Available online at www.sciencedirect.

com

Hydrometallurgy 90 (2008) 103 114 www.elsevier.com/locate/hydromet

Saltpeter extraction and modelling of caliche mineral heap leaching

John A. Valencia a , David A. Mndez a , Jessica Y. Cueto b , Luis A. Cisternas a,b,
a b

Department of Chemical Engineering, Universidad de Antofagasta, Antofagasta, Chile Center for Mining Scientific Research and Technology (CICITEM), Antofagasta, Chile

Received 11 April 2007; received in revised form 26 September 2007; accepted 2 October 2007 Available online 11 October 2007

Abstract The leaching of heaped Caliche minerals represents a valid alternative for the saltpeter extraction, giving good recovery levels with excellent economic projections and technical results when compared with existing technology. This study reports on experimental tests of the leaching of this mineral in columns in which we determined the recovery of nitrate and magnesium at different heights through the bed of the mineral. The columns were sampled at three different heights, recovering the strong solution from sample ports at each height and calculating the recoveries of the target materials at each height. These variables were analyzed as a function of time and irrigation ratio, thus obtaining empirical kinetic expressions. A simple mathematical model was constructed which represented the leaching process for the solutes, considering the variation in height of the bed of the leached mineral. The model showed a good fit for predicting the concentration of nitrate and magnesium. 2007 Elsevier B.V. All rights reserved.
Keywords: Heap leaching; Caliche; Sodium nitrate

1. Introduction Saltpeter is a white salt which is translucent and bright, composed primarily of sodium nitrate. It forms thin crusts on the surfaces of rocks and walls of the mineral and often forms a superficial horizon of some soils in Chile (Pokorny and Maturana, 1997), Spain, Iran, Egypt, and India. Saltpeter is commercially important as a fertilizer, food preservative, in glass manufacture, and in some medicines as a diuretic. It was historically important in the manufacture of gunpowder, and is currently used in explosives, fireworks (rockets), and matches, as well as in metallurgical smelting agents. It is an important raw

Corresponding author. Department of Chemical Engineering, Universidad de Antofagasta, Antofagasta, Chile. E-mail address: lcisternas@uantof.cl (L.A. Cisternas). 0304-386X/$ - see front matter 2007 Elsevier B.V. All rights reserved. doi:10.1016/j.hydromet.2007.10.001

material for obtaining nitrogen in the manufacture of certain compounds such as nitric acid, and as an oxidizing agent in many industrial chemical processes. The main procedures for the extraction of nitrate from caliche mineral were worked out by Pedro Gamboni (a Chilean) in 1853. Twenty years later the English mechanical engineer James T. Humberstone adapted a countercurrent leaching process originally developed by Shank for obtaining soda ash by the Le Blanc process. The latter method allows processing caliche minerals containing 15% nitrates. After the invention of synthetic saltpeter, the Shank process became uneconomical, and was replaced by a cold leaching process developed by Guggenheim in the 1920's. Presently, this process is only used at two worksites, including Maria Elena and Pedro de Valdivia in northern Chile. More detailed information in this area has been presented by the study of Wisniak and Garces (2001).

104

J.A. Valencia et al. / Hydrometallurgy 90 (2008) 103114

Nomenclature A a b Cj,i hi H kj,i L l m n qi qi Rji Ri rr R Si Vsi VGi Vji x zi Column cross-sectional area Parameter Eq. (2) Parameter Eq. (2) Concentration of species j in section i Height of section i Initial height of each section Solution constant of species j in section i, m/h Total height of the column Mean particle size, mm Parameter of Rosin Ramler model Constant in Eq. (11) Flow of solution exiting section i Flow of solution exiting column Radius of soluble particle j in section i Particle radius in section i Irrigation ratio, m3/ton Initial radius of soluble particles Section i Cylindrical volume of section i Volume of insoluble particles in section i Volume of soluble particles j in section i Particle size, mm Adimensional height section i

Greek letters A dimensional concentration Equation parameter 18 Equation parameter 17 Density i Relative porosity of section i Adimensional radius i Delay time in section i Adimensional time Subscript i Section j Soluble species Superscript e Experimental s Saturation

A current existing demand on the world market for low grade saltpeter in fertilizer manufacture has promoted the exploitation of caliche mineral in the Tarapac and Antofagasta regions of Chile. This led to modifications in the original methods of extracting

nitrate. Since 1990 the leaching process has been used on heaps of the mineral at Pampa Blanca, Chile, similarly to that used with some other metals (Dixon and Hendrix, 1993; Qin et al., 2007; Wan and LeVier, 2003; Cariaga et al., 2005), with the leaching agent for caliche

