Hannes.Tammet@ut.ee
Laboratory of Environmental Physics Institute of Physics, University of Tartu
Quiet nucleation of atmospheric aerosol and intermediate ions
Quiet nucleation of atmospheric aerosol and intermediate ions
15th Finnish-Estonian air ion and atmospheric aerosol workshop Hyytiälä 20110524
Sources of knowledge about growth and charging of nanoparticles
Kerminen, V.-M., and Kulmala, M.: Analytical formulae connecting the
“real” and the “apparent”
nucleation rate and the nuclei number
concentration for
atmospheric nucleation events, J. Aerosol Sci., 33, 609–622, 2002.
Tammet H. and
Kulmala M.: Simulation tool for atmospheric aerosol nucleation bursts, J. Aerosol Sci., 36: 173–196, 2005.
Verheggen, B. and Mozurkewich, M.:
An inverse modeling procedure to determine particle growth and
nucleation rates from measured aerosol size distributions, Atmos. Chem. Phys.,
Long quiet periods may happen between burst events. The particles of secondary
aerosol are mortal and would disappear when no supply. How they are regenerated?
Many research papers are written about burst events of atmospheric aerosol nucleation. Not so much
about nucleation during quiet periods between the burst events. Why?
A reason: concentration of intermediate ions sufficiently exceeds the noise level of
common instruments
only during burst events.
Extra noise as in BSMA, lowest contour of 100 cm–3
Measurement with SIGMA, noise from BSMA
Extra noise as in BSMA, lowest contour of 20 cm–3
Measurement with SIGMA, noise from BSMA
Measurement with SIGMA, lowest contour of 20 cm–3
Measurement with SIGMA without extra noise
Low noise instrument
SIGMA:
Tammet, H. (2011) Symmetric inclined grid mobility analyzer for the measurement of charged clusters and fine nanoparticles in atmospheric air. Aerosol Sci. Technol., 45, 468–479.
http://dx.doi.org/10.1080/02786826.2010.546818
Air inlet
Air outlet through Repelli
ng electro
de Attracti
ng electro
des Sheath
air filter
Repelli ng electro
de
Sheath air filter
Attracti ng electrod
es
Repellin g electrod
e
Shield electrod
e Inlet
gate
Air ion trajecto
ry
Electrometricfilter for positive ions Filter batteries Electrometricfilter for negative ions Filter batteries
Shield electro
de
Repellin g electrod
e WORSE HALF OF
MEASUREMENTS
BETTER HALF OF MEASUREMENTS
NOISE
(10 min cycles)
Charged nanoparticles are air ions
Particles and cluster ions
Ion or particle
Molecule or growth unit
Quantum retardation of sticking:
internal enegy levels of a cluster will not be excited and the impact
is elastic-specular
Particle or molecular cluster ? to grow, or not to grow ?
does not grow, molecules will
bounce back
grows,
molecules will stick
1.5 or 1.6 nm
CLUSTER PARTICLE
Introduction to modeling
An aim is to make the mathematical model easy to understand.
GDE is not used and equations will be derived from scratch.
Empiric information is coming from measurements of
intermediate ions. Quiet periods are characterized by very low concentration of nanoparticles and nearly steady state of
aerosol parameters. This allows to accept assumptions:
the size range is restricted with d = 1.5 – 7.5 nm,
the nanoparticles can be neutral or singly charged,
attachment flux of ions does not depend on polarity,
nanoparticle-nanoparticle coagulation is insignificant,
all processes are in the steady state.
Extra comment:
c c
N c
c c I
N c
c c I
0 0
0
0
Assumption: all surfaces are away Law of balance:
genesis = destruction
Flux of ions
to particles
Particle growth through a diameter margin
dd = GR(d) dt d
od
J = GR n
Symbols:
diameter crossing rate,
–
apparent nucleation rate, transit rate, cm–3s–1dt d dN
J ( )
→ dN / dt = GR n dN = n dd = n GR dt d –
particle diameter (d =d
p), nm,dd d d dN
n ( )
)
(
density of concentration distribution, cm–3nm–1–
dt d dd
GR ( ) –
growth rate, nm s–1,N(d) –
number concentration of particles in diameter range of0...d
, cm–3,(a well known equation)
NB: particle growth rate may essentially differ from the population growth rate.
