Aluminothermic Processes 1

Aluminothermic Processes
In this keyword, only the general aspects of aluminothermic processes are discussed: The technical
production of the individual metals and alloys is described in the corresponding keywords, e.g.:
ferroboron → Boron and Boron Alloys; chromium and ferrochromium → Chromium and Chromium Alloys;
ferromolybdenum → Molybdenum and Molybdenum Compounds; ferroniobium → Niobium and Niobium
Compounds; ferrotitanium → Titanium and Titanium Alloys; ferrotungsten → Tungsten, Tungsten Alloys, and
Tungsten Compounds; ferrovanadium → Vanadium and Vanadium Compounds
Fred W. Hall, formerly London & Scandinavian Metallurgical Co. Ltd., Rotherham, United Kingdom

1. Introduction . . . . . . . . . . . . ..... 1 3.1. Pot Design and Construction . . .... 8

2. Basic Metallurgy . . . . . . . . . ..... 2 3.2. Charge and Charge Preparation ... 9
2.1. Basic Thermochemistry . . . . ..... 2 3.3. Feeding . . . . . . . . . . . . . . . . .... 9
2.2. Fundamental Reactions . . . . ..... 4 3.4. Further Processing . . . . . . . . .... 10
2.3. Auxiliary Reactions . . . . . . . ..... 4
3.5. Safety Precautions . . . . . . . . . .... 10
2.4. Fluxes . . . . . . . . . . . . . . . . ..... 5
2.5. Formation of Intermetallic Com- 4. Other Uses of Aluminothermic Pro-
pounds . . . . . . . . . . . . . . . ..... 5 cesses . . . . . . . . . . . . . . . . . .... 10
2.6. Thermochemistry in Practice ..... 6 5. Possible Further Developments .... 10
3. Commercial Production . . . . ..... 7 6. References . . . . . . . . . . . . . . .... 11

1. Introduction in the steel and superalloy industries. Because

The usual method of reduction by carbon is in-
applicable to the reduction of refractory oxides
and ores. The products commonly covered un-
der this title are usually required to be low in
carbon and they are avid carbide formers. Also,
no conventional refractories would stand up to
the long reduction times at the high tempera-
tures required. For such products, advantage is
taken of the high temperatures achieved and the
rapidity of metallothermic reactions using reac-
tive metals, notably Al.
Early examples of such reactions are
Berzelius’ reduction of K2 TaF7 by Na and
Wöhler’s reduction of AlCl3 by Na, in 1825
and 1828, respectively. However, the father
of aluminothermic processes was undoubtedly
Goldschmidt who, in 1898, described the pro-
duction of low-carbon, high-melting-point met-
als, without extraneous heat, by feeding an
exothermic mix into an already ignited first por-
tion [1].
The term “aluminothermic processes” can
cover a wide field. For the present purposes it
is taken, as is generally understood, to be the
manufacture, by Al reduction of refractory ox-
ides or ores, of metals and alloys mainly used
some important alloys are produced aluminosil-
icothermically side by side with those produced
aluminothermically, they are included in the sur-
vey.
The main alloys and metals under discussion
are Cr metal [7440-47-3], FeB [11108-67-1],
FeMo [11121-95-2], FeNb [11108-69-3], FeTi
[12719-90-3], FeV [12604-58-9] and FeW
[12604-57-8]. There are many minor variations,
such as CrC, where superior purity or the ab-
sence of Fe is required for alloys used in super-
alloy production, and FeNb from Nb2 O5 instead
of ore, where extra purity is required, again for
superalloys.
Aluminothermic Mn [7439-96-5], formerly
of importance, has been superseded by elec-
trolytic Mn, and aluminothermic production of
FeTi has largely given way to induction furnace
melting of Fe and Ti scraps. The latter method
also now substitutes for the former, dangerous,
highly exothermic production of 55 – 60 % TiAl
and 55 – 60 % ZrAl. When possible, combined
electroaluminothermic methods can be used.
Because most basic reactions require extra
heat this is provided by Al and a vigorous oxi-
dant. Aluminum is an expensive fuel. At an Al

