The Effects of Aluminum Content, Temperature and

Impurities on the Electrical Conductivity of Synthetic

Bayer Liquors

G. R. BROWNE AND C. W. P. FINN

The electrical conductivity of pure sodium aluminate solutions with compositions in the

range 93 to 128 gl-~ free soda and O to 125 gl-~ alumina was investigated over the

temperature range 40 to 70 ~ Under these conditions, the solution electrical conductivity

was found to vary as a linear function of temperature. For a given free soda content, the

temperature coefficient of conductivity varied as an approximate guadratic function of the

alumina content. A single equation was developed relating solution conductivity to free soda

content in the range 105 to 128 gl-~ alumina, 40 to 100 gl-t over the temperature range 40 to

70 ~ The addition of sodium oxalate and sodium succinate at levels up to 20 gl-~ was

found to have no measurable effect on solution conductivity. However, the addition of

sodium carbonate at levels of 40 gl-~ had a significant effect, generally decreasing the

conductivity. The magnitude of the effect is a complex function of solution composition and

temperature. The technique was used to follow the precipitation of alumina from synthetic

Bayer liquor continuously for 4 h. Possible applications to industrial practice, of the

relationship determined, are briefly discussed.

WITH large amounts of capital tied up in Bayer

process streams and with increasing economic pres-

sures, the viability of the process becomes more depen-

dent on improved efficiency. Maintenance of optimum

process conditions generally requires application of

on-line process control techniques. One parameter of

Bayer liquors which lends itself to continuous monitor-

ing is solution electrical conductivity. Bayer process

liquors are highly conductive and changes in the

conductivity of isothermally decomposing solutions are

related to the stoichiometric release of highly conduct-

ing OH - ions ~ according to the equation

2Na A1 (OH)4a q ~ A1203 9 3H20 $+ Na+OH-aq [1]

This concept is not new 2-9 but published data describing

its quantitative application to solutions of industrial

composition are inherently limited in range due to the

requirement to maintain operating conditions and gen-

erally serve as deviation indicators under conditions of

constant temperature and total soda content.

Laboratory application of this technique is demon-

strated in the literature by the authors ~~ and Kelly I~ over

a limited range of temperature, total soda and alumina

contents.

The aim of this investigation was to extended the

range of electrical conductivity data to cover solution

conditions commonly found during the precipitation

stage of the Bayer process and to investigate the effects

of some significant impurities on conductivity.

G. R. BROWNE, formerly Post-Graduate Student, School of

Mining and Mineral Technology, Kalgoorlie, Western Australia, is

presently Metallurgist, Kambalda Nickel Operations, Western Mining

Corporation. C. W. P. FINN, formerly Alcoa Foundation Lecturer in

Thermodynamics, School of Mining and Mineral Technology, Kal-

goorlie, Western Australia, is presently Associate Professor of

Pyrometallurgy, University of the Witwatersrand, Johannesburg,

Republic of South Africa.

Manuscript submitted February 21, 1980.

METALLURGICAL TRANSACTIONS B

EXPERIMENTAL

Preparation of Solutions

Pure sodium aluminate solutions were prepared by

dissolving electrical grade AI wire in analytical grade

NaOH solution. The dissolution reaction is highly

exothermic and was conducted in covered polypro-

pylene beakers, cooled and diluted to 1 liter. Impurities

were added to the prepared pure solutions as analytical

reagent grade sodium carbonate and sodium oxalate

and technical grade sodium succinate. The prepared

solutions were stable with respect to alumina trihydrate

precipitation indefinitely in the absence of seed mate-

rial.

Composition and Temperature Range

Bayer liquors are described in terms of their free soda

content (expressed as Na20) alumina content (expressed

as A1203) and total soda content (expressed as Na20).

The difference between total soda and free soda is a

measure of impurity content (mainly Na2CO3).

In this study, the range of solution compositions and

temperature used maintained consistency with pub-

lished data 1-8 describing the precipitation stage of the

Bayer Process.

