andrew francis- modeling the melt dispersion mechanism for nanoparticle combustion
TRANSCRIPT
-
8/3/2019 Andrew Francis- Modeling the Melt Dispersion Mechanism for Nanoparticle Combustion
1/42
MODELING THE MELT DISPERSION MECHANISM FOR NANOPARTICLECOMBUSTION
BY
ANDREW FRANCIS
A THESIS
IN
MECHANICAL ENGINEERING
Submitted to the Graduate Faculty
of Texas Tech University inPartial Fulfillment of
the Requirements forthe Degree of
MASTER OF SCIENCE
IN
MECHANICAL ENGINEERING
Valery Levitas
Chairperson of the Committee
Michelle PantoyaCo-Chair
Walt Oler
Fred HartmeisterDean of the Graduate School
December 2007
-
8/3/2019 Andrew Francis- Modeling the Melt Dispersion Mechanism for Nanoparticle Combustion
2/42
Texas Tech University, Andrew Francis, December 2007
ii
ACKNOWLEDGEMENTS
There are truly too many people for me to mention everyone who has helped me
to complete this work. I would like to thank my friends and family for always supporting
me in all aspects of my life. To my wife, Marka, thank you for all of your love and
support, and for always being a Godly example in everything that you do. I am
extremely grateful to Dr. Michelle Pantoya for giving me the opportunity to continue my
intellectual growth and contributing so much to my education. Dr. Valery Levitas, I
really appreciate your patience with me and the time you have spent helping me with this
project. Most of all, I would like to thank Christ for His gift that makes life worth living.
-
8/3/2019 Andrew Francis- Modeling the Melt Dispersion Mechanism for Nanoparticle Combustion
3/42
Texas Tech University, Andrew Francis, December 2007
iii
TABLE OF CONTENTS
Acknowledgements ii
Abstract...... iv
List of Tables..... v
List of Figures........ vi
I: Introduction.. 1
1.1 Thermites..... 1
1.2 Nanocomposite thermites..... 2
1.3 Melt dispersion mechanism..... 5
II: Theory... 7
2.1 Equation for flame rate vs. particle parameters... 7
2.2 Distribution function for particle parameters....... 11
III: Results and Discussion 14
3.1 Effect of distribution of particles geometricparameters on flame propagation rate.. 14
3.2 Effect of passivation temperature (T0) and relative
particle radius (M) on flame propagation rate.. 18
3.3 Effect of alumina shell strength ( u), its distribution, and relativeparticle radius (M) on flame propagation rate.. 22
3.4 Effect of mixing particles with different relative particleradius (M) on flame propagation rate... 26
IV: Conclusions. 31
References.. 33
-
8/3/2019 Andrew Francis- Modeling the Melt Dispersion Mechanism for Nanoparticle Combustion
4/42
Texas Tech University, Andrew Francis, December 2007
iv
ABSTRACT
Thermite particles have long been known to increase in reactivity as they decrease
in size. However, during fast heating (106- 10
8K/s) of Al nanothermites, the diffusion
mechanism that explains micron size thermite reactions cannot explain the extremely fast
ignition times and much higher flame propagation velocities. A new mechanism known
as the melt dispersion mechanism has recently been introduced to explain the fast
oxidation of these Al nanothermites. A model has been created dependant upon key
parameters to predict the reactivity of Al nanothermites. In this study, flame propagation
velocities are statistically evaluated in terms of an integral that employs a probability
density function (pdf) for key parameters and a flame velocity equation dependent on
relative particle size (Al core radius divided by oxide shell thickness), oxide shell
formation temperature, and oxide shell strength. It is shown that flame propagation
velocity depends sensitively on relative particle size, relative particle size distribution,
oxide shell formation temperature, and shell strength. It is also dependant upon particle
size, and oxide shell thickness but not as sensitively. Both single and bimodal particle
sizes were studied. Combining smaller nanoparticles with larger nanoparticles in a
bimodal mixture significantly increases the flame propagation velocity as compared to a
composite consisting of only the larger particles. The results presented here suggest that
better reproducibility of the flame velocity may be achieved experimentally by selecting a
material with a narrow relative particle size distribution. A combination of increased
oxide shell formation temperature and increased oxide shell strength could be used to
maximize the flame velocity in particles with increased relative particle size.
