Solvent Extraction Plants: Thiele Diagram & Theoretical Design Aspects

Solvent Extraction Plants: Thiele Diagram & Theoretical Design Aspects

It has been explained in an earlier talk that solvent extraction in the copper industry is used to convert impure, and frequently very dilute copper solutions into very pure and concentrated solutions from which pure copper can be extracted by a variety of means, usually electrochemical. These impure solutions emanate from in-situ leaching, dump leaching, heap leaching, vat leaching and agitation leaching or from mine waters and bleed streams: equally though they could be generated by direct leaching or roast-leaching of concentrates, so that solvent extraction and electrowinning can be applied to the treatment of most copper-bearing materials if desired. Hence, it is important that all mining and metallurgical personnel be aware of the potential of this unit process; and for the extractive metallurgist in particular, it is important that he understand the mechanisms and practicalities by which he can evaluate the benefits of solvent extraction and thereby design and optimise a total metal recovery process.

This talk, supported by a later one dealing with pilot plant testing, concentrates on advising the metallurgist how to translate the possibly rather vague thoughts of himself or others into a realistically quantified solvent extraction flowsheet.

Unlike many of the lectures on chemical engineering or metallurgy which one is obliged to attend in the course of academic qualification, this talk will begin by illustrating a typical end product, from which illustration it will be clear just how many variables are involved in a metallurgical flowsheet. It will also precipitate the realisation that somebody – namely the metallurgical engineer – is responsible for generating facts and for optimising all these variables such that ideas and possibilities are frozen into a process flowsheet which can be used in final plant design. The practical reasons behind the generation of apparently academic isotherms and distribution curves will become clear as will be the effects, of variables such as pH, copper concentration, raffinate levels, electrolyte composition, diluent and reagent choice, reagent concentration, etc.

Solvent Extraction Flowsheet Example

The bulk of present-day solvent extraction processes are associated with dump or heap leaching to give very dilute solutions although there are examples of vat leaching and of agitation leaching of flotation tailings. These examples will be mentioned later in another talk.

For the purpose of this example therefore – to avoid duplication – another project is chosen which was conceived and developed as far as part construction, before being suspended for economic reasons. Reference will, therefore, be made to the Tenke-Fungurume project of Zaire. This project is useful for illustrative purposes because it posed all the usual problems plus a variety of others not normally encountered – or certainly not all simultaneously.

The orebody in question has an average head grade of about 5.5 percent copper and 0. 35 percent cobalt; however, the mining programme developed showed considerable fluctuation around these figures. The copper mineralisation was malachite, chrysocolla and azurite with some chalcocite giving a sulphide copper content of about 0.4 percent: the cobalt mineralisation was mainly the oxide heterogenite (CoO.Co2O3) with some sulphide linnaeite (Co3S4).

Deeper mineralisation, of course, changed the ratio of oxides to sulphides very significantly. For a large number of reasons, including poor flotation response of the oxides, low sulphur content of the sulphides, high acid consumption by gangue (including dolomite), the flowsheet in figure (i) was decided upon.

simplified-flowsheet-for-tenke-fungurume-ore-processing

From the illustration, it can be seen that the apparently simple flowsheet requires a large number of data before the process can be engineered, and many of these data will have to be generated in the laboratory or the pilot plant. Table (i) lists some of the variables and questions to be answered which will affect design of the solvent extraction plant.

Table (i)

Solvent Extraction Data Needed for Flowsheet

  1. the volume flow of leach liquor and copper tenor
  2. volume flow of organic
  3. the acidity/pH of the leach liquor
  4. the raffinate level required
  5. temperature of the closed circuit,
  6. levels of impurities in leach liquor
  7. the number of extraction stages
  8. the number of strip stages
  9. concentration of S.X. extractant in organic
  10. composition of strip liquor/spent electrolyte
  11. required copper tenor of advance electrolyte
  12. suspended solids in thickener overflow
  13. gypsum content of leach liquor.

Dealing with these queries in turn, one can comment:

Volume flow of leach liquor, i.e. relative to the solids being processed, can actually be varied to suit the metallurgists’ requirements, since it can be anywhere between a minimum – equivalent to the volume of water entering with the solids feed plus wash water from the solids liquid separation stage – and a maximum depending on how big one is prepared to make leach tanks thickener equipment, pumps, etc., consistent, of course, with the ability to maintain solids in suspension during the leach. For a given ore grade and leach efficiency, the copper tenor in leach liquor is directly proportional to the solids:liquid ratio and in the case of fig (i), could have been 6 gpl copper at 10 percent solids in leach or up to 55 gpl copper at 50 percent solids in leach.

