Adsorption of Surfactants and Polymers on Iron Oxides: Implications For Flotation and Agglomeration of Iron Ore

Transcription

1 DOCTORA L T H E S I S Adsorption of Surfactants and Polymers on Iron Oxides: Implications For Flotation and Agglomeration of Iron Ore Elisaveta Potapova

2

3 Adsorption of surfactants and polymers on iron oxides: implications for flotation and agglomeration of iron ore Elisaveta Potapova Luleå University of Technology Department of Civil, Environmental and Natural Resources Engineering Division of Sustainable Process Engineering September 2011

4 Cover illustration: schematic illustrations of surfactan adsorption on a mineral particle (top), mineral flotation (bottom left), and a wet agglomerate (bottom right). Printed by Universitetstryckeriet, Luleå 2011 ISSN: ISBN Luleå

5 ABSTRACT Iron ore pellets are an important refined product used as a raw material in the production of steel. In order to meet the requirements of the processes for iron production, the iron ore is upgraded in a number of steps including, among others, reverse flotation. Under certain circumstances the flotation collector may inadvertently adsorb on the iron ore particles increasing the hydrophobicity of the iron ore concentrate, which in turn has been shown to have an adverse effect on pellet strength. To minimize the influence of the collector on pellet properties, it is important to understand the mechanism of collector adsorption on iron oxides and how different factors may affect the extent of adsorption. In Papers I-III, the adsorption of a commercial anionic carboxylate collector Atrac 1563 and a number of model compounds on synthetic iron oxides was studied in-situ using attenuated total reflectance Fourier transforms infrared (ATR-FTIR) spectroscopy. The effect of surfactant concentration, ph, ionic strength, calcium ions and sodium silicate on surfactant adsorption was investigated. The adsorption mechanism of anionic surfactants on iron oxides at ph 8.5 in the absence and presence of other ions was elucidated. Whereas silicate species were shown to reduce surfactant adsorption, calcium ions were found to facilitate the adsorption and precipitation of the surfactant on magnetite even in the presence of sodium silicate. This implies that a high concentration of calcium in the process water could possibly enhance the contamination of the iron ore with the flotation collector. In Paper III, the effect of calcium, silicate and a carboxylate surfactant on the zeta-potential and wetting properties of magnetite was investigated. It was concluded that a high content of calcium ions in the process water could reduce the dispersing effect of silicate in flotation of apatite from magnetite. Whereas treatment with calcium chloride and sodium silicate made magnetite more hydrophilic, subsequent adsorption of the anionic surfactant increased the water contact angle of magnetite. The hydrophobic areas on the magnetite surface could result in incorporation of air bubbles inside the iron ore pellets produced by wet agglomeration, lowering pellet strength. Based on the adsorption studies, it was concluded that calcium ions could be detrimental for both flotation and agglomeration. Since water softening could result in further dissolution of calcium-containing minerals, an alternative method of handling surfactant coatings on i

6 magnetite surfaces was proposed in Paper IV. It was shown that the wettability of the magnetite surface after surfactant adsorption could be restored by modifying the surface with polyacrylate or sodium silicate. In Paper V, the results obtained using synthetic magnetite were verified for natural magnetite. It was illustrated that the conclusions made for the model system regarding the detrimental effect of calcium ions were applicable to the natural magnetite particles and commercial flotation reagents. It was confirmed that polyacrylate and soluble silicate could be successfully used to improve the wettability of the flotated magnetite concentrate. The fact that polyacrylate improved the wettability of magnetite more efficiently at the increased concentration of calcium ions indicates that this polymer is a good candidate for applications in hard water. Finally, it was concluded that in-situ ATR-FTIR spectroscopy in combination with zetapotential and contact angle measurements could be successfully applied for studying surface phenomena related to mineral processing. ii

7 ACKNOWLEDGEMENTS First, I would like to acknowledge the financial support of this work provided by the Hjalmar Lundbohm Research Centre (HLRC). Secondly, I would like to express my gratitude to the people who made these four years an exciting and fruitful journey: my supervisor, Prof. Jonas Hedlund, for his guidance and trust in me; my assistant supervisor, Dr. Mattias Grahn, for all his help and encouragement, no matter what; and Assoc. Prof. Allan Holmgren for being able to solve any problem and answer any question. Further, I am grateful to Dr. Seija Forsmo and Dr. Andreas Fredriksson for valuable advice and feedback about my work from an industrial perspective. The assistance of Dr. Johanne Mouzon, Lic. Eng. Ivan Carabante, and Lic. Eng. Iftekhar Uddin Bhuiyan in working with the new SEM and of Dr. Annamaria Vilinska in zetapotential and contact angle measurements is highly appreciated. I would like to thank my close colleagues Lic. Eng. Ivan Carabante, Dr. Xiaofang Yang, Lic. Eng. Magnus Westerstrand, and Lic. Eng. Richard Jolsterå for their co-operation, sharing ideas and experiences. Ulf Mattila and Oniel Albino, thank you for saving me from the scariest PhD nightmare a broken computer during the writing of the thesis. Dear administrators, thank you for being helpful and patient when settling all the administrative issues and for not talking work and football during coffee breaks. I would like to thank my colleagues at the former Department of Chemical Engineering and Geosciences and especially my colleagues at the Division of Sustainable Process Engineering for being such great people. You are the best colleagues I could ever have! A big hug goes to all my friends outside the department, outside the university and outside Sweden for being there when I needed you and for making my life a fantastic, unforgettable adventure. A final and very special thank you goes to my family, who have always supported me from a distance, and especially to my mom for her guidance through the moments of confusion. I love you! iii

8 iv

9 LIST OF PAPERS This thesis is based on the following papers: Paper I: Studies of collector adsorption on iron oxides by in-situ ATR-FTIR spectroscopy E. Potapova, I. Carabante, M. Grahn, A. Holmgren, and J. Hedlund Industrial and Engineering Chemistry Research 49 (2010) Paper II: The effect of calcium ions and sodium silicate on the adsorption of anionic flotation collector on magnetite studied by ATR-FTIR spectroscopy E. Potapova, M. Grahn, A. Holmgren, and J. Hedlund Journal of Colloid and Interface Science 345 (2010) Paper III: The effect of calcium ions, sodium silicate and surfactant on charge and wettability of magnetite E. Potapova, X. Yang, M. Grahn, A. Holmgren, S. P. E. Forsmo, A. Fredriksson, and J. Hedlund Colloids and Surfaces A: Physicochemical and Engineering Aspects 386 (2011) Paper IV: The effect of polymer adsorption on the wetting properties of partially hydrophobized magnetite E. Potapova, M. Grahn, A. Holmgren, and J. Hedlund Submitted to Journal of Colloid and Interface Science v

10 Paper V: Interfacial properties of natural magnetite particles compared with their synthetic analogue E. Potapova, X. Yang, M. Westerstrand, M. Grahn, A. Holmgren, and J. Hedlund Full-length paper to be submitted to Minerals Engineering and accepted for presentation at the Flotation 2011 Conference in Cape Town, South Africa Author s contribution to the appended papers Papers I, II, and IV: All experimental work and evaluation, and almost all writing. Paper III: Approximately one-third of experimental work, two-thirds of evaluation, and almost all writing. Paper V: Approximately half of experimental work and evaluation and almost all writing. vi

11 CONTENTS INTRODUCTION... 1 SCOPE OF THE PRESENT WORK... 3 BACKGROUND... 5 Upgrading of iron ore... 5 Froth flotation... 6 Flotation of iron oxides... 8 Flotation effect on wet agglomeration of iron ore... 9 Surface wettability and contact angle measurements Adsorption at the solid/liquid interface Attenuated total reflectance Fourier transform infrared spectroscopy (ATR-FTIR) EXPERIMENTAL PART Materials Methods Film preparation ATR-FTIR spectroscopy Contact angle Zeta-potential RESULTS AND DISCUSSION Characterization of iron oxides (Papers I, II, V) Surfactant adsorption and factors affecting the adsorption (Papers I-III) Adsorption mechanism Factors affecting surfactant adsorption on iron oxides The effect of surfactant adsorption on the properties of the magnetite surface (Paper III) 32 Zeta-potential Contact angle Verification for natural magnetite (Paper V) Summary and implications for flotation and agglomeration of iron ore Restoring magnetite wettability after surfactant adsorption (Papers IV, V) Modification with sodium silicate vii

12 Modification with hydrophilic polymers Verification for the flotated magnetite concentrate (Paper V) Summary and implications for agglomeration of iron ore CONCLUSIONS FUTURE WORK BIBLIOGRAPHY viii

13 INTRODUCTION Adsorption of surfactants and polymers on mineral surfaces is important for many industrial applications such as detergency, coatings, flocculation of fibres and fine mineral particles, dispersion of pigments, stabilization of colloidal suspensions in cosmetics and pharmaceuticals, flotation, and agglomeration. Undesired adsorption of surfactants and polymers can be of interest, too, in cases where it has an adverse effect on the performance of a certain process. For instance, interaction of fulvic and humic acids with iron oxides is a subject of many research publications, since the adsorption of these natural polymers has been shown to impair the remediation of arsenic-contaminated soils using iron oxides [1]. Another example of surface contamination that has received much attention in the literature is the surfactant coating on iron ore concentrates upon flotation. Flotation of iron ore is performed in order to reduce the amount of certain elements, such as phosphorous, in the concentrate to an acceptable level for iron production. Phosphorouscontaining minerals, like apatite, associated with the iron ore are separated by reverse flotation using anionic carboxylate surfactants [2]. Any contamination of the iron ore concentrate with a surfactant decreases wettability of the concentrate and thus has an adverse effect on the subsequent pelletizing process and strength of the iron ore pellets produced [3-5]. In order to minimize any negative effects of surfactant adsorption on the surface properties of the iron ore concentrate after flotation, it is important to elucidate the mechanism of interactions between anionic carboxylate surfactants and iron oxides and to identify the factors that may affect these interactions. This information can further provide an idea about the possibilities of reducing surfactant adsorption on iron oxides or to restore the wettability of the iron ore concentrate after flotation. Several different natural polymers and their derivatives have been proposed as depressants of iron oxides in reverse flotation of iron ore. The depression phenomenon is complex and not fully understood but the major mechanisms are believed to be by blocking surface sites for collector adsorption and by co-adsorption resulting in a hydrophilic surface. Additionally, a number of organic polymeric binders for agglomeration of iron ore have been developed in the last decades (see references [3-14] in [6]). The main advantage of organic binders compared to traditional inorganic binders, like bentonite, is that the former are completely eliminated 1

14 during the heat treatment of iron ore pellets and thus do not introduce any contaminants to the final product. Therefore, studies on interactions of different types of polymers with iron oxides, especially in combination with anionic surfactants, could further extend the possibilities of using polymers as depressants and binders in iron ore beneficiation and agglomeration. Studies on the interactions between anionic carboxylate surfactants and iron oxides are rather sparse. Even less common are investigations involving co-adsorption of anionic surfactants and polymers. The existing studies primarily involve ex-situ methods, batch adsorption and flotation experiments. Application of in-situ techniques like attenuated total reflectance Fourier transform infrared spectroscopy (ATR-FTIR) could provide insight into the mechanism of interaction between surfactants, polymers and iron oxides at the solid-liquid interface. Further, with this technique, the adsorption and desorption kinetics may be monitored in-situ, as may any possible changes in the adsorption mode at different experimental conditions. ATR elements coated with thin films of synthetic analogues of natural mineral particles are commonly applied in the adsorption studies by ATR-FTIR spectroscopy [7] to achieve a high signal-to-noise ratio and to simplify interpretation of the spectroscopic results. However, a possible difference in interfacial properties of synthetic and natural materials is an important issue to consider and might require verification of the results, obtained using synthetic particles, for their natural analogue. 2

15 SCOPE OF THE PRESENT WORK The scope of the present work may be divided into two main parts: Studying the adsorption of surfactants on iron oxides and different factors that can affect the adsorption; Investigating the possibilities of restoring surface wettability after surfactant adsorption. To achieve the first goal, a method based on ATR-FTIR spectroscopy for in-situ studies of the adsorption of organic and inorganic species from aqueous solutions on thin films of synthetic iron oxides was developed. Adsorption and desorption of different surfactants on iron oxides were investigated in order to elucidate the mechanism of interaction and to study the stability of the surface complexes formed. The effect of ph, surfactant concentration, ionic strength, presence of calcium ions and sodium silicate on the adsorption of surfactants on iron oxides was also studied. In the next step, the change of the charge and wettability of the iron oxide surface upon adsorption of calcium ions, sodium silicate, and an anionic carboxylate surfactant was investigated. Based on the information collected in the first part of the work, several means to reduce the effect of surfactant adsorption on the wettability of the iron oxide surface were evaluated, including treatment with sodium silicate and hydrophilic polymers. Finally, the results obtained using synthetic iron oxide particles were verified for mineral magnetite concentrate. 3

16 4

17 BACKGROUND Upgrading of iron ore Being the most widely used metal in the world, iron is found in nature mainly in the form of oxide and sulphide ores. Due to their wide occurrence in nature and their high iron content [2], the most industrially important ores are: hematite ( -Fe 2 O 3 ), magnetite (Fe 3 O 4 ) and goethite ( -FeOOH). After extraction from the deposit, the iron ore is subjected to grinding and enrichment to produce an iron ore concentrate with a required chemical composition and particle size distribution. The main purpose of the ore enrichment is to separate the valuable ironcontaining mineral from the waste minerals (gangue) and to reduce the amount of certain elements (like silicon, phosphorus, aluminium and sulphur) in the concentrate to an acceptable level for the iron production. Concentration of the iron ore can be achieved by gravity separation and/or magnetic separation, sometimes followed by froth flotation in order to further reduce the silica and phosphorous content of the ore [2]. In order to make iron ore concentrates suitable for the blast furnace, fine iron ore particles have to be agglomerated. Two commercial agglomeration processes exist today: sintering and pelletizing. Pelletizing is a more energy-efficient process than sintering and requires less than half the amount of fuel [8]. The pelletizing process starts with balling of wet, so-called green pellets from the iron ore concentrate. This is done in balling drums using bentonite as a binder. Different additives can be introduced to the pellet feed to produce pellets with required properties. Balling of the green pellets is followed by screening where the desired size fraction is separated from the under-size fraction, which is returned to the balling drum, and the over-size fraction, which is first crushed and then returned to the balling drum. Finally, the green iron ore pellets are dried, oxidized and sintered to give the final product. Here, depending on the type of iron-containing mineral, fuel consumption can vary significantly. When magnetite ore is used, a highly exothermic oxidation reaction takes place. In this reaction, magnetite is converted to hematite, accompanied by a heat release that 5

18 accounts for more than two thirds of the total energy required for the subsequent sintering of pellets [9]. The quality of the final pellets produced is highly dependent on the green pellets strength and the pellet size distribution. Breakage of the green pellets results in creation of crumbs and fines that increase the packing density of the pellet bed during drying, oxidation and sintering, thus reducing the pellet bed permeability to air, which is undesirable since it negatively affects both the production capacity and pellet quality [3]. Of all the process steps, froth flotation has probably the largest impact on the surface properties of iron ore concentrate, which also affects pelletization. Froth flotation Froth flotation is based on the difference in surface properties of minerals, namely, their affinity to air and water. Separation of two minerals by flotation can occur if the surface of one of the minerals is hydrophobic and the surface of the other mineral is hydrophilic. Upon introduction of air to the flotation cell, hydrophobic particles will be floated by the air bubbles attached to the particle surface while hydrophilic particles will remain in the water. Fig. 1 illustrates the principle of froth flotation. Figure 1. The principle of froth flotation. 6

