85 0.85 3.867 -6.568
199.7 4.37 0.89 0.89 4.219 -6.797
190.0 4.37 0.92 0.91 4.540 -7.021
289.9 6.79 0.41 0.41 1.909 -6.249
279.1 6.79 0.45 0.44 2.082 -6.151
269.6 6.79 0.49 0.48 2.280 -6.124
259.8 6.79 0.53 0.52 2.473 -6.093
249.9 6.79 0.58 0.57 2.703 -6.110
8OI11Ia>
Irr
3«u
ö8m
EÖ
oo1«
oCO
CO
Tjt
S
ma
tOen
w1
HM
1—1
(M
Tf
CO
•*fo
lO
CO
(M
coo
1—1
^f
^f
1—1
ON
CO
»o
t-
OS
1—1
tvcO^Pt-OOOOCOCOTPCN
fflWICOONOlONiOSO
OOOOb-COlO^f^COCNi-Hi-H
w^ffliooooo^HijoHiflnowo
TfinoOOOOOOHWtOBifMOOOîO^O
~
-OHM'tb-OffiOOSCiO^MMH
oo
•*
K3
00*
<N
CN
00M
oa
oo
oo
cocococococot—cococococococococococococococococococot^cocococococococococoioioio
COCOt-HOSlOi-lOîCNCNOO'^^-liOOrHCO
(MOît-OOCNlOCNi-lOOCOOSi-ICO-^-^fCOi-H
oiHTfSH^sœoNNorjœaoHW
^POîlOCOOîCNCli-l
WNMNWHOO
OO'fSOOOOOOHN«)
00'-fTt<CO-3,l"-C01O
hmwa
O)O
!Û
<ooooN^N»nNNoqooooaoqHWMioooo
«SMSN!OOiaOHH«TfT)iioiiOHinoONNH!OHiOO»',*1,iO<ONSoOONmoOH'*
!OBr-NMOOOONn«MMC)«nW^'il"*iOiOO<ONSOO«NN«NNNNMWWW^^
ÖÖÖÖÖÖÖÖÖÖÖÖÖÖÖÖÖÖÖÖÖÖÖÖÖÖÖ
woonoowNoi»OHMN^ir)(ûoi«wœ»SH!OH(OHn^ioioinr>(-oooiNioson
«©NNooooo>iNnnMM!onnM^^^iow«»NNoqN«NiNiNC)iN«wnnn'fTf
ÖÖÖÖÖÖÖÖÖÖÖÖÖÖÖÖÖÖÖÖÖÖÖÖÖÖÖ
r^NNNSNSooooooajooooooœooooooixioooooooooo»»^^^^^^^^^^0!^0!
tofflœtdœœœoœoiœœoJoiœffldœdaoœoaœœ^SÏÏSHHHHH
oco^ior^w(Noœo^Oi-i^r^oot--icoco^^^iotocoooooooorHcocoi>oocN-^,t>05r>-
OlOOOOOi-H^hO
MW(NHOO>OON(N
lO
CO
00
co
o>o
oa
ai
oo
oooooo
N!û
iO
•*oo
C0
CN
t-H
i—lOOi—icOi—icOt-HCOOOC005050
oaooNNHHOoiaiiohNoooo
CO
lO
^t<
CO
NCNNNNHHnnMWniNNNCN(N(N(NCNNNN«HH«MMnM(NlNNNNNNNN
00
4.7. Appendix 59
Table 4.3: (continued)
T EtfÖl «Hsor«ï£- toQ kTÏ
[K] [mol kg"1]HSOJ "hSO
4
221.0 15.23 0.47 0.47 2.994 -5.865
211.2 15.23 0.51 0.51 3.170 -5.893
201.3 15.23 0.55 0.55 3.355 -5.951
191.3 15.23 0.59 0.60 3.546 -6.039
181.5 15.23 0.64 0.64 3.771 -6.190
"The first two columns indicate the temperature and concentration of the H2SO4 solutions.
aHSO-: experimentally obtained degree of dissociation (see eq 4.10); off^L- is the degree of
dissociation calculated from our Pitzer model using the new Än(T)-formulation; Q is the molal
equilibrium quotient (see eq 4.3); 7 is the activity coefficient product (see eq 4.3).
60 CHAPTER 4. THERMODYNAMIC PROCESSES
Table 4.1: Composition and Volume of the Investigated Aqueous Droplets.
H2so4 (NH4)2S04 Volume
[mol kg-1] [mol kg-1] [10-3cm3]
0.54 0 10
1.13 0 0.5
2.55 0 0.5
4.37 0 0.5
6.79 0 0.5
9.84 0 10
15.23 0 10
0 0.99 0.5
0 1.95 0.5
0 3.17 0.5
0 3.88 1
0 5.35 1
Table 4.2: Fit Parameters to Derive \nKu{T) Using Eq 4.I6.
K AHn/R [K] %/R &/R [K"1]
1.0576 lO-2 -2231.620793 -24.7273 -0.11967
Numerical check: lnü:(273K) = -3.954.
4.7. Appendix 61
Table 4.4: Temperature Dependent Parameters (p) and Temperature Independent Parameters (aca, u)Ca>
b) of the Ion Interaction Model?.
p X AHn/R [K] cl/R S/Ä IK"1]0(0)^H+,HS07
-0.12672773 122.83352564 0.79298257 4.49770xlO-3
3{1)MH+,HS07
1.43843400 140.47112887 -15.44397569 -0.49178043
c(0)H+.HS07
1.08965 xlO"3 -1.36879879 -9.58811 xlO-3 -2.78 xlO-6
c(1)H+.HS07
0.31617615 4.38280351 -0.50421020 2.74165 xlO-3
/3(0)PH+,S02-
0.12485773 9.86603452 -0.59448781 -6.1367X10-4
tf(1)
PH+,S02--0.46260131 358.49482175 17.75619395 5.849624xl0~2
CH+,S02"5.10014X10-3 -3.21170454 -1.264739xl0~2 -4.46 xlO-6
CH+,S02--0.27604369 83.11418947 4.55227587 6.408755xl0-2
^H+,HS07: 1.2 ^H+,so2_: 1.2
Cl,H+,HS07: 0.91291829 WH+,S02-: 1.91623572
aH+,HS07: 2.0 aH+,S02_: 2.0
"Units are as follows: ßej and /%, in kg mol 1; cid and c)J in kg2 mol 2; atca, ^ca, and bCl
in kg1/2 mol-1/2; the temperature dependent parameters are calculated in the following way:
p = lnx + (AFj0i-c°ro + è^T2)(i-^)-(^-^ro)ln^-i^(r-To).
4.7.2 Derivation of the thermodynamic dissociation constant of HS04
The chemical potentials ß% (i = H+, SO2, HS04 ) of the species involved in the dissociation
reaction HSOJ ^ SO|" + H+ can be written as:
ßt(T) = ßl(T) + RT In at(T) (4.18)
where ß\ is the standard chemical potential of the species 1, that is the chemical potential in a
hypothetical 1 mol kg-1 aqueous solution with ideal properties. The activity, at, of each species
is the product of its activity coefficient, 7,, and its molality mt, such that at(T) = li(T)m^ '.
ml is by definition 1 mol kg-1.In chemical equilibrium, the Gibbs free energy for the dissociation reaction is
AGn(T) = /iSO2-(T) + MH+(T)-/iHSO7(T) = 0- (4.19)
Replacing ß% from eqs 4.18 and 4.12 into eq 4.19 yields the temperature dependent thermody¬
namic dissociation constant of HSO4 , Kn(T):
Kn(T)mS02- (T) • mH+ (T) \ /7soJ- (T) • 7h+ (T)
'
mhso; -(T)-mt 7hso7(T)
(4.20)
62 CHAPTER 4. THERMODYNAMIC PROCESSES
Equation 4.20 shows that in order to derive Kn(T) not only dissociation data are required, but
also knowledge of the activity coefficients of the involved ions.
4.7.3 Extended Pitzer Ion Interaction Model
The Gibbs free energy of an aqueous electrolyte solution can be written as
G 1
WvRT RT«0*w - a*1) + J2m^ ~ /4) (4.21)
= Qlnow-I- y^mjlnai, (4.22)i
where R is the universal gas constant, T is the temperature, ww is the mass of the solvent (1
kg of water), Q is number of moles in one kilogram of water (55.51 mol), and mi, ßi, ß], and ai
are defined as above.
For an ideal solution, the activity coefficient 7» = 1 and, thus, the Gibbs free energy of an ideal
solution becomes
Qideal
wZrt= mna^ + ^miln^i). (4.23)
Since the excess Gibbs free energy is the difference between the Gibbs free energies of a real and
an ideal solution, one finds using eqs 4.22 and 4.23
G G G(424)
wwRT wwRT wwRT
= n(ln a* - Ino^ + ^mi In7i. (4.25)i
Instead of water activity, the osmotic coefficient, $, is often used in thermodynamic treatments
and is defined as
$ = --^-lnaw. (4.26)
To derive the water activity in the ideal solution, d^0,1, the Gibbs-Duhem relation is applied to
eq 4.23:
fid In a**"* + Y^mdm (^) = 0, (4.27)i
and integration yields
= "è?"1*- (4-28)In a***'
4.7. Appendix 63
Thus, eq 4.25 can be written as
Gex
WwRT= ^m,(l-$ + ln7,). (4.29)
The water activity, Ow, and the activity coefficients, % (i = H+, S04_, HSO4 ), can be derived
from the following derivatives of the excess Gibbs free energy:
i / piCex \
Using eqs 4.26, 4.28, and 4.30 the expression for the osmotic coefficient can be written as:
,_! _
' J-(*q. (4.32)
The statistical mechanics of electrolytic solutions suggest that Gex can be expressed in terms
of a virial expansion in concentration (Pitzer, 1991). In our model the three-body interaction
(H+, SO|", HSO4 ) and the interaction between HSOJ and SO2- axe neglected and, therefore,
Gex can be written as:
/~<ex
—— ~ /(/) +Vmcma (2Bca(I) + ZC^I)) , (4.33)ca
where m, is the molality of the ith ion (i = a or c, i. e. anion or cation, respectively). Z is
given by Z = ^2tml \ z% |, where zt is the charge of the ith ion. /(J) is the Debye-Hückel
term, representing the long-range electrostatic interaction, and Bca(I) and Cca(I) describe the
short-range interactions in binary solutions. fîca(7) is the two body interaction term (one cation
interacts with one anion) and Cca(I) is the three body interaction term (two identical ions
interact with one other ion: era, aac). AU given parameters depend on temperature and on the
molal ionic strength, /, which is given by i" = 1/2 £^ mtz%.According to eq 4.31 the activity coefficient of a cation or an anion can be derived by taking the
derivatives from eq 4.33:
In 7c = z^F + J2ma(2Bca + ZCca), (4.34)a
In 7a = 22F + ^mc(2Bca + ZCco), (4.35)c
where the summations run over cations (c) and anions (a). The quantity F includes derivatives
of the long-range electrostatic interaction term, /, and of B and C with respect to the ionic
strength I:
F=lf' + Yl mc«(B'ca + ^C'ca) (4-36)
64 CHAPTER 4. THERMODYNAMIC PROCESSES
According to eq 4.32 the osmotic coefficient is calculated from the derivative of eq 4.33 with
respect to u;w:
$-1 = -,
where
If*[I) + £ m°m° [ß-(J) + ZCcaV)] (4.37)
f*W = ^(If'(I)-f(I)), (4-38)
B%(I) = BM + IB'^I), (4.39)
Ct{I) = Cca(/) + ^(/). (4.40)
The differential equations 4.38-4.40 are solved numerically using the foUowing analytical func¬
tions fitted to available data sets:
f*W = -jL£-r> (4-41)1+6/2
Bcl(I) = ßW+ßtte-^, (4.42)
Ccl(I) = cM + cUe-»^, (4.43)
where A^ is the temperature dependent Debye-Hückel parameter given by Pitzer (Pitzer, 1991),and ßca, , ßca , Cca ,
and CcJ are temperature dependent fit parameters, and b, a, and u are
temperature independent fit parameters. Note, that the parameters Cca and cca represent the
average of the two different three-body interactions caa and cca, which cannot be separated
numerically. The analytical functions /* and I?*, in eqs 4.41 and 4.42 axe chosen according to
Pitzer (Pitzer, 1991) to yield the best fit to the data of a large number of aqueous solutions. We
chose 0%, also to depend on the ionic strength. The parameters of our Pitzer ion interaction
model are given in table 4.4.
The solutions of the differential equations 4.38, 4.39, and 4.40 give /(/), Bca{I), Cca(I), and
their corresponding derivatives. These functions axe required to calculate Ow using eq 4.37 or to
derive the activity coefficients employing eqs 4.34 and 4.35.
4.7. Appendix 65
Here ends the publication:
Thermodynamic Dissociation Constant of the Bisulfate Ion from Raman
and Ion Interaction Modeling Studies of Aqueous Sulfuric Acid at Low
Temperatures
D. A. Knopf*, B. P. Luo, U. K. Krieger, and Thomas Koop
Institute for Atmospheric and Climate Science, Swiss Federal Institute of Technology, Hongger-
berg HPP, 8093 Zurich, Switzerland
* To whom correspondence should be addressed. Email: Daniel.Knopf@iac.umnw.ethz.ch.
Reproduced with permission from the Journal of Physical Chemistry Part
A, 2003. Unpublished work copyright 2003 American Chemical Society.
66 CHAPTER 4. THERMODYNAMIC PROCESSES
w©iL© L@©r /
Blank leaf
4.8. HCl solubility in H2S04/H20 solutions 67
4.8 HCl solubility in H2S04/H20 solutions
The newly derived thermodynamic dissociation constant has an effect on the calculated the
amount of H+ in aqueous H2SO4 solutions. The H+ concentration in solution has an impact
on the solubility of trace gases into liquid H2SO4/H2O (see section 2.1.3). A change in sol¬
ubihty, expressed by the effective Henry's law constant, #£, corresponds to a change of the
heterogeneous reaction rate coefficient (see appendix C). Here, the solubihty of HCl, H^cv into
H2SO4/H2O solution is discussed in further detail.
Figure 4.12 shows HCl solubilities derived from model predictions and experimental studies for
four different aqueous H2SO4 solutions. At low temperatures the predictions of the Pitzer model
of this study coincides with the ones of the AIM model (Clegg et al., 1998) only in the case
of a H2SO4 solution 5.5 mol kg-1 in concentration. The solubilities shown in Fig. 4.12 for the
other H2SO4 solutions derived by the Pitzer model in this study axe up to a factor of three
lower than the predictions of the AIM model (Clegg et al., 1998) over the entire temperature
range. The experimental data of Zhang et al. (1993a) axe also up to a factor of three lower than
the AIM predictions. The experimentally obtained data coincides with the predictions of the
Pitzer model of this study in the case of a 8.35 mol kg-1 H2SO4 solution. But for an aqueous
H2SO4 solution 10.2 mol kg-1 in concentration the experimentally obtained data is closer to the
predictions of the AIM model (Clegg et al., 1998). In the case of a H2SO4 solution 15.23 mol
kg-1 in concentration most of the experimentally obtained data points correspond to the AIM
model except the data of Zhang et al. (1993a) which agrees with the solubility values derived in
this study.
The vapor pressure measurements of HCl over mixtures of HCl in aqueous H2SO4 result in an
absolute determination of #hC1 whereas in reactive uptake experiments the value H^cl-^D{is obtained. Therefore, an accurate determination of the solubility from reactive uptake
experiments requires the knowledge of the liquid-phase diffusion coefficient, D\. D\ is a function
of viscosity, which itself depends strongly on temperature (Williams and Long, 1995). The
uncertainty in D\ could be a possible explanation for the scatter within the experimental data
derived by reactive uptake experiments (Hanson and Ravishankara, 1993; Williams and Golden,
1993; Elrod et al., 1995; Hanson, 1998; Robinson et al., 1998) and, hence, their deviations
from the models. Carslaw et al. (1995a) expect an uncertainty in the experimentally obtained
solubihty values of a factor of three due to an uncertainty in the viscosity of up to a factor of
10.
As discussed by Elrod et al. (1995) the measurements of Zhang et al. (1993a) probably suffered
from instrument calibration errors. The tendency of lower solubilities of Zhang et al. (1993a)
compared to the AIM predictions is similar to the solubihty values derived by the Pitzer model
of this work. The solubility values of this study and Zhang et al. (1993a) even coincide for the
H2SO4 solution with a concentration of 15.23 mol kg-1.At this point, it cannot be concluded which of the two models is more accurate in predictingthe HCl uptake by aqueous H2SO4 aerosols. The scatter within the experimental data sets and
the alternating agreement of the experimentally obtained solubilities with both models does
not allow a final decision. However, it should be noted that the Pitzer model presented in this
work is thermodynamically more consistent than AIM.
68 CHAPTER 4. THERMODYNAMIC PROCESSES
Temperature [K] Temperature [K].180 200 220 240 260 280 300 180 200 220 240 260 280 300
10 I i —r——i—>—i—>—i ' r——i—>—i i i—> i 110
10
"E 109CO
L_ 10'
10
,8 -
o
E io7
o
.x 10'
X
10'
,6
,5
10*
-i—>—i—>—i
5.5 mol kg"1 .
\ \'
* V* \.
«
* V.* V.* \r.* X.* X.* X. AIM
/^*!Zhang et al.xOx>. Pitzer model
data * ^Cv. /* X^'t
*^^*.*^%
^.-
10.2 mol kg'1.
8.35 mol kg".
i i
i
L - - -
10"
10°
1180 200 220 240 260 280 300 180 200 220 240 260 280 300
Temperature [K] Temperature [K]
Figure 4.12: HCl solubilities for the indicated concentrations are shown as a function of temperature.
The dotted line represent HHCl predictions of the AIM model (Carslaw et al, 1995a). The dashed line
shows solubility data given by Zhang et al (1993a). The solid line is calculated by the Pitzer model of
this study using the newly derived Ku(T). The open circles represent solubility values derived by vapor
measurements. The solid symbols represent solubility values obtained by reactive uptake measurements.
In the case of the H2SO4 solutions 5.5, 8.35, and 10.2 mol kg-1 in concentration the following denotation
is applied: o Hanson and Ravishankara (1993); • Hanson (1998); Williams and Golden (1993);Elrod et al (1995). In the case of the 15.23 mol kg~1 H2SO4 solution the following denotation is applied:
• Hanson and Ravishankara (1993) ( data of a 15.1 mol kg-1 H2SO4 solution); Williams and Golden
(1993).
4.9. Analysis of H2SO4/H2O Raman spectra 69
Because HCl and HBr have similar molecular properties, one would expect lower solubility
predictions in aqueous H2SO4 solutions for HBr, when using the newly derived H2SO4/H2Ointeraction parameters within a ternary H2S04/HBr/H20 Pitzer model. The original AIM
model (Carslaw et al., 1995a) derived higher HBr solubility predictions than the experimentallyobtained HBr solubility data (Abbatt, 1995; Abbatt and Nowak, 1997; Williams and Long,
1995; Kleffmann et al., 2000). Therefore, Massucci et al. (1999) revised the original AIM
model (Carslaw et al., 1995a) in order to get a better agreement with the experimentallyobtained data. However, the solubihty values derived by vapor pressure measurements are still
a little lower than the predictions of the revised AIM model (Massucci et al., 1999). Therefore,
the newly derived thermodynamic constant should be implemented within the AIM model
(Massucci et al., 1999) to reanalyze HBr solubilities, possibly leading to a better agreement
with experimental data.
If the HCl solubility is indeed a factor of up to three lower than assumed in previous studies
(Carslaw et al., 1995a) this will have consequences also for the heterogeneous reaction rate coeffi¬
cient of HCl on aqueous H2SO4 particles. The heterogeneous reaction rate coefficient will change
linearly with a change in the effective Henry's law constant (see appendix C, Eq. C.I). A lower
solubility of HCl results in a lower reaction probability of HCl with CIONO2 on aqueous H2SO4
aerosol particles and, therefore, leads to a lower amount of activated CI2. Becker et al. (1998)have shown that the high ozone loss rates at the end of January in the Arctic obtained by the
MATCH analysis (von der Gathen et al., 1995), cannot be simulated with their photochemicalbox model. This model includes 11 heterogeneous reactions on NAT and ice, 3 reactions on SAT,and 8 reactions on H2SO4/HNO3/H2O solutions, all based on the analytical expressions of the
predictions of the AIM model (Carslaw et al., 1995b,a). Sensitivity studies of ozone loss rates
within the model of Becker et al. (1998) show no strong dependence on details of the heteroge¬
neous chemistry, i. e. particle formation, temperature dependence, and the negligence of chlorine
deactivation. However, a sensitivity study with respect to heterogeneous reaction rates was not
performed. The HCl solubilities derived in this study imply lower heterogeneous reaction rates
and, thus, even lower chlorine activation will be obtained in the box model calculations. This
will enhance the deviations between box model simulations and the MATCH-analysis in Becker
et al. (1998). Therefore, it can be concluded that there must be other reasons than wrong
HCl solubilities for the observed differences such as additional chemical reactions or dynamicalinfluences like exchange of stratospheric and tropospheric air masses.
4.9 Analysis of H2SO4/H2O Raman spectra
In this section Raman spectra of H2SO4/H2O as a function of temperature and concentration
will be discussed.
The temperature dependence of liquid-phase Raman spectra is shown in Fig. 4.13 for a
H2SO4/H2O solution with a concentration of 6.79 mol kg-1. It can be seen that the inten¬
sity of the i/i(HS04 ) vibration band decreases with decreasing temperature due to the increase
70 CHAPTER 4. THERMODYNAMIC PROCESSES
t—|—i—|—i—|—i—|—i—i——r
"i(HSO„")
^^-^'-
-
-: i i ^-i
500 1000 1500 2000 2500 3000 3500 4000
Raman shift [cm'1]
Figure 4.13: Raman spectra of a H2SO4/H2O droplet 0.5 ßL in volume and a concentration of 6.79
mol kg~l. Spectra are shown from 290 K in 10 K steps. Individual spectra are shifted vertically for better
visibility. The Raman spectra are normalized to the ui(SÖ4~) vibration band.
of the dissociation of the bisulfate ion (see previous sections). The shapes of the vi(H2O) and
1/3(^0) vibration bands change with decreasing temperature due to the reorientation of the
hydrogen bonds between the water molecules. In the literature the reasons for the change in
the shape of the water vibration bands is still discussed. In pure water it is assumed that the
v\(H2O) and 1/3(^0) vibration bands change due to temperature dependent intra- and inter-
molecular coupling and Fermi resonances (Ratcliffe and Irish, 1982; Zhelyaskov et al., 1988).Since in the presented study an aqueous acidic solution is considered the explanation for the
change in the water vibration bands is more complex and, thus, will not be discussed in this
work. The Raman spectra which are taken at temperatures lower than 220 K are supercooledwith respect to SAH and SAT.