J.A. Valencia et al. / Hydrometallurgy 90 (2008) 103114

105

being water. The leaching of piled minerals represents a valid alternative, with excellent economic and technological possibilities and good levels of recovery when compared with current extraction technology and transformation of the product. The leaching of heaps consists of feeding the solvent onto a mass of mineral having a determined grade, so that the solvent percolates through the mineral, carrying away the solute. The enriched solution is recovered from the base of the mineral mass which rests on an impermeable mineral or plastic-covered base. The leaching of caliche differs specifically from the leaching of copper and precious metals minerals. With caliche there are several soluble species while in the case of copper and gold minerals there are few. In the case of copper and gold minerals, dissolution occurs through a chemical reaction and the resulting solutions are diluted; in the case of caliche, dissolution is by simple solubility and the final solutions must be concentrated. The objective of the present research is to study the process of leaching of caliche heaps, with particular attention to the recovery of nitrate and magnesium. The latter is important because it is a control variable in the caliche leaching process. The objective was pursued by establishing experimental tests and then developing a mathematical model representing the leaching process which would allow better understanding of the phenomena involved, and leading to new possibilities for improving the technology. 2. Experiment 2.1. Materials and equipment Testing was carried out in columns using caliche mineral from the Antofagasta Region of Chile. Specific chemical analyses were carried out on the types and grade values of materials composing the mineral. Analytical methodology applied included volumetric oxidation reduction for iodine (I2), gravimetry for the insolubles and sulphate (SO4), atomic absorption spectrometry (AA) for sodium, potassium, calcium, and managanese (AA; Varian Corp., model 220FS instrument), precipitation volumetry for chlorine (Cl), nitrate (NO3) and perchlorate (KClO4) were measured by molecular absorption using a UNICAM Corp. model UV2 instrument, carbonate (CO3) by acidbase volumetry, and moisture by difference in weight. The results are shown in Table 1 (raw column). Mineralogical composition was determined using X-ray diffraction using a Siemens model 5000 automated and computerized diffractometer. This analysis provided a general impression of the species more or

less abundant in the sample; abundant species were those occurring at above 5% and the less abundant species those occurring below this percentage. The goniometer used was a vertical BragaBrentano, with a wave-length radiation of 15,406 (CuK1). The secondary monochromator was of graphite, with 1 mm/ 1 mm/0.1 mm slits. The scanning range was between 3 and 79 (2), time interval 1.0 s, and the database used was that of the International Center of Diffraction Data (ICDD). The main species observed included: nitrate (NaNO3), halite (NaCl), sodium anorthite ((Ca,Na) (SiAl)4O8) and quartz (SiO2). The minor species included: anhydrite (CaSO 4 ), glauberite (Na 2 Ca (SO4)2), loeweite ([Na12 Mg7(SO4)13] 15H2O), calcite (CaCO3), polyhalite (K2Ca2 Mg(SO4)4 2H2O), probertite (NaCa(B5O7)(OH)4 3H2O), gypsum (CaSO4 2H2O) and illitemontmorillonite (insoluble clay). A granulometric analysis of the mineral was carried out using the following Tyler screens: 1, 3/4, 1/2, 5, 8, 20, 30, 70 and 100. The results obtained are shown in Fig. 1, together with the results of fits to the distribution of particle size proposed in the RosinRamler model (see Macas-Garcia et al., 2004), where x represents the particle size in millimeters (mm), l is the mean size of the particle and m is the adjustable parameter characteristic of the particle distribution. h xm i Accumulated k Retanied exp 1 l In the case of the caliche analyzed, the value of l was 12.156 mm, and of m was 1.624, with a correlation factor R2 = 0.9943 and an estimated standard error of 0.0282. Preliminary leaching tests were carried out in columns which demonstrated that large amounts of fine material
Table 1 Results of chemical analyses of the caliche mineral for column leaching in the present study in untreated form (RAW) and classified into particles N 9.53 mm (CLASS.) Element I2 Insoluble Na K Ca Mg Cl SO4 NO3 CO3 KClO4 Moisture Ionic balance Unit ppm % % % % % % % % % % % Raw 647 54.94 9.36 0.99 2.10 0.78 4.60 14.97 10.15 0.03 b 0.1 2.23 0.39 Class 509 60.19 8.08 0.73 1.50 0.70 3.85 12.55 8.80 0.05 b 0.1 3.20 0.94