19 Two types of flotation processes can be distinguished based on the floated fraction: direct flotation refers to the process in which the valuable mineral is transferred to the floated fraction, leaving the gangue in the slurry while in reverse flotation, the gangue is floated and the valuable mineral remains in the slurry [10]. Most minerals are not hydrophobic by nature so their surface has to be modified by a flotation collector, which selectively adsorbs on the surface of a mineral to be floated, rendering it hydrophobic and thus easily attached to the hydrophobic air bubbles. Flotation collectors are heteropolar organic molecules containing both a non-polar hydrophobic hydrocarbon group and a polar head group. Depending on the properties of the head group, flotation collectors can be classified as ionic or non-ionic. Ionic collectors become ionized upon dissolution in water and are further subdivided into anionic (e.g. carboxylates, sulphonates, xanthates [11]), cationic (e.g. amines, quaternary ammonium salts, pyridinium salts [12]) and amphoteric or zwitterionic (e.g. amino acids, glycines, quaternary ammonium sulphonates [12]) based on the charge of the head group after dissociation. Non-ionic collectors contain a head group that does not dissociate in water, e.g. a polyoxyethylene glycol group [13]. Among all the collectors, anionic collectors are most widely used in mineral flotation [10]. For instance, fatty acids and petroleum sulphonates are applied in non-sulphide mineral flotation while xanthates are commonly used for most sulphide ores [14]. In order to prevent the air bubbles holding mineral particles from bursting when they reach the air-water interface, a frother is added to the flotation cell to facilitate the formation of a stable froth, which is further transferred from the flotation cell surface to the collecting launder. Additionally, different modifiers are typically used in order to increase flotation selectivity [10]. Modifiers can either enhance or reduce the effect of a collector on a certain mineral and are therefore referred to as activators and depressants. Activators are usually soluble salts that become ionized in solution and interact with the mineral surface altering its chemical nature and making it more favourable for collector adsorption [10]. The action of depressants is more complex and not always fully understood. However, one of the main mechanisms is blocking of the surface sites by adsorbing the depressant to prevent collector adsorption [14]. Dispersants may also be added to the flotation system to facilitate liberation of different small-size mineral fractions (slime) from the surface of larger ore particles, thus facilitating increased floatability of the larger particles, which in turn improves the recovery. Finally, ph regulators are added to control the ph one of the key variables in the flotation process that affects the surface 7

20 properties of the minerals and the speciation of both the flotation chemicals and naturally occurring inorganic ions (e.g. carbonates, sulphates) in the process water. Apart from modifiers added deliberately, different inorganic ions naturally present in the process water may affect the flotation performance [15] by activating or depressing the flotation of a certain mineral, changing collector solubility and zeta-potential of the mineral surface. For instance, pyrite can be activated in the presence of copper ions [16] and activation of magnetite for flotation with fatty acids can occur in the presence of calcium ions [17]. Flotation of iron oxides The choice of flotation process in iron ore beneficiation depends on the nature of the gangue associated with the iron-containing mineral, which can be siliceous or acidic (rich in silica) and calcareous or basic (rich in calcium oxide) [10]. When iron oxide is to be separated from the siliceous gangue, either direct or reverse flotation can be applied. Anionic collectors such as fatty acids and alkyl sulphates and sulphonates [18] are most commonly used for flotation of iron oxides from siliceous gangue minerals at ph values where the surface of iron oxide is positively charged. An example of a process utilizing fatty acid flotation for the concentration of hematite is the Republic mine process in the state of Michigan in the USA [18]. Quartz and silicate minerals can be floated from iron oxides with cationic collectors, primarily amines, when their surface is negatively charged. In order to increase flotation selectivity, iron oxides can be successfully depressed by starch or dextrin [18]. For instance, the Empire and Tilden mines (Michigan, USA) and the Griffith Mine (Ontario, Canada) have been using ether amines to float the siliceous gangue from the iron oxides [18]. This type of flotation is also utilized in Brazil, Chile, India, Mexico, Russia, and South Africa [19]. Calcareous phosphate gangue minerals can be floated from iron oxides with modified fatty acids. Selectivity is improved when sodium silicate or starch is used as a depressant [2]. The Swedish mining company LKAB has been using reverse froth flotation with an anionic fatty acid based collector for dephosphorization of magnetite concentrate. In order to improve flotation selectivity and phosphorous recovery, sodium silicate is added as a dispersant/depressant. A distinctive feature of the flotation of calcareous ores is the presence of calcium ions in the process water [20]. Calcium is known to facilitate precipitation of fatty acids [21] which may result in unnecessary increases in fatty acid collector consumption. 8

21 Moreover, high concentrations of calcium ions have been shown to activate magnetite for flotation with a fatty acid collector by adsorbing on the magnetite surface and changing its charge [17]. Flotation effect on wet agglomeration of iron ore Wet agglomeration implies that fine particles in agglomerates are held together by a liquid, which acts as a binder. The amount of liquid in the structure of agglomerates determines agglomerate strength and is characterized by the liquid saturation (S) (Eq. 1): 100 F 1 P S 100 F L, (1) where F liquid content in the agglomerate; fractional porosity; P density of particles; L density of liquid. Wet agglomerates can be in different liquid saturation states (see Fig. 2). Figure 2. States of liquid saturation in wet agglomerates [22-24]. According to the capillary theory [25] developed for wet agglomerates with a freely movable binder (like water), the tensile strength of agglomerates increases with the increase in liquid saturation due to the development of the capillary forces. The tensile strength reaches maximum in the capillary state (liquid saturation 80-90%) when all the pores inside the agglomerate are filled with liquid and concave menisci are formed at the pore openings (see Fig. 2). Complete wetting of the surface is required for full development of the capillary forces. The relation between the tensile strength ( c) of wet agglomerates in the capillary state and surface wettability is described by the Rumpf equation [25]: 1 1 c a cos LS, (2) d 9

22 where a constant; fractional porosity; liquid surface tension; d average particle size; LS liquid-solid contact angle. Strictly speaking, the Rumpf equation is only valid for agglomerates produced using a freely movable binder and is not applicable to iron ore pellets balled with water and bentonite clay [26, 27]. However, as illustrated below, similar trends as those described by the Rumpf equation are observed for the wet strength of iron ore pellets. Flotation of the iron ore prior to agglomeration may affect several parameters in the Rumpf equation. The presence of a flotation collector in the water reduces the surface tension of the water, which has been shown to decrease the wet strength of iron ore pellets [3, 22, 28]. Adsorption of flotation collector on the surface of the concentrate makes the surface more hydrophobic and could be expected to further reduce pellet wet strength. Although no experimental studies investigating the dependency of the agglomerate strength on the contact angle of the feed have been found, an adverse effect of flotation collector adsorption on the wet strength of iron ore pellets is commonly reported [3-5]. Additionally, collector adsorption increases the affinity of the concentrate surface for air, which results in attachment of air bubbles to the surface of the concentrate, followed by incorporation of bubbles inside green pellets, increasing pellet porosity and decreasing the liquid saturation [3]. However, the authors conclude that the decrease in pellet wet strength upon adding flotation collector was not due to the decreased liquid saturation, but was due to the fact that air bubbles inside the green pellets behaved like large plastic particles, increasing plastic deformations in pellets and weakening the pellet structure. To reduce the disturbances in balling circuits, variation in the properties of the pellet feed should be minimized. Together with moisture content and fineness, wettability of the iron ore concentrate should be monitored, so that necessary process adjustments could be made in both flotation and pelletization. Surface wettability and contact angle measurements Wetting of a solid surface occurs due to adhesion forces between the surface and the wetting liquid, which act against the cohesive forces within the liquid and make the liquid spread over the surface at a certain contact angle (see Fig. 3). 10

23 Figure 3. Schematic illustration of the contact angle at the solid-liquid-gas contact line. The solid-liquid contact angle ( LS) of a liquid drop on a polished, flat, solid surface is defined by the Young equation [29]: SG cos, (3) LS LG LS where SG, LS, and LG are the solid-gas, solid-liquid and liquid-gas interface tension, respectively. For each pair of liquid and solid characterized by certain SG and LG, the solid-liquid contact angle is determined by the liquid-solid interfacial free energy ( LS). According to the van Oss theory [30], the liquid-solid interfacial free energy can be divided further into the apolar Lifshitz-van der Waals part ( LW ) and the polar part, with the latter comprising Lewis acid ( + ) and Lewis base ( - ) components: LW LW 2. (4) LS SG LG SG LG SG LG SG Iron oxides have a large amount of acid and base sites [31], contributing to the polar component of the surface free energy, and are consequently hydrophilic. For instance, a contact angle of 25 ± 5 was reported [32] for water on the polished surface of natural magnetite, measured using a static sessile drop method. However, it is not always possible to obtain a completely smooth surface for contact angle measurements, which leads to the problem of high variation in the results reported for the same iron oxide. Additionally, chemical heterogeneity, introduced by impurities present in the natural mineral samples, for example, may have a significant effect on the measured contact angle. Iveson et al. have shown that the contact angle of the mixed hematite-goethite ores varied from 0 to 74 depending on the relative content of these two minerals [33]. Depending on the particle size and morphology, different techniques are used for contact angle measurements [34, 35]. Optical tensiometry methods (e.g. a static sessile drop method) are based on capturing and analyzing images of a liquid drop placed on a surface and are suitable mainly for measuring the contact angle on flat surfaces. However, the static sessile drop 11 LG

24 method can also be used to assess the wettability of colloid particles, providing that a closely packed layer of particles can be formed [36]. The contact angle of natural mineral powders is commonly estimated by the Washburn method, which is an example of force tensiometry methods and is based on measuring the sorption of a wetting liquid by a powder material upon immersion. In this method, the packing of particles is also important since it may affect the penetration rate of the wetting liquid and thus the measured value of the contact angle [37]. Although it might be a challenge to obtain a true value of contact angle for non-ideal systems such as porous films and mineral powders, contact angle measurements can be successfully used to characterize the changes in the wettability of these materials upon adsorption of reagents related to flotation and pelletization. Adsorption at the solid/liquid interface Adsorption is a process of accumulation of adsorbate species from a bulk gas or liquid on the surface of an adsorbent. In the case of interactions of surfactants and polymers with mineral surfaces, it is the adsorption at the liquid/solid interface that is of interest. When a solid surface is placed in contact with a polar liquid (like water), the surface may acquire a net surface charge due to ionization of the surface groups, adsorption of ions from solution or dissolution of ions comprising the surface [38]. Consequently, an electrical double layer may be formed, due to the concentration of oppositely charged counter-ions at the charged surface to maintain charge-neutrality (see Fig. 4). Figure 4. Schematic figure of the electrical double layer at a liquid/solid interface. 12

25 In Fig. 4, represents the so-called Stern layer where the counter-ions have the highest concentration and are held close to the surface. Beyond the Stern layer, the concentration of counter-ions decreases until it reaches the bulk concentration. The charge of the surface in the slip plane just outside the Stern layer is referred to as zeta-potential and can be estimated from electrokinetic measurements. Considering an iron oxide surface in contact with water, a fully hydroxylated surface should be expected. The net charge of the iron oxide surface is dependent on protonation/deprotonation of the hydroxyl groups when the ph of the solution changes (see Eq. 5). FeOH H H 2 FeOH FeO (5) The ph at which the net charge of the surface is zero is referred to as the point of zero charge (PZC). For iron oxides, the PZC is usually observed at ph 7-8 [39]. Above this ph, the surface is charged negatively, whereas below the PZC the surface bears a positive charge. The PZC of a surface can be determined by a potentiometric titration in an indifferent electrolyte. When the charge of a surface is measured by electrophoresis, the ph at which the zetapotential is equal to zero is termed the isoelectric point (IEP). In the absence of specific adsorption of non-potential-determining ions, the values of the PZC and the IEP should be the same. The zeta-potential plays an important role in the adsorption of ionic species at mineral-water interfaces. The change in the zeta-potential upon adsorption can be used as an indication of the type of forces involved in adsorption [40]. If adsorption takes place only through electrostatic interaction, the absolute value of the zeta-potential will decrease upon adsorption and will eventually reach zero when the surface is fully saturated with the adsorbate. However, if apart from electrostatic interaction, specific adsorption occurs due to affinity of certain species to the surface, the zeta-potential of the surface can go through zero and then become reversed. Another indication of specific adsorption is the shift of the PZC and the IEP of a surface in the opposite directions upon adsorption. Depending on the forces contributing to adsorption, specific adsorption can be either physical or chemical. Chemical adsorption refers to when an adsorbate forms a covalent bond with the surface of the adsorbent while physical adsorption implies contribution of weaker forces such as hydrogen bonding and van der Waals interactions. 13

26 Quantitatively, adsorption of a certain compound on a solid surface is described by an adsorption isotherm. It is obtained by plotting the measured amount of the adsorbate on the surface against the equilibrium concentration of adsorbate in solution. Different adsorption models have been developed to describe experimental adsorption data; the most common models used for describing adsorption at the solid-liquid interface are the Langmuir and the Freundlich models [38]. The Langmuir adsorption isotherm (Eq. 6) is based on the assumption of localised monolayer adsorption and that the heat of adsorption is independent of surface coverage. x x max ac 1 ac (6) In this equation, x x max is the fraction of the surface covered with the adsorbate; C is the equilibrium concentration of the adsorbate in solution and a is the adsorption constant. The Freundlich adsorption isotherm (Eq. 7) can be derived from the Langmuir isotherm by introducing an exponential change to the heat of adsorption with surface coverage. Thus, this model implies adsorption on an energetically heterogeneous surface. The different adsorption sites may be grouped patchwise, with sites having the same heat of adsorption grouped together. x 1 / n kc (7) m In this equation, x is the amount of the adsorbate adsorbed on a specific mass m of the adsorbent; k and n are empirical constants. Both the Langmuir and the Freundlich isotherms are applicable to the adsorption of surfactants on mineral surfaces. However, due to specific properties of surfactant molecules (e.g. their ability to form micelles or adsorbed multi layers) the adsorption of these molecules can be characterized by other types of isotherms. For instance, adsorption of ionic surfactants on oppositely charged surfaces is frequently described by an S-shaped isotherm when plotted using a logarithmic scale and referred to as a Somasundaran-Fuerstenau isotherm [41]. This isotherm has four characteristic regions as illustrated in Fig. 5 [42]. Region I represents adsorption at low surfactant concentrations due to electrostatic forces between the surfactant species and oppositely charged surface sites. In Region II, surfactant 14

27 species on the surfaces begin to form two-dimensional surface aggregates due to hydrophobic interactions between the hydrocarbon chains in surfactant molecules. Since the electrostatic interactions are still active in this region, the adsorption density shows a sharp increase. In Region III, the surface charge is fully neutralized by the adsorbed surfactant species and electrostatic forces do not contribute to adsorption any longer. However, interaction between hydrophobic chains in surfactant species still occurs, further increasing adsorption density, though at a lower rate. In region IV, the surfactant concentration in solution reaches the critical micelle concentration (CMC) and any increase in concentration contributes primarily to formation of micelles in solution without changing the adsorption density on the surface much. Figure 5. Somasundaran-Fuerstenau isotherm [42]. Polymer adsorption on solid surfaces is commonly characterized by a so-called high-affinity adsorption isotherm, exhibiting a sharp increase in surface loading at a very low polymer concentration, which is followed by a plateau at higher concentrations [43]. This type of adsorption isotherm has been reported for polyelectrolytes adsorbed on the surfaces of the same [44] and opposite charge [45] as well as for the adsorption of non-charged polymers [46]. A specific feature of polymer adsorption is that it can hardly be reversed by dilution [43] due to the fact that a polymer molecule is bound to the surface through a number of segments, which have to be detached from the surface in order to desorb the polymer molecule. The ability of a polymer molecule to adopt a large number of configurations both in solution and at the solid/liquid interface makes polymer adsorption rather complex as compared to adsorption of small molecules and ions. Numerous theoretical models describing polymer adsorption have been proposed and can be found elsewhere [43, 47]. 15

