See discussions, stats, and author profiles for this publication at: https://www.researchgate.

net/publication/263030634

The Evolution of Søderberg Aluminum Cell Technology in North and South

America

Article  in  JOM: the journal of the Minerals, Metals & Materials Society · December 2013
DOI: 10.1007/s11837-013-0855-1

CITATIONS READS

7 982

2 authors:

Mike Barber Alton Tabereaux

Canadian Museum of Nature / Musée canadien de la nature Consultant
8 PUBLICATIONS   20 CITATIONS    97 PUBLICATIONS   685 CITATIONS   

SEE PROFILE SEE PROFILE

Some of the authors of this publication are also working on these related projects:

Aluminum electrolysis View project

Soderberg Technology View project

All content following this page was uploaded by Alton Tabereaux on 03 May 2022.

The user has requested enhancement of the downloaded file.

JOM, Vol. 66, No. 2, 2014
DOI: 10.1007/s11837-013-0855-1
Ó 2014 The Minerals, Metals & Materials Society

The Evolution of Søderberg Aluminum Cell Technology in North

and South America

MIKE BARBER1 and ALTON T. TABEREAUX2,3

1.—2468 Wyndale Crescent, Ottawa, ON K1H 7A6, Canada. 2.—1585 Wilson Lake Shores, Muscle
Shoals, AL 35661, USA. 3.—e-mail: attaber@aol.com

In the 1940s, the horizontal stud Søderberg cells were considered to be

superior to the small 30-kA Hall prebake cells, which at that time operated at
a lower current efficiency and higher energy consumption. The amperage was
first increased on the Søderberg cells from 30 kA to 50–60 kA and then later to
90–120 kA by basically increasing the anode length and the number of anode
studs. Due to the increase in demand for aluminum metal, the less expensive
Søderberg smelters proliferated in the 1940s to the 1970s in North and South
America. In the 1970s, 24 Søderberg smelters located in North and South
America had a primary aluminum capacity over 3 million tpy. The largest
operating Søderberg smelter, Companhia Brasileira de Aluminio, has a plant
capacity over 470000 tpy and started the last new Søderberg potline in 2007.
However, poor magnetics inherent with end-to-end Søderberg cell busbar
design limited any further increase in amperage while the side-to-side pre-
bake cells were able to operate more efficiently at 200 kA and higher. Com-
pared with prebake technology, Søderberg cells are now less efficient and have
higher production costs, they are more difficult to automate and they have the
greatest environmental and health challenges. Health studies from the middle
of the 1970s showing a clear link between Søderberg tar fume exposure and
the incidence of various types of cancer lead companies to propose a program
of replacement. As a result, today there are only five Søderberg smelters
operating in North and South America with a capacity of <1 million tpy.

INTRODUCTION proved to be difficult, and by 1937 the design had

The continuous self-baking Søderberg anode was
seen as a major advancement for the electrometal-
lurgical industry when it was invented by Carl W.
Søderberg in 1918. The first Søderberg anodes in
aluminum cells in the early 1920s had a circular
cross section, but operational difficulties with this
configuration led to the development of rectangular
anodes in the early 1930s. The real starting point of
the Søderberg revolution for use in aluminum elec-
trolysis cells came in 1931 at the Pechiney alumi-
num smelter at Riouperoux, near Grenoble in
France. This cell had two Søderberg anodes with
horizontal studs, or conducting pins, and it operated
initially at 22 kA although Torchet1 claimed to have
operated pots at up to 60 kA in his patent applica-
tion. The plant also developed a hooding system for
improved gas collection and working conditions. The
operation of a cell with two Søderberg anodes
been changed to a single anode.2 The horizontal
stud Søderberg (HSS) cell was such a success that
17 plants worldwide were built using the new
technology by 1937. At this point, the electrical
current in Søderberg cells had increased to about
30 kA.
In the early years, Søderberg-type pots had a
number of advantages over prebake cells, including
the following:
 Cost savings have been linked to anode prepara-
tion including both reduced capital costs for the
anode plant (no forming, baking and rodding
facilities are required), as well as reduced oper-
ating costs.
 The Søderberg anode plant is much simpler to
build; therefore, a smelter can be brought into
operation quickly.
 Anode plant production has greater flexibility; it