106

J.A. Valencia et al. / Hydrometallurgy 90 (2008) 103114

Fig. 1. Experimentally obtained granulometric particle size distribution (PSD) of caliche mineral of the present study compared with the theoretical RosinRamler distribution (Macas-Garcia et al., 2004).

was formed, causing problems due to channeling, and to transport of the fines. In order to avoid these problems we decided to classify the mineral to be leached, loading the columns with material having particle sizes greater than 3/8 inch (9.53 mm). In order to determine the effect of the classification procedure, the chemical composition of the mineral was re-determined, producing the results presented in Table 1. No significant differences were found between the initial results (raw) and those presented (class.) in Table 1. The preliminary tests were also used to estimate the time required for the mineral to be completely wetted prior to beginning the leaching proper; the time required for this was 10 h for the total mineral loaded. 2.2. Procedure Leaching tests were carried out following methods typically used by the industry (NEN 7343, 1995), but using water as the leaching solution. The procedures for storage and transport of the solids used were based on CEMA (Conveyor Equipment Manufacturer Association) and ASTM (American Society for Testing and Materials) norms. CEMA indicates that the criterion of the dimension of the experimental model can be a cylindrical recipient with a diameter of about 20 times the particle diameter. Thus for carrying out the tests a column was used with an internal diameter of 200 mm since more than 89% of the mineral loaded had a diameter of less than 20 mm. In order to simulate a leaching process for columns of different heights, a column was prepared which was divided into three different heights of the mineral, where

samples of the leaching solution could be removed from the column at each of the three heights. Enriched solution samples extracted from each height could then

Fig. 2. Schematic diagram of equipment used in the experiment.

J.A. Valencia et al. / Hydrometallurgy 90 (2008) 103114

107

be analyzed for concentrations and recoveries of nitrate and magnesium. Divisions of the mineral at the different heights in the column were established by installing a perforated plexiglass plate and a drain bib at each height from which samples of the leaching solution could be extracted. A pad of glass wool was installed at each level of the mineral to prevent the movement of fines from one level to the next. The experimental scheme is shown in Fig. 2, and Table 2 lists the disposition of the mineral modules which define the different heights in the column. A low flow rate of water (4.8 l/h m2) was delivered by a peristaltic pump in order to minimize transport of fine particles by the leaching flow. The tests were carried out at room temperature, using potable water at a temperature of about 25C. The mineral in the columns was irrigated at first until completely wetting the entire bed and then the solution was maintained without draining up to 10 hours, which was the period selected based on preliminary tests, so that the dissolving fluid at the exit port from each height would be the same as that entering and would reduce the effect of absorption of moisture by the mineral. This was then followed by continuous dropwise irrigation. The water was fed into the first module, from which it was drained (solution from height 1), then directed into the second module (solution from height 2), and finally into the third (height 3). Continuous monitoring was carried out, taking samples every 12 hours from the outputs of each module. After 250 h of irrigation, entry of water into module #1 was stopped and the leaching solution was allowed to flow until each module had been drained. Representative bed samples were then taken from each module in order to make mineralogical analyses of the residue, determine which salts had been leached, and making other analyses of the solutions. Table 3 shows the results of chemical analyses of the leached mineral residue left in the column. This sample was prepared from a composite of subsamples of residues at each of the levels. Assays of nitrate and magnesium for each level are given in Table 4. The results of the chemical analyses of the leached mineral showed that since most of the soluble species had been removed, the percentage of insolubles in the residue
Table 2 Characteristics of leached modules Height (modules) Column height, mm Mass of caliche loaded, Kg. 1 2 3 910 1820 2730 27.20 54.40 81.60

Table 3 Results of chemical analyses on residue of leached caliche mineral remaining in the leaching column Element I2 Insolubles Na K Ca Mg Cl SO4 NO3 CO3 KClO4 Moisture Ionic balance Unit ppm % % % % % % % % % % % Content 37 88.86 0.09 0.26 2.38 0.14 0.01 6.23 0.06 0.00 b 0.1 17.54 1.41

