developments in electrochemical science, engineering ... · anthracite + pitch alumina powder or...

"Developments in electrochemical science, engineering design, control, and materials

enhancing aluminium smelting technology"

Barry Welch

Barry@Barry.co.nz

Aluminium smelting has advanced considerably

2

From cells producing less < 1 tonne Al / year in the 1890’s

... to cells producing

>1250 tonnes / year today!

In China they have Run cells at 700,000A – 5.1 tonnes/day 1800 tonnes / year

Al Electrowinning is not the simple process as per texts! i.e. Al2O3 + 3/2C ≠ 3/2CO2 + 2Al with Edecomp = 1.18V

& CO not being formed electrochemically as it is “polarized”

• There are between 5 & 8 products released from the anode !

– CO2(g), CO(g), COS(g), SO2 (g), HF(g) , Cdust , COF2(g) ,CF4 (g), C2F6(g)

– & anode gases react with materials of construction

• The Electrochemical Cathode product (Na-Al alloy) undergoes numerous reactions with materials of construction. Products include:

– NaCx , NaCN, NaOH, Na2CO3, Al4C3, Si, NaxAlpSiyOz

• The cell is dependent on work practices to Remove Al, replace & protect Anodes.

• Al2O3 has to be replenished

3

Liquid Aluminium

Consumable Carbon Anode

Al2O3 dissolved in molten Na3AlF6

+

-

Importance of historically following the changes

• Understanding why we introduced – Work practices

– Control actions

• Assessing how technology & scale-up have impacted them

• Raising questions as to the relevance of retaining the old ways – Manual versus Robotics

– Need or relevance of Data processing simplifications & manipulations

• E.g.“Resistance smoothing”

4

Design & limited Theory make mistakes

5

CAPEX

• Cell cost

– Low productivity

• < 18 kg Al /m2 Potroom /day

– Short cell life

• Services

– Anode production

– Cranes and transport vehicles

– Emissions capture and recovery

OPEX

• Electrical energy

– priced on competitive demand

– greater human usage

– Limited generation capacity with low generation emissions

• Alumina

– high degree of refinement for metal applications

– hi transport cost

• Anode carbon

– Purity and structural constraints limited supply.

• Labour

– Limited automation of work practices

• repairs and maintenance

– Dusty abrasive atmosphere

The Challenges For Al Smelting

6

• Emission – Prevent HF

release – Minimize SOx

– Minimize COx

– Avoid CFx FOCx

• Waste Minimization – Prevent /Destroy

toxic materials in spent cathodes

Responsible Citizen

Energy Lowering - Faradaic Efficiency (CE%) Improvements

7

Cell Designs have Changed

8

9

Sὄderberg Technology dominates at end of WWII

…and energy consumption has been reduced

10

kWh/lb.

kWh/lb.

kWh /lb.

kWh /lb.

kWh /lb.

kWh /lb.

kWh /lb.

54 kWh/kg

Al

37 kWh/kg

Al

17.6 kWh/kg

Al

20.4 kWh/kg Al

16,28 kWh/kg Al

27 kWh/kg Al

Problems with cell Life

11

0

500

1000

1500

2000

2500

3000

3500

1890 1910 1930 1950 1970 1990 2010

Typ

ical

ce

ll lif

e (

day

s)

Year

Trends in Cell life

All Materials of construction of Cell lining are porous!

• They are wetted by liquid electrolyte! – And this enables reactions down a

thermal gradient!

• Sodium which is Co-deposited

with aluminium is the most damaging reactant - Especially with disordered carbon structures, With the product undergoing a molar volume expansion.. – Na(Al) + x C ↔Na Cx – X Increases with the graphitisation

of the carbon and reduces the expansion ,

– but being weakly bound the sodium can undergo other chemical reactions.

12

Graphitised

ordered carbon

structure

Anthracitic carbon

quality zone

Binder carbon quality

zone

Pitch Carbon

Graphitized Carbon

Cathode Technology change 1950 to 1975 Applying Materials Science

Paste Rammed cells Life <700 days

Preformed & Heat-treated

Blocks (& higher) AlF3 bath

(life> 2000 days)

Alumina Powder

Anthracite + Pitch

Alumina Powder or Refractories

Gas or electrically calcined Anthracite + pitch baked to 1100 C

Blending Materials science with modern economic management!