Volume Flow of Organic is determined by the chemical performance of the extractant being used under the conditions prevailing – which are influenced by pH, buffering action of impurities, temperature, etc. – and the concentration of the extractant in the diluent or carrier kerosene. There is a minimum amount of extractant which is tolerable because it must be present in an amount at least equal to that required to chemically complex the copper present and to result in the desired raffinate level. Obviously, the correct amount relative to the copper present can be achieved by high flow and low concentration or low flow and high concentration.

In the case of the flowsheet of figure (i), the flowrate necessary was 1.8 times the aqueous flow, i.e. an organic : aqueous ratio of 1.8:1.

The Acidity/pH of the leach liquor is determined by optimisation of leaching conditions with due consideration of the effect on solvent extraction. Leaching times, and hence equipment sizing, can normally be reduced by increased acidity : similarly, copper extraction efficiency is frequently enhanced. However, there is a negative effect on solvent extraction performance and usually acid consumption by gangue increases, entrainment losses from solids-liquid separation increase and there is an increased cost in subsequent effluent neutralisation.

In the case of the flowsheet shown in figure (i), there is also acid loss to the cobalt plant. The acid tenor chosen was 10 gpl for design purposes, largely based on leaching requirements and the need to avoid evolution of CO2 gas in thickeners by continuing dissolution of gangue carbonates. Probably, actual plant operations would have been somewhat lower or circuit changes may have introduced a 2-stage counter-current leach.

The raffinate level required in terms of copper tenor is obviously as low as possible, consistent with the cost of high efficiency solvent extraction, and also taking due consideration of the overall circuit. In dump leaching, one could argue that copper in raffinate is returned to the dump and eventually recovered; this is true but it means that there is a delay in recovering copper already leached. In vat or agitation leaching, raffinate can be used in washing (with some being bled to tailings for volume control) and hence there will be direct loss of raffinate and the contained copper. Subsequent neutralisation of the tailings consumes more lime to compound the cost of copper and SO4 losses.

In the case of the flowsheet of figure (i) the copper is recovered via the cobalt plant and hence about 1 gpl was considered suitable, but even this incurs expenditure of lime in order to precipitate and recycle.

Temperature of the closed circuit is normally not a controlled parameter but temperature does affect kinetics of solvent extraction, phase disengagement, extractant stability and kerosene diluent loss. In dump leaching temperatures might be 10 – 20°C but in agitation leaching it can be higher. In the case of figure (i) it could have been in the 40 – 50°C range because of the closed circuit, high heats of reaction in leaching, heat of dilution of large quantities of acid and heat generation in the electrolytic section.

Whilst the thermodynamics of solvent extraction may be little affected by temperature it is clear that there are many implications in terms of kinetics and plant design.

Levels of impurities in leach liquor are very important from two points of view in particular: the first is the buffering effect which can enhance solvent extraction performance whilst the second is physical in terms of fluid viscosity and entrainment of impure leach liquor into the tankhouse.

The implications are that real solutions are needed for laboratory/pilot plant work and that, in the main plant, probably a small bleed of liquor will be needed from the tankhouse to maintain electrolyte purity. If left entirely alone, one must presume that the levels of impurities in the electrolyte would eventually equalise with those in the leach circuit.

In dump leaching, save for iron, this may be of little importance but in closed circuits such as fig (i) impurity levels could be 18 – 20 gpl due to the high gangue content of the ore.

The number of extraction stages required is obviously important because of capital equipment costs and reagent inventory. The number of stages required is a function of the thermodynamic performance of the reagent and leach solution and the efficiency of each mixing stage. The lower the raffinate tenor required, the more the number of stages but there is an economic limit – which is obvious from the extraction isotherm – beyond which one would not attempt to lower the raffinate level.

It is very important therefore to use a strong reagent which, at equilibrium, quickly results in a low raffinate level. In the case of the flowsheet shown, 2 stages were proven to be sufficient, but only because of the particular circuitry involved. With Acorga’s second generation extractants, 2 stages are now adequate for most circuits.

The number of strip stages is important for similar capital cost reasons. Again the number required is a function of the thermodynamic properties of the liquors concerned and it is the efficient removal of copper from the loaded organic, down to a low level, which provides a stripped organic liquor, suitable to equilibrate with a very lean raffinate in the extraction stage.

