Minerals ~ring, Vol. 4, No. 5/6, lIP. 563-571, 1991Printed in G1'eat Britain

0892-6875/9\ $3.00 + 0.00~ \99\ Perpmon Press pic

BENEFICIATION OF A HIGH DOLOMITIC PHOSPHATE ORE:A BENCH SCALE OPTIMIZATION STUDY

K. ZHONG1, T. V. VASUDEVAN and P. SOMASUNDARAN2

H~nry Krumb School of Mines, Columbia University, New York, NY 10027, U.S.A.Current address: Wuhan Chemical Engineering Institute, Wuhan, China P .R.C.

2 To whom correspondence should be addressed(Received 7 August 1990; revision accepted 26 September 1990)

ABSTRACT

A three stage flotation scheme was employed to beneficiate a low grade. highdolomitic phosphate ore from Florida. The reagent scheme used. developedon the basis of our previous fundamental studies. consisted of oleic acid addedas collector. mineral oil and ethoxylated sulfonate as co-collectors and 1.hydroxy ethylidene 1.1- diphosphonic acid (Dequest 2010) as francolitedepressant. The three stage scheme was designed to remove fine dolomite(minus 210 microns) in the first stage. quartz in the secolrd and coarse dolomite(plus 210 microns) in the filral stage. Using a systematic optimizatiolrprocedure. which included a comprehensive factorial design. a 420xl05 micronsphosphate ore assaying 9.8 wt% P20S' 1.9 wt% MgO and 58 wt% acidinsolubles was beneficiated to a product analyzing 31 wl% P20S' 0.9 wt% MgOand 6.0 wt% acid insolubles at a phosphate recovery of 71 wt%.

KeywordsOptimization; beneficiation; phosphates; carbonates; francolite; dolomite;flotation; oleic acid; ethoxylated sulfonate; phosphonic acids

INTRODUCTION

Development of a suitable process to beneficiate low grade, high dolomitic phosphate oresis important for extending the rapidly depleting reserves of Florida phosphates. Thecurrently used commercial process of double flotation [1] is not adequate for reducing thedolomitic impurity level to less than 1.0 weight percent MgO in the concentrate, asstipulated by the phosphate industry. During the past decade, studies conducted byinvestigators in the Florida phosphate industry [2,3,4,5] and the Tennessee Valley Authority[6] have resulted in the development of a number of processes. However., all these studiesoriginated on a bench scale and the fundamentals involved in the separation scheme havenot been established. This leads to serious limitations for optimization of these processesand also their use for beneficiating ores of different origin and composition. Recently,Moudgil and Chanchani [7,8] and Ince [9] have conducted some fundamental studies whichresulted in the development of two processes for the removal of dolomite from apatite.Extension of these fundamental studies to beneficiate the natural ores on a bench scale wasalso reported [10]. However, a systematic optimization of the important variables was notattempted and no guidelines are yet available for processing such complex ores.

In this study, beneficiation of a high dolomitic Florida phosphate Ore on a bench scale usinga three stage flowsheet is discussed. The reagent scheme employed included the use ofethoxylated sulfonate as a co-collector along with the conventionally used oleic acid -

563

S64 L bmNG et aI.

mineral oil mixture, and a phosphonic acid depressant. These reagents were identifiedthrough fundamental studies and showed promising results in microflotation tests [II].Ethoxylated sulfonate was added due to its high salt tolerance since non-selectiveprecipitation of metal- surfactant complexes is reported to be an important reason for theloss in selectivity observed during the separation of salt-type minerals [12]. Anotherimportant feature of the current study is the use of a comprehensive factorial design anda systematic optimization procedure employed in each stage.

Work described in this paper summarizes the results obtained from the optimization study

EXPERIMENTALMaterials

Phosohate ore: High dolomitic phosphate ore was supplied by the Florida Institute ofPhosphate Research (FIPR). The as-received ore sample was deslimed and screened using710 and 105 microns screens. The undersize material was discarded while the oversizefraction was crushed and ground, followed by desliming and screening using a 105 micronssieve. The oversize fraction obtained from the latter step had a high dolomite content and,therefore, was mixed with the 110xl05 microns fraction of the as-received ore. Thecomposite material, used as the flotation feed, assayed 11.0 wt% PzOs, 2.2 wt% MgO and54 wt% acid insolubles. A 420xlOS microns fraction, also used as flotation feed, wasobtained by screening the composite 710xl05 microns feed. The 420xl05 microns fractionassayed 9.8 wt% PzOs, 1.9 wt% MgO and 58.3 wt% acid insolubles.

