RC Nunnington, KU Maske, AFS Schoukens, TR Curr, and NA Barcza,

Developments in refractories for plasma technology,

Institute of Refractory Engineers' Bi-ennial Colloquium, CSIR, Pretoria, 6 March 1985, 38 pp.

.)

DEVELOPMENTS IN REFRACTORIES FOR PLASMA TECHNOLOGY

By R.C. Nunnington, K.U. Maske, A.F.S. Schoukens,

T.R. Curr, and N.A. Barcza
(Council for Mineral Technology, Randburg)

....
•
 A B S T RA CT

Potential operational problems and associated refractory problems

encountered in d.c. transferred-arc plasma furnaces are discussed,
and reference is made to the methods by which they were overcome in
pilot-scale furnaces.

As many of these problems are common to a.c. open-bath furnaces, they can
can be remedied by the use of existing technology, for instance by the
introduction of water cooling to minimize erosion of the sidewalls. This
cooling can be extended further to facilitate the formation of a protective
layer of frozen slag or unreacted feed material at the side walls, which
effectively separates the refractory from the molten metal and slag in the
bath.

However, owing to the nature of the energy-transfer mechanism in the d.c.

transferred-arc furnace, particular care is needed in the design of the
hearth and the roof of the furnace, and this requires a significant
engineering input. The use of special composite refractories containing
graphite or metallic phases are discussed in this context .

..,
 1. INTRODUCTION
Studies of thermal plasma-arc technology and its potential advantages
over conventional processes for high-temoerature extractive metallurgy
were started at the Council for Mineral Technology (Mintekl in the
mid-1970s. This work was undertaken because the effect of this
rapidly developing technology on conventional processes and equipment
like the submerged-arc electric smelting furnace needed to be evaluated.
The work was regarded as being in the national interest, because it
involved the determination of the amenability of local materials to
processing, technically and economically, in the various types of plasma
systems that were being developed. Mintek also approached industry, in
particular ferro-a 11 oy producers, to draw their attention to these new
deve 1opmen ts.

Project work was undertaken initially at Tetronics Research and

Development Limited (TRD) in England, since no plasma pilot-plant
facility was available locally. The transferred plasma-arc system at
TRD comprises a water-cooled metallic plasma device that is
mechanically precessed to spread the arc-contact area over the surface
of the bath. The operation is open arc (as opposed to submerged arc),
and a controlled input of fine feed material, which enters the furnace
via entry ports in the roof, fall under gravity to the surface of the
bath where melting, dissolution, and subsequent reduction take place.
The smelting of chromium and manganese ores and the melting and
refining of off-grade metal fines arising from. product handling at
existing plants were carried out successfully on this facility at
powers up to 0,75 MW.

The need for a local pilot-plant facility became apparent as a result of

those successful pilot-plant trials overseas, and Mintek started work
on the specification and design of a multi-purpose pilot plant in 1980.
 rience
The trans ferre d-arc concept was chosen because of the expe
of plasma
gained at TRD. This paper descr ibes the vario us types
r refer ence
systems and their inter facin g with furna ces with parti cula
which was
to the Mintek faci litie s. The development of refra ctori es,
technology
necessary so that the problems assoc iated with this new
could be solve d, are descr ibed in deta il.

2. DESCRIPTION OF THERMAL PLASMAS

is reached
A thermal plasma is esse ntial ly a state of matt er that
e parti ally
when suffi cien t energy is supp lied to a gas for it to becom
ioniz ed and elec trica lly conductive. Plasmas have been class ified
sses curre ntly
according to their temperatures and press ures and proce
rge plasmas,
of inter est in pyrometallurgy involve the use of arc-d ischa
between 4000
which opera te near atmospheric press ure at temperatures
argon or
and 10 OOO K, and the use of a parti ally ioniz ed gas like
nitro gen to stab ilize the arc. The use of a stab ilizi ng gas is an
meta llurg y, and
esse ntial featu re of the plasma system used in process
arc furna ces.
disti ngui shes plasma furnaces from conv entio nal elec tric-

3. MEANS OF GENERATING THERMAL PLASMAS

cules or atoms
A large amount of energy is requi red to sepa rate gas mole
used in
into ions and elect rons . The energy for gaseous plasmas
tric field
meta llurg ical appl icati ons is supp lied by an extcrna1 elec
circu it. .The
that heats the plasma and foms part of the elec trica l
the process is
relea se of this energy and its subsequent adso rptio n into
is used as a
the essence of the plasma furna ce, in which the plasma
medium to conv ert elec trica l energy into them al energy.

ces, and
Dire ct or alter natin g curre nt can be used in plasma furna
.: are two
hence two or more elec trica l conta cts are requ ired. There
 basic electri cal config uration s: non-tra nsferre d and transfe rred arc
(Fig. 1 ). In the non-tra nsferre d-arc mode, the plasma is generated
between two electro des (positi ve and negativ e) located inside the
device , which is usually called a plasma torch. In the transfe rred-
arc system, the plasma arc is transfe rred from the device (conunonly
•
connected as the negativ e) to an externa l electro de, which usually
takes the fonn of the materi al to be processed (e.g. a molten bath).

Cathode

jacket column
Plasma tail flame

Non-tran sferred-a rc Transferr ed-arc

system system

Fig. 1. Schemati c of non·transferred·ar~ and transferre d·arc systems

3. 1. The Non-tr ansferr ed-arc Plasma System

Non-tr ansferr ed-arc plasma devices were developed origin ally from arc
1
gas heaters .and are typifie d by the Westinghouse arc gas heater
(Fig. 2). The arc heater consis ts basica lly of a closely spaced pair
of tubula r water-c ooled copper electro des within which an electr ic arc
is magne tically rotated at extremely high veloci ties. Process gas is
,. injecte d through the gap between the two electro des and drags the arc
to the interio r of the arc chamber, while the superheated gas is
.
directe d into the proces s.Another example of the non-tr ansfer red-arc
2
system is the SKF plasma genera tor (Fig. 3).
 Magne tic field Replac eable
coils electro des

Plasm a
tail nam~

Water for Gas Gas

coolin g

Fig. 2. Schem atic of Westin ghouse arc heater'

Magn etic
coil

Electr ode
2
Fig. 3. Schem atic of SKF tubula r arc heater device

3.2. The Tran sferr ed-ar c Plasma System

. trica l circu it
The p'rocess mate rial forms an integ ral part of the elec
a d.c. elec trica l
in Tran sferr ed-ar c systems, which normally oper ate from
power supply. The elect rode commonly forms the cathode (nega tive
d is the anode
conn ectio n), and the bath to wl;iich the a.re is trans ferre
water-cooled
(pos itive conn ectio n). The cathode device can be of the
w grap hite
copp er-to rch configuratic~, or it can cons ist of a hollo
elect rode .

number of
This system shows grea t pote ntial for appl icati on to a
y is most
meta llurg ical react ions because the inten sive thermal energ
it is
effec tivel y direc ted at, and trans ferre d to, the mate rial
 processing. Examples of transferred-arc plasma devices are the plasma
torches of the Plasma Energy Corporation 3 (Fig. 4) and Tetronics
Research and Development Ltd 4 (Fig. 5).

