Dispersion-flocculation studies on hematite-clay systems

International Journal o f Mineral Processing, 11(1983) 285--302

285

Elsevier Science Publishers B.V., Amsterdam -- Printed in The Netherlands

DISPERSION-FLOCCULATION STUDIES ON HEMATITE-CLAY SYSTEMS

B. GURURAJ, J.P. SHARMA, A. BALDAWA, S.C.D. ARORA, N. PRASAD* and A.K. BISWAS**

Department of Metallurgical Engineering, Indian Institute of Technology, Kanpur-208016 (India) (Received July 16, 1982; revised and accepted April 18, 1983)

ABSTRACT

Gururaj, B., Sharma, J.P., Baldawa, A., Arora, S.C.D., Prasad, N. and Biswas, A.K., 1983. Dispersion-flocculation studies on hematite-clay systems. Int. J. Miner. Process., 11 :285--302. Studies have been made on the separability of clay minerals such as kaolinite, illite and montmorillonite from hematite in dispersant-starch flocculant systems. The grossly different dispersibility of hematite from that of clay minerals aided separation by selective dispersion and floceulation. Moderate success has been achieved with a selectivity index nearing 4.0 (average recovery values around 80%). The studies have been extended to hematite recovery and clay rejection from the slimes of the Barsua iron ore washing plant owned by Rourkela Steel Plant, India. Limited success achieved in the starch selective fiocculation method has been attributed to the difficulties associated with fine grain size, clay mineralogy and liberation. The ore exhibits the p h e n o m e n o n of differential grinding. Hematite-rich coarser particles in the slime can be separated by differential settling in dispersant systems followed by selective floeculation in low-starch systems.

INTRODUC~ON

Indian iron ores contain fine-grained clay minerals that present considerable difficulties in beneficiation and lowering of aluminium content in the blast furnace feed. Attempts to wash Indian iron ores for the purpose of removal of alumina-through selective dispersion and settling-have met with only limited success (Viswanathan and Paranjpe, 1968; Yu, 1968). Selective flocculation has been considered to be one of the most promising avenues for beneficiating fine particles. A useful review records the underlying principles and commercial applications of this beneficiation tech*In-charge, Raw Materials Group, R & D Centre for Iron and Steel, Steel Authority of India Ltd., Ranchi, India. ** To whom all communications should be addressed.

0301-7516/83/$03.00

© 1983 Elsevier Science Publishers B.V.

286 nique (Read and HoUick, 1976). The Selectivity Index (S.I.) is the geometric mean of Rvm/(lOO-Rvm) and Rlvm/(lOO--Rlvm) where Rvm and Rlvm are the percentage recoveries of the valuable mineral in the concentrate and of the less valuable mineral in the gangue (Gaudin, 1939). There exists very little scientific literature on the separability of pure hematite fines from kaolinite, montmorillonite and illite fines of different sizes through (a) selective dispersion and settling and (b) selective flocculation. Our studies were designed to bridge this gap in the literature and also to elucidate the inter-relationships between stages of cleaning, grade, recovery and selectivity index. The studies were extended towards hematite recovery and clay rejection from the slimes of the Barsua iron-ore washing plant owned by Rourkela Steel Plant, India. The aluminium-containing phases in the said slime were earlier characterised as -2pm grains of kaolinite, illite and montmorillonite (Gururaj et al., 1979) -- the detail matching with Kitchener's (1977) description of a typical natural slime ' h o t easily handled by conventional mineral processing techniques". MATERIALS The following materials were used: (a) Hematite, kaolinite, montmorillonite and illite - - p u r i t y established by X-ray analysis; hematite was supplied by Thomas Baker & Co., and the rest by Industrial Minerals and Chemical Co. Pvt. Ltd., Bombay. The minerals were separately ground in ceramic jars and size~fractionated. (b) Barsua iron ore (-2 ram) slime ground to below 200 mesh and then size-fractionated by hydraulic settling. Densitimetric sub-fractionation by acetylene tetrabromide and hematite-dissolution of subfractions by Jackson's Citrate-Bicarbonate-Dithionite (CBD) method (1975)provided liberation data. While 0--74 #m material assayed 70.4% Fe203, 10.0% SiO2 and 10.16% A1203, the 1--20 pm fraction assayed 70.7% Fe203, 10.6% SiO2 and 14% A1203 and the 1--8 pm fraction assayed 65.6% Fe203, 8.8%SIO2 and 18.2% A1203. (c) Commercial starch and starch-phosphate: starch solutions were prepared by different methods such as "causticizing": "homogenizing", "causticizing-homogenizing" (15 rains, in blender at 16 000 rpm) and "modified causticizing-homogenizing" (6 rains ageing of the causticized gel and 5 mins homogenizing). Starch phosphate was prepared by phosphorylating starch with mono- and di-sodium hydrogen phosphate at 130--135°C (Whistler, 1964). Whereas raw starch assayed 0.012% phosphorus, the phosphorylated starch assayed 3.4% P. (d) Sodium silicate, sodium pyrophosphate and versicol W-13 as dispersants; sodium fluoride, sodium fluorosilicate, calcium chloride, hydrochloric acid and caustic soda as modifiers.