Conductivity Measurements

The conductivity of the solutions was measured using

a Philips PW 9501/01 Conductivity Bridge in earthed

mode with an external reference resistance of 1.000~2

(_+ 0.01). The conductivity cell was a Philips PR

9514/10 immersion probe with a cell constant of 1.00

cm-l (_+ 0.01) at a measuring frequency 200 Hz. Con-

ductivity bridge output was monitored on a Metrohm

Labograph chart recorder.

ISSN 0360-2141/81/0911-0487500.75/0

9 1981 AMERICAN SOCIETY FOR METALS AND VOLUME 12B, SEPTEMBER 1981--487

THE METALLURGICAL SOCIETY OF AIME

Table I. Range of Variables Used in Conductivity Measurements

Variable Units Range

Free soda gNa20/1 93-128

Alumina gAl203/1 0-125

Sodium carbonate gNa2CO3/1 40

Sodium oxalate

gNa2C20~/1

Sodium succinate gNa2C4H404/1 20

Total soda gNa20/1 93-175

Temperature ~ 40-70

Temperature Measurement and Control

Solution temperature was measured with a copper

constantan thermocouple in a stainless steel sheath im-

mersed in the solution. Thermocouple output was

monitored with a Metrohm Labograph chart recorder.

The thermocouple was calibrated against a standard-

ized 76 mm immersion mercury-in-glass thermometer

accurate to ___ 0.05 ~

Solution temperature was maintained within __+ 0.1 ~

with a Grant SX forced circulation water bath. The

bath surface was covered with polypropylene beads to

minimize evaporation.

Solution Containment

On the basis of published corrosion data '2 all equip-

ment in contact with the test solutions was fabricated

from Type 304 stainless steel. This choice of material

provided:

a) a high degree of resistance to corrosion which

prevented solution contamination,

b) excellent heat transfer between the test solutions

and thermostated water bath,

c) electrical earthing of the test solutions.

Solution Agitation

Solution conductivity values increased with stirrer

speed up to a critical value beyond which it became

constant. All reported values were taken at speeds

exceeding this minimum value. Vortex breakers were

installed in the solution container to reduce the stirrer

speed required to reach this plateau region.

Solution Decomposition Test

To test the usefulness of this technique under con-

ditions in which alumina trihydrate is precipitating, a

solution decomposition test of 4 h duration was con-

ducted under the following conditions:

Free Soda (Na20) = 116 g1-1

Initial Alumina (A1203) = 122 gl-

Temperature = 69.7 ~ + 0.1

Initial volume = 1.000 1 ___ 0.005

Prior to the seed addition of 100 g of sandy alumina

trihydrate, solution temperature and conductivity were

stabilized and a 5 ml sample taken for analysis by the

modified Bushey titration method./TM

After seed addition, the stirrer was stopped for 1 min

at half hour intervals, 10 ml samples were withdrawn,

filtered through dry Whatman No. 540 paper to remove

precipitate, and 5 ml of clear solution were analyzed by

the modified Bushey method. ]3,~4

Conductivity and temperature were continuously reg-

istered on a chart recorder. Instrument calibration was

checked at 15 min intervals with a calibrated potentio-

meter. No instrument drift was observed. On comple-

tion of the test, the alumina trihydrate produced was

recovered by filtration, washed with distilled water,

dried at 105 ~ and weighed. The conductivity probe

was checked using an optical microscope at 1000 times

magnification. No evidence of contamination by pre-

cipitate was observed. Electrical calibration against 1.0

molar KCI at 25 ~ before and after the test showed

that the cell constant did not vary.

RESULTS

Pure Sodium Aluminate Solutions

Conductivity

temperature and composition for

pure sodium aluminate solutions are shown in Fig. 1.

Data from Glastonbury 9 at 25 ~ is included in Fig. 1 for

comparison. Conductivity is a function of temperature,

free soda and alumina content of the solution. In

general, conductivity increases with temperature and

free soda content and decreases with alumina content

with other parameters held constant.