c 2005 Wiley-VCH Verlag GmbH & Co. KGaA, Weinheim

10.1002/14356007.a01 447
2 Aluminothermic Processes
powder price of 1400 $/t and a small-scale elec-
tricity cost of 4.9 C/kW h the heating effect of
Al combustion costs about three times that of
electricity.
Although a minor proportion of production,
some FeB, FeNb, FeTi, and FeV is produced in
the arc furnace. In view of the small tonnages
involved, such productions usually occur only
where an arc furnace exists for other purposes.
Apart from the necessity of metallothermic
reduction advantages of the process as compared
to conventional reduction processes are:
– Very rapid reaction with less heat loss by ra-
diation and convection
– Much less gas volume
– Easy accomodation of diverse productions
– Small plant investment (important, consider-
ing the small production).
A disadvantage of metallothermic production
is that no refining of the metal is possible. It is
important to leave the slag in situ to allow metal
drops to settle into the regulus, so that one is
faced with a large depth of slag quickly crusting
over. Alternatively, if the slag is tapped, requir-
ing subsequent ore dressing processes to recover
included metal, the metal regulus quickly be-
comes sluggish. Therefore, the composition of
the mix must be regulated very carefully to pro-
duce the optimum combination of oxygen and
reactant metal levels in the product, and all in-
gredients must be low in harmful impurities. Ob-
vious ones are C, S, and P. Other impurities in
the various ores used for different products are
As, Sn, Pb, Sb, and Cu. Nitrogen is not usually
a problem because the metal regulus formed is
protected by the slag layer. Chromium metal,
used in superalloys and electric heating ele-
ments, is an extreme case, where, apart from low
impurities generally, certain impurities, such as
Ga, Pb, and B, must be kept down to a few parts
per million.
The worldwide – excluding the former USSR
and China – production of metal and alloys by
aluminothermic processes is shown in Table 1.
Table 1. Production of metal and alloys by aluminothermic processes
t/a Expressed as
Cr 4550 Cr
FeB 1350 18 % B
FeMo ∗ 40500 65 % Mo
FeNb 16300 65 % Nb
FeV 25000 80 % V
FeW 4000 80 % W
∗ The FeMo represents about one third of the total Mo used in the
steel industry; the remaining two thirds are used as MoO3 .
2. Basic Metallurgy
A metallothermic reduction of a metal com-
pound is possible when the reductant metal has
a greater affinity for the nonmetal element of
the compound than the desired metal. In various
branches of metallurgy the nonmetal may be a
halide, sulfur, or oxygen. The reductant metal is
usually Al, Si, Mg, Ca, or Na. For instance, Ti
metal is produced by reduction of halides with
Na or Mg and V metal by reduction of V2 O5
with Ca in a bomb reaction (to prevent vapor-
ization of the Ca). Interesting examples are the
reduction of NiO and MoO3 by Fe in the arc fur-
nace. Aluminothermic processes, as previously
defined, depend upon the high affinity of Al for
oxygen as compared to that of many other met-
als, an example being:
Cr2 O3 + 2 Al → 2 Cr + Al2 O3
2.1. Basic Thermochemistry
Aluminum has a high affinity for oxygen, and
the ease of reduction of a metal oxide depends
on the difference in oxygen affinity of Al and the
metal. The affinity for oxygen is indicated by its
heat of formation, ∆H, expressed for compar-
ison purposes as ∆H per mol/oxygen (kJ/mol
O). Table 2 gives a selection of ∆H values at the
standard temperature of 298 K.
 Aluminothermic Processes 3
Table 2. Heats of formation of oxides [2]
which is related to the heat of formation by the
Oxide -∆H 298 , -∆H 298 , formula:
J/mol oxide kJ/mol O
∆G0T = ∆H − T ∆S
CaO 634.7
634.7
MgO 601.7 where S is the entropy and is always zero or
601.7
Al2 O3 559.4 positive because it corresponds with the increas-
1678.5 ing atomic disorder of the system with tempera-
ZrO2 550.6
1101.5 ture. Figure 1 shows the free energy of formation
TiO2 945.4 472.7 of the oxides of interest at high temperatures.
SiO2 911.0 455.5
B2 O 3 424.1
All oxides become less stable as temperature in-
1272.8 creases. The more negative the free energy of
Na2 O 415.3 415.3
Ta2 O5 409.5
formation at a given temperature the greater the
2047.3 tendency for that metal to reduce an oxide of less
Nb2 O5 380.2
1900.8 negative free energy.
Cr2 O3 376.8 The free energies of reaction of oxides with
1130.4
V2 O5 310.2 Al and Si are given in Figures 2 and 3.
1551.6
WO3 843.2 280.9
Fe3 O4 279.3
1117.5
Fe2 O3 821.9 273.8
FeO 264.6 264.6
MoO3 745.7 248.7

Figure 2. Change in free energy of reaction in the reduction

of oxides by aluminum

Although the free energy of reaction is a

Figure 1. Change in free energy of oxide formation reac-
tions [3]
A quantitative measure for the stability of an
oxide is the Gibbs free energy of formation, ∆G,
quantitative measure, not all data necessary for
such complex reactions as commercial metal-
lothermic reactions are readily available and it
is usual for the practicing metallurgist to rely
on heats of formation for calculation of heats of
reaction.
4 Aluminothermic Processes

A guide for a successful aluminothermic re-

action, permitting adequate separation of metal
and slag, is a -∆H 298 higher than 300 kJ/mol [4].
This is confirmed in practice: additional heat is
required for Cr production, whereas the reac-
tion producing 80 % FeV is so vigorous that it
requires cooling. This guide does not apply to
silicothermic reactions.