-
8/3/2019 Andrew Francis- Modeling the Melt Dispersion Mechanism for Nanoparticle Combustion
5/42
Texas Tech University, Andrew Francis, December 2007
v
LIST OF TABLES
1. Material parameters at melt temperature T=Tm... 8
2. Experimental data corresponding with Figure 4...... 10
-
8/3/2019 Andrew Francis- Modeling the Melt Dispersion Mechanism for Nanoparticle Combustion
6/42
Texas Tech University, Andrew Francis, December 2007
vi
LIST OF FIGURES
1. Heat of reaction of Thermitesand High Explosives.... 2
2. Flame propagation speeds (burn rate) for nano-Al mixed
with MoO3 loose powders 4
3. Oxidation of micron (a) and nano (b) Al particles....... 6
4. Correlation between experimental data and melt dispersion model.... 10
5. Relative flame velocity as a function of relative
standard deviation for variations inR.............................................................................. 14
6. Relative flame velocity as a function of relative
standard deviation for variations in ... 16
7. Relative flame velocity as a function of relative
particle size for various ..... 17
8. Relative flame velocity as a function of relativestandard deviation for T0 = 300 K... 19
9. Relative flame velocity as a function of relative
standard deviation for T0 = 400 K... 20
10. Relative flame velocity as a function of relative
standard deviation for T0 = 500 K... 20
11. Relative flame velocity as a function of relative
standard deviation for T0 = 600 K... 21
12. Relative flame velocity as a function of relativestandard deviation for T0 = 700 K... 21
13. Relative flame velocity as a function of relative standard deviation foru = th 23
14. Relative flame velocity as a function of relative standard deviation foru = th/1.1.. 23
15. Relative flame velocity as a function of relative standard deviation foru = th/1.2.. 24
16. Relative flame velocity as a function of relative standard deviation foru = 1.1th... 24
-
8/3/2019 Andrew Francis- Modeling the Melt Dispersion Mechanism for Nanoparticle Combustion
7/42
Texas Tech University, Andrew Francis, December 2007
vii
17. Relative flame velocity as a function of relative standard deviation foru = 1.2th....... 25
18. Relative flame velocity as a function of mass fraction for M2 = 40........ 28
19.Relative flame velocity as a function of mass fraction for M2 = 80.... 28
20.Relative flame velocity as a function of mass fraction for M2 = 150.......... 29
21.Relative flame velocity as a function of mass fraction for M2 = 1000........ 29
22. Relative flame velocity as a function of mass fraction for M1 = 20........ 30
-
8/3/2019 Andrew Francis- Modeling the Melt Dispersion Mechanism for Nanoparticle Combustion
8/42
Texas Tech University, Andrew Francis, December 2007
1
CHAPTER I
INTRODUCTION
1.1 Thermites
A traditional thermite reaction is an exothermic reduction-oxidation reaction
between a metal (e.g., Al, Mg, B, Ti, and Zr) and a metallic or non-metallic oxide (e.g.,
Fe2O3, CuO, MoO3, and C2F4) [21]. The aluminum (Al) particles in a thermite are
covered by a thin but growing oxide shell () on the order of nanometers (10-9 m) in
thickness. For the reaction to take place, oxygen from the oxidizer must diffuse through
the oxide shell to the metal, and particles from the metal must diffuse to the oxidizer. As
a result, traditional thermite reactions are rate limited by the diffusive rates of the fuel and
oxidizer through the oxide shell. This reaction is relatively slow when compared to the
reaction of a high explosive (HE). A high explosive reaction is also a reduction-
oxidation reaction, but does not depend on diffusion rates and instead is controlled by the
energy required to break atomic bonds. The reaction of a HE occurs so fast that it
detonates, creating a supersonic shock wave that compresses the unreacted material in
front of the shock wave increasing the temperature to the point of ignition [10]. The HE
cyclotetramethylene-tetranitramine (HMX) detonates at the velocity of 9.1 km/s which is
more than three times the speed of sound in HMX [9][14]. The unconfined thermite 1-3
m Al+MoO3 is shown to have a flame propagation velocity of 4.1 m/s, 10
3
times less
than the speed of sound in Al [22][11]. Consequently, the time for a traditional thermite
to completely react is on the order of seconds in contrast to that of a HE, which is on the
-
8/3/2019 Andrew Francis- Modeling the Melt Dispersion Mechanism for Nanoparticle Combustion
9/42
Texas Tech University, Andrew Francis, December 2007
2
order of picoseconds (10-12
s) [7][5]. However, the amount of energy released by a
traditional thermite reaction is larger than that in a HE reaction. Figure 1 shows the heat
of reaction for three traditional thermite reactions and three common HE reactions.
0 1000 2000 3000 4000 5000
Al/Bi2O3
Al/Fe2O3
Al/MoO3
TNT
RDX
HMX
Heat of Reaction
cal/g
cal/cc
Figure 1. Heat of reaction of Thermites [6]and High Explosives [19].
1.2 Nanocomposite thermites
Nanocomposite thermites, sometimes referred to as metastable intermolecular
composites (MICs), are of great interest because they give off large amounts of energy
compared to HEs, provide much faster energy release when compared to traditional
thermites, and allow for greater control over material homogeneity and material
properties [5]. The reactivity of thermites, evaluated by ignition delay time (tig) and
-
8/3/2019 Andrew Francis- Modeling the Melt Dispersion Mechanism for Nanoparticle Combustion
10/42
Texas Tech University, Andrew Francis, December 2007
3
flame propagation velocity (C), increases as the diameter of the Al fuel particle
decreases. When the diameter of the Al particle is reduced from the traditional 1-100 m
range to the 20-120 nm range, the flame propagation velocities change from the order of
cm/s to 0.9-1 km/s, and the ignition delay times are 1000 times faster [1][20][3][17].
While undergoing fast heating (106- 10
8K/s), the rise time (tf) for temperature
and pressure in a nanocomposite thermite, which characterizes the reaction time, is on the
order of 10 s [3][2]. For self-diffusion coefficients (D) of oxygen and Al in -alumina
at 800-950C, D = 10-19 and 10-18 cm2/s respectively. The diffusion length
nmxDtl fd 510)62(2 == , which is five orders of magnitude smaller than the oxide
shell thickness [12]. If diffusion was the mechanism through which this reaction takes
place, the reaction time would have to be 1010
times slower, or the time of diffusion
would have to be 1010 times faster.
There is also other evidence that shows the diffusion mechanism which explains
microcomposite thermite reactions does not hold true for nanocomposite thermites. First,
for diffusion-controlled oxidation, the flame propagation rate is proportional to one over
the square of the particle diameter (d-2
) [4]. However, for particle diameters smaller than
80 nm, flame speed is independent of particle size (See Figure 2) [3].
-
8/3/2019 Andrew Francis- Modeling the Melt Dispersion Mechanism for Nanoparticle Combustion
11/42
Texas Tech University, Andrew Francis, December 2007
4
Figure 2. Flame propagation speeds (burn rate) for nano-Al mixed with MoO3 loose
powders [17]
Second, for diffusion controlled oxidation, ignition delay time is a power function
ofd. For particle sizes d < 120 nm, ignition delay time is independent of particle size [8].
Third, the flame propagation rates for microthermites increase with sample density, but
for nanothermites it decreases [16].