increased from 60.19 to 88.86%. The percentage recovery of nitrate reached 95.62%, magnesium 94.59%, sulfate 68.28%, iodine 95.35% and chloride 99.83%. The X-ray analysis showed the species in major abundance in the residue were quartz (SiO2), bassanite (CaSO40.5H2O) and sodium anorthite [(Ca, Na)(Al, Si)4O8]. The lesser abundant species included orthoclase (KAlSi3O8), calcite (CaCO3), Mg-vermiculite [(Mg2.36Fe0.48Al0.16)(Al1.28 Si2.72)O10(OH)2(H2O)6 Mg], montmorillonite [Ca0.2(Al, Mg)2Si4O10(OH)24H2O], probertite [NaCa(B5O7)(OH)4 3H2O], kaolinite [Al2Si2O5(OH)4], loeweite [Na12 Mg7 (SO4)1315H2O], moscovite [KAl2(Si3Al)O10(OH,F)2] and gypsum (CaSO42H2O). When comparing the mineralogical results from the leached residue with those in the original mineral sample, it can be seen in that the species containing the nitrate disappeared from the more and less abundant species resulting in a total recovery of nitrate of over 95%. In the case of the magnesium, the species which contained this element in the raw mineral were loeweite and polyhalite. The first of these remained present in the residue (probably partially dissolved), but the polyhalite was completely dissolved. It was apparent that the dissolution kinetics of loeweite and polihalite were different, governing the solution behavior of the magnesium as noted in assays of the strong solution from different levels, which gave two different solubility rate curves for magnesium.
3. Results and discussion 3.1. Leaching recovery as a function of irrigation time The percentage recovery of nitrate and magnesium were determined from the initial grade value of the mineral and from the final analysis of the leached residues in each module. The trends in percentages of recovery of nitrate are shown in Fig. 3 as a function of irrigation time.

108

J.A. Valencia et al. / Hydrometallurgy 90 (2008) 103114

Table 4 Results of column leaching tests and estimations of percentages of recovery as a function of residue assays Height Irrigation time, h 250 336 420 Irrigation Ratio m3/ton 1.36 0.92 0.76 Height of residue mm 600 1200 1800 Mass of residue kg 17.39 34.74 34.74 Residue assays,% NO3 1.41 0.60 0.60 Mg 0.046 0.058 0.141 %, Recovery NO3 89.75 95.65 95.62 Mg 95.80 94.68 87.16

1 2 3

The nitrate recovery curves showed that excellent percentages of recovery were obtained at heights 2 and 3 (Table 4) when compared with the recovery obtained at the lowest height. The best dissolution was observed at the lowest height (Height 1) during the initial hours of irrigation. The lesser amount of recovery obtained at the first height can be explained by the fact that the leaching time was less than at the others levels, resulting in less dissolution of the nitrate at the lower level. This is shown by the lower levels of nitrate in the leached residues in Table 4, it is also possible to determine the final height of each mineral leached, as this decreases by 34% compared with the initial, and the mass of the gravel decreases by 36% compared with the initial mass of mineral loaded. These percentages are approximately the same for each height of mineral. The concept of the kinetic as mentioned above is more clearly demonstrated by analysis of the magnesium results (Fig. 4). That is, solution of the magnesium was more rapid in the lower portion of the column. Nevertheless, there was a difference between the magnitude of solution of the magnesium and that of nitrate, as in the case of magnesium the recovery is greater at Height 1 although the irrigation time at this height was less than in the others as shown in Table 4. 3.2. Leaching recovery as a function of irrigation ratio A more detailed analysis of the behavior of the recovery of nitrate and magnesium is obtained when it is observed as a

function of irrigation ratio. The term irrigation ratio, rr, represents the quotient between the volume of the leaching solution fed and the tonnage of mineral being irrigated, is widely used in the mining industry instead of leaching time, since the operational conditions may vary and the leaching time may be a variable, which is difficult to quantify. In contrast, the volume of leaching solvent is easily quantified. The irrigation ratios obtained in the test were less at Heights 2 and 3, since they represented double and triple the mass that contained at Height 1. With respect to the recovery of nitrate as a function of irrigation ratio (Fig. 5), it can be noted that at the greatest height, the kinetic begins slowly until reaching an irrigation ratio near 0.3 m3/ton, then beginning to accelerate until reaching and exceeding the kinetic curves of Heights 2 and 3. The contrary case is observed at the lowest level (Height 1) which shows a rapid kinetic with the lowest irrigation ratios, but as these increase, the kinetic slows, becoming practically constant for irrigation ratios greater than 1.2 m3/ton. As might be expected, in the middle level (Height 2), there is intermediate behavior between the two extremes mentioned. Nevertheless, the greatest recovery is obtained from the middle level, and this can be attributed to the fact that a sufficient irrigation ratio is not obtained at the greatest height (Height 3) for obtaining a greater nitrate recovery5. The behavior of the recovery of magnesium as a function of the irrigation ratio (Fig. 6) at the lowest level (Height 1) has the more rapid kinetic, which is subsequently reached by the

Fig. 3. Percentage recovery of nitrate (NO3) from different heights in the column as a function of irrigation time.