287 EXPERIMENTAL EQUIPMENT AND METHODS Column studies for settling were undertaken with - 2 0 + 5 micron particles of pure minerals. Settling experiments were carried o u t in a cylindrical column of 50 cm height, 10 cm diameter with 5 outlets at equal spacings of 10 cm: the outlet tubes were 5 mm in diameter. A suspension of 10 mg cm -3 concentration of sized particles was thoroughly stirred, taken into the column and allowed to settle. Around 40 cm 3 pulp samples were withdrawn from different depths and at different intervals of time, solid contents of these samples were estimated gravimetrically. Typical plots showed concentration profiles at different time intervals from which 'Stability Numbers' Sh,tn of the suspensions for depth h and time tn were c o m p u t e d as defined by Zanoni et al. (1975). Flocculation experiments on pure minerals were done in a 3-cm-dia., 16.5-cm-high cylindrical jar in which the height o f the 100 cm 3 pulp was 11.5 cm, and the sampling port height from the base was 3.5 cm. Accurately weighed (around 0.9--1.0 g) mineral powder was dispersed in the pulp with dispersant for 1 min. Then the dose of causticised starch was added and mixed; 3 minutes' time was allowed for settling of floc. The collecting port was opened to collect the unflocculated/unsettled particles which were dried and weighed. In some experiments causticized and homogenized starch solutions were stored for 24 hrs prior to use and only 1 min time was allowed for settling of flocs. Modifiers such as CaCI2, acid or alkali, dispersants etc. were added and mixed for 1 min each prior to addition of starch. Flocculation experiments with synthetic mixtures (50 : 50) of pure minerals (1--8 pm size) and Barsua ore slimes were performed in a flocculating column o f 11 cm internal diameter and 24.5 cm height. The tapping port was at a distance of 19.5 cm from the t o p of the flocculating column. Suspensions of t w o pure minerals were mixed to constitute the total pulp of 1600 cm 3. The final pulp concentrations -- for all experiments on synthetic mixtures or slimes -- were adjusted to be 1%. Sodium silicate or pyrophosphate solution was added to the pulp and mixed for 7 min. This was followed by one minute's conditioning with each of the successively added reagents such as NaF, starch solution etc. The flocs were allowed to settle for 1 min. Dry flocs and the unflocculated fractions were weighed and chemically assayed. Some three-stage flocculation experiments were conducted by re-dispersing and re-flocculating the flocs F1 to F2 and then to F3 or "A". For each fractionation, grades of the fractions and the mineral recoveries were computed to yield the successive selectivity indices. The three unflocculated portions were mixed and subjected to flocculation giving floc "B". The final unflocculated residue was collected and labelled as "C". Some two-stage and six-stage flocculation experiments were also conducted. In each stage, additional and identical dosages of reagents were used. All the fractions were dried, weighed and analysed in terms of Fe203, SiO2 and Al~O3 content.

288

Some preferential settling experiments were done in the apparatus described before with starvation dosage (very low concentration) of starch, when settled particles could not be said to correspond to stable flocs. Another set of settling experiments was conducted in a 12-1 stainless-steel cylindrical column of 63 cm height in the cylindrical portion, 12 cm internal diameter with a conical bottom. In the cylindrical portion, there were 3 taps T1, T2, and T3 at heights of 41 cm, 21 cm and 7.5 cm respectively, from the base of the cylindrical portion, and the conical part was fitted with tap T4. The central shaft was fitted with a stirrer at the base of the cylindrical part. For one experiment, an additional stirrer was fixed on the shaft 53 cm above the b o t t o m stirrer. A sample o f 1 5 0 - - 2 0 0 g lot of Barsua slime was mixed and stirred with a total of 12 liters of water for 10 min. Then the stirrer was stopped and one minute's settling time was allowed. Fractions were collected through T1, T2, T3 and T4 in succession, T4 collecting the most quickly settling heavy fraction. The solid contents in these fractions were filtered, dried, weighed and chemically analysed. While the first experiment was performed with two stirrers operating at 1 5 5 0 RPM, the second experiment was done with only one stirrer at the bottom, operating at 2 2 5 0 RPM.

LEGEND

18

16

o -~--5

0 mt 1 mt mt

• --

15 mt

A -? -

30 mt 60mr

For depth h c m ond time interv(]t t n minutes, stobitity number

Sh'tn =

Area OPEn On.xlO 0 AreaOPAB

u12 (J

}-.

:El0

A1

g ~a

m l

tn

$6 u z

3 4

0/

i

0

I

i

10

20 DEPTH

30 IN CM

Z,0 --

J

,

f

50 p-

Fig.1. Settling characteristics of kaolinite with dispersant (Na2SiO~).

30.7 12.1 7.1 24.4 15.3 14.4

50.5 22.2 10.0 70.2 26.5 10.0

53.9 58.8 6.7 61.3 54.4 7.5

84.2 28.3 10.0 91.3 31.1 10.8

VW13 73.3 49.9 -75.0 57.9 --

80.6 73.5 31.9 83.5 73.3 27.5

S.P. 56.1 53.4 8.3 69.9 47.9 7.5

S.S.

N.D.

S.S.

N.D.

S.P.

Montmorillonite

Hematite

98.6 93.9 83.9 99.0 95.5 90.4

VW13 84.2 85.5 41.7 95.4 89.0 45.3

N.D. 88.9 75.2 44.4 83.4 79.6 51.5

S.P.

Kaolinite

92.8 63.4 11.7 94.1 69.0 13.5

S.S.

99.9 82.8 25.1 99.9 92.1 23.2

VW13

73.3 62.4 29.1 86.0 69.3 41.8

N.D.

Illite

72.2 52.6 34.4 76.3 59.4 61.2

S.P.

66.7 53.6 19.5 72.9 56.4 22.7

S.S.

92.8 69.7 38.9 92.8 79.8 47.9

VW13

N.D. m e a n s n o d i s p e r s a n t ; S.P. a n d S.S. m e a n 4 X 10 -4 M/I s o l u t i o n s o f s o d i u m p y r o p h o s p h a t e a n d s o d i u m silicate r e s p e c t i v e l y , V W 1 3 m e a n s 4 p p m o f Versicol W13.