(Published online January 3, 2014) 223

224
is easy to start small and later to scale up
production.
 No handling and cleaning of hot anode butts is
required.
 There is no need to handle large amounts of anode
crust material or to prepare anode cover material.
 Fewer disturbances of cell heat balance occur due
to changing anodes. Early Hall prebake cells were
difficult to operate due to the large number of
anodes to change, up to 65.
 VS Søderberg cells were easier to apply gas skirts
to collect pot fumes compared with early prebake
cells which initially had no cell hooding.3
THE EARLY YEARS: 1930s to 1940s
The Søderberg self-baking anode was first dem-
onstrated in aluminum cell tests made in the plant
of the Aluminum Company of America (Alcoa) at
Baden, NC in 1924. The test was abandoned due to
its higher energy consumption and impurities com-
pared with prebake cells. The first potline of
Søderberg cells in the United States started oper-
ating at Alcoa Tennessee smelter in Maryville, TN
in 1927. These Søderberg cells had one round
(221 cm diameter) Søderberg anode electrode as
well as a circular (290 cm) cathode (shown in
Fig. 1). The operational performance of these round
Søderberg cells proved to be not competitive with
the improved Hall prebake cells at that time, and
thus, they were shut down after a very brief period.
No more Søderberg cells were built in the United
States until 1939 when Alcoa built a potline of
40 kA HS Søderberg cells at the Tennessee plant.
The first HS Søderberg cells were introduced to
Canada in 1936 when two prototype HSS cells were
tested at the Alcan Arvida. The cells ran success-
fully, and the following year, half a potline of HSS
cells was built that operated at 30 kA. In the same
year, an Alcan expert visited Europe and on his
Fig. 1. Circular Søderberg cathodes at the Alcoa Tennessee smelter
pre-1940 (Source: Alcoa).
Barber and Tabereaux
return recommended 40-kA HSS lines to be built at
Arvida and Shawinigan smelters. By 1938, two
potrooms at Arvida and one potroom at Shawinigan
were operating HSS cells at 40 kA. The results must
have been good because in the following year, the
decision was taken to build a 42-kA potroom at
Arvida.
THE SECOND WORLD WAR
The critical moment in the expansion of
Søderberg technology occurred with the outbreak of
WW II on September 3, 1939 as it almost immedi-
ately created a huge new demand for aluminum. In
both Canada and the United States, planners real-
ized that new production facilities would have to be
built as quickly as possible to supply the military
needs for the war.
Alcoa, Reynolds, and Kaiser Aluminum
Companies
In the 1940s, the U.S. federal government
embarked on a program that expanded aluminum
production capacity twofold over the next 4 years to
meet the demands of the military. The government-
owned Defense Plant Corporation (DPC) spent more
than $180 million on nine aluminum reduction
plants between 1941 and 1945, some of which were
operated by Alcoa and the others operated by
industry newcomers. During the war years, the
aluminum capacity was split roughly evenly
between Søderberg and prebake aluminum smelt-
ers. In 1946, the government-owned Søderberg
plants were first leased and later sold to Reynolds
(Longview and Listerhill smelters) and Kaiser
(Tacoma smelter).
The combined annual production of the initial
four Søderberg smelters (14 potlines and 1927 cells)
in the United States in the 1940s was 218000 tonnes
per year. A summary of the key performance indi-
cators (KPI) is provided in Table I. for the early
Søderberg cells in the United States during the
1940s:
 The HS Søderberg smelters initially started
operating at 40 kA; in the 1950s, the HSS cell
anodes were enlarged and the potline amperage
increased to about 50 kA. In the 1960s, the potline
amperage at Reynolds Longview smelter was
increased to 70 kA.
 The average annual production capacity per
potline was 15000 tonnes.
 The average energy consumption was 18.5 kWh/
kg Al.
 The average current efficiency was 84.9%.
 The HS Søderberg potlines at Reynolds Listerhill
(shown in Fig. 2) operated for 43 years, the HS
Søderberg potlines at Tacoma operated for
53 years, and the HS Søderberg potlines at
Longview operated 60 years.
The Evolution of Søderberg Aluminum Cell Technology in North and South America 225

Table I. The first four Søderberg aluminum smelters in United States in the 1940s

Startup Closure Pot Lines Cells 1940s 1970s Capacity, kWh/ CE Pot Studs
Company Location date date Years type # # kA kA tpy kg % V #
Alcoa Alcoa, TN 1939 1946 7 HSS 3 360 43 50 38676 19.3 85.0 5.50 18
Alcoa Alcoa, TN 1946 1975 29 VSS 3 360 43 60 37766 19.0 83.0 5.30 24
Reynolds Longview, WA 1941 2001 60 HSS 3 372 43 70 40440 18.0 86.0 5.19 30
Reynolds Listerhill, AL 1942 1985 43 HSS 3 595 43 55 63931 18.2 85.0 5.20 20
Kaiser Tacoma, WA 1947 2000 53 HSS 2 240 50 60 38000 18.1 85.5 5.20 20
Total 192 14 1927 218813
Average 44.4 18.5 84.9 5.28 22.4

Fig. 2. Operator breaking the side crust of a 40-kA HS Søderberg

cell at the Reynolds Listerhill smelter (Source: Reynolds Aluminum).

Alcan Aluminum
Before 1940, Alcan only had 10 prebake and
Søderberg potlines in operation 6 potlines in Arvida
and 4 old potlines in Shawinigan, with a total pro- Fig. 3. HS Søderberg potlines at the Alcan Arvida aluminum smelter
duction of only 75000 tonnes. They did not have in 1943 (Source: Rio Tinto Alcan).
enough operating experience with the new 40-kA
HS Søderberg cells at this point, so the first phase of
production increased was achieved by building six
potlines of 42-kA prebake cells in 1940. Work on the
Søderberg cells continued and by 1941, when the
greatest expansion in war-time aluminum started,
the technology was considered to be mature.
Søderberg technology was thought to be better
adapted to rapid expansion as the anode plant
requirements were much simpler. So when the call
came to increase production for the war effort, Alcan
responded by building 21 potrooms of Søderberg
cells (shown in Fig. 3) during the 1941–1943 period
that operated at 42–45 kA except for the last line
built at Isle Maligne, which operated at 50 kA, the
highest achieved by Alcan at this point. The 15 HS
Søderberg potlines at Arvida, including lines 46–57,
lines 22 and 23, and line 26 (lines 22 and 23 had
been combined into a single potline). Isle Maligne
had one HS Søderberg potline (shown in Fig. 4).
In the early 1940s, the combined annual produc-
tion for the five Søderberg smelters (24 potlines and
3775 cell) in Quebec was initially around 300000 Fig. 4. Operator breaking the side crust of a HS Søderberg cell at
tonnes per year of which all were produced with the Alcan Isle Maligne smelter (Source: Rio Tinto Alcan).
226 Barber and Tabereaux