2.3. Auxiliary Reactions

Many reactions require additional heat to pro-
duce fully liquid metal and slag. Except for pro-
ducing pure Cr, the first step is to make an addi-
tion of iron oxide plus reactant, giving a higher
average heat of reaction. This also has the ad-
vantage of producing significantly lower melt-
ing points (Table 4).
◦
Table 4. Melting points of some metals (in C) and of alloys of
Figure 3. Change in free energy of reaction in the reduction these metal with iron
of oxides by silicon V 1 902 80 % FeV 1 770
B 2 180 16 % FeB 1 550
Nb 2 468 65 % FeNb 1 560
Mo 2 620 65 % FeMo 1 900
W 3 410 80 % FeW 2 500
2.2. Fundamental Reactions
Although Ca and Mg may be the most suitable In silicothermic reactions, a proportion of Al
reactants, they are not used in the reductions un- is usually used to increase the exothermicity par-
der review. Their basic costs are high compared ticularly, because the products, e.g., FeW and
to Al and Si and 16 parts by weight O require FeMo, have such high melting points.
40 parts Ca, 24 parts Mg, 18 parts Al, or 14 parts Because the alloy specifications have become
Si. Furthermore, the boiling points of Ca and traditional, although not necessarily ideal, there
Mg are 1 484 ◦ C and 1 090 ◦ C, respectively. The is a limit to the amount of iron oxide and Al
heats of reaction of the oxides of interest with to be added, and, indeed, it may be more eco-
Al and Si are given in Table 3. nomical to increase the heat content of the mix-
Table 3. Heats of reaction per mole reductant
ture furhter, by using highly oxidic compounds
plus Al. The reactions and the heats of reaction
-∆H 298 , kJ (in kJ/mol Al) for the additives most commonly
3/4 TiO2 + Al → 3/4 Ti + 1/2 Al2 O3 130.2 used are:
3/4 SiO2 + Al → 3/4 Si + 1/2 Al2 O3 156.0
1/2 B2 O3 + Al → B + 1/2 Al2 O3 202.6 K2 Cr2 O7 + 4 Al → 2 Cr + 2 Al2 O3 + K2 O
3/10 Nb2 O5 + Al → 3/5 Nb + 1/2 Al2 O3 268.8
1/2 Cr2 O3 + Al → Cr + 1/2 Al2 O3 273.8 ∆H 298 = − 422.0 kJ/mol Al
3/10 V2 O5 + Al → 3/5 V + 1/2 Al2 O3 373.9
1/1 Wo3 + Al → 1/2 W + 1/2 Al2 O3 417.4
3/8 Fe3 O4 + Al → 9/8 Fe + 1/2 Al2 O3 420.2 3 BaO2 + 2 Al → 3 BaO · Al2 O3
1/2 Fe2 O3 + Al → Fe + 1/2 Al2 O3 428.3 ∆H 298 = - 812.0 kJ/mol Al
3/2 FeO + Al → 3/2 Fe + 1/2 Al2 O3 442.1
1/2 MoO3 + Al → 1/2Mo + 1/2 Al2 O3 466.4
2/3 WO3 + Si → 2/3 W + SiO2 348.8 NaClO3 + 2 Al → Al2 O3 + NaCl
1/2 Fe3 O4 + Si → 2/3 Fe + SiO2 352.1
2/3 Fe2 O3 + Si → 4/3 Fe + SiO2 363.0 ∆H 298 = - 869.2 kJ/mol Al
2 FeO + Si → 2 Fe + SiO2 381.8
2/3 MoO3 + Si → 2/3 Mo + SiO2 414.1 3 KClO4 + 8 Al → 4 Al2 O3 + 3 KCl
∆H 298 = - 839.0 kJ/mol Al
 Aluminothermic Processes 5