.60

"50

~ .40-

> .30-

o E .20

"10"

SOLUTION

COMPOSI TI ON

(~pD

3 105 0

4 93,0

1~ 41.0

19 O 25.5

I 93.17 38.2

11 llS 63,5

105 60.7

r~ 93.0

57.0

it, i 128

100

15 i 105 77J

128

t25

17 116 97.5

15 I 93.0 81.2

tg) 116

12~

20i 105 103

Data at 25~

after GLASTONBU~Y (10)

~6 20 3b z,b 50 6b 70

T(~

Fig. 1--Solution conductivity

temperature for pure sodium alu-

minate solutions.

488--VOLUME 12B, SEPTEMBER 1981 METALLURGICAL TRANSACTIONS B

Sodium Aluminate Solutions

Containing Impurities

The conductivity characteristics of sodium aluminate

solutions with either sodium oxalate or sodium suc-

cinate additions show no significant deviation from the

pure solution data up to addition of 20 gl- l which

exceeds the normal equilibrium values in industrial

solutionsJ

However, the effect of 40 gl- l of sodium carbonate

was very significant as shown in Fig. 2. Generally, the

effect of sodium carbonate addition was a reduction in

solution conductivity at a given level of temperature,

free soda and alumina, this effect being most pro-

nounced at low temperatures and alumina contents.

Solution Decomposition Test

Figure 3 shows the variation of conductivity with

time during the precipitation test as well as the dis-

solved alumina content remaining in solution deter-

mined from both conductivity and titration. The equi-

librium alumina content of 68 gl- ~ was calculated from

the data of Pearson. l

The final alumina trihydrate product weighted 178 g

of m1203 9 3H20 of which 100g g was the original seed.

Thus 78 g of trihydrate equivalent to 51 g of A1203

precipitated during the test. Since the original solution

.60-

.50-

.40-

.30

o ~ '20-

.10-

50LUTION

COMPOSITION

(gpt)

Line 5 ~'-

105

2 116 0

3 93.0

4 128

5 105

106

128

116

li~

-- r

41.0

38.2

60.5

63,5

60.7 ~A~

57.0

77.1 ~

100

125

97.5

81.2

125

103

lb 2b 3'0 4'0 5'0 6b 7b

T(~

Fig.

2--Solution conductivity

temperature for sodium aluminate

solutions containing 40 gl -l of sodium carbonate.

130-

120

~-110-

gloo

u 90-

E 80-

f~ 7o-

m 60-

50-

40-

-- by titfimetric OnQlysis

EquJ[JbrJum A[umJno. Content

1 2_ 3 4

T!me

(hrs)

Fig. 3--Solution alumina content

time during precipitation of

AI203 9 H20.

contained 122 g of

A120 3

then 71 g remains per liter of

solution which compares favorably with the concen-

tration determined by both conductivity and titration.

No correction was made for the small amount of

precipitate removed during sampling as after the stirrer

was stopped for sampling the solids settled rapidly.

DISCUSSION

i) Pure Solutions

Least squares regression analysis of the conductivity

temperature data for the pure solutions produced a

series of linear equations of the form:

K = m T + b

[2]

whereK = solution electrical conductivity (s cm-t),

m = (OK/OT)S,A

= temperature coefficient of con-

ductivity (s ~ cm- ~~ t), and b = apparent conduc-

tivity at 0 ~ (s cm-l).

The values of these regression coefficients are con-

tained in Table II. The uniqueness of each line was

evaluated by considering the difference between either

one or both of the regression coefficients m and b for

adjacent lines. In the majority of cases, the difference

was significant at the 99 pct confidence level. The

minimum degree of significance obtained was at the 90

pct confidence level.