28 Attenuated total reflectance Fourier transform infrared spectroscopy (ATR-FTIR) Fourier transform infrared (FTIR) spectroscopy is based on the ability of molecules to undergo transitions from one vibrational energy state to another by absorbing infrared radiation [48]. In order for absorption to occur, the transition must involve a change in the dipole moment of a vibrational mode. Each molecule can only absorb radiation of certain frequencies, that is, the natural vibrational frequencies of the molecule, resulting in a number of absorption bands located at different frequencies in the spectrum. In infrared spectroscopy, the frequency is traditionally expressed in wavenumbers (cm -1 ). The most commonly used types of vibrational modes are stretching and bending. A stretching vibration is characterized by a change in the length of a bond between atoms while a bending vibration involves the change in the angle between bonds. The amount of infrared radiation absorbed (A, Absorbance) by a certain species is described by the Lambert-Beer law (Eq. 8). I 0 A log l C (8) I In this equation, I 0 is the initial intensity of the radiation and I is the intensity of the radiation after interaction with the sample; is the molar absorptivity of the species at a certain wavelength; l is the path length of the radiation in the sample; C is the concentration of the species of interest in the sample. Thereby, the Lambert-Beer law illustrates that the intensity of a specific absorption band is proportional to the amount of the corresponding group in the sample and can thus be used for quantitative studies. FTIR spectroscopy enables both ex-situ and in-situ studies. Attenuated total reflectance FTIR spectroscopy (ATR-FTIR) is a technique well suited to in-situ studies, providing an opportunity to study the interactions at the solid-liquid interface without changing the surface characteristics of the sample [49]. The ATR technique is based on the phenomenon of attenuated total reflectance, which is schematically illustrated in Fig

29 Figure 6. Schematic figure of an ATR waveguide illustrating the ATR phenomenon. In this technique, the IR beam with the initial intensity I 0 passes through a waveguide, having a high refractive index n 2 and being surrounded by a medium with lower refractive index n 1. The difference in the refractive indices results in attenuated total reflection of the beam inside the waveguide, provided that the incident angle fulfils the relation shown in Eq. 9. sin n n 1 (9) 2 At each point of reflection, an evanescent wave of the IR radiation is formed perpendicular to the waveguide. The wave can interact with the surrounding medium in the vicinity of the waveguide resulting in attenuation in the intensity of the totally reflected beam. The amount of radiation of a certain wavelength ( ) absorbed by the surrounding medium depends on the penetration depth d p (see Fig. 6), which is defined by Eq. 10. d p 2 n 2 sin 2 n n / 2 (10) The penetration depth is, by definition, the distance from the interface where the intensity of the electric field (E) of the wave has declined to a value equal to: E = E 0 e -1 (11) In this equation, E 0 is the intensity of the electric field at the surface of the waveguide. The values of the penetration depth typically vary in the range from some hundred nanometres to a few micrometres, making the ATR-FTIR spectroscopy a surface-sensitive 17

30 technique. Furthermore, the short penetration depth significantly reduces the absorption of IR radiation by water, facilitating studies of aqueous systems. In-situ spectroscopic measurements open up possibilities for following the adsorption process in real time and to obtain information about adsorption and desorption kinetics and equilibria, surface complexes formed at the solid/liquid interface and the orientation of adsorbed species. ATR-FTIR spectroscopy has been extensively used to study the adsorption of surfactants and polymers on mineral surfaces [7, 49-51]. The fact that adsorption can be performed either directly on a bare waveguide or on a waveguide coated by a thin layer of adsorbent makes the technique applicable to a wide variety of systems [52]. 18

31 EXPERIMENTAL PART Materials Three different iron oxide materials were used in the experimental work (see Table 1). Table 1. Iron oxide materials used in the experimental work. Iron oxide Synthesis method/origin Paper(s) Synthetic hematite Matijevic [53] I Synthetic magnetite Massart and Cabuil [54] II-IV Mineral magnetite * LKAB, Kiruna, Sweden V * Cleaned by magnetic separation and flotation, stored at the ambient conditions for two years. The iron oxides were characterized using X-ray diffraction (XRD), scanning electron microscopy (SEM), electrophoresis, gas adsorption, and contact angle measurements. Adsorption of a commercial flotation collector, Atrac 1563 from Akzo Nobel, and four model compounds was investigated in this work (see Fig. 7). Figure 7. Chemical structures of Atrac 1563 (a), ethyl oleate (b), maleic acid (c), poly (ethylene glycol) monooleate (PEGMO) (d), and dodecyloxyethoxyethoxyethoxyethyl maleate 19

32 (e). R represents a linear alkyl chain in fatty acids or a C 19 H 29 chain in resin acids, R CH 3 (CH 2 ) 7 CH=CH(CH 2 ) 7, R CH 3 (CH 2 ) 11. Commercial flotation collector Atrac 1563 has a complex chemical composition: % ethoxylated tall oil ester of maleic acid, and 1-5 % maleic anhydride (Akzo Nobel material safety data sheet). Since the exact composition and chemical structure of Atrac 1563 are unknown, four different reagents, as shown in Fig. 7b-e, were evaluated as model compounds to be used in the experimental work instead of Atrac Two types of soluble silicate were used in the experiments. Water glass, i.e. an aqueous solution of sodium silicate, in this case with a SiO 2 :Na 2 O weight ratio of 3.25, is used as a dispersant/depressant in the flotation of iron ore. Sodium metasilicate (Na 2 SiO 3 9H 2 O) was used as an analytical grade alternative of water glass. In the experiments described in Papers IV and V, adsorption of four different polymers was investigated (see Table 2). Table 2. Polymers used for surface modification of magnetite. Average Polymer name Structural formula molecular weight Supplier Dispex A40 (ammonium polyacrylate) Dispex N40 (sodium polyacrylate) 4000 BASF 4000 BASF ATC 4150 (aliphatic quaternary polyamine) Eka chemicals Soluble starch * N/A Merck * 1 wt % aqueous starch solution containing 0.5 wt % NaOH was heated to 84 C for 10 minutes and then cooled to room temperature [55]. 20

33 Methods The main instrumental techniques used in the present work were ATR-FTIR spectroscopy, contact angle and zeta-potential measurements. Film preparation For the spectroscopic and contact angle measurements, the appropriate substrate was coated with a film of synthetic iron oxide. For the experiments described in Paper I, both sides of the waveguide were coated with a hematite film by means of dip-coating. In the experiments described in Papers II-V where synthetic magnetite was used, only one side of the waveguide was coated with a film by spreading a certain amount of magnetite dispersion and air-drying it at room temperature. The reason for this was to prevent the magnetite film from absorbing too much IR radiation in the spectroscopic measurements. ATR-FTIR spectroscopy Spectral data were collected using a Bruker IFS 66v/S spectrometer equipped with a liquid nitrogen cooled mercury-cadmium-telluride (MCT) detector and a deuterated triglycine sulphate (DTGS) detector, a vertical ATR accessory and a stainless steel sample cell (see Fig. 8). Trapezoidal ZnSe crystals (Crystran Ltd) with 45 cut edges and dimensions of 50x20x2 mm were used as ATR waveguides. Figure 8. Schematic illustration of the experimental setup. Thick solid lines represent liquid flow whereas the dashed arrows indicate the IR beam. Adsorption measurements were performed in-situ at room temperature with a continuous flow of working solution pumped through the cell with recirculation, except for the desorption experiments in which the solution was not recirculated. The ph during the adsorption experiments was kept constant by a Mettler Toledo T70 titrator. 21

34 Contact angle The static sessile drop method was used to determine the contact angle of the synthetic magnetite nanoparticles (Papers III-V). Contact angle measurements were performed using a Fibro 1121/1122 DAT-Dynamic Absorption and Contact Angle Tester equipped with a CCD camera. The measurement was performed by placing a 4 L water droplet onto the magnetitecoated substrate using a microsyringe. A series of images were taken and analysed using the DAT 3.6 software. To investigate the effect of different reagents on the wettability of the synthetic magnetite particles, consecutive adsorption of the reagents was performed on the magnetite film in the same way as in the spectroscopic measurements. Between the adsorption steps, the contact angle of the magnetite film was measured. The contact angle of the natural magnetite particles (Paper V) was determined by the Washburn method using a Krüss K100 force tensiometer. Liquid sorption by the magnetite powder was recorded as a function of immersion time, and Krüss LabDesk 3.1 software was used to calculate the contact angle applying the Washburn equation. First, the capillary constant of the Washburn equation was estimated for each sample using n-hexane. Thereafter, the contact angle of the magnetite powder was measured using deionized water. The values of the capillary constant and the contact angle were calculated as an average of three replicates. To investigate the effect of different reagents on the wettability of the natural magnetite particles (Paper V), batch adsorption was performed using suspensions containing 10 g magnetite per ca 40 ml solution at ph 9 and room temperature. After adsorption, the solution was decanted and magnetite was dried in an oven overnight at 50 C. Zeta-potential The zeta-potential of both synthetic and natural iron oxides as a function of ph was determined by electrophoresis using a ZetaCompact instrument equipped with a chargecoupled device (CCD) tracking camera. The electrophoretic mobility data was further processed by the Zeta4 software applying the Smoluchowski equation. For the case of magnetite concentrate, the measurements were performed using the m fraction of the magnetite slurry collected at the LKAB concentrating plant in Kiruna, Sweden, after flotation. The required size fraction of the particles was separated by vacuum filtration. Further experimental details are available in the appended papers. 22

35 RESULTS AND DISCUSSION Characterization of iron oxides (Papers I, II, V) The X-ray diffraction data of the iron oxides used in the present study (Fig. 9) confirmed pure crystalline phases of hematite (a) and magnetite (b, c), without any other phases present in amounts detectable by XRD. The peak width decreases in the sequence synthetic magnetite > synthetic hematite > natural magnetite, reflecting the increasing particle size of the iron oxide materials (10 nm and 130 nm, as determined by SEM, see below, and < 45 m [56], respectively). Figure 9. XRD patterns of a synthetic hematite (a), a synthetic magnetite (b), and a natural magnetite (c). The reflections originating from the corresponding iron oxides are indexed with the appropriate Miller indices. 23

36 SEM images in Fig. 10 show examples of cross-sections of the films of synthetic hematite (a) and magnetite (b) on a ZnSe substrate, which were used in the spectroscopic and contact angle measurements. In both cases, porous films were formed with a thickness of ca 1 m and nm, respectively. Fig. 10 illustrates that the synthetic iron oxide crystals had a uniform spherical habit and were slightly aggregated. The particles in the magnetite concentrate in Fig. 10c exhibited high variation in both size and shape. The figure illustrates that the coarse magnetite particles were covered by very fine particles (less than 1 m in size), some of which, according to the EDS results, had a high content of silicon and aluminium and could be the remains of aluminosilicate minerals, which are present in the iron ore before concentration. Figure 10. Side view SEM images of a hematite film (a) and a magnetite film (b) on a ZnSe crystal, a top view SEM image of the mineral magnetite particles on a carbon tape (c), and a close-up SEM image of a magnetite particle shown in Fig. 10c (d). 24

37 Fig. 11 shows the zeta-potential of the iron oxides as a function of ph. The fraction of the mineral magnetite concentrate used for the zeta-potential measurements was found to be mainly comprised of other minerals that magnetite and will not be discussed here. The IEP for the synthetic magnetite (empty triangles) was observed at ph 7, as per the literature [39], whereas the IEP for the hematite particles (filled diamonds) was observed around ph 5, which is lower than expected and could be caused by the adsorption of chloride [57] or carbonate [58] ions on the surface. Figure 11. Zeta-potential as a function of ph of the synthetic hematite in 10 mm KNO 3 ( ) and of the synthetic magnetite in 10 mm NaCl ( ). Regarding the wetting properties of the iron oxides used in this work, the contact angle of the synthetic magnetite was 15-25, whereas the magnetite concentrate had a contact angle of The lower wettability of the magnetite concentrate was likely due to the hydrophobic flotation collector species that have been reported to be present on the surface of the concentrate after flotation [3]. However, the inconsistency in the obtained values could also be due to the difference in particle size as well as the measuring techniques used. Table 3 summarizes the morphological properties of the iron oxides used in this study. Table 3. Morphological properties of the iron oxide materials. Property Synthetic hematite Synthetic magnetite Mineral magnetite Particle size 130 nm 5-15 nm 85% -45 m [56] Particle shape Spherical Spherical Irregular BET (N 2 ) surface area, m 2 g [56] 25

38 Surfactant adsorption and factors affecting the adsorption (Papers I-III) In the present work, the adsorption of one commercial flotation collector (Atrac 1563) and four model compounds (PEGMO, maleic acid ester, ethyl oleate, and maleic acid) on synthetic iron oxides was investigated. However, only three of the compounds were found to show similar adsorption behaviour: Atrac 1563, PEGMO, and the maleic acid ester. The adsorption of these three compounds will be discussed below. Adsorption mechanism Fig. 12 shows the spectra of Atrac 1563, PEGMO, and the maleic acid ester, as-received and adsorbed on synthetic hematite and magnetite at ph 8.5. Figure 12. ATR-FTIR spectra of Atrac 1563 (1), PEGMO (2) and the maleic acid ester (3) as-received and spread over an uncoated ZnSe crystal (a); of Atrac 1563 (1) and PEGMO (2) adsorbed on hematite from a 10 mg L -1 solution at ph 8.5, and maleic acid ester (3) adsorbed on magnetite from a 25 mg L -1 solution containing 0.01 M NaCl at ph 8.5 (b). As illustrated in Fig. 12, similar absorption bands are observed in the spectra of the surfactants, confirming structural resemblance of the head groups in these molecules. Assignment of the main absorption bands in the spectra of the surfactants is presented in Table 4. More detailed discussions of the spectral features displayed by the surfactants used in the present study are given in Papers I, II, and V. 26

39 Adsorption of the surfactants on synthetic iron oxides was performed from aqueous solutions at ph 8.5, i.e. at an optimum ph for the flotation of apatite from iron oxide [2]. At this ph, the free carboxylic groups in Atrac 1563 and maleic acid ester become deprotonated, forming a negatively charged carboxylate ion as indicated by two new bands originating from the symmetric and asymmetric stretching vibrations of the carboxylate ion (v s (COO - ) and v as (COO - ), respectively) in the spectra of these compounds adsorbed on the iron oxides (spectra (1) and (3) in Fig. 12b). No bands associated with the carboxylate ion were found in the spectrum of PEGMO (spectrum (2) in Fig. 12b) suggesting that the ester bond in PEGMO does not break upon adsorption on hematite at the conditions studied. Table 4. Assignment of absorption bands originating from Atrac 1563, PEGMO, and maleic acid ester adsorbed on synthetic hematite and magnetite in-situ at ph 8.5. The numbers in parentheses represent the position of the corresponding absorption bands in the same compounds as-received, spread over a ZnSe substrate. Peak position, cm -1 Atrac 1563 PEGMO Maleic acid ester Peak assignment 1724 (1736) 1740 (1736) 1724 (1728) (C=O) in ester [59] (1709) - (1715) (C=O) in acid [48] (1645) - (1643) (C=C) [60] as (COO - ) [60] s (COO - ) [60] 1171 (1159) 1175 (1173) 1178 (1161) (C-O) in esters [61] (1115) 1104 (1105) (C-O-C) [62] (1070) - (C-OH) [63] At ph 8.5 the surface of the iron oxides was characterized by a negative zeta-potential (see Fig. 11) and, consequently, no considerable adsorption of anionic carboxylate surfactants on iron oxides would be expected at this ph due to electrostatic repulsion between the negatively charged carboxylate ions and the surface bearing the same charge. In the present work, no adsorption of maleic acid on hematite took place at ph 8.5, which agrees with the results reported by Hwang and Lenhart [64]. However, both in the previous studies on oleatehematite systems [65, 66] and in this work (spectra (1) and (3) in Fig. 12), the anionic carboxylate surfactants exhibited considerable adsorption on iron oxides even at ph values 27