slag, so as to permit good separation of slag

Figure 4. Melting point curves of binary alumina slags
In some cases additional heat was applied
formerly extraneously by preheating the charge.
This has been given up because of practical diffi-
culties of heating a mixture of low heat conduc-
tivity and the danger of premature ignition, with
loss of the charge and damage to the preheating
equipment. The cheap nitrates are not used now
because products low in nitrogen are required
and the metals in question are avid nitride for-
mers. BaO2 is useful because it not only acts as
an oxidant, but is a strong base and assists the
progress of the reaction by combining with the
Al2 O3 produced. However, its high molecular
mass and its one available oxygen atom make it
a very expensive oxidant, and its use is confined
to highly exothermic mixes, e.g., for TiAl and
ZrAl, where it helps to smooth the reaction. The
common oxidant is NaClO3 , but KClO4 is to be
preferred because it is more stable and NaClO3
is hygroscopic. In Cr mixes, K2 Cr2 O7 is used
and is preferred to CrO3 on health grounds as
well as expense.
and metal. In the case of aluminosilico reactions,
residual iron oxide and Al2 O3 are already good
fluxes, and lime also is used because the strongly
basic CaO, apart from fluxing, lowers the activ-
ity of Al2 O3 and SiO2 .
In aluminothermic reactions residual oxides
may play a part, the compound Al2 O3 · TiO2
having a melting point of 1590 ◦ C; Na2 O and
K2 O, not completely vaporized, also assist. The
usual flux is CaO which has a pronounced effect
on the melting point of Al2 O3 . It is not used
in Cr manufacture as it spoils the slag for use
in refractories and calcium chromite is a stable
compound.
Figure 4 shows the effect on melting point of
additions of several oxides to Al2 O3 .
The energy required to fuse the flux is par-
tially compensated by the heats of formation of
the compound oxide (Table 5).
Table 5. Heats of formation of compound oxides
Compound −∆H 298 , kJ/mol
Al2 O3 · SiO2 192.6
2Al2 O3 · B2 O3 69.5
BaO · Al2 O3 100.5
CaO · Al2 O3 15.1
CaO · SiO2 89.2
2.5. Formation of Intermetallic
Compounds
The heat of reaction of the mixture can be in-
creased by the formation of intermetallic com-
pounds. Table 6 gives some examples. However,
this is restricted in that most of the higher heats
of formation belong to unwanted compounds be-
cause the usual aim is to keep residual elements,
such as Al and Si, to a minimum. Of the alloys
considered it is only in the case of FeTi and to a
lesser extent in that of FeB that aluminides have
some significance.
Table 6. Heats of formation of intermetallic compounds

Compound ∆H 298 , kJ/mol

2.4. Fluxes FeTi 40.6

FeB 29.3
FeAl3 112.2
Both Al2 O3 and SiO2 have high melting points, FeSi 77.0
2050 ◦ C and 1722 ◦ C, respectively. The aim in TiAl 80.8
CaSi 150.7
metallothermic reactions is to produce a fluid AlB 67.0
6 Aluminothermic Processes
All reactions are equilibria. Commercial al-
loys must be low in oxygen and most alloys can
be produced at 1 % Al or Si without appreciable
oxygen content. This is not the case with FeTi,
where the equilibrium cannot be driven further
than Al levels of 4 – 6 %.
2.6. Thermochemistry in Practice
A commercial metallothermic reaction must
provide sufficient heat to drive the reaction as
far to the right side of the equation as possible,
plus that heat required to bring the products to
a temperature sufficiently high to permit sepa-
ration of metal and slag and to provide for heat
losses. The amount of total heat must be the min-
imum required, because heat is expensive and
because excess heat supplied to an exothermic
reaction moves the equilibrium in the undesired
direction.
Heats of formation have not always been
determined at the high operating temperatures.
Working on heats of formation at 298 K requires
knowing the specific heats, latent heats, and, in
charges employing sodium salts and chlorates,
the heats of vaporization. Slags can be complex
so that little idea can be had of their heats of
formation. The amount of heat losses cannot
be forecast and varies with the size of charge.
Losses can be established and compensated for
only by trial and error.
Therefore, available information is used as
much as possible, accompanied by intelligent
extrapolations of such information. Final adjust-
ments to the charge are a matter of trial and error.
A very simple example is:
Cr2 O3 + 2 Al → 2 Cr + Al2 O3 ∆H 298 = − 548 kJ
Heat required for products at 2000 ◦ C:
Al2 O3 H 2273 −H 298 356 kJ
2 Cr H2273 −H 298 184 kJ
amounts 540 kJ
to:
According to this, there is a heat balance, so
there should be a satisfactory result. However,
it is found in practice that 125 kJ extra heat are
required, provided by Al plus K2 Cr2 O7 and/or
KClO4 . This 125 kJ (for a certain size of charge)
is a measure of the heat losses. More would be
required for small charges and less for bigger
charges. It is interesting that this conforms with
the guide.
Chemical composition of ores is some guide
to improving the heat balance. Generally the
higher the state of the oxide in the ore, the higher
the heat of reaction with Al or Si. A classical
example was that of Mn ore where the reac-
tion 3 MnO + 2 Al → 3 Mn + Al2 O3 did not sup-
ply sufficient heat. A preoxidation to Mn3 O4
produced a satisfactory result. Preoxidation also
is found beneficial to some ilmenites.
There are many other factors affecting the fine
control of an exothermic reaction, and it is not
always economic to indulge in laboratory con-
trol of all of them. They mainly affect the speed
of the reaction, which should be an optimum to
minimize heat losses but not to create excess heat
with its adverse effect on the equilibrium.
Granulometry. If the charge is to be reacted
in bulk then the ideal would be a charge of
concrete-type granulometry, giving maximum
contact and minimum air inclusion. However,
to provide effective heat control, charges usually
are fed. There should be an optimum particle size
of the ingredients, therefore, coarse enough to
minimize dust losses but fine enough to give an
adequate speed of reaction. Aluminum powder
can be controlled, because suppliers will supply
within reasonable size ranges. In some cases,
such as the highly exothermic FeV reaction, Al
pellets and turnings can be used and they are
used also to some extent in the very competi-
tive production of the cheap alloy FeTi. These
are crude raw materials and different parcels re-
quire slight adjustments to the mix. Ores are a
different matter. These must usually be used as
they arise; the coarser they are in any one pro-
duction, the more the amount of heat required.
Where ores are extremely fine, leading to high
dust losses, it may be economical to pelletize.
Mineralogic Structure of Ores. Complex
ores, such as pyrochlore, can vary widely in
structure, and different parcels require adjust-
ments to the mix.
Bulk Density. This is related to granulome-
try, but even in Cr2 O3 , which would be expected
to be uniform from parcel to parcel and where
grain sizes are of a few microns, variations in