Reagents:

Collectors: A mixture of fatty acids containing 70 wt% oleic acid, was obtained from Fluka,A. G. Extra heavy white mineral oil and ethoxylated sulfonate, used as co-collectors, werepurchased from Sonneborn and sons Inc., and PPG Inc., respectively.

Modifier: I, hydroxyethylidine-I,I-diphosphonic acid (Dequest 20 I 0), obtained fromMonsanto chemical co., was used as the phosphate depressant.

Other Reagents: Nitric acid and potassium hydroxide, used as pH modifying reagents, weresupplied by Fisher Scientific company. Calcium, magnesium and phosphorous standards,used for chemical analysis, were purchased from Environmental Resources Associates.

Methods

Aooaratus: Bench scale flotation was carried out using a Denver. Model D-l flotationmachine provided with a 6.9 cm diameter impeller and a 1.5 litre cell. An agitationintensity of 1500 rpm was maintained during conditioning and flotation.

, f

Procedure: Five hundred grams of feed were conditioned at the desired pulp density fora known time interval after which make-up water was added in order to maintain a pulpdensity of 36 weight percent during flotation. All experiments were carried out using tapwater and flotation was continued till completion.

Analysis: P20S and MgO contents of the feed and notation products were determined asfollows: 5 mls of concentrated hydrochloric acid was added to I gram of solids, and heate.dto dryness. 5 mls of concentrated nitric acid and 5 mls of concentrated hydrochloric acidwere then added, heated to expel the fumes, made up to a known volume ( 500 mls or Ilit.), and the solution analyzed by inductively coupled plasma emission spectrosCODY.

Beneficiation of a phosphate ore S6S

EXPERIMENTAL DESIGN, RESULTS AND DISCUSSION

Design of Flowsheet

Dolomite and Quartz are the main gangue minerals in the phosphate ore and since these twominerals exhibit vastly different flotation behavior, in terms of their response to the fattyacid collector, it is difficult to remove both the minerals in a single stage. Also, the feedanalysis indicated that the major fraction (50 wt%) of the dolomite present was in the coarsesize range (plus 210 microns). Since coarse dolomite is known to display poorer flotationresponse than the finer particles [13], an attempt was made to remove the two fractions indifferent stages. Therefore, a three stage beneficiation scheme was conceived in whichdolomite would be removed in two stages and Quartz in a third stage.

Preliminary experiments conducted led to a flowsheet and reagent scheme described inFigure 1. The flowsheet described in the figure was designed to achieve beneficiation byremoving the majority of the fine dolomite during the first stage, quartz and some portionof coarse dolomite during the second and the remaining coarse dolomite during the thirdstage.

Feed 500 aCondit/ooing (pH = 5.5)

Reagent (In sequence) TIme. min.

Deq.2010 2EO. sulfonate 1Oleic acid 2Minerai oil 1

FlotationStage I

Float 1(dolomite)

Waste water (Fresh water)

Conditioning (pH= 7-7.5)

Reagent (In seq.) Time,mln

EO. sulfonate 1Oleic acid 1Mineral oil 1

FlotationStage II

Sink (quartz + dolomite)Acid washing(pH 4-4.5), 2 min.

C<XJditioning (pH = 5-5.5)

Reagent (In seq.) TIme. min.

Deq.2010 1EO. sulfonate 1Oleic acid 1Mineral 011 1

FlotationStage III

Float 2 Sink 1(dolomite) (Francolite)

Fig.l Flowsheet of three stage flotation scheme

S66 K. ZHoNG et aI.

~lo~~ion of ~)omite from Francolite - First staReIn this stage, fine dolomite was removed by flotation using a mixture of oleic acid andmineral oil (1:1 by weight) as collector and co-collector respectively. Ethoxylated sulfonatewas used as an additional co-collector, while Dequest 2010 was used as the francolitedepressant. Effects of pH and conditioning pulp density were examined in addition to theeffect of reagent dosage using the factorial design.