The instrumental setup can also be used to record Raman spectra of small droplets with a
diameter of about 50 /uu. Figure 4.14 shows Raman spectra of a H2SO4/H2O droplet with a
concentration of 3.04 mol kg-1 at different temperatures which is similar to the Raman spectra
of Fig. 4.5. A comparison between the two sets of Raman spectra shown in figures 4.5 and 4.14
cannot be performed quantitatively, since the H2SO4 concentration of the solutions differs about
0.5 mol kg-1. But the similar quality of both Raman spectra indicates that even spectra taken
from small droplets can be used for a quantitative analysis.
4.9. Analysis of H2SO4/H2O Raman spectra 71
1—1—1—1——!—1—1—1—1—if—r
Figure 4.14: Raman spectra of a H2SO4/H2O droplet with a volume of 6.5-10~b ßl and a concentration
of 3.04 mol kg~l. Spectra are shown from 290 K in 10 K steps until freezing occurs (190 K). The dashed
line corresponds to a Raman spectrum of the frozen H2SO4/H2O droplet. Individual spectra are shifted
vertically for better visibility. The Raman spectra are normalized to the v\ (SO4- ) vibration band.
Figure 4.15 shows liquid-phase Raman spectra of a H2SO4/H2O droplet at 250 K for varying
H2SO4 concentrations. As the concentration increases, the amount of HSOJ increases too as
indicated by the rise of the ^(HSOJ) vibration band at about 1050 cm-1. The increase in con¬
centration can also be seen in a strong decrease in the vibration bands corresponding to water
in the range of 2800-3700 cm-1. Note that the Raman spectra of solutions with concentrations
of 0.54-2.55 mol kg-1 were supercooled with respect to ice.
The line intensity area of a particular vibration band is proportional to the molecular number
of the species in the solution. Thus, for a quantitative analysis of the concentration of a species
in solution the line intensity areas must be obtained. This is done by simultaneously fitting
Lorentzian functions to the peaks. Figure 4.16 shows such a multiple Lorentzian fit to a Raman
spectrum. It can be seen that the characteristic vibration bands of the corresponding molecules
can be determined easily.The various Raman spectra recorded as a function of temperature and concentration can be used
to derive a relation between the Raman spectra and the corresponding H2SO4 concentrations.
This has been done by defining R, the ratio of the integrated line intensities of the SO|" and
HSO4 vibration bands in the range of 800-1300 cm-1 (with a negligible intensity stemming from
72 CHAPTER 4. THERMODYNAMIC PROCESSES
500 1000 1500 2000 2500 3000 3500 4000
500 1000 1500 2000 2500 3000
Raman shift [cm"1]
3500 4000
Figure 4.15: Liquid-phase Raman spectra of H2SO4/H2O droplets 0.5 ßl in volume and varying concen¬
trations given in molality at 250 K. The spectrum of the solution with a concentration of 0.54 m°l kg-1is recorded at 260 K. Individual spectra are shifted vertically for better visibility. The Raman spectra are
normalized to the î>i(S04~) vibration band at 980 cmT1.
4.9. Analysis of H2SO4/H2O Raman spectra 73
1
-i(S042-) , -
Bc3
-
co
-1—»
JO
-
(0c
.",(HS041
-
1a ^(S048) .
/ VH ^(hso;>V4(HS04-) / / \
/: •. .• \ *s(HsO J .
700 800 900 1000 1100 1200 1300 1400
Raman shift [cm1]
Figure 4.16: Raman spectrum of a H2SO4/H2O droplet 0.5 ßl in volume and a concentration of 6.79
mol kg"1 at 230.5 K. The dotted lines show the Lorentzian functions, which are used to fit the single
vibration bands of the spectrum.
the i*2(H30+) vibration band) and the integrated line intensities of the v\(H.20) and 1/3(^0)vibration bands in the range of 2500-4000 cm-1:
R =
1300
£ /"(SOl") + /"(HSOJ) + /"(H30+)^=800
4000
£ ^(H2o)i/=2500
(4.44)
Figure 4.17 shows R as function of temperature and concentration. R is constant for a par¬
ticular solution concentration over the investigated temperature range within the experimental
uncertainty of about 10 %.
Kamenz (1999) performed Raman spectroscopic measurements of aqueous H2SO4 bulk samples.
He derived a relation between the H2SO4 concentration and the Raman spectrum, expressed as
a weighted ratio, Ryj. The ratio Ry, is derived by the integrated line intensities of the SO|_ and
HSOJ vibration bands in the range of 800-1300 cm-1, Is, and the integrated line intensities of
the i/i(H.20) and 1/3(^0) vibration bands in the range of 2500-4000 cm-1, Ih, each multiplied
74 CHAPTER 4. THERMODYNAMIC PROCESSES
0.7
0.6
0.5
0.4
0.3
0.2
0.1
0.0
180 200 220 240 260 280 300
I
320 3400.7
I
niiiiiiiïnïiiHminium
0.6
0.5
0.4
0.3
0.2
0.1
0.0180 200 220 240 260 280 300 320 340
Temperature [K]
Figure 4.17: R is plotted as function of temperature for various H2SO4 concentrations. From bottom
to top: 0.54, 1.13, 2.55, 6.79, 9.84, 15.23 mol kg-1.
with their corresponding molar masses:
'IsulfRw
98 L
98 • Is + 18 • Ih' (4.45)
where Isuif — 98 • Is and Is = 98/s + 18//,. Kamenz (1999) studied Rw as a function of
concentration for a temperature of 285 K. The R-values of this study have been converted to
iî^-values in order to compare the data of Kamenz (1999) with those of this study. Figure
4.18 shows Rw as function of molal concentration. The i^-values derived by Kamenz (1999)and the i^-values obtained in this study agree within the experimental uncertainty3 Since the
data of Kamenz (1999) were obtained only at 285 K the i^-values derived in this study in
the temperature range of 180-324 K provide a significant improvement. Therefore, the analysis
of the relation between Raman spectrum and H2SO4 concentration has been expanded to this
temperature range by fitting the experimental data to Eq. 4.45.
The ratio Ru, for a given solution concentration can be obtained from the following function:
Rvim) = 0.17478(±0.00988) • mO-53359(±o.028i4)j (4.46)
3This and the fact that R is independent of temperature for each concentration (see Pig. 4.17) indicates that
no concentration changes have occurred in the investigated droplets during the cooling experiments.
4.10. Analysis of (WH4J2SO4/H2O Raman spectra 75
Concentration [mol kg ]
Figure 4.18: Rw, the weighted ratio of SO\~ containing molecules to the sum of SO\~ containing
molecules and water molecules, is plotted as function of concentration. The squares with corresponding
error bars are R^,-values derived in this study obtained in a temperature range of 180-324 K. The triangles
represent data of Kamenz (1999) valid for a temperature of 285 K. The solid line represents a fit of the
Rw-values derived in this study.
where m is the H2SO4 concentration in mol kg-1. The derivation of the H2SO4 concentration
from a Raman spectrum of a solution of unknown concentration can be obtained using the
following function, which is the inverse function of Eq. 4.46:
miRu,) = 27.05586(±1.2436) • ^93228(^.09104) > (4.47)
The error in concentration associated with Eq. 4.47 is about 7.5 %. This value was determined
by analyzing the fitting procedure with a Gauss error propagation law.
4.10 Analysis of (NH4)2S04/H20 Raman spectra
Figure 4.19 shows Raman spectra of an aqueous (NH4)2SÜ4 droplet with a concentration of
5.35 mol kg-1 for varying temperatures. The measurements reach well into the temperature
regime which is supercooled with respect to solid (NH4)2S04 and ice (see Fig. 2.2). Below
253 K the droplet is slightly higher supersaturated with respect to solid (NHi)2S04 than with
respect to ice. The Raman spectrum at a temperature of 226 K is still a spectrum of liquid
76 CHAPTER 4. THERMODYNAMIC PROCESSES
500 1000 1500 2000 2500 3000 3500 4000
Raman shift [cm"1]
Figure 4.19: Raman spectra of a fiVHt^504/02 0 droplet 1.0 ßl in volume and a concentration of 5.35
mol kg-1. Spectra are shown from 296 K in 10 K steps until freezing occurs (216 K). Individual spectra
are shifted vertically for better visibility. The Raman spectra are normalized to the ^(SO^-) vibration
band.
(NH4)2S04/H20. At a temperature of 216 K the strong signal of ice can be observed in the
spectrum. In addition, the shift of the ^(SO^-) normal vibration to lower wavenumbers, the
splitting up of the ^(SO^-) normal vibration, and the decrease in band width of the ^(SOf-)normal vibration indicates the possible formation of solid (NH4)2S04. This indication will be
discussed in further detail in the next section. From the presented Raman spectra it cannot be
concluded which solid phase, ice or (NH4)2S04, nucleated first.
The variation of the integrated fine intensities corresponding to the SO4- ion with temperature
is not as pronounced as in the case for aqueous H2SO4 solutions. The reason for this is the
complete dissolution of (NH4)2S04 into NH^ and SO|" by the reaction (NH4)2S04 <=* 2 NHj+ SO4-. This is corroborated by the Raman spectra which show no signal of HSOJ.Figure 4.20 shows liquid-phase Raman spectra of (NH4)2S04/H20 droplets at 250 K for varying
concentrations. The change in concentration due to water uptake can only be seen in an increase
of the normal vibration of water {y\(H.20) and uz{ß.iO)).
4.11. The ferroelectric phase transition of CJVH4J2SO4 77
500 1000 1500 2000 2500 3000 3500 4000
'E
CO
^—»
!o>_
COc
4-»
c
5.35_-—-\
3.88 /-^"^X
,
3.17 ^--*\
.
1.95 / \
0.99 / \
. n r
500 1000 1500 2000 2500 3000 3500 4000
Raman shift [cm"1]
Figure 4.20: Liquid-phase Raman spectra of (NH4)2S04/H20 droplets 0.5-1 ßl in volume and varying
concentrations given in molality at 250 K. Individual spectra are shifted vertically for better visibility.
The Raman spectra are normalized to the ^(SO^-) vibration band at 980 cm"1.
4.11 The ferroelectric phase transition of (NH^SC^
Sofid (NH4)2S04 belongs to the group of ferroelectric substances. The characteristic property
of these substances is the distribution of the electric dipole moments in at least two sublattices
78 CHAPTER 4. THERMODYNAMIC PROCESSES
(I, II). Below a critical temperature, the Curie temperature, Tc, the structural symmetry in a
ferroelectrical crystal is lowered leading to an appearance of a spontaneous electrical polariza¬
tion of the crystal. These polar properties axe lost above Tc in the so-called paraelectric phase.
Since the electrical polarization changes during the phase transition, the dielectric constant also
alters significantly at Tc.
(NH4)2S04 undergoes a structural phase transition at 223 K associated with a change in space
group from D^/Pnam in the paraelectric phase to C\vIPna2\ in the ferroelectric phase. Sev¬
eral models have been proposed to describe the microscopic mechanism of the transition in the
crystal. Up to now there is a discussion in the literature about the microphysical mechanism
of the phase transition and, also, wether it is a first or second order phase transition. Below,
three commonly discussed models will be presented and their strengths and weaknesses will be
indicated.
O'Reilly and Tsang (1967a) suggested that the ferroelectricity of (NHi)2S04 is due to a dis¬
tortion of NH4 ions and that the transition results from an ordering of the distorted ions with
respect to the mirror plane of the crystal. But such order could not be confirmed by nuclear
magnetic resonance experiments (Miller et al., 1962) and neutron diffraction measurements
(Schlemper and Hamilton, 1966). Furthermore, this order-disorder type mechanism is not able
to explain other features of the phase transition.
An alternative model is that of Sawada and Takagi (1975) who proposed a phase transition of
displacive type. This model attributes the net spontaneous polarization to the shifts of the two
NH4 ions and the SO4- ion along the c-axis from the equilibrium positions in the paraelectric
phase. The basic idea of this model is that a mixed mode of translational and rotational vibra¬
tions of the ions is responsible for the displacement of the NH4 ions. However, an experimental
verification of these vibrations has not been observed in any kind of vibrational spectra (Jainet al., 1973; Iqbal and Christoe, 1976b; Petzelt et al., 1974) and the validity of the basic idea
was also criticized (Jain and Bist, 1974).The third approach was proposed by Jain et al. (1986) and Bajpai and Jain (1987) who de¬
scribe the phase transition as one of a molecular distortion type. This mechanism can take
place only in crystals that have at least one molecular unit and it occurs mainly as a result of
the change in the structure and symmetry of molecular unit(s) rather than in their positions or
orientation. Jain et al. (1986) showed that the distortion in the SO4- ion triggers the transition
and, therefore, can serve as an order parameter for the transition. The external vibration modes
(lattice vibrations) of SO4- and the translational and vibrational modes of NH4" can be found at
frequencies lower than 450 cm-1. This frequency range was investigated by Iqbal and Christoe
(1976a), Iqbal and Christoe (1976b), and Unruh et al. (1978) who corroborate the important
role of the SO|" ions for this phase transition. This model accounts for many properties of the
crystal, including the dielectric anomaly and the heat of the transition which is in agreement
with measured values of Shomate (1945), Hoshino et al. (1958), and Higashigaki and Chihara
(1981). At Tc the SO|" ion undergoes a sudden change in its internal structure, whereas the
NH4 ions do not change significantly at Tc, but undergo a continuous change in a region ± 10
K around Tc. The specific isobaric heat capacity of (NH4)2S04 is shown as line A in Fig. 4.21
(Shomate, 1945). The heat capacity shows the characteristics of a lambda phase transition. This
kind of transition is often assigned to a second order phase transition in ferroelectric substances
due to an order-disorder mechanism. But in the case of (NH4)2S04 it is believed that the fer-
4.11. The ferroelectric phase transition of (NH4)2SÖ4 79
roelectric phase transition is of first order type (Hoshino et al., 1958; Jain et al., 1973; O'Reilly
and Tsang, 1967b). The increase of the specific heat capacity with temperature approaching Tc
(see Fig. 4.21) is due to the slight reorientation of the NHj ions. At Tc the sudden distortion
of the S04_ ions leads to an infinite value of the specific heat capacity which corresponds to a
first order phase transition. Since a deuteration of the NHj-group does not change the Curie
temperature of the phase transition, this further supports that the ferroelectric phase transition
is of first order type (Hoshino et al., 1958; Jain et al., 1973; O'Reilly and Tsang, 1967b).
160
ia>
8 80
40
0
100 200 300
Temperature [KJ
Figure 4.21: The specific isobaric heat capacity is plotted as a function of temperature (Shomate, 1945).
A: (NH4)2S04; B: NH4Al(S04)2; C: NHiAl(S04)2 H20.
Torrie et al. (1972) derived the Raman-active and infrared-active vibration modes for the param¬
agnetic and ferroelectric phase of solid (NH4)2S04 from Raman spectroscopic and infrared spec¬
troscopic measurements. Here it will be focused on the Raman spectroscopic measurements.
Figure 4.22 and 4.23 show the investigated Raman vibration bands of the SO4- molecule in
(NH4)2S04 obtained in the current work. These bands reveal significant changes when going
from the paraelectric to the ferroelectric phase: The ^(SO^-) vibration band shifts to lower
wave numbers at the transition temperature. The ^(SO^-) vibration band is fourfold degen¬
erated in the liquid phase (see appendix B.l). In the solid paraelectric phase the ^(SO^-)vibration band is twofold degenerated but within the ferroelectric phase the degeneration is
completely cancelled (see Fig. 4.23) which is in agreement with the quoted vibration modes of
Torrie et al. (1972). Table 4.5 compares the given positions of the vibration modes of Torrie
et al. (1972) with the positions measured in this study. Both experimental data sets agree within
the experimental uncertainty. In this work, the reorientation of the NH|(I) and NH^(II) ions
could not be investigated since the low frequency range was not experimentally analyzed.
"T~7
'
I
%--
fJL
„ _
T
l J&' y^? i
80 CHAPTER 4. THERMODYNAMIC PROCESSES
<
'
*N
«T \ \.
*-*
'c : 1 \
3
^ / / - I(01-*~»
2 i&_
Ä
£-(0c<D I4-»
_C
940 950 960 970 980 990
Raman shift [cm"1]
1000
Figure 4.22: The shift in the line intensity of v\ (SO\~) during the ferroelectric transition is shown.
The solid lines correspond to the vi(SO\~) normal vibration in the paraelectric phase (T > 223 K) and
the dotted lines represent the vi(SO\~) normal vibration in the ferroelectric phase (T < 223 K).
Conclusion
The Raman spectra obtained in this study confirm the abrupt molecular distortion of the SO4-ions at Tc. These observations support the model by Jain et al. (1973, 1986), who also sug¬
gested that the phase transition is triggered by the SO4- ions. Therefore, it can be concluded
that the phase transition is of the molecular distortion type (Jain et al., 1986). Due to the
pronounced role of the SO4- ions at Tc and the infinite value of the isobaric heat capacity at
Tc it can also be assumed that the ferroelectric phase transition is a first order transition. In
this study the ferroelectric phase transition was observed to occur at 223.1±0.1 K. Furthermore,
the temperature hysteresis of the phase transition was found to be smaller than 0.3 K which is
in agreement with studies of Hoshino et al. (1958) and Iqbal and Christoe (1976b). Thus, the
ferroelectric phase transition is not kinetically inhibited. This makes the transition an ideal "in
situ" temperature calibration point for spectroscopic aerosol experiments.
4.11. The ferroelectric phase transition of (IVH4J2SO4 81
«
.
<n
c'
3
λ m
re
«-»
/r '^ \re // % / *
A
// v ' * ' \^ -
CO if£- ifCD 1/ ,
Ç\ \
•\>
1 1—
""
1020 1040 1060 1080 1100 1120 1140 1160 1180
Raman shift [cm"1]
Figure 4.23: The splitting of the doublett of the vs(SO\~) vibration band during the ferroelectric tran¬
sition is shown. The solid lines correspond to the 1/3 (SO\~) normal vibrations in the paraelectric phase
(T > 223 K) and the dotted lines represent the vz(SO\~) normal vibrations in the ferroelectric phase (T
< 223 K).
Table 4.5: Raman frequencies of the SO\~ ion in solid fJVÏÏJ)2S04 at 223.1 K (paraelectric phase)
and at 223 K (ferroelectric phase). Vibration, character, maximum position determined by Torrie et al.
(1972) and maximum position derived in this work are presented.
normal vibration character max. position
V [cm-1]Torrie et al. (1972)
max. position
V [cm-1]this work
paraelectric v\ Ai 977 ±2 972 ±4
phase vz F2 1062 ± 10 1061 ± 4
vz F2 1106 ± 10 1090 ± 4
ferroelectric v\ Ai 972 ±2 970 ±4
phase vz F2 1043 ± 10 1053 ± 4
vz F2 1077 ± 10 1083 ± 4
vz F2 1124 ± 10 1115 ±4
vz F2 1147 ± 10 1138 ± 4
Seite Leer /
Blank leaf
Chapter 5
Kinetic processes in UT/LS aerosol
particles
This chapter presents the results of studies on atmospheric kinetic processes: the nucleation
of solid phases in liquid stratospheric aerosol particles and the nucleation of ice in aque¬
ous (NH4)2S04 aerosols. A theoretical and experimental analysis of the derivation of homo¬
geneous nucleation rate coefficients of NAD and NAT from binary HNO3/H2O and ternary
HNO3/H2SO4/H2O solutions is given. Sections 5.1 to 5.6 are identical to the publication "Ho¬
mogeneous nucleation of NAD and NAT in liquid stratospheric aerosols: insufficient to explain
denitrification" published in Atmospheric Chemistry and Physics, 2, 207-214, 2002. Section 5.7
deals with a newly proposed pseudo-heterogeneous nucleation mechanism, i. e. the nucleation is
induced at the surface of the particle (Tabazadeh et al., 2002a,b; Djikaev et al., 2002). There¬
fore, the experimentally derived volume-based nucleation data will be reanalyzed with respect to
surface-induced nucleation. The last section discusses results of homogeneous ice nucleation ex¬
periments on aqueous (NHi)2S04 solutions. The experimental data of this work are comparedto homogeneous ice nucleation rate coefficients obtained by a different experimental method.
The newly proposed pseudo-heterogeneous nucleation mechanism will also be considered in the
analysis of the ice nucleation data.
83
84 CHAPTER 5. KINETIC PROCESSES
Seite Leer /
Blank leaf
85
Homogeneous nucleation of NAD and NAT in liquid stratospheric aerosols:
insufficient to explain denitrification
D. A. Knopf*, T. Koop, B. P. Luo, U. G. Weers, and T. Peter
Institute for Atmospheric and Climate Science, Swiss Federal Institute of Technology,
Honggerberg HPP, 8093 Zurich, Switzerland
* To whom correspondence should be addressed. Email: Daniel.Knopf@iac.umnw.ethz.ch.
Published in Atmospheric Chemistry and Physics, 2, 207-214, 2002.
86 CHAPTER 5. KINETIC PROCESSES
Seite Leer /
Blank leaf
5.1. Abstract 87
5.1 Abstract
The nucleation of NAD and NAT from HN03/H20 and HNO3/H2SO4/H2O solution droplets
is investigated both theoretically and experimentally with respect to the formation of polar
stratospheric clouds (PSCs). Our analysis shows that homogeneous NAD and NAT nucleation
from liquid aerosols is insufficient to explain the number densities of large nitric acid containing
particles recently observed in the Arctic stratosphere. This conclusion is based on new droplet
freezing experiments employing optical microscopy combined with Raman spectroscopy. The
homogeneous nucleation rate coefficients of NAD and NAT in liquid aerosols under polar strato¬
spheric conditions derived from the experiments axe < 2 x 10-5 cm-3s-1 and < 8 x 10~2
cm-3s-1, respectively. These nucleation rate coefficients are smaller by orders of magnitude
than the value of ~ 103 cm-3 s-1 used in a recent denitrification modelling study that is based
on a linear extrapolation of laboratory nucleation data to stratospheric conditions (Tabazadehet al., Science, 291, 2591-2594, 2001). We show that this hnear extrapolation is in disagree¬
ment with thermodynamics and with experimental data and, therefore, must not be used in
microphysical models of PSCs. Our analysis of the experimental data yields maximum hourly
production rates of nitric acid hydrate particles per cm3 of air of about 3 x 10-10 cm-3(air) h-1
under polar stratospheric conditions. Assuming PSC particle production to proceed at this rate
for two months we arrive at particle number densities of < 5 x 10-7 cm-3, much smaller than
the value of ~ 10-4 cm-3 reported in recent field observations. In addition, the nitric acid
hydrate production rate inferred from our data is much smaller than that required to reproducethe observed denitrification in the modelling study mentioned above. This clearly shows that
homogeneous nucleation of NAD and NAT from liquid supercooled ternary solution aerosols
cannot explain the observed polar denitrification.