Stripping of loaded organic is done with spent electrolyte and is enhanced by low copper content and high acidity of the latter. These two values must, however, be consistent with electrodeposition of good quality copper and with low anode corrosion in the tankhouse. High quality copper is normally made at spent copper tenors of say 25 gpl upwards (depending on current density, circulation rates, smoothing agents, temperature, impurity levels, etc.) and acid levels can be up to about 200 gpl. Normal strip liquors analyse about 30 gpl copper and 160 – 180 gpl acid.

Two strip stages were chosen in the circuit discussed and this is still fairly normal.

Concentration of S.X. extractants in organic must be considered from the point of view of extraction efficiency and loading capacity, the cost of entrainment losses and possibly adverse effects on phase disengagement (and hence on settler sizing). It must be remembered though that whilst improved thermodynamic performance can be achieved by increased organic concentration, frequently a similar overall effect can be achieved by increasing the volume of organic at the low concentration relative to the volume of aqueous copper liquor, i.e. by increasing the O/A ratio. This latter method obviously increases overall flow rates and hence mixer and settler sizes, but it is often necessary because there is a maximum extractant concentration which is tolerable because of solubility.

Which procedure to follow can be predicted chemically or thermodynamically from laboratory investigations, but effects on plant sizing and entrainment losses must be found from pilot plant work.

The loading capacities of earlier S.X. extractants were about 0.25 gpl (and less) copper per percent extractant in kerosene whereas Acorga’s second generation extractants load to in excess of 0.54 gpl copper per percent extractant in kerosene.

The flowsheet shown would have used a 40 percent concentration of extractant in kerosene which is very high – in fact much higher than – normal – but tolerable in the situation of very high temperature.

Composition of strip liquor/spent electrolyte has been largely explained in point 2.8 above, being limited to a large extent by tankhouse tolerances. From the point of view of the S.X. plant, however, strongly acidic solutions at very low copper tenor are the most advantageous since this is the converse of extraction and is predictable from the standard S. X. equation.

standard-s-x-equation

i.e. strong acid favours the formation of CuSO4.

In the case of the flowsheet here, 30 gpl copper was chosen as the limit dictated by electrowinning and 150 gpl acid was deemed to strip the organic adequately.

Required copper tenor of advance electrolyte is the factor which determines the flowrate of strip liquor through the solvent extraction plant, consistent with temperature control requirements of the tankhouse and the sizing of coalescers or filter columns to remove entrained organic ahead of tankhouse operations.

Normally, the strip liquor entering solvent extraction at nominally 30 gpl copper, is boosted to between 40 and 50 gpl copper before returning to the tankhouse, i.e. an increase of 10 or 20 gpl. Since the difference in copper tenor between loaded and stripped organic can be say 5 gpl (depending on reagent concentration, pH of pregnant feed liquor and acidity of the strip liquor), it follows that the organic to aqueous ratio will be up to 4:1.

In figure (i) the O/A ratio would have been 3.3:1 to give advance electrolyte at 45 gpl copper.

Suspended solids in thickener overflow should be kept to a minimum, say 10-50 ppm, in order to avoid the formation of a “crud” or “scum” in the solvent extraction settlers which would lead to losses of organic liquor by entrapment. Actually, some crud is always formed and centrifuges are normally installed to recover organic liquor from it. Dissolved or colloidal silica can also be a problem.

Various types of filter or clariflocculators can be used in the case of agitation leaching but for dump leaching, there is frequently no need for filters, and a settling pond is adequate.

Figure (i) would have used clariflocculators.

Gypsum content of the leach liquor is very important when treating ores with high calcium carbonate content, i.e. limestone or dolomite, because supersaturation can be reached with subsequent precipitation and scaling in the solvent extraction system. Whilst, like item 2.12, there is little or no effect on thermodynamics of the reactions, there can be obvious physical problems.

Typical SX Circuit Profile Using Acorga Reagent

Having shown how solvent extraction fits into the overall process flowsheet, it is logical to now show the solvent extraction operation in slightly more detail. Following from figure (i), the solvent plant itself is illustrated in figure (ii), with performance data based on a first generation extractant.

circuit-profile-for-high-capacity-using-s-x-plant-using-first-generation-extractant

It must be remembered that in this special case the feed acidity and copper tenor are both very high but that also 40 percent concentration of extractant was necessary. The illustration is not, therefore, meant to depict any more usual operation but rather is included to complete figure (i) and to illustrate the transfer of copper round a circuit.