the supplier’s final calcination temperature can
affect the bulk density greatly and hence the re-
covery.
Mass of Charge. The percentage of heat
losses varies inversely with the size of charge.
These losses cannot be measured, but when the
mass of a charge is increased, less heat per unit
charge is required; this decrease can be obtained
by an intelligent estimate finalized by trial and
error.
Rate of Feeding. This is very much an art
and efficient feeding can minimize some of the
above mentioned difficulties. The general aim
is to maintain a cover of unreacted mix on the
surface of the liquid slag, so as to minimize radi-
ation losses but not to feed too fast and produce
too hot a reaction.
It can be seen from the above that whereas
thermochemistry plays a vital role in provid-
ing the basic knowledge for any metallothermic
production, there are other “fine tuning” adjust-
ments that are a matter of experience.
3. Commercial Production
There are two methods of promoting the reac-
tion. One is to fill the reaction vessel with the
charge and ignite it via an easily ignitable mix-
ture, such as BaO2 or Na2 O2 plus aluminum;
the reaction then proceeds from top to bottom
of the charge. This is a wasteful method because
only approximately one third of the capacity of
the reaction vessel is used. The mix calculation
must be very accurate because no control of the
speed of reaction can be exercised. Because the
charge is always covered by liquid slag, dust
losses should be low, but there can be consid-
erable loss by splashing owing to the violence
of the reaction. The short reaction time should
mean low heat losses, but this advantage is coun-
teracted by the intense radiation from an always
liquid surface. This method is seldom used, but
there are some reactions that are so exothermic
that they are dangerous and practically impossi-
ble to feed, such as those for the minor products
55 – 60 % TiAl and ZrAl. This “firing down”
method is used sometimes also for FeW produc-
tion, where the extra heat supplied to achieve the
Aluminothermic Processes 7
high melting point requires a highly exothermic
reaction near the limit of feedability.
The second method, almost universal, is to
use a charge slightly more exothermic than that
for “firing down,” to ignite a small portion in
the reaction vessel, and to feed the remaining
charge. By doing so, the full capacity of the ves-
sel is used so that vessel preparation costs per
unit production are approx. one third those of the
“firing-down” method. Another great advantage
is that the speed of the reaction can be controlled
by varying the feeding rate so that it gets neither
too hot nor too cold. As mentioned previously,
this is quite an art, but it is surprising how soon
a skilled worker becomes used to visual control
of the temperature of the reaction.
There are also two types of reaction vessel.
One is a mobile steel pot, lined with refractory.
The other is a sand bed into which cylindrical
holes are dug and lined with firebrick. This lat-
ter method usually is confined to aluminosilico
reactions, such as those for FeW and FeMo.
Pot reactions are carried out in firing cham-
bers, where the under pressure, as compared to
that outside, is kept to the minimum required
for complete exhaust of fume consistent with
the charge feeder’s need to observe the reaction
but to keep dust losses to a minimum. Sand bed
reactions are usually under a movable hood con-
nected to the main flue. Whereas tall chimneys
were once the solution to disposal of fume, mod-
ern conceptions of pollution control demand ef-
ficient dust collection. Owing to the volatility
of MoO3 , this production demands its separate
dust collector, the recovery from which makes
the difference between profit and loss on the op-
eration. For other productions, because of the
small tonnage of individual products and be-
cause the fume is largely Al2 O3 , CaO, NaOH,
etc., with low contents of valuable oxide, a com-
munal collector is used. Collectors are of many
types, but the most efficient is the reverse-jet fil-
ter. Because of the presence of CaO, NaOH, etc.,
the filter, when not in use, must always be kept
above 100 ◦ C to prevent caking of the fume on
the bags.
Figure 5 shows the arrangement, in principle,
for feeding, reaction in a chamber, and fume ex-
traction. After firing, the reaction pots must be
left to cool to allow final settling out of metal
droplets and eventual solidification. Depending
on the size of the reaction, solidification can take
8 Aluminothermic Processes
a long time, but 24 h, which often fits in with the
overall routine of the production, is usually suf-
ficient. The pot is then stripped, leaving a block
of metal with adherent block of slag. The whole
is then carefully quenched in a water tank, which
causes the slag block to separate and embrittles
the metal to some extent, facilitating subsequent
crushing.
Figure 5. Reaction chamber with feeding hopper, reaction
vessel, and dust collector
a) Feeding hopper; b) Feed-regulating valve; c) Feeding
chute; d) Separable refractory-lined steel plate cylinder;
e) Refractory-lined base; f) Optional magnesite inner lining;
g) Firebrick-lined chamber; h) Filter; i) Offtake to chimney
The sides of the metal block are surrounded
by an adherent mixture of metal and slag created
by penetration into the refractory lining. This
easily breaks away and is consumed in the next
reaction. The top surface, containing pockets of
very adherent slag, requires removal by auto-
matic hammers and, in the last resort, by hand
chipping hammers. The bottom of the block usu-
ally requires less attention. Final cleaning of the
whole block is by shot blasting.
The block is coarsely broken up by a mobile
hammer of the pile driver type and crushed to de-
sired size in jaw crushers and/or rotary crushers
with intermediate screenings. There is immense
variety in customer’s size requirements, ranging
from, say, 15-cm lumps for bulk users down to
graded powders for welding rod manufacturers.
An increasing demand is for bags of 6 mm and
down alloy, each containing a fixed weight of
the desired alloying element.
3.1. Pot Design and Construction
Although there is theoretically no limit to the
size of the charge, as compared to steel produc-
tion, the weights are small for the following rea-
sons:
Costs. Overall, raw materials are expensive.
It is the aim of a producer to try to keep pace
evenly with sales, holding modest stocks of
raw materials and finished product. The to-
tal annual production of any one producer is
small and diverse, with, in some cases, several
specifications for one product.
Handling Facilities. The total weight of a
refractory-lined pot full of metal and slag is
about three times that of the metal produced.
Crane size is limited by cost and the small
building required for carrying out the reac-
tions, and for total operation lift trucks are
convenient.
Preliminary Metal Breaking. The small ton-
nages involved preclude a very expensive,
large, preliminary breaking plant. Some al-
loys, particularly FeMo, are very tough and
metal block thickness is usually limited to
30 – 35 cm.
Metal Specification. Some residual specifica-
tions are very tight, notably that for Al in Cr
metal. A mishap in the weighing of the charge
can produce an off-grade block that may be in
stock for a long time.
Refractory Failure. Although rare, there can
be an occasional failure of the refractory, pro-
ducing breakout of metal and slag at high eco-
nomic loss.
Therefore, metal block weights usually vary
between 1000 and 2500 kg and, depending on
the density of metal and slag, two pot sizes suf-
fice. A small pot of ca. 200 kg metal capac-
ity is useful for intermediate size investigations
on new raw materials. A typical pot covering
1000 – 1750 kg metal would have internal di-
mensions of ca. 100 cm diameter and 125 cm
height and one for 2500 kg would have the same
height but be of 120 cm diameter in order to
avoid too thick a metal block.
The outer construction consists of a plate
steel cylinder attached to a heavy steel base by
quick-acting lugs. The base may be on wheels
or adapted for lift truck operation. The lining