Factorial Desi2n:A five-factor, three-level fractional factorial design [14] was employed in the first stage.Details of various factors and their levels are shown in Table I. Selectivity index as definedbelow was used as the performance criterion [15]:

51 - % P20S Recovery + % MgO Rejection - 100(1)

where suffix I represents the first stage

TABLE 1 Factors and levels for stage I

Code Factor Level

.1 0 +1

A 600 1200 1800Oleic acid/Mineral oil, g/T

(50:50 by weight)

B Ethoxylated sulfonate, gfT 0 100 200

c Pulp density, wtO/o 45 55 65

D pH 5.0 5.5 6.0

E Dequest 2010, 9/T 0 20 40

Variance analysis of the effect of various factors was carried out and the results obtainedare summarized in Table 2. The parameter F, which represents the effect of a factor oran interaction in a normalized form, is obtained by dividing the residual mean of square ofa factor or an interaction by that of the error. F-test indicates the relative significance ofa factor over that of the experimental error. A comparison of the F values calculated fordifferent factors showed that at 95% confidence level (Fo.os = 3.74) only the effect ofethoxylated sulfonate was significant (F= 10.2). This suggests that non-selectiveprecipitation of metal-surfactant complexes on minerals is an important reason for the lossin selectivity observed during separation of salt-type minerals and that by using salt-tolerant reagents selectivity can be improved.

Qotjmjzation - First Sta2e: .Optimization for the first stage was carried out following the path of steepest ascent [16,17].Since it was found from the factorial experiments that only the ethoxylated sulfonate levelwas significant, optimization of only this factor was attempted while the other parameterswere maintained at constant levels. The results of the experiments, summarized in Table3, showed the selectivity index to increase with ethoxylated sulfonate dosage in the testedrange of 50 to 150 gfT. However, increase in ethoxylated sulfonate dosage above 100 glTcaused excessive foaming making the operation difficult. Hence, from a practical point ofview, 100 gfT ethoxylated sulfonate would represent the optimum dosage.

Beneficiation of a phosphate ore 567

TABLE 2 Variance analysis for stage I

Code Factor/Inter.

Sum of

square

Degrees of

freedom

Mean of

squares

F

A O.A/M.O 446 2 223 1.29

B EO. S 3508 2 1754 10.20

c Pulp Density 30 2 15 0.09

D pH 357 2 179 1.04

E Deq.2010 194 2 97 0.56

2758 16 172

Total 7293 26

TABLE 3 Results of optimization for stage I

Oleic acid/Minerai oil (1:1) = 600 9/t

Dequest 2010 = 40 9/t

pH = 5.5

Pulp density = 55 wt%

EO Prod. Weight

g/t %Assay, 0/0

P20S MgO Ins

Dis., %

P205 MgO Ins511

50 Float 4.3 3.8 16.8 3.0 1.5 39.6 0.2 38.1

Float75 6.6 3.6 13.1 2.0 2.2 41.1 0.3 38.9

100 Float 7.3 5.5 15.0 2.0 3.6 49.2 0.3 45.6

125 Float 8.6 7.2 12.3 3.5 5.6 51.4 0.6 45.8

150 Float 11.1 9.3 11.5 6.0 9.9 58.4 1.4 48.5

Experiments described thus far were conducted using a 710xlO5 microns feed. Since420xlO5 microns fraction is more commonly used in the Florida phosphate plants, anattempt was made to determine the effect of feed size on the flotation performance. Theresults obtained showed that the optimum condition for flotation established with the710xlO5 microns fraction held valid also for the 420xlO5 microns feed although theselectivity index value obtained with the latter was higher (Table 4). Therefore, with thepractical application of the process in mind, experiments under stages II and III werecarried out with the 420xlO5 microns fraction.

Flotation of francolite from Quartz - Second StageThe main objective of this stage was to reject the Quartz in the sink. Flotation of francoliteusing oleic acid/mineral oil mixture and ethoxylated sulfonate was carried out at the naturalpH (7 - 7.5) of the suspension. Since preliminary results showed that a portion of the coarseM F ""1I.-r

568 K. ZHoNG et aI.

dolomite was also rejected along with Quartz, the selectivity criteria used included dolomiteremoval efficiency also.