40 above the IEP suggesting that the adsorption of anionic carboxylate surfactants on iron oxides is not exclusively determined by the electrostatic forces. Based on the results from adsorption of maleic acid, it may be concluded that the carboxylate function is not likely to be responsible for the adsorption of the carboxylate surfactants onto the iron oxides above their IEP. Similar to the ability of non-ionic surfactants (like PEGMO) to adsorb on solid surfaces via the polar head group [42], the adsorption of Atrac 1563 and maleic acid ester on iron oxides above their IEP could be determined by the presence of polar, but not charged, groups such as ester carbonyl, hydroxyl, and ethoxy-groups. The suggested mechanism of surfactant adsorption on iron oxides at ph values above the IEP is illustrated in Fig. 13. Figure 13. Proposed adsorption mechanism of Atrac 1563 (a), PEGMO (b), and maleic acid ester (c) on iron oxides from aqueous solutions at ph 8.5. R represents a linear alkyl chain in fatty acids or a C 19 H 29 chain in resin acids, R CH 3 (CH 2 ) 7 CH=CH(CH 2 ) 7, R CH 3 (CH 2 ) 11. Dashed ovals indicate the moieties interacting with the surface. 28

41 Additionally, hydrophobic interaction between the hydrocarbon chains of the surfactants could possibly contribute to the adsorption, as indicated by the shift of the CH 2 asymmetric stretching vibration band in the spectra of Atrac 1563, PEGMO, and maleic acid ester with the increase of surfactant loading on the surface (not shown) [67]. Thus, a conclusion can be made that both the hydrophobic tail and the polar head group determine the ability of a surfactant to adsorb on iron oxides. The desorption experiments (Fig. 13 in Paper I) showed that the adsorbed species of Atrac 1563 could be removed from the hematite surface only partially, even at increased ph, implying rather strong interaction between the surfactant and the iron oxide. It is important to mention here that carboxylate ions can be expected to facilitate the adsorption of the surfactants on iron oxides below the IEP when the net charge of the surface is positive. The contribution of electrostatic forces to the adsorption of surfactants containing free carboxylic groups explains their strong adsorption dependency on the surface charge of the iron oxide and consequently on ph and ionic strength [66], as will be discussed later. Factors affecting surfactant adsorption on iron oxides Surfactant adsorption on a solid surface can be affected by many factors, including surfactant concentration, ph, temperature and presence of inorganic ions. In this chapter, the effect of surfactant concentration, ph, and total concentration of ions (ionic strength) on surfactant adsorption onto iron oxides is discussed. The results of adsorption of an anionic carboxylate surfactant on magnetite in the presence of calcium ions and sodium silicate are also presented. Surfactant concentration. Due to the fact that the absorbance of infrared radiation is proportional to the concentration of the absorbing species according to the Lambert-Beer law (Equation 8), the intensity of the bands in a spectrum of a surfactant adsorbed on iron oxide can be assumed to be proportional to the amount of surfactant on the surface. This assumption is reasonable as long as all the adsorbed species have transition dipole moments of similar value. The absorbance of the C-H symmetric stretching vibration band in the spectra of PEGMO and Atrac 1563 adsorbed on hematite at ph 8.5 plotted as a function of surfactant concentration in solution (see Fig. 8 and 12 in Paper I, respectively) was in good agreement with the Freundlich adsorption model (Equation 7). This type of adsorption implies that the heat of adsorption changes depending on the surface coverage [68], as discussed above. 29

42 Ionic strength. For the adsorption of the maleic acid ester on magnetite at ph 8.5, a ten-fold increase in ionic strength (from 10-2 to 10-1 M NaCl) resulted in a 20-25% increase in the intensity of the bands originating from the surfactant adsorbed on magnetite (see Fig. 7 in Paper II), indicating the contribution of electrostatic forces to adsorption. This can be regarded as further evidence for the formation of outer-sphere surface complexes. Calcium chloride and sodium silicate. Fig. 14 illustrates the effect of calcium chloride and sodium silicate on the adsorption of maleic acid ester on magnetite. Figure 14. Intensity of the ester C=O stretching vibrations band as a function of time during in-situ adsorption of maleic acid ester on magnetite at ph 8.5 from a 25 mg L -1 aqueous solution without Ca 2+ and Na 2 SiO 3 added ( ), with 4 mm Ca 2+ ( ), 0.4 mm Na 2 SiO 3 ( ), and with 4 mm Ca 2+ and 0.4 mm Na 2 SiO 3 ( ). Background electrolyte: 10 mm NaCl. The adsorption of maleic acid ester on magnetite in the presence of calcium ions (open triangles in Fig. 14) increased dramatically compared to when no calcium ions were added (open circles in Fig. 14). This result agrees with the findings reported by Rao et al. [17] that activation of magnetite for flotation with anionic collector occurred in the presence of calcium ions. Calcium ions are also known to facilitate precipitation of fatty acids by forming calcium soaps [21], which may also adsorb on the surface of magnetite [4, 17]. Considering the effect of sodium silicate, competitive adsorption of silicate and surfactant species on magnetite was observed resulting in a three-fold decrease in surfactant adsorption (filled circles in Fig. 14) as compared to when no silicate was added (open circles in Fig. 14). However, desorption experiments (see Fig. 10 in Paper II) revealed higher stability of the 30

43 surfactant-magnetite complex as compared to the silicate-magnetite complex, suggesting that silicate species in solution are not likely to replace the surfactant molecules already adsorbed on magnetite. Similar results were recently reported by Roonasi et al. for silicate-oleate adsorption on magnetite [69]. The adsorption behaviour in the silicate-surfactant-magnetite system changed significantly with the introduction of calcium ions. Despite the fact that silicate adsorption slightly increased in the presence of calcium ions (see Fig. 5 in Paper II), almost no silicate adsorption was observed when the surfactant was added to the system, resulting in nearly as high adsorption of the surfactant (filled triangles in Fig. 14) as with calcium ions only (open triangles in Fig. 14). Thus, the depressing activity of sodium silicate on surfactant adsorption was almost completely suppressed in the presence of calcium ions. One explanation for such behaviour could be a much higher affinity of the surfactant for the calcium ions as compared to that of sodium silicate, which is not surprising since carboxylate surfactants are known to adsorb on calcium sites on apatite and other calcareous minerals [70]. ph change. Fig. 15 illustrates the adsorption of maleic acid ester on magnetite as a function of ph. Figure 15. Intensity of the ester C=O stretching vibration band originating from the maleic acid ester adsorbed on magnetite in-situ from a 25 mg L -1 solution at different ph. The ph was gradually decreased from ph 10. The surfactant was allowed to adsorb for 5 hours at each ph. The background electrolyte was 10 mm NaCl. The spectral data indicates that the amount of surfactant on the magnetite surface decreased with increasing ph, as typically observed for the adsorption of anionic surfactants on the 31

44 surfaces bearing the same charge. As the ph decreases, the surface charge first becomes less negative and then turns positive (see Fig. 11) thus making the surface more electrostatically favourable for adsorption of the negatively charged deprotonated surfactant species. An increased precipitation of the surfactant on the magnetite surface at acidic ph could further contribute to the surfactant loading on the surface. When the magnetite surface was pretreated with calcium ions and sodium silicate prior to surfactant adsorption, the adsorption of the surfactant in the ph range went through a maximum at ph 8.5, in concert with the results reported by Morgan [71] for oleate adsorption on hematite and explained by the formation of an acid-soap complex [(RCOO) 2 H] - [66, 72]. Fig. 6b in Paper III further illustrates that surfactant adsorption was denser at ph 9.5 than at ph 7.5, which opposes the trend in Fig. 15. Such behaviour could be explained by the affinity of the surfactant towards calcium ions, which are expected to be present on the magnetite surface in a larger amount at higher ph, as becomes evident from the zeta-potential results presented in Fig. 16a. An increased calcium-surfactant precipitation at higher ph would contribute to this behaviour. The effect of surfactant adsorption on the properties of the magnetite surface (Paper III) In this chapter, the effect of adsorption of an anionic carboxylate surfactant (maleic acid ester) onto synthetic magnetite in the presence of calcium ions and sodium silicate is discussed. Zeta-potential Fig. 16 shows the zeta-potential of synthetic magnetite as a function of ph in the presence of calcium chloride, sodium silicate, and maleic acid ester. Whereas calcium ions were capable of reversing the zeta-potential of magnetite at ph values above the IEP (empty squares in Fig. 16a), sodium silicate exhibited the opposite effect, making the magnetite surface more negatively charged and shifting the IEP to lower ph (empty triangles in Fig. 16a). Considering the combined effect of calcium ions and silicate species, the resulting zeta-potential of the magnetite particles was determined by the ratio of these compounds in solution (filled diamonds and empty squares in Fig. 16b). As the calcium-to-silicate ratio increased, the IEP of 32

45 the magnetite particles shifted to higher ph values and the zeta-potential above the IEP became less negative. When maleic acid ester was added to the solution containing calcium chloride and sodium silicate (filled triangles in Fig. 16b), the zeta-potential of the magnetite particles became slightly more negative as compared to that with only calcium and silicate, probably due to the adsorption of the surfactant on the magnetite surface via positively charged calcium ions. The adsorption of the surfactant in a bi-layer structure due to hydrophobic chain-chain interactions could also result in an additional negative charge introduced by the deprotonated surfactant head groups oriented towards the solution in the second adsorbed layer [73, 74]. Figure 16. Zeta-potential as a function of ph: (a) of the magnetite crystals in 10 mm NaCl ( ), 3.3 mm CaCl 2 ( ), and 1 mm Na 2 SiO 3 ( ); (b) of the magnetite crystals in 3.3 mm CaCl 2 and 0.4 mm Na 2 SiO 3 ( ), in 3.3 mm CaCl 2 and 1 mm Na 2 SiO 3 ( ), 3.3 mm CaCl 2, 0.4 mm Na 2 SiO 3, and 25 mg L -1 maleic acid ester ( ), of the maleic acid ester (no magnetite crystals) in a 15 mg L -1 aqueous solution containing 10 mm NaCl and 2.4 mm CaCl 2 ( ). The zeta-potential of the magnetite particles in the presence of surfactant, calcium, and silicate was nearly constant in the entire ph range studied, with the IEP expected to be below ph 5, suggesting specific interaction between the surfactant and magnetite. Similar results were reported by Rao et al. [73] for oleate adsorption on fluorite and were explained by the adsorption of calcium oleate precipitate, characterized by a strongly negative and nearly constant zeta-potential at ph Regarding the maleic acid ester, it is difficult to say whether calcium-surfactant complexes were formed at the surface or already in solution, followed by the adsorption of the calciumsurfactant complexes onto magnetite. The zeta-potential of the surfactant in solution 33

46 containing calcium chloride (without magnetite particles, empty triangles in Fig. 16b) showed similar dependency on ph as the zeta-potential of magnetite in solution containing calcium chloride, sodium silicate, and maleic acid ester (filled triangles in Fig. 16b). However, the values of the zeta-potential in the latter case were significantly less negative, confirming the proposed mechanism of surfactant adsorption on magnetite in the form of ternary complexes with calcium ions. Contact angle Table 5 illustrates the effect of calcium chloride, sodium silicate, and surfactants on the wettability of synthetic magnetite. Table 5. Water contact angle of the as-synthesized synthetic magnetite and magnetite after consecutive conditioning with calcium ions, sodium silicate and a surfactant. The background electrolyte was 10 mm NaCl. The values reported were measured 1 second after a drop of water was deposited on the surface and are presented as an average value ± one standard deviation. As-synthesized Treatment 4 mm CaCl mm Na 2 SiO 3 25 mg L -1 surfactant magnetite 20 ± 3 15 ± 4 10 * 43 ± 8 (Atrac 1563) Contact angle, 44 ± 3 (maleic acid 22 ± 3 19 ± 2 10 * ester) * The exact value of the contact angle could not be estimated since the contact angle after silicate adsorption was below the detection limit of the instrument (10 ). Whereas treatment with sodium silicate improved magnetite wettability as discussed in detail in Paper III, adsorption of the surfactants resulted in an increased hydrophobicity of synthetic magnetite. Nearly the same contact angle was obtained after treatment with Atrac 1563 or maleic acid ester, suggesting that these compounds had a similar effect on magnetite wettability at the conditions studied. A higher variation of the contact angle for the case of Atrac 1563 could possibly be the result of a complex chemical composition of the surfactant. The results in Table 5 further indicate that sodium silicate did not prevent the adsorption of the surfactants on magnetite in the presence of calcium ions, in agreement with the spectroscopic result discussed above. 34

47 Verification for natural magnetite (Paper V) In order to test whether the conclusion regarding surfactant adsorption on synthetic magnetite in the presence of calcium ions and sodium silicate were applicable to natural magnetite particles, adsorption of Atrac 1563 and water glass in the presence and absence of calcium ions was performed on the mineral magnetite (see Fig. 9, 10 and Table 3 for material characterization). Fig. 17 shows the results of the contact angle measurements performed after batch adsorption experiments. Figure 17. Water contact angle of the natural magnetite particles after adsorption of 1 mg g -1 water glass at ph 9 for 1 h, followed by the adsorption of Atrac 1563 for 20 minutes at the same ph. Adsorption of both compounds was performed in the presence of either 10 mm NaCl ( ) or 4 mm CaCl 2 ( ). The points at 0 mg g -1 represent the contact angle of the magnetite concentrate after adsorption of 1 mg g -1 water glass at ph 9 for 1 h. In the absence of calcium ions, the contact angle of the natural magnetite did not change upon the adsorption of Atrac 1563 indicating that no or very little adsorption took place on magnetite pretreated with water glass. However, the adsorption of Atrac 1563 in the presence of calcium ions resulted in an increased contact angle of the magnetite particles, despite the pretreatment with water glass, due to the activation of the magnetite surface for surfactant adsorption by calcium ions. These findings agree with the spectroscopic and contact angle results obtained using synthetic magnetite. 35

48 Summary and implications for flotation and agglomeration of iron ore Based on the results discussed above, the following conclusions regarding surfactant adsorption on iron oxides and its effect on the surface properties of iron oxides can be made: 1. Anionic carboxylate surfactants were capable of adsorbing on iron oxides at ph values above the IEP of iron oxides. The adsorption increased in the presence of cations, especially calcium, and could be reduced by preconditioning with sodium silicate, but only in the absence of calcium ions. Desorption of the surfactants from the surface was only partial, even at elevated ph (up to ph 10). 2. The zeta-potential and the IEP of magnetite particles in the presence of calcium chloride and sodium silicate was determined by the relative content of these compounds. Adsorption of an anionic carboxylate surfactant did not have any drastic effect on the zeta-potential of magnetite. 3. Magnetite wettability improved after treatment with calcium chloride and sodium silicate, whereas subsequent surfactant adsorption made the magnetite surface more hydrophobic. For the flotation of iron ore, these results imply that a certain amount of the flotation collector would likely adsorb on magnetite increasing the required collector dosage, especially at high concentrations of calcium ions in the process water. Calcium ions may also have an adverse effect on the dispersing performance of water glass. The adsorbed flotation collector on the magnetite surface could facilitate flotation of magnetite to a certain extent, resulting in reduced flotation selectivity. The fact that adsorbed collector species can hardly be removed from the magnetite surface by rinsing with water suggests that a certain amount of flotation collector would remain on the surface of the magnetite concentrate after flotation and would be carried over to the balling drums. This would affect the pelletizing process negatively and would reduce the strength of the pellets produced, as discussed above. Contamination of magnetite with flotation collector may be expected to increase with increased concentration of calcium ions in the process water. Consequently, the second part of the present work was focused on finding the means to minimize the effect of adsorbed flotation collector on magnetite wettability. 36