energy from hydroelectric generating stations. Alu-
minum production continued to increase with
higher amperage and modernization.
The KPIs are shown in Table II for the early
Søderberg cells during this period in Canada:
 In the 1940s, the Søderberg potlines initially
operated 42–45 kA; the HSS cells were modern-
ized in the 1960s with amperages up to 68 kA.
 The annual production of the 15 potlines at
Arvida during this period was around
200,000 tonnes/year, or 14,000 tonnes per potline.
 The average energy consumption was high,
18.5 kWh/kg Al.
 The average current efficiency was 84%.
 The oldest Søderberg potlines at Beauharnois,
Shawinigan and Arvida operated from 63 years to
72 years.
Unfortunately, the boom did not even last until the
end of the war. By early 1944, it was evident that
aluminum production had exceeded demand as
ingots started to pile up in the smelters. Arvida
alone had a stockpile of 60000 tonnes. By April
1944, Alcan had started to shut potlines in Arvida,
but still the ingots accumulated. At the end of 1944,
Alcan had closed 15 potlines in Arvida, 3 potlines in
Shawinigan and the entire plants at Beauharnois
and La Tuque were closed; La Tuque was closed
permanently. And still the cut-backs continued;
before the end of 1945, Shawinigan was completely
shut down including the original Hall prebake cells,
and five additional potlines in Arvida were shut
down. Alcan’s annual production was at that time
similar to what it had been at the start of the war.
THE GOLDEN YEARS: 1950 TO 1970
Canada
By 1946, signs of economic recovery were grow-
ing, and Alcan, slowly at first, but then with
increasing haste, started activating their closed
capacity. Amazingly by early 1949, all the shuttered
capacity had been started, and still the demand for
aluminum increased.
Prior to this period, the majority of HS Søderberg
potlines operated at 42–45 kA, and stud planting
and pulling was done manually with simple devices.
Vertical stud Søderberg (VSS) technology was
eventually developed to operate Søderberg cells at
higher amperages, e.g., >100 kA. They had longer
anodes with additional steel studs and required
costly overhead cranes to plant and pull studs. VSS
plants were also seen as having better working
conditions.
Plans were made for further expansion and for
operating cells at a much higher current. Pechiney,
Balco, Reynolds, and Elektrokemisk had started
working on improving VSS cell technology in the
late 1940s, attempting to reach 100 kA. In 1949,
Alcan built and tested a VSS cell at Arvida operat-
ing at 42 kA. The test cell worked well and an
audacious plan was made to scale up the current in
cells by a factor of more than two, aiming for
100 kA. This design was used in 1952 at Isle
Maligne. The first half line started in May 1952, but
immediately it ran into severe operating problems
due to unexpected magnetic problems. The pot had
been built with a single-entry busbar (like all the
earlier, smaller cells). The busbar design had to be
modified to a double-entry symmetrical arrange-
ment before the second half of the line was built,
and this modified cell design was used for the new
smelter at Kitimat BC in 1954.4 Unfortunately, the
modifications were not a complete success as potline
405 was closed in 1976 and potlines 1 and 2 in
Kitimat probably have the highest specific energy
consumption of any pot in the world. When potline
405 closed, it was running at 82–83% current effi-
ciency. The operating problems in Kitimat were so
severe that it was decided to change technologies for
the second stage of construction.
Alcan representatives visited the Pechiney alu-
minum smelter at Saint Jean-de-Maurienne,
France and saw the operation of their 100-kA VSS
pots. The operation was so much better than Kiti-
mat potlines 1 and 2 that the decision was taken to
use the Pechiney VSS cell technology for the rest of
the Kitimat potlines. Thus, Kitimat potlines 3, 4,
and 5 were started in 1955–1956, using the 100-kA
VSS design, but with the potline amperage at
105 kA. The current was increased to 125 kA in the
late 1960s following a program to widen the anodes
from 80 in. to 100 in. (from 203 cm to 254 cm). Cell
start-up of a 125-kA VSS cell is shown in Fig. 5 and

Table II. The first five Søderberg aluminum smelters in Canada in the 1940s

Startup Closure Pot Lines Cells 1940s 1970s Capacity, kWh/ CE Pot Studs
Company Location date date Years type # # kA kA tpy kg % V #
Alcan Arvida 1941 2004 63 HSS 15 2000 43 62 212365 18.6 84.0 5.25 26
Alcan Schawinigan 1/1942 2014 72 HSS 4 536 43 68 57591 18.4 85.0 5.25 26
Alcan Beauharnois 1/1943 2009 66 HSS 2 268 43 68 28626 18.5 84.5 5.25 26
Alcan Tuque 11/1942 1944 2 HSS 2 268 40 42 26156 18.9 83.0 5.25 26
Alcan Isle Maligne 7/1943 2000 57 HSS 1 134 43 62 14400 18.1 85.0 5.15 26
L 403
Total 260 24 3,206 339138
Average 52.6 67.6 18.5 83.9 5.20 30.9
The Evolution of Søderberg Aluminum Cell Technology in North and South America 227