is of moistened, crushed slag from chromium
or ferrovanadium production, because this is a
reasonably pure corundum. To ease cleaning of
the metal block, magnesite tiles may be used in
the metal zone, particularly for the hotter reac-
tions. After preparation of the refractory base, ei-
ther from bricks or ramming of crushed slag, the
sides are rammed around a tapered, withdraw-
able template. Ramming is by hand, by pneu-
matic or electric hammer, or by jolting on a
foundy jolt molding machine. The lining must
then be dried for several hours to remove mois-
ture. It is taken up to a dull red heat.
Basic sand bed operation is to construct a
well in the sand bed from unmortared firebricks
which takes the whole molten charge. After suf-
ficient time is allowed for metal to settle out, a
channel is made through the sand at the slag-
metal interface level and the slap tapped off. A
variation is to use a similar well in the sand bed
sufficient to accomodate only the metal, topped
by a firebrick-lined steel shell to accommodate
the slag, with ample sand seal at the junction. Af-
ter the reaction, the slag is tapped at the junction
and the shell used several times.
3.2. Charge and Charge Preparation
Ideally, there would be an optimum size for raw
materials, fine enough for a successful reaction
but as coarse as possible to minimize dust losses.
Fortunately, most oxides and ores are of reason-
able size in these respects. Wolfram ores may
have to be ground sometimes and it is known
that very fine ores, such as pyrochlore, have been
pelletized. Chromium oxide, at a grain size of a
few µm, can be expected to give high dust losses
but the grains have a strong tendency to self-
adherence so that the particles are much larger.
Even so, there is a slightly noticeable difference
in recovery when different oxides are used, this
varying directly with the final calcining of the
oxide during its manufacture. The calcination
increases grain size and density.
Auxiliary oxidants are of satisfactory grain
size with the proviso that NaClO3 tends to ag-
glomerate on storage and must be sieved down
to max. 1 mm before use.
The minimum size of Al is limited by dust
losses and the risk of dust explosions. It should
not be below ca. BSS 240 = 64 µm. The size used
Aluminothermic Processes 9
in exothermic reactions then depends on the re-
action. For chromium it is on the order of aver-
age 0.25 mm, for FeB and FeNb, 0.45 mm, and
the Al powder can be partially substituted by
cheaper foil powder. In the highly exothermic
FeV reaction, where the V2 O5 is in the form
of flakes or lumps, only a little coarse powder is
used to assist the start of the reaction, the bulk be-
ing pellets, chopped wire, and/or pure turnings.
Despite the fine size of ilmenite and rutile, the
same applies to FeTi production. Owing to the
relative cheapness of the ores, Al is the most ex-
pensive ingredient, and cheap aluminum is used
at the expense of Ti recovery.
In aluminosilico reductions only powders are
used, in the case of FeMo because the ore is fine,
in the case of FeW to obtain maximum speed of
reaction so as to achieve the high temperature re-
quired. The reductants are Al powder and 75 %
FeSi, on the order of 0.25 mm average.
The amount of flux used is kept to a minimum
in aluminothermic charges. Where sodium salts
are used as auxiliary oxidants, residual Na2 O
already lowers the melting point of the slag. A
calcium salt is the usual flux because it has such
a significant effect on the melting point of Al2 O3
(see Fig. 4). Instead of CaO, CaF2 often is used
as its low melting point of 1418 ◦ C assists in
promoting a smooth reaction. The amount used
is often 1 – 2 % of the total charge, never more
than 5 %.
Significant quantities of CaO – CaF2 are used
in aluminosilico thermic reactions because it is
important for their success to tie the SiO2 pro-
duced. As can be seen from Table 4, CaO has a
strong affinity for SiO2 .
After the components have been weighed out
accurately, they are mixed completely in a gen-
tly acting mixer, such as a drum with lifters, V
blender, or rotating cube. Duration of mixing can
be determined only by practice and can be de-
termined initially by analyzing samples taken at