5.2 Introduction
Polar stratospheric cloud (PSC) particles activate chlorine from reservoir to reactive species
by heterogeneous reactions on their surfaces. Field measurements have shown that PSCs can
be composed of liquid supercooled ternary solutions (STS) and nitric acid trihydrate (NAT)
(Schreiner et al., 1999; Voigt et al., 2000). In addition, nitric acid dihydrate (NAD) has been
suggested to exist in PSCs based on laboratory experiments (Worsnop et al., 1993). Large
HN03-containing PSC particles can lead to significant denitrification of the polar stratosphere
by sedimentation (Fahey et al., 2001). However, the mechanisms of how such large particles
come about have not yet been established (Tolbert and Toon, 2001). It has been suggested that
large nitric acid hydrate particles leading to denitrification could be produced by homogeneous
nucleation of NAD and NAT from liquid STS (Tabazadeh et al., 2001), based on an extrapolation
of laboratory aerosol nucleation data (Salcedo et al., 2001) to stratospheric conditions. However,
the employed extrapolation is in disagreement with bulk nucleation experiments performed at
stratospheric conditions (Koop et al., 1997b). For example, according to the nucleation formula¬
tion an aqueous ternary solution of 41.2 wt% HNO3 and 3.9 wt% H2SO4 and 1 cm3 in volume is
predicted to freeze at 249.0 K. In contrast, in experiments such samples did not freeze down to
88 CHAPTER 5. KINETIC PROCESSES
temperatures of 190 K (Koop et al., 1995). Also, the formulation is in disagreement with aerosol
experiments of aqueous nitric acid solutions (Bertram and Sloan, 1998b,a; Bertram et al., 2000a;
Salcedo et al., 2001). To resolve these inconsistencies we investigate here the nucleation kinetics
of NAD and NAT in liquid binary HN03/H20 and ternary HNO3/H2SO4/H2O solutions both
theoretically and experimentally. First, we reexamine the physics of the nitric acid hydrate
nucleation formulation used by Tabazadeh et al. (2001). Second, we present new experimentaldata on NAD and NAT nucleation from STS droplets under stratospheric conditions. Third,
we use these data together with previously published data sets to deduce upper limits of ho¬
mogeneous nucleation rate coefficients of NAD and NAT. Finally, from the inferred nucleation
rate coefficients we derive maximum production rates of sofid nitric acid particles under polar
stratospheric conditions.
5.3 Nucleation formulation analysis
Salcedo et al. (2001) and Tabazadeh et al. (2001) have employed classical nucleation theory to
describe the experimentally observed homogeneous nucleation rate coefficients, Jhom-, of nitric
acid hydrates (NAX; X = D or T):
Jhom(T) = n-Hq f — j exp-AGact(T)
RT(5.1)
where nuq is the HNO3 molecular number density in the liquid, R is the universal gas constant,
k is the Boltzmann constant, and h is the Planck constant. AGact is the activation energy
required to form a critical cluster in the solution. According to classical nucleation theory this
activation energy depends on the saturation ratio of the respective nitric acid hydrate («Snax):
AGact(T) = yTra3 (T)Vsoi
[RTln(SNAX)AGdif (T) (5.2)
Here, asi is the interfacial tension between the solid and liquid phase, vsoi is the molar volume
of NAX in the critical cluster, and AGdif is the HNO3 diffusion activation energy across the
boundary between the cluster and the solution. Since measured values for crsi and AGdif are not
available (MacKenzie, 1997), AGact can be determined from experimentally observed nucleation
rate coefficients by solving Eq. 5.1 for AGad,:
AGact(T) = -RT Inh Jhom(T)kT ring
(5.3)
Figure 5.1 shows values of AGact in aqueous nitric acid solutions as plotted by Salcedo et al.
(2001) as function of the NAD and NAT saturation ratio derived from their experimental data
using Eq. (5.3). The laboratory data reveal a hnear relationship between AGact and Snax
(solid lines in Fig. 5.1) in the experimentally observed range of saturation ratios (Snad=11_
30, 5nat=52-107). Since stratospheric saturation ratios (shaded areas in Fig. 5.1) are much
smaller than the experimentally investigated range, Tabazadeh et al. (2001) used a linear ex¬
trapolation (dotted lines in Fig. 5.1) to infer AGact-values for NAD and NAT at stratospheric
5.3. Nucleation formulation analysis 89
200
190
^ 180
170
160
40
T-1 35
o
1 30
D
1 25
J 20
<
15
^f
-
10 20 30 0 30 60 90 120
'NAD 'NAT
Figure 5.1: AGact as function of the NAD and NAT saturation ratios derived from laboratory nucleation
data using Eq. (5.3) (Salcedo et al, 2001). All data points were derived from experiments with droplets
consisting of binary aqueous nitric acid solutions of varying composition, (a): : 57 wt%, 60 wt%, and
64 wt% HNOz Salcedo et al (2001); : 64 wt% HN03 (Bertram and Sloan, 1998b). (b): ; 54 wt%
HNO3 (Salcedo et al., 2001); : 54 wt% HNO3 (Bertram and Sloan, 1998a). The color coding indicates at
which temperature the data were obtained. The shaded regions indicate typical NAD and NAT saturation
ratios at polar stratospheric conditions. The solid lines show the linear relationship between AGact and
Snax observed by Salcedo et al (2001), and the dotted lines are the linear extrapolations to stratospheric
conditions used in Tabazadeh et al (2001).
conditions. However, applying such an extrapolation is physically unreasonable because accord¬
ing to Eq. (5.2), AGact increases towards infinity for -Snax approaching unity. In contrast, the
linear extrapolation leads to a AGoct-value of about 30 kcal mol-1 in each case. Note, that an
underestimation of AGact by 1 kcal mol-1 increases the corresponding homogeneous nucleation
rate coefficient by a factor of 14. Therefore, the Hnear extrapolation underestimates AGact and,
consequently, largely overestimates the homogeneous nucleation rate coefficient at low satura¬
tion ratios. In Fig. 5.2 we elucidate the effects of the hnear extrapolation on the homogeneous
nucleation rate coefficients of NAD and NAT in aqueous nitric acid solutions. The nucleation
formulation produces finite nucleation rate coefficients along the NAD and NAT melting point
curves (where Snax=1) an(l does so even for values of -Snax < 1 (not shown in Fig. 5.2). This is
thermodynamically impossible for any spontaneous process, since the formation of an unstable
crystal (similar to nucleating ice above 273.15 K) would lead to an increase of the total Gibbs
free energy of the system. This nucleation formulation produces unrealistically high nucleation
rate coefficients of about 108 cm-3 s-1 at the top of the NAD and NAT melting curves (dark
yellow region at T — 230 K in Fig. 5.2a and T = 250 K in Fig. 5.2b) in disagreement with
numerous experimental studies (Anthony et al., 1997; Koop et al., 1997b; Bertram and Sloan,
1998b,a; Bertram et al., 2000a; Salcedo et al., 2001). In addition, at stratospheric tempera¬
tures (180-200 K) and saturation ratios (between the solid and dotted lines in Fig. 5.2) the
homogeneous nucleation rate coefficient increases with temperature. This behavior is due to the
90 CHAPTER 5. KINETIC PROCESSES
OT
10
10
15
11
E 107ifL 3^
10°E° 1
-f 10_1
10-5
260ICE \ »»w~V°)
240 V.
_\NAM
*T 220
/ NAD^tr—
200
'/ i1
180 !/ A
ICE NAT,
0 20 40 60 80 0
HN03 [wt%]
20 40 60 80
HNO3 [wt%]
Figure 5.2: Homogeneous nucleation rate coefficients of NAD (a) and NAT (b) in binary HNO3/H2O
solutions as function of temperature and concentration using the formulation of Tabazadeh et al (2001).
Solid lines show the melting point curves of the different solid phases (S = 1). The regions between dotted
and solid lines indicate typical polar stratospheric temperatures (< 200 K) and saturation ratios (< 4- 7
for NAD and < 23.5 for NAT). Black asterisks correspond to the experimental data shown in Fig. 5.1.
fact that in the formulation AGact depends solely on SnaX) independently of the temperature
(Tabazadeh et al., 2001, note 21). Only in the proximity of the experimental data (black aster¬
isks in Fig. 5.2) a reasonable temperature and concentration dependency of Jhom is observed. We
conclude that the linear relationship between AGact and Snax should not be used outside the
range of available experimental data and, therefore, should not be extrapolated to stratosphericconditions.
5.4 Experimental
Freezing experiments with HNO3/H2O and HNO3/H2SO4/H2O droplets were performed in or¬
der to determine homogeneous nucleation rate coefficients of NAD and NAT at stratospheric
saturation ratios. We chose to investigate large droplets (0.12-0.27 cm in diameter) because
smaller droplets (2 x 10-5-8.5 x 10-3 cm in diameter) do not freeze at stratospheric tempera¬
tures and saturation ratios (Anthony et al., 1997; Bertram and Sloan, 1998b,a; Bertram et al.,
2000a; Salcedo et al., 2001). The droplets were deposited with a micropipette on a hydrophobi-
cally coated quartz plate inside a laminar flow clean bench. Either a Teflon plate or an o-ring,
each covered by a thin layer of high-vacuum-grease, served as a spacer for a second quartz plate
which sealed the droplets against ambient air. The inner diameter of the spacer depended on
the investigated droplet volume and varied between 0.3-0.6 cm and the spacer thickness ranged
between 0.125-0.175 cm. The total volume of the cell was about 8.8 x 10~3-5 x 10-2 cm3. The
volume of the droplets varied between 10-3-10-2 cm3. Therefore, even at room temperature the
5.4. Expérimental 91
number of water and HNO3 molecules in the gas phase of the cell is neghgible when compared
to the number of condensed water and HNO3 molecules in the droplets. Hence, the composition
of the droplets stays constant during a freezing experiment. The preparation of the droplet cell
took about 15 s. The number of molecules which may evaporate during that time is neghgible
to the total number of molecules in the condensed phase. This was confirmed by checking the
melting points of the droplets after freezing which were found to be in agreement with the phase
diagram. After sealing the droplets against ambient air with a second plate, the droplet cell
was placed on a temperature stage attached to a Confocal Raman Microscope (see Fig. 5.3). In
video
analysis
laser
_^
532 nm
microscope
spectrographCCD detector
temperature stage
grating:1800 g/mm
aerosol cell
Figure 5.3: Sketch of the experimental setup.
this setup the droplets' temperature can be varied between 170-295 K. The temperature was
calibrated by measuring the melting points of heptane (182.55 K), octane (216.35 K), decane
(243.45 K), dodecane (263.5 K), and water (273.15 K) in the cell. Phase changes (i.e. freezing or
melting) are observed visually with the microscope part of the setup. In addition, the crystalline
solids formed upon freezing were identified by Raman spectroscopy using a Nd:YAG-laser at a
wavelength of 532 nm for illumination. The backscattered light is reflected onto a grating (1800
mm-1) and focused on the CCD detector of the spectrograph. The resulting spectral resolution
is about 2-4 cm-1 within the observed range of 500-4500 cm-1. Figure 5.4 shows Raman spec¬
tra of droplets (10-2 cm3) with an HNOs:H20 mole ratio of 1:2 and 1:3 corresponding to the
stoichiometry of NAD and NAT, respectively. In each case the spectra were recorded during a
cooling cycle (red spectra) and a warming cycle (blue spectra) at about the same temperature.
To avoid any possible temperature bias the droplets were not illuminated by laser fight duringthe course of the freezing experiments reported below. Spectra were taken only after the droplets
were frozen.
Table 1 shows the composition, volume, total number of different performed experiments, and
total number of individual droplets. The droplets were prepared from stock solutions which
92 CHAPTER 5. KINETIC PROCESSES
-i 1 r
HN03:H201:2 (a):
yy
hquid/\ 211 K.
JL
500 1000 1500 2000 2500 3000 3500 4000
0.95
co
c3
2
CO
x
0.3
0.25 -
0.2
500
T 1 1 •"
HN03:H201:3
- 1 -
(b>:
jf
liquid » 211 K
frozen
1000 1500 2000 2500 3000 3500 4000
Raman shift [cm"1]
Figure 5.4: Raman spectra of droplets with a volume of 10~2 cm3, (a): Red line: spectrum of a liquid
droplet with a HNO3.H2O mole ratio of 1:2 at 211 K; blue line: spectrum of a frozen droplet at 212
K. (b): Red line: spectrum of a liquid droplet with a HNO3.H2O mole ratio of 1:3 at 211 K; blue line:
spectrum of a frozen droplet at 211 K. The spectra are normalized with respect to the ui(NO^) vibration
band at ~1040 cmT1.
5.4. Experimental 93
Table 5.1: Composition, volume, total number ofperformed experiments, and total number of individual
droplets. The symbols refer to the ones in Fig. 5
Solution HN03 H2SO4 H20 Volume Symbol #Exp. # Drop.
[wt%] [wt%] [wt%] [10-3cm-3]1 63.6 0 36.4 5-10 X 16 16
2 53.8 0 46.2 1-10 + 28 28
3 32.2 13.8 54.0 10 * 22 5
4 38.3 7.6 54.1 10 • 16 4
were titrated against a 1 M NaOH solution. In an experimental run the droplets were cooled
at a rate of dT/dt = —10 Kmin-1 until nucleation occurred. The ternary solution droplets
(solution 3 and 4, Table 5.1) did not freeze above 178 K during such runs. Hence, the tem¬
perature in these experiments was decreased stepwise (by 5-10 K) keeping the temperature
constant for several minutes after each step. All experiments were recorded on tape together
with the experimental time and droplet temperature. The video tapes were analyzed after¬
wards to determine the number of nucleation events, n, as a function of time and temperature.
The upper limit of the homogeneous nucleation rate coefficient J^ can be derived from the
experimental data using the following formula:
n*
«CCO =
£y,.t,(D' (5"4)
i
where tt(T) = fT' [dT"/dt)~ldT' is the time interval that the ith droplet with volume Vt re¬
mained liquid between T and T*. T* is either the nucleation temperature of the droplet or
the lowest investigated temperature, and (dT/dt)t is the cooling rate applied in the particular
experiment, n* is the upper fiducial limit of n determined by Poisson statistics at a confidence
level of 0.999 (Koop et al., 1997b) - i.e., if the experiments were repeated an infinite number of
times the observed number of nucleation events will be smaller than n* in 99.9 % of the cases.
Equation (5.4) yields a conservative (i.e. the highest possible) Jhom-vahie, which is in agreementwith the experimental data.
In detail, Eq. (5.4) is conservative for the following reasons: First, the time interval, U(T), that
a droplet stays liquid below T is always smaller than the time interval it would stay liquid at
T (assuming that Jhom monotonically increases with decreasing temperature in the investigated
temperature range). Second, instead of using the actual number of nucleation events, n, we
employed n*, which represents a conservative value for n because n* > n in all cases. Third,
it cannot be ruled out that heterogeneous nucleation of NAD and NAT occurred in the large
droplets. Even in this case, the observed nucleation rate is always an upper limit for the ho¬
mogeneous nucleation rate, independently of whether heterogeneous nucleation occurred or not.
All experimental data were analyzed using Eq. (5.4). The derived Jj^-values were used to
calculate lower limits of the activation energy, AG^, according to Eq. (5.3).
94 CHAPTER 5. KINETIC PROCESSES
5.5 Results and discussion
In Fig. 5.5 the resulting AGact-values axe shown as function of temperature and saturation
ratio. The different symbols in Fig. 5.5 correspond to those in Table 5.1. In addition, we have
reanalyzed published bulk experiments (Koop et al., 1995, 1997b) to determine the upper hmit
for Jhom. according to J^^iT) = n*/(V t) ,where n* is the same as above, V is the volume
of the solution, and t is the time the solution remained liquid at temperature T (Koop et al.,
1997b). The corresponding AG^-values were obtained using Eq. (5.3) and axe shown as open
symbols in Fig. 5.5. The solid symbols in Fig. 5.5 represent the same data as in Fig. 5.1.
Furthermore, we have added the aerosol nucleation data by Bertram et al. (2000a). We note,
that we have used a nucleation rate coefficient of J — 4.4 x 109 cm-3 s-1 for these data, slightly
lower than the one in the original publication (A.K. Bertram, personal communication). Figure
5.5 clearly reveals that the newly derived AGacrvalues are significantly higher than the hnear
extrapolation formulation at stratospheric conditions. Since our data points axe lower limits of
AGact (thus, upper limits of Jhom) the actual values of AGact are likely to be even higher than
those shown in Fig. 5.5. Clearly, the hnear extrapolation used in Tabazadeh et al. (2001) is not
in agreement with our new droplet data nor with bulk experiments published previously. In the
following, we use the combined experimental data (Koop et al., 1995,1997b; Bertram and Sloan,
1998b,a; Bertram et al., 2000a; Salcedo et al., 2001, and this work) to derive upper homogeneous
nucleation rate coefficients of NAD and NAT at stratospheric conditions. Figure 5.7a shows the
composition and corresponding NAD and NAT saturation ratios of STS droplets at 50 mbar
(approx. 20 km altitude) for mixing ratios of 5 ppmv H2O, 10 ppbv HNO3, and 0.5 ppbv H2SO4
(Carslaw et al., 1994). In Fig. 5.7b, NAD and NAT nucleation rate coefficients axe shown for
the conditions displayed in panel (a). In the region of highest saturation ratios (shaded region
in Fig. 5.7) circles and squares represent maximum nucleation rate coefficients derived from
experimental data as follows: In Fig. 5.5 for one temperature (e.g. 191.5 K) all data points are
selected by color, and then interpolated as function of saturation ratio using 5nax read off Fig.
5.7a. From the AGoct-value obtained in this way we derive J^, using Eq. (5.1). Blue and red
arrows mark the temperature where 5nad—1 and 5nat=1j respectively, i.e. where the nucleation
rate coefficients must decrease to zero. Solid lines in Fig. 5.7b represent homogeneous nucleation
rate coefficients calculated using the formulation of Tabazadeh et al. (2001). Figure 5.7c shows
the corresponding NAD and NAT particle production rates for the conditions displayed in panel
(a) using the nucleation rate coefficients shown in panel (b). Solid lines axe calculated with the
equations given by Tabazadeh et al. (2001) taking into account that the total aerosol volume
increases with decreasing temperature (Carslaw et al., 1994). Stars are values taken directly
from Fig. 1 in Tabazadeh et al. (2001). The circles and squares represent the production rates
calculated using the experimentally derived upper nucleation rate coefficients shown in Fig.
5.7b. Figure 5.7c reveals that the maxima of the resulting production rates of the formulation
by Tabazadeh et al. (2001) (solid lines) are too large by a factor of 108 for NAD and 104 for
NAT when compared to the experimentally derived production rates.
5.5. Results and discussion 95
200
190
^ 180
170
160
oo
o<3
o
E
"5o
0 10 20 30 0 30 60 90 120
'NAD
40
K «
(a')
35
»Y^ê*
-
30
•
-..*
-
25 -... *-
°NAT
(b')
t#A
% ^&A*\x
^fefc:*.* -
**
4 8 12 0 10 20 30 40
SNAD SNAT
Figure 5.5: AGact as function of the NAD and NAT saturation ratios derived from laboratory nucleation
data using Eq. 5.3. Large droplet data: x : 63.6 wt% HN03; +: 53.8 wt% HN03; *: 32.2 wt% HNO3 and
13.8 wt% H2SO4; •: 38.3 wt% HNO3 and 7.6 wt% H2S04 (all this work). Bulk solution data: A: binary
HNO3/H2O solutions of varying composition (Koop et al, 1997b). O: ternary HNO3/H2SO4/H2O solu¬
tions of varying composition (Koop et al, 1995, 1997b). Aerosol data: (a): : 57 wt%, 60 wt%, and 64
wt% HNO3 (Salcedo et al, 2001); : 64 wt% HN03 (Bertram and Sloan, 1998b); A: binary HN03/H20
aerosol of varying composition (Bertram et al., 2000a). (b): : 54 wt% HNO3 (Salcedo et al, 2001);
: 54 wt% HNO3 (Bertram and Sloan, 1998a). The color coding indicates at which temperature the
data were obtained. The solid lines indicate the linear relationship between AGact and Snax observed by
Salcedo et al (2001), and the dotted lines are the linear extrapolations to stratospheric conditions used
in Tabazadeh et al (2001). (a') and (b') show an enlarged view of the top left corner of panels (a) and
(b), respectively.
96 CHAPTER 5. KINETIC PROCESSES
Correction of Fig. 5.5:1
200
190
^ 180
170
160
ou
oo
<l
40
35
l (a)
30:
25""-\kj.
n""^Sn^e
20 ^V,15
0 10 20 30 0
°NAD
40
(a')
35
a\^
-
30
r- ^\.
-
75 -•-. "V
30 60 90 120
^NAT
II (b1)
;^„"
i i
*.
4 8
°NAD
12 0 10 20 30 40
'NAT
Figure 5.6: AGact as function of the NAD and NAT saturation ratios derived from laboratory nucleation
data using Eq. 5.3. Large droplet data: x: 63.6 wt% HN03; +: 53.8 wt% HN03; *: 32.2 wt% HN03 and
13.8 wt% H2SO4; •: 38.3 wt% HN03 and 7.6 wt% H2S04 (all this work). Bulk solution data: A: binary
HNO3/H2O solutions of varying composition (Koop et al, 1997b). O: ternary HNO3/H2SO4/H2O solu¬
tions of varying composition (Koop et al, 1995, 1997b). Aerosol data: (a): : 57 wt%, 60 wt%, and 64
wt% HNO3 (Salcedo et al, 2001); : 64 wt% HNO3 (Bertram and Sloan, 1998b); A: binary HN03/H20
aerosol of varying composition (Bertram et al, 2000a). (b): : 54 wt% HNO3 (Salcedo et al, 2001);
: 54 wt% HNO3 (Bertram and Sloan, 1998a). The color coding indicates at which temperature the
data were obtained. The solid lines indicate the linear relationship between AGact o,nd Snax observed by
Salcedo et al (2001), and the dotted lines are the linear extrapolations to stratospheric conditions used
in Tabazadeh et al. (2001). (a') and (b') show an enlarged view of the top left corner of panels (a) and
(b), respectively.
1In the original figure 5.5a the nucleation data of the aqueous 63.6wt% HNO3 droplets are missing.