tapping metal from a VSS cell at the Kitimat

smelter is shown in Fig. 6.
There followed a lull in construction activities due
mostly to unfavorable economic conditions. The last
potlines 7 and 8 at Kitimat were built from 1964 to
1967 using essentially the same cell design that had
been used in lines 3–5. The first building, 8B, was
started with smaller anodes, but the next two
buildings, 8A and 7B, were started with wider
Søderberg anodes. These were to be the last new
Søderberg potlines to be built in Canada.
In the 1960s, a major program was undertaken in
Quebec to modernize and improve the productivity of
the HSS cells. The anodes were widened from 106 cm
to 157 cm and the current was boosted from 42 kA up to
Fig. 5. Pot startup of a VS Søderberg cell at the Kitimat smelter
(Source: Rio Tinto Alcan).
60 kA. This represented a 40% increase in productiv-
ity. For the 23 HS potrooms in Quebec, that amounted
to nearly 140,000 extra tonnes per year. As well,
improvements were made in pot-tending equipment
and better ventilation and gas scrubbing systems.
At the same time, British Aluminium was build-
ing a VSS aluminum smelter at Baie Comeau Que-
bec. The first potline started operation at the end of
1957 with a capacity of nearly 50,000 tonnes per
year. It used the same 100-kA pot that BACO
started at Lochaber, Scotland.
The combined annual production of the Søderberg
smelters at Kitimat and Baie Comeau (9.5 potlines
and 1454 cells) in the 1960s was about 445000 ton-
nes per year as shown in Table III.
 During this period, the Søderberg cells at Kitimat
and Baie Comeau operated from 105 kA to 135 kA.
 The average annual production capacity was
45,000 tonnes per potline at Kitimat and 55,000
tonnes per potline at Baie Comeau.
 The average energy consumption was high:
Fig. 6. Tapping metal from HSS cells at the Alcan Kitimat smelter 17.9 kWh/kg Al and Kitimat and 16.0 kWh/kg at
(Source: Rio Tinto Alcan). Baie Comeau.

Table III. Søderberg aluminum smelters started operation in Canada in the 1950s and 1960s

Startup Closure Pot Lines Cells Current Capacity, kW h/ CE Pot Studs

Company Location date date Years type # # kA tpy kg % V #
Alcan Isle Maligne 5/1951 2000 49 HSS 1 134 43 14400 18.1 83.0 5.15 26
L404
Alcan Isle Maligne 4/1952 1976 24 VSS 1 171 106 43694 19.1 82.0 5.25 60
Line 405
Alcan Kitimat L 1, 2 1954 2014* 60 VSS 2 342 105–113 99975 19.0 86.0 5.45 60
Alcan Kitimat L 3, 4, 5 1956 2014* 58 VSS 3 360 125–136 116413 18.6 86.0 5.40 58
Alcan Kitimat L 7, 8 1964 -1967 2010 46 VSS 1.5 210 126–135 68451 18.6 86.0 5.40 54
Reynolds Baie Comeau 1957 2014* 57 VSS 3 542 114 164383 15.8 90.5 4.80 54
Total 294 11.5 1759 507318
Average 121.9 18.0 87.1 5.30 52
*Planned.
228 Barber and Tabereaux

 The average current efficiency range was from
86.0% to 90.5%.
 The Søderberg potlines at Kitimat have operated
for 60 years and Baie Comeau operated 57 years,
but both smelters are in the process of closing all
Søderberg potlines.
From 1941 to 1980, when the prebake plant at
Grande Baie, Quebec was built, Alcan had built no
new prebake cells in Canada. It was not until 2000
that Alcan’s prebake primary aluminum production
exceeded its Søderberg production.
United States
The situation was the same in the United States as
the aluminum industry’s cycle of rapid wartime build-
up in the United States was repeated during the Kor-
ean War in the 1950s. The United States government
encouraged existing U.S. companies to construct addi-
tional reduction plant facilities due to a deficit in alu-
Fig. 7. A 150-kA side-by-side HS Søderberg cell at the Reynolds
Arkadelphia smelter in the 1980s (Source: Reynolds Aluminum).
minum availability for military purposes. The majority
of the aluminum plants were built using Søderberg
technology as at that time it was considered to offer
lower operating costs and a higher metal purity than
the alternative choice, prebake cell technology. The
atmospheric emissions from Søderberg cells were not a
concern at that time. Additionally, Søderberg cells were
less costly than prebake cells, which require anode
baking furnaces as well as anode rodding facilities.
All of these new smelters operated at higher
amperages, from 70 kA to 153 kA using longer HS
or VS Søderberg anodes with additional steel studs.
The largest HS Søderberg cell ever developed was
the 150-kA cells used at Reynolds Arkadelphia, AR,
and Corpus Christi, TX, which had Søderberg an-
odes that were 10 m in length (shown in Fig. 7).
Most new smelters were located in regions of the
United States that had inexpensive hydroelectric
power, e.g., Tennessee Valley Authority (TVA) and
Bonneville Power Administration (BPA). However,
Alcoa Point Comfort, Reynolds Corpus Christi, and
Kaiser Chalmette smelters used natural gas to
generate electricity until they closed between 1982
and 1985 due to escalating costs for natural gas.
From 1951 to 1959, 1.16 million tonnes of primary
aluminum capacity was added from six new
Søderberg smelters in the United States: Reynolds
Corpus Christi, Reynolds Arkadelphia, and Reynolds
Massena; Martin Marietta Dales; Anaconda Colum-
bia Falls; and Alcoa Point Comfort. Potlines were
added at two existing smelters: Reynolds Longview
and Reynolds Listerhill as shown in Table IV.
The combined annual production of the Søderberg
smelters in the United States (39 potlines and 5252
cells) in the 1950s and 1960s was about 1500000
tonnes per year. A summary of the KPIs for the
Søderberg potlines built in the United States in
the 1960s and 1970s shown in Table V include
the following:

Table IV. New Søderberg aluminum smelters in United States in the 1950s

Startup Closure Pot Lines Cells Current Capacity, kWh/ CE Pot Studs
Company Location date date Years type # # kA tpy kg % V #
Kaiser Chalmette, 1954 1983 32 HSS 8 1120 76 225000 16.6 90.0 5.01 36
LA
Reynolds Corpus Christi, 952 1982 30 HSS 2 280 143 73000 16.6 90.0 5.01 64
TX
Reynolds Arkadelphia, 1953 1985 32 HSS 1 160 153 62000 16.6 90.0 5.01 64
AR
Martin Columbia Falls, 1955 2009 54 VSS 2 240 108 69340 16.7 91.0 5.01 50
Marietta MT
Reynolds Longview, WA 1957 2001 44 HSS 3 504 102 138279 16.4 91.5 5.04 40
Reynolds Listerhill, AL 1958 1985 27 HSS 3 522 92 127059 16.6 90.0 5.01 32
Martin Dales, WA 1958 2000 42 VSS 2 300 108 82000 16.0 88.5 4.75 50
Marietta
Alcoa Point Comfort, 1959 1978 19 VSS 7 416 70 26000 17.0 88.0 5.01 32
TX
Reynolds Massena, NY 1959 * 54 HSS 3 504 92 123000 16.3 91.5 5.01 34
Total 334 31 4046 1159678
Average 104.9 16.5 90.1 5.00 44.7
The Evolution of Søderberg Aluminum Cell Technology in North and South America 229

Table V. New Søderberg aluminum smelters in United States in the 1960s

Startup Closure Pot Lines Cells Current Capacity, kWh/ CE Pot Studs
Company Location date date Years type # # kA tpy kg % V #
Anaconda Columbia Falls, 1965 2009 44 VSS 1 120 108 34670 16.7 91.0 5.01 50
MT
Kaiser Chalmette, LA 1967 1984 17 HSS 1 160 93 39000 16.6 90.0 5.01 36
Anaconda Columbia Falls, 1968 2009 41 VSS 2 240 108 69340 16.7 91.0 5.01 50
MT
Kaiser Tacoma 1968 2000 32 HSS 1 160 93 39000 16.6 90.0 5.01 40
Martin Goldendale, 1971 2003 32 VSS 3 526 122 168000 15.1 89.0 4.50 54
Marietta WA
Total 166 8 1206 350009
Average 104.8 16.3 90.2 4.90 46.0

 The majority of the Søderberg cells operated at
100 kA; two HS Søderberg plants operated from
143 kA to 153 kA.
 The average annual production capacity was
41000 tonnes per potline.
 The average energy consumption was high:
17 DCkWh/kg Al.
 The average current efficiency was 90.2%.
 The Søderberg potlines at Massena, NY, contin-
ued to operate after 54 years, but it is currently is
in the process of closing potlines.
Further expansion of primary aluminum smelters
continued in the 1960s although at a slower pace as
shown in Table VI. Two new Søderberg smelters
were added by Anaconda at Columbia Falls, MT,
and Martin Marietta at Goldendale, WA; additional
Søderberg potlines were added at the Kaiser
Chalmette and Tacoma smelters.
South America
The primary aluminum industry has also taken
root in South America. A small smelter initially
owned by Elquisa started operations in Ouro Preto
MG, Brazil, in 1945 using HS Søderberg technology
supplied by Elektrokemisk, but it closed in July
1946 due to financial problems. The smelter was
bought by Alcan in 1950 and was restarted in
August 1951 with a capacity of 1800 tpa. In 1956,
the production was expanded to 6000 tpa, and in
1958, a second potline Reducao II with a capacity of
23000 tpa was started using an HS Søderberg pot
design from Alcan Arvida. In June 1955, a new VS
Søderberg plant was started by Companhia
Brasileira de Aluminio (CBA) located at Aluminio
near Sao Paulo, Brazil, with a capacity of 10000 tpa
using Søderberg technology from Montecatini.
By 1960, the region was ready for rapid expansion;
in 1963, a 65000-tpa VS Søderberg smelter was built
at Vera Cruz, Mexico, using Alcoa technology; in
1965, a 30000-tpa VS Søderberg smelter was built at
Paranam, Surinam, using Alcoa technology; and the
VS Søderberg potlines at the CBA plant in Aluminio,
Brazil, were expanded. These plants were followed
soon by the construction of a new 96000-tpa VS
Søderberg plant at Poco de Caldas, Brazil, in 1970
using Alcoa technology; a new 28000-tpa HS Søder-
berg potline at Aratu, Brazil, in 1970; and further
expansions of VS Søderberg potlines by CBA in 1970
and in 1977 and a third 26500-tpa HS Søderberg
potline was built at Ouro Preto smelter in 1978.
The combined annual production of the Søderberg
smelters in the South America (18 potlines and 2676
cells) in the 1990s was about 1244000 tonnes per
year. The KPI shown in Table VI for the Søderberg
cells during in South America were:
 Seventy-two percent of the Søderberg cells oper-
ated >100 kA; VSS cells at Paranam smelter
operated at 160 kA.
 The average production per potline was 33000
tonnes per potline.
 The average energy consumption was 15.2 kWh/
kg Al with low values from 14.5 kWh/kg to
14.9 kWh/kg Al at the CBA and Aratu smelter.
 The average current efficiency was 88.8%.
 The VS Søderberg potlines at CBA and one HS
Søderberg potline at Ouro Preto are continuing to
operate.
TECHNOLOGY IMPROVEMENTS
The amperage was increased in many Søderberg
potlines by increasing the anode area to reduce the
cell resistance by making anodes wider or longer
and adding additional studs to cells. The relation-
ship between Søderberg anode (bottom) area and
amperage is in the range of 111000 cm2/cell per 100
amperes for VSS cells and 136000 cm2/cell per
100 kA for HSS cells (shown in Fig. 8); there are 42–
45 studs/cell per 100 kA. The average anode current
density was 0.90 A/cm2 for HS cells and 0.74 A/cm2
for VS cells.
Of the many technological improvements in the
Søderberg plants, one of the most significant was
certainly the introduction of lithium into the elec-
trolyte chemistry. Interestingly, Charles Martin
Hall mentioned the possible use of lithium in his
1886 patent.5 Starting in the 1960s, lithium was
230 Barber and Tabereaux