49 Restoring magnetite wettability after surfactant adsorption (Papers IV, V) Based on the discussion above, the best way to reduce the adsorption of flotation collector on magnetite would probably be by decreasing the concentration of free calcium (and magnesium) ions in the process water. Application of different chelating agents [17, 75-78] and ion exchangers [79] has been proposed for that purpose. However, the removal of calcium ions from the process water would result in further dissolution of sparingly soluble calcareous minerals, again increasing the calcium concentration. Alternatively, modification of the magnetite surface after flotation could be performed in order to increase surface wettability. In the present work, two types of hydrophilizing agents were investigated, namely, hydrophilic polymers and sodium silicate. Their effect on the wettability of synthetic magnetite after surfactant adsorption is presented in this chapter. Modification with sodium silicate Since sodium silicate is known to have a depressing effect on iron oxides in flotation with anionic carboxylate collectors, the ability of sodium silicate to improve magnetite wettability after surfactant adsorption was investigated (see Table 6). Table 6. Water contact angle of synthetic magnetite after consecutive adsorption of calcium chloride, sodium silicate, and a surfactant, followed by treatment with sodium silicate in the presence of calcium chloride for 24 hours. Concentrations of the reagents were the same as in the experiments described in Table 5. Adsorption was performed at ph 8.5. Treatment CaCl 2, Na 2 SiO 3, and surfactant CaCl 2 and Na 2 SiO 3 42 ± 2 (maleic acid ester) 21 ± 1 Contact angle, 49 ± 3 (Atrac 1563) 16 ± 1 A considerable decrease in the magnetite contact angle was achieved after 24 hours of conditioning with sodium silicate in the presence of calcium ions. The observed effect could be caused by desorption of surfactant due to the difference in concentration at the surface and in solution or by substitution of the surfactant species for silicate. However, the latter would contradict the results reported by Roonasi et al. [69] for competitive adsorption of sodium 37

50 oleate and sodium silicate on magnetite, stating that silicate in solution could not easily replace oleate adsorbed on the magnetite surface. Modification with hydrophilic polymers Three types of polymers, viz. cationic, anionic, and non-ionic (Table 2), were investigated regarding their ability to adsorb on surfactant-coated magnetite and improve the wettability of the magnetite surface. Although all the polymers tested adsorbed on magnetite (Fig. 2 in Paper IV) independent of their charge and functionality, only anionic ammonium polyacrylate could increase magnetite wettability after surfactant adsorption (see Table 7). Table 7. Water contact angle of synthetic magnetite after consecutive adsorption of calcium chloride, sodium silicate, and a surfactant, followed by treatment with a polymer, and storage in air for 24 h. Concentrations of the reagents were the same as in the experiments described in Table 5. Polymer concentration was 12.5 mg L -1. Adsorption was performed at ph 8.5. Treatment Contact angle, CaCl 2, Na 2 SiO 3, Maleic acid ester Atrac 1563 and surfactant 46 ± 5 43 ± 8 Cationic aliphatic Polymer Starch Anionic ammonium polyacrylate polyamine 68 ± 2 40 ± 4 24 ± 6 20 ± 6 24 h in air 49 ± ± 3 20 ± 4 24 ± 8 Spectroscopic results (Fig. 2 in Paper IV) did not provide any evidence for the detachment of the surfactant from the magnetite surface upon polyacrylate adsorption, since no negative absorption bands originating from the surfactant were present in the spectra of polyacrylate adsorbed on magnetite. Accordingly, the decrease in the contact angle of synthetic magnetite upon polyacrylate adsorption was most likely due to shielding of the hydrophobic surfactant moieties from the water phase by long, flexible polymer chains able to form loops on the surface, especially in the presence of calcium ions [45], as illustrated in Fig. 18. The high density of the carboxylic groups in the polyacrylate chain makes it highly hydrophilic, resulting in an improved wettability of the magnetite surface. A similar phenomenon was reported by Somasundaran and Cleverdon [80] for amine/cationic PAM adsorption on quartz. The authors concluded that the polymer interacted 38

51 with the surface without affecting surfactant adsorption and that depression of quartz was achieved due to masking of the surfactant by the polymer. Figure 18. Schematic illustration of polyacrylate adsorption on magnetite pretreated with surfactant. For the sake of simplicity, iron oxide surface sites and other adsorbed species are not shown. Considering the mechanism of interaction of polyacrylate with the magnetite surface, polymer adsorption likely took place via calcium ions [81], similar to the adsorption of carboxylate surfactants. Calcium ions have been shown to facilitate polyacrylate adsorption on oxides [45] due to reduced electrostatic repulsion both between the carboxylate ions in the polymer chain and the negatively charged oxide surface, and between the carboxylate ions within the molecule. Such intramolecular bridging results in a more coiled conformation of polyacrylate chains both in solution and on the surface, increasing packing efficiency of the polymer species on magnetite. The conclusion regarding the mode of adsorption of polyacrylate on magnetite was further confirmed by the zeta-potential measurements (Fig. 5 in Paper IV), which showed that the zeta-potential of calcium-polyacrylate complex in solution was nearly the same as the zetapotential of the magnetite particles treated with calcium chloride, sodium silicate, anionic surfactant, and polyacrylate, suggesting that polyacrylate was adsorbed on magnetite as a ternary complex with calcium ions. 39

52 Verification for the flotated magnetite concentrate (Paper V) To investigate whether soluble silicate and polyacrylate were efficient in improving the wettability of magnetite concentrate after flotation, the contact angle of the concentrate was measured by the Washburn method before and after adsorption of water glass and sodium polyacrylate. Fig. 19 illustrates the effect of treatment with water glass on the wettability of the magnetite concentrate. Figure 19. Water contact angle of the magnetite concentrate upon modification of the surface with water glass in 10 mm NaCl at ph 9 for 9 h. Prior to water glass adsorption, the concentrate was preconditioned with 10 mm NaCl at ph 9 for 1 hour. The points at 0 mg g -1 represent the contact angle of the magnetite concentrate after conditioning with 10 mm NaCl at ph 9 for 1 hour. In concert with the results obtained for the synthetic magnetite, wettability of magnetite concentrate after flotation was significantly improved by water glass adsorption, with a contact angle of 28 ± 3 obtained at the highest water glass dosage. As could be expected, the hydrophilizing effect increased with increased water glass concentration. The adsorption of sodium polyacrylate on magnetite concentrate was performed in the presence of calcium ions, based on the conclusion about the adsorption mode of polyacrylate on magnetite discussed above. The wettability of the concentrate slightly improved upon polymer adsorption at the concentration of 0.04 mg g -1 (Fig. 9 in Paper V). However, the increase in polymer concentration at constant concentration of calcium ions did not lead to a further decrease in the contact angle of the magnetite concentrate. Since calcium ions are 40

53 known to facilitate interaction of polyacrylate with metal oxides [45], polyacrylate adsorption in the presence of calcium ions could be determined not only by polymer concentration in solution but also by the calcium-to-polymer ratio. Fig. 20 illustrates the effect of calcium chloride concentration on the wettability of magnetite concentrate at constant concentration of sodium polyacrylate. Figure 20. Water contact angle of the magnetite concentrate upon modification of the surface with 0.04 mg g -1 sodium polyacrylate at ph 9 for 1 hour measured with the Washburn technique. The point at 0 mm represents the contact angle of the magnetite concentrate treated with sodium polyacrylate in the presence of 10 mm NaCl and without calcium ions. Without calcium ions, treatment with the polymer did not have any effect on the wettability of the magnetite concentrate. On adding 4 mm CaCl 2 at the same concentration of polyacrylate, the contact angle of the magnetite concentrate decreased as discussed above. Even better results were achieved when the CaCl 2 concentration was increased to 6 mm, indicating that the efficiency of polyacrylate in improving magnetite wettability was affected by the concentration of calcium ions. 41

54 Summary and implications for agglomeration of iron ore To summarize the results of surface modification of magnetite after surfactant adsorption, the following conclusions can be drawn: 1. Soluble silicate and polyacrylate were shown to improve wettability of synthetic magnetite after surfactant adsorption and of mineral magnetite concentrate cleaned by flotation. 2. The effect of sodium polyacrylate on magnetite wettability improved in the presence of calcium ions. 3. The adsorption of sodium polyacrylate did not have any major effect on the zetapotential of synthetic magnetite particles. Consequently, both water glass and sodium polyacrylate may be used to improve the wettability of magnetite concentrate after flotation and prior to agglomeration. Improved wetting of magnetite concentrate could be expected to facilitate agglomeration and increase the strength of the pellets produced. Compared to water glass, treatment with polyacrylate would require less time and would not introduce any impurities to the final product after sintering. The fact that calcium ions facilitate adsorption of polyacrylate on magnetite makes the polymer suitable for the application in processes utilizing process water rich in calcium. Since the zeta-potential of magnetite was not affected by polyacrylate adsorption to any considerable extent, treatment with the polymer would not impair the electrostatic interaction between the magnetite concentrate and bentonite binder in agglomeration. Considering environmental issues, polyacrylate with a molecular weight similar to the one used in this study (M w 4500 and 4000, respectively) has not been found to have any adverse effect on the environment [82]. Despite low biodegradability [83], polyacrylate and its products of degradation are not toxic to aquatic, terrestrial, and mammalian species [82, 84]. In hard water, polyacrylate can precipitate in the form of calcium polyacrylate when all the carboxylic groups in the polymer become neutralized by calcium ions [82], resulting in polymer removal from the aqueous phase. Accordingly, polyacrylate seems to be a good candidate for use in improving wettability of flotated magnetite concentrate prior to agglomeration. 42

55 CONCLUSIONS A versatile method based on ATR-FTIR spectroscopy was developed and successfully used for in-situ studies of the adsorption of surfactants, polymers, and inorganic compounds on thin films of synthetic iron oxides. Using the developed method, the adsorption mechanism of several surfactants on iron oxides was elucidated. The dramatic effect of calcium ions on the adsorption of carboxylate surfactants on magnetite was for the first time confirmed in-situ. It was also illustrated that soluble silicate could reduce surfactant adsorption on magnetite but only in the absence of calcium ions. Among other factors that affected surfactant adsorption on iron oxides were surfactant concentration, ph, and ionic strength. Variation of conditioning time with sodium silicate was not found to have any considerable effect on surfactant adsorption on magnetite in the presence of calcium ions. Surfactant adsorption considerably decreased magnetite wettability. Once adsorbed, surfactant species could not be completely removed from the surface by rinsing with water, even at elevated ph. Treatment with soluble silicate and polyacrylate proved to be a feasible means for restoring wettability of magnetite after surfactant adsorption. The effectiveness of polyacrylate improved in the presence of calcium ions, making this polymer a good candidate for applications in hard water. The results obtained using synthetic iron oxides were verified for natural magnetite particles, suggesting that in-situ ATR-FTIR spectroscopy in combination with zeta-potential and contact angle measurements on synthetic materials could be successfully applied to studying surface phenomena related to mineral processing. 43

56 44

57 FUTURE WORK Since it has previously been shown that storage of the magnetite concentrate prior to agglomeration helps to improve green pellet quality, it would be interesting to investigate the aging of flotation collector and other species (e.g. silicate) present on the magnetite surface. To confirm or disprove the proposed adsorption mechanism of silicate, surfactant, and polymer on magnetite in the presence of calcium ions, it would be useful to study the distribution of these species on the magnetite surface using ATR-FTIR microscopy. To test whether treatment of the flotated magnetite concentrate with water glass or polyacrylate could improve green pellet quality, small-scale balling of the concentrate could be performed, followed by porosity, strength, and plasticity studies of the green pellets. Interactions between bentonite binder and magnetite modified with surfactant and/or polyacrylate could also be studied in order to obtain information about possible effects of these species adsorbed on magnetite concentrate on bentonite performance as a binder. 45

58 46

59 BIBLIOGRAPHY [1] L. Weng, W.H. Van Riemsdijk, T. Hiemstra, Effects of fulvic and humic acids on arsenate adsorption to goethite: Experiments and modeling, Environ. Sci. Technol. 43 (2009) [2] K.H. Rao, K.S.E. Forssberg, Chemistry of iron oxide flotation, in: M.C. Fuerstenau, G. Jameson, R.-H. Yoon (Eds.), Froth flotation: a century of innovation, Society for Mining, Metallurgy and Exploration, Inc., Littleton, 2007, pp [3] S.P.E. Forsmo, S.E. Forsmo, B.M.T. Björkman, P.O. Samskog, Studies on the influence of a flotation collector reagent on iron ore green pellet properties, Powder Technol. 182 (2008) [4] J.O. Gustafsson, G. Adolfsson, Adsorption of carboxylate collectors on magnetite and their influence on the pelletizing process, in: H. Hoberg, H. von Blottnitz (Eds.), XX International Mineral Processing Congress, GDMB Gessellschaft fur Bergbau, Metallurgie, Rohstoff- und Umwelttechnik, Aachen, Germany, 1997, pp [5] I. Iwasaki, J.D. Zetterström, E.M. Kalar, Removal of fatty acid coatings from iron oxide surfaces and its effect on the duplex flotation process and on pelletizing, Trans. Soc. Min. Eng. AIME 238 (1967) [6] G. Qiu, T. Jiang, H. Li, D. Wang, Functions and molecular structure of organic binders for iron ore pelletization, Colloids and Surfaces A: Physicochemical and Engineering Aspects 224 (2003) [7] R.S.C. Smart, W. Skinner, A.R. Gerson, J. Mielczarski, S. Chryssoulis, A.R. Pratt, R. Lastra, G.A. Hope, X. Wang, K. Fa, J.D. Miller, Surface characterization and new tools for research, in: M.C. Fuerstenau, G. Jameson, R.-H. Yoon (Eds.), Froth flotation: a century of innovation, Society for Mining, Metallurgy, and Exploration, Inc., Littleton, 2007, pp [8] Sustainability report, LKAB, 2010, pp [9] S.P.E. Forsmo, S.E. Forsmo, P.O. Samskog, B.M.T. Björkman, Mechanisms in oxidation and sintering of magnetite iron ore green pellets, Powder Technol. 183 (2008) [10] B.A. Wills, Mineral processing technology: an introduction to the practical aspects of ore treatment and mineral recovery, 5th ed., Pergamon Press, Oxford,

60 [11] V.A. Glembotskii, V.I. Klassen, I.N. Plaksin, Flotation, Primary Source, New York, [12] R.W. Smith, Cationic and amphoteric collectors, in: P. Somasundaran, B.M. Moudgil (Eds.), Reagents in mineral technology, Marcel Dekker, Inc., New York, 1988, pp [13] J.S. Clunie, B.T. Ingram, Adsorption of non-ionic surfactants, in: G.D. Parfitt, C.H. Rochester (Eds.), Adsorption from solution at the solid/liquid interface, Academic Press, London, 1983, pp [14] D.R. Nagaraj, S.A. Ravishankar, Flotation reagents - a critical overview from an industry perspective, in: M.C. Fuerstenau, G. Jameson, R.-H. Yoon (Eds.), Froth flotation: a century of innovation, Society for Mining, Metallurgy, and Exploration, Inc., Littleton, 2007, pp [15] J.A. Rajala, R.W. Smith, Ionic strength, collector chain length and temperature interactions in alkyl sulfate flotation of hematite, Transactions of the American Institute of Mining, Metallurgical, and Petroleum Engineers 274 (1983) [16] N.P. Finkelstein, The activation of sulphide minerals for flotation: A review, Int. J. Miner. Process. 52 (1997) [17] K.H. Rao, P.O. Samskog, K.S.E. Forssberg, Pulp chemistry of flotation of phosphate gangue from magnetite fines, Trans. Inst. Min. Metall. C 99 (1990) [18] D.C. Yang, Reagents in iron ore processing, in: P. Somasundaran, B.M. Moudgil (Eds.), Reagents in mineral technology, Marcel Dekker, Inc., New York, 1988, pp [19] A.E.C. Peres, A.C. Araujo, H. El-Shall, P. Zhang, N.A. Abdel-Khalek, Plant practice:nonsulfide minerals, in: M.C. Fuerstenau, G. Jameson, R.-H. Yoon (Eds.), Froth flotation: a century of innovation, Society for Mining, Metallurgy, and Exploration, Inc., Littleton, 2007, pp [20] H. El-Shall, P. Zhang, N.A. Abdel-Khalek, P. Somasundaran, Phosphate flotation practice, in: M.C. Fuerstenau, G. Jameson, R.-H. Yoon (Eds.), Froth flotation: a century of innovation, Society for Mining, Metallurgy, and Exploration, Inc., Littleton, 2007, pp [21] J. Leja, Surface chemistry of froth flotation, Plenum Press, New York, [22] D.M. Newitt, J.M. Conway-Jones, A contribution to the theory and practice of granulation, Trans. Inst. Chem. Eng. 36 (1958) [23] C.G. Barlow, Granulation of powders, Chem. Eng. 36 (1968)