0

500

1000

1500

2000

2500

3000

3500

1890 1910 1930 1950 1970 1990 2010

Typ

ical

ce

ll lif

e (

day

s)

Year

Trends in Cell life

14

Summary of Cathode Materials

• Carbon materials optimization has conflicting requirements

• Complicated by jointing difficulties – Differential properties

• Galvanic & electrochemical corrosion

• Some possible design solutions with materials selection?

– Permeability modification?

15

2Al + (x+3)SiO2 (refractory Slag) + Bath ⇨ Al2O3*x SiO2 +3Si +Heat

+ molar Volume expansion

Thermite HEAVE CAUSED BY METAL FLOOD TO SUBCATHODE

Parallel Current paths Note: 4Al + 3C = Al4C3 generates 0.09 to 0.15V

A

l

A

l

Aluminium

Electrolytic Path

I2 = [(V+~0.1) / Rbath]

Electronic

Path

I1 = [ V/Rc ]

R bath

-

+

16

Growth of Al4C3 deposit in sub cathodic pores once solubility product exceeded

Note: 4Al + 3C = Al4C3 generates 0.09 to 0.15V

Na in Al

A

l

A

l

Al3+

Na +

Al 3+

Al 3+ C 4-

Ping!!

17

-

+ Anode reactions: Na ⇒ Na+ + e Al ⇒ Al3+ + 3e NaCx ⇒ Na+ +xC e

Cathode reaction: 3C +12e + 4Al3+ ⇒ Al4C3

Bath reaction: Na+ + Na3AlF6 ⇒ 6NaF + Al3+

NaCx

Na on Cathode block also causes a Growth of Al4C3 deposit once solubility product exceeded

4[Al3+] + 3[C4-] = Al4C3

Na Co-deposited with Al - or intercalated has same effect!

Na (al)

A

l

A

l

Al 3+

Al 3+

C 4-

Ping!!

Na+

Na+

Na+

Na+

Overall Cell Reaction: 12Na(Al) + 4Na3AlF6 + 3C ⇒ 24 NaF + Al4C3

18

“The cells are dependent on work practices to Remove Al -Metal Tapping”

Ladle was used for low

kA Cells

Tap-hole also used, but slowly phased out…

19

Vacuum assisted tapping

Crucible removed by manpower or….

Alumina “control” Technology up to ~1950 ( & beyond)! Volumetric side addition of the mass needed for 6 to 8 hours

20

***Ledge was pseudo-eutectic of alumina cryolite Temperature cycling lead to formation of a hard bottom alumina crust on the cathode surface!

• Three steps

– Spread Al2O3 along side of cell

– Crust break

– Wait for “anode effect”

From 1950 to 1975 – automated mechanically assisted (timed) Alumina feeders

21

• Minimum feed about 20 kg

• Typical feed 60 to 100 kg – Time interval 30 to 90

minutes typically – Mass of feed usually exceeds

the feed zone saturation solubility limit

• Normally forms sludge/muck

• More alert and than it is possible

to dissolve added during anode effect termination

1959 to 1980 – Explosive Changes

• All companies had fundamental & applied R&D activities – Explosion in fundamental knowledge , especially bath chemistry and

electrode processes

• Materials science for the space age, structure-property relationships – Better carbons & composite carbide refractories for cell lining

• Analogue control instrumentation – Solid state transistors replacing Valve

• ⇒Digital control & computing – Modelling developments

22

The industry emphasis was also to retrofit improvements. And this has continued. This led to simplifications in both Data processing and theoretical explanation. These simplifications need to be rectified for fine tune advancements

More recent advances through applying fundamental and applied knowledge (1955 to 2000)

23

Demand feed Logic –from late 1960’s Enabled smaller Al2O3 additions reducing sludge formation

• Alumina feed philosophy- Overfeed the alumina, then wait for the voltage to rise a predetermined amount

• alternative wait for the Rate of increase in voltage to exceed a certain value

24

• Early Limitations - quality of signal because of high potline AEF (including pre programmed ones) – Need to get accurate voltage rise

signal of <100 mV in order to ensure do not risk AE’s • Enter the “Resistance

Smoothing” approach

• Aim: – To compensate for impact on cell

voltage of short term fluctuations in line current for Ohmic contributions

• “Normalized Voltage”

Vn = (Vactual – Bemf)*Iavge/Iactual + B emf

With a wise word of warning

25

Mid I970’s – Signal smoothing (Tabereaux et al LM 1983 p479)

Resistance control. A common value of 1.65 V used for B EMF not converted back to voltage. R cell = [Vcell -1.65]/I actual

But the control changes are Voltages not resistances!