increasingly added (usually in the form of Li2CO3

Studs

4.65, 4.42 28, 30

2800, 30000 15.4, 14.9 88.2, 88.7 4.45, 4.42 32, 58

56.5
58

64
50

50
chips or pellets) to Søderberg cells to decrease their

#
fluoride emissions and increase current efficiency.
Although its use was often questioned in the alu-

4.34

5.00
4.80

5.18

4.50
Pot

minum industry, it was eventually used in all

V

Søderberg plants in Canada as well as all HS

Søderberg plants in the United States.
Lithium was used to reduce the vapor pressure
(along with reducing the AlF3 content in bath) and
88.5

89.0

88.0
88.0

91.3

88.8
CE
%

the liquidus temperature of the electrolyte and

thereby reduce the emission of fluorides from cells;
it gave reductions of 25–35% and thus allowed
23500, 26500 15.7, 14.9
kWh/

Søderberg plants to meet environmental regula-

14.5

16.9
16.3

16.0

15.2
kg

tions. In addition, lithium ions increased the elec-

trical conductivity of the electrolyte, which
permitted cells to operate at a wider ac-distance
Capacity,

and/or higher amperage with increased production.

1244170
472585

30000
65300

96000
tpy

A major disadvantage of using lithium as an addi-

tive in the electrolyte is that it results in a signifi-
cant concentration of lithium in the aluminum
metal produced in these cells, 10–20 ppm. The pre-
70, 90, 128
Current

sence of lithium in aluminum has been shown to be

64.5, 61

61, 125

110.5
160
112

128

deleterious for various uses of primary aluminum,

kA

e.g., production of magnesium-containing alumi-

num alloy sheet or plate as well as other products.
Consequently, in 1984 Alcan invented the ‘‘TAC’’
142, 168

176, 96
Cells

1508

2676

process (treatment of aluminum in crucibles) that

222

288
76
#

effectively removes lithium from aluminum by

reaction with AlF3 powder injected directly
Lines

into aluminum metal in large metal transfer

2, 7

18
#
2

1
2

crucibles.6
While substantial improvements in paste formu-
1970, 1982 2010, 2010 40, 28 HSS, VSS

lation and the gas exhaust systems had been made

type
HSS

VSS

VSS
VSS

VSS
Pot

in many plants, it was recognized that workers were

still being exposed to dust, heat, and polycyclic
aromatic hydrocarbons (PAH) emissions. As a
Table VI. Søderberg aluminum smelters in South America

55, 35*
Years

result, companies embarked on programs to develop

316
52*

43*
34
29

and implement new mobile equipment with air-

conditioned cabs. This represented very complex
engineering challenges. The old potrooms tended to
Closure

be small and crowded, leaving little space for vehicle

date

1999
1992
*

access. Wheel-breakers had been introduced in the

VS potrooms in the late 1960s (see Fig. 9), and in
the late 1970s work started on the HS vehicle fleet.
1958, 1978

Alumino, Sao Paulo, 1961–2007

Startup

New machines were introduced for crust breaking,

date

1965
1963

1970

trimming, stud planting, and pulling, and finally,

the most complex, a machine for raising the rods
connecting the studs to the anode busbar.
By the 1970s, the expansion of Søderberg cell
Paranam, Surname
Vera Cruz, Mexico

Artu, Saramenha,

Pocos de Caldas,

technology had reached 23 Søderberg smelters

Ouro Preto,
Location

located in North and South America, which had a

Brazil

Brazil

Brazil

Brazil

combined primary aluminum capacity over 3 million

tpa (shown in Fig. 10).

THE DECLINE OF SØDERBERG CELLS

Although Søderberg plants would continue to be
Vera Cruz
Company

built in the Americas after 1980 (Aratu VS Søderberg

Alumino

Average

in 1982, and a continued expansion program by CBA

Alcan

Alcan
Alcoa

Alcoa

Total
CBA

stretching from 1982 until 2007), the glory years of

Søderberg cell technology were over.
The Evolution of Søderberg Aluminum Cell Technology in North and South America 231

Fig. 8. The relationship between anode area and amperage in Søderberg cells.