The inability of diffusion controlled oxidation to explain the reactivity of
nanocomposite thermites indicates that other models are needed to help explain
experimental observations. One model presented by Levitas et al is described in terms of
a dispersion mechanism and shows potential for explaining experimental observations
associated with nano-particle combustion. The objective of this study is to explore how
500
700
900
1100
0 50 100 150
Al Particle Diameter (nm)
BurnRate(m/s)
Independent of particlediameter
500
700
900
1100
0 50 100 150
Al Particle Diameter (nm)
BurnRate(m/s)
Independent of particlediameter
-
8/3/2019 Andrew Francis- Modeling the Melt Dispersion Mechanism for Nanoparticle Combustion
12/42
Texas Tech University, Andrew Francis, December 2007
5
varying particle parameters impact the operational regime for the melt dispersion
mechanism.
1.3 Melt dispersion mechanism
For relatively slow heating rates (~103
K/s) nanothermites behave like
microthermites, reacting slowly, signifying that the reaction is rate limited by diffusion.
For both micro and nanothermites the oxide shell grows as metallic and oxidizer
molecules diffuse through the oxide shell and react with their counterparts. In both cases,
either the oxide shell breaks before melting or slow damage to the oxide shell leads to
slow flow of the liquid Al through cracks in the oxide shell inhibiting oxide spallation
[18].
Fast heating of nanothermites (106 108 K/s) creates substantial internal stresses
due to the difference in thermal expansion of Al and Al2O3. Also, the melting
temperature of nano-Al decreases with particle diameter. This allows the Al core of the
fuel particles to begin to melt during fast heating [11]. The volume of Al increases by
6% when it melts [23]. The pressure inside the Al particle causes tensile hoop stress (h)
in the oxide shell. Due to the small thickness of the oxide shell (1 8 nm), it is almost
defect free and therefore its ultimate strength (u) approaches the theoretical maximum
strength of alumina (th) estimated at 11.33 GPa [11]. The combination of large volume
change, differences in thermal expansion coefficients, and high oxide shell strength
causes high pressures within the Al core (1-2 GPa) [11]. When -h = u the oxide shell
spallates causing an imbalance in the pressure within the Al particle and exposed surface.
-
8/3/2019 Andrew Francis- Modeling the Melt Dispersion Mechanism for Nanoparticle Combustion
13/42
Texas Tech University, Andrew Francis, December 2007
6
The pressure at the surface of the Al sphere immediately following spallation is estimated
to be 10 MPa due to gas pressure and surface tension, while internally the pressure is still
1-2 GPa [11]. This results in an unloading wave that propagates to the center of the
particle creating a tensile pressure of similar value. This tensile pressure disperses
molten atomic scale clusters of Al radially outward at a velocity of 100 250 m/s [11].
The clusters then react quickly, on the order of picoseconds (10-12
s), with the oxidizer
because they no longer have an oxide shell [11].
Aluminum meltsbefore shellfracture
Small size moltenaluminum clustersdisperse from anunloading wave at highvelocity
b) Nano:
a) Micron:
Initial alumina shell and aluminumcore
Growingalumina shell
fractures inalumina shell
Healeddefectivealumina shell
Characteristic time
Spallatingalumina shell
Figure 3. Oxidation of micron (a) and nano (b) Al particles. Adapted from [11].
-
8/3/2019 Andrew Francis- Modeling the Melt Dispersion Mechanism for Nanoparticle Combustion
14/42
Texas Tech University, Andrew Francis, December 2007
7
CHAPTER II
THEORY
2.1 Equation for flame rate vs. particle parameters
Levitas et al showed that the major parameter controlling the reactivity of nano-Al
thermites is the volume fraction of Al melt (f) at the time of oxide shell spallation, which
occurs when the maximum tensile stress at the internal surface of the shell reaches the
ultimate tensile strength of the oxide, -h = u. The maximum tensile stress (h) in the
alumina oxide shell at r=R was derived by Levitas et al [11] for the case of a two-layer
sphere with an internal radius R and external radius R~
using elasticity theory. The first
term in Equation 1 represents the stress due to the thermal expansion of solid Al, and Al
melting in the particle core. The second term represents the stress caused by surface
tension at the aluminum-alumina interface. The third term accounts for the surface
tension in the oxide shell at the alumina-gas interface and the pressure of the gas on the
oxide shell.
))2(32()2()2(4))(2(6 212122
2122
3
21212
3
hRH
KKGKGmRp
RH
KGm
H
KKGm gii +++
++
++
=
[1]
where
))()()1()(( 11011m
m
m
m
s
m
si fTTfTTfTT +++= , [2]
)( 022 TTi = , [3]
sm KffKK 111 )1( += , [4]
-
8/3/2019 Andrew Francis- Modeling the Melt Dispersion Mechanism for Nanoparticle Combustion
15/42
Texas Tech University, Andrew Francis, December 2007
8
MRRRm /11/1/~
+=+== , [5]
))1((43 23
1221
3 KmKGKKmH ++= . [6]
Here Mis the relative Al core radius,R/,Kis the bulk modulus, G is the shear modulus,
is the linear thermal expansion coefficient, f is the volume fraction of Al melt in the
particle core, Tmis the Al melting temperature, T0is the oxide shell passivation
temperature, 3 m is the volumetric expansion during the melting of Al,1
and 2 are the
surface tensions at the aluminum-alumina interface and alumina-gas interface that
appears during the reaction, gp is the pressure of the gas outside the oxide shell,
subscriptss and m are for the solid and melt phases of Al, and subscripts 1 and 2 are for
Al and Al2O3, respectively. The values of the parameters in Equation 1 are given in
Table 1 [11].
Table 1. Material parameters at melt temperature T=Tm [11]
Substituting Equation 1 into oxide shell fracture criterion -h = u , one obtains the
equation for the volume fraction of melt necessary to cause fracture of the shell.