Plasma gas

Water
cooling

Molten bath

Transferred arc

Fig. 4. Plasma Energy Corpo_ration's transferred-arc torch with tubular electrode 3

Cathode cooling In~

[Deionized water)
Out-

. - Argon gas

In~
Nozzle cooling
'=>ut-

• •

,
Ndule I '---- Tungsten cathode

'
··~.r~
•
)•\
Fig. 5. T.R.0."s water-cooled translerr.ed-arc torch with rod cathode electrode'
•
4. PROCESSES
A large number of metallu rgical processes have been carried out in a
variety of laborato ry and industr ial plasma furnaces . P·lasma
5
applica tions in metallu rgy were recently reviewed by Barcza and Stewart
6
and by Reid . Most of these furnaces are essenti ally electric -arc
furnaces in which the carbon electrod es have been replaced by one or
more plasma torches and, consequ ently, the conventional electric arcs
have been changed to plasma arcs by the use of controll ed input of gas.

Plasma-arc technology offers a potentia l alterna tive method for some

processes that are traditio nally carried out in electric -arc and
submerged-arc furnace s. The plasma furnace possesses specific
advantages over conventional furnace s, and is being employed
commercially in applica tions in which use can be made of these
advantages. Neverth eless, the potenti al advantages of plasma technology
have not been realized fully, and are still being explored. For instanc e,

the throughput of a plasma furnace can be higher than that of a

conventional furnace because of the higher power densitie s that can be
achieved; hence the same throughput can be achieved by a plasma furnace
that is much smaller than a conventional system. In contras t to

conventional furnace s, most plasma furnaces can process fine feed

materia l directly without prior agglome ration, and can use less-exp ensive
coal fines as the reducing agent in smelting process es. In addition ,
metal contain ing lower concent rations of oxygen can be produced because
air is substan tially excluded from the system as a result of the
introdu ction of the plasma gas, and low carbon levels can be maintained
because the graphite electrod es used in plasma furnaces are usually at
a sufficie nt distance from the melt.

6
 5. MINTEK'S PLASMA FACILITIES AND GENERAL FURNACE-DESIGN PHILOSOPHY
Four d.c. transfer red-arc plasma furnaces rated respecti vely at 50 kVA,
100 kVA, 200 kW, and 3,2 MVA have been installe d in the pilot bays at
Mintek. These furnaces have the same basic configu ration., although
they differ in size and design details.

A schematic diagram of Mintek's 100 kVA furnace is shown in Fig. 6.

7
The furnace , which has been describe d previou sly , comprises a 100 kVA
d.c. power supply, a hollow graphite electrod e (cathode) provided with
a gas supply, a refracto ry-lined steel shell, a flat roof with a
refracto ry lining, and three cylindr ical steel anodes projecti ng through
the refracto ry hearth. The furnace is operated typicall y at 70 kW (90 V,

800 A) at arc lengths of between 50 and 150mm. Argon and nitrogen are

used as the plasma gas at flowrate s between 5 and 15 l/min. In the

numerous smelting campaigns carried out in this facility~ the furnace

\'/JS tapped every 1 to 2 hours.

Feed port
Ceramic wool
Electrode
seal

Nitrogen
purge

Alumina
tube --~--Magnesia lining
'Taphole
~:i---
zone
Thermoco uples Metal
I I
o
o
'
''
I'
.
I
I

- .·. .
0,5m
~

Fig. 6. Mintek's 100kVA'd.c . transferred -arc furnace

7
 By 1981, Mintek's involvement in plasma processes had reached the stage
where it was necessary for the d.c. transferred-arc furnace to be proved
at the pilot scale. It was then decided that a 3,2 MVA furnace should be
built, since it would provide useful scale-up data for industry at the
1 to 2 MW power levels.

The criteria for refractory design were based rather on the proving of
the thennal efficiency of this type of reactor than on the adoption of
the approach involving high throughput versus high heat loss, which is
used in current open-arc technology. The details of the refractory
lining are shown in Fig. 7, which indicates the use of substantial
insulating material, particularly in the furnace hearth.

Rammed, fired, and tar-

Magnesia spine! brick
impregnated magnesia
(90% MgO)

Position of
off-gas ducts
High-alumina castable
(900/o AliOJ)

Insulating brick
(340Jo Ali03)

•
......
1:,,

anode

1,0 !11

Fig. 7. Original lining of 3,2 MVA plasma furnace

8
 The respective average projected temperatures for the side wall and
the hearth shell were 120 and 140°C for heat flows of 11,0 kW per metre
of side-wall height and 1,0 kW per square metre of hearth. During the

final stages of the commissioning run in May 1983, hot spots were
noticed around the gas off-takes. It was later confirmed that these
hot spots had developed because.the brickwork in this area had collapsed.

Subsequent examination of the lining indicated that much of the insultating

material behind the hot-face magnesia brick had fused (Fig. 8), resulting
in the destabilization of approximately one-third of the hot-face lining.

Fig. 8. Photograph showing fusion of insulation and resultant collapse of hot-face brickwork

At that stage, successful'test work on the smaller 100 kVA scale had led to
the realization that a h~gh-power·, high-throughpu t operation was necessary
for reliable furnace -performance in smaller vessels with a more conductive
lining. The philosophy is similar to that used· in the design of ultra-high
... ~;;.. power (UHP) steelmaking furnaces, i.e. that extra losse.s".._of heat energy are
more than compensated for by high throughputs and short turn-around times,
provided that it is borne in mind that the plasma operation is essentially ..
not a batch process except for the intermittent tapping.
q
 It was also noted that substantial thermal shock damage
(spalling) had occurred in the hot face brickwork in the upper
sidewalls. The 3,2 MVA furnace was subsequently relined with a
hot-face of chromate bonded magnesia ramming, but subsequent
courses of direct-bonded magnesia-chromite and magnesia ramming
were used as the safety lining in the sidewalls. Details of the
lining are given in Fig. 9.