5.5. Results and discussion 97
S -1
5 10°
to
o3
10s I* *
Ij
103 ; i i i i i i t lui»» i| i i i \ | i
'
'°"U ' " » - i —i L i i—l
186 187 188 189 190 191 192 193 194 195
Temperature [K]
10'
196
Figure 5.7: (a) The composition (green and orange lines) and the saturation ratios (red and blue lines)
of STS aerosols as a function of temperature at 50 mbar with 5 ppmv H2O, 10 ppbv HNO3, and 0.5 ppbv
H2SO4 (Carslaw et al, 1994). The shaded region indicates the temperature range where the S^p^x-values
have their maximum, (b): Upper limits for the nucleation rate coefficients of NAD (squares) and NAT
(circles) in STS droplets, derived from experimental data for the conditions shown in panel (a). For
comparison, solid lines indicate the homogeneous nucleation rate coefficient in STS droplets for the same
conditions calculated using the formulation of Tabazadeh et al (2001). (c): Hourly production rates of
NAD and NATparticles (squares and circles, respectively) per cm3 of air derivedfrom the nucleation rate
coefficients shown in panel (b). The increase of the total aerosol volume with decreasing temperature was
taken into account (Carslaw et al, 1994). Also shown as solid lines are the NAD and NAT production
rates for the same conditions calculated using the formulation of Tabazadeh et al (2001). Stars show
values for similar conditions taken directly from Fig. 1 of Tabazadeh et al. (2001). (We note that we can
reproduce the stars by assuming a constant total aerosol volume of 5.9 x 10~12 cm3.) Arrows in (b) and
(c) mark the temperature where the saturation ratio of NAD (blue) and NAT (red) equals one.
98 CHAPTER 5. KINETIC PROCESSES
5.6 Conclusions
Salcedo et al. (2001) have investigated the nucleation of NAD and NAT from binary aqueous
nitric acid droplets. We consider their experimental data to be sound and the observed linear
relationship between the activation energy, AGaCi, and the respective nitric acid hydrate
saturation ratio, £nax> to be valid in the experimentally observed range of saturation ratios
(5nad=H-30, S'nat=:52-107). However, the theoretical arguments and experimental data
presented above show that the linear relationship between AGact and -Snax is not valid
at stratospheric saturation ratios. Therefore, the linear relationship must not be used in
microphysical models of PSCs.
The analysis of experimental data presented above shows homogeneous NAD and NAT nucle¬
ation rate coefficients to be exceedingly low (< 2 x 10-5 cm-3 s-1 and < 8 x 10-2 cm-3 s-1, re¬
spectively) in STS aerosols under polar stratospheric conditions, in agreement with earlier studies
(Koop et al., 1995, 1997b). These nucleation rate coefficients are smaller by orders of magnitudethan those used in a recent modelling study of stratospheric denitrification (Tabazadeh et al.,
2001). In that study, it was asserted that homogeneous NAD and NAT nucleation from STS
aerosols is sufficient to explain the denitrification observed in the Arctic and Antarctic strato¬
sphere. NAT particle number densities that are in agreement with recent field observations (~10~4 cm-3, (Fahey et al., 2001)) were obtained by converting all NAD particles into NAT parti¬
cles in the simulation. This was achieved by adding the NAD and NAT homogeneous nucleation
rate coefficients. The corresponding particle production rates were about ~ 10-5 cm-3 (air) h-1.
In contrast, using the upper limits for the particle production rates (Fig. 5.7c) derived in this
study and assuming the maximum saturation ratios to persist for two months, we arrive at par¬
ticle number densities of ~ 5 x 10-7 cm-3, much smaller than reported by Fahey et al. (2001).
Furthermore, Tabazadeh et al. (2001) state that NAT particle production rates smaller than ~
10-5 cm-3(air) h-1 are unimportant to denitrification. Even if we combine the NAD and NAT
production rates of Fig. 5.7c the maximum possible value in agreement with the laboratory data
is only ~ 3xl0-10 cm-3(air)h-1. This clearly shows that homogeneous nucleation of NAD
and NAT from liquid supercooled ternary solution aerosols cannot explain the observed polar
denitrification. Therefore, other NAD/NAT formation mechanisms such as heterogeneous NAT
nucleation on ice particles are required to explain polar denitrification (Waibel et al., 1999).
5.6. Conclusions 99
Here ends the publication:
Homogeneous nucleation of NAD and NAT in liquid stratospheric aerosols:
insufficient to explain denitrification
D. A. Knopf*, T. Koop, B. P. Luo, U. G. Weers, and T. Peter
Institute for Atmospheric and Climate Science, Swiss Federal Institute of Technology,
Honggerberg HPP, 8093 Zurich, Switzerland
* To whom correspondence should be addressed. Email: Daniel.Knopf@iac.umnw.ethz.ch.
Published in Atmospheric Chemistry and Physics, 2, 207-214, 2002.
100 CHAPTER 5. KINETIC PROCESSES
Seite Leer /
Blank leaf
5.7. Pseudo-heterogeneous nucleation of PSCs 101
5.7 Pseudo-heterogeneous nucleation of PSCs
Tabazadeh et al. (2002a) claim to have found evidence that pseudo-heterogeneous nucleation
occurs in laboratory nucleation experiments. Therefore, the experimental data shown in section
5.5 are reanalyzed with respect to a potential surface-induced nucleation pathway. In section
2.2.2 the derivation of the surface-based homogeneous nucleation rate coefficient, J/fom, and
the conversion of the volume-based homogeneous nucleation rate coefficient, J)^, into Jhsom is
given. Before reanalyzing the experimentally obtained nucleation data shown in the previous
1015
10" Iw
CM
I
o
10-3
10
10*
-15
260
240
220
200
180
0 20 40 60 80 0 20 40 60 80
HN03 [wt%] HN03 [wt%]
Figure 5.8: The surface-based homogeneous nucleation rate coefficients of NAD (a) and NAT (b) in
binary HNO3/H2 O solutions as a function of temperature and concentration using the formulation of
Tabazadeh et al. (2002a) taken from Knopf et al (2003) The solid lines indicate the melting point
curves of the different solid phases (S = 1). The regions between dotted and solid lines represent typical
polar stratospheric temperatures (< 200 K) and saturation ratios (< 4-7 for NAD and < 23.5 for NAT).
sections (see Fig. 5.7), the parameterization of the pseudo-heterogeneous nucleation of NAD
and NAT given by Tabazadeh et al. (2002a) will be discussed.
Tabazadeh et al. (2002a) give AG^AD and AGf£AT as a function of temperature and HNO3
mole fraction inside the droplets (see appendix E.2). The surface-based homogeneous nucleation
rate coefficient can be derived from Eq. 2.32 using AG^AD and AGf^AT . Figure 5.8 shows
surface-based homogeneous nucleation rate coefficients of NAD (panel a) and NAT (panel b) in
binary HNO3/H2O solutions as a function of temperature and concentration (Knöpfet al., 2003)based on the formulations of Tabazadeh et al. (2002a). The derived J^AD and J^AT-valuesshow a physically unreasonable behavior: As S approaches unity, i. e. the melting curve (solidlines in Fig. 5.8), the surface-based nucleation formulation of Tabazadeh et al. (2002a) predicts
high jf^AD and J^AT-values. This is at odds with CNT, since for S approaching unity AG*d
increases towards infinity (see Eq. 2.26). Hence, Jhgm and Jh'om -values must vanish when
the concentration approaches the melting curves in Fig. 5.8. Furthermore, the temperature
dependence of Jhom at about 60 wt% HNO3 is unusually small, indicating that J^^changes only by three orders of magnitude over a temperature range of 65 K, which seems
102 CHAPTER 5. KINETIC PROCESSES
unlikely. More importantly, the increase in the surface-based nucleation rate coefficient with
increasing HNO3 concentration contradicts nucleation experiments showing that no freezing
occurs between 72 and 75 wt% (Bertram et al., 2000a). Hence, it must be concluded that
the current parameterization for pseudo-heterogeneous nucleation of NAD and NAT given by
Tabazadeh et al. (2002a) has severe deficiencies.
In the following a reanalysis of the experimentally obtained nucleation data of this work with
respect to a pseudo-heterogeneous nucleation mechanism is presented. The surface-based
homogeneous nucleation rate coefficients derived from these data will be compared to the
nucleation rate coefficients given by Tabazadeh et al. (2002a).If nucleation starts at the surface of a particle it must be assured that the molecular surface
layer is not contaminated by foreign (e. g. organic) molecules (Tabazadeh, 2003). The present
analysis is performed under the assumption that the droplet surface, i. e. the vapor-liquid
interface, is not contaminated by external species. This assumption is based on the following
arguments. A droplet of 10 ul in volume contains about 2-1014 molecular surface sites and
about 3-1021 molecules in the volume. Therefore, about 1014 molecules are needed to occupy
the surface layer. Even surface active species dissolve into the aqueous phase (Jungwirth, 2003).
Using mass spectroscopy Middlebrook et al. (1997) detect 0.02 wt% of organic contaminants
in 0.2 /mi particles in a laboratory environment. These impurities were assigned to organic
substances such as formaldehyde, ethylene, acetylene and butane, which have solubilities in
water of about 0.001-0.01 % (Howard and Meylan, 1997). Formaldehyde is very soluble in
water (Saxena and Hildemann, 1996) and, therefore, will be distributed in the volume of
the droplets rather than at their surface. From the solubilities and the molecule number
in a droplet of 10 ul in volume it can be concluded that at least 3-1017 organic molecules
can be dissolved in the volume. In equilibrium, the organic molecules will be distributed
equally throughout the droplet, hence, only a fraction of the organic molecules will be sitting
on the droplet surface. Assuming 0.1 ppbv of organics in the gas phase and a 7 of 0.1
(Middlebrook et al., 1997), the flux of organics impinging on the droplet (r = 0.13 cm)with subsequent dissolution within the liquid is about 1011 s-1. Therefore, it would take
more than 106 s to dissolve the 3-1017 organic molecules. Based on the above arguments, it
is assumed in the following section that the droplet surfaces of this study can be treated as clean.
The solid line in Fig. 5.9 represents Jhom in a solution with a HNO3 mole fraction of 0.333
(Tabazadeh et al., 2002a). The surface-based homogeneous nucleation rate coefficients derived
from experiments presented above (dotted lines) are upper limits (see appendix D.l), i. e.
the surface-based homogeneous nucleation rate coefficient is expected to be even lower. The
filled circles represent reanalyzed bulk sample experiments of Koop et al. (1997b), where the
vapor-liquid interface was taken as the surface area. In the investigated temperature range the
difference between the surface-based homogeneous nucleation rate coefficients derived from the
experiments of this work and the surface-based homogeneous nucleation rate coefficients derived
from the formulation of Tabazadeh et al. (2002a) is at least four orders of magnitude. There¬
fore, jf^AD obtained by using AGf;fAD of Tabazadeh et al. (2002a) strongly overestimates
5.7. Pseudo-heterogeneous nucleation of PSCs 103
(0
CM
"E£
3=0)O
O
to
rr
c
g's©
o3
170 180 190 200 210 220 230
Temperature [K]
Figure 5.9: The solid line shows the surface-based homogeneous nucleation rate coefficients of NAD in a
solution with a HNO3 mole fraction of 0.333 derived using the formulations of Tabazadeh et al. (2002a).
The vertical bars encompass the temperature range of the experimental data used by Tabazadeh et al.
(2002a) to derive the parameterization. The dotted line represents NAD nucleation data of large droplets
from a solution with a mole fraction of 0.333. The filled circles represent binary HNO3/H2O solutions
with a mole fraction of 0.333 taken from Koop et al. (1997b)
the surface-based homogeneous nucleation rate coefficient with respect to the surface-based
homogeneous nucleation rate coefficients derived from the experiments of this work.
The solid and dotted lines in Fig. 5.10 represent the surface-based homogeneous nucleation rate
coefficients of NAD in a solution with a HNO3 mole fraction of 0.246 (Tabazadeh et al., 2002a)and the upper limits of the surface-based homogeneous nucleation rate coefficients derived from
experiments of this study. Reanalyzed bulk experiments are also shown (Koop et al., 1997b),which are in agreement with the nucleation rate coefficients derived in this work. Above a
temperature of 195 K the modelled nucleation rates overestimate the experimentally obtained
nucleation rate coefficients by up to four orders of magnitude. In this temperature range the
nucleation formulation of Tabazadeh et al. (2002a) is not able to simulate the Jhom -values
derived from experiments of this work. Below 190 K the experimentally obtained nucleation
rate coefficients increase strongly with temperature and approaches the Jham -values derived
by using the formulations of Tabazadeh et al. (2002a). Since the data derived in experiments
are only upper limits of the homogeneous nucleation rate coefficient, the difference between
both data sets could be even larger.The solid and dotted line in Fig. 5.11 represent the surface-based homogeneous nucleation rate
coefficients for NAT in a solution with a HNO3 mole fraction of 0.246 given by Tabazadeh
et al. (2002a) and the Jhom -values obtained in experiments. The reanalyzed bulk experiments
104 CHAPTER 5. KINETIC PROCESSES
105
104
103
102
10
1
lO"1
lO"2
'U160 170 180 190200 210 220 230 24^°
Temperature [K]
Figure 5.10: The solid line shows surface-based homogeneous nucleation rate coefficients of NAD m a
solution with a HNO3 mole fraction of 0.246 derived using the formulations of Tabazadeh et al. (2002a).
The vertical bars envelop the temperature range of the experimental data used by Tabazadeh et al (2002a)
to derive the parameterization. The dotted line represents NAD nucleation data of large droplets from a
solution with a mole fraction of 0.246. The filled circles represent binary HNO3/H2O solutions with a
mole fraction of 0.246 taken from Koop et al. (1997b)
ci -\j Ann
(Koop et al., 1997b) are in agreement with the Jhom -values obtained in this work. Since the
9 NAT 9 NATdotted line represents upper limits of Jhom there is no contradiction between Jf^,m -values
C MAT
derived by using the formulations of Tabazadeh et al. (2002a) and Jhom -values derived from
the experiments of this study. However, as shown in Fig. 5.8 the surface-based nucleation model
suffers from an erroneous temperature and concentration dependence. Hence, the coincidence*7 NAT
of the Jfrem -values obtained by the parameterization of Tabazadeh et al. (2002a) and the
ç MAT1
^hom ~values derived from the experiments of this study at ~185 K is only by chance.
Figure 5.12a shows the composition and corresponding NAD and NAT saturation ratios of
STS droplets at 50 mbar (approx. 20 km altitude) for mixing ratios of 5 ppmv H2O, 10
ppbv HNO3, and 0.5 ppbv H2S04 (Carslaw et al., 1994). In Fig. 5.12b, NAD and NAT
surface-based nucleation rate coefficients axe shown for the conditions displayed in panel (a).These surface-based homogeneous nucleation rate coefficients are analyzed as explained in
section 5.4, but the volume of the zth droplet, K, in Eq. 5.4 was substituted by the surface
of the ith droplet, St. In the region of highest saturation ratios (shaded region in Fig. 5.12)arrows represent the range of the upper limits of surface-based nucleation rate coefficients
derived from volume-based nucleation rate coefficients of Fig. 5.7 using Eq. 2.34. The radii
of the investigated aqueous HNO3 droplets vary between ~3-10~5-0.248 cm. Hence, for
each upper limit of the volume-based homogeneous nucleation rate coefficient two values for
5.7. Pseudo-heterogeneous nucleation of PSCs 105
160 170 180 190 200 210 220 230
Temperature [K]
240
Figure 5.11: The solid line shows surface-based homogeneous nucleation rate coefficients of NAT in a
solution with a HNO3 mole fraction of 0.246 derived using the formulations of Tabazadeh et al. (2002a).
The horizontal bars encompass the temperature range of the experimental data used by Tabazadeh et al.
(2002a) to derive the parameterization. The dotted line represents NAT nucleation data of large droplets
from a solution with a mole fraction of 0.246. The filled circles represent binary HNO3/H2O solutions
with a mole fraction of 0.246 taken from Koop et al. (1997b)
the surface-based nucleation rate coefficient are derived corresponding to the minimum and
maximum radius (upper and lower end of the arrows in Fig. 5.12). Figure 5.12c shows the
production rates per cm3 air and hour derived from the surface-based nucleation rate coefficients
given in panel (b). The particle volume per cm3 air was taken from Carslaw et al. (1994). An
average aerosol number density of about 10 cm-1 was taken and the corresponding total surface
was calculated assuming monodisperse aerosol droplets.The maximum surface-based nucleation rate coefficients and production rates of NAD given
by Tabazadeh et al. (2002a) (blue solid lines in Fig. 5.12) are about 7 orders of magnitude
larger than the corresponding data derived from experiments of this work (highest values of
blue arrows in Fig. 5.12). This could be expected due to the large differences in the nucleation
rate coefficients shown in Fig. 5.9 and 5.10. The production rate of NAD derived in this
study is about 10~9 cm-3 h-1 and, therefore, too low to account for the observed particlenumber densities of 10~4 cm-3 of large nitric acid containing particles (Fahey et al., 2001).The maximum surface-based nucleation rate coefficients and production rates of NAT given
by Tabazadeh et al. (2002a) agree quite well since the modelled surface-based nucleation rate
coefficients and the experimentally derived upper limits of the nucleation rate coefficients do
not contradict each other as shown in Fig. 5.11. However, this agreement is only by accident,
because Fig. 5.8b indicates that the surface-based nucleation rate coefficients derived by the
106 CHAPTER 5. KINETIC PROCESSES
186 187 188 189 190 191 192 193 194 195 196
25
20
15
10
5
10
10'
103
105
107
102
lO"4
106
186 187 188 189 190 191 192 193 194 195 196
Temperature [K]
Figure 5.12: (a) The composition (green and orange lines) and the saturation ratios (red and blue
lines) of STS aerosols as a function of temperature at 50 mbar with 5 ppmv H2O, 10 ppbv HNO3, and
0.5 ppbv H2SO4 (Carslaw et al., 1994). The shaded region indicates the temperature range where the
SjsiAX-values have their maximum, (b): The both-way ending blue and red arrows represent upper limits
for the surface-based nucleation rate coefficients of NAD and NAT in STS droplets, respectively, derived
from experimental data for the conditions shown w panel (a). The range of the J^om-values is due to
the different aerosol radii of the various nucleation data sets. For comparison, solid lines indicate the
surface-based homogeneous nucleation rate coefficient in STS droplets for the same conditions calculated
using the formulation of Tabazadeh et al (2002a) (c): Hourly production rates of NAD and NAT
particles (blue arrows and red arrows, respectively) per err? of air derivedfrom the surface-based nucleation
rate coefficients shown in panel (b) The total aerosol surface as a function of temperature was taken
into account (Carslaw et aL, 1994). An average particle number density of 10 cm~3 was assumed. Also
shown as solid lines are the NAD and NAT production rates for the same conditions calculated using the
formulation of Tabazadeh et al. (2002a). Small arrows in (b) and (c) mark the temperature where the
saturation ratio of NAD (blue) and NAT (red) equals one.
t' r^
5.8. Homogeneous ice nucleation in (HH4J2SO4/JT2O droplets 107
formulation of Tabazadeh et al. (2002a) exhibit an unreasonable temperature and concentration
dependency. Here, a maximum NAT production rate of about 2-10-6 cm-3 h_1 is derived.
Microphysical sensitivity studies show that an hourly production rate below 10-5 cm-3 h_1
has no significant influence on the overall stratospheric particle number density of nitric acid
containing particles (Tabazadeh et al., 2001, 2002a; Mann et al., 2002). Hence, also the NAT
production rate is too low to account for the observed particle number density of large nitric
acid containing particles. Furthermore, it must also be considered that the production rates
derived here are only upper limits and, hence, the production rate in the atmosphere could be
much lower.
Conclusion
It has been shown that the new parameterization of surface-based AGa^ and AGn^ given
by Tabazadeh et al. (2002a) does not lead to a physically reasonable behavior of the surface-based
homogeneous nucleation rate coefficients. The paxameterizations also suggest that nucleation
of NAD occurs in highly concentrated HNO3 solutions which, however, is not corroborated by
experiments (Bertram et al., 2000a). Furthermore, the modelled J-values show unreasonable
temperature dependence, e. g. when approaching the melting points of the respective solid. The
parameterization of Tabazadeh et al. (2002a) for a pseudo-heterogeneous nucleation is not capa¬
ble to describe the data set obtained in this work and Bertram et al. (2000a) and the nucleation
data sets of Salcedo et al. (2001), Bertram and Sloan (1998b), and Bertram and Sloan (1998a).The surface-based production rates derived experimentally in this work from droplets under
stratospheric conditions are up to 7 orders of magnitude lower than the predicted ones. These
production rates are too low to explain the observed particle number density of large nitric acid
containing particles of about 10~4 cm-3 (Fahey et al., 2001) and, hence, the subsequent den¬
itrification of the polar vortex. Furthermore, from the experimentally obtained surface-based
homogeneous nucleation rate coefficients it cannot be concluded that a pseudo-heterogeneous
phase transition occurred at all in the present experiments.
Since the volume-based homogeneous nucleation mechanism and the proposed pseudo-
heterogeneous nucleation mechanism are not able to explain the particle number densities of
the observed large nitric acid containing particles, other formation mechanisms must exist, such
as heterogeneous NAT nucleation on ice particles (Waibel et al., 1999).