Cell Amperage – more recent years

Firs

t Po

int

fed

~1

96

5

Ind

ivid

ual

An

od

e A

dju

st~1

98

0

~ 1

80

kA M

ag C

om

p P

FPB

19

76

26

AP

27

-30

, Len

gth

ened

cel

ls f

or

cap

acit

y cr

eep

19

86

to

30

0kA

AP

3X

cap

acit

y cr

eep

th

ru R

etro

fits

of

larg

er

ano

des

& h

igh

er C

urr

ent

den

sity

19

90

20

05

Ch

ines

e lo

nge

r ce

lls 2

00

3 t

o 2

01

5, t

o 6

00

kA

CE% Improvements Design + Chemistry + Control

1950 Centre break, side-by-side

Cells 3% AlF3 985 oC

7% AlF3 T = 975 o C

Demand feeding needed for operating

at 7%

Point Fed Cells

10 to 12 % AlF3 955 to 965 oC

Some of CE% loss due to current

bypass / shorting! Some by Na in

Cathode!

27

Raising CE% through applying fundamental knowledge

• 1955 Grjotheim

– Demonstrated that Al dissolved in pure cryolite forming a sodium fog

• & Na flame

• Bredig et al. ORNL 1959

– demonstrated that the NaF was a good solvent for Na

• Petit Foster & Frank 1976

– presented curves showing how the Na in Al & Na3AlF6 was reduced by AlF3 additions.

– Hence lowering metal solubility and the back reaction.

AlF3 additions limited by the adverse impact on alumina solubility, and electrolyte resistance or ACD

28

~2005 High Current – Strategy basis changes!

2. Operating in a different zone of the anode potential curve

30

1. Control curve modified through: Low Anode Cathode distance High AlF3 concentration

3. Back EMF set at 1.65V brings in error

• V = Enernst + α + β*f(I) + I* Rohmic

• R = δV/δI = Rohmic + δ/ δ I {β*f(I)}

Hi AlF3 &CD Zone

Many cell conditions have changed over the years! Do our design, work ,control ,& operating and practices match the

new requirements?

31

0

0.000002

0.000004

0.000006

0.000008

0.00001

0.000012

0.000014

0

1

2

3

4

5

6

0 0.2 0.4 0.6 0.8 1

Re

sist

ance

Ce

ll V

olt

s

Fraction Target Current

Errors in Resistance calculations

R Traditional

Using 1.65V as Bemf is no longer appropriate – errors too great!

Bath volume halved in the last 25 years! Alumina removal rate in cell up 25%

And we have much more information from the cells in the computer - but do have we automatic analysis, and do we use it for problem diagnostics??

Al Electrowinning is not the simple process as per texts! i.e. Al2O3 + 3/2C ≠ 3/2CO2 + 2Al with Edecomp = 1.18V

& CO not being formed electrochemically as it is “polarized”

• There are between 5 & 8 products released from the anode !

– CO2(g), CO(g), COS(g), SO2 (g), HF(g) , Cdust , COF2(g) ,CF4 (g), C2F6(g)

– & anode gases react with materials of construction

• The Electrochemical Cathode product (Na-Al alloy) undergoes numerous reactions with materials of construction. Products include:

– NaCx , NaCN, NaOH, Na2CO3, Al4C3, Si, NaxAlpSiyOz

• The cell is dependent on work practices to Remove Al, replace & protect Anodes.

• Al2O3 has to be replenished

32

Liquid Aluminium

Consumable Carbon Anode

Al2O3 dissolved in molten Na3AlF6

+

-

Some Reactions occurring in an Operating Cell – depending on anode Potential Gradient?