Magnesia spine! brick

High-alumina castable (90% MgO)
(90% AJi03)

Magnesia-chromite brick
(30% Cr203)
Inspection
port

Chemically bonded magnesia

ramming (95% MgO)

Chemically bonded magnesia

ramming (850/o MgO)

Tar-impregnated magnesia ,,.__ _ _ Stainless-

brick (96% MgO) steel anode
!,Orn
r
.
Fig. 9. Details of revised lining for 3,2 MVA plasma furnace
.

By then, the design temperature for the shell had risen to 258°C
for a much higher heat loss of 38 kW per metre of side-wall
height. similarly, the temperature of the hearth shell was
predicted as being 16Q°C for a heat loss of 1,4 kW per square
.metre of hearth-shell area. However, side-wall temperatures were
not considered to be high enough to warrant the use of wa~er
cooling. A 10-·day smelting campaign was carried out with the
above lining strategy ~n which some 85 toris of slag and metal
were tapped at an average tapping ternperature·of 165o c. Average
0

lining wear of the side-wall refractory was subsequently measured

and found to be 30 mm, and no evidence of damage due to therrno-
rnechanical stress.
 Recent process-che mistry testwork was centred on the 200 k'vl furnace
illustrated in Fig. 10. This furnace uses the same power supply as the
3,2 MVA furnace, and therefore it can be used in operations to test whether
higher power densities can be used.

Vi'~"od'
Baked S0derberg electrode Off-gas

I
G"mmmg
paste Feed port port port
Magn~sia
~
Electrical

ring
\J L7
440mm
el
Alumina fibre

Magnesia
ramming
Pyrometer
'sight tube' IJIE .,
Taphole

Metal heel

Datum line

Mild- - -
steel
anode

•
0,5m

Fig. 10. Schematic arrangement of 200kVA furnace

6. SPECIFIC REFRACTORY PROBLEMS ENCOUNTERED AT MINTEK AND HOW T.HESE

WERE OVERCOME
Th~ specific problems related to-~efractories can be cl~ssified in terms

of their location in the furnace and their origin: i.e. electrical,

chemical, or thermo-mechanical. These are summarized schematical ly in
Fig. 11.
 G

D\ J
wt 'j"
Roof erosion
(spalling
o,;d,.;o")

Radiation --+----.,f----4t

Metal

Melting
of anode

Fig. 11. Refractory problems encountered In a d.c. transferred-a rc plasma furnace

Electric al problems include melting of the anode, stray arcing to the

shell, and arc instabi lity leading to hot spots at the side wall. Chemical
erosion is encountered at the side wall, specific ally in the region of the
slag-me tal interfac e.

Refracto ry wear in plasma- furnace roofs is due mainly to high-tem perature

fluxing and to the high thermo-mechanical loads imposed by the high levels
of upward reflecte d radiativ e heat energy.

6.1. Anode Melt-down

Several configu rations are possible in the hearth of a d.c. transfer red-arc
plasma furnace. Perhaps the most vulnera ble configu ration is that in which
one or more steel anode bars are embedded in the hearth refracto ry to
provide the necessa ry electric al contact with .the molten bath. The followin g·
are possible disadva ntages of this ._system.
(a) The arc become~ attach~d to one particu lar anode, causing rapid melt-

down of that anode and the break-o ut of slag and m~tal from the •

furnace . Once a run through has occurre d, refracto ry damage in this

area preclude s the insertio n of a new anode bar.
 (b) To prevent melt-down after tapping of the furnace, a residual
metal heel must remain in the furnace at all times. This heel
can cause concentration or dilution effects in the tapped
material, and generally restricts the ability for process-
chemistry variables to be changed during a campaign.

s~
A1 ternative hearth designs have been adopted by ASEA of Sweden
and Didier 10
of West Germany. These involve the use of an
electrically conducting refractory hearth. The ASEA design is
shown schematically in Fig. 12.

Hollow graphite electrode

Off-gas port

Tapping
Steel
cladding
or bars De-slagging port

Metal
anode
plate
8
Magnesia
brick
sump
or ramming·
or
water
cooling

Magnesia-graphite
and insulation bricks

Fig. 12. The ASEA furnace 9

The upper course consists of high-fired magnesia brick with metal

cl adding on two faces. The cl ad<;li'hg a11 O'i·IS e l ectri city to be conducted to
the lower courses, which consist of alternating layers of graphite ~nd
insulation brick. This combihation allows for a continuous conducting
path to the air- or water-cooled anode plate, and minimizes heat losses

13
 from the furnace hearth. However, this is a complex design, which must
allow for the exact matching of joint tolerances and thermal expansion
to avoid the opening-up of joints, leading to metal penetration or loss
of electrical continuity.
The Didier design also uses metallic conductor s, but in the form
of metal rods connected to an external steel plate attached to
the underside of the furnace. As with the ASEA design, the
hearth refractor y consists of successiv e layers of graphite-
containin g refractory , with a working hearth of magnesia-
containin g chromic oxide.
The concept of a conductin g hearth was also adopted by Mintek,
but simpler hearth designs were used that require the use of only
one refractor y. Two different designs were tested successfu lly
on the 100 and 200 kW furnace facilitie s.

6.1.1. Extended collector-p late type hearth

The first type of hearth is a hybrid that utilizes metallic
conductors but not electricall y conducting refractorie s. It is

shown schematical ly in Fig. 13.

Collector Bus bars

pins (mild steel)
(mild steel) 4 of 100 mm width
24 of IOmm
diameter

Collector plate attachment

(mild steel) to furnace shell

0,5m

Fig. 13. Extend_ed collector-plate type conducting hearth

~circular· steel collector plate, to which 20 to 30 steel

collector pins are welded, is embedded in a rammed hearth that
·-
.can be of magnesia or alumina.ma terial. The collector pins

penetrate to the bath of molten metal, and have a total cross- '" .·I
sectional area designed to carry more than the full current I
 pins are selected so that the energy flow by thermal conduction
will be restricte d in the hearth area directly beneath the arc-
attachment zone. This limits the flow of energy that the collector
plate is required to dissipate so that reasonabl e temperatures can
be maintained. The internal metal collector plate then distribut es
the thermal load and collects the current load from each pin. It

should be noted that the anode attachment is made directly to the

furnace shell at the lower side wall, which further minimizes the
possibili ty of metal run-out.
This design differs somewha t from that propose d by Didier, where
the metal rods in the furnace hearth provide the necessar y
electric al connect ion between the metal bath and a graphite sub-
hearth~ which is in contact with the externa l collecto r plate

attached to the furnace bottom.