TABLE 4 Optimization results for 420 x 105 microns feed

EO.S.

g/t

Prod. Weight%

51,Assay, %

P20S MgO Ins

Dis., %P 205 MgO Ins

5.7 1.6

98.4

100.0

75 Sink

Total

94.3

100.0

46.6

2.6 15.2 1.5

9.6 1.0 63.8

9.2 1.8 60.2

8.3

91.7

100.0

8.0 13.4 2.5

10.1 0.8 64.5

9.9 1.9 59.4

6.7

93.3

100.0

100 Sink

Total

52.4

8.1 6.3 13.0 2.5

9.4 0.8 65.6

9.2 1:8 60.5

6.0 58.3 0.3

125 Sink

Total

91.9

100.0

94.0 41.7 99.7

100.0100.0 100.0

52.3

Thus, SII = [2 x % PzOs recovery + % (quartz + dolomite) rejection - 200] / 2 (2)

The type of separation attempted in this stage is widely practised in the Florida phosphateindustry and is relatively simple. Therefore, the factorial design employed was limited totwo factors (oleic acid/mineral oil, A; ethoxylated sulfonate, B) and two levels(40 and 60g/T of A; 20 and 40 g/T of B). Results from the factorial tests showed that effects of oleicacid/mineral oil and ethoxylated sulfonate, and their interactions, were significant.Optimization studies were conducted following the path of steepest ascent. Optimumreagent dosages found were 40 g/T ethoxylated sulfonate and 118 g/T oleic acid/mineraloil mixture. The maximum selectivity achieved was 49.5 which corresponds to a phosphaterecovery of 89 wt% and, dolomite and Quartz removal efficiencies of 26 and 95 wt%,respectively.

Coarse Dolomite Flotation - Third Sta26In this stage. oleic acid - mineral oil mixture and ethoxylated sulfonate were used ascollectors for dolomite and Dequest 2010 as a francolite depressant. as in~e first stage.Since the feed to this stage was obtained after the original feed was processed through twostages the amount and. therefore. the pulp density used in the third stage was relativelylower (25 wt%). Also. since the effects of pH and pulp density were not found to besignificant in the first stage. these factors were maintained at fixed levels. serectivity index. . ,used was the same as 10 the f 1.rst stage;

A three-factor, two-level factorial design was used and various factors and levels employedare summarized in Table 5. Variance analysis, carried out as described in:~~age I, showedthat at 95% confidence level (FO.O5 - 10) both ethoxylated sulfonate (F- ~) and Dequest2010 (F - 199) dosages were significant variables (Table 6). The import.ce of Dequest2010 dosage can be attributed to the enhanced hydrophobicity of francolite ijstage III feed,caused by oleate treatment in stage II. "?

48.2

51.8

100.0

0.1

99.9

100.0

59.1

40.9

100.0

0.4

99.6

100.0

Beneficiation of a phosphate ore S69

TABLE 5 Factors and levels for stale III

Cod. Factor Level, gfT

.1 +1

A Dequest 2010 20 10

B EO.S. 80 40

c O.A./M.O. 150 200

TABLE 6 Variance analysis for stage III

Factor/Inter.

Sum of

squaresDegreesof freedom

Mean of

squares

F

A Deq.2010 1193 1 1193 198.8

B EO.S 126 1 126 21.0

c O.A.jM.O 25 1 25 4.2

AxB 17 1 17 2.8

Error 18 3 6

Total 1379 '7

Optimization, carried out following the path of steepest ascent, showed that the optimumreagent scheme was oleic acid/mineral oil at 182 g/T, ethoxylated sulfonate at 57 g/T andDequest 2010 at 57 g/T. Selectivity index obtained under the optimum condition was 43.8which corresponds to a P20S recovery of 92.4 wt% and a dolomite rejection of 51.4 wt%.

Overall Flowsheet

Beneficiation of 420xlOS microns flotation feed using a three stage beneficiation scheme,under optimized conditions, produced a phosphate concentrate containing 28 wt% P20S and1.3 wt% MgO. Marginally lower P20S level in the phosphate concentrate was easilyincreased to the level desired by the phosphate industry (> 30 wt P20S) by employingcleaning operations subsequent to stages II and III which removed the entrapped Quartz andthe dolomite gangue, respectively. The concentrate thus obtained using the three stagebeneficiation scheme, which includes cleaning steps subsequent to stages II and III,contained 31 wt% P20S, 0.9 wt% MgO and 6.0 wt% acid insolubles. The phosphate recoverywas 71 wt%.

IC.. ZH>NO el aI.S70

CONCLUSIONS

A reagent scheme which included the use of ethoxylated sulfonate as a co-collector, to theconventionally used oleic acid/mineral oil mixture, and a phosphonic acid francolitedepressant (DeQuest 2010) was employed to beneficiate a high dolomitic Florida phosphateore. Using a three stage beneficiation scheme, which included two cleaning steps, aflotation feed containing 9.8 wt% P2°.5.' 1.9 wt% MgO and 58 wt% insolubles wasconcentrated to a product assaying 31 wt% P20S, 0.9 wt% MgO and 6 wt% insolubles at aphosphate recovery of 71 wt%.