The value of the temperature coefficient of electrical

conductivity (m) varies with both the solution-free soda

(S) and alumina (A) contents, but the value of the

METALLURGICAL TRANSACTIONS B VOLUME 12B, SEPTEMBER 1981--489

Table II. Pure Solutions

Free Soda 93 gl-

Alumina, gl- ) 0 25.5 38.2 57.0 81.2

I03) 5.00 4.66 4,70 3.70 3.71

b 0.185 0.125 0.104 0.098 0.024

r 0.99 0.99 0.99 0.99 0.99

Free Soda 105 gl -

AI203, gl-' 0 41 60.7 77.1 103

m (x 103) 5.32 5.07 4.73 4.12 3.42

b 0.191 0.126 0.065 0.041 (-)0.097

r 0.99

Free Soda 116 gl-

AlzO3, gl -I 0 51 63.5 97.5 125

m (x 103) 5.41 5.40 5.24 3.96 3.86

b 0.203 0.098 0.062 0.018 ( - )0,011

r 0.99

Free Soda 128 gl- i

A1203, gl -I 0 41 60.7 100 125

m (• 103) 5.81 5.70 5,55 4.78 4.43

b 0.194 0.136 0.036 0.013 0.0023

r 0.99

ordinate mtercept (b) is relatively independent of S over

most of the range of A.

From this series of equations a single equation was

developed (Appendix). It predicts the alumina content

(A) of a pure sodium aluminate solution from its

electrical conductivity, temperature and free soda con-

tent.

(M4T

B4)

S 2 +

(MsT + B 5) S

(M6T

B6)

A =

(M1T + B1)S 2

+(M2 T +

B2)S +(M3T+ B3)

[31

where

M t = 7.89 X 10 -s

M 2 = --1.78 X 10 -s

M 3 = 9.76 x 10 -4

M 4 =

-5.27 X l0 -6

M s = 1.23 • 10 -3

M s = -6.50 X 10 -2

B] = 3.50 x 10 -6

B 2 = 8.17 x 10 -4

B 3 = -4.93 X 10 -2

B4 = 2,75 x 10 -4

B 5 = -6.34 • 10 -2

B~ = 3.83

The mean absolute prediction error was 1.5 pct, max-

imum absolute error 4.7 pct and error variance 2.8 pct.

Sodium Aluminate Solutions Containing

40 gl-t of Sodium Carbonate

For all but the highest concentration of free soda (128

gl-~) the addition of 40 gl-~ of sodium carbonate to

pure sodium aluminate solutions lowered their electrical

conductivity, in some cases by over 20 pct as shown in

Fig. 4 to 7. Least squares regression analysis of con-

ductivity

temperature data in Fig. 1 gave a series of

linear equations of the form given in Eq. [2]. The values

of the regression coefficient are given in Table III. In all

cases, the change in conductivity from the pure solution

values was significant at the 99 pct confidence level with

respect to either or both the regression coefficients.

Quantitative description of the effect of Na2CO 3

I>Z 20

.~ 12-

~ -

g- a-

-8-

Free 5cx:Io = 93 Ogpl

folal Soda = 116gpl

20 46 66 8C) 1()0 120 1/*0

Solution Alumina Content (gpl)

Fig. 4--Decrease in solution conductivity

alumina content for 40

g/- ~ of Na~ CO 3 and 93 g/- ] free soda.

24-

Fr6~, ~ ~ 105gt~

To~ol Soda : 12B~

(..9

g 12-

"5 .

tn 8-

~ T:~~

._~

I1J

-8-

' '(3 ' 8b ' '

I).0

0 "20 4 60 100 120

Solution Alumina Content (gpl)

Fig. 5--Decrease in solution conductivity

a]umina content for 40

gl-~ of Na2CO 3 and 105 gl- ' free soda.

490--VOLUME 12B, SEPTEMBER 1981 METALLURGICAL TRANSACTIONS B

addition at the 40 gl- l level over all ranges of free soda,

alumina and temperature was attempted with limited

Success.

An equation for the decrease in solution conductivity

24"

20,

16.

~- 4

~4-

~ -

--8-

Free 5oda, = 116gpl

Tct(zl 5oda = 139aj:~

o 2'0 16o

Solution Alumina Content (gp0

Fig. 6---Decrease in solution conductivity vs alumina content for 40

gl-I

of Na2CO3 and 116 gl-1 free soda.