A number of factors contributed to the decline, as

follows:
 Energy costs Most of the Søderberg plants had
been built when energy costs were low. As a
result, extra money was not invested to imple-
ment large busbar cross sections as high voltage
drops were offset by lower capital costs. In
addition, almost all Søderberg potrooms had an
end-to-end cell arrangement, which increased
busbar lengths, and thus voltage drops. And
finally, Søderberg cells have a lower current
efficiency than modern prebake cells, so their
energy costs per tonne of aluminum are higher.
 Amperage levels Once amperage levels started to
exceed 120–150 kA, MHD effects became too strong
to be adequately compensated for in an end-to-end
pot cell arrangement, and consequently, as Søder- Fig. 9. A wheel breaker machine breaking the side crust at the Alcan
berg cells became larger, their operation became Kitimat VS Søderberg smelter (Source: Rio Tinto Alcan).
more difficult and unsustainable.
 Health In 1975, the Canadian Federal Health
Protection Branch signaled the presence of an working atmosphere during anode baking and stud
unusual number of bladder cancer cases in the replacement (as shown in Fig. 11). By comparison,
Chicoutimi area, including Arvida.7 In 1979, Gibbs the PAH components of coal tar pitch in prebake
and Horowitz8 published a report linking Søder- anodes are destroyed when heated to 1150–1200°C
berg potroom workers to an increased risk of lung in anode baking furnaces.
cancer. In both cases, tar fumes from the Søderberg  Environment In the 1990s, the aluminum indus-
anodes were identified as being the probably toxic try also became aware of the harmful effect of the
agent. More recent studies in France, Canada, and perfluorocarbons (PFC), which are potent green-
Norway have confirmed the elevated risk of blad- house gases that develop during anode effects.
der cancer for Søderberg potroom workers. The tar Originally Søderberg smelters operated with a
fumes were found to contain PAHs. Some of its high anode effect frequency, typically 1–2 AE/
components such as benzo[a]pyrene (BaP) were pot-day, due to its large side-break and alumina
identified as being mutagenic and highly carcino- feeding operations at 2–4 h intervals. Eventu-
genic. The sources of PAH is the coal tar pitch ally, some Søderberg potlines have been retrofit-
binder of the anode. The PAH problem is unique to ted with point feeder technology to reduce anode
the Søderberg operation as they are emitted to the effects and CO2 equivalent emissions.
232 Barber and Tabereaux

Fig. 10. The expansion of primary aluminum production from Søderberg smelters, 1940–2010.

 Manpower Søderberg potlines typically use twice as

many work hours per tonne of metal as prebake
plants. In addition, due to a lack of space and the
nature of the work, it has proved very challenging
and expensive to mechanize or automate work.
The cumulative shutdown of Søderberg capacity in
North and South America from 1940 to 2013 (shown
in Fig. 12) indicates that there have been four major
crisis periods.

Period 1
In 1974, the OPEC Middle East oil embargo initi-
ated a worldwide energy crisis. Between October
1973 and January 1974, world oil prices quadrupled,
affecting the price of electrical generation. This was
followed a couple of years later by an increase in
natural gas prices. As a result, aluminum capacity
was closed in the United States, Japan, and in Wes-
tern Europe, and new capacity added since the late
1970s has been only in regions with low-cost energy.

Period 2
This period corresponds to a major increase in
power costs by the TVA as it affected smelters in the Fig. 11. Stud pulling from a VS Søderberg cell with ‘‘wet’’ anode
southeast United States with the closure of 18 pot- paste (Source: Reynolds Aluminum).
lines.