J.A. Valencia et al. / Hydrometallurgy 90 (2008) 103114

109

Fig. 4. Percentage recovery of magnesium (Mg) from the column at different heights as a function of irrigation time.

kinetics of the other levels, although these do not produce the same recovery as the first, since they never achieve an irrigation ratio of greater than 1.0 m3/ton. The experimental data were fit to empirical expressions, finding that the following models provided satisfactory fits for all the heights: Recovery 1 exp a rrb 2

Where rr is the irrigation ratio. The parameters a and b of empirical expression 2 are given in Table 5, together with the coefficient of fit R2 and a standard error estimated for each of the fits. The parameters were obtained by linearizing expression 2, and fitting the data to a straight line. 3.3. Leaching model Models have been found in the literature which have analyzed the heap leaching process, considering unreacted core model with generation of ash (see for example da Silva, 2004 and KNona and Liddell, 2005). Therefore those works

have not included the variation of particle size without generation of ash, with which is generated a decrease in the height of the bed as we wish to demonstrate as follows. We will consider that the column is cylindrical in section, and represents an element of volume within the heap. This cylinder is then subdivided into i sections, with the volume of each section represented by VS,i equal to the volume of a cylinder, i.e. VS,i = A hi, where hi is the height of the cylindrical section and A represents the transverse area of the bed. The length of the column L, can vary as the reaction proceeds as will be seen below. Fig. 7 shows the division of the column into sections which have an input flow of q0 to its upper portion and output flow qL from its lower portion. Dissolution of one or various species occurs in each section. The model developed below considers the following assumptions: 1. A plug flow through the mineral bed is assumed, although due to the mathematical complexity of the resolution, the model is represented by completely mixed reactors in series with a time delay. In this way, for each given section, a

Fig. 5. Experimental recovery of nitrate (NO3) and kinetic fit, as a function of irrigation ratio.

110

J.A. Valencia et al. / Hydrometallurgy 90 (2008) 103114

Fig. 6. Experimental recovery of magnesium (Mg) and kinetic fit, as a function of irrigation ratio.

2.

3.

4.

5.

6.

completely mixed reactors with a time delay was considered. In this work only 1 section for height was necessary. It is assumed that the solid to be leached occurs as particles of pure mineral without mixtures of insoluble materials or other soluble species, also, that the insoluble particles do not contain soluble materials and thus do not take part in the leaching process. For the soluble particles, it is assumed that dissolving effects the shrinking of spherical particles, without the production of ash, so that the particles progressively decrease in size, and with this there is a variation in the height of the bed, which decreases with leaching of the material. The leaching of nitrate from caliche generates little residue, and few deposits (this differs from the leaching of copper sulfide minerals where residues are generated and deposits are formed due to the formation of sulfur on the gangue and even on the mineral). For simplicity it is considered that the porosity of the bed is constant, even though the porosity of the mineral changes due to removal of soluble species, and there are variations in the height of the bed. The geometry of the soluble mineral particles was considered to be spherical. The temperature for any effect was considered constant, that is, no consideration was given to salt solubility, changes in density, and kinetics of the dissolution resulting from temperature changes. The size distribution of the mineral particles to be leached was not considered. A single mean particle size was

assumed. This was done in order to obtain a simple leaching model, and considering that the leaching process is carried out considering uniform particle sizes to avoid, on one hand, obstructing the passage of the solution by fine particles, and on the other hand, decreasing the percentage of recovery due to the presence of larger particles. 7. The saturation concentration of a determinate species depends on the other dissolved species and therefore its the changes in the time. The saturation concentration for magnesium was assumed to be 20 g/L. and for NO3 250 g/ L as determined by experimentally saturating a solution with the initial mineral. Fig. 8 shows a section which assumes the existence of spherical particles of gangue (insolubles) and soluble species having the same initial radius. The section has a total volume of VS,I, which represents the sum of the volume of the insoluble particles VG,i, plus the volume of the soluble particles, PJ j1 Vj;i , and the volume of the interstitial solution, Vi.

Table 5 Parameters for the recovery fit of nitrate (NO3) and magnesium leaching of caliche mineral in columns at different heights Height Nitrate a 1 2 3 2.01 3.94 3.91 b 0.70 1.19 1.32 R
2

Magnesium a 2.25 3.01 2.59 b 0.99 1.20 1.20 R2 0.99 1.00 1.00

0.96 0.99 0.99

Fig. 7. Division of the leaching column into sections.

