My final project focused on the (sort of) combustion reaction of aluminum and gallium as it pertains to powering underwater vehicles.

1. 2.28 Final
Presentation
ADRIANNA RODRIGUEZ
COURSE 2, MIT 2016

2. Problem: Powering AUVs
 Autonomous Underwater Vehicles
 Remus 600: depth of 600-1500 m, max 2.1 m/s, 5 kWh battery, mission length 24
hours
 1/3 of body volume available
 Currently powered by… batteries. No onboard power generation
Image Source: http://www.unmannedsystemstechnology.com/wp-content/uploads/2013/02/Remus-600-S-AUV1.jpg
32.4
cm
3.25 m

3. Can this be done differently?
 Underwater vehicles do not have ready access to air
 Primary use is military purposes, needs to remain stealthy. Minimize any
bubbling or gas production
 Needs to have highest energy density possible
 At 600 m, pressure is 6.13 MPa. At 1500 m, 15.8 MPa.
𝜋𝑟2 𝐿 = 𝜋 0.15 2 ∗ 1 = 0.071 𝑚3 Available for power system
 Essentially unlimited heat sink, ocean is at ~22 C
 Unlimited source of (salt) water
Constraints
However…

4. Aluminum Fuel
 Aluminum reacts with water, oxidation reaction, produces aluminum
oxides and hydrogen gas
 Three possible reactions depending on reaction temperature, all produce
the same amount of hydrogen, just using decreasing quantities of water as
temperature increases
2 𝐴𝑙 + 6 𝐻2 𝑂 → 2 𝐴𝑙(𝑂𝐻)3+3 𝐻2
2 𝐴𝑙 + 4 𝐻2 𝑂 → 2 𝐴𝑙 𝑂(𝑂𝐻) + 3 𝐻2
2 𝐴𝑙 + 2 𝐻2 𝑂 → 2 𝐴𝑙2 𝑂3 + 3 𝐻2
 Very energy dense, highly exothermic
 Aluminum products are non-toxic
 Hydrogen gas can be captured & burned (producing only water) for
additional energy

5.  Aluminum (ex. Aluminum foil, soda can) will not react at room temperature
 Thin layer of alumina (Al2O3) forms on contact with air, prevents water
from reacting
 Need to either prevent alumina layer from forming, or else destroy it
Aluminum Reaction

6. Aluminum Reaction (cont.)
Eutectic (Gallium-Indium)
 Eutectic has a low melting point,
about 6 C, causes separations along
grain boundaries
 In liquid water and at room
temperature, eutectic (immiscible in
water) is liquid, reaction ensues
 Problem: Ga-In is expensive
Inorganic Salt Solution
SEM images of
multicomponent aluminium
alloy after preparation [1]
 Solution creates pitting of alumina
surface
 Highly temperature dependent and
takes longer to achieve good yield
 With very strong solutions, integrity of
the reaction vessel will degrade over
time
Reaction of aluminum in
NaCl and KCl solutions [2]

7. Aluminum Reaction (cont.)
Liquid Aluminum
 Alumina layer does not form on liquid
aluminum (or Al-Ga alloys), even when
exposed to air
 Problem: separating waste products
from unreacted aluminum (and
recyclable gallium) is very difficult
Powders
Schematic of liquid
aluminum reactor [2]
 At very small (μ-n) powder sizes,
aluminum can react at as low as 560-
800 C in atmospheric pressure and
room-temperature water
 Reaction is largely surface area
dependent, so smaller particles react
faster
SEM of 80- nm diameter
aluminum particles [3]

8. How Powdered Aluminum Reacts
 Still a lot of uncertainty about the specific mechanisms by which the
reaction proceeds
 Basic idea: use same-sized particles mixed thoroughly with water, ignite,
react (produces flame)
 High surface area and reactivity, short reaction times, lower ignition
temperatures, highly spherical
SEM of 80- nm dia. aluminum particles [3]
Most detailed literature
focused on 30-100 nm

9. Characterizing the Reaction:
Diameter Dependence
 High dependency on diameter
 Smaller diameter particles have less heat
capacity and heat up faster
 Oxide film might also be able to absorb
more water relative to volume on smaller
particles
Plot of ignition temperature as a function of sample size. The data
plotted was collected by the source author from multiple other
sources. [4]
Particle dia. (nm) Oxide layer thickness
(nm)
mass % active Al
38 3.1 54.3%
50 2.1 68%
80 1.9 81%
130 2.2 84% [3]