Period 3
Period 3 includes a number of factors. Skyroc-
keting electrical power prices resulted due to the
California electricity US Western energy crisis of
2000 and 2001 as California had a shortage of
electricity. Only two of the original 10 aluminum
smelters in the Northwest are still operating. In
addition, old capacity in Quebec was shut down as
new replacement smelters were built, and in Suri-
nam, the Alcoa Paranam plant was closed due to low
rainfall affecting hydroelectric power generation.
Period 4
Period 4 corresponds to the 2008–2009 global
recession that began with a marked global economic
decline beginning in August 2007 as a liquidity
crisis and took a particularly sharp downward turn
in September 2008. It resulted in the plummet in
the price of primarily aluminum from US $3,000/mt
(July 2008) to $1,340/mt (March 2009). This was
followed by record high metal inventories. LME
aluminum inventories more than tripled from 1.2
The Evolution of Søderberg Aluminum Cell Technology in North and South America
Fig. 12. The decline in primary aluminum production from Søderberg smelters in North and South America, 1940–2013.
Fig. 13. The world price for primary aluminum and smelter costs
from 1983 to 2013.
Fig. 14. Sumitomo side breaker bars on a VS Søderberg cell
(Source: Reynolds Aluminum).
233
million tonnes pre-crisis to more than 4.5 million
tonnes by the middle of 2009. Stockpiles in ware-
house climbed to record levels above 5 million ton-
nes in 2011–2013 and consequently depressed
aluminum prices. The LME 3-month price was
US$1880 in September 2013.
The major recession periods, 1980s, 2000 and
2008 with higher operating costs and continued
weakness in aluminum prices initiated the closure
of Søderberg smelters with dated technology and
higher costs in North and South America as noted in
Fig. 13.
THE FUTURE OF SØDERBERG CELLS
Of the 108 Søderberg potlines that once operated
in the Americas, only 22 potlines are still in opera-
tion, and there have already been announcements
concerning the closing of 12 potlines. Of the
remaining 12 Søderberg potlines, 11 are located in
Brazil with 9 of them being operated by CBA. By
2015, all aluminum smelters in Canada using
Søderberg technology will be closed. At that point,
all aluminum smelters in Canada will use the more
efficient and environmental cleaner modern pre-
bake anode technology.
Major aluminum companies have developed
improved Søderberg technology to reduce worker
exposure to PAH compounds in the workplace as
well as the emission of fluorides into the atmo-
sphere. For example, in the late 1970s, the Sumi-
tomo aluminum company marketed and sold their
Søderberg pot technology, which consisted of ‘‘dry’’
anode paste with a lower pitch content and the
introduction of pot crust bar breakers for automat-
ing alumina additions (shown in Fig. 14) as well as
graphitic cathode blocks and larger anodes to
reduce energy consumption.9 Dry scrubbers were
introduced to capture and recycle fluoride emis-
sions, and lithium was introduced as an additive to
234
Fig. 15. Removal of large chunks of Søderberg anode from a VSS
cell at the Kitimat smelter (Source: Rio Tinto Alcan).
the electrolyte to reduce fluoride emissions. Much
work was done improving the capture efficiency of
the exhaust systems.
With Søderberg anodes, baked carbon quality was
always an issue, especially after the conversion
to dry anode paste. And as the anodes became big-
ger, maintaining homogeneous quality became more
difficult. At the same time, the anodes became more
sensitive to changes in raw materials, particularly
coke quality. This resulted in increased suscepti-
bility to air burn, heavy dust formation, cracking,
and even the separation of large chunks of carbon
from the anode block. Dust skimming and the
removal of loose carbon became an inevitable part of
daily operations (shown in Fig. 15).
Changes were made in Søderberg paste formula-
tion and anode operating practices to lower tar fume
emissions and reduce their toxicity. Workers were
better protected as well. The use of high-perfor-
mance respiratory protection became mandatory.
Where possible, workers were inside closed cabins
provided with air conditioning and high-efficiency
filtering. To the extent possible, tasks were auto-
mated. Yet despite very significant improvements,
Søderberg plants continued to fall further behind
the performance of the new generation of prebake
plants.
These improvements lowered the PAH emissions
and made feeding the pot easier. More recently,
successful and valuable improvements have been
reported in Søderberg pot design and operation in
some countries, mainly in Norway, Spain, and
Russia. Improvements to Søderberg cell technology
were made at the Elkem Lista smelter in Norway in
the 1990s and were implemented recently at the
Alcoa La Coruna and Aviles smelters in Spain to
improve current efficiency and reduce energy
View publication stats
Barber and Tabereaux
consumption. However, the major focus of the
technology is to improve the environmental perfor-
mance by (I) better alumina feeding control tech-
nology using point feeders to reduce the frequency
of anode effects and the greenhouse gas emissions,
and (II) introducing a closed anode top to nearly
eliminate the PAH emissions.10–12 Rusal has
developed different methods for hooding and sealing
Søderberg pots and implemented point feeders and
a colloidal anode with pitch content close to that of a
prebaked anode in Søderberg potlines at the Kraz-
noyarsk aluminum smelter in Russia.13
However, Søderberg smelters are facing chal-
lenging environmental requirements as the indus-
try expects even more stringent emission
requirements in the years to come. If environmental
and working atmosphere issues are addressed and
resolved to the satisfaction of regulatory officials,
and if economic conditions improve with higher
metal prices, then Søderberg cells will certainly
continue to operate for years to come—the last
survivors of a once highly successful technology.
ACKNOWLEDGEMENTS
The book, ‘‘Global Mission—The Story of Alcan’’
was consulted extensively when writing the section
on Alcan history.14 In addition, we would like to
thank Claude Gilbert and others at the Arvida
Research and Development Centre for their help in
providing details of Alcan’s operating history.
REFERENCES
1. P. Torchet, U.S. Patent 2,073,356 (9 March 1937).
2. H. Morsel, Histoire Technique de la Production d’Alumin-
ium (Grenoble: Presses Universities de Grenoble, 1991).
3. J.P. McGeer, JOM 38, 27 (1986).
4. G. Holywell, Personal Communications (2013).
5. C.M. Hall, U.S. Patent 400,664 (1889).
6. G. Dube, U.S. Patent 4,470,846 (11 September 1984).
7. D.T. Wigle, Lancet 2, 83 (1977).
8. G.W. Gibbs and I. Horowitz, J. Occup. Med. 21, 347 (May
1979).
9. G. Holywell, The Canadian Metallurgical & Materials
Landscape 1960 to 2011 (Montreal, QC: The Metallurgy and
Materials Society of the Canadian Institute of Mining,
Metallurgy and Petroleum, 2011), pp. 177–201.
10. T.B. Pedersen, A.K. Syrdal, and A. Saethre, Light Metals
1995, ed. J.W. Evans (Warrendale, PA: TMS, 1995),
pp. 253–256.
11. T.B. Pedersen, M. Jensen, K. Kalgraf, W. Larsen, and A.T.
Olsen, Light Metals 2008, ed. D.H. DeYoung (Warrendale,
PA: TMS, 2008), pp. 239–244.
12. A.K. Syrdal, Light Metals 2002, ed. W. Schneider (Warren-
dale, PA: TMS, 2002), pp. 319–325.
13. V. Mann, Light Metals 2006, ed. T.J. Galloway (Warrendale,
PA: TMS, 2006), pp. 181–183.
14. D.C. Campbell, Global Mission—The Story of Alcan, Vol. I to
1950 (Toronto, ON: Ontario Publication Company,
1985–1990).