02 =++ CBfAf [7]
-
8/3/2019 Andrew Francis- Modeling the Melt Dispersion Mechanism for Nanoparticle Combustion
16/42
Texas Tech University, Andrew Francis, December 2007
9
where
22
3)2(6 KMGmKA m += , [8]
us
m
g
KmGKMKGTKKMm
KGpMKmB
)34()()2(6
)32)(2)1((
2
3
222
3
222
2
++++
+++=
[9]
)))1((43(
))2(32)(2)1((
)2(4)2(6
2
3
2
3
2
2222
2
221
3
22
3
KmKGmKKM
KGKGKpMm
KGmKTGKMmC
SSu
SSg
S
++
++++
++=
[10]
Here K is the difference in bulk moduli between liquid and solid Al, SK is the bulk
modulus of solid Al, T is the difference between the melting temperature of Al, Tm, and
the passivation temperature, T0, at which the initial oxide shell was formed, and is
the difference in linear thermal expansion coefficients between solid Al and alumina [13].
The flame velocity can be expressed in terms of the volume of the melt as follows
AACBBVfVV 2/4( 2maxmax +== for 10 < f , [11]
maxVV = for f =1,
where Vmax is the maximum velocity (950 m/s) that can be achieved in the experimental
setup under study [13]. Levitas et al used Equations 7 through 11 to plot the relative
flame velocity (V/Vmax) as a function ofMfor different values ofu, beginning with u =
th = 11.33 GPa. Figure 4 shows a comparison between these theoretical values and
experimental data given in Table 2 that was previously obtained.
-
8/3/2019 Andrew Francis- Modeling the Melt Dispersion Mechanism for Nanoparticle Combustion
17/42
Texas Tech University, Andrew Francis, December 2007
10
Table 2. Experimental data corresponding to Figure 4 [13]
Figure 4. Correlation between experimental data and melt dispersion model [13].
It is evident that the experimental data correlates well with the melt dispersion
theory.
Diameter Shell Thickness Average flame velocity
nm nm M m/s
44 2 10.00 950
50 1 24.00 789.3
80 2 19.00 948
110 1.5 35.67 721
121 2 29.25 765
120 4 14.00 947
-
8/3/2019 Andrew Francis- Modeling the Melt Dispersion Mechanism for Nanoparticle Combustion
18/42
Texas Tech University, Andrew Francis, December 2007
11
2.2 Distribution function for particle parameters
This study explores the effects of varying key parameters within the melt
dispersion mechanism. This analysis is based on probability density functions (pdfs) for
the particle size distribution (p), oxide shell thickness distribution (t), and oxide shell
strength distribution (s). The equations for these normal distributions used for this
analysis are given by
=
2
2
1exp
2
1),(
MM
M
Mxxp
, [12]
=2
2
1exp
2
1),(
xxt , [13]
=
2
2
1exp
2
1),(
u
xxs , [14]
wherex is the integration variable, and M , , and are the standard deviations
corresponding to parameters M, , and u, respectively. These pdfs give the probabilityp
dx, t dx, ands dx that a given particle with a parameterx will lie in the interval dx aboutx.
The integral of each pdf fromx = to is unity. However, the only valid values
for the parameters M, , and u used in our analysis are positive (>0), so each of the pdfs
can only be integrated from 0 to . Each pdf can be multiplied by a variable that
depends onx. The variables of interest in this analysis are the total volume of the
spherical Al core (v) and the volume of the Al core that is molten at oxide shell spallation
(vm).
-
8/3/2019 Andrew Francis- Modeling the Melt Dispersion Mechanism for Nanoparticle Combustion
19/42
Texas Tech University, Andrew Francis, December 2007
12
33
3
4Mv = [15]
vffMvm
== 33
3
4 [16]
Integration with respect tox from 0 to then returns the mean value of variables v and
vm.
dxxvxpvp )()(0
= [17]
dxxvxpv mmp
)()(0
= [18]
dxxvxtvt )()(0
= [19]
dxxvxtv mmt )()(
0
= [20]
dxxvxsvs )()(0
= [21]
dxxvxsv mms )()(
0
= [22]
Here superscriptsp, t, ands correspond to parameters M, , and u, respectively.
Dividing the average volume of molten Al at oxide shell spallation by the average total
volume of the spherical Al core ( vvm ) gives the average fraction of Al melt ( f ) needed
-
8/3/2019 Andrew Francis- Modeling the Melt Dispersion Mechanism for Nanoparticle Combustion
20/42
Texas Tech University, Andrew Francis, December 2007
13
to cause oxide shell spallation which is equivalent to relative flame velocity, V/Vmax,from
Equation 11. Dividingmv by v also cancels any deviation from unity caused by making
0 the lower limit of integration.
maxVVvvfp
m
pp == [23]
maxVVvvft
m
tt == [24]
maxVVvvfs
m
ss == [25]
Equation 23 can be manipulated to account for relative flame velocity varying due
to distributions of bothR and . In order to do this, both the numerator and the
denominator of Equation 23 is multiplied by the pdf of oxide shell thickness distribution
(t(y, ), shown in Equation 13. A new integration variable (y) is introduced to designate
the integration variable related to .. The equation, shown in Equation 26, must then be
integrated twice.
( ) ( )
( ) ( )dydxyx
ee
dxdyTyxfyxee
VVvvfy
M
Mx
u
y
M
Mx
II
m
IIII
M
M
===
0 0
33
5.0
2
)(
5.0
2
)(
0
0
33
0
5.0
2
)(
5.0
2
)(
max
3
4
22
),,,(3
4
22
2
2
2
2
2
2
2
2
[26]
-
8/3/2019 Andrew Francis- Modeling the Melt Dispersion Mechanism for Nanoparticle Combustion
21/42
Texas Tech University, Andrew Francis, December 2007
14
CHAPTER III
RESULTS AND DISCUSSION
3.1 Effect of distribution of particles geometric parameters on flame propagation
rate
Values of relative flame velocitywere plotted versus relative standard deviation
( M/M) for values ofMranging from 5 to 500 in Figure 5 using Equation 23.
( )
( )
===
0
33
5.0
2
)(
0
33
0
5.0
2
)(
max
3
4
2
),,,(
3
4
2
2
2
2
2
dxxe
dxTxfxe
VVvvf
M
Mx
u
M
Mx
p
m
pp
M
M
[23]
Figure 5. Relative flame velocity as a function of relative standard deviation for variations inR.