5.8 Homogeneous ice nucleation in (NH^SC^/t^O droplets
In this section the results of homogeneous ice nucleation experiments with aqueous (NH4)2S04
droplets are discussed. The experimentally obtained homogeneous ice nucleation rate coeffi¬
cients are compared to experimentally derived homogeneous ice nucleation rate coefficients of
Hung et al. (2002), JHung- Hung et al. (2002) and Hung and Martin (2001) tried to derive
a single formulation for the homogeneous ice nucleation rate coefficient using nucleation data
108 CHAPTER 5. KINETIC PROCESSES
sets obtained from several different techniques, including optical microscopy (OM) (Bertramet al., 2000b) (similar to the technique presented here), differential scanning calorimetry (DSC)
(Bertram et al., 2000b), continuous flow thermal diffusion chamber (CFD) (Chen et al., 2000),and several aerosol flow tube studies employing infrared spectroscopy for the detection of ice
nucleation (AFT-IR) (Prenni et al., 2001; Czizco and Abbatt, 1999; Chelf and Martin, 2001).After a reexamination of the results of Hung and Martin (2001) by Hung et al. (2002), the
authors conclude that Chelf and Martin (2001) studied an aerosol containing both crystalline
and aqueous particles. The crystalline phase of (NH4)2S04 is detected by a change in the in¬
frared spectrum due to the ferroelectric phase transition (see section 4.11). The existence of
crystalline particles shifts the spectroscopically derived composition of the aerosol to higher con¬
centrations, which has been corrected by Hung et al. (2002). Since the freezing temperatures of
the (NH4)2S04 aerosol derived by Chelf and Martin (2001) are in agreement with the freezing
temperatures measured by Czizco and Abbatt (1999), Hung et al. (2002) suggest that also the
aerosol of Czizco and Abbatt (1999) consists of a mixture of crystalline and aqueous particles
and, thus, must be corrected in composition. In the spectra of Prenni et al. (2001) a signal of
the ferroelectric phase transition is absent, thus, it can be assumed that the particles nucleated
homogeneously. Prenni et al. (2001) set the composition of the aerosol particles by passing
the aerosol through a conditioning flow tube, in which the particles were in equilibrium with
the ice-coated tube walls. The nucleation data of Prenni et al. (2001) corresponds closer to
the J-values obtained by OM, DSC, and CFD experiments. If the temperature of the aerosol
particles of Prenni et al. (2001) is determined using the features of the IR spectrum, similar
to the procedure described by Hung et al. (2002), the freezing temperatures come closer to the
freezing values of Hung et al. (2002). A more difficult task is the reconciliation of AFT-IR, OM,
DSC, and CFD measurements. Hung et al. (2002) are not able to find a common J-function
which describes the J-values obtained by the different experimental methods. Therefore, in this
study further experiments were performed employing the OM-technique to measure additional
J-values. The possibility of a pseudo-heterogeneous nucleation mechanism will be considered
in the analysis. Volume-based Jffun9-values of Hung et al. (2002), J#uns, and J^TO-values of
this study, and the corresponding surface-based JHung-values of Hung et al. (2002), JHung, and
J^^-values of this study will be compared to each other.
Figure 5.13 shows volume-based J^unp-values of Hung et al. (2002) and upper limits of J^omderived in this study. The experimentally derived freezing points and J^^-values of this work
axe in agreement with the freezing points and J^om-values presented in Bertram et al. (2000b).The upper limits derived in this work imply that the "true" J^^-values for a defined (NH4)2S04mole fraction are smaller than those of the corresponding dashed line. Figure 5.13 indicates that
an agreement between AFT-IR and OM-technique is obtained only for low (NH4)2S04 mole frac¬
tions (below 0.02). (NH4)2S04 solutions with higher mole fractions (above 0.02) disagree.
Since Hung et al. (2002) cannot explain the discrepancies between the different data sets, pseudo-
heterogeneous nucleation was taken into account in the present analysis as a possible nucleation
pathway (Tabazadeh et al., 2002a,b). In this reanalysis of the ice nucleation data of aqueous
(NH4)2S04 droplets derived from the experiments of this work it was assumed that the droplet
surface is not contaminated by external species (see discussion in section 5.7). The J^m-valuesderived by Hung et al. (2002) and the J^j-values obtained in this study are reanalyzed with
5.8. Homogeneous ice nucleation in (JVH4J2SO4/IÎ2O droplets 109
11
10
<o
E ' 0.070
1
F 8 "
10x:
-3 1
01
\
O 7
6
T I I I
0.11
205 210 215 220 225 230 235 240
Temperature [K]
Figure 5.13: The solid line indicate volume-based homogeneous nucleation rate coefficients by Hung
et al (2002). The (NH4)2SOi mole fractions which correspond to the J-values are plotted besides the
lines. The data for XHNO3 = 0 (1. e. pure water) are taken from the model of Tabazadeh et al. (2000). The
dashed lines represent upper limits of the volume-based homogeneous nucleation rate coefficients obtained
in this study.
respect to surface-induced nucleation using Eq. 2.34. A mean diameter of 300 nm and about
40 /im for the droplet radii of Hung et al. (2002) and of this work, respectively, were chosen.
Figure 5.14 shows surface-based homogeneous nucleation rate coefficients of Hung et al. (2002),
Jfiungi and ^fom"vames derived in this study for different (NH4)2S04 mole fractions. J-¬values obtained in this work again represent upper limits of the nucleation rate coefficient (see
Fig. 5.13). As in the volume-based case there is no contradiction between the values of JHung
and Jhsom-values of this study for low mole fractions (below a mole fraction of 0.02). In the
case of a mole fraction of 0.07 both data sets are in agreement for temperatures higher than
218 K. For lower temperatures there is still a discrepancy between the measured nucleation rate
coefficients.
The difference between the data sets could have their origin in the following possible reasons.
First, it appears that the freezing points of higher concentrated particles of Hung et al. (2002)were obtained at warmer temperatures when compared to the nucleation rate coefficients derived
in the OM experiments (Bertram et al., 2000b), DSC measurements (Bertram et al., 2000b),and the freezing points of this study. As Hung et al. (2002) state that using the spectroscopic
features of the AFT-IR spectra of Prenni et al. (2001) to obtain the freezing temperatures, i.
110 CHAPTER 5. KINETIC PROCESSES
205 210 215 220 225 230 235 240
Temperature [K]
Figure 5.14: The solid line indicate surface-based homogeneous nucleation rate coefficients by Hung
et al (2002). The (NH4)2SÛ4 mole fractions which correspond to the J-values are plotted besides the
lines. The data for xhno3 = 0 (i. e. pure water) are taken from the model of Tabazadeh et al. (2000). The
dashed lines represent upper limits of the surface-based homogeneous nucleation rate coefficients obtained
m this study.
e. the procedure given by Hung et al. (2002), the freezing temperatures of Prenni et al. (2001)
come closer to the data of Hung et al. (2002). Therefore, the difference in freezing temperatures
could lie in a temperature bias due to the temperature retrieval procedure given by Hung et al.
(2002).Second, an erroneous droplet composition could be the reason for the difference in the data sets.
The deconvolution of the aerosol composition from infrared-spectra is a difficult task. The com¬
position of the particles used in the OM-techniques axe determined by measuring their melting
points. The melting points of the particles can be easily detected visually with the microscope
and, therefore, are measured directly without complicated algorithms.
Third, Hung et al. (2002) have taken into account that the aerosol used in the AFT-IR ex¬
periments consist of solid and aqueous (NH4)2S04 particles. A small fraction of these solid
particles can nucleate ice heterogeneously. These few ice particles can deplete the surrounding
gas phase water partial pressure thereby growing into large particles with significant ice signal
in the infrared spectra. Hence, heterogeneous ice nucleation leads to larger J-values at higher
temperatures.
5.8. Homogeneous ice nucleation in (NHi^SOi/R^O droplets 111
Conclusion
The volume-based homogeneous ice nucleation rate coefficients derived in this work axe in agree¬
ment with a previous study of Bertram et al. (2000b) which also employed the OM-technique.
The differences between the data sets obtained using the OM-technique (this study, Bertram
et al. (2000b)) and the AFT-IR-technique ((Czizco and Abbatt, 1999; Chen et al., 2000; Prenni
et al., 2001; Chelf and Martin, 2001; Hung and Martin, 2001; Hung et al., 2002) was further
consolidated in the range of higher (NH4)2S04 mole fractions. It has also been shown that
pseudo-heterogeneous nucleation cannot explain the differences between the data sets of both
methods. There is no evidence that surface-induced nucleation has occurred at all in the ex¬
periments. Therefore, the causes for these discrepancies still remain speculative. The reasons
could be a false temperature and composition determination within the AFT-IR experiments.
Also, heterogeneous ice nucleation in the mixed (NH4)2S04 aerosol could have occurred. Further
experiments with the ability to cover a larger range of homogeneous nucleation rate coefficients
axe necessary to answer the remaining ambiguities.
Seite Leer /
Blank leaf
Chapter 6
Final remarks
6.1 Summary and conclusion
During the course of this thesis thermodynamic properties of and kinetic processes in aqueous
solutions were investigated. The main topics included:
• the dissociation of the bisulfate ion in H2SO4/H2O solutions,
• the nucleation of NAD and NAT in HN03/H20 and HNO3/H2SO4/H2O solutions,
• the nucleation of ice in (NH4)2S04/H20 solutions.
The experimental temperatures and concentrations were chosen such that they represent the
conditions experienced by aerosol droplets in the UT/LS. For these purposes an experimental
setup has been built to investigate aerosol properties optically and by Raman spectroscopy. A
temperature stage was constructed and adapted to a Confocal Raman Microscope. The stage
allows an accurate adjustment of the temperature of the investigated aerosol droplets from 180
to 330 K. Furthermore, a sample preparation procedure was developed for the production of
droplets in the 5 /xm-1.5 mm size range. Finally, Raman spectroscopy has been proved to be
a sensitive tool for the quantitative investigation of composition changes in aqueous droplets
with diameters of 50 /xm-1.5 mm.
Thermodynamic properties
The dissociation of the bisulfate ion (HSO4 ^ SO4- + H+) has been studied in aqueous H2SO4
solutions. For this purpose, Raman spectra of aqueous H2SO4 and (NH4)2S04 droplets with
concentration of 0.54-15.23 mol kg-1 and 0.99-5.35 mol kg-1, respectively, were recorded in a
113
114 CHAPTER 6. FINAL REMARKS
temperature range of 180-326 K. The experimentally derived degree of dissociation of HSOJincreases with decreasing temperature for all investigated aqueous H2SO4 solutions. This is in
contrast to the widely used atmospheric thermodynamic model AIM (Clegg et al., 1998), which
underestimates the degree of dissociation by up to a factor of 5. The experimentally obtained
low-temperature data of the dissociation of the bisulfate ion was implemented in a Pitzer ion
interaction model of the H2S04/H20-system. This Pitzer ion interaction model was used to
derive a new formulation of the thermodynamic dissociation constant for the bisulfate ion in
the temperature range of 180-473 K. The thermodynamic dissociation constant derived in this
work is shown to be thermodynamically consistent with the Nernst heat theorem, in contrast
to the dissociation constant used in AIM.
Relevant atmospheric properties of the aerosol particles such as water activity and trace gas
solubility are affected by the newly derived thermodynamic dissociation constant of the HSO4ion. For 1.13-15.23 mol kg-1 H2SO4/H2O solutions the Pitzer model presented in this study
predicts water activities up to 10 % lower than the corresponding values derived with the AIM
model (Clegg et al., 1998). Predicted activity coefficients differ by up to 2 orders of magnitudewhen compared to the ones obtained from the AIM model. Heterogeneous reaction rates in
H2SO4/H2O depend on the solubility of the involved trace gases. Predicted HCl solubilities
axe up to 3 orders of magnitude lower when compared to the solubilities derived with the AIM
model (Carslaw et al., 1995a). Lower HCl solubilities and, hence, decreased heterogeneousreaction rates result in lower chlorine activation.
Kinetic processes
PSC formation mechanisms were studied both theoretically and experimentally by analyzingnucleation parameterizations given in literature and by measuring upper limits of the homo¬
geneous nucleation rate coefficients of NAD and NAT in aqueous HNO3 and HNO3/H2SO4droplets. The main focus of the present study was to investigate possible formation processes
of large nitric acid containing particles observed in particle number densities of 10-4 cm-3 in
the lower stratosphere in the winter 1999/2000 (Fahey et al., 2001). These large particles lead
to a strong denitrification of the polar stratosphere, which inhibits the deactivation of ozone
destroying agents. Tabazadeh et al. (2001) suggested that homogeneous nucleation of NAD
and NAT in STS aerosols is sufficiently fast to obtain large nitric acid containing particles in
number densities of 10-4 cm-3. Knopf et al. (2002) have shown that the parameterization used
in Tabazadeh et al. (2001) yields unreasonable homogeneous nucleation rate coefficients under
stratospheric conditions. Therefore, this parameterization must not be used in microphysicalaerosol models which are applied to stratospheric conditions (Knopf et al., 2002). In contrast
to the parameterization, the experimentally derived homogeneous nucleation rate coefficients of
NAD and NAT are exceedingly low (< 2-10-5 cm_3s_1 and < 8-10-2 cm_3s_1, respectively) in
STS aerosols under polar stratospheric conditions. This is in agreement with earlier nucleation
studies of NAD and NAT in aqueous binary HNO3 and ternary HNO3/H2SO4 bulk solutions
(Koop et al., 1995, 1997b). The experimentally derived upper limits of NAD and NAT
homogeneous nucleation rate coefficients yield maximum hourly particle production rates of
6.1. Summary and conclusion 115
~3-10~10 cm_3(air)h-1 under stratospheric conditions. Thus, the production rates used in
the microphysical modelling study of stratospheric denitrification of Tabazadeh et al. (2001)axe 5 orders of magnitude too high. If maximum saturation ratios axe assumed to persist for
two months, the experimentally derived production rates yield particle number densities of
~5-10-7 cm-3, which are much smaller than the reported values by Fahey et al. (2001). The
experimental nucleation data of this PhD thesis clearly show that homogeneous nucleation
pathway cannot be responsible for the occurrence of large nitric acid containing particles in
particle number densities of 10-4 cm-3.
More recently, Tabazadeh et al. (2002a) and Djikaev et al. (2002) suggested a pseudo-
heterogeneous nucleation mechanism, (i. e. the nucleation is induced at the aerosol surface,)which may lead to the observed particle number densities of large nitric acid containing particlesof 10-4 cm-3. The pseudo-heterogeneous parameterization of Tabazadeh et al. (2002a) yieldsNAD production rates of about ~5-10-2 cm_3(air) h_1 in the polar stratosphere. This NAD
production rate is sufficient to obtain the observed particle number densities of 10~4 cm-3.
However, the present study has shown that the pseudo-heterogeneous parameterization is
at odds with classical nucleation theory and experimentally obtained data. Therefore, the
stratospheric pseudo-heterogeneous nucleation rate coefficients derived from the parameteri¬zation of Tabazadeh et al. (2002a) must not be used in microphysical modelling studies. The
experimentally obtained surface-based nucleation rates of NAD and NAT of this work are up to
7 orders of magnitude lower than the values derived by the parameterization of Tabazadeh et al.
(2002a). The experimentally derived hourly surface-based production rates of NAD and NAT
axe <10-9 cm_3(air) h_1 and <2-10-6 cm_3(air) h-1, respectively. Microphysical sensitivitystudies show that production rates below ~10-5 cm-3 (air) h_1 do not have a significantinfluence on the overall stratospheric particle number densities of nitric acid containing particles
(Tabazadeh et al., 2001, 2002a; Mann et al., 2002). Therefore, the experimentally derived
surface-based production rates exclude the pseudo-heterogeneous nucleation pathway as a
formation mechanism of the large nitric acid containing particles observed in particle number
densities of 10~4 cm-3 (Fahey et al., 2001).The theoretical and experimental analysis of the homogeneous and pseudo-heterogeneousnucleation mechanism indicate that neither nucleation mechanisms can explain the observation
of the particle number densities of the large nitric acid containing particles and the subsequentdenitrification of the polar vortex. Hence, other nucleation mechanisms must be responsible,for example heterogeneous nucleation of NAT on ice particles as suggested by Waibel et al.
(1999). Another recently proposed formation mechanism is the mother cloud/NAT-rockmechanism of Fueglistaler et al. (2002). This mechanism suggests that type la PSCs can serve
as mother clouds for the formation of the large nitric acid containing particles. Individual NAT
particles at the cloud base fall into lower stratospheric layers undepleted in gas phase HNO3
and, thus, rapidly accelerate due to a positive feedback between their growth and sedimentation.
Cirrus ice cloud formation was investigated by measuring upper limits of the homogeneous ice
nucleation rate coefficients in aqueous (NH4)2S04 droplets. At upper tropospheric conditions,for example at 215 K, the experimentally derived upper limits of the homogeneous ice nucleation
rate coefficients of this work are lower than ~106 cm-3 s_1 for an aqueous (NH4)2S04 droplet
116 CHAPTER 6. FINAL REMARKS
with a (NH4)2S04 mole fraction of 0.07. The nucleation data derived in this work are in
agreement with experimentally derived homogeneous ice nucleation rate coefficients of Bertram
et al. (2000b), but disagree with experimentally obtained data of AFT-IR studies (Czizco and
Abbatt, 1999; Chelf and Martin, 2001; Prenni et al., 2001; Hung and Martin, 2001) for higher
(NH4)2S04 mole fractions. At temperatures lower than 225 K and (NEÏ4)2S04 mole fractions
higher than 0.05 the microscope experiments yield significantly lower homogeneous ice nucleation
rate coefficients when compared to those obtained by the AFT-IR studies. The newly derived
upper limits of the homogeneous ice nucleation rate coefficients of this work does not lead to
a reconciliation of the different data sets. The difference must be due to the experimental and
theoretical retrieval procedures. The analysis of ice nucleation in aqueous (NH4)2S04 droplets
with respect to a pseudo-heterogeneous nucleation pathway does not reconcile the nucleation
data of the microscope and AFT-IR experiments either. The discrepancy at temperatures lower
than 225 K and (NH4)2SÛ4 mole fractions larger than 0.05 remains.
6.2 Outlook
The results of the investigations of thermodynamic and kinetic processes performed in this work
raise a number o open questions and topics which should be investigated in the future.
Thermodynamic processes
Pitzer modellingIn the context of thermodynamic processes in aqueous aerosol particles the thermodynamic dis¬
sociation constant derived in this work should be implemented in ternary NH3/H2SO4/H2O and
HNO3/H2SO4/H2O Pitzer ion interaction models. Predictions of the water activity and NH3
solubility in NH3/H2SO4/H2O solutions may experience significant changes due to the newlyderived binary interaction parameters of H2SO4/H2O. The changes in the solution properties
may be important for upper tropospheric aerosols with respect to homogeneous and heteroge¬
neous nucleation as a pathway for cirrus ice cloud formation and heterogeneous reactions. STS
aerosols consist of HNO3/H2SO4/H2O solutions. The influence of the newly derived thermody¬namic dissociation constant on HNO3, H2O, and HCl uptake should be investigated. This may
have further implications on nucleation of NAD and NAT in STS aerosols due to slight changesin the aerosol composition at stratospheric conditions.
It has been recognized that bromine may be responsible for 25 % of the ozone destruction
(WMO, 1998). Therefore, solubilities of HBr in H2S04/H20 and HNO3/H2SO4/H2O dropletsshould be calculated using the modified Pitzer ion interaction model. The predicted solubilities
should be compared with experimental data (Abbatt, 1995; Abbatt and Nowak, 1997; Williams
and Long, 1995; Kleffmann et al., 2000).
6.2. Outlook 117
Kinetic processes
Formation mechanisms of PSCs
The formation mechanism of observed large nitric acid containing particles with number
densities of 10-4 cm-3 is still unresolved. Recently, it was suggested that pseudo-heterogeneousnucleation might be a suitable nucleation pathway for the observed large nitric acid containing
particles. In a recent discussion (Tabazadeh, 2003) it was claimed that pseudo-heterogeneousnucleation cannot be investigated in laboratory environment due to contamination of the
employed aerosol particles. To resolve the influences of surface contamination of the particleson the measured nucleation rate coefficients, nucleation rate coefficients derived from purposelycontaminated particles should be compared to those from clean particles. Furthermore,nucleation rate coefficients of particles varying widely in volume and in surface area should be
measured to determine the dominant nucleation mechanism.
Heterogeneous nucleation mechanisms of NAD and NAT should be investigated experimentally.
Heterogeneous nuclei such as meteoritic material (Murphy et al., 1998) should be introduced
into binary and ternary HNO3/H2O and H2SO4/HNO3/H2O solutions to determine nucleation
rate coefficients (Biermann et al., 1996). Also, the gas-to-solid nucleation of NAD, NAT, and
ice (deposition mode) should be further investigated to derive corresponding heterogeneousnucleation rate coefficients. Raman spectroscopy could be used to identify the solid phasewhich forms on the nuclei. The measured heterogeneous nucleation rate coefficients togetherwith data on the prevalence of the investigated nuclei could be used to derive stratospheric
production rates that can be evaluated with respect to the observed particle number densities
of NAT in the stratosphere.
Pseudo-heterogeneous nucleation
If pseudo-heterogeneous nucleation occurs, the surface enrichment of solution species should be
studied as a function of droplet radius. Stuart and Berne (1999) investigated the propensity of
chlorine molecules as a function of droplet curvature. A similar molecular dynamics study should
be performed for typical particle compositions of UT/LS aerosol such as NH3/H2SO4/H2O and
HNO3/H2SO4/H2O. This could increase our understanding on which molecules may initiate or
inhibit the nucleation on the surface of atmospheric aerosol droplets.
Organics
Murphy et al. (1998) have shown that aerosol particles in the upper troposphere usuallycontain also a fraction of organic compounds. These organics may influence homogeneous and
heterogeneous nucleation mechanisms of solid phases. This should be investigated by nucleation
experiments employing aerosol droplets which are doped with organics.
Heterogeneous nucleation
Because there axe only few studies of heterogeneous ice nucleation in concentrated inorganic
118 CHAPTER 6. FINAL REMARKS
solutions (Zuberi et al., 2002) further systems such as aqueous (NH4)3H(S04)2, NH4HSO4,
and NH4NO3 solutions should be investigated. Since heterogeneous ice nucleation can occur
on solid nuclei immersed in liquid aerosols (immersion mode) or on solid particles from the gas
phase directly (deposition mode) both formation pathways should be studied in the experiments.
Homogeneous ice nucleation
The discrepancies of the homogeneous ice nucleation rate coefficients in (NH4)2S04/H20solutions derived by optical microscopy experiments and AFT-IR experiments shows the
importance to build a technique which is capable of measuring a wide range of homoge¬
neous nucleation rate coefficients. Such a setup would be able to reconcile the different
data sets available in the literature (Bertram et al., 2000b; Chen et al., 2000; Prenni et al.,
2001; Czizco and Abbatt, 1999; Chelf and Martin, 2001; Hung and Martin, 2001, and this work).
Appendix A
Experimental
A.l Electrical circuit for the operation of the inkjet-cartridge
Single droplets are generated by an inkjet-cartridge (Hewlett Packard, Model HP 51604). A
sketch of the structure and functionality is given in Fig. A.l. The functionality of the inkjet-
cartridge is based on a thermo-electrical principle. A resistor wire heats strongly the liquid
for a short moment. A gas bubble forms and pushes a definite amount of liquid through the
nozzle. The inkjet-cartridge is operated by a positive square pulse signal of about 23 V and
A nozzle
F"
j surfaceInk-
" ' '
1channel
u1circuit path
1 silicium-
resistorsubstrafe
Bdroplet
gas bubble H
Figure A.l: Sketch of the working principle of the inkjet-cartridge. Panel A shows the internal structure
of the inkjet-cartridge. Panel B shows the inkjet-cartridge in operation. This figure was adapted from
Düwel (2003).