J.A. Valencia et al. / Hydrometallurgy 90 (2008) 103114

111

The relation between volume Vi and VS,i is given by the porosity, i. Vi ni VS ;i Vi J P

j1

particle dissolves radially, thus causing a lowering in the height of the bed. A
J drj;i dhi 4p X Nj;i rj2;i dt dt 1 ni j1

ni VG;i

5 Vj;i Vi

It is assumed that the particles are spherical, and thus, the volume of each type of particle that will be leached in the section of the column is: 4 Vj;i Nj;i prj3;i 3 6

Where Nj,i is the number of particles of species j contained in volume VS,i , whose particle radius is rj,i. The porosity of the bed may change with time, in relation to change in the volume of the solution Vi and less species which are dissolved Vj,i. Nevertheless, in the present study it is assumed that the porosity is constant and equal at a mean value during the leaching. This supposition is required to maintain the simplicity of the equations of the model. Deriving Vj,i with time, following Eq. (6), we obtain: dVj;i drj;i 4pNj;i rj2;i dt dt 7

Eq. (9) shows the change of hi depends of the instantaneous particle dissolution area and the change of particle size, where the initial condition for height is hi(0) = Hi. It is reasonable to consider that each section has the same initial height and therefore hi(0) = H. Later will be demonstrated particle size decreases over time, and therefore we expect hi decreases. The variation in radius of leached particles is related to the supposition that these particles dissolve without leaving ash, and their size becomes reduced radially. Simultaneously, the rate in mass reduction of each particle is proportional to the surface area available for the dissolution and to the n power of the difference in concentration between the particle surface (saturation concentration, Cs j ) and the concentration in the bed (Cj,i), following a solution mechanism shown in Fig. 9 for a single particle. If we define the mass of a single particle as follows: 4 q pr 3 3 j j;i 10

With a constant particle density ji, the change in the mass of the particle is:  n drj;i kj 4prj2;i Cjs Cj;i 11 4pqj rj2;i dt Simplifying,  n drj;i kj Cjs Cj;i dt

Using Eq. (5) and replacing Eq. (7) it can be shown that:
J drj;i dVi 4pni X Nj;i rj2;i dt 1 ni j1 dt

qj

12

For convenience, it is taken into account that Vi = iAhi. Thus, in changing the volume of the solution, we also change the height of the cylindrical section hi, since it is considered that the

Where kj is a dissolution constant or mass transfer coefficient in m/h, n is a constant (a value of 0.6 was assumed in this work), Cjs is the saturation concentration of j and Cj,i is the concentration of chemical species j in volume Vi, and where

Fig. 8. A Distribution of the material in a cylindrical section of the column prior to initiation of leaching of species, t = 0. B Distribution of material in a cylindrical section of the column after a time period during which leaching of species has begun, t 0.

112

J.A. Valencia et al. / Hydrometallurgy 90 (2008) 103114

the initial condition for size is rj,i(0) = Rj,i. In general we can consider that all the sections have the same initial particle size, this mean rj,i(0) = Rj. Also in this work all soluble species j have the same initial particle size and therefore rj,i(0) =R. In Eq. (12), the negative sign implies that the mass is becoming smaller, and is contributing to the volume of solution Vi. Based on the equations obtained, it is possible to conceive of each cylindrical section of the column as a completely mixed reactor which functions in a quasi-steady state, due to the change in the volume of the solution. For a section i, the material balance for each dissolved species can be written assuming the controlling volume is the solution that carries away the dissolved species, thus having a positive sign for the expression of the reaction speed. The material balance is as follows: d Vi Cj;i qi1 Cj;i1 dt  n 4pNj;i kj rj2;i Cjs Cj;i qi Cj;i

aj;i

n 1 d aj;i dzi 1 zi aj;i1 kj;i r2 aj;i j;i 1 aj;i ni ni ds ds

18

with aj;i 0 1 where gj;i 4pNj;i R3 j 1 ni AH 4pNj;i kj R2 j Cj qni bj

sn1

zi 0 1 rj;i 0 1

kj Cjs AH qj Rj q

19

kj;i

13

In this equation, qi represents the flow of solution which exits the present section i, while qi 1 is that which comes from the preceding section. Considering that qi 1 = qi = q and Vi = iAhi, the definitive expression remains in the following form: ni ACj;i dCj;i dhi ni Ahi qCj;i1 4pNj;i kj rj2;i dt dt   Cjs Cj;i
n

qCj;i

14

Simulation were developed that indicate that the volumetric flow rate changes between 0.07% and 0.36%, confirming the previous supposition. In this last equation, we also have to consider a delay i for Cj,I, specific for each section of the column. Since the variation in height depends on changes in the volume of the dissolved species, we must write this same material balance for all the species susceptible to being dissolved. Eq. (14) shows an integral-differential equation which is no linear. This equation gives Cj, i(t) under initial condition Cj,i(0) = Cjs. Replacing it in Eq. (12) rj,i(t) may be obtained and then replace this last function in Eq. (9) to obtain hi(t). Eqs. (9), (12) and (14) can be re-written as a function of the following adimensional variables: aj;i C j;i rj;i hi qt zi s rj;i AH Cjs H Rj 15