Several smelting trails were carried out successfu lly by the use of
this configura tion at currents of over 1000 A and power densities
2
up to 0,6.MW/m. Chromate-bonded magnesia ramming was used as the
hearth refractor y.

Little or no damage to the collector plate was observed after

completion of the runs. Perhaps this was to be expected since any
collector pin will tend to re-solidi fy as the arc attachment moves
to favour other collector pins. Even if inadverte nt over-powering
of the furnace should occur, there is little chance of a furnace
break-out with this design of anode contact with the bath.

6. 1.2. Conducting refractor y hearth

The second type of conducting hearth uses electrica lly conducting
resin-bonded magnesia-graphite refractor ies. These refractor ies
·"!:'.

are in the form of a rammable for the small-sca le 'pot-test ' and
100 kVA furnaces, or interlock ing bricks with an integral collector -
plate design for the 200 ·kV/ furnace (Figs. 14 and 15 respectiv ely).

Only one type of conducting refractor y is used in the bricked

conducting hearth without any steel cladding , thus giving improved
joint tightness .
 8

CJCJ
Pre fired Rammed
magnesia-----+-+!-'- magnesia-graphite
-<1--------i-
crucible hearth refractory
Mild-steel
Collector ------r-tF=~~"~~~=:U--+-- collector
pins plate

Mild-
steel
anode

0,25m

Fig. 14. Conducting refractory hearth for small-scale 'pot-test' operation

Anode attachment
Collector to furnace shell
plate

0,5m

Fig. 15. Magnesia-graphite bricked conducting hearth

16
 Electrical properties. The excellent combination of properties
that can be achieved by the introduction of substantial
proportions of flake graphite as a matrix component in magnesia-
11 1
based refractories is well-documented - "'t. However, although
these properties are of relevance in a conducting hearth refractory,
it is the electrical properties imparted by the graphite which
are the most important, and the graphite flakes must be distributed so
that a continuous network is formed in the brick matrix (Fig. 16).

Fig. 16. The optical microstructure of magnesia-graphite conducting hearth refractory. The
continuity of the graphite network (G) around the grains of periclase (P) can be seen

Electrical continuity is ensured by maximization of the number of

graphite particle-parti cle contacts. This is achieved by the use
of a fine graphite particle size (typically SO.per cent smaller
than 75µm), which gives a high ratio of surface area to volume,
and ~ mixing technique that results in the coating ~f individual

magnesia grains with a layer of graphite flakes.

17
 In order that the conductiv ity of these materials could be
assessed, the electrica l resistivi ty of rammed and cured specimens
of 150mm length and 50mm diameter were determined. A simple

measuring technique was used, which had been developed for the
15
determina tion of the electrica l resistivi ty of electrode paste •
The measurements obtained on fused grain magnesia-graphite
materials containin g up to 40 per cent graphite are shown in Fig. 17.

10-2
8
I =Test pieces containing
6
synthetic graphite
4
2
2 =Test pieces containing
E 10-J natural flake graphite
c: 8
6
?;-
4
:~
2
"'
·;;;
u 10-4
....
c; 8
u
·.::: 6
uu 4
Ei 2
10- $
8
6
4
2
10-6
10 20 30 40
Graphite content, OJo by mass

Fig. 17. Electrical resistivity versus graphite content for materials containing natural flake and
synthetic graphites at room temperature

As expected, the electrica l resistivi ty of magnesia -graphite

materials decreases with increasin g graphite content, irrespect ive
of the type of graphite used. However, a saturatio n level is
reached when the graphite content is above approximately 30 per cent,
i.e. when the minimum number of electrica l contacts required to
realize the maximum possible conductiv ity in the matrix is achieved.

At equivalen t graphite contents, natural flake graphite is

18
apparently marginally more effective as a matrix conductor than
1
synthetic 1 graphite (ground electrode stubs), possibly because
of its greater degree of graphitizat ion as the theoretical
atomic-packing density is approached.

Compatibili ty with process chemistry. One reservation regarding

the use of carbon-cont aining refractorie s in the hearth of a plasma
furnace is the dissolution of carbon in the metal or alloy in the
bath. This problem is particularl y relevant during the smelting
or refining of low-carbon ferro-alloy s, and is th~ reason why the
use of graphite alone as a hearth material is avoided.

The likelihood of carbon pick-up from the graphite was evaluated

in a series of 10kg pot-test experiments in the small-scale pot-
test reactor (Fig. 14). Table 1 gives the chemical analysis of·
silica-mang anese, medium-carbon ferromanganese, and silicon fines
as received, and after remelting. A magnesia-30 per cent graphite
hearth was used.

TABLE
Analyses of fines as received and after re-
melting. Composition, % by mass

Si Mn Fe c Cr Ca

Fe-Mn-Si fines As received 15,4 65,0 13,3 1 ,8 - -

Remelted 16,5 67,9 14 ,2 1 '8 - -
Medium-C Fe-Mn
fines As received 2,4 80 '1 15,5 1 '4 - -
Remelted 2,7 79 ,2 15,7 1 '4 - -
Si fines As received 95,7 0,09 0 ,8E 0 '14 0,02 0,23
Remelted 97,0 0,04 0 ,s: 0 '10 0,06 0,06

No pick-up of carbon occurred in any of the remelted alloys, and the

silicon fines were even refined slightly. This indicates the
suitability of this type of hearth for these remelting operations.

19
 On the larger 100 kVA scale, successful chromite-smelti n9 trials
lasting many days were carried out using a rammed hearth containing
30 per cent graphite. Further trials are in progress in the larger
200 kW furnace using the bricked version as a preliminary stage
so that the suitability for scale-up can be assessed.
Initial results from these trials, which involved the remelting of ORI,
have indicated that power densities of 2,0 MW/m2 and above, at feed
rates exceeding 500 kg/hr, are possible with the bricked hearth design.
Use of alternative materials. Other materials are being

investigated for use as electrically conducting hearth refractories both

as the bulk-refractory oxide phase as well as the matrix-conduct ing
phase. These include alumina and partially stabilized zirconia as
the oxide phase, and finely ground ferro~alloy fines as the

conducting phase.