119
120 APPENDIX A. EXPERIMENTAL
0-30 V
0-3 A
0-10 V
APM
PF
4I Osn.
PF
Figure A.2: Sketch of the electric circuit used for the operation of the inkjet-cartridge. The first part of
the circuit is the source, followed by the amplification and the signal inversion.
6 /xs in duration. Figure A.2 shows the electric circuit needed to drive the inkjet-cartridge. A
pulse generator is used for the production of a square pulse of 10 V height and 6 ßs duration.
This signal has to be amplified to 23 V. Since the load resistor of the inkjet-cartridge is about
65 Cl a power supply with a minimum of 25 V and 0.4 A is necessary. A homemade power
supply yielding up to 30 V and up to 3 A is used. The amplification circuit consist of a n-
channel Power Metal-Oxide-Semiconductor Field-Effect-Transistor(BUZll) and a resistor, R,
of 100 Q. Then follows the inverter, which transforms the incoming negative pulse signal to
a positive pulse signal. The inversion is done using a p-channel (MTP12P10) and a n-channel
(MTP12N10) Power Field-Effect-Transistor. Parallel to the inkjet-cartridge an oscilloscope is
applied to monitor the resulting pulse signal. The pulse generator can be operated in single
pulse mode, hence, generating single droplets. The operation of the inkjet-cartridge with pulse
frequencies of up to 100 Hz is also possible producing higher droplet numbers.
Appendix B
Raman spectroscopy
B.l Assignments of the normal vibrations of the investigatedRaman spectra
The number and the symmetry species (also called irreducible representations) of the normal
vibrations depend on the point group of the molecule or the molecule fragment. A detailed dis¬
cussion about point groups and their significance for the selection rules of the normal vibrations
can be found in Engelke (1985) and Schrader (1995). All assignments in the following tables
are taken from the work of Querry et al. (1974) and Cox et al. (1981). The assignments of the
vibration modes can be derived unambiguously by polarization measurements of IR and Raman
spectra. The following abbreviations for the characterization of the normal vibrations were used:
v. stretching vibration
Ö: bending vibration
Indices:
s: symmetricas: antisymmetricd: degenerate
The SO4- ion and the NH4 ion have a tetrahedral structure and belong to the point group T^
with the symmetry species IA1+IE+2F2. All normal vibrations are Raman-active, whereas
the vibrations of the symmetry species F2 are IR-active.
The HSOJ ion has a pyramidal structure and belongs to the point group C3„. This yields the
symmetry species 3Ai+3E. All normal vibrations are Raman-active and IR-active.
The H2O molecule belongs to the point group C2^. This yields 2A1+B2 symmetry species for
a molecule consisting of three atoms. All normal vibrations are Raman and IR-active. In the
121
S024- ion NHJ ion
v [cm-1] v [cm-1]
Vs 980 3033
Od 451 1685
Vd 1105 3134
Öd 613 1397
122 APPENDIX B. RAMAN SPECTROSCOPY
Vibration, symmetry species, mode, and maximum position of the SO4 ion and NH4 ion.
normal vibration symmetry species vibration mode maximum position maximum position
v\ Ai
V2 E
vz F2
Vj Fj2
Vibration, symmetry species, mode, and maximum position of the HSOJ ion.
normal vibration symmetry species vibration mode maximum position
v [cm-1]
vi Ai vs 1047
v2 E bas 417
1/3 Ai v 1341
Vz E Vas 1230
v4 Ai os 885
v4 E ôs 593
Vibration, symmetry species, mode, and maximum position of the H2O molecule.
normal vibration symmetry species vibration mode maximum position
v [cm-1]
1/1 Ai us 3219
v2 Ai b~s 1640
vz Bi v_as3445
liquid phase appears an additional intermolecular vibration, vr,, at 590 cm-1 due to hydrogen
bonding (Querry et al., 1974).
The H30+ ion belongs to the point group C3,,. This yields 3Ai+3E symmetry species for a
molecule consisting of four atoms. All normal vibrations are Raman-active and IR-active.
The H2SO4 molecule has a tetrahedral structure and belongs to the point group 02^. This yields
the symmetry species 4Ai-|-A2+2Bi-r-2B2. All normal vibrations are Raman and IR-active.
B.l. Assignments of the normal vibrations of the investigated Raman spectra 123
Vibration, symmetry species, mode, and maximum position of the HsO+ molecule.
normal vibration symmetry species vibration mode maximum position
V [cm-1]
vi Ai v3 2650-3380
V2 Ai Ss 1134
vz E vd 2650-3380
V4 E Sd 1670
Vibration, symmetry species, and maximum position of the H2SO4 molecule.
normal vibration symmetry species maximum position
v [cm-1]
V! Ai 905
1*2 Ai 381
v2 A2 417
1/3 Ai 1140
v3 Bi 1190
1/3 B2 1370
V4 Ai 741
1/4 Bi 564
v4 B2 965
Seite Leer /Blank leaf
Appendix C
Heterogeneous chemistry
The uptake of a gas into a liquid followed by reaction involves a number of physical as well as
chemical processes (Finlayson-Pitts and Pitts, 2000):1. The transport of the gas to the surface, which is determined by the gas-phase diffusion
coefficient, Dg.2. The uptake at the interface, controlled by the mass accommodation coefficient, a.
3. The diffusion into the bulk, which depends on the liquid diffusion coefficient, Di.
4. The chemical reaction in the bulk.
Assuming a fast gas transport, high solubility, and fast reaction, the following diffuso-reactive
uptake coefficient can be obtained (Hanson and Ravishankara, 1993; Finlayson-Pitts and Pitts,
2000):
- = - + ?-—, (CI)7 a 4JRTJJ*v/ÂF
where R is the universal gas constant, T is absolute temperature, v is the mean thermal velocity
of the gas-phase molecules at T, H* is the effective Henry's law constant, and A;1 = /^[X]^ is
the pseudo-first-order loss rate coefficient for the reaction of [Y] with [X] in the liquid. Thus,
for a particular value of a, Eq. C.l indicates that the reaction in the bulk of the aerosol scales
linearly with H*.
125
Seite Leer /
Blank leaf
Appendix D
Nucleation rate coefficients and
production rates
D.l Derivation of upper nucleation limits
This section is based mainly on the work of Koop et al. (1997b). Since nucleation is a stochas¬
tic process, similar to radioactive decay, one can give an expectation value for the number of
nucleation events which should occur within a given sample number and particular observation
time. Due to the stochastic nature of nucleation the successful formation of a critical nucleus
does not depend on the previous trials. Also, different nucleation processes are independent of
each other. An experiment must be repeated several times, or, alternatively, a large number of
equal samples must be observed simultaneously to obtain statistical information (Koop et al.,
1997b).For a large number of molecules, m, in a sample or droplet, the probability, P, to observe k
nucleation events can be expressed by a Poisson distribution:
Pfc(f) = ^~e_mP' (D-1}
where p is the probability that a molecule becomes the center of a critical cluster. For small p
the nucleation rate for the whole sample can be introduced asw = mp/t within the observation
time, t. If we assume ntot equal samples and nuq is the number of samples which have not
nucleated within a specific time, t, the probability of observing k nucleation events within t is
Pk{t) = M!e-- ^jom i ln
ntot
k\"
ntotk\ V" lnhq(t)\.'
The total observation time in an experiment consists of the time, for which each droplet stays
liquid, £hg,i, and the time after which individual droplets nucleate, tnuCtt.
nhq nnuc
Hot = / jHXq% + y
jtnuCil , \D.oJ
i=0 »=0
127
128 APPENDIX D. NUCLEATION RATE COEFFICIENTS AND PRODUCTION RATES
where nnuc is the number of droplets, that nucleate after times tnuCti. It is found that nnuc =
Li)ttat. If we assume higher or lower values of oj (i. e. the uncertainty in w) than the measured
w-value, the probability of the occurrence of nnuc nucleation events is never zero. Therefore, an
upper fiducial limit, uup, can be given such that less than n\fuc nucleation events occur with a
given probability x (x is also called "confidence level"), if ujup was the true nucleation rate:
OO nnuc i . \k
x= £ Pfe(W = l-e-^-£^fp-. (D.4)
==fl,nuc+l fc—U
Even if no nucleation events occurred during the experiments, an upper limit, nn%c, of possible
nucleation events due to the statistical uncertainty in u can be given:
<L = wupttot = In (j~^j • (D-5)
Equation D.5 yields for zero observed nucleation events, nnXuc = 0, an upper limit of the number
of nucleation events of nffuc = 6.908 for a confidence level of 0.999. In other words, if the
experiment would be repeated an infinite number of times the maximum number of observable
nucleation events are smaller than n^„c within a probability of 0.999. For ntfuc > 0, Eq. D.4
must be solved numerically. Values for n\fuc for given nnXuc are presented in Table D.I.
Using these upper limits for the number of nucleation events in the analysis of the experimental
data leads to upper limits of the nucleation rates. These upper limits of the nucleation rate can be
understood as the most conservative values for the determined nucleation rates, i. e. the highest
estimation of a particular nucleation rate. The "true" nucleation rate is expected to be even
lower, since the upper limit of the nucleation rate includes also possible heterogeneous nucleation
processes, which have a lower nucleation barrier and, hence, nucleate at higher temperatures.
The upper limit of J can be used to derive lower limits of AG1^ using Eq. 2.29. But these
AGatf'-values cannot be used to derive quantitative values of the Gibbs free energy of activation
for diffusion or the surface tensions using Eq. 2.26 and 2.27, since the AG^-values do not
represent the "true" activation values of the solution.
D.2 Derivation of stratospheric production rates of NAD and
NAT
Here, a more detailed description of the analysis of the data presented in Fig. 5.7 of Knopf et al.
(2002) is given:The experimentally derived data analyzed as described in section 5.4 and all available different
nucleation data sets (Koop et al., 1995,1997a; Bertram and Sloan, 1998b,a; Bertram et al., 2000a;
Salcedo et al., 2001) plotted in Fig. 5.5 are used within this evaluation. For each temperature
all available data on AGact were gathered and the highest value for each saturation ratio was
used to derive AGact as a function of the NAD or NAT saturation ratio. The stratospheric
saturation ratio corresponding to the temperature was taken according to Fig. 5.7a. This yields
the lowest value of Jj^. This procedure was repeated for all plotted temperatures for which
D.2. Derivation of stratospheric production rates of NAD and NAT 129
Table D.l: Upper and lower fiducial limits for selected numbers of nucleation events, nnUC, at a confi¬
dence Level of x = 0.999 as calculated using Poisson statistics (Koop et al, 1997b).
UlowHot nnuc ^upHot
nda 0 6.908
0.001 1 9.233
0.045 2 11.229
0.191 3 13.062
0.429 4 14.794
0.739 5 16.455
1.107 6 18.062
1.520 7 19.626
1.971 8 21.156
2.452 9 22.657
2.961 10 24.134
5.794 15 31.244
8.958 20 38.042
12.337 25 44.636
15.869 30 51.083
19.518 35 57.418
23.260 40 63.662
27.078 45 69.833
30.959 50 75.942
38.878 60 88.007
46.963 70 99.909
55.180 80 111.682
63.506 90 123.348
71.921 100 134.924
159.130 200 247.675
433.739 500 573.028
a, not defined.
several nucleation data sets were available. The corresponding production rates shown in Fig.
5.7c were obtained by multiplying the aerosol volume density given by Carslaw et al. (1994) and
the time of one hour with the homogeneous nucleation rate coefficients shown in Fig. 5.7b.
Seite Leer /
Blank leaf
Appendix E
Parametrizations of NAD and NAT
nucleation mechanisms
E.l Homogeneous nucleation parametrization ofNAD and NAT
Here, the parameterizations used by Tabazadeh et al. (2001) are given to obtain Fig. 5.2 of
Knopf et al. (2002). The laboratory nucleation rates of Bertram and Sloan (1998b), Bertram
and Sloan (1998a), and Salcedo et al. (2001) are parameterized in the below given way by
Tabazadeh et al. (2001). The nucleation activation energies are given by (Salcedo et al., 2001),but were extrapolated to stratospheric saturation ratios by Tabazadeh et al. (2001). In the case
of NAD the following formulations were used:
AG£Ad((Snad) = (28.8 ±0.2)-(0.37 ±0.01)5nad. (E.l)
The homogeneous nucleation rate coefficients can be derived by
Jnat = 1-138 -nrVTrSexp-AGNAD
RT(E.2)
where r is the radius of the aerosol particle, which is assumed to be about 0.82 /mi.
In the case of NAT nucleation Tabazadeh et al. (2001) gives the following expressions including
the nucleation activation energy derived by Salcedo et al. (2001):
AG^T(SNAT) = (30.9 ±0.3)-(0.14 ±0.0004)5nat, (E.3)
which is used to derive homogeneous NAT nucleation rate coefficients by
Jnat = 9.269 • 103VTr3exp-AG^ad
RT(E.4)
131
132APPEJVDJX E. PARAMETRIZATIONS OF NAD AND NAT NUCLEATION MECHANISMS
E.2 Pseudo-heterogeneous nucleation activation energies of
NAD and NAT
The following parametrizations are given by Tabazadeh et al. (2002a) to obtain the pseudo-
heterogeneous nucleation rate coefficients of NAD and NAT shown in Fig. 5.8. The activation
energy for the nucleation of NAD on the surface of a binary HNO3/H2O solution is given by:
^GÎaAD(xnN03,T) = 11.5593 + 0.0804214T
- (71.5133 - 0.256724 T) XHNO3, (E.5)
where xhno<< is the mole fraction of the solution and T is the temperature in Kelvin. SinceC WAT
AGact was onry obtained for a mole fraction value of 0.246, the concentration dependence of
AGact (XHN03)T) is used to obtain a concentration dependence of AG^ (xhno3>T):
AGact (XHN03,T) =
AGg,NAD(0 246 r)AGact (zHN03,T), (E.6)
where AG^NAD(0.246, T) is given by
AGf£AT(0.246, T) = -45.2429 + 0.364844 T. (E.7)
These nucleation activation energies can be used to calculate surface-based homogeneous nucle¬
ation rate coefficients of NAD and NAT by using Eq. 2.32.
List of Figures
1.1 Aerosol formation mechanisms and aerosol size distribution 2
1.2 Sketch of the standard atmosphere 4
1.3 Composition of aerosol particles of the lower stratosphere 5
1.4 Volume density of PSC particles as function of temperature 6
1.5 Radiative forcing of atmospheric agents (IPCC) 8
2.1 Phase diagram of H2S04/H20 17
2.2 Phase diagram of (NH4)2S04/H20 18
2.3 Phase diagram of HN03/H20 19
2.4 Gibbs free energy for the formation of a critical cluster 22
2.5 Surface tensions of a liquid sitting on its solid 23
2.6 Nucleation of a solid at the surface of its liquid 24
2.7 Term diagram of the Raman scattering process 25
2.8 Intensity distribution within a Raman spectrum 26
3.1 Sketch of the atomizer 30
3.2 Sketch of the single droplet generator using an inkjet-cartridge 30
3.3 Sketch of the experimental setup 33
3.4 Sketch of the temperature stage 34
4.1 Sketch of the experimental setup 41
4.2 Laser influence on Raman spectra 42
133
134 List of Figures
4.3 Raman spectra of (NH4)2S04/H20 droplets 0.99 and 5.35 mol kg1 in concentration 44
4.4 Phase diagram of H2S04/H20 45
4.5 Temperature dependent Raman spectra of a H2SO4/H2O droplet 2.55 mol kg-1in concentration 46
4.6 Ratios of mS02-/mHS0- in H2SO4/H2O solutions 47
4.7 Degree of dissociation of the HSOJ ion 48
4.8 The thermodynamic dissociation constant of the HSOJ ion 50
4.9 Gibbs free energy, reaction enthalpy, and reaction entropy of the dissociation of
HSOJ 51
4.10 Activity coefficients and water activity of a 1.13 mol kg-1 H2SO4/H2O solution .54
4.11 Activity coefficients and water activity of a 9.84 mol kg-1 H2SO4/H2O solution .55
4.12 HCl solubilities of H2SO4 solutions 5.5, 8.35, 10.2, and 15.32 mol kg-1 in concen¬
tration 68
4.13 Temperature dependent Raman spectra of a H2SO4/H2O droplet 6.79 mol kg-1in concentration 70
4.14 Temperature dependent Raman spectra of a H2SO4/H2O droplet 3.04 mol kg-1in concentration 71
4.15 Concentration dependent Raman spectra of H2SO4/H2O droplets 72
4.16 Lorentzian fit functions 73
4.17 Ratio R as a function of temperature 74
4.18 Ratio Rw as a function of concentration 75
4.19 Temperature dependent Raman spectra of a (NH4)2S04/H20 droplet 5.35 mol
kg-1 in concentration 76
4.20 Concentration dependent (NH4)2S04/H20 Raman spectra 77
4.21 Heat capacity of the ferroelectric phase transition 79
4.22 Effect of the ferroelectric phase transition on the ^i(S04~) normal vibration ... 80
4.23 Effect of the ferroelectric phase transition on the 1^3 (SO^-) normal vibrations . .81
5.1 AGact of NAD and NAT as a function of saturation ratio 89
List of Figures 135
5.2 Homogeneous nucleation rate coefficients of NAD and NAT using the formulations
of Tabazadeh et al. (2001) 90
5.3 Sketch of the experimental setup 91
5.4 Raman spectra of HNO3/H2O droplets 92
5.5 AGact-values of this work as a function of the NAD and NAT saturation ratios .95
5.6 AGact-values of this work as a function of the NAD and NAT saturation ratios .96
5.7 Composition, homogeneous nucleation rate coefficients, and production rates of
STS aerosols 97
5.8 Pseudo-heterogeneous NAD and NAT nucleation rate coefficients as a function of
temperature and concentration 101
5.9 Surface-based homogeneous nucleation rate coefficients of NAD in a solution with
a HNO3 mole fraction of 0.333 103
5.10 Surface-based homogeneous nucleation rate coefficients of NAD in a solution with
a HNO3 mole fraction of 0.246 104
5.11 Surface-based homogeneous nucleation rate coefficients of NAT in a solution with
a HNO3 mole fraction of 0.246 105
5.12 Composition, surface-based homogeneous nucleation rate coefficients, and corre¬
sponding production rates of STS aerosols 106
5.13 Homogeneous nucleation rate coefficients in (NH4)2S04 droplets 109
5.14 Surface-based homogeneous nucleation rate coefficients in (NH4)2S04 droplets . .110
A.l Sketch of inkjet-cartridge 119
A.2 Electric circuit to operate inkjet-cartridge 120
Seite Leer /
Blank leaf
Bibliography
Abbatt, J. P. D., 1995. Interactions of HBr, HCl, and HOCl with supercooled sulfuric acid
solutions of stratospheric compositions. J. Geophys. Res. 100 (7), 14009-140017.
Abbatt, J. P. D., Beyer, K. D., Fucaloro, A. F., McMahon, J. R., Woolridge, P. J., Zhang, R.,
Molina, M. J., 1992. Interaction of HCl Vapor with Water-ice: Implications for the Strato¬
sphere. J. Geophys. Res. 97, 15819-15826.
Abbatt, J. P. D., Nowak, J. B., 1997. Heterogeneous Interactions of HBr and HOCl with Cold
Sulfuric Acid Solutions: Implications for Arctic Boundary Layer Bromine Chemistry. J. Phys.
Chem. A 101, 2131-2137.
Albrecht, B., 1989. Aerosols, cloud microphysics and fractional cloudiness. Science 245, 1227-
1230.
Anthony, S. E., Onash, T. B., Tisdale, R. T., Disselkamp, R. S., Tolbert, M. A., Wilson, J. C,
May 1997. Laboratory studies of ternary HNO3/H2SO4/H2O particles: Implications for polar
stratospheric cloud formation. J. Geophys. Res. 102 (D9), 10777-10784.
Atkins, P. W., 1994. Physikalische Chemie, 2nd Edition. VCH, Weinheim.
Bajpai, P. K., Jain, Y. S., 1987. The phase transition in ammonium sulphate: dynamics of
deuterated NH+ ions and the heat of transition. J. Phys. C: Solid State Phys. 20, 387-393.
Becker, G., Müller, R., McKenna, D. S., Rex, M., Carslaw, K. S., December 1998. Ozone loss
rates in the Arctic stratosphere in the winter 1991/92: Model calculations compared with
Match results. Geophys. Res. Lett. 25 (23), 4325-4328.
Berry, R. S., Stuart, A. R., Ross, J., 2000. Physical Chemistry (Topics in Physical Chemistry),2nd Edition. John Wiley & Sons, New York.
Bertram, A. K., Dickens, D. B., Sloan, J. J., April 2000a. Supercooling of type 1 polar strato¬
spheric clouds: The freezing of submicron nitirc acid aerosols having HNO3 mole fractions
less than 0.5. J. Geophys. Res. 105 (D7), 9283-9290.
Bertram, A. K., Koop, T., Molina, L. T., J., M. M., 2000b. Ice Formation in (NH4)2S04-H20Particles. J. Phys. Chem. A 104, 584-588.
137
138 BIBLIOGRAPHY
Bertram, A. K., Sloan, J. J., June 1998a. The nucleation rate constants and freezing mechanism
of nitric acid trihydrate aerosol under stratospheric conditions. J. Geophys. Res. 103 (Dil),
13261-13265.
Bertram, A. K., Sloan, J. J., January 1998b. Temperature-dependent nucleation rate constants
and freezing behavior of submicron nitirc acid dihydrate aerosol particles under stratospheric
conditions. J. Geophys. Res. 103 (D3), 3553-3561.
Beyer, K. D., Hansen, A. R., October 2002. Phase Diagram of the Nitric Acid/Water System:
Implications for Polar Stratospheric Clouds. J. Phys. Chem. A 106 (43), 10275-10284.
Biermann, U. M., Presper, T., Koop, T., Mösinger, J., Crutzen, P. J., Peter, T., 1996. The
unsuitability of meteoritic and other nulcei for polar stratospheric cloud freezing. Geophys.
Res. Lett. 23 (13), 1693-1696.
Bolsaitis, P., Elliott, J. F., 1990. Thermodynamic Activities and Equilibrium Partial Pressures
for Aqueous Sulfuric Acid Solutions. J. Chem. Eng. Data 35, 69-85.
Borrmann, S., Solomon, S., Avallone, L., Toohey, D., Baumgardner, D., 1997a. On the orrurrence
of CIO in cirrus clouds and volcanic aerosol in the troposphere region. Geophys. Res. Lett.
24 (16), 2011-2015.
Borrmann, S., Solomon, S., Dye, J. E., Luo, B. P., 1997b. The potential of cirrus clouds for
heterogeneous chlorine activation. Geophys. Res. Lett. 23 (16), 2133-2137.