$40t3 o

S4ot s

$30tl S3ots $3ot3o S~otl

8h, tn

S t a b i l i t y n u m b e r o f m i n e r a l s u s p e n s i o n s w i t h a n d w i t h o u t d i s p e r s a n t s (Sh,tn h in c m a n d t n in m i n u t e s )

TABLE I

(30 ~O

290

T A B L E II S o m e flocculation results E x p t . Flocculation s y s t e m s 1 2 3 4

H, 5 - - 2 0 p p m , SS M, 5 - - 2 0 H, 5--20 H, 5 - - 2 0

urn. 100 um urn, pro,

ST (C) 0 - - 2 0 0 and 400 p p m

5

H, 0--2 urn, ST (MCH)

ST (C) STP

Results O p t i m u m 400 p p m ST, 100 p p m SS better Very little flocculation O p t i m u m 40 p p m ST (C), non-ageing better O p t i m u m STP 10 p p m , non-ageing better, little flocculation for M O p t i m u m 40 p p m ST (MCH), 7.3 pH better t h a n 10.4

( A l l s u b s e q u e n t expts, were with MCH starch)

6

H, 1--8 u m ST (MCH)

7

H, 2--20 u m

O p t i m u m 20 p p m ST (MCH), effect of SS and Ca 2. small O p t i m u m 40 p p m ST, 7.2 pH better at 100 p p m SS (Fig.2).

Single-stage selective flocculation o f 50 : 50 m i x t u r e s 8 H/K, 1--8 u m , 40 p p m ST O p t i m a l grade at 100 p p m SS: G 72, R 78, SI 2.86 9 H/K, 1--8 urn, 40 p p m ST Satisfactory grade and recovery at 100 p p m SS, 50 p p m SF: G 78, R 78, S.I. 3.58 (Fig.3) 10 H/K, 1--8 urn, 40 p p m ST O p t i m a l grade at 50 p p m SP: G 70, R 76, S.I. 2.6 11 H/K, 1--8 urn, 40 p p m ST O p t i m a l grade at 50 p p m SP and 50 p p m SF: G 76, R 74, S.I. 3.1 12 H/K, 1--8 urn, ST 40 p p m , Steady fall in recovery and grade with inpH 10.5, SSF 0 - 100 p p m crease in SSF c o n c e n t r a t i o n . 13 H/M, 1--8 urn, ST 0 - - 1 0 0 p p m , O p t i m u m ST 40 p p m : G 65.3, R 55.8, pH 7.3, no dispersant S.I. 1.73 14 H/M, p H 7--10.5, ST 40 p p m O p t i m u m pH 9.5: G 66.8, R 58.1, S.I. 1.85 15 H/M, SS 100 p p m , ST 40 p p m , O p t i m u m SF 50 p p m : G 76, R. 73.7, S.I. SF 0 - - 1 0 0 p p m 3.12 (Fig.3) 16 H/M, pH 9.5, ST 40 p p m , O p t i m u m SS 100 p p m : G 71.7, R 64.1, SS 0 - - 1 5 0 p p m S.I. 2.3 17 H/I, 1--8 urn, pH 7.3, O p t i m u m ST 40 p p m : G 67, R 61.4, S.I. no dispersant 1.92 18 H/I, 1- -8 urn, ST 40 p p m O p t i m u m pH 9.5: G 68.3, R 62.7, S.I. 2.03 pH 7--10.5 19 H/I, 1--8 urn, ST 40 p p m , pH 9.5, O p t i m u m SS 100 p p m : G 73.7, R 69.3, SS 0 - - 1 5 0 p p m S.I. 2.6 20 H/I, 1--8 urn, ST 40, SS 100 p p m , O p t i m u m SF 50 p p m : G 77.6, R 76.0, pH 9.5, SF 0 - - 1 0 0 p p m S.I. 3.36 (Fig.3) Three-stage flocculation 21

H/K, 1--8 u m pH 7.3 Starch 20 p p m No dispersant

Reciprocal relationship b e t w e e n grade a n d recovery (Fig.4). Best grade 75% a n d S.L2.67

(Abbreviations: H = hematite; K = kaolinite; M = montmorillonite; I = illite; ST = starch; C = causticized; HM = h o m o g e n i z e d ; C H = c a u s t i c i z e d - h o m o g e n i z e d ; MCH = modified

c a u s t i c i z e d - h o m o g e n i z e d ; STP = starch p h o s p h a t e ; SS = s o d i u m silicate; SP = s o d i u m p y r o p h o s p h a t e ; SF = s o d i u m fluoride; SSF = s o d i u m silico-fluoride; G = grade in % h e m a tite; R = p e r c e n t recovery o f h e m a t i t e ; S.I. = selectivity index).

291 EXPERIMENTAL

RESULTS

Settling tests yielded particle concentration-height plots such as the one given in Fig.1 from which "stability numbers" Sh,tn of the suspension for depth h cm and time tn min. were c o m p u t e d as defined by Zanoni et al. (1975) and illustrated in Fig.1. Sh, tn values for different minerals in dispersant solutions are presented in Table I. Results of some key experiments for flocculation of pure minerals and selective flocculation of mineral pairs are presented in Table II. Starch flocculates hematite appreciably but montmorillonite only feebly. The m e t h o d of preparation of starch solutions was found to be important; mere homogenisation did not give satisfactory results. For causticised starch, a 40 p p m concentration was optimal for flocculation of hematite. Ageing of solutions gave inferior results. Starch phosphate was found to be a strong flocculant b u t ageing gave inferior and less reproducible results. The modified causticizing-homogenizing (MCH) method for preparation of starch solution gave the best reproducible results. 0--2 ~m hematite particles were found to be less flocculable than 5--20 pm particles, particularly in the presence of sodium silicate. For 1--8 pm particles, a 20 p p m concentration of starch was optimal and the effects of Ca 2÷ and sodium silicate were n o t pronounced. For 2--20 p m hematite particles, optimal concentration of MCH starch in the presence of 100 ppm sodium silicate and varying pH was around 40 ppm (Fig.2). 100 O

9O

2

-

O

80

u

o 70

i60

~ so ~ ~o 30, o

pH :

7.2 ,

z~ pH : 10.5,

20

N o 2 S i O 3 : 100 p p m

Na2SiO 3 : 1 0 0 pprn

10--

0

I 0

I

I

20 40 60 Starch concentration in p p m

I 80

100

F i g . 2 . E f f e c t o f s t a r c h c o n c e n t r a t i o n and pH o n f l o c c u l a t i o n o f pure h e m a t i t e .