Solvent Extraction Process SX: Hydrometallurgical Extraction of Copper

Figure (iii) shows another profile using a similar feed liquor and a second generation Acorga extractant. Whilst the copper, and acid tenors around the circuit are not exactly the same as in fig. (ii), the illustrated performance is actually achieved with only 15 percent extractant concentration (i.e. less than half) at an O/A ratio of 2.1:1.

Although these figures merely illustrate the transfer of copper around the circuit, and in view of the different conditions should not be used directly for comparative purposes, it is clear that the second generation extractants would now allow the project flowsheet to be treated substantially differently.

typical-circuit-profile-with-acorga-2nd-generation-extractant

Solvent Extraction Isotherms

Having explained how solvent extractions can be included in a hydrometallurgical flowsheet and outlined some of the large number of interacting variables to be considered, the most basic requirement for solvent extraction design is an understanding of distribution coefficients and equilibrium isotherms. These data and curves are obtained in the laboratory – and to some extent by computer if enough basic data are in the computer programme. Acorga, for example, is able to predict S.X. performance over a small range of reagent concentrations, by computer extrapolation of practical data obtained at one concentration.

Distribution curves or equilibrium isotherms merely illustrate the distribution of a dissolved component, between two phases : in the particular case of solvent extraction of copper, it merely shows how copper distributes between the aqueous and organic phases under the different conditions experienced in the process, but an understanding of this is sufficient for preliminary design of process flowsheet. Of course, final proof with real solutions and the generation of engineering data, are the function of a pilot plant. Even with a pilot plant, however, its efficiency is calculated by comparison with laboratory curves and data.

Solvent Extraction Distribution Curves

The objective of solvent extraction in the metallurgical industry is clearly to extract one metal ion and to reject the others, either to recover the metal extracted or merely to remove it as an impurity.

Figure (iv) shows the distribution of several common metals between a sulphate medium and di 2.EHPA as a function of equilibrium pH.

extraction-of-metals-from-sulphate-medium

It can be seen that there is potential for separating copper from say cobalt and nickel at pH 3 where a high extraction of copper would result relative to that for the other two metals mentioned: however, the reaction would only be of value if iron and zinc were first removed by some other means.

Further, this type of curve is based on equilibrium pH and does not allow for the decrease in pH which accompanies extraction and hence intra stage neutralisation would be needed to maintain the extraction efficiencies indicated.

The newer copper-specific extractants, now used industrially, exhibit much steeper curves for copper and also function at much lower pH values, as seen in figure (v).

extraction-of-copper-with-acorga

Copper-specific extractants and distribution isotherms

With the development of copper-specific extractants, the above suites of equilibrium pH distribution curves assumed less significance in the copper industry and only the distribution of copper was important, in terms of efficiency of extraction at pH conditions likely to pertain in leaching operations. Since the extractants developed were efficient at lower pH values, the way was clear for commercialisation of solvent extraction techniques.

The type of curve used here is known as the distribution isotherm and it shows the relative distribution of copper between organic and aqueous layers at prevailing pH, i.e. only the feed pH is fixed and the remaining data is at whatever pH results from the extraction. Figure (vi) shows the general form of an extraction isotherm, i.e. removing copper from aqueous solution by the organic phase.

typical-extraction-isotherm

These curves are generated in the laboratory by taking aliquots of feed liquor and mixing each with a measured amount of organic reagent, such that a range of say 8 samples is available representing various O/A ratios. The range should cover, says:

samples-range-ratio

Mixing should be vigorous and for about 15 minutes, in order that equilibrium can be reached in each case before the phases are allowed to separate for analysis. The copper tenor of each aqueous aliquot is used with the copper tenor of the corresponding organic layer, in order to provide points on the isotherm being generated.

A strip isotherm (fig v) is generated in a similar fashion by contacting loaded organic phase with various proportions of the strip liquor (electrolyte) but in this case usually 4 points are adequate,

proportion-of-strip-liquor

The strip isotherm shows the equilibrium data when copper is being removed from the organic phase by strong acid or electrolyte.

typical-strip-isotherm

These isotherms, of the types shown in figures (vi) and (vii), form the basis of preliminary design for any solvent extraction plant. It will not be possible for the commercial plant to perform better than predicted by the isotherms (unless conditions change, in which case a new isotherm is needed) and in all cases efficiency will dictate that it performs somewhat worse.