Fig. 18 presents the results of electrical-res tivity tests that

\vere done at tempera tu1·es up to 1500°C under argon on a. fused-
grai n magnesia-26 per cent graphite material, a commercial
alurnina~26 per cent graphite brick, and a fused-grain magnesia-
35 per cent ground ferrochromium fines material.

10- J

8
6
4
2
E Alumina;-26070 graphite
c: 10-.
>: 8
>
6
.,,
-~
~ 4
-;:; • I
I
·;:
u 2
CJ
<J 10- ~
Lli
8
6
4

2
10- ,,
···:-> ·- ·II

0 250 500 750 1000 1250 1500

Temperature, °C

Fig. 18. Electrical resistivity versus temperature for magnesia-graphite, alumina-graphite, and
magnesia-ground ferrochromium materials
 i.

All three curves have a chara cteris tic shape , which is simila r to
that for semi-c onduc tors. A rapid decrea se in resiSt ivity occurs
above about 750°C, followed by a flatte ning- off above 1150 to
1200°C. This is believ ed to be due to carbon izing of the phenolic
resin bond that is common to all three mater ials, since mater ials
of simila r composition with a non-carbon yieldi ng bond (e.g. Calgon)
do not exhib it this type of curve and, in fact, exhib it little
change in restiv ity over the whole tempe rature range.
the
Also, the resis tivit y value s on cooli ng did not follow
heati ng-up curve and showe d only a sligh t incre ase in
resis tivit y when coole d to room temp eratu re. It would

appea r that, as the resin carbo nizes, the electr ical contac t
between the graph ite flakes improves, and contin ues to improve
until carbo nizati on is complete at 1150 to 1200°C. This indica tes

the importance of resin conten t and qualit y (i.e. carbon yield) in

electr ically conducting refrac tories of this type. In gener al,

these result s show that many combinations of oxide phase and

condu cting- matrix phase are possib le, and can be matched to
partic ular proces s-chem istry requir ement s. This is partic ularly

impor tant when the conducting matrix phase is delibe rately made
metal lic because, if the purity of the mater ial in the brick
matrix is equal to, or higher than that of the molten metal or
alloy in the bath, contam ination due to dissol ution of the
refrac tory matrix can be avoide d. For example, it is highly

necess ary for the fonnat ion of extrem ely stable carbid es in
titaniu m alloys to be avoide d, since even very small quant ities
of dissol ved carbon from a graph ite matrix may make ·it diffic ult.
for alloy specif icatio ns to be met.

6.2. Stray Arcing

A phenomenon that has been encountered in the d.c. transf erred- arc
ed
furnac e is 'stray • arcing , in which arcing from the cathod e is direct
via
to the furnac e shell via the roof components instea d of to the anode
the bath (Fig. 11). Stray arcing often occurs when the feeding of the
raw materia ls begins, which effectiv ely increase s the arc resistan ce.
This is obviously not desirab le, since it results in over-he ating of the
furnace shell and a loss of power input to the molten bath.

A more serious effect of stray arcing occurs when it is directed to a

water-co oled plasma torch, since it can cause holing of the outer water-
cooled copper tube with subsequent leaking of water into the furnace.

At Mintek the problem of stray arcing to the roof was overcome mainly by
electric al means and partly by changes in refracto ry design. The

electric al method involved electric al insultat ion of the roof from the
furnace itself. Hence, the roof shell is allowed to rise electric ally to

a potentia l that could be of the order of 110 V for the 200 kW' furnace.

Safety precauti ons are necessar y to prevent personnel from acciden tally
contacti ng the roof while working on the furnace . The change in

refracto ry design involved the development of a water-co oled electrod e

seal or 'stuffin g box•, which physica lly isolates the electrod e from
the roof shell (Fig. 19).
Graphite
electrode

Resilient insulation

fi - - Cooling water in
1
~,1--- Compressiv e
j insulation
Cooling water 1-,...+;::;:; ::=--- Purge gas in
out

Magnesia spinet brickwork

Rammed, fired, and tar-
impregnated magnesia
plug

Fig. 19. Schematic arrangement of electrode seal and refractory plug

??
Nitrogen is forced into the furnace and passes over the surface of the
hot electrode. This has the added advantage that the ingress of air
is suppressed and, hence, the oxidation of the graphite is reduced. The
refractory plug on which the seal is seated consists of a ra1M1ed and
fired 95 per cent magnesia material, which was impregnated with tar and
coked. At first, a 90 per cent alumina castable was tried, but this
could not withstand the thermal shock resulting from the high levels of
radiative heat energy reflected from the surface of the bath.

The stray arcing associated with the water-cooled plasma torch has been
overcome successfully by the installation of graphite sleeves round the
outer copper tube 16 (Fig. 20).

- - - Clamp assembly

Graphite
sheath

...... - _j. .. '

'

Fig. 20. Water-cooled plasma torch with protective graphite sleeve''

23
The choice of graphite is possibly surprisi ng at first, in view of its
high conduct ivity. However, insulati ng ceramics like alumina or
zirconia cannot be used because of the tremendously high thermal gradien ts
through the sleeve, which can be as high as 1700 to 1750°C over a radial
thicknes s of 10 to 20mm. The alternat ive approach is the use of a

sleeve that is highly electric ally conducting and the provisio n of a

disperse d current path back to the.wate r-cooled copper tubes so that the
arc does not attach directly onto the copper. Graphite is the only

materia l current ly availab le that can meet both requirem ents, i.e.
electric al conduct ivity and resistan ce to thermal shock. Open-arc

experiments with this design failed to induce any damage to the torch
due to stray arcing, even when the stray arc was deliber ately introduc ed
to the graphite sleeve as a dead short.

6 . 3. Are F1a re

In d.c. single-e lectrode (or water-co oled plasma-torch) systems , the

problem of 'arc flare' has been eliminat ed largely by the use of a
centrall y located electrod e and by gas stabiliz ation of the plasma
column.