0.2 0.4 0.6 0.8 1
0.5
0.6
0.7
0.8
0.9
1
30
40
20
10
60
100
500
M=5
RelativeFlameVelocity(V/V
max)
Relative Standard Deviation (/)
-
8/3/2019 Andrew Francis- Modeling the Melt Dispersion Mechanism for Nanoparticle Combustion
22/42
Texas Tech University, Andrew Francis, December 2007
15
For this case, Tm =933.67 K, T0 = 300 K, =2 nm and u=11.33 GPa. The only
parameter that varies is M, which is assumed to have a standard normal distribution with
standard deviation m. Since Mis equal toR/, and is held constant,R alone varies.
The actual upper integration limit used to evaluate this equation and all subsequent
integrals with an upper limit of was the nominal value of the varying parameter plus
seven standard deviations (e.g. for this case M+7 M). Above this the effects were
negligible (below .00001%).
Figure 6 shows the relationship between V/Vmaxand relative standard deviation
derived from Equation 24.
( )
( )
===
0
33
5.0
2
)(
0
33
0
5.0
2
)(
max
3
4
2
),,,(3
4
2
2
2
2
2
dxxMe
dxTxMfxMe
VVvvfx
u
x
t
m
tt
[24]
Values ofTm, T0, u, and were kept the same as before, but this timeR was held
constant for each value ofM, and was assumed to have a standard normal distribution.
-
8/3/2019 Andrew Francis- Modeling the Melt Dispersion Mechanism for Nanoparticle Combustion
23/42
Texas Tech University, Andrew Francis, December 2007
16
Figure 6. Relative flame velocity as a function of relative standard deviation for variations in .
It is evident that for the same nominal value ofM, variations inR affect the
relative flame velocity more than variations in . This is due to the fact ifR varies, the
range ofMvalues is twice as large as when varies. For example, for relative standard
deviation of .5, whenR varies, the range ofMvalues is .5Mto 1.5M, but when varies,
Mranges from .75Mto 1.25M. Also, when is allowed to vary, it is well below the
critical value of equal to 7.7 nm below which the oxide shell strength approaches
maximum strength [11].
Equation 26 can be used to model relative flame velocity assumed to vary due to
distributions of bothR and . Simplifying Equation 26 results in
0.2 0.4 0.6 0.8 1
0.5
0.6
0.7
0.8
0.9
1
30
40
20
60
100
500RelativeFlameVeloc
ity(V/Vmax)
M=5 & 10
Relative Standard Deviation (/)
25
-
8/3/2019 Andrew Francis- Modeling the Melt Dispersion Mechanism for Nanoparticle Combustion
24/42
Texas Tech University, Andrew Francis, December 2007
17
dyyedxxe
dxdyTyxfyxee
VVvvfyMx
M
u
yMx
MII
m
IIII
M
M
===
0 0
32
)(
32
)(
0
0
33
0
2
)(
2
)(
max
2
2
2
2
2
2
2
2
3
2
),,,(3
2
[27]
Figure 7 shows how V/Vmax varies with Mfor various values of.
Figure 7. Relative flame velocity as a function of relative particle size for various .
Because of the small change in flame propagation velocity as increases beyond
the scope of the melt dispersion mechanism, relative flame propagation velocity is
assumed to vary uniformly at the average relative flame velocity between oxide shell
thicknesses = 2 nm and 200 nm (i.e. = 3.8 nm) above M=18.9. Below M=18.9,
50 100 150 200
0.5
0.6
0.7
0.8
0.9
1
Relative particle size (M)
Relativeflamevelocity(V/Vmax)
=2 nm
=3.8 nm
=200 nm
18.9
=3.8nm
-
8/3/2019 Andrew Francis- Modeling the Melt Dispersion Mechanism for Nanoparticle Combustion
25/42
Texas Tech University, Andrew Francis, December 2007
18
relative flame velocity is not affected by changes in . Based on these assumptions,
Equation 27 is simplified to
dyyedxxe
dyyeTxfxexe
VVvvfyMx
M
y
u
MxMx
MII
m
IIII
M
MM
+===
0 0
32
)(
32
)(
3
0
2
)(
0 9.18
0
32
)(
32
)(
max
2
2
2
2
2
2
2
2
2
2
3
2
),,8.3,(3
29.18
[28]
Reducing this equation results in
+
===
0
32
)(
0 9.18
0
32
)(
32
)(
max
2
2
2
2
2
2
9.18
),,8.3,(
dxxe
Txfxexe
VVvvfM
MM
Mx
u
MxMx
II
m
IIII
. [29]
which is independent of variations in .
3.2 Effect of passivation temperature (T0) and relative particle radius (M) on flame
propagation rate
Another major parameter affecting flame velocity is oxide shell formation
temperature, T0. The effects of increasing T0 for different values ofMwere studied using
Equation 23.
( )
( )
===
0
33
5.0
2
)(
0
33
0
5.0
2
)(
max
3
4
2
),,,(3
4
2
2
2
2
2
dxxe
dxTxfxe
VVvvf
M
Mx
u
M
Mx
p
m
pp
M
M
[23]
-
8/3/2019 Andrew Francis- Modeling the Melt Dispersion Mechanism for Nanoparticle Combustion
26/42
Texas Tech University, Andrew Francis, December 2007
19
Relative flame velocity was plotted against relative standard deviation ofR for the
same values ofTm, , and u as before for values of T0 ranging from 300 K to 700 K.
Figure 8. Relative flame velocity as a function of relative standard deviation forT0 = 300 K.
0.2 0.4 0.6 0.8 1
0.5
0.6
0.7
0.8
0.9
1
30
40
20
10
60
100
500
M=5T0 = 300 K
RelativeFlameVelocity(V/Vmax)
Relative Standard Deviation (M/M)
-
8/3/2019 Andrew Francis- Modeling the Melt Dispersion Mechanism for Nanoparticle Combustion
27/42
Texas Tech University, Andrew Francis, December 2007
20
Figure 9. Relative flame velocity as a function of relative standard deviation forT0 = 400 K.