intervals.
The charge is then dropped into a feeding
hopper.
3.3. Feeding
When a charge has been prepared accurately to
produce the desired result, success depends on
the efficiency of feeding. The usual method is to
10 Aluminothermic Processes
control the exit valve of the feeding hopper ac-
cording to the visual observation of the speed of
the reaction, making slight adjustments to com-
pensate for momentary coldness or hotness. Ad-
ditional finer control is often applied by means
of a handrake in the feeding chute. A fast reac-
tion, such as that for FeV, often requires cooling,
and a second small hopper may be used to feed
a portion of crushed FeV slag. A FeW reaction
is fed as fast as possible.
3.4. Further Processing
Pot stripping, block cleaning, and crushing re-
quire no further description than described in the
introduction to Chapter 3, except to emphasize
the importance of avoiding cross contamination.
This is particularly the case with FeB. A small
amount of B is very beneficial to a steel designed
for it but an unknown amount can be disastrous.
3.5. Safety Precautions
It must be realized that exothermic production,
inadequately supervised, can be a very danger-
ous operation. Aluminum powder and oxidants
are themselves fire hazards, and risk is magni-
fied, when they are in contact, maybe to explo-
sive proportions.
Aluminum powder and oxidants must be
stored separately and the latter kept clear of car-
bonaceous material. Only sufficient for a single
charge should be brought to the mixing depart-
ment, and, during weighing, the oxide or ore
should be sandwiched between the Al powder
and the auxiliary oxidant to avoid direct, con-
centrated contact.
Mixing plants should be grounded to avoid
buildup of static electricity. During firing, over-
feeding can cause violent eruptions. Lumpy
NaClO3 can cause explosions. In case of an inad-
vertent lining failure, a large enough well should
be provided to accept all the charge.
The main causes of disastrous explosions are
Al dust explosions, either primary, because of
careless handling, or secondary, over an overfed
reaction. It also has been known for the metal
and slag, apparently solid, to be quenched while
still liquid internally, creating a disastrous ex-
plosion.
Employees must be protected from inhala-
tion of dust and physical contact with charge
components. This is particularly the case with
hexavalent compounds of Cr, which can cause
the well-known chrome ulcers.
Housekeeping must be exceptional. Fine dust
accumulated on roof beams has been known to
combust spontaneously.
4. Other Uses of Aluminothermic
Processes
A significant use of an aluminothermic process
is the welding of railway and tramway line joints,
where the reaction takes place in a small cru-
cible, and the steel produced is tapped onto the
joint. In a similar manner, heavy cracked steel
components, such as rolls, are welded and worn
parts are built up and remachined. Developments
over the last few years are the repair of ingot
mold stools, where the steel stream has worn a
cavity, and of cracked ingot mold lugs.
The use of exothermic reactions in incendiary
bombs, where the intense Fe2 O3 – Al reaction
ignites the magnesium casing, is well known.
Other military uses are for flares, which in the
case of sea rescue give off colored smokes.
Milder reactions, damped down by insulating
materials, are used in feeder head compounds,
which are spread over the surface of liquid steel
ingots in order to conserve the heat and minimize
piping. For steel castings, similar compositions
are formed into sleeves inserted in the sand mold
to insure liquidity over the whole of the pouring
time. Where large amounts of alloying addition
are made to steel in the ladle, the cooling effect
of the addition may be counteracted by the use
of exothermic briquettes composed of alloy, Al,
and oxidant. Minor uses are for jointing of Al
and Cu cables by a CuO – Al reaction, and as
slow reactions for hand warmers and food cans
for explorers.
5. Possible Further Developments
Wherever alloys with low content are required,
aluminothermic processes are necessary, and
they have been brought to such an efficiency
that little further improvement can be foreseen.