292

~" so g c

pH : 9.5, s t a r c h : 40 N a ~ ] 3:100 ppm mont~or.,oo~te . . . . 0 Grade .. . Kaolinite A Recovery Kaolinite iL~te ~-:.;~Grade

I o so I/ /

/

;

i,

A, Rec,overy, ,

5

0

0

,

I I L I

I

,

I

~

20

40 60 80 NoF c o n c e n t r a t i o n in ppm

100

Fig.3. Effect of different dosages o f NaF on the grade and recovery of hematite in flocculated part in a synthetic mixture (50 : 50) of pure hematite and clay mineral (1--8 #m) (kaolinite/montmorfllonite/illite). 80

I

I

I

I

4

I

E m c¢

oo.

.c

30

_.a 70

i

¢J .E

E 60

2

pH of pu[p: 7.3 Starch : 20 ppm o R e c o v e r y - grade A Selectivity index - r e c o v e r y

so

I

4O % recovery

I

I

I

I

50 60 of hematite in flocculated

7O

part

Fig.4. Effect of cleaning on the recovery, grade and selectivity index m hematite-kaolinite (50 : 50) mixture of 1--8 ~m particle size.

Single-stage selective flocculation experiments were performed on the 1--8 ~ m size particles of the following mineral pairs: hematite-kaolinite, hematite-montmorillonite and hematite-illite. Better results were obtained with starch concentrations around 40 p p m in the presence of 100 p p m sodi-

293 un silicate, 50 ppm sodium fluoride and pH 10.5. The best result obtained was for the 1--8 ~m hematite-kaolinite system (Fig.3) where both grade and recovery of hematite were 78% and the selectivity index was 3.58. Comparable values were obtained for montmorillonite and illite (Fig.3). Threestage flocculation experiments without dispersant revealed reciprocal relationship between grade and recovery. Best values of grade (75%) corresponded to 47.6% recovery and S.I. 2.67 (Fig.4). ~" Some floc-flotation experiments were performed on a 1--8 pm synthetic mixture of hematite (5/8th of the total weight) and the three clay minerals (1/8th of the total weight for each mineral). Magnafloc 292-a cationic flocculant patented by Allied Colloids Ltd., U.K. was used (1 ppm concentration) at pH 8.5. The floc was floated with n-dodecylamine hydrochloride collector. Dependence of recovery of F%O3, the grade and selectivity index on collector concentration are shown in Fig. 5. At 10 -5 m/1 concentration of collector, optimal grade (74.6%) and selectivity (1..95) were obtained and hematite recovery was 74.6%. Increasing cationic collector concentration reduced recovery of Fe203 in the unfloated part -- Fe203 probably tending to float along with clay. Since the separation involved the efficiency and selectivity of two steps (viz. flocculation and flotation) instead of one, S.I. values were inferior to those obtained for flocculation of mixtures of pure minerals. Selective flocculation was attempted on 1--20 pm Barsua slime by starch (Fig.6) and starch phosphate (Fig.7). Since finer particles are likely to correspond to better liberation, subsequent experiments were performed with 1--8 ~m particles without modifi-

- 1

0

0

~

12.0

SO

~,

t~

1.0 ~ 'r.

"~ 70

,.

B eJ

g

60

g ~

50

10-6 5x10-6 10-s 5x~0-s Collector(N-OodecylAmineHydroChloride) concentrotion

Fig.5.

Effect

in mo[es/titre

o f collector concentration on the flocculation o f 1 - - 8 ~ m synthetic mix-

ture of hematite, illite, kaolinite and montmorillonite. Flocculant 1 ppm, pH 8.5, pulp concentration 1%.

294

80 I-

30 )25

; °°I-I

~

/

__

I

I

,or ~,3o~o

o~

~,

o

I/ 0[ 0

. I

I 10 Starch

;

--I 2.o ~

~1o ~ s . I .....

-

-~o.s

I

I I I I I I I0 20 30 40 50 c o n c e n t r a t i o n in pprn

Fig. 6. E f f e c t o f c o m m e r c i a l starch c o n c e n t r a t i o n o n f l o c c u l a t i o n o f 1 - - 2 0 / ~ m Barsua

slime, pH 10.5, pulp density 1%, Na=SiO~ 100 ppm, NaF 50 ppm. g. ~oo

~o

--

3.5

--

3.0

--

2.5

so

~ 80 o

.E 613 2.0 -o .E

,? so ~ m

~, Grade

:.. ~0

o

Recovery-

~, 3o - -

• •

S.I.F=S i S.[.FeA[

u

o

1.s L; 1.o

20 TM 0

0

~

~

0 -

3

10-

.~

0

o

i 0

I

i

I

lO 20 Starch concentration

i

0.5

I 3o in ppm

Fig.7. Effect of starch phosphate concentration on the flocculation of 1--20 ~m Barsua slime, p H 1 0 . 5 , pulp density 1%, N a 2 S i O ~ 1 0 0 p p m , N a F 50 ppm. ers and m s p e r s a n t s w h i c h t e n d e d t o overdisperse t h e fine particles. F o r three-stage p r o c e s s i n g , t h e o p t i m a l starch c o n c e n t r a t i o n s f o r giving best iron r e c o v e r y and selectivity indices were 20 p p m and 10 p p m , respectively (Fig.8). T h e d e p e n d e n c e o f grade and r e c o v e r y o n n u m b e r o f f l o c c u l a t i o n stages is s h o w n in Fig.9.