Figure 10. Relative flame velocity as a function of relative standard deviation forT0 = 500 K.
0.2 0.4 0.6 0.8 1
0.6
0.7
0.8
0.9
1
T0 = 400 K
30
40
20
10
60
100
500
M=5
RelativeFlameVelocity(V/Vmax)
Relative Standard Deviation (M/M)
0.2 0.4 0.6 0.8 1
0.7
0.75
0.8
0.85
0.9
0.95
1
T0 = 500 K
30
40
20
10
60
100
500
M=5
RelativeFlameVelocity(V/Vmax)
Relative Standard Deviation (M/M)
-
8/3/2019 Andrew Francis- Modeling the Melt Dispersion Mechanism for Nanoparticle Combustion
28/42
Texas Tech University, Andrew Francis, December 2007
21
Figure 11. Relative flame velocity as a function of relative standard deviation forT0 = 600 K.
Figure 12. Relative flame velocity as a function of relative standard deviation forT0 = 700 K.
0.2 0.4 0.6 0.8 1
0.92
0.94
0.96
0.98
1
T0 = 700 K
40
60
100
M=5 10 2030
500
Relative Standard Deviation (M/M)
RelativeFlameVelocity(V/Vm
ax)
150
300
200
0.2 0.4 0.6 0.8 1
0.8
0.85
0.9
0.95
1
30
40
20
10
60
100
500RelativeFlameVelocity(V/Vmax)
M=5
Relative Standard Deviation (M/M)
T0 = 600 K
-
8/3/2019 Andrew Francis- Modeling the Melt Dispersion Mechanism for Nanoparticle Combustion
29/42
Texas Tech University, Andrew Francis, December 2007
22
The relative flame velocity increases dramatically with an increase in T0. As a
comparison, for T0 =300 K, complete melting of Al begins for values of nmM 6.19 ,
but for T0 = 700 K, complete melting begins for nmM 4.112 . This is due to the
decrease in difference between T0 and the melting temperature of Al (Tm). This smaller
difference causes less pressure build up in the Al particle due to differences in inelastic
strains of Al and Al2O3 (see Equations 1, 2, and 3) and allows a higher percentage of Al
to melt before oxide shell spallation.
3.3 Effect of alumina shell strength ( u), its distribution, and relativeparticle radius(M) on flame propagation rate
The effects of variations in oxide shell ultimate strength, u, for different values
ofMwere studied using Equation 25.
( )
( )
===
0
33
5.0
2
)(
0
33
0
5.0
2
)(
max
3
4
2
),,,(3
4
2
2
2
2
2
dxMe
dxTxMfMe
VVvvf u
u
x
x
s
m
ss
25
Relative flame velocity was plotted against relative standard deviation ofu for
Tm=933.67 K, =2 nm, T0=300 K, and various values ofMin Figures 13 to 17.
-
8/3/2019 Andrew Francis- Modeling the Melt Dispersion Mechanism for Nanoparticle Combustion
30/42
Texas Tech University, Andrew Francis, December 2007
23
Figure 13. Relative flame velocity as a function of relative standard deviation foru = th.
Figure 14. Relative flame velocity as a function of relative standard deviation foru = th/1.1
0.1 0.2 0.3 0.4
0.5
0.6
0.7
0.8
0.9
1
RelativeFlameVelocity(V/Vmax)
Relative Standard Deviation (sg/sg)
25
40
10
60
100
500
20
M=5
u = th =11.33 GPa
16
0.1 0.2 0.3 0.4
0.4
0.5
0.6
0.7
0.8
0.9
1
RelativeFlameVelocity
(V/Vmax)
Relative Standard Deviation (sg/sg)
25
40
10
60
100
500
20
M=5
u = th/1.1
16
-
8/3/2019 Andrew Francis- Modeling the Melt Dispersion Mechanism for Nanoparticle Combustion
31/42
Texas Tech University, Andrew Francis, December 2007
24
Figure 15. Relative flame velocity as a function of relative standard deviation foru = th/1.2
Figure 16. Relative flame velocity as a function of relative standard deviation foru = 1.1th.
0.1 0.2 0.3 0.4
0.3
0.4
0.5
0.6
0.7
0.8
0.9
1
RelativeFlameVelocity(V/Vmax)
Relative Standard Deviation (sg/sg)
25
40
10
60
100
500
16
20
M=5
u = th /1.2
0.1 0.2 0.3 0.4
0.6
0.7
0.8
0.9
1
RelativeFlameVelocity(V/Vma
x)
Relative Standard Deviation (sg/sg)
40
60
10
100
500
1000
16
25
M=5
u
= 1.1th
-
8/3/2019 Andrew Francis- Modeling the Melt Dispersion Mechanism for Nanoparticle Combustion
32/42
Texas Tech University, Andrew Francis, December 2007
25
Figure 17. Relative flame velocity as a function of relative standard deviation foru = 1.2th.
The flame velocity also varies directly with u. When u increases, more pressure
is required to cause oxide shell spallation so more Al melts before -h = u. This increase
in pressure causes a greater imbalance of pressure in the Al sphere immediately following
spallation which produces larger tensile stress in the unloading wave and promotes
dispersion of any solid Al left in the core [12]. Particles with a ratio of M=1000 reach a
higher relative velocity than M=40 particles by increasing u by only 20%, from th to
1.2th. For higher values ofu, the relative flame velocity increases much more rapidly
with increasing M than it does at lower values ofu.