Use of Isotherms in Process Design 

by the McCabe-Thiele Diagram

It is now proposed to demonstrate how the extraction and strip isotherms can be used to both design a solvent extraction circuit and to predict the performance of that circuit.

The problem in the randomly chosen example to be shown, is how to treat an aqueous liquor containing 12 gpl Cu++ and about 1.6 gpl free H2SO4. The objective is to operate an electrolytic tankhouse with a pregnant electrolyte at 45 gpl Cu++ and a spent electrolyte at 30 gpl Cu++ and 160 gpl acid.

First, the relevant extraction and strip isotherms are produced in the laboratory by the method described earlier, using stripped organic at equilibrium with the aforementioned spent electrolyte. The initial choice of organic phase is largely a matter of experience but at a loading mentioned previously, i.e. about 0.5 gpl Cu per percent extractant in diluent, a concentration of about 24 percent should be needed in theory: actually, since no system is perfect or loads to 100 percent loading, in this case 30 percent is provisionally assumed. It is also important at this early stage to recognise that the organic phase is never stripped down to zero copper content, in other words there is always a circulating load. The copper content of stripped organic is a function of many parameters including those in table (ii).

Table (ii)

Parameters affecting copper content of stripped organic

  1. the organic extractant chosen
  2. the organic concentration
  3. acidity of strip liquor
  4. copper content of strip liquor
  5. number of strip stages
  6. temperature

In this example, the stripped organic would actually still contain about 6.3 gpl Cu++ : one would expect much lower values with lower concentration of organic, with weaker extractants, stronger strip acid, etc. and it could, in fact, in some cases, be as low as say 0.1 gpl copper. O/A ratios in excess of 1:1 can be used to effect copper transfer in greater amounts.

The important value to consider is the transfer capacity of the system in gpl copper, i.e. the difference in gpl copper between loaded and stripped organic per volume of reagent used. But before jumping to premature conclusions that a weak extractant operating between say 0.1 gpl and 5 gpl is as good as a strong extractant operating between say 5 gpl and 10 gpl, it must be recognised that only the stronger extractants can operate efficiently at the higher values of copper and acid tenor so that a comparison is not relevant at this stage.

extraction-isotherm

Figure (viii) shows the extraction isotherm generated for the conditions specified above, using the provisionally assumed 30 volume percent extractant in diluent. Because this 30 percent is still slightly inadequate to load all the copper in addition to the 6.3 gpl which recirculates, i.e. a total of 18.3 gpl, an increase in ratio of organic volume to aqueous volume is used, i.e. an increase in O/A ratio.

We must now consider the concept of an operating line. If the volumes of organic and aqueous were to be equal, then an increase in organic tenor of 1 gpl copper would be matched by a decrease in the aqueous tenor also of 1 gpl: if say we had twice the volume of organic then it would only need an increase of 0.5 gpl in the organic to effect a decrease of 1 gpl in the aqueous tenor. In the first case, the operating line has a slope of 1 and in the other it has a slope of ½. Therefore, it is obvious that by choosing the O/A ratio, the concentration in the organic layer will be varied for any aqueous feed tenor.

The origin of the operating line is important to locate and this is defined by the stripped organic level and the raffinate level, which latter is obtained by trial and error, or iteration.

In the example of fig (viii), an O/A ratio of 1.43:1 is illustrated so that the slope of the operating line is defined accordingly, passing through the stripped organic value of 6.3 gpl copper and a raffinate level of that required.

In order to predict stage performance, a vertical line is drawn from the aqueous feed tenor to meet the operating line, and then horizontally to meet the equilibrium isotherm : this point on the isotherm represents the gpl copper in the aqueous and organic phases leaving the first stage of extraction.

In practice, as stated, the reaction will not reach equilibrium exactly and say 90 – 95 percent efficiency can be assumed. This is therefore allowed for in figure (viii) by stopping the horizontal line short of the isotherm by that factor (see later). Again this is a first approximation since the true 90 percent would be represented by the diagonal of the rectangle completed about the first stage, relative to the extended diagonal touching the isotherm : this naturally also is a matter of trial and error but readily achieved with practice. The example shows that the first (E.1) stage of extraction in the multistage system would result in a loaded organic of 14.5 gpl copper and an intermediate aqueous raffinate of 4.75 gpl copper.