Comprehensive factorial design followed by optimization was carried out for all threestages. In the first stage, where the majority of the fine dolomite particles were removed,only the effect of ethoxylated sulfonate was found to be significant. However, in the thirdstage, employed for the removal of coarse dolomite, effects of both ethoxylated sulfonateand Dequest 2010 were found to be significant. The effectiveness of ethoxylated sulfonatewas attributed to its salt-tolerance property due to which non-selective precipitation ofmetal-surfactant complexes is minimized. The importance of Dequest 2010 in stage III wassuggested to be due to the enhanced hydrophobicity of francolite in the feed to that stage.It was shown that by using an appropriate combination of reagents with conditions anddosage optimized systematically, a low grade complex phosphate ore can be beneficiated toa high grade concentrate at a reasonably good phosphate recovery.

ACKNOWLEDGEMENTS

The authors wish to acknowledge the Florida Institute of Phosphate Research (FIPR Project# 83-02-037) and the National Science Foundation (NSF Project # MSM-86-17183) fortheir financial support of this work.

REFERENCES

Crago, A. Process of Concentrating Phosphate Minerals U. S. Patent No. 1.193.640(1940).

I.

Lawver, J.E., Murowchick, B.L. & Snow, R.E. Beneficiation of South Florida HighCarbonate Phosphates 15MA-. Technical Economic Con!erel,ces, Orlando, Florida,Preprints TA/78/1 (1978).

2.

3. Snow. R.E Beneficiation of Phosphate Ores U.S Patenl No. 4.144.969 (1979).

Dufour, P., Pelletier, B., Predali, J.J. & Ranchin, G. Beneficiation of South FloridaPhosphate Rock with High Carbonate Content Proceedings, Second InternaJionalSymposium on Phosphorous Compounds, Boston, p. 247 (1980).

4

Lawver, J.E. Wiegel, R.L. Snow, R.E. & Hwang, C.L. Phosphate ReservesEnhancement by Beneficiation Mining Congress Journal, 68, p. 27 (1982).

s.

Lehr, I.R. & Hsieh, S.S. Beneficiation of High Carbonate Phosphate Ores U. S.Patent 4'.287,053 (1981).

6.

7. Moudgil, B.M. & Chanchani, R. Flotation of Apatite and Dolomite using SodiumOleate as Collector Minerals and Metallurgical Processing, 2(1), p.13 (1985).

8. Moudgil, B.M. &; Chanchani, R. Selective Flotation of Dolomite from Francoliteusing Two-stage Conditioning Minerals and Metallurgical Processing, 2(1), p.19(1985).

511Beneficiation of a phosphate ore

9. Ince, D.E. Effect of Sodium Chloride on the Selective Flotation of Dolomite fromApatite Ph.D Thesis. University 01 Florida (1987).

10. Moudgil, B. M., Ince, D., Vasudevan, T. V. & Sober, D. Bench-Scale Optimizationof the Two-Stage Conditioning Process for Apatite-Dolomite Separation Mineralsand Metallurgical Processing, 7(1), p. 53 (1990).

11 Somasundaran, P. Beneficiation of Dolomitic Phosphates, Final Report (Jan. 1989),Florida Institute of Phosphate Research, Bartow, Florida.

12. Ananthapadmanabhan, K. P. & Somasundaran, P. Role of Dissolved Mineral Speciesin Calcite-Apatite Flotation Minerals and Metallurgical Processing, 1(2), p. 36(1984).

13, Moudgil, B.M. Separation of Dolomite from South Florida Phosphate Rock - PhaseII, Final Report (July 1987), Florida Institute of Phosphate Research, Bartow,Florida.

14. Anon. Normal Statistical Methods (in chinese), p. 68, Institute of Mathematics ofthe Academy of Sciences of China (1975).

15 Sresty, G. C. & Somasundaran, P. Selective Flocculation of Synthetic MineralMixtures Using Modified Polymers Int. J. Min. Process., 6, p. 303 (1980).

16. Cochran, W.G. & Cox, G.M. Experimental Designs 2nd Edn, p. 357, John Wiley &Sons, New York (1957).

7. Somasundaran, P. & Prickett, G. O. Optimization of Flotation Operation usingStatistical Methods Trans AIME, 244, p.369 (1969).