295

100 3.5 90 ~-

u 80-

30

¢~ 70 5 ._c 60 ¢£

12.5 2.0'-

~ so

~1.5

2 -~ o

30-

~

211--

o

100

I 0

i/1

/

A o • •

I

Grode Recovery S'I'FeSi S.I.FeAI

I

~

.0 0.5

t

o

5 10 15 20 25 Starch concentration in ppm

Fig.8. Effect of starch concentration on three-stage flocculation of 1--8 #m Barsua slime. pH 7.0, pulp density 2%, Na2SiO 3 0 ppm, NaF 50 ppm. ~oo o ou

90

":~ 80 ~

E 2 70 c~

pH : 8.5 Grade, 5 ppm starch o Recovery, 5 ppm starch • Grade, 10 ppm starch • Recovery, 10 ppm starch

60

g L

~ 50

I

J

I

t

2 3 4 5 Number of floccu[otion stages

6

Fig.9. Effect of number of stages on the flocculation of 1--8 #m Barsua slime by MCH starch at 8.5 pH. Pulp density 2%, Na2SiO 3 0 ppm, NaF 50 ppm. It is e v i d e n t t h a t selective f l o c c u l a t i o n o f Barsua slime gives a m a x i m u m S.I. o f 2 - - 2 . 5 and 6 6 - - 7 2 % iron r e c o v e r y ; t h e r e c o v e r y figure can be increased o n l y at t h e c o s t o f grade. " F l o c c u l a t i o n " e x p e r i m e n t s o n 0 - - 7 4 ~ m slime w i t h starvation d o s a g e starch ( s u c h as 2 p p m ) in s o l u t i o n are better described as settling experim e n t s p r o d u c i n g l o o s e - s t r u c t u r e d flocs. S o m e b l a c k Fe203 particles also

59.0

60.6

51.2

76.7

0

0.2

1.0

2.0

87.6

88.8

86.2

87.4

4.6 (17.2)* 4.5 (16.3) 5.6 (19.5) 6.7 (22.4) 4.9

5.9

3.6

3.6

86.0

66.0

67.1

66.0

49.5

23.3

30.0

28.0

AI203

Fe203

% SiO.

% Fe203

% AI:O 3

% Recovery

Assay

41.5

28.8

30.0

26.0

SiO 2

TABLE IV

II-1 II-2 II-3 II-4

14.3 9.0 14.8 61.9

12.3 26.0 15.4 46.3 100.0

100.0

One stirrer expt. fractions

I-1 I-2 I-3 I-4

Two-stirrer expt. fractions

wt.%

72.0 74.4 71.2 80.0

67.2 72.8 72.0 85.6

15.6 14.0 15.0 14.1

16.0 14.7 14.4 8.8

12.2 11.6 11.8 5.6

11.6 11.3 11.9 7.2

10O.O 86.2 77.6 64

100.0 89.3 65.1 50.9

77.8 78.2 80.0

79.3 82.0 85.6

11.4 10.2 8.8

Fe-recovery % % Fe203 Assay % AI:O 3 Assay

SiO 2

F%O3

A1203

Cumulative data for binary cut

2.9

2.2

2.2

2.8

Fe/Si

Assay %

Data on c o l u m n settling experiments (no reagent used)

2.5

2.5

2.2

2.7

Fe/A1

Selectivity indices

*Bracketted figures denoting A|203 % in corresponding gangue indicate good alumina rejection.

Wt.% floc or settled material A+ B

Starch conc. ppm

Data o n low starch concentration flocculation/settling experiments (pH 7.0, Na~SiO 3 15 p p m , NaF 50 p p m )

TABLE III

Fe/Al 1.2, Fe/Si 1.56

Fe/AI and Fe/Si 1.3 Both 1.4

Selectivity indices

¢O O~

297

underwent selective settling. 86% recovery of iron was achieved by this method (Table III). Settling tests in the 12-1 settling column without dispersant gave similar results (Table IV). A grade of 85% Fe203 is attainable by the preferential settling method but a recovery of more than 50% needs the use of dispersant and starvation dosage of starch. Some typical Grade-Recovery data for selective flocculation/settling experiments on the Barsua slime are summarised in Fig.10. 90

e ¢

® 2 p p m starch

•

e ~7, uJ

SELECTIVE FLOCCULATION • 1-20 ~um 70.7%Fe203 ~ 1-8 /~m 65.6'1.F¢203 (VIDE FIGS.6 & 8)

O

2&

219~{ I S0

SETTLING O - 7 4 p m 70.4°1. Fe203 ® WITH OlSPERSANTS~LITTLESTARCH ( VIDE TABLE 111) WITHOUT OlSRERSANT OR STARCH (VIDE TABLE Iv ) ATWO-STIRRERS D ONE-STIRRER EXRTS

I

I I B0 Fe 2 03 RECOVERY'/.

?10

i

~ 80

I

I go

Fig.10. Grade-recovery data for typical selective floecu]ation-sett]ing experiments on Rarsua slime. DISCUSSION AND CONCLUSIONS

In most cases, kaolinite was found to be the best dispersed (stable)mineral and hematite the poorest (see Table I). This could be due to density effect, charge effect or both. For montmorillonite and illite, Na4P207 was found to be a better dispersant compared to Na2SiO3. Such a superiority was not established for kaolinite and hematite. Versicol W13 was found to have a special dispersive effect on montmorillonite. A lower dosage of dispersant such as 50--100 ppm may be tried for satisfactory and somewhat selective dispersion of clay minerals and deliberately quicker settling rate for hematite. Slightly higher dispersant dosages may be suitable for dispersion of all minerals including hematite if selective flocculation of the latter is intended. A very high dosage of dispersant may, however, prevent adsorption of flocculant on the hematite surface. Thus over-dispersion may jeopardise selectivity in separation. Use of starch as flocculant is specific, helping in selective flocculation of hematite rather than the clay minerals. Khosla et al. (1983) have shown that starch is chemically and somewhat irreversibly adsorbed on hematite. The anchorage is exothermic and probably through the phosphoryl group