61 [24] B.C. Hancock, P. York, R.C. Rowe, An assessment of substrate-binder interactions in model wet masses. 1: Mixer torque rheometry, Int. J. Pharm. 102 (1994) [25] H. Rumpf, The strength of granules and agglomerates, in: W.A. Knepper (Ed.), Agglomeration, Interscience, New York, 1962, pp [26] S.P.E. Forsmo, A.J. Apelqvist, B.M.T. Björkman, P.O. Samskog, Binding mechanisms in wet iron ore green pellets with a bentonite binder, Powder Technol. 169 (2006) [27] S.M. Iveson, J.D. Litster, K. Hapgood, B.J. Ennis, Nucleation, growth and breakage phenomena in agitated wet granulation processes: A review, Powder Technol. 117 (2001) [28] W. Pietsch, Size enlargement by agglomeration, John Wiley & Sons, Chichester, [29] T. Young, An essay on the cohesion of fluids, Philos. Trans. R. Soc. London 95 (1805) [30] C.J. van Oss, M.K. Chaudhury, R.J. Good, Monopolar surfaces, Adv. Colloid Interface Sci. 28 (1987) [31] H.H. Kung, Transition Metal Oxides: Surface Chemistry and Catalysis, Elsevier, New York, [32] Y. Wang, J. Ren, The flotation of quartz from iron minerals with a combined quaternary ammonium salt, Int. J. Miner. Process. 77 (2005) [33] S.M. Iveson, S. Holt, S. Biggs, Advancing contact angle of iron ores as a function of their hematite and goethite content: Implications for pelletising and sintering, Int. J. Miner. Process. 74 (2004) [34] T.T. Chau, A review of techniques for measurement of contact angles and their applicability on mineral surfaces, Miner. Eng. 22 (2009) [35] T.T. Chau, W.J. Bruckard, P.T.L. Koh, A.V. Nguyen, A review of factors that affect contact angle and implications for flotation practice, Adv. Colloid Interface Sci. 150 (2009) [36] J. Shang, M. Flury, J.B. Harsh, R.L. Zollars, Comparison of different methods to measure contact angles of soil colloids, J. Colloid Interface Sci. 328 (2008) [37] S. Kirchberg, Y. Abdin, G. Ziegmann, Influence of particle shape and size on the wetting behavior of soft magnetic micropowders, Powder Technol. 207 (2011) [38] D.J. Shaw, Introduction to colloid and surface chemistry, 4th ed., Butterworth- Heinemann, Oxford,

62 [39] U. Schwertmann, R.M. Cornell, Iron oxides in the laboratory: preparation and characterization, 2nd ed., Wiley-VCH, Weinheim, [40] J. Lyklema, Adsorption of small ions, in: G.D. Parfitt, C.H. Rochester (Eds.), Adsorption from solution at the solid/liquid interface, Academic press, London, 1983, pp [41] L.K. Koopal, E.M. Lee, M.R. Böhmer, Adsorption of Cationic and Anionic Surfactants on Charged Metal Oxide Surfaces, J. Colloid Interface Sci. 170 (1995) [42] R. Zhang, P. Somasundaran, Advances in adsorption of surfactants and their mixtures at solid/solution interfaces, Adv. Colloid Interface Sci (2006) [43] G.J. Fleer, J. Lyklema, Adsorption of polymers, in: G.D. Parfitt, C.H. Rochester (Eds.), Adsorption from solution at the solid/liquid interface, Academic Press, London, 1983, pp [44] A. Pettersson, G. Marino, A. Pursiheimo, J.B. Rosenholm, Electrosteric stabilization of Al 2 O 3, ZrO 2, and 3Y-ZrO 2 suspensions: Effect of dissociation and type of polyelectrolyte, J. Colloid Interface Sci. 228 (2000) [45] K. Vermöhlen, H. Lewandowski, H.D. Narres, M.J. Schwuger, Adsorption of polyelectrolytes onto oxides - The influence of ionic strength, molar mass, and Ca 2+ ions, Colloids and Surfaces A: Physicochemical and Engineering Aspects 163 (2000) [46] L.K. Koopal, The effect of polymer polydispersity on the adsorption isotherm, J. Colloid Interface Sci. 83 (1981) [47] F.T. Hesselink, Adsorption of polyelectrolytes from dilute solution, in: G.D. Parfitt, C.H. Rochester (Eds.), Adsorption from solution at the solid/liquid interface, Academic Press, London, 1983, pp [48] N.B. Colthup, L.H. Daly, S.E. Wiberly, Introduction to Infrared and Raman Spectroscopy, 3rd ed., Academic Press, London, [49] E.W. Giesekke, A review of spectroscopic techniques applied to the study of interactions between minerals and reagents in flotation systems, Int. J. Miner. Process. 11 (1983) [50] W. Van Bronswijk, L.J. Kirwan, P.D. Fawell, In situ adsorption densities of polyacrylates on hematite nano-particle films as determined by ATR-FTIR spectroscopy, Vib. Spectrosc 41 (2006) [51] A. Omoike, J. Chorover, Spectroscopic study of extracellular polymeric substances from Bacillus subtilis: Aqueous chemistry and adsorption effects, Biomacromolecules 5 (2004)

63 [52] A.R. Hind, S.K. Bhargava, A. McKinnon, At the solid/liquid interface: FTIR/ATR - The tool of choice, Adv. Colloid Interface Sci. 93 (2001) [53] E. Matijevic, Production of monodispersed colloidal particles, Annu. Rev. Mater. Sci. 15 (1985) [54] R. Massart, V. Cabuil, Effect of some parameters on the formation of colloidal magnetite in alkaline medium: Yield and particle size control, J. Chim. Phys. 84 (1987) [55] H.O. Lien, J.G. Morrow, Beneficiation of lean iron ores solely by selective flocculation and desliming, CIM Bull. 71 (1978) [56] S.P.E. Forsmo, Oxidation of magnetite concentrate powders during storage and drying, Int. J. Miner. Process. 75 (2005) [57] K. Kandori, Y. Kawashima, T. Ishikawa, Zeta-potential of monodispersed cubic haematite particles containing chloride ions, J. Mater. Sci. Lett. 12 (1993) [58] W.A. Zeltner, M.A. Anderson, Surface charge development at the goethite/aqueous solution interface: Effects of CO2 adsorption, Langmuir 4 (1988) [59] D.G. Cameron, J. Umemura, P.T.T. Wong, H.H. Mantsch, A fourier transform infrared study of the coagel to micelle transitions of sodium laurate and sodium oleate, Colloids and surfaces 4 (1982) [60] K.D. Dobson, A.J. McQuillan, In situ infrared spectroscopic analysis of the adsorption of aliphatic carboxylic acids to TiO 2, ZrO 2, Al 2 O 3, and Ta 2 O 5 from aqueous solutions, Spectrochimica Acta - Part A: Molecular and Biomolecular Spectroscopy 55 (1999) [61] W.L. Driessen, W.L. Groeneveld, F.W. Van Der Wey, Complexes with ligands containing the carbonyl group. Part II. Metal(II) methyl formate, ethyl acetate and diethyl malonate solvates, Recl. Trav. Chim. Pays-Bas 89 (1970) [62] P.C.J. Beentjes, J. Van Den Brand, J.H.W. De Wit, Interaction of ester and acid groups containing organic compounds with iron oxide surfaces, J. Adhes. Sci. Technol. 20 (2006) [63] H.H. Zeiss, M. Tsutsui, The carbon-oxygen absorption band in the infrared spectra of alcohols, J. Am. Chem. Soc. 75 (1953) [64] Y.S. Hwang, J.J. Lenhart, Adsorption of C4-dicarboxylic acids at the hematite/water interface, Langmuir 24 (2008)

64 [65] W.Q. Gong, A. Parentich, L.H. Little, L.J. Warren, Diffuse reflectance infrared Fourier transform spectroscopic study of the adsorption mechanism of oleate on haematite, Colloids and surfaces 60 (1991) [66] R.D. Kulkarni, P. Somasundaran, Flotation chemistry of hematite/oleate system, Colloids and surfaces 1 (1980) [67] J.J. Kellar, C.A. Young, K. Knutson, J.D. Miller, Thermotropic phase transition of adsorbed oleate species at a fluorite surface by in situ FT-IR/IRS spectroscopy, J. Colloid Interface Sci. 144 (1991) [68] D.D. Do, Adsorption analysis: equilibria and kinetics, Imperial College Press, London, [69] P. Roonasi, X. Yang, A. Holmgren, Competition between sodium oleate and sodium silicate for a silicate/oleate modified magnetite surface studied by in situ ATR-FTIR spectroscopy, J. Colloid Interface Sci. 343 (2010) [70] Y. Lu, J. Drelich, J.D. Miller, Oleate adsorption at an apatite surface studied by ex-situ FTIR internal reflection spectroscopy, J. Colloid Interface Sci. 202 (1998) [71] L.J. Morgan, K.P. Ananthapadmanabhan, P. Somasundaran, Oleate adsorption on hematite: Problems and methods, Int. J. Miner. Process. 18 (1986) [72] K.P. Ananthapadmanabhan, P. Somasundaran, Acid-soap formation in aqueous oleate solutions, J. Colloid Interface Sci. 122 (1988) [73] K.H. Rao, J.M. Cases, P. De Donato, K.S.E. Forssberg, Mechanism of oleate interaction on salt-type minerals. IV. Adsorption, electrokinetic, and diffuse reflectance FT-IR studies of natural fluorite in the presence of sodium oleate, J. Colloid Interface Sci. 145 (1991) [74] T. Wakamatsu, D.W. Fuerstenau, Effect of alkyl sulfonates on the wettability of alumina, Trans. Soc. Mining. Eng. AIME 254 (1973) [75] A. Yehia, M.A. Youseff, T.R. Boulos, Different alternatives for minimizing the collector consumption in phosphate fatty acid flotation, Miner. Eng. 3 (1990) [76] J. Engesser, Effect of water chemistry, water treatment and Blaine on magnetite filtering and magnetite agglomeration with bentonite clay, Miner. Metall. Process 20 (2003) [77] Q. Liu, Y. Zhang, Effect of calcium ions and citric acid on the flotation separation of chalcopyrite from galena using dextrin, Miner. Eng. 13 (2000) [78] T.C. Eisele, S.K. Kawatra, S.J. Ripke, Water chemistry effects in iron ore concentrate agglomeration feed, Miner. Process. Extr. Metall. Rev. 26 (2005)

65 [79] W. Von Rybinski, M.J. Schwuger, P.A. Schulz, B. Dobias, Zeolites for increased selectivity and recovery in flotation processes, Reagents in the Minerals Industry (1984) [80] P. Somasundaran, J. Cleverdon, A study of polymer/surfactant interaction at the mineral/solution interface, Colloids and surfaces 13 (1985) [81] F. Jones, J.B. Farrow, W. Van Bronswijk, An infrared study of a polyacrylate flocculant adsorbed on hematite, Langmuir 14 (1998) [82] M.B. Freeman, T.M. Bender, An environmental fate and safety assessment for a low molecular weight polyacrylate detergent additive, Environ. Technol. 14 (1993) [83] R.J. Larson, E.A. Bookland, R.T. Williams, K.M. Yocom, D.A. Saucy, M.B. Freeman, G. Swift, Biodegradation of acrylic acid polymers and oligomers by mixed microbial communities in activated sludge, Journal of Environmental Polymer Degradation 5 (1997) [84] B.D. Cook, P.R. Bloom, T.R. Halbach, Fate of a polyacrylate polymer during composting of simulated municipal solid waste, Journal of Environmental Quality 26 (1997)

66 54

67 PAPER I Studies of collector adsorption on iron oxides by in-situ ATR-FTIR spectroscopy E. Potapova, I. Carabante, M. Grahn, A. Holmgren, and J. Hedlund Industrial and Engineering Chemistry Research 49 (2010)

68

69 Ind. Eng. Chem. Res. 2010, 49, Studies of Collector Adsorption on Iron Oxides by in Situ ATR-FTIR Spectroscopy E. Potapova,* I. Carabante, M. Grahn, A. Holmgren, and J. Hedlund DiVision of Chemical Engineering, Luleå UniVersity of Technology, SE Luleå, Sweden In this work, the adsorption of three model collectors, viz., poly(ethylene glycol) monooleate (PEGMO), ethyl oleate, and maleic acid, as well as the commercial fatty-acid-type collector Atrac 1563, was studied in situ on synthetic hematite using attenuated total reflectance Fourier transform infrared (ATR-FTIR) spectroscopy. The adsorption behavior of the studied compounds on hematite was determined to a large extent by the polar headgroup. Adsorption of Atrac and PEGMO as a function of concentration showed good agreement with the Freundlich adsorption model, suggesting energetically heterogeneous adsorption. In situ desorption experiments revealed that a large fraction of the Atrac was weakly attached to the hematite surface, as it was partially removed by flushing with water at ph 8.5 and 10. These results suggest that a separate washing unit after the flotation step could be beneficial in reducing the contamination of iron ore by flotation chemicals. Introduction Iron ore pellets are an important refined product used as a raw material in the manufacturing of steel. The production of iron ore pellets comprises several stages: grinding and upgrading of the iron ore; balling of wet, so-called, green pellets; and drying, sintering, and oxidation of the green pellets to the final product, to be transported to iron or steel plants. LKAB is a Swedish mining company whose pelletizing plants utilize magnetite iron ore from two deposits located in northern Sweden: Kiruna and Malmberget. The Kiruna ore is a mixture of magnetite and apatite having a phosphorus content of ca. 1 wt %. To reduce the phosphorus content to an acceptable level for the blast furnace process 1 (i.e., to less than 0.025%), the ore is subjected to reverse flotation with an anionic fatty-acidtype collector reagent (Atrac 1563) with methyl isobutyl carbinol (MIBC) used as a frother. To increase flotation selectivity and phosphorus recovery, sodium silicate is added to the system. The sodium silicate is used in fairly small amounts ( gt -1 ) and thus acts primarily as a dispersant, and it has not been found to prevent collector adsorption on the magnetite surface to any great extent. 2,3 Ideally, the collector should adsorb only on the apatite gangue, rendering it hydrophobic and thus easily floated from the magnetite. However, unwanted adsorption of the flotation reagents on magnetite also occurs. It has been estimated that the amount of Atrac adsorbed on the magnetite fed to balling circuits is 10-30gt Once the collector is adsorbed on the surface of magnetite, it is difficult to eliminate. 5 Therefore, the collector should be added in sufficiently small amounts so that the apatite surface is rendered hydrophobic whereas the adsorption on magnetite is minimized. However, in the real process, the Atrac dosage is adjusted based on the phosphorus content in the pellet feed, and there is a slight degree of overdosage to ensure that the desired phosphate levels are achieved. Typically, the dosage varies in the range of g per tonne of magnetite concentrate. 6 After the flotation step, the pulp passes through magnetic separation and filtration steps where it is subjected to repeated * To whom correspondence should be addressed. elisaveta.potapova@ltu.se. Tel.: dilutions and thickenings that can have a mild washing effect on the ore, partly removing the flotation reagents from the magnetite surface. However, no separate washing unit is used for the purification of magnetite from the flotation chemicals at LKAB. It has been found that adsorption of the flotation collector agent on the iron ore has a negative effect on the balling process, as the collector adsorbed on the surface of magnetite makes the particles more hydrophobic, which can lead to the attachment of air bubbles onto the surface of the iron ore particles. 6 The air bubbles incorporated into green pellets decrease the green pellet strength in both the wet and dry states. A low wet strength tends to cause a wide size distribution of the green pellets leaving the balling drums. The undersize fraction is recirculated back to the balling drum, thus leading to increased energy consumption and decreased capacity of the pelletizing plant. Breakage of the pellets at the stage of drying and induration causes dust formation and decreased pellet bed permeability, resulting in lower production volumes and aggravated pellet quality. 7 To minimize the influence of the collector on the pelletizing process, it is important to understand the mechanism by which the collector interacts with the iron oxide. For instance, it has been shown that the presence of Ca 2+ ions in the process water increases the adsorption of flotation collector reagent on magnetite. 8 In that work, the adsorption of the collector OS 130 on magnetite was studied in the batch experiments by measuring the residual concentration of the collector in solution after adsorption using the method of Gregory. 9 According to the suggested 8 mechanism, positively charged calcium ions adsorb at negatively charged surface sites on the magnetite surface, which results in a more positively charged surface, rendering it more favorable for the adsorption of negatively charged collector species. Spectroscopic techniques have been used extensively for studying the adsorption of fatty-acid-based collectors onto mineral surfaces Fourier transform infrared (FTIR) spectroscopy is widely applied because it provides the possibility of identifying complexes formed at the surface. 13 Ex situ FTIR techniques, such as diffuse reflectance infrared Fourier transform (DRIFT) spectroscopy and reflection absorption infrared spec /ie901343f 2010 American Chemical Society Published on Web 01/07/2010