Bregman, B., Wang, P. H., Lelieveld, J., February 2002. Chemical ozone loss in the tropopause
region on subvisible ice clouds, calculated with a chemistry-transport model. J. Geophys. Res.
107 (D3), ACH 5-1-5-12.
Carslaw, K. S., Clegg, S. L., Brimblecombe, P., 1995a. A Thermodynamic Model of the System
HCI-HNO3-H2SO4-H2O, Including Solubilities of HBr, from > 200 K to 328 K. J. Phys. Chem.
99, 11557-11574, http://www.hpcl.uea.ac.uk/ e770/aim.html.
Carslaw, K. S., Luo, B., Clegg, S. L., Peter, T., Brimblecombe, P., Crutzen, P. J., November
1994. Stratospheric aerosol growth and HNO3 gas phase depletion from coupled HNO3 and
water uptake by liquid particles. Geophys. Res. Lett. 21 (23), 2479-2482.
Carslaw, K. S., Luo, B. P., Peter, T., 1995b. An analytic expression for the composition of
aqueous HNO3-H2SO4 stratospheric aerosols including gas phase removal of HNO3. Geophys.
Res. Lett. 22 (14), 1877-1880.
Carslaw, K. S., Peter, T., Clegg, S. L., 1997. Modeling the composition of liquid stratospheric
aerosols. Rev. Geophys. 35, 125-154.
Chelf, J. H., Martin, S. T., January 2001. Homogeneous ice nucleation in aqueous ammonium
sulfate aerosol particles. J. Geophys. Res. 106 (1), 1215-1226.
BIBLIOGRAPHY 139
Chen, H., Irish, D. E., 1971. A Raman Spectral Study of Bisulfate-Sulfate Systems. II. Consti¬
tution, Equilibria, and Ultrafast Proton Transfer in Sulfuric Acid. J. Phys. Chem. 75 (17),
2672-2681.
Chen, Y., DeMott, P. J., Kreidenweis, S. M., Rogers, D. C, Sherman, D. E., November 2000. Ice
formation by sulfate and sulfuric acid aerosol particles under upper-tropospheric conditions.
J. Atmos. Sei. 57 (22), 3752-3766.
Clegg, S. L., Brimblecombe, P., 1995. Application of a Mutlicomponent Thermodynamic Model
to Activities and Thermal Properties of 0-40 mol kg-1 Aqueous Sulfuric Acid from < 200 to
328 K. J. Chem. Eng. Data 40, 43-64.
Clegg, S. L., Brimblecombe, P., Wexler, A. S., 1998. A thermodynamic model of the system
H+-NH|-S042--N0J-H20 at tropospheric temperatures. J. Phys. Chem. A 102, 2137-2154,
http://www.hpcl.uea.ac.uk/ e770/aim.html.
Clegg, S. L., Rard, J. A., Pitzer, K. S., 1994. Thermodynamic Properties of 0-6 mol kg-1
Aqueous Sulfuric Acid from 273.15 to 328.15 K. J. Chem. Soc, Faraday Trans. 90 (13), 1875-
1894.
Colberg, C, 2002. Experimente an levitierten H2SO4/NH3/H2O-Aerosolteilchen: Atmosphrische
Relevanz von Letovizit. Swiss Federal Institute of Technology, ETH, Zürich.
Cox, R. A., Haldna, U. L., Idler, K. L., Yates, K., 1981. Resolution of Raman spectra of aqueous
sulfuric acid mixtures using principal factor analysis. Can. J. Chem. 59, 2591-2598.
Czizco, D. J., Abbatt, J. P. D., June 1999. Deliquescence, efflorescence, and supercooling of
ammonium sulfate aerosols at low temperature: Implications for cirrus cloud formation and
aerosol phase in the atmosphere. J. Geophys. Res. 104 (11), 13781-13790.
Das, A., Dev, S., Shangpliang, H., Nonglait, K., Ismail, K., 1997. Electrical Conductance and
Viscosity of Concentrated H2SO4/H2O Binary Systems at Low Temperatures: Correlation
with Phase Transitions. J. Phys. Chem. B 101, 4166-4170.
Dawson, B. S. W., Irish, D. E., Toogood, G. E., 1986. Vibrational Spectral Studies of Solutions
at Elevated Temperatures and Pressures. A Raman Spectral Study of Ammonium Hydrogen
Sulfate Solutions and the HSOj-SO|~. J. Phys. Chem. 90, 334-341.
Debye, P., Hückel, E., 1923a. Zur Theorie der Elektrolyte. Phys. Z. 24 (I), 185.
Debye, P., Hückel, E., 1923b. Zur Theorie der Elektrolyte. Phys. Z. 24 (II), 305.
Defay, R., Prigogine, I., 1966. Surface Tension and Adsorption. Longmans, London.
Dickson, A. G., Weolowski, D. J., Palmer, D. A., Mesmer, R. E., 1990. Dissociation Constant of
Bisulfate Ion in Aqueous Sodium Chloride Solutions to 250 °C. J. Phys. Chem. 94, 7978-7985.
Djikaev, Y. S., Tabazadeh, A., Hamill, P., Reiss, H., 2002. Thermodynamic Conditions for
the Surface-Stimulated Crystallization of Atmospheric Droplets. J. Phys. Chem. A 106 (43),
10247-10253.
140 BIBLIOGRAPHY
Donovan, J. R., Salamone, J. M., 1983. Sulfuric Acid and Sulfur Trioxide. In: Grayson, M.
(Ed.), Sulfonation and Sulfation to Thorium and Thorium Compounds, 3rd Edition. Vol. 22
of Encyclopedia of Chemical Technology. Wiley & Sons, New York, pp. 190-232.
Düwel, I., 2003. Beobachtung der laserinduzierten Einzeltropfenverdampfung mit Hilfe der
laserinduzierten Fluoreszenz. Master's thesis, Ruprecht-Karls-University of Heidelberg.
Dye, J. E., Baumgardner, D., Gandrud, B. W., Kawa, S. R., Kelly, K. K., Loewenstein, M.,
Ferry, G. V., Chan, K. R., Gary, B. L., May 1992. Particle Size Distributions in Arctic Polar
Stratospheric Clouds, Growth and Freezing of Sulfuric Acid Droplets, and Implications for
Cloud Formation. J. Geophys. Res. 97 (8), 8015-8034.
Elrod, M. J., Koch, R. E., Kim, J. E., Molina, M. J., 1995. HCl Vapour Pressures and Reac¬
tion Probabihties for C10N02 + HCl on Liquid H2SO4-HNO3-HCI-H2O Solutions. Faraday
Discuss. 100, 269-278.
Engelke, F., 1985. Aufbau der Moleküle. Teubner-Studienbücher, Stuttgart.
Engineering, D., 2001. Bob-module. Turner, Oregon 97392, www.decadenet.com.
Fahey, D. W., Gao, R. S., Carslaw, K. S., Kettleborough, J., Popp, P. J., Northway, M. J., Hole-
cek, J. C, Ciciora, S. C, McLaughlin, R. J., Thompson, T. L., Winkler, R. H., Baumgardner,
D. G., Gandrud, B., Wennberg, P. O., Dhaniyala, S., McKinney, K., Peter, T., Salawitch,
R. J., Bui, T. P., Elkins, J. W., Webster, C. R., Atlas, E. L., Jost, H., Wilson, J. C, Herman,
R. L., Kleinböhl, A., von König, M., February 2001. The Detection of Large HN03-Containing
Particles in the Winter Arctic Stratosphere. Science 291, 1026-1030.
Finlayson-Pitts, B. J., Pitts, J. N. J., 2000. Chemistry of the Upper and Lower Atmosphere.
Academic Press, San Diego, California.
Fowler, L. D., Randall, D. A., 1996. Liquid and ice cloud microphysics in the csu general
circulation model, part iii: Sensitivity to model assumption. J. Climate 9.
Fueglistaler, S., Luo, B. P., Voigt, C, Carslaw, K. S., Peter, T., 2002. NAT-rock formation by
mother clouds: a microphysical model study. Atmos. Chem. Phys. 2, 93-98.
Gable, C. M., Betz, H. F., Maron, S. H., 1950. Phase Equilibria of the System Sulfur Trioxide-
Water. J. Am. Chem. Soc. 72 (4), 1445-1448.
Giauque, W. F., Hornung, E. W., Kunzler, J. E., Rubin, T. R., 1960. The Thermodynamic
Properties of Aqueous Sulfuric Acid Solutions and Hydrates from 15 to 300°K. J. Am. Chem.
Soc. 82 (1), 62-70.
Gierens, K., Schumann, U., Helten, M., Smit, H., Marenco, A., 1999. A distribution law for
relative humidity in the upper troposphere and lower stratosphere derived from three years
of MOZAIC measurements. Ann. Geophysicae 17, 1218-1226.
Hamill, P., Toon, O. B., Dec 1991. Polar Stratospheric Clouds and the Ozone Hole. Phys. Today
44 (12), 34-42.
BIBLIOGRAPHY 141
Hanson, D., Mauersberger, K., August 1988a. Laboratory studies of the nitric acid trihydrate:
Implications for the south polar stratosphere. Geophys. Res. Lett. 15 (8), 855-858.
Hanson, D., Mauersberger, K., October 1988b. Vapor-Pressures of HNO3/H2O Solutions at
Low-Temperatures. J. Phys. Chem. 92 (21), 6167-6170.
Hanson, D. R., 1998. Reaction of CIONO3 with H20 and HCl in Sulfuric Acid and HNO3/
H2S04/H20 Mixtures. J. Phys. Chem. A 102, 4794-4807.
Hanson, D. R., Ravishankara, A. R., 1991. The Reaction Probalities of CIONO2 and N2O5 on
40% and 75% Sulfuric Acid Solutions. J. Geophys. Res. 96, 17307-17314.
Hanson, D. R., Ravishankara, A. R., 1992. Investigation of the Reactive and Nonreactive Pro¬
cesses Involvbing CIONO2 and HCl on Water and Nitirc Acid Doped Ice. J. Phys. Chem. 96,
2682-2691.
Hanson, D. R., Ravishankara, A. R., 1993. Uptake of HCl and HOCl onto Sulfuric Acid: Solu¬
bilities, Diffusivities, Reaction. J. Phys. Chem. 97 (47), 12309-12319.
Harned, H. S., Hamer, W. J., January 1935. The Thermodynamics of Aqueous Sulfuric Acid
Solutions from Electromotive Force Measurements. J. Am. Chem. Soc. 57, 27-35.
Hayes, A. C, Kruus, P., Adams, W. A., 1984. Raman Spectroscopic Study of Aqueous
(NH4)2S04 and ZnS04 Solutions. J. Sol. Chem. 13 (1), 61-75.
Heintzenberg, J., 1989. Fine particles in the global troposphere-a review. Tellus 41B, 149-160.
Helsel, R., 1995. Graphical Programming: A Tutorial for HP VEE. Prentice Hall PTR, New
Jersey.
Heymsfield, A. J., Miloshevich, L. M., Twohy, C, Sachse, G., Oltmans, S., 1998. Upper tro¬
pospheric relative humidity observations and implications for cirrus ice nucleation. Geophys.
Res. Lett. 25 (9), 1343-1350.
Higashigaki, Y., Chihara, H., 1981. Phase-Transition in Ammonium-D4 Sulfate as Studied by
Heat-Capacity Measurements between 3 K and 300 K. Bull. Chem. Soc. Japan 54 (3), 901-908.
Hoshino, S., Vedam, K., Okaya, Y., Pepinsky, R., October 1958. Dielectric and Thermal Study
of (nh4)2so4 and (NH4)2BeF4 Transitions. Phys. Rev. 112 (2), 405-412.
Houghton, J., Ding, Y., Griggs, D. J., Noguer, M., van der Linden, P. J., Dai, X., Maskell, K.,
Johnson, C. A., 2001. Climate Change 2001: The Scientific Basis. Contribution of Working
Group I to the Third Assessment Report of the Intergovernmental Panel on Climate Change.
Cambridge University Press, United Kingdom and New York.
Howard, P. H., Meylan, W. M., 1997. Handbook of Physical Properties of Organic Chemicals.
CRC Press, New York.
Hung, H.-M., Malinowski, A., Martin, S. T., 2002. Ice Nucleation Kinetics of Aerosols Containing
Aqueous and Solid Ammonium Sulfate Particles. J. Phys. Chem. A 106, 293-306.
142 BIBLIOGRAPHY
Hung, H.-M., Martin, S. T., September 2001. Apparent freezing temperatures modeled for several
experimental apparatus. J. Geophys. Res. 106 (17), 20379-20394.
Iqbal, Z., Christoe, C. W., 1976a. Raman Scattering Study of Ferroelectric Phase Transition in
Ammonium sulfate. Solid State Commun. 18, 269-273.
Iqbal, Z., Christoe, C. W., 1976b. Raman Scattering Study of Phase Transition in Ammonium
sulfate. Ferroelectrics 12, 177-179.
Jaenicke, R., 1993. Tropospheric aerosols. In: Hobbs, P. V. (Ed.), Aerosol-Cloud-Climate Inter¬
actions. Hobbs. Academic Press, San Diego, pp. 1-31.
Jain, Y. S., Bajpai, P. K., Bhattacharjee, R., D., C, 1986. Phase transition and temperature
dependence of the molecular distortion of ions in ammonium sulphate. J. Phys. C: Sohd State
Phys. (19), 3789-3796.
Jain, Y. S., Bist, H. D., 1974. Comments on the paper 'The origin of the ferroelectric phase
transition in ammonium sulphate'. Solid. State. Commun. 15, 1229-1230.
Jain, Y. S., Bist, H. D., Upreti, G. C, October 1973. The Ferroelctric Phase Transition in
Ammonium Sulphate. Chem. Phys. Lett. 22 (3), 572-575.
Jensen, E. J., Toon, O. B., Westphal, D. L., Kinne, S., Heymsfield, A. J., 1994a. Microphysical
modeling of cirrus, 1, Comparison with 1986 FIRE IFO measurements. J. Geophys. Res.
99 (5), 10421-10442.
Jensen, E. J., Toon, O. B., Westphal, D. L., Kinne, S., Heymsfield, A. J., 1994b. Microphysical
modeling of cirrus, 2, Sensitivity studies. J. Geophys. Res. 99 (5), 104443-1045.
Ji, K., Petit, J. C, June 1993. Calorimetric Identification of a New Nitric-Acid Hydrate able to
Play a Role in the Heterogeneous Chemistry of the Stratosphere. C. R. Acad. Sei. II 316 (12),1743-1748.
Junge, C. E., 1961. Vertical Profiles of Condensation Nuclei in the Stratosphere. J. Meteorol.
18, 501-509.
Junge, C. E., Manson, J. E., 1961. Stratospheric Aerosol Studies. J. Geophys. Res. 66 (7),2163-2182.
Jungwirth, P., 2003. Interactive comment on "Commentary on "Homogeneous nucleation of
NAD and NAT in liquid stratospheric aerosols: insufficient to explain denitrification" by
Knopf et al." by A. Tabazadeh. Atmos. Chem. Phys. Discuss. 3, S103-S104.
Kamenz, T., May 1999. Das Tröpfchenfeldexperiment zur Simulation der Gasaufnahme
stratosphärischer Aerosolpartikel: Oberflächenakustische und ramanspektroskopische Unter¬
suchungen. Ph.D. thesis, Ruprecht-Karls-University, Heidelberg.
Kiefer, W., Haarer, D., Spiess, H. W., 1995. Spektroskopie amorpher und kristalliner Festkrper.
Spiess, H. W., Darmstadt.
BIBLIOGRAPHY 143
Kleffmann, J., Becker, K. H., Bröske, R., Rothe, D., Wiesen, P., 2000. Solubility of HBr in
H2S04/H20 and HNO3/H2SO4/H2O Solutions. J. Phys. Chem. A 104, 8489-8495.
Knopf, D. A., Koop, T., Luo, B. P., Weers, U. G., Peter, T., 2002. Homogeneous nucleation of
NAD and NAT in liquid stratospheric aerosols: insufficient to explain denitrification. Atmos.
Chem. Phys. 2, 207-214.
Knopf, D. A., Luo, B. P., Krieger, U. K, Koop, T., Peter, T., 2003. Experimental and Theo¬
retical Analysis with Respect to Surface-Induced Nucleation Madrid, EAC 2003 Conference,
accepted.
Knopf, D. A., Zink, P., Schreiner, J., Mauersberger, K., 2001. Calibration of an Aerosol Compo¬
sition Mass Spectrometer with Sulfuric Acid Water Aerosol. Aerosol Sei. Technol. 35, 924-928.
Koop, T., Biermann, U. M., Luo, B., Crutzen, P. J., Peter, T., April 1995. Do stratospheric
aerosol droplets freeze above the ice frost point? Geophys. Res. Lett. 22 (8), 917-920.
Koop, T., Carslaw, K. S., Peter, T., 1997a. Thermodynamic stability and phase transition of
PSC particles. Geophys. Res. Lett. 24 (17), 2199-2202.
Koop, T., Luo, B. P., Biermann, U. M., Crutzen, P. J., Peter, T., 1997b. Freezing of
HNO3/H2SO4/H2O Solutions at Stratospheric Temperatures: Nucleation Statistics and Ex¬
periments. J. Phys. Chem. A 101, 1117-1133.
Koop, T., Luo, B. P., Tsias, A., Peter, T., August 2000. Water activity as the determinant for
homogeneous ice nucleation in aqueous solutions. Nature 406, 611-614.
Landsberg, G., Mandelstam, L., 1928. Eine neue Erscheinung bei der Lichtzerstreuung in Krys-
tallen. Die Naturwissenschaften 16, 557-558.
Lide, D. R. (Ed.), 1998. CRC Handbook of Chemistry and Physics, 79th Edition. CRC Press,
New York.
Lohmann, U., Feichter, J., 1997. Impact of sulphate aerosols on albedo and lifetime of clouds:
A sensitivity study with the echam4 gem. J. Geophys. Res. 102, 13685-13700.
Luo, B. P., Carslaw, K. S., Peter, T., Clegg, S. L., 1995. Vapor pressures of
H2S04/HN03/HCl/HBr/H20 solutions to low stratospheric temperatures. Geophys. Res.
Lett. 22, 247-250.
Luo, B. P., Clegg, S. L., Peter, T., Müller, R., Crutzen, P. J., January 1994. HCl solubility and
liquid diffusion in aqueous sulfuric acid under stratospheric conditions. Geophys. Res. Lett.
21 (1), 49-52.
MacKenzie, A. R., 1997. Solid-Liquid) Kelvin Equation and the Theory of Interfacial Tension
Components Commensurate? J. Phys. Chem. B 101, 1817-1823.
Mann, G. W., Davies, S., Carslaw, K. S., Chipperfield, M. P., Kettleborough, J., 2002. Polar vor¬
tex concentricity as a controlling factor in Arctic denitrification. J. Geophys. Res. 107 (D22),13-1-13-11.
144 BIBLIOGRAPHY
Marshall, W. L., Jones, E. V., 1966. Second Dissociation Constant of Sulfuric Acid from 25 to
350° Evaluated from Solubilities of Calcium Sulfate in Sulfuric Acid Solutions. J. Phys. Chem.
70 (12), 4028-4040.
Martin, S. T., 2000. Phase Transitions of Aqueous Atmospheric Particles. Chemical Reviews
100, 3403-3453.
Massucci, M., Clegg, S. L., Brimblecombe, P., 1996. Equilibrium Vapor Pressure of H2O above
Aqueous H2SO4 at Low Temperature. J. Chem. Eng. Data 41, 765-778.
Massucci, M., Clegg, S. L., Brimblecombe, P., 1999. Equilibrium partial pressures,
thermodynamic properties of aqueous and solid phases, and CI2 production from
aqueous HCl and HNO3 and their mixtures. J. Phys. Chem. A 103A, 4209-4226,
http://www.hpcl.uea.ac.uk/ e770/aim.html.
Meilinger, S. K., Koop, T., Luo, B. P., Huthwelker, T., Carslaw, K. S., Krieger, U. K., Crutzen,
P. J., Peter, T., November 1995. Size-dependent stratospheric droplet composition in lee wave
temperature fluctuations and their potential role in PSC freezing. Geophys. Res. Lett. 22 (22),
3031-3034.
Middlebrook, A. M., Iraci, L. T., McNeill, L. S., Koehler, B. G., Wilson, M. A., Saasted, O. W.,
Tolbert, M. A., Hanson, D. R., 1993. Fourier transfrom infrared studies of thin H2SO4/H2Ofilms: Formation, water uptake, and solid-liquid phase changes. J. Geophys. Res. 98, 20473-
20481.
Middlebrook, A. M., Thomson, David, S., Murphy, D. M., September 1997. On the Purity of
Laboratory-Generated Sulfuric Acid Droplets and Ambient Particles Studied by Laser Mass
Spectrometry. Aerosol Sei. Technol. 27 (3), 293-307.
Miller, S. R., Brenman, M., Waugh, J. S., Blinc, R., 1962. Nuclear Spin-Lattice Relaxation in
some Ferroelectric Ammonium Salts. Phys. Rev. 126 (2), 528-.
Murphy, D. M., Thomson, D. S., Mahoney, T. M. J., November 1998. In situ measurements of
organics, meteoritic material, mercury, and other elements in aerosols at 5 to 19 kilometers.
Science 282, 1664-1669.
Nernst, W., 1906. Über die Berechnung chemischer Gleichgewichte aus thermischen Messungen.
Nachr. Ges. Wiss. Göttingen, Math.-Phys. Kl. (1), 1-40.
Note 1, .m® is assumed to be equal to the total H2SO4 molality, although this assumption
is not required to derive aHSO- •
Note 2, .In Clegg et al. (1998) K\\(T) was transformed from molality to mole fraction scale.
O'Reilly, D. E., Tsang, T., 1967a. Deuteron Magnetic Resonance and Proton Relaxation Times
in Ferroelectric Ammonium Sulfate. J. Chem. Phys. 46 (4), 1291-1300.
O'Reilly, D. E., Tsang, T., February 1967b. First-Order Ferroelectric Transition in (NH4)2S04.J. Chem. Phys. 46 (4), 1301-1304.
BIBLIOGRAPHY 145
Peter, T., 1997. Microphysics and Heterogeneous Chemistry of Polar Stratospheric Clouds.
Annu. Rev. Phys. Chem. 48, 785822.
Petzelt, J., Grigas, J., Mayerova, I., 1974. Fax Infrared Properties of the Pseudoproper Ferro¬
electric Ammonium Sulfate. Ferroelectrics 6, 225-234.