298 of amylopectin. Causticizing and homogenizing at high RPM are necessary for the rupture of starch granules and liberation of amylopectin, when the dosage required for optimal flocculation is reduced. Therefore, preparation of starch solution is critically important. Ageing probably causes bio-degradation, the phenomenon being manifest with starch phosphate also which is otherwise a stronger flocculating reagent, presumably on account of its higher phosphoryl content (LaMer, 1964). Starch concentrations of 20--40 ppm seem to be optimal (see Table II), producing flocculated globules of hematite (instead of powdery sediments at lower starch concentrations) which do not easily break during boiling. High proportions of starch may form more than half a monolayer coating which is not conducive for flocculation (Healy and LaMer, 1964). High concentration of starch is also known to promote flocculation of quartz and silicates and thus decrease selectivity (Friend et al., 1973; Sresty et al., 1978, 1980). Solutions of 100 ppm sodium silicate give better results than 400 ppm concentration solutions. Higher dosage of dispersant presumably over-disperses hematite and decreases adsorption of starch and causes formation of fluffy or porous flocs which settle very slowly. Since the aluminium-containing grains in Indian iron ores were found to be 2 pm or less (Gururaj et al., 1979), the flocculation behaviour of fine mineral particles had to be studied. Flocculation of hematite particles less than 2 t~m in size was found to be poor (see Table II and also Colombo, 1977) and adversely affected due to sodium silicate dispersant; flocculation did not improve with the use of Ca2÷. For 0--2 pm particles of a hematiteclay mixture, the use of a dispersant such as Na2SiO3 remained problematical: without it, the clay particles were not adequately dispersed and with it, hematite particles were difficult to flocculate (see Table II). Even at coarser sizes such as 8 or 20 pm, selective flocculation of hematite particles from hematite-clay mixtures was of moderate and not very high selectivity index. This is probably due to the lattice structure of clay minerals. The positive charges in the lateral edges of clay particles (which may bear net negative charges) are assumed to be responsible for the coating of negatively charged coal by clay even though clay may have an overall negative charge like coal (Brown, 1962). A similar coating mechanism of clay over hematite, even though likely charged, would explain some heterocoagulation and decrease in the selectivity of flocculation. For pH values 2--7, clay and hematite particles may bear opposite charges when heterocoagulation would be heightened. Another reason for moderate to poor selectivity may be adsorption of starch on clay particles activated by Ca 2÷ and Mg2÷. Flocculation of kaolinite by polyacrylamides is augmented by Ca2÷ ion (Michaels and Morelos, 1955). Ca 2÷ and Mg2÷ are adsorbed on quartz (Clarke and Cooke, 1968), activate adsorption of starch (Iwasaki and Lai, 1965) and induce its flocculability at high pH (Iwasaki et al., 1980). Recently, Drzymala and Fuerstenau (1981) have shown that polyacrylic acid adsorption producing flocculation of quartz is activated by metal ions like Fe 3÷ which may originate from

299 hematite, and that activation can be prevented by disodium EDTA, sodium hexametaphosphate or potassium fluoride. In our experiments on hematite-clay systems, even though distilled water was used, the filtrate from the pulp in flocculation experiments contained around 80 ppm Ca and 12 ppm Mg as ions (no iron was detectable). It is conceivable that these ions originating from clay minerals (Iwasaki et al., 1980) may selfactivate the said minerals with reference to starch adsorption, resulting in poor hematite-clay separation which may not be significantly improved by repeated flocculation. This hypothesis is strongly supported by Rao's observation (1980) that a kaolinite and montmoriUonite-rich sludge is easily flocculated by an anionic polyacrylamide after pre-treatment with a sufficient quantity of Ca 2÷ and Mg2÷. Iwasaki (1980) recommended the use of sequestering agents like silicate or polyphosphates to prevent the role of Ca 2÷ and Mg2÷ in inducing flocculation of quartz by starch. Limited use of sodium silicate also helps in selective dispersion of clay minerals as noted by us. Sodium fluoride depresses clay particles in terms of adsorption of flocculant (Read, 1971, 1972). Sodium silico-fluoride apparently depresses both clay and hematite, reducing the flocculability of the latter. Best separations of 1--8 pm hematite-clay particles were possible with 40 ppm starch (MCH), 100 ppm sodium silicate and 50 ppm sodium fluoride. With 76--78% recovery and grade values, the selectivity indices for separation of kaolinite, montmorillonite and illite (paired with hematite), were 3.58, 3.12 and 3.36 respectively (see Table II). Repeated flocculation gave better grade but poorer recovery (Fig.4). This reciprocal relationship between grade and recovery has been noted earlier by Sresty and Somasundaran {1977), Sresty et al. (1978) and Dicks and Morrow (1978). Processing of data in some literature shows that the selectivity index for separation of hematite by selective flocculation from orthoclase is 4.5 (Read, 1971, 1972), from quartz 3.0--5.6 in successive stages of flocculation (Sresty and Somasundaran, 1977, 1980; Sresty et al., 1978) and from rutile 6.4 (RineUi and Marabini, 1979). In our experiments, clay minerals gave somewhat lower values, 3.12--3.58. The possible reasons for poorer selectivity have been earlier adduced. However, considerable scope remains for further research on the use of diverse flocculants and dispersants. For example, while starch has been shown to be much stronger as a flocculant for hematite than polystyrene sulfonate (Sresty et al., 1977) and polyacrylamide (Read and Hollick, 1976), polyacrylamides are known to produce open--structural flocs from which mechanically occluded impurities can be more easily elutriated (Read and Hollick, 1976). Strongly adsorbing flocculants do not necessarily produce the best selectivity in separation. S t u d i e s o n Barsua slime