10. Difference Between μ- and n-size
 When examining n- versus μ-sized particles, the ignition temperature is
much closer to the melting point of Al for the nano scale, but it’s closer to
alumina melting point at about 1 μm
 Possible idea is that the entire core of the particle is melting, which relaxes
the stresses on/within the alumina boundary layer. For nano-scale
particles:
 Heavy diameter dependence indicates diffusion-controlled burning
 For smaller particles, combustion can be (crudely) modeled like droplet
combustion
 Bburn ~ 1.08, 𝑚 = 41.29 kg/m2s

11.  Diameter of the particles also affected the
packing density in one study (smaller particles
had a greater packing density)
 This has a direct influence on the equivalence
ratio/burn rate
Characterizing the Reaction:
Diameter Dependence
Plot of mass burning per unit area as a function of
particle diameter. The shape implies a D-1 relationship. [5]
Table showing effect of particle diameter on packing density and ER.
[3]

12. Burn Rate: Lab Setting
 Again, diameter effects the equivalence
ratio and the “well-mixed-ness” of the
aluminum fuel within the water
 Reaction propagated faster down the tube
with smaller particles
 On the order of 0.01 m/s for 100-nm
particles around 1 MPa
 Increases to 0.1 m/s for 10 MPa

13. Pressure Dependence
 Experiments showed clear sensitivity to
ambient pressure during reactions
 Modeled as Arrhenius equation, exponent is on
the order of 0.5
 Pressure correlated to burn rate, but not to
reaction completeness
Plot of linear flame front speed versus pressure for 38-nm
particle at stoichiometric mix. [3]
Reaction completeness
plotted as a function of
pressure showed
weak/no correlation.
Efficiencies were as low
as 85% and generally
around 95% [5]

14. Pressure Dependence
 Increased pressure could help the
water diffuse through the alumina
surface, which would instigate surface
reactions at the Al core
 Increased pressure leads to an
increase in flame temperature,
increased flame temperature increases
burn rate
 Increased flame temperature would
explain brighter Al2O3 flame

15. Is There an Accurate Model?
 Reaction is probably controlled by a combination of diffusion and kinetics
 Even in laboratory testing, powder diameters could have enough variation
that some particles would be diffusion-controlled and some would be
more first order
 Literature is particularly inconclusive on a good model for alumina
behavior (diffusion through the layer, internal forces leading to breaks, etc)

16. Powering the Remus 600
 Operates at 600-1500 m water depth, meaning 6.13-15.18 MPa of water
pressure
 Based on this pressure, for 38-nm fuel particles, the flame speed is between
0.11 -0.16 m/s
 Remus can travel as fast as 4 knots (2.06 m/s)
 Assume that Remus is permitted to cruise for some period of time at 1500 m
depth and at 0.16 m/s
 Reaction is very efficient, powders finish combusting almost instantly
0.16
m/s

17. Powering the Remus 600
 5 kWh/24h = 210 W = 210 J/s power demand
 ~0.07 m3 of free space, this can fit 191 kg Al fuel powder. For 38-nm
particles, only 54% of this is usable aluminum (the rest is the alumina film)
– 103 kg Al fuel
 ΔhR,f = 48.41 MJ/kg * 103 kg = 4989 MJ potential energy
 4989 MJ/210 J/s = >5000 hours of electricity
 At 3m length, if a reaction tube is designed to span the length of the
Remus, all fuel will be fully reacted upon expulsion.
Currently, the Remus has a 5 kWh battery and a 24 hour
mission lifetime.

18. Sources
1. http://www.sciencedirect.com/science/article/pii/S0925838805000265
2. https://www1.eere.energy.gov/hydrogenandfuelcells/pdfs/aluminum_wat
er_hydrogen.pdf
3. http://www.sciencedirect.com/science/article/pii/S1540748906003191
4. http://onlinelibrary.wiley.com/doi/10.1002/prep.200400083/epdf
5. http://www.tandfonline.com/doi/pdf/10.1080/00102200802414873