It should be observed that parameter j,i represents the fraction of the volume occupied by the soluble species j in section i. Also, parameter j,i represents two ratios, the ratio of dissolving speed of the soluble species to the velocity of the solution, and the ratio between the saturation concentration and the density of the soluble species. Then the equations can be interpreted as follows: the height (Eq. (16)) of the pile decreases as the fraction of the volume of each species decreases, and that the area of each particle decreases at the speed of decrease in the radius of the soluble particles. The radii of the soluble particles decrease (Eq. (17)) as the concentration of the solution decreases, and depends on the ratio between the velocity of solution and percolation of the solution. We determined the change in concentration Cj, solving Eqs. (9), (12) and (14) using the RungeKutta method (Bequette, 1998). For the case of nitrate obtained from the experiments carried out and from the model (Fig. (10)), the parameters which were input for resolution of the model were: the dissolution constant 0.2 (m/h)(kg/m3)0.4 for the solution of the nitrate and 0.01 (m/h) (kg/m3)0.4 for the magnesium, which were obtained through iterations of the model until finding a value which fit the data of the model to experimentally obtained values; the initial mineral particle size to be leached was 6.35 10 3 m, the feed flow to the columns was 1.14 104 m3/h, the porosity of the bed was considered to be 0.2, which was

From which is obtained

J d r j;i dzi X g j;i r 2 j;i ds ds j1

16

n d r j;i bj 1 aj;i ds

17

Fig. 9. Transient balance of material of a particle in being dissolved.

J.A. Valencia et al. / Hydrometallurgy 90 (2008) 103114

113

Fig. 10. Results from the variable height model showing the experimental concentrations of nitrate (NO3) for the three heights analyzed.

estimated considering that the initial porosity of the bed was 0.4. The initial value of the porosity of the mineral was obtained by direct measurement of the quotient of the volume of the recipient and the volume of the mineral which was calculated from the mass of the mineral loaded and its density. Only one section for height was used in this work. The lag times were 0 for NO3 and Mg in height 1, 30 and 12.5 for NO3 and Mg in height 2, 70 and 25 for NO3 and Mg in height 3. These lag times are only a fraction of the experimental lag times (25 and 0 for NO3 and Mg in height 1, 100 and 25 for NO3 and Mg in height 2, and 225 and 50 for NO3 and Mg in height 3). Therefore it is clear that saturation is really the cause lag time response in the effluent concentration curves of Figs. 10 and 11. The results are shown in Figs. 10 and 11. In general there is a good agreement between the model and the experimental values. The lag times occur since the solution becomes saturated with the soluble species and thus the leaching occurs in the segments of the columns, also, although probably less significant, due to the percolation time in the column. To this, we note that as the

solution lowers through the bed it increases in concentration until reaching saturation, without having extracted all the nitrate and magnesium in the upper part of the column. It should be noted that the results of the model fit better with the nitrate experimental data, which is attributable to the fact that the nitrate closely follows the assumptions of the model. In the case of the magnesium concentrations, Fig. 11, since this species was found to be contained within 2 mineralogical species and as they do not have the same solution kinetic, the model has more difficulties to adapt to its experimental data. When analyzing the specific case of the leaching of nitrate following the model, it is possible to verify the decrease in height and radius of the particles resulting from this phenomenon. This is also verified experimentally (Tables 2 and 4). Nevertheless the decrease in height resulting in this model only represent the dissolution of NO3 and Mg and therefore does not prove to be the same as in a real scenario (34.1%) since the former does not consider the change in porosity and the dissolution of other salts contained in the mineral which produce greater decreases in the height, since

Fig. 11. Results from the variable height model showing the experimental concentrations of Magnesium (Mg) for the three heights analyzed.