An earlier study on flocculation of Indian hematitic slimes by starch (Subramanian and Natarajan, 1979) was not directed towards selectivity

300 analysis. Limited success has been achieved b y us towards beneficiation of hematite slimes from clayey gangue b y selective flocculation. Results are inferior to those obtained with pure clay-hematite artificial mixtures. Yarar and Kitchener (1970) and Read (1971) had earlier pointed o u t that natural grinding o f ores and even co-grinding of pure minerals cause contamination through smearing, and thus the selectivity of flocculation as experienced in artificial mixtures is rarely achievable. Better iron recovery with selective flocculation from 1--8 pm particles compared to 1--20 g m particles is evidently due to better liberation. Jackson's (1975) CBD m e t h o d of dissolution of hematite material was applied to 0--8 and 8--15 p m acetylene tetrabromide- " f l o a t " and "sink" materials, and X-ray and chemical analysis were done for the residues. The results revealed p o o r liberation and hence fractionation of kaolinite and montmorillonite (liberation of illite was better). While one-sixth to one-third of the total A1 and Si was dissolved in the hematite-leaching CBD solution, a b o u t 10% of the iron remained undissolved. Thus complex and occlusive occurrences of Fe and A1, viz., iron alumino-silicates, lattice substitution in the goethitic structure by A1 etc. cannot be ruled out. It is understandable, therefore, that the difficulty in liberation would naturally limit the selectivity index values for selective flocculation of Barsua slime to those dictated by liberation (around 2.0) rather than those achieved for pure mineral systems (above 3.0). Selective flocculation and floc-flotation have been reported to be industrially successful in Tilden Mine Operations (Paananen and Turcotte, 1978), for Mesabi range ores (Dicks and Morrow, 1978) and non-magnetic taconite ores (Frommer, 1968). But in such cases, quartz is the principal gangue material -- which can be easily separated b y selective flocculation of hematite by starch (Sresty and Somasundaran, 1980) -- and nearly complete liberation has been reported at 25 pm. As a matter of fact, coarser and liberated quartz particles at 85% minus 500 mesh grind could be easily removed by cationic flotation (Paananen and Turcotte, 1978). On the other hand, for Barsua slime, substantial liberation of clay particles is possible only around 2 um (Gururaj et al., 1979), and at this size or even around 20 pm, selective flocculation of hematite from liberated clay minerals does not give a S.I. as high as that for quartz. Use of larger quantities of dispersant and flocculant in selective flocculation of near-micron size slimes poses a dilemma in so far as little dispersant means less dispersion and hence low selectivity, and on the other hand, larger proportions of dispersant cause overdispersion o f all phases needing more flocculant (for the process) which again might worsen selectivity. Selective dispersion and settling seems to be a promising avenue for limited beneficiation of Barsua iron ore and slime. Inferior dispersibility of hematite -- compared to the better dispersion o f the clay minerals -- coupled with its higher density, allow preferential settling of liberated and lowclay hematite particles. Such liberated or low-clay particles are abundant in Barsua slime. Comminution of Barsua ore results in differential grinding

301 and produces partially liberated particles, n o t only in the millimeter-but also in th e micron-size range. It has been recently pr opos e d to improve the grade o f the millimeter range industrial concentrates obtained by scrubbing and washing in Barsua Mine, th r o u g h the simultaneous use o f a selective polymeric dispersant which would disperse and r e m ove t he adhering clay particles (S.A.I.L., 1982). While this process would improve the grade o f the hematite-concentrate in terms o f lower clay cont ent , it would also increase the p r o p o r t i o n o f slime and iron-loss in the slime, t h e r e b y accentuating the need o f efforts for iron-recovery f r om slime, as pr opos ed and discussed in this paper. Table IV illustrates t h a t some recovery o f iron is possible by plain differential settling w i t h o u t dispersant. Grade and recovery are improvable t hrough t h e use o f dispersant and little starch (Table III). Differential settling in dispersant solution followed by settling-flocculation in starvation dose starch (2 ppm) seem to yield t he best grade and recovery (Fig.10): 66--86% Fe203 recovery, c o n c e n t r a t e assay 87--89 % Fe203 and 4.6--6.7% A1203, S.I.Fe,A1 2.5--2.7. Such steps m a y be commercially executed in thickeners, cyclones, heavy media cyclones or suitable elutriators. ACKNOWLEDGEMENT The authors are grateful to t he Steel A u t h o r i t y o f India Ltd., for sponsoring this research project and permitting publication o f the results. S.A.I.L. provided fellowships t o t he first t hr ee authors. REFERENCES Brown, D.J., 1962. Coal Flotation. In: Froth Flotation. 50th Anniversary Volume, Am. Inst. Min. Met. Eng., pp. 5--7. Clarke, S.W. and Cooke, S.R.B., 1968. Adsorption of calcium, magnesium and sodium ion by quartz. Trans. Am. Inst. Min. Metall., 241: 334--341. Colombo, A.F., 1977. Selective flocculation-flotation of Western Mesabi oxidised taconite, Proc. 50th, Annu. Meet. Min. Eng. Section Am. Inst. Min. Met. Eng., 8: 1--18. Dicks, M. and Morrow, J.B., 1978. Application of the selective flocculation-silica flotation process to the Mesabi Range Ore. Paper presented at 1978 AIME Annual Meeting, Denver, Colorado, 78-B-6, pp. 1--14. Drzymala, J. and Fuerstenau, D.W., 1981. Selective flocculation of hematite in the hematite-quartz-ferric ion-polyacrylic acid system. Part 1, activation and deactivation of quartz. Int. J. Miner. Process., 8: 265--277. Friend, J.P. and Kitchener, J.A., 1973. Some physico-chemical aspects of the separation of finely divided minerals by selective flocculation. Chem. Eng. Sci., 28: 1071--1080. Friend, J.P., Iskra, J. and Kitchener, J.A., 1973. Cleaning a selectively flocculated mineral slurry. Trans. Inst. Min. Metall. (Sect C; Miner. Process. Extra. Metall.), 82: 235. Frommer, D.W., 1968. Preparation of non-magnetic taconites for flotation by selective flocculation. Proc. 8th Int. Miner. Process. Congr., Leningrad, U.S.S.R. Paper D-9. Gaudin, A.M., 1939. Principles of Mineral Dressing. McGraw Hill, New York, Chap. 10, p.235. Gururaj, B., Prasad, N., Ramachandran, T.R. and Biswas, A.K., 1979. Studies on composition and beneficiation of a f'me-grained alumina-rich Indian iron ore. 13th Int. Miner. Process. Congr., Warsaw, June 4--9, 1979. Proceedings published in 1981 by Elsevier, Amsterdam, Vol. 2, part A: 447--471.