However, if skewing of the arc is allowed to take place at the bath

surface , energy transfe r can be directed towards the side wall of the
furnace , where rapid localize d erosion
. . ries can occur
of the refracto

(Fig. 11). Several conditio ns can lead to skewing of the arc-atta chment
point, e.g. the use of extended arcs, insuffic ient flow of gas,
unsuitab le geometry of the electrod e tip,.and asymmetrical electric al
connect ions at the anode and the cathode where electrom agnetic
·'
interact ions with the plasma column will force it away from a •
centrali zed position . In larger furnaces these conditio ns can be
 and the
remedied to a certa in exten t by a sound engin eering approach
where
contr ol of process varia bles. However, in small -scale react ors,
wall is much
the critic al distan ce between the electr ode tip and the side
can resul t
short er, even sligh t exposure of the side walls to arc flare
in sever e refra ctory damage.

tively
This problem was large ly overcome at Mintek by the use of selec
es of high
placed water -coole d panel s used in conju nctio n with refra ctori
try. An
condu ctivit y, and by caref ul atten tion to elect rode- tip geome
scale
example is given in Fig. 21, which shows a react or of pot-t est
rammed
with an enclosed coolin g jacke t of the helic al-co il type, a
il-po int'
resin-bonded magn esia-g raphi te refra ctory , and a machined 'penc
electr ode tip.

C =:r-Roof
:=...r,:;; ;::;;-._;:;;

Water-
Pencil 1. . .- - cooling 0,25m
electro de jacket ·

Ir Graphi te
block
'--~~~~~~
~~ ........

oint electro de tip

• Fig. 21. Pot-tes t reactor with water cooling and pencil-p

· ....

Succe ssful trials were carri ed out with this system, up to ten
•
indiv idual experiments being carrie d out with a singl e rammed
lining

25
with neglig ible fettlin g. Previo usly, a prefir ed 95 per cent

MgO-magnesia crucib le was used in each experiment.

The pencil -point electr ode tip ensure s that the arc attach ment at the
cathod e is confined to a limite d surfac e area at the tip, and cannot
wander around the periph ery as it does with an electr ode of larger
diame ter. Typic ally, the electr ode tip has a diame ter of 15mm and a

centra l hole of Smm, whereas the standa rd electr ode used up to 200 kVA
r
level has a diame ter of 50mm and a centra l hole of 10mm. The smalle
centra l hole requir es that higher gas veloc ities should be generated
to mainta in a simila r gas flowra te. The higher gas veloc ities also
help to centra lize the arc. The use of water- cooled magne sia-gr aphite

mater ials is discus sed in more detail in a later sectio n.

6.4. Chemical Erosion of Side-w all Refra ctorie s

of
Although chemical erosio n can be discus sed as an indivi dual mechanism
refrac tory wear, it is usuall y associ ated with, and worsened by, the
arc-fl are problem discus sed above. Consequently, the methods used for
s
the-Te ductio n of wear due to chemical erosio n are simila r to the method
that can be used for the minim ization of damage due to arc flare. In
additi on to select ive water -cooli ng of the side-w all refrac tories , raw
,
mater ials can be fed select ively into the furnac e around the side walls
partic ularly into hot-sp ot region s, to modify the composition of the
the
slag locall y and to increa se the visco sity of the slag. T.his limits
abilit y of the slag to penet rate into the open pores of the refrac tory
hot face.

Although chemical erosio n is gener ally due to chemical intera tions

between the slag and the refrac tory phase s, it can also be a direct
resul t of the reduct ion of the oxide refrac tory by silico n, titaniu m,

26
aluminium, and carbon species that are dissolved in the alloy, as was
realized when inconsisten t silicon levels were obtained during the
smelting of certain ferro-alloy s. This inconsisten cy was invariably associated

with severe magnesia refractory erosion and a brilliant white flame

at the exhaust port. The erosion was in the form of under-cuttin g or
notching, and comprised up to 80 per cent of the original refractory
thickness over a side-wall depth of only 10 to 20mm. Upon analysis,

the condensed fume, which had rapidly blocked the off-gas port, was
cound to contain over 90 per cent MgO, which confirmed that the intense
white flame was associated with the combustion of magnesium-metal vapour.

Some of the possible reduction reactions taking place in the furnace are
shmvn be 1ow.

+ Mg ( vap ) ' ( 1)
MgO(ref) +Si -:: SiO(vap)

MgO(ref) + 12Si t ~Si0 2 (slag) + Mg(vap), (2)

MgO(ref) +~Si+ CaO(slag) t H2Ca0.Si0 t(slag) + Mg(vap)' (3)

and 1M + CO (vap) fA
+ '19 (vap)'
(4)
g0 (ref) + C +_

where the subscripts ref and vap refer to the refractory and the vapour
respectivel y.

Plots of the change in free energy as a function of temperatu.re for these

reactions (based on one unit mole of magnesium vapour) are plotted in
Fi g. 22.

-·.

27
 1 =React ion line MgO(r<fJ + Si~SiO<••Pl + Mg<••Pl
2 =React ion line MgO(r<f) + JSi~JSi02(slag) + Mg<••pl
3 =React ion line MgO(n:I) +!Si+ CaO(,lag) ~j(2Ca0 · Si0 2 )(•I••> + M&<••Pl
4 = Reactio n Ii ne M gO(n:f) + C ~ CO<••P> + M&<••pl
---Rea ction line M&.r aim)~Mgo ••ml
200 '
PM,= 0,001 atm
150
--- --- ---
--- ---
100
--- --- - --- -
0 50
<l

1600 1700 1800 1900 2000 2100 2200 2300 2400

-50 I 1877K 2050K ~

. ~
(1600°C ) (1777 °C)

Temper ature, K

carbon or silicon (Data from Barin and

Fig. 22. Free-en ergy diagram for the reductio n of MgO by
Knacke 17 )

tempe rature
This simple thermodynamic approach indic ates that, at a bath
ive change
of 1600°C, none of the react ions are favoured, because a posit
in free energy is indic ated. However, at a tempe rature of 1777°C, the
magnesium
reduc tion of magnesia by silico n in the presence ·of CaO to form
absence
vapour and dicalc ium silic ate is feasi ble·( react ion 3). In the
ion
of CaO, or where silico n monoxide is formed, the minimum react
2 and
temperatures are much highe r, i.e. 2177 and 1997°C for react ions
is the
1 respe ctive ly. Signi fican tly, the carbo n-red uctio n line ·(4)
red.
next most favou rable , a minimum tempe rature of 1857°C being requi
favou rable
At tempe rature s above 1907°C, reduc tion by carbon becomes more
than reduc tion by silico n.

on of
It will be noted that all the above react ions involve the.fo rmati
of 1
one or more gaseous produ cts, and a magnesium parti al press ure
atmosphere. Where one or both of the parti al press ures of these speci es
are lower than 1 atmosphere, the equilibriu m will be shifted to the right
of the equation. One effect of this is that the reaction becomes
thermodynamically feasible at a lower temperatu re. Lines reflectin g the

change in equilibriu m for reduced magnesium partial pressure are also

•
shown in Fig. 22.