~ 20-

16-

o c

12-

80-

~ T=60~ Free Sod~

(5)

= 128c1~

Sodo

= 151gp[

o 2'o 46 6'o 16o

Solution Alumina Content (gpl)

Fig. 7--Decrease in solution conductivity

alumina content for 40

gl -I of Na2CO3 and 128 glmin ~ free soda.

Table III. Carbonate Solutions,

40 g1-1

Free Soda, 93 gl-' Total Soda, 116 gl-

Alumina, gl- ~ 0 25.5 38.2 57.0 81.2

m (• 103) 4.94 4.80 4.92 4.21 4.16

b 0.150 0.086 0.066 0.063 -0.009

9 0.99

Free Soda, 93 gl- ~ Total Soda, 128 gl- t

Alumina, gl- k 0

41 60.7 77.1 103

m (• 103) 4.24 4.40 4.32 3.99 3.59

b 0.230 0.144 0.071 0.034 - 0.030

9 0.99

Free Soda, 93 gl -t Total Soda, 139 gl-

Alumina, gl- i 0 51.0 63.5 97.5 125

m (• 103) 5.52 5.61 5.37 4.31 4.40

b 0.163 0.065 0.035 - 0.009 - 0.049

9 0.99

Free Soda, 93 gl -I Total Soda, 151 gl 1

Alumina, gl- i 0 41.0 60.7 100 125

m (• 103) 10.2 8.74 7.69 5.12 3.75

b -0.078 -0.055 -0.049 -0.017 +0.032

r 0.99

Table IV. Solution Decomposition Test

Conductivity AI203 Content, gl- i

Time, h s - ~ cm- t

Conductivity Titration

0 0.260 123 122

0.5 0.286 102 102

1.0 0.349 83.5 87.0

2.0 0.370 75.0 77.5

3.0 0.390 73.0 76.2

4.0 0.398 71.0 73.0

in terms of S 2, S, A 2, A and T was obtained 15 but the

prediction was poor with deviations between predicted

and actual of between 5 and 50 pct.

APPLICABILITY

The laboratory application of technique of monitor-

ing solution electrical conductivity as a means of

following temporal variations in the composition of a

pure decomposing sodium aluminate solution is well

illustrated by Fig. 3. One important factor in this test

was that the measured conductivity was not affected by

the presence of seed particles.

The development of nonintrusive, corrosion resistant

conductivity probes, as reported by several

authors4,6,~6increases the applicability of the process to

industrial conditons. An inherent feature of Bayer

Process streams, particularly at the precipitation stage,

is that the composition changes only slowly, so it is

important that optimum conditions are strictly main-

tained. If the electrical conductivity is characterised

METALLURGICAL TRANSACTIONS B VOLUME 12B, SEPTEMBER 1981--491

over the narrow range of solution compositions likely to

be encountered, and free and total soda are essentially

constant, measurement of conductivity and tempera-

tures will give directly the solution alumina content.

CONCLUSIONS

1) Solution electrical conductivity of pure synthetic

Bayer process liquors is a predictable function of

temperature, free soda and alumina contents.

2) Additions of sodium oxalate or sodium succinate

at concentrations up to 20 gl- l have no measurable

effect on solution conductivity.

3) Addition of sodium carbonate at a concentration

of 40 gl- ~ has a significant effect, generally reducing

solution conductivity. No satisfactory mathematical

prediction equation was found.

4) Solution conductivity can be used to follow the

isothermal decomposition of pure sodium aluminate

solution in the laboratory with a high degree of

accuracy.

APPENDIX

Development of Equation 2

The lines in Fig. 1 each fit equations of the form

K -- mT + b

[1]

Least squares linear regression of m against A at each

value of S and of b against A at each value of S yield

equations of the form

(

O~--)S A

m = +mo

and

= + bo

with a high degree of correlation. This yielded coef-

ficients at each level of S.