0.1 0.2 0.3 0.4
0.6
0.7
0.8
0.9
1
RelativeFlameVelocity(V/V
max)
Relative Standard Deviation (sg/sg)
u = 1.2th
40
60
10
100
500
1000
16
25
M=5
-
8/3/2019 Andrew Francis- Modeling the Melt Dispersion Mechanism for Nanoparticle Combustion
33/42
Texas Tech University, Andrew Francis, December 2007
26
3.4 Effect of mixing particles with different relative particle
radius (M) on flame propagation rate
The effect of mixing particles with multiple Mvalues (M1 and M2) was also
examined using the MDM model. The volume of the Al contained in the core of each
particle is given by
3
1
3
113
4MvA = [30]
3
2
3
223
4MvA = , [31]
where subscripts 1 and 2 represent particles M1 and M2, respectively. The total amount of
Al within the mixture is given by
AA vnvnv 2211 +=
[32]
where n represents the number of particles of each relative size in the mixture. Using
Equation 11 to find the volume fraction of melt required for oxide shell spallation, the
total amount of Al that melts within the mixture can be given by
),(),( 22221111 MfvnMfvnvAA
m +=
. [33]
Because of how extremely small these particles are, it is impractical to count the
number of each size of particle. Before mixing the particles with different Mvalues, the
mass of each can be determined and from that, one can find the total mass of Al involved
in the reaction along with the mass percent of each particle size. This can be
accomplished by using the following equations
)( 111111SSAA vvncmm +== , [34]
-
8/3/2019 Andrew Francis- Modeling the Melt Dispersion Mechanism for Nanoparticle Combustion
34/42
Texas Tech University, Andrew Francis, December 2007
27
)()1( 222222SSAA vvnmcm +== , [35]
SSAA vv
cmn
1111
1 +
=, [36]
SSAAvv
mcn
2222
2
)1(
+
=
, [37]
where m is the total mass of all the particles, m1and m2 are the masses of particles with
relative size M1and M2,c is the mass fraction ofM1 particles, (1-c) is the mass fraction of
M2 particles, A
and S
are the densities of Al and Al2O3, respectively, for each relative
particle size. After determining v and vm,the volume fraction of Al melt at oxide shell
spallation is by definition
maxVVvvf m ==
. [38]
Relative flame velocity was plotted in Figures 18 through 21 as a function of the
mass fraction of the particle with the smallerMvalue (M1). For this case, Tm =933.67 K,
T0 = 300 K, u=11.33 GPa, and =2 nm forM1and =6 nm forM2. The density of Al
and Al2O3 are 2.7g/cm3
and 3.05g/cm3, respectively.
-
8/3/2019 Andrew Francis- Modeling the Melt Dispersion Mechanism for Nanoparticle Combustion
35/42
Texas Tech University, Andrew Francis, December 2007
28
Figure 18. Relative flame velocity as a function of mass fraction for M2 = 40.
Figure 19. Relative flame velocity as a function of mass fraction for M2 = 80.
0.2 0.4 0.6 0.8 1
0.7
0.75
0.8
0.85
0.9
0.95
RelativeFlameVelocity
(V/Vmax)
M2 = 40
30
40
25
35
M1=20
Mass Fraction M1 (c)
0.2 0.4 0.6 0.8 1
0.6
0.7
0.8
0.9
RelativeFlameVelocity(V/Vmax)
Mass Fraction M1 (c)
60
80
M2 = 80
30
45
M1=20
-
8/3/2019 Andrew Francis- Modeling the Melt Dispersion Mechanism for Nanoparticle Combustion
36/42
Texas Tech University, Andrew Francis, December 2007
29
Figure 20. Relative flame velocity as a function of mass fraction for M2 = 150.
Figure 21. Relative flame velocity as a function of mass fraction for M2 = 1000.
0.2 0.4 0.6 0.8 1
0.6
0.7
0.8
0.9
RelativeFlameVelocity(
V/Vmax)
M2 = 150
50
100
35
150
M1=20
Mass Fraction M1
(c)
0.2 0.4 0.6 0.8 1
0.5
0.6
0.7
0.8
0.9
RelativeFlameVelocity(V/Vm
ax)
Mass Fraction M1
(c)
250
1000
M2 = 1000
40
100
M1=20
-
8/3/2019 Andrew Francis- Modeling the Melt Dispersion Mechanism for Nanoparticle Combustion
37/42
Texas Tech University, Andrew Francis, December 2007
30
Mixing particles that have lowerMvalues with particles that have higherM
values will greatly increase the mixtures flame velocity compared to the flame velocity
of the particles with the higherMvalue. Moore et al (2006) showed that a mixture of Al
particles that consisted of 70% particles withR=38 nm (M=8.3) and 30% particles with
R=2 and 10 m provided essentially the same flame velocity as 100% nanoparticles [15].
Figure 22 shows the relationship between the mixtures with M1=20 taken from Figures 18
through 21.
Figure 22. Relative flame velocity as a function of mass fraction for M1 = 20.
It is evident that as the percentage of smaller particles increases, the flame
velocity approaches the flame velocity of the smaller particles.
0.2 0.4 0.6 0.8 1
0.5
0.6
0.7
0.8
0.9
M1 = 20
M2=40
80
150
1000
Relative
FlameVelocity(V/Vmax)
Mass fraction M1 (c)
-
8/3/2019 Andrew Francis- Modeling the Melt Dispersion Mechanism for Nanoparticle Combustion
38/42
Texas Tech University, Andrew Francis, December 2007
31
CHAPTER IV
CONCLUSIONS
The effects of key parameters on relative flame velocity within the scope of the
melt dispersion mechanism were examined. Flame propagation velocities were
statistically evaluated using probability density functions for particle size, oxide shell
thickness, and oxide shell strength. It was shown that flame propagation velocity
depends sensitively on relative particle size, the relative particle size distribution, and
oxide shell formation temperature. It was also shown to depend upon particle size, oxide
shell thickness, and oxide shell strength, but not as sensitively. Variations in particle size
were shown to have a greater effect on flame velocity than variations in oxide shell
thickness. Mixing smaller nanoparticles with larger nanoparticles showed to increase the
mixtures flame propagation velocity compared to the larger nanoparticles and allowed it
to approach the maximum flame velocity.