Improved refractories may permit further trans-
fer of some reactions to the arc furnace, saving
Al plus auxiliary oxidant, but at present, those
which can be envisaged possibly to stand up
to the longer exposure are very expensive and
have to compete with the self-produced refrac-
tory from an exothermic reaction.
Plasma processes are a possibility, the
very high temperatures encouraging the oxide–
carbon reaction, but up to now there has been no
success in driving the reaction to low C levels.
Here again, refractories are a problem.
A relaxation of maximum impurity levels
would be of mutual economic benefit to alloy
maker and steel maker. Stringent specifications
for superalloys are obvious, but when the small
proportion of alloy used in many steels and the
wealth of purification methods used in bulk steel
production are considered, then some relaxation
in impurity levels should be possible, enabling
use of cheaper, more impure ores. Higher Al
and/or Si levels would give better recovery and,
Aluminothermic Processes 11
where present specifications are very low, lower
oxygen levels.
6. References
1. H. Goldschmidt, Justus Liebigs Ann. Chem.
301 (1898) 19 – 28.
2. O. Kubaschewski, C. B. Alcock: Metallurgical
Thermochemistry, 5th ed., Pergamon Press,
London-New York 1979.
3. N. P. Elyutin et al: Production of Ferroalloys,
2nd ed., Israel Programme for Scientific
Translations, S. Monon, Jerusalem 1961.
4. W. Dautzenberg, in: Dürrer & Frank, (G.
Volkert, K.-D. Frank eds.): Metallurgie der
Ferrolegierungen, 2nd ed., Springer Verlag,
Berlin-Göttingen-Heidelberg-New York 1972.
5. E. M. Levin et al.: Phase Diagrams for
Ceramists, American Ceramic Society Inc.,
Columbus, 1964, Supplement 1969.
6. W. G. Moffatt: Handbook of Binary Phase
Diagrams, General Electric Corp.,
Schenectady 1973 (plus updates).