Al Producing Reactions in the Smelting Cell (ΔGo

reaction) (ΔHo

reaction) Enabling Potential

Completion Potential

kJ at 960oC

kJ at 960oC Volts Volts

0.75Al2O3*H2O + 3C + 1/2AlF3 = 2Al + 3CO(g) + 3/2HF(g) 1537 0.86 2.65

Al2O3 + 3C + 3S = 3COS(g) + 2Al 1219 0.910 2.10

Al2O3 + 3C = 2Al + 3CO(g) 1351 1.072 2.333

Al2O3 + 3/2C = 2Al + 3CO2(g) 689 1099 1.189 1.899

1/2Al2O3 + 3/2C + Na3AlF6(l) = 2Al + 3/2COF2(g) + 3NaF(l) 1078 1438 (1.85V*) Note : the chemical reaction

2COF2(g) +C = 2CO(g) + CF4(g) Is enabled, as soon as surface intermediates are formed. “Under-potential evolution”

ΔGo= -46kJ 123 (~1.74*V)

2Na3AlF6 + 3/3C = 3/2CF4(g) + 2Al + 6NaF 1483 1940 2.563 3.356

* Imposes limit on Electrode Potentials

The difference between actual and completion electrode potentials can be provided

by heat transfer to the electrode interface

This reaction almost certainly never occurs in

present installed cells!

33

Enabled is one thing, completing reaction/product

formation is another!

For Al Smelting Anodes. • Sufficient energy must cross the

interface boundary to COMPLETE product formation!

– ΔH reaction needed -here

represented as an equivalent

Voltage even though energy in

excess of Δ G can be either as

V or heat!

• A mixture of products always

formed electrochemically at

anode! • An indication Heat transfer to the

carbon anode may be limiting!

CO+CF4 Enabled Via COF2 ~ 1.82 V

CO+CF4 Complete ~ 2.35 V

CO Complete

~2.33V

CO enabled ~ 1.1 V

CO2 enabled ~ 1.2 V

CO2 complete ~ 1.89 V

DDC

Independent

CF4 Enabled ~2.45 V

Independent CF4 complete ~ 2.95 V

34

Another Work-practice- Replacing and protecting Anodes. .. Time consuming, introduces disturbances and operating problems!

35

In a multi-electrode system - the current & Reactions between individual anode pairs is not controlled,

only the total current through the cell!

36

Fast-forward to today – Hi Productivity • Cells for higher productivity and

lower energy consumption

• Much reduced electrolyte volumes so rate of change of concentration is faster!

• Working potential band for avoiding anode effects is reduced

• Alumina solubility is reduced by higher aluminium fluoride concentrations

• Power supplies are more stable and anode effect frequenciess substantially reduced – Do we still need resistance

smoothing?

• The control band is tighter

37

Maximum PFC’s zone* Direct CF4 Enabled

Background PFC’s zone* Extra CO + CF4 (via COF2) enabled

T <~ 975oC AlF3 < ~7% iD < ~0,73 Al2O3 > ~1.8%

T < 975 -955oC AlF3 < ~7 -12% iD < ~0.73 – 0.88 Al2O3 > ~2%

T < 955 -935oC AlF3 < ~12 -16% iD < ~0.9 – 1.1 Al2O3 > ~2.2%

Cell control band! 1.75V

~1.80V

2.35V

Better signal Processing needed to avoid PFC’s

• Change due to electrode process – not resistance!

• Need to limit rise

• Monitor under/overfeed duration to stop going to too low alumina concentration

• Uniform feeding & mixing more important!

38

Individual Anode Current Monitoring – an added control tool!

• Helps determine changes in spatial conditions in cells. – Minimize background PFC’s

– Highlight equipment & work-practice errors

• Marginal gains in CE%

– But greater gains can be derived by lowering Voltage for Minimum energy/kg

• But a revision & update of control strategy also needed

39

Nothing is New – 1987 !!!!

Minimizing energy means less emissions & greater efficiency!

But there are controlling laws that are invariably ignored!

& CE% improvements are of less value than lowering voltage!