c –
concentration of small ions, cm–3Particle growth through a diameter interval
d
a= d – h/2 d
b= d + h/2
Inflow Leakage Outflow
d
Extrasource
(analog: classic problem about water tank and pipes)
Steady state balance:
Inflow + Extrasource – Outflow – Leakage = 0
or
Outflow = Inflow + Extrasource – Leakage
(GDE : Inflow + Extrasource – Outflow – Leakage = Increment)
Equation of steady state balance
Inflow J(d
a) = GR(d
a) n(d
a), Outflow J(d
b) = GR(d
b) n(d
b), Leakage = ,
ddabS ( d ) n ( d ) dd Extrasource =
ddabE ( d ) dd
ba b
a
d d d
a d a
b
b
n d GR d n d E d dd S d n d dd
d
GR ( ) ( ) ( ) ( ) ( ) ( ) ( )
General steady state balance equation (integral form):
d
a= d – h/2 d
b= d + h/2
Inflow Leakage Outflow
d
Extrasource
Outflow = Inflow + Extrasource – Leakage dt
d dn d
d n
S ( )
) ( ) 1
(
relative depletion rate orsink
of particles s–1,(incl. CoagS as a component)
–
state = CSTChargingComparison with Lehtinen et al. (2007)
ba b
a
d d d
a d a
b
b
n d GR d n d E d dd S d n d dd
d
GR ( ) ( ) ( ) ( ) ( ) ( ) ( )
Balance equation:
Substitute GR n with J, assume E = 0, consider da = const & db = argument:
dconst
S d n d dd const
d
J ( ) ( ) ( )
Equation (4) in Lehtinen et al. (2007):
J GR
d CoagS dd
dJ
pp
)
(
Differences: different notations of sink and two simplifications
E = 0 &
additional components of sink are neglected, substituten
withJ/GR
:) ) (
( ) ( )
( J d
d GR
d S dd
d
dJ )
( ) ) (
( S d n d
dd d
dJ
calculate derivative:
Sink of nanoparticles on background aerosol
The background aerosol can be replaced with an amount of monodisperse particles in simple numerical models. The diameter of particles is
assumed dbkg = 200 nm that is close to the maximum in the distribution of coagulation sink. The concentration Nbkg can be roughly estimated
according to the sink of small ions. The coagulation sink is calculated as
S
bkg= K(d, d
bkg) N
bkgThe coagulation coefficient
K (d, d
bkg)
depends on the nanoparticle charge and the sink could be specified according to the charge.Notations: neutral nanoparticles – index 0, charged nanoparticles – index 1.
Sink of neutral nanoparticles
S
bkg0= K
0(d, d
bkg) N
bkgSink of charged nanoparticles
S
bkg1= K
1(d, d
bkg) N
bkgDifference is small and neglecting of the charge would not induce large errors.
Charging and discharging of particles
0 +
–
+
–
+
– 1
0
1 0
ion-to-neutral-particle attachment coefficient
(a special case of
ion-to-opposite-charged-particle attachment coefficient
or the recombination coefficient
TWO
TWO ONE ONE
Sink of nanoparticles due to the small air ions
When a neutral particle encounters a small air ion then it converts to a charged particle and number of neutral
particles is decreased. We expect concentrations of positive and negative ions c equal and the sink is
S
ion0= 2 β
o(d) c
A charged particle can be neutralized with an ion of
opposite polarity. The sink of charged nanoparticles on small ions is
S
ion1= β
1(d) c
Extrasource of nanoparticles
Some amount of neutral particles appear as a result of
recombination the charged nanoparticles of the same size with small ions of opposite polarity:
E
0(d) = 2 β
1(d) c n
1(d)
The ion attachment source of charged particles of one polarity is
E
1(d) = β
0(d) c n
0(d)
E
0 is usually a minor component in the balance of neutral particles whileE
1 is an important component in the balance of charged particles.If the rate ion-induced nucleation is zero, then all charged
nanoparticles are coming from the extrasource.
Numerical solving of balance equations
ba b
a
d d d
a d a
b
b
n d GR d n d E d dd S d n d dd
d
GR ( ) ( ) ( ) ( ) ( ) ( ) ( )
b ab
a a
b ab
d
db
Y d dd Y d d d d d d
a
( ) ( )( ) where
2
) ( )
) (
(
abY d
aY d
bd
Y
A small step can be made using the midpoint rule and few iterations:
) ( )
( )
1 Y d
b Y d
ab
a
d
d )
...
3, 2,
The first mean value theorem states for any continuous
Y = Y(d):
da db da db da db da
da db da db da db
Step by step: d
GR
orn
can be computed step by step moving upwards or downwardsa
b
d
d
h G
0a GR
0( d
a) G
1b GR
1( d
b)
Abbreviations:
, , , etc.
Itemized numerical model of steady state growth of nanometer particles
h n
c S
h cn
n G
n
G
0b 0b
0a 0a 2
1ab 1ab (
bkg0ab 2
0ab)
0abh n
c S
h cn
n G n
G
1b 1b
1a 1a
0ab 0ab (
bkg1ab
1ab)
1abEquations:
Example of a specific problem:
Given –
nucleation ratesJ
0 andJ
1 or values of distribution functionsn
0 andn
1 at first diameter, and growth ratesGR
0 at all sizes.) , (
) ,
(
1 01
0
K d d G K d d
G
u
uTwo degrees of freedom
Growth rates or values of a distribution function can be computed step by step starting form four initial values of G
0, G
1, n
0, and n
1. If the distribution of intermediate ions is measured then one initial value (n
1) is known. The ratio G
0/G
1is always known and the
number of unknown initial values is reduced to two. These two may be presented by G
0and n
0at some point or by any pair of parameters that are unambigyosly related with G
0and n
0.