302 Healy, T.W. and LaMer, V.K., 1964. The energetics of flocculation and re-dispersion by polymers. J. Colloid Sci., 1 9 : 3 2 3 and J. Phys. Chem., 1962, 66: 1935. Iwasaki, I. and Lai, R.W., 1965. Starches and starch products as depressants in soap flotation of activated silica from iron ores. Trans. Am. Inst. Min. Met. Eng., 239: 364. Iwasaki, I., Smith, K.A., Lipp, R.J. and Sata, H., 1980. Effect o f calcium and magnesium ions on selective desliming and cationic flotation of quartz from iron ores, Proc. Int. Syrup. on Fine Particle Proceedings, Las Vegas, Nevada, Feb. 24--28, 1980, Chap. 54, pp. 1057--1082. Also: Selective flocculation of fine-grained iron-bearing materials. Principles. NSF Workshop Report on Beneficiation o f Mineral Fines, Chap. 20, p. 257. Jackson, M.L., 1975. Sodium citrate-bicarbonate-dithionite (CBD) method for removal of free iron oxides from soil or clay, In: Soil Chemical Analysis - - Advanced Course. 2nd Edition, 10th Printing, University of Wisconsin, pp. 44--48. Khosla, N.K., Bhagat, R.P., Gandhi, K.S. and Blswas, A.K., 1982. Calorimetric and other interaction studies on mineral-starch adsorption systems. Colloids and Surfaces, in press. Kitchener, J.A., 1977. NATO, ASI on the Scientific Basis of Floccuiation, edited by K.J. Ires, paper 12. Christ's College, Cambridge, England, pp. 284--285. LaMer, V.K., 1964. J. Colloid Sci., 19: 291--293. Michaels, A.S. and Morelos, O., 1955. Poly-electrolyte adsorption by kaolinite. Ind. Eng. Chem., 47: 1801. Paananen, A.D. and Turcotte, W.A., 1978. Factor influencing selective flocculationdesliming practice at the Tilden Mine. Paper presented at AIME Meeting, Feb. 1978, pp. 1--8. Rao, S.R., 1980. Flocculation and De-watering of Alberta Oil Sand Tailings. Int. J. Miner. Process., 7 : 245--253. Read, A.D., 1971. Selective flocculation separations involving hematite. Trans. Inst. Min. Metall., Sect. C, 80: 24--31. Read, A.D., 1972. The use of high-molecular weight poly-acrylamides in the selective flocculation separation o f a mineral mixture. Br. Polymer J., 4: 253--264. Read, A.D. and Hollick, C.J., 1976. Selective flocculation techniques for recovery of fine particles. Miner. Sci. Eng., 8(3): 202--213. Rinelli, G. and Marabini, A.M., 1979. A new reagent system for the selective flocculation of rutile. 13th Int. Miner. Process. Congr., Warsaw, Poland, June 1979. Proceedings published in 1981 b y Elsevier, Amsterdam, Vol. 2, Part A: 316--345. S.A.I.L. (Steel Authority of India Ltd.), 1977. Barsua Iron Mine Report, O.M.Q. Department, April 1977. S.A.I.L., 1982, Private Communication. Sresty, G.C. and Somasundaran, P., 1977. Beneficiation o f mineral slimes using modified polymers as selective flocculants. Proc. 12th Int. Miner. Process. Congr., Sao Paulo, Brazil. Sresty, G.C. and Somasundaran, P., 1980. Selective floccuiation of synthetic mineral mixtures using modified polymers. Int. J. Miner. Process., 6: 303--320. Sresty, G.C., Raja, A. and Somasundaran, P., 1978. Selective floccuiation of mineral slimes using polymers. In: Recent Advances in Separation Science, Vol. 4. CRC Press. Subramanian, S. and Natarajan, K.A., 1979. Floccuiation studies on hematite ore fines using starches. Trans. Indian Inst. Met., 32(2): 157--165. Viswanathan, S. and Paranjpe, V.G., 1968. Washing and agglomeration of iron ores Trans. Ind. Inst. Met., 21(2): 71. Whistler, R.L., 1964. Methods in Carbohydrate Chemistry, Vol. 4. Academic Press, New York, N.Y., 295 pp. Yarar, B. and Kitchener, J.A., 1970. Selective flocculation of minerals: 1 - - basic principles; 2 - - experimental investigation of quartz, calcite and galena. Trans. Inst. Min. Metall., Sect. C, 79: 23--33. Yu, A.T., 1968. Noamundi - - India's new iron ore complex. Min. Eng., 20(11): 70. Zanoni, A.E., Asce, M. and Blomqulst, M.W., 1975. Column settling tests for flocculant suspensions. J. Environmental Eng., Div. ASCE, 101 (EE 3): 309--318.