Cell-Cell Volts Volts (at Cell) Energy in = Vat cell*I

Vcell-cell = V at cell + V external I*V external = I2 * R external - Wasted Energy

Lowering energy through Voltage Reduction kWh/kg = 298* V cell-cell /CE%

• The Aim: – To reduce energy

• The First Question - how is the energy used at the cell?

41

Utilisation of energy at the cell (/t Al)

42

• Preheat all input Materials – impure alumina

– Gross carbon anode

– make up fluorides

– Anode cover material

• Thermal energy to maintain the reactor temperature

– The cell heat loss

• Enthalpies of all

conversion reactions

– making aluminium forming both CO and CO2 gases

– forming COS

– evolving HF

1.89 t Al2O3

~7.6 kg Na2O

~ 20 kg H2O(l)

~80 kg Recycle fluorides

All at ~25oC

500 to 600 kg C

10 to 20 kg S

All at 25oC

150 to 250 kg

Anode cover Material

~ 15 to 30 kg for bath

Make All at ~25oC

12 to 20 kg AlF3 at 25oC

Heat losses include impact at anode change

43

Process Outputs (Reaction Products) ( Typical Values / tonne Al Produced)

Energy In = V x I x t

~ 12,700 to 14,500 kWh

Cell Gases 1110 to 1350 kg

CO2

120 to 300 kg CO

~ 30 kg HF

~ 20 kg COS

960oC

~ 100 kg Cover @ ~600oC

~ 100 kg [Gross-Net] C @ ~900oC

Bath tap

~ 15 kg liquid bath

@ ~960oC

Metal tap

1 ton liquid Al

@ ~960oC

43

Materials Preheat Energy Demand

Material kWh/ t Al

Range Comment & Reduction Options Gains?

Secondary Alumina 690 ± 50 Partial pre-heat? Geothermal energy source above 400°C ?

- 300

Gross Carbon 215 ± 30

Better shape design for retaining covering material? Better anode setting pattern for covering? Reduce carbon reactivity towards CO formation? Heat source to achieve temperature above 450°C?

-25 -140

AlF3 & Anode cover Material

25

± 10 Develop a recyclable Alternative?

-10

Does Iceland have an opportunity to get geothermal energy at the required temperature? What would be the new design of the pre-heater and anode handling?

44

Reactions conversion Energy Demand

Reaction kWh/ kg Al

Range Comment & Reduction Options Gains?

Aluminium production xAl2O3(s) + 3C = 2x Al +3COx (g)

5810* ± 100 Depends on Net Carbon consumption & fraction of metal formed with CO(g) as net reaction product* Reduced by more calcined less reactive anode's of uniform reactivity. Better control of cell stability and conditions.

-90 -100

Formation of Sulphur products, HF, & Na that enters cathode

20 ± 10

Little flexibility as controlled by raw materials

0

• Based on lowest net carbon achieved (392 kg C/t Al) and CO /CO2 ratios measured in operating cells .

45

The concluding challenges for Energy

• How can you eliminate the extra energy required through materials impurities – Especially S & HF?

• How can cell heat-loss be reduced? – Can the need for ledge be

removed?

• Can the feed materials be preheated using “free” thermal energy?

46

In conclusion

• Times have changed – Simplifications are not necessary and now harmful

• Many academic interpretations have been made for experimental designs and conditions that ignore limitations of operations. – Academics must work closely with ,but with open eyes open

– And a multi discilinary approach is needed!

• Robotics, and automation are lagging – Including ones that lead to energy savings

But success stories exist!

47

Note: Some CO(g) is formed directly electrochemically despite the literature implying otherwise! (Rest by “Back-Reaction”)

48

CO limited by Surface coverage/Desorption

CO2

Cell Heat loss

• Reductions constrained by the need to maintain side ledge. – Benefits of increasing side

ledge diminish for large cells.

– Depends on whether operating for high productivity(& CE%) or minimum energy.

• Significant further reductions require totally new cell design concepts and operating practices – automation of work practices

– robotics for precision

49

4

4.5

5

5.5

6

6.5

7

7.5

8

8.5

0 100 200 300 400 500 600

kWh

loss

/ k

g A

l

Operating Current (kA)

Trends in Cell Heat loss (kWh/kg Al)