Some examples of necessary initial information:
growth rate at a certain size and a nucleation rate,
growth rates at two different sizes,
ratio of growth rates for two sizes and a nucleation rate.
ratio of growth rates for two sizes
and value of n
0at a certain size.
Test data
characteristic of quiet nucleation
Measurements with the SIGMA in the city of Tartu (April 2010 – February 2011) were sorted by the
instrumental noise and the worse half of data was
deleted. Next the data were sorted by concentration of intermediate ions and the half of measurements with high concentration was deleted. Remained 16240 five-minute records are expected to belong to the quiet phase of nucleation.
d : nm dN1/dd : cm–3nm–1
(average of 16240 records of both + and– intermediate ions)
N noise
OK
Fitting the measurements by means of the numerical model
J0 = 5.0 cm–3s–1,
J1 = 0.00133 cm–3s–1, dbkg = 200 nm,
Nbkg = 2224 cm–3.
nm Sbkg:1/h GR CST 1.5 7.38 1.26 0.402 2.0 4.48 2.85 0.686 2.5 3.11 3.74 0.693 3.0 2.33 3.73 0.679 3.5 1.82 3.55 0.680 4.0 1.47 3.54 0.693 4.5 1.22 3.55 0.711 5.0 1.03 3.56 0.751 5.5 0.88 3.59 0.770 6.0 0.76 3.64 0.790 6.5 0.67 3.70 0.809
d : nm dN/dd : cm–3nm–1
(average of + and– ions)
Nbkg is estimated according to the small ion depletion. J0 and J1 are chosen by method of trial and error.
NB: the method does not provide unambiguous results.
Alternative approach
Use any numeric model of nanometer aerosol dynamics, decide
steady state conditions, adjust growth parameters, and integrate over a long period at least of few hours Example (simulation tool)
J0= 13 cm-3s-1, J1 = 0.07 cm-3s-1 d = 1.5 2.5 4.5 6.5 nm GR = 0.8 3.6 3.5 3.8 nm/h dN1/dd : cm–3nm–1
Automated fitting of intermediate ion measurements
Given: measurements of intermediate ions
n
1(d)
on a set of diameters
(d
1, d
2,…, d
m)
Assume and iterate 2…5 times:
n
0b n
0a n
1b/ n
1a
G n c n h S c n h
G n
a a ab bkg abb
b 1 1 0 0 1 1 1
1
1
1 ( )
1 1
0 0
) , (
) ,
( G
d d K
d d G K
u
u n G cn h S c n h
n G
a a ab bkg abb
b 0 0 1 1 0 0 0
0
0
1 2 ( 2 )
2 2 ,
1 1 1
0
0 0 a b
b ab ab a
n n n
n
n n
Fitting the measurements adjusting the growth rate
J0 = 5 cm–3s–1, J1 = 0.013 cm–3s–1, dbkg = 200 nm,
Nbkg = 2224 cm–3. nm Sb:1/h CST 1.5 7.38 0.401 2.0 4.48 0.682 2.5 3.11 0.686 3.0 2.33 0.669 3.5 1.82 0.668 4.0 1.47 0.678 4.5 1.22 0.692 5.0 1.03 0.728 5.5 0.88 0.742 6.0 0.76 0.758 6.5 0.67 0.772
d : nm GR : nm h–1
WARNING: the solution is ambiguous. Different
assumptions about
J
0 andJ
1 are possibleRestrictions on the free parameters
(when fitting the test distribution)
PRIOR INFORMATION?
ANALOG OF REGULARIZATION?
3 variants of GR0(d1) 3 variants of J0(d1) 3 variants of GR0(d1)
3 variants of J0(d1)
Effect of guess about J
0(1.5 nm)
while required relation is
GR
0(3 nm) = GR
0(7 nm)
(fitting the test distribution)
Sink, growth rate and transit rate compared with Lehtinen et al. (2007)
d : nm S : 1/h,
GR0 : nm/h J0(d) : cm–3s–1
Conclusions
SIGMA provides low-noise measurements of intermediate ions.
The integral equation of steady state balance derived in a straigth-forward way enables to design correct numerical algorithms with ease.
Measurement of intermediate ions is not sufficient to getunambiguous solution of balance equation. Additionally the values of two scalar parameters are required. Some combinations are:
growth rate at a certain size and a value of
n
for neutral particles, growth rates at two different sizes, ratio of growth rates at two different sizes and a nucleation rate.
The nucleation of 3 nm neutral particles at Tartu about J = 0.5 cm–3s–1is considerable contribution into the atmospheric aerosol generation.
The nucleation rate of 3 nm charged particles at Tartu about 0.002…0.005 cm–3s–1 indicates the minor contributionof ion-induced nucleation during periods of quiet nucleation.