70 1494 Ind. Eng. Chem. Res., Vol. 49, No. 4, 2010 Figure 1. Schematic image of an ATR waveguide illustrating the ATR phenomenon. For the sake of clarity, the thickness of the sample (in this case, iron oxide film) and the penetration depth d p are enlarged. troscopy (RAIRS), imply drying of the sample after adsorption, which can affect surface complexes and, hence, the validity of the results. On the contrary, FTIR attenuated total reflectance (FTIR-ATR) spectroscopy facilitates in situ studies of both adsorption kinetics 14 and complexes formed on the surface in the presence of water. 13,15 In the ATR technique (Figure 1), a sample with a certain refractive index, n 1, is placed in a close contact with a waveguide characterized by a high refractive index, n 2. The IR beam is passed through the waveguide at a certain incident angle θ. In the case when sin θ is larger than the ratio n 1 /n 2, the beam is totally reflected inside the waveguide, forming a perpendicular evanescent wave of the IR radiation at each point of reflection. This wave propagates through the sample in the vicinity of the waveguide and interacts with the sample, causing partial absorption of the radiation by the sample material and thus a reduction of the intensity of the totally reflected beam. The penetration depth, d p, is defined as the distance from the interface at which the electric field of the wave is equal to E 0 e -1 (see Figure 1), where E 0 is the electric field of the wave at the interface. Typical values of the penetration depth are from some hundred nanometers to a few micrometers, depending on the refractive indices of both the waveguide and the sample, as well as on the wavelength and incident angle of the radiation. Such a range of penetration depths makes FTIR- ATR spectroscopy a very surface-sensitive technique, providing the possibility for in situ studies of surfaces in contact with aqueous solutions. During the past decade, the FTIR-ATR technique has been developing, and new applications in surface chemistry have evolved, for example, as a tool for studying adsorption 13,16-19 and diffusion in thin films or at interfaces. This technique has also been used in studies of catalytic reactions, 25,26 as well as in sensor applications. 27,28 FTIR-ATR spectroscopy has also been applied for studying the adsorption of surfactants at mineral surfaces, both qualitatively 29,30 and quantitatively. 14,31-33 Our group has developed the ATR technique for studies of adsorption in zeolite films 17,27,34-37 and on mineral surfaces. 14,18,38-41 The studies have been both qualitative 35-38,40,41 and quantitative, 14,17,18,27,34,39 and even the molecular orientations of adsorbates have been determined. 17,18,35,38,41 Although magnetite (Fe 3 O 4 ) is the main iron-containing mineral in the ore utilized by LKAB, it was found to become partly oxidized during storage in air, forming first maghemite (γ-fe 2 O 3 ), which has the same crystal structure but mostly Fe 3+ on the surface, and then hematite (R-Fe 2 O 3 ), which has a different crystal structure than magnetite and maghemite. 42 This phenomenon was described earlier for submicrometer-sized magnetite particles by both Haneda and Morrish 43 and Gedikoglu. 44 In solution, the oxidation of magnetite was also observed and explained by the leaching of Fe 2+ ions from the surface. 45 Thus, when magnetite ore is subjected to grinding and flotation, apparent oxidation of the surface can occur both in water and in air. In this work, the adsorption of flotation collector reagents on iron oxide was studied in situ using FTIR-ATR spectroscopy for the purpose of obtaining essential information about the interaction between the flotation collector reagents and the iron oxide surface in the presence of water. This information is crucial for the improvement of the pelletizing process, as it could suggest possibilities to reduce the unwanted collector adsorption on the iron ore surface and improve green-pellet strength. Hematite was chosen as the adsorbent because it was reported to be the final product of the surface oxidation of magnetite. In addition, reports on the adsorption properties of collector-type molecules on hematite were found to be very scant in the literature. Materials and Methods Materials. Hematite crystals were synthesized from an FeCl 3 solution according to the method described by Matijević. 46 The obtained hematite crystals were purified by repeated (five times) centrifugation at rpm for 30 min and redispersion in a 0.06 M aqueous solution of acetic acid (glacial, >99.7%, Alfa Aesar). The crystals were stored as a2wt% suspension in a 0.06 M aqueous solution of acetic acid at ph 3. For powder X-ray diffraction analysis, the suspension was freeze-dried, yielding a fine hematite powder. The flotation collector reagent, Atrac 1563 (Akzo Nobel, Sweden), was provided by LKAB. Atrac 1563 is a yellow viscous liquid with a complex chemical composition: % ethoxylated tall oil ester of maleic acid and 1-5% maleic anhydride (Akzo Nobel material safety data sheet). Tall oil is a byproduct of the Kraft pulp manufacturing process and is a mixture of mainly fatty acids (e.g., oleic acid) and resin acids (e.g., abietic acid). A similar collector for the froth flotation of oxide and salt-type minerals that is a combination of a monoester of a dicarboxylic acid and a monocarboxylic acid is described in a patent. 47 As described in the patent, the first component is an aliphatic monocarboxylic acid containing 8-22 carbon atoms bonded to a dicarboxylic acid containing 4-8 carbon atoms through an alkylene oxide group with 2-4 carbon atoms, thus resulting in a molecule with two ester carbonyls and one free carboxylic group at the end of the molecule. Monocarboxylic acid with 6-24 carbon atoms is added to increase the selectivity and/or yield of the monoester. Poly(ethylene glycol) monooleate (PEGMO) with a typical number-average molecular weight (M n ) of 460 (Aldrich), maleic acid (Fluka, g 99%), and ethyl oleate (Aldrich, 98%) were used as model collector reagents. From the average molar mass, the length of the poly(ethylene glycol) chain in the PEGMO model collector reagent was estimated to be about four ethylene glycol units. Oleic acid esters were chosen because oleic acid is one of the main components of tall oil. Ethoxylated tall oil could thus be modeled by using PEGMO, which has the same structure as ethoxylated oleic acid. Maleic acid was chosen as one of the model compounds because it is the tail group of the molecules in Atrac. Ethyl oleate was used as the third model compound to study the effect of the poly(ethylene glycol) chain on the adsorption properties of oleate. Working solutions of Atrac and the model collector reagents were prepared in the following way: First, a 0.1 g L -1 stock solution of the compound in distilled water was prepared. In

71 the next step, the required amount of the stock solution was mixed in distilled water to give final solutions of the desired concentration (1-25 mg L -1 ). All aqueous solutions were prepared using distilled water. The distilled water used for the spectroscopic measurements was first boiled for 1 h, and then argon was bubbled through the water to minimize the amount of dissolved carbon dioxide. The ph was controlled during the experiments by a Mettler Toledo T70 titrator using a 0.05 M aqueous solution of sodium hydroxide (per analysis, Merck). Dip-Coating. Hematite films were deposited on ZnSe substrates using a Nima DC-multi 8 dip-coater. Prior to deposition, the substrates were washed in acetone (g99.5%, VWR), ethanol (99.7%, Solveco Chemicals AB), and distilled water (10 min in each). A 2 wt % hematite suspension was prepared by dispersing the hematite crystals in a 6.27 M aqueous solution of acetic acid. The substrates were immersed in the suspension and withdrawn at a speed of 5 mm min -1, and the film thickness was controlled by the number of dips. To prepare a film with a thickness of ca. 1 μm, eight dips were needed. Scanning Electron Microscopy (SEM). SEM images of the hematite film on a ZnSe substrate were obtained using a Philips XL 30 microscope with a LaB 6 filament. The samples were mounted on alumina stubs using carbon glue and subsequently sputtered with a thin layer (ca. 10 nm) of gold to provide conductivity. X-ray Diffraction. X-ray diffraction patterns of both hematite powder and film were collected with a Siemens D5000 diffractometer running in Bragg-Brentano geometry using Cu K R radiation. To analyze the film, the hematite-covered ZnSe substrate was mounted with carbon glue onto a custom-made alumina holder. Zeta-Potential Measurements. The point of zero charge (PZC) of the hematite crystals used in this work was determined by electrophoresis using a ZetaCompact instrument equipped with a charge-coupled device (CCD) tracking camera. The obtained electrophoretic mobility data were further processed by the Zeta4 software applying the Smoluchowski equation. The samples were prepared in the following way: One drop of the hematite suspension was dispersed in 1 L of 0.01 M potassium nitrate. The ph of the samples (10 samples, 100 ml each) was adjusted using potassium hydroxide and nitric acid. The samples spanned the ph range from 2 to 11. For each sample, the measurement was repeated three times, and the final PZC was calculated as an average of the obtained values. FTIR-ATR Spectroscopy. Infrared spectra were recorded using a Bruker IFS 66v/S spectrometer equipped with a liquidnitrogen-cooled mercury cadmium telluride (MCT) detector. ZnSe ATR crystals (Crystran Ltd.) in the form of a trapeze with 45 cut edges and dimensions of mm were used in this study. Measurements of adsorption on the hematite-coated ATR crystals were performed in situ in a cell with a flow pumped through on both sides of the ATR crystal; see Figure 2. The incidence angle of the infrared beam was set to 45. All adsorption experiments on hematite were performed at room temperature using water solutions of model collectors or Atrac at ph 8.5 (the ph used in the flotation process at LKAB) pumped continuously through the cell at a flow rate of 10 ml min -1 with recirculation. Prior to the adsorption measurements, the hematite film was flushed with a weakly alkaline solution (ph 8.5) for 2 h to remove the residues of acetic acid and carbonate species from the surface. A background spectrum was recorded afterward with water at ph 8.5 in contact with a hematite-coated ZnSe substrate. Spectra of the model collectors Ind. Eng. Chem. Res., Vol. 49, No. 4, Figure 2. Schematic image of the FTIR-ATR flow cell. For the sake of clarity, the thickness of the iron oxide film is enlarged. Figure 3. Images of (a) uncoated and (b) hematite-coated ZnSe ATR crystals. and Atrac in pure form were recorded in argon atmosphere using a bare ZnSe substrate with a droplet of a collector spread over its surface. ZnSe in argon atmosphere was recorded as a singlebeam background spectrum. All background and sample spectra were obtained by averaging 500 scans at a resolution of 4 cm -1. Data processing was performed using Bruker Opus 4.2 software. Results and Discussion Film and Powder Characterization. Figure 3 shows a photograph of ZnSe crystals before and after being coated with a hematite film. As is evident from the uniform red color of the coated ATR crystal, the obtained film appeared to be quite uniform, continuous, and even along the crystal surface and stable under the conditions used for the in situ experiments, because the visual appearance of the coated crystal did not change after the experiments. SEM images (Figure 4) showed that the hematite crystals had a uniform spherical habit with a diameter of ca. 130 nm and that the crystals were distributed evenly over the ATR crystal surface, forming a porous film with an average thickness of ca. 1 μm. Figure 5 shows XRD patterns of freeze-dried synthetic hematite powder and a hematite film on a ZnSe crystal. The XRD pattern of the powder shows that the synthesized material contained pure, randomly oriented hematite crystals without any other iron oxide phase present in amounts detectable by XRD. By comparing the relative reflection intensities in the powder and in the film, it can be concluded that the crystals in the film were also randomly oriented. Figure 6 shows the ζ-potential of the synthetic hematite crystals used in this work as a function of ph. The point of

72 1496 Ind. Eng. Chem. Res., Vol. 49, No. 4, 2010 Figure 6. ζ-potential of the hematite crystals as a function of ph. Figure 4. (a) Top- and (b) side-view SEM images of a hematite film on a ZnSe crystal. Figure 5. XRD patterns of hematite powder (top) and hematite film on a ZnSe crystal (bottom). The peak labeled with an asterisk (*) emanates from the ZnSe substrate. zero charge (PZC) is at about ph 4.8. According to data compiled by Fuerstenau 48 and Cromieres, 49 the PZC of hematite has been reported to occur within a broad range of ph values, viz., from 4.8 to Factors that can influence the value of the PZC are 1 the technique used to determine the ζ-potential, impurities (especially in the case of minerals), sample preparation procedure, and species adsorbed on the surface (e.g., carbonates). 50,51 In the present study, the PZC of hematite was probably affected by the carbonate species adsorbed on the surface, which are known to lower the PZCs of iron oxides by ca. 1 ph unit. 52 The presence of the carbonate species was confirmed by spectra (not shown) recorded during flushing of the hematite film with distilled water at ph 8.5 prior to the adsorption experiments. Weak negative bands at ca and 1340 cm -1 originating from the outer-sphere carbonate species 52 were observed in the spectra, indicating the desorption of carbonates from the hematite surface. Desorption of carbonates had ceased already after 30 min of flushing, suggesting that carbonate species in solution and on the surface were in equilibrium. Adsorption of Model Compounds. To better understand the adsorption mechanism of Atrac on hematite, three model compounds of Atrac were studied. As it is known that Atrac contains maleic acid esterified with an ethoxylated tall oil (Akzo Nobel material safety data sheet), poly(ethylene glycol) monooleate (PEGMO) and maleic acid were chosen as model compounds. From the number-average molecular weight, M n ) 460, it was estimated that an average poly(ethylene glycol) chain in PEGMO contained four repeated ethoxy units (-CH 2 -O-CH 2 -). Ethyl oleate was used as an additional model compound to study the effect of the poly(ethylene glycol) chain on the adsorption properties of oleate. PEGMO is a nonionic surfactant 53 characterized by a higher solubility in water than the corresponding fatty acid because of the poly(ethylene glycol) chain. PEGMO can be expected to interact with the iron oxide surface in three different ways, viz., through the ester carbonyl, the ether oxygen linkages, or the tail hydroxyl group. Figure 7 shows spectra of a droplet of pure PEGMO on ZnSe and PEGMO adsorbed on hematite from aqueous solution. Strong absorption bands at 2922 and 2854 cm -1 in Figure 7a originate from asymmetric and symmetric stretching vibrations of the C-H bonds (ν as and ν s ). 54 Long-chain carboxylic acids (e.g., oleic acid) contain significantly more methylene groups than methyl groups and are thus characterized by much stronger absorption bands corresponding to the CH 2 groups compared to the CH 3 groups. The latter can be observed only as shoulders. C-H deformation in methyl and methylene groups is found around 1458 cm The absorption band observed in pure PEGMO at 1736 cm -1 (Figure 7a) is associated with stretching vibrations of the CdO ester bond. 56 It is only slightly shifted upon adsorption on hematite (Figure 7b), indicating that no significant interaction occurs between the ester carbonyl and the surface. The absence of bands around 1570 and 1430 cm -1, corresponding, respectively, to asymmetric and symmetric