Continuing the stepping-off process as shown, always allowing for the stage efficiency, it can be seen that the aqueous copper tenor drops from 12 gpl in 3 stages to 0.3 gpl, whilst the loaded organic rises from 6.3 gpl to 14.5 gpl. Naturally, it will be unlikely that the chosen raffinate level will be achieved in an integer number of stages and therefore a trial and error method must be used, changing the raffinate and hence the origin of the operating line : alternatively, a different O/A ratio could be chosen to change the slope of the operating line. The value of doing such an exercise in too much detail is of course questionable in view of assumptions on stage efficiency, etc.

It is apparent though, that this stepping-off procedure, allowance for stage efficiency, and iteration regarding raffinate levels and O/A ratios, can be very tedious to other than a skilled technologist and with this in mind, I.C.I./Acorga has developed the computer programme referred to previously. The programme can operate in design mode or performance mode. In either case the extraction and strip isotherms are needed as input and then the computer will either decide the O/A ratio, number of stages, etc., to achieve a chosen raffinate, or alternatively will estimate the best achievable raffinate for a given set of plant parameters.

To complete the overall picture, a stripping isotherm is needed, prepared as described earlier, and this also is used in a McCabe-Thiele construction. Fig (ix) shows the construction for the example in question.

strip-isotherm

In strip, Acorga extractants readily achieve stage efficiencies of 100 percent and these are assumed in the McCabe-Thiele construction. The construction technique is otherwise identical to that for extraction.

Parameters Which Influence Isotherm Shape 

and Their Influence on McCabe-Thiele Constructions

It can be readily appreciated from the foregoing, and particularly from figs (viii) and (ix), that design and predicted performance of a solvent extraction system are largely determined by the McCabe-Thiele constructions. Some of the more important variables which influence the isotherm shape and hence the actual construction are summarised in table (iii) and commented upon later.

Table (iii)

Parameters affecting extraction isotherms

and McCabe-Thiele Constructions

  1. Type/Strength of extractant
  2. Shape of isotherm
  3. Concentration of extractant
  4. pH of feed liquor and buffering actions
  5. Copper content of feed liquor
  6. Stripped organic level
  7. O/A ratio and slope of operating line
  8. Diluent choice
  9. Approach to equilibrium or stage efficiency

Type & Strength of the Extractant

Figure (x) shows the effect on the extraction isotherm associated with a particular feed liquor when a first generation extractant is replaced by a second generation extractant such as Acorga P.5100.

The steepness of the curve is increased markedly and it can be seen that the same raffinate level is achievable in 2 stages of extraction compared to 3 stages with the original extractant, using much the same O/A ratio.

(It will be noticed that the operating lines have different origins due to the different tenors of copper in the stripped organic.)

The steepness of the isotherm can be affected by a number of parameters but the strength of the extractant, as demonstrated in fig (x) is the most significant and basic influence.

comparison-of-acorga

Shape of the Isotherm, “S” effect

Apart from the steepness of the isotherm, it is possible to experience a modification to the shape, in the form of an “S”.

An isotherm exhibiting this effect is most unsatisfactory if low raffinate levels are required, because of the decreasing efficacy of the final extraction stages. Figure (xi) illustrates the effect and shows that it would be uneconomic – or, at best, very costly -to achieve a low raffinate level using so many stages of extraction.

the-s-curve-extraction-isotherm

Such an “S” effect is frequently found with weak extractants and low pH of feed.

Concentration of Extractant

Any change in conditions, e.g. feed concentration, acidity, extractant type, extractant concentration, etc., demands that a new isotherm be generated. Figure (xii) illustrates the effect of increasing the concentration of P.5100 from 13 v/o to 17 v/o, and without actually constructing the McCabe-Thiele diagram, it is obvious that the number of stages to give the same raffinate (for a given O/A ratio) will be decreased. As will be seen later, the operating lines will have different origins (about 2.8 gpl in organic phase and about 3.5 gpl in organic phase, for the 13 v/o and 17 v/o respectively), due to different stripping performance but in fig. (xii), the 13 v/o extractant would also need an O/A ratio of at least 3:1 in order to function at the 12 gpl feed tenor and this would demand very high flowrates and plant size.

variation-of-p-5100-extraction-isotherm-shape-with-extractant-concentration

pH of feed liquor and buffering action

Because of the general reaction:

ph-of-feed-liquor-and-buffering-action-general-reaction

the extraction of copper by these organic reagents is favoured by low acidity : the loading capacity of the extractant is decreased at low pH values. This effect is illustrated in fig (xiii) for 15 v/o P.5100 over the pH range 0.5 to 3.5 although for second generation reagents like Acorga P.5100, the effect is considerably less than for the first generation extractants : these latter could only load about 0.25 gpl copper per v/o extractant, even at pH2 and performance fell markedly at lower pH values.