114

J.A. Valencia et al. / Hydrometallurgy 90 (2008) 103114

Table 6 Deviation of the relative error of the concentrations obtained by the model, versus the experimental concentrations Height 1 2 3 NO3 0.79 0.19 0.11 Mg 0.42 0.14 0.20

the caliche is a mineral whose composition contains more than 40% material soluble in water at room temperature. The model supposes that the extraction occurs as a function of the radius of the mineral particle, and due to the difference in saturation concentration and solution in the interstices of the bed. The resolution of the preceding is adjusted to the behavior of the solution contained within the bed, validating the proposed supposition. The relative error of Eq. 20, produced by the model is shown in Table 6, where Cn is the concentration in the model and Ce n is the experimental concentration, and N is the number of data. Thus the relative mean error for the NO3 is about 15% for the last two heights, while that for Mg is about 25%. The effluent concentration in the first height has high deviation, but it can be considered that the behavior is delayed. These errors can be considered acceptable, considering the simplicity of the model, and the simplifications carried out.
N e 1X jCn Cn j Relative error e N n Cn

mineralogical species, while the Mg is present as two soluble minerals. The analysis of the residue shows that practically all of the NO3 (nitrate) was leached, while one of the Mg minerals was completely leached (polyhalite), and part of the other (loeweite) was not leached. This study is the first approximation of modeling with variation in height, which could be complemented in future studies by including the size distribution of the mineral loaded, variation in porosity in the column, and the interactive effects of solution of one species on another. Of these factors, probably the most important is the variation in porosity over time, since the particles change significantly in size during the leaching process. Acknowledgments The authors wish to thank CONICYT for support through Fondecyt Project 1020892. Luis Cisternas thanks Mr. Patricio Pinto Gallardo (SKM Minmetal) for his expert technical assistance in carrying out the leaching experiments. References
Bequette, B.W., 1998. Process dynamics: modeling, Analysis and Simulation. Prentice Hall, Inc. Cariaga, E., Concha, F., Seplveda, M., 2005. Flow through porous media with applications to heap leaching of copper ores. Chemical Engineering J. 111 (23), 151165. Dixon, D.G., Hendrix, J.L., 1993. A general model for leaching of one or more solid reactants from porous ore particles. Metallurgical Trasactions B 24B 157168. da Silva, G., 2004. Relative importance of diffusion and reaction control during the bacterial and ferric sulphate leaching of zinc sulphide. Hydrometallurgy 73 313324. KNona, C., Liddell, 2005. Shrinking core models in hydrometallurgy: What students are not being told about the pseudo-steady approximation. Hydrometallurgy 79 6268. Macas-Garca, A., Cuerda-Correa, E., Daz-Dez, M.A., 2004. Application of the RosinRammler and GatesGaudinSchuhmann models to the particle size distribution analysis of agglomerated cork. Materials Characterization 52 159164. NEN 7343, 1995. Leaching Characteristics of Building and Solid Waste Material, Leaching Tests, Determination of the Leaching of Inorganic Components from Granular Materials with the Column Test, Netherlands Normalization, Delft, Netherlands. Pokorny, L., Maturana, I., 1997. Sodium Nitrate, Fourth edition. KirkOthmer ECT Encyclopedia of Chemical Technology, Vol. 22, pp. 383393. Qin, W., Li, W., Lan, L., Qiu, G., 2007. Simulated small-scale pilot plant heap leaching of low-grade oxide zinc ore with integrated selective extraction of zinc. Minerals Engineering 20 (7), 694700. Wisniak, J., Garcs, I., 2001. The rise and fall of the Saltpeter (sodium nitrate) industry. Indian Journal of Chemical Technology 8 (5), 427438. Wan, R., LeVier, K.M., 2003. Solution chemistry factors for gold thiosulfate heap leaching. International Journal of Mineral Processing 72 (14), 311322.

20

4. Conclusions The tests of column leaching provide examples of a valid technique for the kinetic analysis of the leaching process in heaps of soluble species. The variation carried out in the development of this study included division of the mineral bed into three different heights which was done in order to understand the variation in concentration of the leaching solution at different heights in the column. It was possible then to evaluate and adjust correlations which were convenient for recovery of the species, and establishing empirical kinetic expressions for the behavior of nitrate and magnesium each height. The results show that it is possible to leach caliche for the extraction of saltpeter using water as a leaching agent to give extractions above 90% with irrigation ratios of 0.8 m3/ton. The height of the pile decreases significantly with leaching (ca. 34%) since the soluble species make up an important percentage of the mineral. The model developed was fit to experimentally obtained values for nitrate and magnesium concentrations, for the three bed heights analyzed. The results are better for leaching NO3 than for Mg; this is explained in part because the NO3 occurs as only one type of