During the reduction of metallic oxides with carbon, carbon monoxide is

generated . This gas, together with the plasma gas nitrogen or argon,
lowers the partial pressure of the magnesium vapour in the furnace
atmosphere. If the partial pressure is lowered, say to 0,3 atmospheres,
reduction reaction 3 will be thermodynamically feasible at only 1592°C.
This is in accord with observati ons made during the furnace campaigns, i.e.
that maximum refractor y erosion is associate d with 'stewing' of the bath,
little slag cover, and with no feed entering the furnace. ln this

instance, the magnesium vapour can be swept away by other furnace gases~

hence, a decreased partial pressure is maintaine d, and reduction of the

magnesia is favoured. If a deep slag cover is present, the total pressure
at the metal-ref ractory interface will· be greater than atmosphere and,

hence, the partial pressure of magnesium will approach atmosphere. This

will result in less favourabl e condition s for the formation of magnesium.

These findings led to the adoption of modified furnace practice, in which

more careful attention to balancing of the feed rate and power input
obviates the need for 'stewing' of the melt prior to tapping.

A further development is the use of a freeze lining to prevent the

molten slag and metal from contactin g the side~wall·refractory. A

layer of 'frozen' slag can be built up by selective water-coo ling of

the side walls. The furnace lining then consi~ts, in effect, of

resolidif ied slag. Alternati vely, feed material can be added

29
deliberately in such a way that it enters the furnace in close
proximity to the selectively water-cooled side walls. A thin layer of

unreacted feed is then built up at the side walls as the level of the
bath rises.

Initial work on the pot-test scale was so successful that prefired

magnesia and alumina crucibles could be reused several times. Fig. 23

shows a solidified plug of material with a surface coating (freeze

lining) of unreacted feed and slag. This plug was removed intact,

leaving an untouched refractory surface in the magnesia crucible.

Fig. 23. Solidified plug showing freeze linir:ig of unreacted feed and slag

However, the rate at which heat can ne removed from the reactor (which
limits the power-input level) is detenntned largely by the low thennal
fi"J·: .
conductivity of the magnesia , i.e. 3,0 W/m/K.

10
 The thennal conductivities of magnesia -graphite materials are much
higher than this 11 , typically about 20 W/m/K for a resin-bonded 80 per
cent magnesia-20 per cent graphite material, which allows for faster
rates of heat removal. The graphite content of these materials imparts
other properties that make them suitable for use in a plasma furnace,
as follows.
(a) Their resistance to thennal shock is excellent due to a minimal
degree of ceramic bonding between the magnesia grains, which
retards crack propagation through the refractory. Also, because
the graphite matrix is slightly compressible, it can absorb any
expansion of the individual magnesia grains.
(b) They have low coefficients of thermal expansion. For instance,
the typical percentage expansions of magnesia and magnesia-20 per
12
cent graphite material at 1200°C are 1 ,6 and 0,6 respectively .
The thennal expansions of magnesia and graphite also helps to
retard crack propagation during thermal cycling.
(c) Their low wettability by molten phases minimizes the penetration
of those phases into the open pores of the hot face.

In addition, where resin or tar bonding is used, low porosity and good
hot strength can be achieved without the need for hard firing, which is
commonly used for magnesi~ brick. This feature makes the use of rarrmed
linings in the 10 to 20 per cent graphite range very attractive for
smaller furnaces and for furnaces with complex lining de~igns that

preclude the use of preformed shapes.

EncouragiQg results were obtained with a resin-bonded fused-grain

magnesia-20 per cen~ flake graphite material, which can be rammed in situ
and requires a curing temperature of only 150°C. The testing of this

31
material was limited to small-scale experiments in which 10kg of feed
material were used, and it was found that up to twelve individual
experiments could be performed with the same lining without the need for
major fettling. The cooling ability of the lining was improved further
by solid-metal cooling fins that were embedded in the refractory.

In conjunction with the research work on these materials, a computer

model is being developed that will enable water-coole d refractory
systems to be designed for larger furnaces requiring a protective freeze
lining. A one-dimensional steady-stat e heat-flow model enables the
design of the cooling panels, and the flowrates, thermal conductivit y of
the refractorie s, lining thicknesses , and freeze-lini ng thicknesses to

be specified. The system on which the model is based is shown schematical ly

in Fig. 24.

\\;.it er _ _ __
in

Freeze
\V ater-cooling
/lining
panel
Cooling
fin Molten
Magnesia- -H.c,..c,.~~ bath
graphite
refractory

·I
Water
out

HF= Heat flow in frozen feed or slag layer

H..,= Heat extracted by water cooling
H R =Heat flow in refractory layer

Fig. 24. Schematic representation of the system us~d In the development of the heat-flow model

., ')
 This concept can be extended further in the design of slag-covered
water-cooled panels at the upper side walls. However, in view of the
danger of stray arcing to the panel, where only a thin slag cover is
present, it is unlikely that water-cooled panels will be used
extensively without a hot-face lining of refractories in d.c.
transferred-arc furnaces in the near future.

6.5. Refractory Damage in Plasma-furnace Roofs

Before the plasma facilities at Mintek were developed, a major concern
was that the use of long (typically longer than 250mm) column plasma
arcs would result in radiation damage to the side walls. In practice,
however, the radiation damage was found to be directed ratber at the
roof by intennittent upward reflection from the bath surface. The amount
of radiation reflected to the furnace roof is governed largely by the
nature of the process (i.e. whether .it involves melting or smelting), the
amount of gas produced in the process, the arc length, the distance from
the bath surface to the roof; and the amount of feed material on the bath
surface.

Several alumina and magnesia castables or rammables were tried in the 100
and 200 kVA furnaces, but heavy spalling and reactions with the slag and
gaseous phases in the furnace atmosphere result in poor roof life. Coked
electrode paste was tried as an alternative roof material but, although this
material could withstand the thennaJ stresses imposed by the radiation,
rapid erosion often occurred in the hottest centre section due.to the
ingress of air through the· various roof ports, particularly where the
sealing was inadequate.

33
 The dual requirements of high mechanical strength and a high degree of
resistance to thet111al shock were met finally by the use of a composite
material developed by the use of existing concrete technology. This
material is based on a 95 per cent alumina castable, and contains 6
per cent by mass stainless-steel fibres measuring 0,5 by 2 by 25mm,
which are produced from a type 304 - 18 per cent chromium, 8 per cent

nickel - steel (Fig. 25).