loading-capacity-gpl

Consequently, the effect of increased acidity of feed on the extraction isotherm is predictably detrimental and isotherms are depressed. However, in the case of the stronger second generation reagents like P.5100, the effect is small and table (iv) shows data from a pilot plant where in 2 stages of extraction, raffinate levels (and hence metallurgical recoveries) were maintained over a wide range of feed acidity.

variations-in-acid-feed-solutions

Copper content of Feed liquor

The effect of increased copper content in feed liquor is illustrated in figure (xiv).

variation-of-extraction-isotherm-shape

For any chosen copper tenor in the organic phase, the copper tenor in the aqueous phase – with which it is in equilibrium – will be lower for the isotherm associated with the lower copper feed tenor. This follows logically from the effect of the acid generated in the reaction, as described previously.

Stripped Organic Level

It has been explained earlier, that the origin of the operating line is determined by the copper tenors of stripped organic and aqueous raffinate. Hence, the effect on the McCabe-Thiele is obvious by reference to any of the constructions in previous illustrations, e.g. Fig (x).

For convenience, figure (xv) is included to show that a small shift in the origin of the operating line (due to an increase in copper tenor of stripped organic), necessitates an increase in O/A ratio if the same raffinate level is to be maintained. This in turn means increased flowrates and hence increased equipment size. Alternatively, if the same O/A ratio is maintained, resulting in the same operating line slope, then increased copper tenor in raffinate must be tolerated.

effect-of-increased-tenor-of-stripped-organic

A complementary diagram, in figure (xvi), shows the stripped organic level attainable as a function of extractant concentration, using different strip liquors. As might be expected, higher strip acidities and lower extraction concentrations minimise the copper tenor of stripped organic. This is confirmation of comments made earlier.

variation-of-minimum-stripped-solvent-copper

O/A ratio and slope of operating line

The O/A ratio determines the slope of the operating line as described previously. Increasing the ratio decreases the slope and minimises the number of stages on a McCabe-Thiele construction : alternatively, it can decrease raffinate levels attainable in the same number of stages.

Taking an extreme and impracticable case as shown in fig (xvii), an increase in O/A ratio from 1.17:1 to 6:1 would reduce the number of stages required from 2 to 1 – albeit at the expense of an enormous plant size.

effect-of-o a-ratio-on-mccabe-thiele

Diluent Choice

The diluent is normally chosen for its flash point and for its influence on phase disengagement, extractant solubility and reaction kinetics: it does not influence the shape of the isotherm but, of course, the reaction kinetics influence approach to equilibrium and hence the McCabe-Thiele construction.

Approach to Equilibrium – Stage Efficiency

This was referred to earlier and featured in figure (viii). Apart from this figure, all the others have – for simplicity – assumed 100 percent stage efficiency.

It is obvious that anything less than 100 percent efficiency will result in more stages being required and this is self-explanatory. Fortunately, Acorga reagents exhibit very rapid reaction kinetics, and, consequently, stage efficiencies are usually considerably greater than 90 percent in extraction and near 100 percent in strip.

Factors affecting approach to equilibrium are, of course, reaction kinetics, available mixer time, O/A ratios within mixers, temperature, phase continuity, intensity of agitation, etc. – and, of course, short circuiting within a mixer.

Construction of a McCabe-Thiele diagram making allowance for stage efficiency is more time consuming than making the assumption of 100 percent, because all constructions are usually done by iteration. However, as mentioned earlier, I.C.I./Acorga have developed a computer programme for exactness and speed.

At this juncture, it is worthwhile to elaborate upon the precise interpretation of McCabe-Thiele diagrams, almost by way of a summary of the paper.

effect-of-stage-efficiency

Figure (xviii) is drawn – in rather a contrived fashion since there are so many interdependent variables – to illustrate the effect of stage efficiencies of less than 100 percent.

Referring to the figure, the heavy continuous line pertains to 100 percent stage efficiency whilst the dotted line is less than 100 percent.