The results presented here suggest that better reproducibility of the flame velocity
may be achieved experimentally by selecting a material with a narrow relative particle
size distribution. It also suggests that a combination of increased oxide shell formation
temperature and increased oxide shell strength could be used to maximize the flame
velocity in particles with increased relative particle size.
Research in the field of nanothermites continues to be very promising and
suggests ways to approach the flame propagation velocities of high explosives with the
much safer, less sensitive, and much higher energy producing nanothermites. Continued
research and experimentation is needed to explore the full extent of the melt dispersion
-
8/3/2019 Andrew Francis- Modeling the Melt Dispersion Mechanism for Nanoparticle Combustion
39/42
Texas Tech University, Andrew Francis, December 2007
32
mechanism, optimize these Al nanothermite reactions, and extend this mechanism to
other nanothermites.
-
8/3/2019 Andrew Francis- Modeling the Melt Dispersion Mechanism for Nanoparticle Combustion
40/42
Texas Tech University, Andrew Francis, December 2007
33
REFERENCES
[1] Asay B. W., Son S. F., Busse, J. R., and Oschwald, D. M. Propellants, Explosives, and Pyrotechnics,29, 216 (2004).
[2] Bazyn, T., Krier, H., and Glumac, N. Combustion and Flame,145, 703 (2006).
[3] Bockmon, B. S., Pantoya, M. L., Son, S. F., Asay, B. W., and Mang, J. T.
Journal of Applied Physics, 98, 064903 (2005).
[4] Bruzostowski, T. A. & Glassman, I.,Heterogeneous Combustion,edited by H. G. Wolfhard, I. Glassman, and L. Green, Jr., Academic,
New York, (1964).
[5] Dlott, D. D.,Energetic materials: Initiation, Decomposition and Combustion,Elsevier, New York, 125-192,(2003).
[6] Fischer, S.H., Grubelich, M.C. Theoretical Energy Release of Thermites,
Intermetallics, and Combustible Metals.Proceedings of the 24thInternational
Pyrotechnics Seminar(1998).
[7] Granier, J. J., Combustion characteristics of A1 nanoparticles and
nanocomposite A1+MoO3 thermites.Doctoral dissertation, Texas TechUniversity (2005).
[8] Granier, J. J. and Pantoya, M. L., Combustion and Flame, 138, 373 (2004).
[9] Khler, J., and Meyer, R.,Explosives,Fourth Edition, VCH Publishers, New
York, (1993).
[10] Kuo, K. K.,Principles of Combustion, Second Edition, Hoboken, New Jersey:John Wiley & Sons, 734-779,(2005).
[11] Levitas, V. I.; Asay, B. W.; Son, S. F.; Pantoya, M. L.Journal of Applied Physics,
101, 083524, (2007).
[12] Levitas, V. I.; Asay, B. W.; Son, S. F.; Pantoya, M. L.Applied Physics Letters,89, 071909,(2006).
[13] Levitas, V. I., Pantoya, M. L., and Dikici, B., Melt Dispersion versus Diffusive
Oxidation Mechanism for Aluminum Nanoparticles: Critical Experiments andControling Parameters (in press).
-
8/3/2019 Andrew Francis- Modeling the Melt Dispersion Mechanism for Nanoparticle Combustion
41/42
Texas Tech University, Andrew Francis, December 2007
34
[14] Menikoff, R. and Kober, E., Compaction waves in granular HMX, Tech. Rep.LA-13546-MS, Los Alamos National Lab., (1999).
[15] Moore, K., Combustion behaviors of bimodal aluminum size distributions in
thermites. Master's thesis, Texas Tech University (2005).
[16] Pantoya, M. L. and Granier, J. J.,Propellants, Explosives, and Pyrotechnics30,53 (2005).
[17] Plantier, K. B., Pantoya, M. L., and Gash, A. E., Combustion and Flame, 140, 299(2005).
[18] Rai, A., Lee, D., Park, K., and Zachariah, M.,Journal of Physical Chemistry B108, 14793 (2004).
[19] Sanders, V.E., Busse, J.R., and Son, S.F. Environmentally Responsible
Percussion Primers for Small and Medium Caliber Ammunition.Proceedings ofthe 39thJANNAF Conference, Colorado Springs, CO (Dec. 2003).
[20] Son, S. F., W. C. Danen, W. C., B. S. Jorgensen B. S.,, B. W. Asay, B. W., J. R.
Busse, J. R., and Pantoya, M. L., Defense Applications of Nanomaterials,ACS Symposium Series 891, 227240 (2005).
[21] Wang, L.L., Munir, Z.A., & Maximov, Y.M.. Review thermite reactions:
Their utililzation in the synthesis and processing of materials,Journal ofMaterials Science, 28, 3693-3708, (1993)
[22] Watson, K., Fast Reaction of Nano-Aluminum: A Study on Fluorination versus
Oxidation, Masters Thesis, Texas Tech University (2007).
[23] Zinovev, V. E.,Handbook of Thermophysical Properties of Metals at HighTemperatures,Nova Science, New York, 140 (1996).
-
8/3/2019 Andrew Francis- Modeling the Melt Dispersion Mechanism for Nanoparticle Combustion
42/42
Texas Tech University, Andrew Francis, December 2007
PERMISSION TO COPY
In presenting this thesis in partial fulfillment of the requirements for a masters
degree at Texas Tech University or Texas Tech University Health Sciences Center, I
agree that the Library and my major department shall make it freely available for research
purposes. Permission to copy this thesis for scholarly purposes may be granted by the
Director of the Library or my major professor. It is understood that any copying or
publication of this thesis for financial gain shall not be allowed without my further
written permission and that any user may be liable for copyright infringement.
Agree (Permission is granted.)
Andrew Boone Francis 11/30/2007Student Signature Date
Disagree (Permission is not granted.)
_______________________________________________ _________________
Student Signature Date