It was found that regression of these coefficients

against S give the best fit when a quadratic least

squares regression was used. This yielded equations of

the form

~A = M~S2 + M2S + M3

m o = M4 S2+MSS+M 6

O-A = B ISZ + B2S + B3

bo = B4S: + BsS

+ B 6

492--VOLUME

12B, SEPTEMBER 1981

Then on substitution of [4] and [5] into [2] and [6] and

[71

into

[31

followed by substitution of

I21

and

~31

into

[11

yields,

K = (M~S 2 + M2S + M3)A + (M,S 2 + MsS + M6)T

+ B~S 2 + B2S + B3)A +

(B4S 2 +

BsS

B6)

[8]

which on solving for A becomes Eq. [3] in the text.

SYMBOLS

A Solution alumina content (gl- t of A1203)

b Intercept in

K vs T

regression equation (~2-~ cm-t)

Bi Prediction equation coefficient

K Solution electrical conductivity (~2-~ cm- i)

m Temperature coefficient of conductivity (~2-

cm-,oc-I)

Mi Prediction equation coefficients

r Correlation coefficient for

K vs

Tregression

S Solution free soda content (gl -~ of Na:O)

T Solution temperature (~

ACKNOWLEDGMENTS

The authors express their appreciation to Western

Mining Corporation for technical assistance and finan-

cial support of one of the authors (GRB), to Alcoa of

Australia Ltd for technical advice and to the staff of the

Western Australian School of Mines, especially Dr. T.

Pyle and Mr. L. Stonehouse for their advice.

REFERENCES

[2] 1. T. G. Pearson:

The Chemical Background of the Aluminium

Industry,

Royal Institute of Chemistry, Monograph No. 3,

London, 1955.

2. M. A. Averbikh, G. H. Birger, V. A. Nechepas, Yu. L. Romanov,

and V. T. Makhov:

Sov.J. Non Ferrous Met., Feb.

1975, vol. 16,

no. 2, pp. 31-33.

[3] 3. E. A. Khottski, V. A. Zarinskii, Yu. A. Tonkushin: July 1965, vol.

3 I, no. 7, pp. 825-27.

4. R. Rosenthal, and R. J. Kidder:

Ind. Eng. Chem.,

June 1961, vol.

53, p. 554.

5. B. Toth, F. Vallo, and E. Pungor:

Banyasz. Kohasz. Lapok Kohasz,

1975, vol.

108

no. 4, pp. 185-88. Translated by T. D. Gedeon for

Alcoa of Australia (W.A.) Ltd.

6. O. Klug, F. Kovacs, F. Farkas, and M. Gambos:

Proc. Conf. Appl.

Phys. Chem.,

vol. 1, pp. 719-26.

7. A. N. Adamson:

Chem. Eng.

June 1970, pp. CE 156-71.

8. A. N. Adamson, E. J. Bloore, and A. R. Carr:

Extractive

Metallurgy of Aluminium,

G. Stroup, ed., vol. 1 (ALUMINA), pp.

23-58, Interscience, New York, 1963.

[4] 9. J. R. Glastonbury: J.

AppL Chem.,

Feb. 1966, vol. 16, pp.

40-42.

10. G. R. Browne and C. W. P. Finn:

MetalL Trans. B,

1977, vol.

8B, p. 349.

[5] 11. J. U Kelly: Ph.D. Thesis, 1962, University Microfilms Ltd, High

Wycomb, England.

12. E. Rabald:

Corrosion Guide,

2nd ed., p. 696, Elsevier, Amsterdam,

1968.

13. H. Bushey:

Anal. Chem.,

1948, vol. 20, p. 169-72.

[6] 14. E. Kirke: Private Communication,

Alcoa of Australia (W.A.) Lid,

1976.

15. G. R. Browne: Graduate Diploma Thesis, Western Australian

Institute of Technology, Department of Metallurgy, 1977.

16. A, R. Timm: National Institute for Metallurgy Report No. 2003,

[7] Johannesburg 1979.

METALLURGICAL TRANSACTIONS B