Taking the former case, aqueous feed enters the SX system (E.1 mixer) at copper tenor Xf. The tenor at which it leaves mixer E.1 is represented by point B on the isotherm – obtained by drawing lines vertically through Xf to meet the operating line (at A), horizontally through A to meet the isotherm at B, (and then for subsequent work, vertically downwards to meet the operating line at C). The final aqueous raffinate tenor leaving mixer stage E.2 is obtained by moving horizontally from C to meet the isotherm at D (and then vertically downward to meet the operating line at E). These points D & E represent raffinate at copper tenor Xr.

Conversely, stripped organic enters stage E.2 at copper tenor Ys, i.e. at point E on the operating line, and leaves at a tenor equivalent to point D (or C) and then loaded organic finally leaves stage E.1 at tenor equivalent to point B (or A), i.e. tenor Y1.

It is assumed to be now understood that the points on the isotherm represent mixer discharge tenors. The operating line introduces the concept and effect of relative volumes of liquors and hence distribution of mass of copper: the upper end of the line (A) represents gpl copper in aqueous feed to stage E.1 and gpl copper in loaded organic leaving E.1, whilst the lower end (E) represents gpl copper in aqueous raffinate leaving the final extraction stage and gpl copper in stripped organic entering this final stage. For example, therefore, aqueous entering stage E.1 at copper tenor Xf mixes with organic leaving stage E.2 at copper tenor equivalent to point D, the mix being represented by point K. In order to reach equilibrium, point K moves to point B on the isotherm with redistribution of copper between organic and aqueous phases. It will be noted that the position of point B is determined by the operating line such that BC over BA (or AK over KC) is the slope or A/O ratio.

For a fuller explanation of the McCabe-Thiele construction, we will now proceed to the situation where stage efficiency is less than 100 percent.

The dotted lines are constructed in a manner similar to the above, but on each approach to the isotherm, the construction stops short by an amount depending upon stage efficiency. Hence the aqueous feed at A stops at B1, drops to the operating lines at C1, approaches the isotherm horizontally as far as and then drops to the operating line at E1 : in this particular case it can be seen that a third extraction stage is necessary to arrive at final point E via point G.

Again, conversely, the tenors of the organic phases leaving stages E.3, E.2 and E.1 correspond to values at points, G, D1 and B1 (or E1 C1 and A).

It is the determination of points B1, D1 and G which needs explanation. I.C.I./Acorga can do this using a computerised programme but it is relatively straightforward to proceed by trial and error, or iteration.

Taking the first stage of extraction, namely E.1, as the example, the organic phase enters from point at the corresponding copper tenor and meets with incoming aqueous liquor, tenor Xf, from point A. The separate phases could be represented jointly by point H. When mixed in stage E.1 mixer, an equilibrium should be reached, represented by point J on the isotherm : however, in this case, with a stage efficiency of only 82 percent, the mix moves towards point J but actually only as far as point B1, such that H-B1 is 82 percent of H-J. This is the geometric representation which is arrived at by trial and error.

It is common, although incorrect, practice to merely apply the stage efficiency factors to the horizontal lines, AB1, C1D1 E1G, etc. This is in fact acceptable at the lower part of the isotherm if it is very steep, i.e. for stages E.4, E.3 for example, but the short method can be very wrong at the upper part of the curve, i.e. stage E.1. In fig (xviii) the short method corresponds to a stage efficiency of only 70 percent whereas, in reality, the stage efficiency is about 82 percent for stage E.1.

Acorga Data Bank

The foregoing notes should have indicated how the solvent extraction process fits into a hydrometallurgical flowsheet, should have highlighted the amount of data required and should also have explained how the metallurgist can go about preliminary design of a solvent extraction system.

There are many variables involved, of which account must be taken, both in the leaching/electrowinning processes and in the solvent extraction process itself, and many of these variables are interacting. Notably, for example, the tenor of spent electrolyte is critically important because this is the strip liquor also.

Leaching and electrowinning conditions are normally dictated by the operating company, although frequently such conditions are modified to be more compatible with solvent extraction or to take advantage of more efficient solvent extraction performance. With solvent extraction plant, however, much help can be received from reagent manufacturers in terms of predicting design and performance at an early stage in the project, and indeed continuing association throughout further more detailed development work is most beneficial.

Acorga now carries a large data file of isotherm and McCabe-Thiele information for most of the solvent extraction conditions likely to be met in practice, and therefore can rapidly give a first assessment of the reagent type and concentration required for any potential project. The number of extraction and strip stages can also be reasonably reliably predicted to assist in capital costing, and consequently very preliminary estimates without testwork are much more realistic.