Fig. 25. Silmple of used castable

The dried and fired properties of this material are given in Table 2,
which shows the high strength attainable.

TABLE 2
Dried and fired properties of 95 per cent alu~ina castable
containing 6 per cent stainless-steel fibres

Property Dri ed at 11 O° C

Bulk density, t/m 3 2,90 2,70

·. Cold ~rushing strength, MP
a 62 31

Pet11lanent linear change, % 0,5

.. •

34
 Successful furnace campaigns were achieved with the roof design shown
schematically in Fig. 26.

Water cooling
Water-cooled Feed
t off-gas ports
Electrode port
port

Mild-
steel
anchors

•
95% alumina castable
containing 6% stainless-steel fibres

0,1 m

Fig. 26. Schematic representation of water-cooled roof consisting of 95 per cent alumina castable
containing 6 per cent stainless-steel fibres

Because of its excellent resistance to damage from thermal shock,

the material lends itself to the use of water cooling, the
ef~iciency of which is aided by the use of mild-steel hanger pins

(anchors) that also help to maintain the structural integrity.

Typical heat losses of 10 to 12 kW to the cooling water were
noted for flowrates of 25 to 30 l/min. This was sufficient. for
the integrity of the fibres to be maintained to within 20 mm of
the hot face.
 7. CONCLUSIONS
It has been shown that many operational and associated refractory
problems that are typical of a.c. open-bath furnaces occur in the
d.c. transferred-arc plasma furnace. However, the energy-transfer
mechanism to the bath, particularly where long arcs are used can
give rise to some problems that are confined to plasma-furnace
operation.

Many of the refractory problems have been solved, not by the

development of new refractories, but by the modification of existing
technology, and by the selection of the refractory as a component of
an engineering system rather than by the consideration of its
suitability as compared with that of alternative materials.

Thennal-plasma technology is reaching the stage where large-scale

pyrometallurgical applications are a reality, and smaller-scale
operations have been in use for more than 10 years. One particular gas-
refonning process using plasma has been in use for some 40 years. The
larger-scale developments, i.e. great'er than 10 MW, are for the melting
of steel scrap and alloy steels, tQe smelting of ferro-alloys, and the
processing of bag-house dust from steel plants. The refractory work
undertaken at Mintek so far has benefited pilot-plant work and industry.
It is hoped that this contribution will be of assistance in the further
commercial realization of suitable plasma systems for the processing of
local materials that are less suited to processing by conventional
methods:

8. ACKNOWLEDGEMENTS
This paper is published by pennissl.on of the Counci 1 for Mineral
Technology. The contributions of Messrs H. Lagendijk, B. Tinniswood
and B. Waldron, who conducted much of the experimental work on the

36 p'
 refractories, and Messrs K.C. Nicol, M.S. Rennie, W.C.J. Cameron, and
B. Groenewald, who assisted with the work relating to control and
measurement, are gratefully acknowledged.

REFERENCES
1. FEY, M.G., and MELILLI, W.J. The application of thermal plasma
systems to economic scale ironmaking. PREPRINT, Sth INTERNATIONAL

SYMPOSIUM ON PLASMA CHEMISTRY, EDINBURGH, SCOTLAND, AUGUST 1981.

pp. 1-21.
2. ANON. Reduction by plasma metallurgy. MetaZZ. Plant Technol.,

vol. 5. 1984. p.100.

3. CAMACHO, S.L. Long-column plasma arc torches for pyrometallurgical
applications. PREPRINT, MINTEK 50, SANDTON, SOUTH AFRICA, MARCH

1984. pp. 1-14.

4. PAGE, D. Some of Tetronics 1
contributions and achievements in
respect of plasma developments. Faringdon, U.K., Tetronics R & D
Limited. pp. 1-13.
5. BARCZA, N.A., and STEWART, A~B. The potential of plasma arc
technology for the production of :fer,ro-alloys. PREPRINT INFACON

83, TOJCIO. pp. 1-24.

6. REID, K.J. Plasma metallurgy in the 80's. PREPRINT, MINTEK 50,

SANDTON, SOUTH AFRICA, MARCH 1984. Pr· 1-23.

7. CURR, T.R., et al. The design.and operation of transferred-arc
plasma systems for pyrometallurgical applications. 6th INTERNATIONAL

SYMPOSIUM ON PLASMA CHEMSITRY. Boulos, M.I., and Munz, R.J. (eds.).

Sherbrooke (Canada) Universitede Sherbrooke, 1983. vol. 1,

pp. 175-180.
8. STENKVIST, S.E. The high powered graphite cathode d.c. arc
plasma-properties and practical applications. PREPRINT MINTEK 50,

SANDTON, SOUTH AFRICA, MARCH 1984. pp. 1-19.

 9. STENKVIST, S.E. A method and device for operating a d.c. arc

furnace. S.A. Patent 82/8759. Assignee ASEA Aktiebolag,

Sweden, 29th Nov. 1982. pp. 1-17.

10. HLAWATSCHEK, H., et al. Hearth bo~tom, more particularly

for d.c. arc furnaces. S.A. Patent 85/1301. Assignee:
Didier-Werk AG, West Germany, 20th Feb. 1985. pp 1-6.

11. JACKSON, B., and WEBSTER, R. Developments continue in

magnesite-carbon refractories for EAF and BOS use. Iron Steel

Int., vol. 54, no. 2. Oct. 1981. pp. 67-78.

12. BROWN, A. The properties of ceramic graphite bodies. Refract. J.,

vol. 58, no. 2. 1983. pp. 7-10.

13. ELSNER, M. New generation of basic carbon-magnesite refractories.

fiiaC.. Hutn., vol. 37, nos. 9-10. 1981. pp. 257-262.

11. NACAMU, R.L., and LALAMA, S.J. Magnesia refractories in basic

oxygen steelmaking. Iron Steelmak., vol. 8, no. 5. May 1981.

pp. 21-25.

15. SCHOUKENS, A.F.S. Council for Mineral. Technology (Mintek),

•
Randburg, Private corrrnunication. 1983.

16. BARCZA, N.A., and MOONEY, J.F. The p~oiection of water-cooled

plasma-generating devices. S.A. Patent 83/1374. Assignee.

Council for Mineral Technology, R.S.A., 28th Oct., 1983.

pp. 1-16 .
•
17. BARIN, 1., and KNACKE, 0. Thennochemical properties of inorganic
.
substances. Berlin, Spri~ger-verlag, 1973.· pp. 116, 119, 162,
674, 688, 689.

/'
38 I'