Pincus, R., Baker, M., November 1994. Effect of precipitation on the albedo susceptibility of
clouds in the marine boundary-layer. Nature 372, 250-252.
Pitzer, K. S., 1991. Activity Coefficients in Electrolyte Solutions, 2nd Edition. CRC Press: Boca
Raton, FL.
Placzek, G., 1934. Rayleigh-Streueung und Raman-Streuung. Vol. VI of II. E. Marx, Leipzig, p.
205.
Prenni, A. J., Wise, M. E., Brooks, S. D., Tolbert, M. A., 2001. Ice nucleation in sulfuric acid
and ammonium sulfate particles. J. Geophys. Res. 106 (3), 3037-3044.
Pruppacher, Hans, R., Klett, J. D., 1997. Microphysics of Clouds and Precipitation. Vol. 18.
Kluwer Academic Publishers, Netherland.
Querry, M. R., Waring, R. C, Holland, W. E., Earls, L. M., Herrman, M. D., Nijm, W. P., Hale,
G. M., January 1974. Optical constants in the infrared for K2SO4, NH4H2PO4, and H2SO4
in water. J. Opt. Soc. Am. 64 (1), 39-46.
Raman, C. V., Krishnan, K. S., 1928. A new type of secondary radiation. Nature 121, 501.
Rard, J. A., Habenschuss, A., Spedding, F. H., 1976. A Review of the Osmotic Coefficients of
Aqueous H2S04 at 25 °C. J. Chem. Eng. Data 21 (3), 374-379.
Ratcliffe, C. I., Irish, D. E., 1982. Vibrational spectral Studies of Solutions at Elevated Temper¬
atures and Pressures. 5. Raman Studies of Liquid Water up to 300 °C. J. Phys. Chem. 86,
4897-4905.
Ravishankara, A. R., Sheperd, T. G., 1999. Scientific Assessment of Ozone Depletion: 1998.
World Meteorological Organization, Geneva, Switzerland, Ch. 7, pp. 7.1-7.59.
Robinson, G. N., Worsnop, D. R., Jayne, J. T., Kolb, C. E., Swartz, E., Davidovits, P., October
1998. Heterogeneous uptake of HCl by sulfuric acid solutions. J. Geophys. Res. 103 (19),
25371-25381.
Salcedo, D., Molina, T., Molina, M. J., 2001. Homogeneous Freezing of Concentrated Aqueous
Nitric Acid Solutions at Polar Stratospheric Temperatures. J. Phys. Chem. A 105, 1433-1439.
Sawada, A., Takagi, Y. andlshibashi, Y., 1975. Ferroelectric Phase-Transition in (NH4)2S04-K2S04 Mixed Crystals. J. Phys. Soc. Japan 38 (5), 1408-1414.
Saxena, P., Hildemann, L. M., 1996. Water-Soluble Organics in Atmospheric Particles: A Crit¬
ical Review of the Literature and Application of Thermodynamics to Identify Candidate
Compounds. J. Atmos. Chem. 24, 57-109.
146 BIBLIOGRAPHY
Schlemper, E. O., Hamilton, W. C, 1966. Neutron-Diffraction Study of Structures of Ferroelec¬
tric and Paraelectric Ammonium Sulfate. J. Chem. Phys. 44 (12), 4498-4509.
Schoeberl, M. R., Hartmann, D. L., 1991. The Dynamics of the Stratospheric Polar Vortex and
its Realtion to Springtime Ozone Depletions. Science 251, 46-52.
Schrader, B., 1995. Infrared and Raman Spectroscopy. VCH, Weinheim.
Schreiner, J., Voigt, C, Kohlmann, A., Arnold, F., Mauersberger, K., Larson, N., 1999. Chemical
Analysis of Polar Stratospheric Cloud Particles. Science 283, 968-970.
Seinfeld, John, H., Pandis, Spyros, N., 1998. Atmospheric Chemistry and Physics. John Wiley
& Sons, New York.
Senior, C. A., Mitchell, J. L. B., 1993. Carbon dioxide and climate: The impact of cloud
parametrization. J. Climate 6, 393-418.
Shomate, C. H., 1945. Specific Heats at Low Temperatures of (NH4)2S04, NH4A1(S04)2 and
NH4A1(S04)2 • 12 H20. J. Am. Chem. Soc. 67, 1096-198.
Smekal, A., 1923. Zur Quantentheorie der Dispersion. Die Naturwissenschaften 11, 873-875.
Solomon, S., Garcia, R. R., Rowland, F. S., Wuebbles, D. J., June 1986. On the depletion of
Antarctic ozone. Nature 321, 755-758.
Stanford, J. L., Davis, J. S., 1974. A century of stratospheric cloud reports: 1870-1972. Bull.
Am. Meteor. Bull 55, 213-219.
Staples, B. R., 1981. Activity and Osmotic Coefficients of Aqueous Sulfuric Acid at 298.15 K.
J. Phys. Chem. Ref. Data 10 (3), 779-798.
Stuart, S. J., Berne, B. J., 1999. Surface Curvature Effects in the Aqueous Ionic Solvation of
the Chloride Ion. J. Phys. Chem. A 103, 10300-10307.
Swartz, E., Shi, Q., Davidovits, P., Jayne, J. T., Worsnop, D. R., Kolb, C. E., 1999. Uptake
of Gas-Phase Ammonia. 2. Uptake by Sulfuric Acid Surfaces. J. Phys. Chem. A 103 (44),
8824-8833.
Tabazadeh, A., 2003. Commentary on "Homogeneous nucleation of NAD and NAT in liquid
stratospheric aerosols: insufficient to explain denitrification" by Knopf et al. Atmos. Chem.
Phys. Discuss. 3, 827-833.
Tabazadeh, A., Djikaev, Y. S., Hamill, P., Reiss, H., 2002a. Laboratory Evidence for Surface
Nucleation of Solid Polar Stratospheric Cloud Particles. J. Phys. Chem. A 106 (43), 10238-
10246.
Tabazadeh, A., Djikaev, Y. S., Reiss, H., December 2002b. Surface crystallization of supercooled
water in clouds. Proc. Nat. Acad. Sei. 99 (25), 1587315878.
BIBLIOGRAPHY 147
Tabazadeh, A., Jensen, E. J., Toon, O. B., Drdla, K., Schoeberl, M. R., 2001. Role of the
Stratospheric Polar Freezing Belt in Denitrification. Science 291, 2591-2594.
Tabazadeh, A., Martin, S. T., Lin, J. S., April 2000. The effect of particle size and nitric acid
uptake on the homogeneous freezing of aqueous sulfuric acid particles. Geophys. Res. Lett.
27 (8), 1111-1114.
Tabazadeh, A., Turco, R. P., Jacobson, M. Z., 1994. A model for studying the composition and
chemical effects of stratospheric aerosols. J. Geophys. Res. 99, 12897-12914.
Tolbert, M. A., Toon, O. B., 2001. Solving the PSC Mystery. Science 292, 61-63.
Tomikawa, K., Kanno, H., 1998. Raman Study of Sulfuric Acid at Low Temperatures. J. Phys.
Chem. A 102, 6082-6088.
Torrie, B. H., Lin, C. C, Binbrek, O. S., Anderson, A., 1972. Raman and Infrared Studies of
the Ferroelectric Transition in Ammonium Sulphate. J. Phys. Chem. Solids 33, 697-709.
Tsias, A., Prenni, A. J., Carslaw, K. S., Onash, T. P., Luo, B. P., Tolbert, M. A., Peter, T.,
1997. Freezing of polar stratospheric clouds in orographically induced strong warming events.
Geophys. Res. Lett. 24 (18), 2303-2306.
Turco, R. P., Toon, O. B., Hamill, P., 1989. Heterogeneous physicochemistry of the polar ozone
hole. J. Geophys. Res. 94, 16493-16510.
Turnbull, D., Fisher, J. C, 1949. Rate of Nucleation in Condensed Systems. J. Chem. Phys.
17 (1), 71-73.
Twomey, S., 1974. Pollution and planetary albedo. Atmos. Environ. 8 (12), 1251-1256.
Unruh, H.-G., Krüger, J., Sailer, E., 1978. Phase Transition Mechanisms Revealed by Optical
Spectroscopy. Ferroelectrics 20, 3-10.
Voigt, C, Schreiner, J., Kohlmann, A., Zink, P., Mauersberger, K., Larsen, N., Deshler, T.,
Kroger, C, Rosen, J., Adriani, A., Cairo, F., di Donfrancesco, G., Viterbini, M., Ovarlez, J.,
Ovarlez, H., David, C, Dörnback, A., 2000. Nitric Acid Trihydrate (NAT) in Polar Strato¬
spheric Clouds. Science 290, 1756-1758.
von der Gathen, P., Rex, M., Harris, N. R. P., Lucie, D., Knudsen, B. M., Braathen, G. O.,
Debacker, H., Fabian, R., Fast, H., Gil, M., Kyro, E., Mikkelsen, I. S., Rummukainen, M.,
Stahelin, J., Varotsos, C, May 1995. Observational evidence for chemical ozone depletion over
the arctic in winter 1991-92. Nature 375, 131-134.
Waibel, A. E., Peter, T., Carslaw, K. S., Oelhaf, H., Wetzel, G., Crutzen, P. J., Pöschl, U.,
Tsias, A., Reimer, E., Fischer, H., 1999. Arctic Ozone Loss Due to Denitrification. Science
283, 2064-2069.
Whitby, K. T., Sverdrup, G. M., 1980. California aerosols: Their Physical and Chemical Char¬
acteristics. Adv. Environ. Sei. Technol. 8, 477-525.
148 BIBLIOGRAPHY
Williams, L. R., Golden, D. M., October 1993. Solubility of HCl in Sulfuric Acid at Stratospheric
Temperatures. Geophys. Res. Lett. 20 (20), 2227-2230.
Williams, L. R., Long, F. S., 1995. Viscosity of Supercooled Sulfuric Acid Solutions. J. Phys.
Chem. 99, 3748-3751.
WMO, 1992. World Meteorological Organization, International Meteorological Vocabulary, 2nd
Edition. Geneva.
WMO, 1998. Scientific Assessment of Ozone Depletion: 1998. World Meteorological Organiza¬
tion, Geneva, Switzerland.
Worsnop, D. R., Fox, L. E., S., Z. M., Wofsy, S. C, 1993. Vapor Pressures of Solid Hydrates of
Nitric Acid: Implications for Polar Stratospheric Clouds. Science 259, 71-74.
Young, T. F., Maranville, L. F., Smith, H. M., 1959. Raman Spectral Investigations of Ionic
Equilibria in Solutions of Strong Electrolytes. In: Hamer, W. J. (Ed.), The Structure of
Electrolyte Solutions. Wiley, New York, pp. 35-63.
Zeleznik, F. J., 1991. Thermodynamic Properties of the Aqueous Sulfuric Acid System to 350
K. J. Phys. Chem. Data 20 (6), 1157-1200.
Zhang, R., Wooldridge, P. J., Molina, M. J., 1993a. Vapor Pressure Measurements for the
H2SO4/HNO3/H2O and H2SO4/HCI/H2O Systems: Incorporation of Stratospheric Acids into
Background Sulfate Aerosols. J. Phys. Chem. 97, 8541-8548.
Zhang, R., Woolridge, P. J., Abbatt, J. D. P., Molina, M. J., 1993b. Physical Chemistry of
the H2SO4/H2O Binary System at Low Temperatures: Stratospheric Implications. J. Phys.
Chem. 97, 7351-7358.
Zhelyaskov, V., Georgiev, G., Nickolov, Z., 1988. Temperature tudy of Intra- and Inter-molecular
Coupling and Fermi Resonance Constants in the Raman Spectra of Liquid Water Using
Fourier Deconvolution. J. Raman Spectrosc. 19, 405-412.
Zuberi, B., Bertram, A., Cassa, C. A., Molina, L. T., Molina, M. J., 2002. Heterogeneous
nucleation of ice in (NH4)2S04-H20 particles with mineral dust immersions. Geophys. Res.
Lett. 29 (10), 142-1-142-4.
Acknowledgements
The development of a Ph.D. student and his doctoral thesis is not possible without the great
support of his colleagues, friends, and family. Here I want to thank all people who participated
directly and indirectly in this Ph.D. doctoral thesis.
I thank
• Thomas Peter to have offered to me the possibihty to join his group at the ETH Zürich.
I appreciated your comments and advices. You established a wonderful mood within your
group where it is a pleasure to work.
• Thomas Koop, alias "Köpi", (alias "ice cream killer") for a wonderful scientific and amica¬
ble relationship. I had fun working with you and I learned very much from you. Certified:
one of the best tutors you can get!
• Ulrich Schurath being the co-examiner of my Ph.D. thesis and coming to Zürich to hold
the PhD defense.
• Beiping Luo, who is able to treat a fish and chicken in a wok as fast as a fortran code.
Beiping, thanks for your cooperation, unlimited ideas, and your humor.
• Uwe G. Weers, formerly known as the python, now alias "Notfallaufnahme" in M 8.1, for
technically support, friendship, and much inspiration.
• Dominik W. Brunner, alias "Chefpathologe" of M 8.1, for not becoming crazy and leaving
our office, good spirit, Badminton-colleague, and dynamical input (PV will be always with
you).
• Uh K. Krieger, specialist for Grisons nut pie, for all the good experimental ideas and
feedbacks.
• Bernhard Zobrist, alias "Chefarzt" of M 8.1, for good vibes, experimental support, and
friendship.
• Christina Colberg, for showing a physicist how Raman spectra can be interpreted, friend¬
ship, good time, and Badminton-colleague (well thought, poorly played!).
149
• Christian Braun, alias "the Hacker", for digital connections within the experiment, soft¬
ware support, friendship, and some beers.
• Peter Isler, for technical support (which was often necessary in a few minutes), for good
techno vibrations, and snowboard support.
• Claudia Marcolli, for showing a physicist what organics all can do.
• Michaela Hegglin, alias "Micca", for good vibes, yogurt, and friendship.
• Marc Wüest, the Plüsch- and Schümlipflümli-specialist, for fastest and most patient IT-
support, friendship, and humor.
• Ruedi Lüthi, for the technical, computer hardware, and infrastructure support.
• Hans Hirter, master of IT-support, for connection between the computers I used.
• Edwin Hausammann, for the technical support. If you have been in our group already two
years ago, I would have had to write two more chapters in my PhD thesis!
• Thomas Huthwelker, alias "Huthi, the man who had the shortest stay in the States" for
always being open for discussions, much experimental input, and ideas.
• Bruno Nussberger, the man with the smooth hands, for developing all the glass apparatus.
• Daniel Lüthi, for Sun/email support.
• Raphael Schefold, for good vibes, and for support in institutional politics.
• Petra Forney, for taking over some of the administration.
• Eva Choffat, for administrative advice and speaking consequently Swiss German.
• all other Ph.D. students for an enjoyable mood.
• all skating and snowboarding people I know (especially Chris Eggers) giving me the right
balance for a successful Ph.D. thesis.
• all the bar people at the skiing region Jochpass, Engelberg, Switzerland, for enjoyable
weekends (Schümlipflümli).
• Ursula, Jose, Barbara, and Michi Sogo for making me feel home in Switzerland from the
beginning on of the Ph.D. thesis.
• my whole family for supporting me the last years and giving me the freedom in choosing
my personal development.
150
Curriculum Vitae
Daniel Alexander Knopf
Education and University Career
2000-2003 Ph.D. doctoral thesis at the Institute for Atmospheric and Climate
Sciences. Swiss Federal Institute of Technology (ETH) Zurich, Switzerland.
Title: Thermodynamic Properties and Nucleation Processes
of Upper Tropospheric and Lower Stratospheric Aerosol Particles.
1998-1999 Diploma thesis in atmospheric physics, at the Max-Planck-Institute
for Nuclear Physics, Division Atmospheric Science, Heidelberg, Germany.
Title: Calibration of an Aerosol Beam Mass Spectrometer
with definite Sulfuric Acid Water Aerosols.
1997-1998 During Graduation/Diploma thesis:
Spanish study at the Ruprecht-Karls-University of Heidelberg, Germany.
1997 Graduated in physics at the Ruprecht-Karls-University of Heidelberg.
Elective subjects: environmental physics and economics.
1996 Certificate for "Interdisciplinary Supplementary Studies in
Environmental Sciences", Ruprecht-Karls-University of Heidelberg.
1992 Begin of physics studies at the Ruprecht-Karls-University of Heidelberg.
1983-1992 Graduate of the Hebel-Grammer-School in Schwetzingen, Germany.
Professional Career
2000-2003 Teaching assistance at the Institute for Atmospheric
and Climate Sciences, ETH Zurich, Switzerland.
1999 Scientific assistant responsible for the "AIDA-Aerosol-Project"
at the Max-Planck-Institute for Nuclear Physics, Heidelberg, Germany.
1998-1999 Scientific assistant at the Max-Planck-Institute for Nuclear Physics,
Heidelberg, responsible for the operation of particle accelerators.
1995-1997 Scientific assistant at SAP Corporation, Walldorf, Germany.
Responsible for the installation, configuration, and error finding
in the software and hardware field of complex TCP/IP networks.
1992-1996 Working student employed for the controlling of computerized metal
processing machines at Vögele Corporation, Mannheim, Germany.
1991-1994 Snowboard teacher.
1991-1992 Coach of a table-tennis team.
Languages
German: mother tongue. English fluent in spoken and written. Spanish and French: flu¬
ent/advanced. Italian: basic.
Publications
Peer-Reviewed Articles
• Knopf, D. A., Luo B. P., Krieger, U. K., Koop, T.: Thermodynamic Dissociation Constant
of the Bisulfate Ion from Raman and Ion Interaction Modeling Studies of Aqueous Sulfuric
Acid at Low Temperatures, J. Phys. Chem. A, 107, 4322-4332, 2003.
• Knopf, D. A., Luo B. P., Krieger, U. K., Koop, T., Peter, T.: Homogeneous nucleation
of NAD and NAT in liquid stratospheric aerosols: insufficient to explain denitrification,
Atmos. Chem. Phys., 2, 207-214, 2002.
• Knopf, D. A., Zink, P., Schreiner J., Mauersberger, K.: Cahbration of an Aerosol Com¬
position Mass Spectrometer with Sulfuric Acid Water Aerosol, Aerosol Sei. Technol, 35,
924-928, 2001.
• Schreiner J., Voigt, C, Zink, P., Kohlmann, A., Knopf, D., Weisser, C, Budz, P., Mauers¬
berger, K.: A Mass Spectrometer System for Analysis of Polar Stratospheric Aerosols,
Rev. Sei. Inst, 73, 446-452, 2002.
• Zink P., Knopf, D. A., Schreiner, J., Mauersberger, K., Möhler, O., Saathof, H., Seifert, M.,
Tiède, R., Schurath, U.: Cryo-chamber simulation of stratospheric H2SO4/H2O particles:
Composition analysis and model comparison, Geophys. Res. Lett., 11, Vol. 29, 46-1-46-4,
2002.
Conference Proceedings and Extended Abstracts
• Knopf, D. A., Luo B. P., Krieger, U. K., Koop, T., Peter, T.: Experimental and Theo¬
retical Analysis with Respect to Surface-Induced Nucleation, Madrid, European Aerosol
Conference 2003, J. Aerosol Science, 2003, accepted.
• Knopf, D. A., Luo B. P., Krieger, U. K., Koop, T.: The Thermodynamic Dissociation
Constant of HSOJ at Atmospheric Conditions, Madrid, European Aerosol Conference
2003, J. Aerosol Science, 2003, accepted.
• Knopf, D. A., Koop, T., Weers, U. G., Krieger, U. K., Peter, T.: Investigation of Ice Nucle¬
ation in Liquid Aerosols Using Raman Microscopy, Leipzig, European Aerosol Conference
2001, J. Aerosol Science, 32, Suppl. 1, 283-284, 2001.
• Möhler, O., Bunz, H., Saathoff, H., Schäfer, S., Seifert, M., Tiède, R., Schurath, U., Knopf,
D., Schreiner, J., Voigt, C, Zink, P., Mauersberger, K.: The Potential of the AIDA Aerosol
Chamber for Investigating PSC Formation and Freezing Mechanisms. BAD TOLZ, Work¬
shop 'Mesoscale Processes in the Stratosphere' (9.11.-11.11.1998) proceedings. Air Pollu¬
tion Report 69, Office for Official Publications of European Communities, Luxembourg,
1999, 171-174.
• Zink, P., Knopf, D., Schreiner, J., Voigt, C, Mauersberger, K., Bunz, H., Möhler, O.,
Saathoff, H., Seifert, M., Tiède, R., Schurath, U.: Growth of Aerosol Particles under
Stratospheric Conditions - Experiments inside the AIDA Aerosol Chamber. BAD TOLZ,
Workshop 'Mesoscale Processes in the Stratosphere' (9.11.-11.11.1998) proceedings. Air
Pollution Report 69, Office for Official Publications of European Communities, Luxem¬
bourg, 1999, 281-284.
Ph.D. Thesis and Diploma Thesis
• Knopf, D. A., Thermodynamic Properties and Nucleation Processes of Upper Tropospheric
and Lower Stratospheric Aerosol Particles, Ph.D. thesis 15103, ETH Zurich, Switzerland,
2003.
• Knopf, D., Kalibration eines Aerosolstrahlmassenspektrometers mit definierten Schwe¬
felsäure-Wasser-Aerosolen, Diploma thesis, Ruprecht-Karls-University of Heidelberg, Ger¬
many, 1999.
Invited Talks
• Thermodynamic Properties and Nucleation Processes of UT/LS Aerosol Particles, Uni¬
versity of British Columbia, Chemistry Department, Vancouver, Canada, 20th of May
2003.
• Equilibrium and Non-equilibrium Processes in Aqueous Aerosols of the UT/LS, Institute
for Meteorology and Climatology, Research Center Karlsruhe, Karlsruhe, Germany, 17th
of March 2003.
• Equilibrium and Non-equilibrium Processes in Aqueous Aerosols of the UT/LS, Max-
Planck Institute for Nuclear Physics, Division Atmospheric Science, Heidelberg, Germany,
18th of March 2003.
Oral and Poster Presentations at Conferences
2002 Conference of the European Geophysical Society, Nice, France.
2001 American Geophysical Union, San Francisco, USA.
2001 Visit of the international and interdisciplinary ETH summer school
"Cortona-Week" for Ph.D. students, Cortona, Italy.
2001 European Aerosol Conference, Leipzig, Germany.
1999 Bunsen-Conference of the German Chemical Society, Dortmund, Germany.
1998 Conference of the German Physical Society in Regensburg, Germany.
1996 Conference "Economy-Energy-Entropy-Ecology",
organized by the European Physical Society, Geneva, Switzerland.
Recommended