73 Ind. Eng. Chem. Res., Vol. 49, No. 4, Figure 8. Intensity of the band originating from the symmetric stretching vibration (ν s) of the C-H bond as a function of concentration of PEGMO in solution (0). The solid line represents the fitted Freundlich adsorption model. Figure 7. Infrared spectra of (a) a droplet of pure PEGMO on ZnSe and (b) PEGMO adsorbed on a hematite film in situ from a 10 mg L -1 aqueous solution. Note that a.u. represents arbitrary units here and elsewhere. stretching of a carboxylate group, 12,57 suggests that the ester bond in PEGMO does not break upon adsorption on hematite. The intense band at 1115 cm -1 (Figure 7a) emanates from the C-O-C stretching vibration in the poly(ethylene glycol) chain. 58 It has a shoulder on its low-frequency side at ca cm -1 probably emanating from the stretching of the C-O bond between the hydroxyl group and the CH 2 group at the end of the poly(ethylene glycol) chain. 59 Both the peak frequency and the shoulder frequency are shifted to lower wavenumbers when PEGMO is adsorbed on hematite (see Figure 7b; 1115 f 1095 cm -1 and 1070 f 1047 cm -1, respectively), suggesting that the poly(ethylene glycol) chain is involved in the bonding to the surface. Figure 8 shows the change in absorbance (measured as peak height) of the 2854 cm -1 band during the adsorption of PEGMO on hematite at different concentrations after 1 h of adsorption at each concentration. The experiment was started at the lowest concentration of 1 mg L -1 ; thereafter, the concentration in solution was increased, and the measurement was continued. Figure 8 shows that the intensity of the band originating from the symmetric stretching vibration (ν s ) of the C-H bond increased with increasing concentration in solution, suggesting an increase of the adsorption of PEGMO on hematite. The experimental data exhibited a poor fit with the Langmuir model of adsorption, with a coefficient of determination (R 2 ) of Plotting the data on a logarithmic scale yielded a straight line with R 2 ) 0.99, indicating that the obtained data were in good agreement with the Freundlich model of adsorption, which implies that the surface of adsorption is heterogeneous, for instance, with different adsorption sites grouped patchwise based on their adsorption energies, as suggested earlier. 60 A desorption experiment (not shown) showed that the intensity of the band originating from the symmetric stretching vibration (ν s )ofthe C-H bond in PEGMO was reduced only by 5% upon flushing the cell with water at ph 8.5 for 1 h, indicating that most of the PEGMO was strongly adsorbed to the hematite surface. Maleic acid was the second model compound studied. The ability of maleic acid to form intramolecular hydrogen bonds makes it easily soluble in water. In aqueous solution at ph 8.5, it is expected to be fully deprotonated. 61 From the spectroscopic measurements it was concluded that maleic acid did not adsorb on hematite to any considerable extent at ph 8.5 in the concentration range studied, viz., 1-25 mg L -1. This could be explained by the fact that, at ph 8.5, the hematite surface is negatively charged and thus repels maleic acid, which, at this ph, contains two negatively charged carboxylate ions. Hwang and Lenhart 60 studied the adsorption of maleic acid on hematite at various ph values and concluded that the adsorption was controlled by the surface charge of hematite, suggesting that electrostatic interaction is the predominant force of adsorption of maleic acid on hematite, which agrees well with our observations. Ethyl oleate was the third model compound of Atrac used in the experiments. It is almost insoluble in water due to the fact that, instead of a hydrophilic poly(ethylene glycol) chain, it has a hydrophobic ethyl group bonded to the carboxylate. This change in chemical composition, of course, also affects the adsorption properties of the molecule. The change in absorbance of the 2852 cm -1 band during the adsorption of ethyl oleate on hematite at different concentrations is shown in Figure 9. The measurements were performed in the same way as for PEGMO. Equilibrium was achieved at each concentration within 2 h. Figure 9 shows that the absorbance of the band originating from the symmetric stretching vibration (ν s ) of the C-H bond is about a factor of 6 lower at all concentrations of ethyl oleate than the absorbance of the same band at corresponding concentrations of PEGMO. This indicates that about 6 times less ethyl oleate is adsorbed compared to PEGMO at the corresponding concentrations. The interaction between ethyl oleate and the surface was probably very weak, as no significant band shifts were observed in the spectrum of ethyl oleate on hematite (not shown) compared to the spectrum of pure ethyl oleate.

74 1498 Ind. Eng. Chem. Res., Vol. 49, No. 4, 2010 Table 1. Assignment of Absorption Bands Originating from Pure Atrac Spread over ZnSe and Atrac Adsorbed on Hematite in Situ froma10mgl -1 Aqueous Solution pure Atrac on ZnSe Atrac adsorbed on hematite peak assignment ν as (CH 2) ν s (CH 2) ν(cdo) in ester ν(cdo) in acid ν as (COO - ) ν s (COO - ) δ (CH 2 and CH 3) ν (C-O) in esters 63 Figure 9. Intensity of the band originating from the symmetric stretching vibration (ν s) of the C-H bond as a function of the concentration of ethyl oleate in solution. Figure 10. Infrared spectra of (a) a droplet of pure Atrac on ZnSe and (b) Atrac adsorbed on a hematite film in situ from a 10 mg L -1 aqueous solution. Atrac Adsorption. Anionic fatty-acid surfactants are believed to interact with iron oxides electrostatically, that is, to adsorb on the positively charged iron oxide surface below its point of zero charge (PZC). 1 Nevertheless, oleate species are also known to chemisorb on hematite at several ph units above the PZC, forming iron oleate. 12,62 Despite the fact that, at ph 8.5, the hematite surface is charged negatively (see Figure 6), adsorption of Atrac on hematite at this ph was still observed. Figure 10 shows the infrared spectra of a droplet of pure Atrac on ZnSe and Atrac adsorbed on hematite from an aqueous solution. Being a multicomponent system, Atrac presents a rather complicated infrared spectrum with several absorption bands in the 1000 and 3000 cm -1 regions (Figure 10). Several bands in the spectrum of pure Atrac (see Figure 10a) are similar to those observed in the spectrum of pure PEGMO (see Figure 7a), including CH 2 stretching vibrations at 2924 and 2855 cm -1 and CH 2 deformation at 1456 cm -1. The absorption bands observed in pure Atrac at 1736 and 1159 cm -1 (see Figure 10a) are associated with stretching vibrations of the CdO 55 and C-O 63 bonds in esters, respectively. Upon complexation of the ester group with a metal ion, ν(cdo) shifts to lower frequency, and ν(c-o) shifts to higher frequency. 64 These shifts are observed in the case of adsorption of Atrac on hematite, with ν(cdo) and ν(c-o) shifting by 14 and 16 cm -1, respectively (see Figure 10b), suggesting rather strong interaction between the ester carbonyls in Atrac and the hematite surface. The stretching vibration of the CdO bond of free carboxylic acids is observed in pure Atrac at 1709 cm -1 (see Figure 10a). 65 Upon deprotonation of the carboxylic group in solution this band disappears, and two new bands are observed at 1568 and 1427 cm -1 in the spectra of Atrac adsorbed on hematite (ν as and ν s, respectively; see Figure 10b). 66 As discussed above, adsorption of maleic acid on hematite was not observed under the experimental conditions used because of the electrostatic repulsion of the carboxylate anion and negatively charged hematite surface, suggesting that the adsorption of Atrac on hematite through carboxylate ions is unlikely and that the most probable interaction is through ester carbonyls connected by an ethoxy group. However, the carboxylate ions in the adsorbed molecules of Atrac are situated in the vicinity of the surface, and the bands originating from them can thus be found in the spectra of Atrac adsorbed on hematite. For band assignments, see Table 1. As one of the main components of Atrac is known to be ethoxylated tall oil, stretching vibrations of the C-O-C group in the ethoxy chain were expected to be observed around 1100 cm However, rather weak absorption bands were found in that wavenumber range in the spectra of both pure Atrac and Atrac adsorbed on hematite (see Figure 10), indicating that the degree of ethoxylation of the tall oil is quite low. This is further supported by the information found in the patent 47 indicating that only one ethoxy group derived from an alkylene oxide with two to four carbon atoms is found in the collector. A shorter poly(ethylene glycol) chain and the possible presence of highly nonpolar resin acids in Atrac result in a lower solubility of Atrac as compared to that of PEGMO. Prior to studying the adsorption of Atrac as a function of concentration on hematite, the adsorption as a function of concentration on an uncoated ZnSe crystal was investigated. In Figure 11, the intensity of the 2855 cm -1 band is plotted as a function of Atrac concentration in solution. The measurements were carried out in a similar way as for PEGMO. At each concentration, steady state was reached within 1 h.

75 Ind. Eng. Chem. Res., Vol. 49, No. 4, Figure 11. Intensity of the band originating from the symmetric stretching vibration (ν s) of the C-H bond as a function of the concentration of Atrac in solution. Figure 12. Intensity of the band originating from the symmetric stretching vibration (ν s) of the C-H bond as a function of the concentration of Atrac in solution (0). The solid line represents the Freundlich adsorption model fitted to the experimental data. Figure 11 illustrates that the absorbance increases with increasing concentration, but not linearly, suggesting that the recorded signal corresponds not to the bulk concentration but to the amount of Atrac adsorbed on ZnSe. The measured absorption intensities are quite low, as expected for a polished ZnSe surface with a very low surface area. Atrac was detected on the crystal even after the cell had been flushed with water for 1 h and the crystal had been dried in air, indicating a rather strong affinity for the ZnSe surface. Figure 12 shows the change in the intensity of the band originating from the symmetric stretching vibration (ν s )ofthe C-H bond during the adsorption of Atrac on hematite as a function of concentration. The measurements were performed in the same way as for PEGMO. Adsorption equilibrium was achieved at each concentration within 3-5 h. Figure 12 demonstrates that the adsorption of Atrac on the hematite surface increased with increasing concentration in solution. A poor fit of the experimental data was observed for the Langmuir model of adsorption with a coefficient of determination (R 2 ) of When plotted on a logarithmic scale, the experimental data resulted in a straight line with R 2 ) 0.99, suggesting that the Freundlich adsorption model fitted the experimental data quite well. Desorption of Atrac from hematite with time was studied in situ by flushing the cell with distilled water at two different ph values; see Figure 13. Prior to desorption, adsorption of Atrac Figure 13. Intensity of the band originating from the symmetric stretching vibration (ν s) of the C-H bond in Atrac as a function of time during the in situ adsorption of Atrac onto hematite from a 25 mg L -1 solution at ph 8.5 (Δ) and desorption by flushing with water at ph 8.5 (() and ph 10 (0). on hematite was performed in situ from a 25 mg L -1 solution at room temperature and ph 8.5 until adsorption equilibrium was reached. After the first desorption experiment at ph 8.5, Atrac was adsorbed again on the same hematite film from the same solution until adsorption equilibrium was reached. After that, desorption at ph 10 was performed. As the two adsorption curves leveled out at approximately the same absorbance values, the second adsorption curve is not shown. As illustrated by Figure 13, the absorbance was reduced rather rapidly, even when the sample was flushed with distilled water at ph 8.5 (i.e., the same ph as used during the adsorption), indicating that Atrac desorbed rather rapidly. However, the data indicate that, after 1 h of flushing, more than 50% of the Atrac still remained on the surface, suggesting that some amount of Atrac was strongly attached to the hematite surface. When the sample was flushed with distilled water at ph 10, the data indicate that Atrac desorbed more rapidly, probably because of the electrostatic repulsion between the carboxylic groups and the, at that ph, highly negatively charged hematite surface. Nevertheless, after 1 h of flushing, the data indicate that still about 40% of the originally adsorbed Atrac remained on the hematite surface. These results suggest that mild washing of the iron ore during magnetic separation and filtration in LKAB s process is not sufficient to completely remove the adsorbed flotation collector from the iron ore, especially if the ph of the water at these steps is below the flotation ph. At LKAB, the yearly average ph of water in the clarifying pond, which provides 80% of the process water, is reported to have varied in the range from 7.8 to 8.1 during the period of time from 1992 to However, the seasonal variation of ph is much higher and covers the ph range from 7.2 to 9.2, with lower values during the period of snowmelt. A decrease of ph during washing as compared to that during flotation (8.5) could lead to further adsorption of the flotation reagent on the surface of iron ore. A separate washing unit operating at a ph higher than the flotation ph before the pelletization plant could be helpful in reducing the amount of Atrac adsorbed on the magnetite surface and possibly improving green-pellet strength. Adsorption Mechanisms of Atrac and Model Compounds. As shown above, the mode of adsorption of carboxylic acids and their derivatives is determined by the tail group of the molecule and its polarity. Maleic acid did not adsorb on hematite at the ph of the experiments likely because of the electrostatic repulsion between the carboxylate anion and the negatively charged surface. Ethyl oleate showed a very weak interaction

76 1500 Ind. Eng. Chem. Res., Vol. 49, No. 4, 2010 Figure 14. Proposed adsorption geometries of (a) PEGMO and (b) Atrac on hematite from aqueous solutions at ph 8.5. R represents alkyl radical in oleic acid [CH 3(CH 2) 7CHdCH(CH 2) 7], and R represents alkyl radical in fatty acids (including oleic acid). Dashed ovals indicate the groups involved in adsorption. These groups are polar without negative ionic entities and should interact with the polar and negatively charged Fe-O surface. with hematite because of the nonpolar ester group. Both PEGMO and Atrac were found to adsorb on hematite at the experimental conditions. Based on the interpretation of the spectroscopic data, PEGMO and Atrac are proposed to adsorb as illustrated in Figure 14. PEGMO showed rather strong absorption intensities in the IR spectra when adsorbed on hematite, with significant shifts of the bands originating from the ethoxy and hydroxyl groups, which indicate that the adsorption of PEGMO probably occurs through the long ethoxy chain and the tail hydroxyl group (Figure 14a), where the latter is not likely to be deprotonated at the ph used in this work. Being a combination of ethoxylated fatty acids and maleic acid, Atrac contains ester carbonyls, ethoxy groups, and a free carboxylic group. The free carboxylic group is deprotonated in solution at ph 8.5 and is not likely to adsorb on the negatively charged hematite surface, whereas the ester carbonyls exhibited rather strong interactions with the hematite surface, as indicated by considerable shifts of the band originating from ester carbonyls in the IR spectra. Thus, the suggested mode of adsorption of Atrac on hematite is probably through ester carbonyls and the short ethoxy chain, as illustrated in Figure 14b. Conclusions Continuous and evenly distributed hematite films of controllable thickness were deposited on ATR crystals by dip-coating. It was shown that the adsorptions of both a flotation agent and the selected model compounds on such films could be monitored in situ by FTIR spectroscopy. Maleic acid does not adsorb on hematite at ph 8.5 because of the repulsion between the negatively charged hematite surface and two carboxylate anions present in maleic acid at this ph, suggesting that electrostatic interactions affect the adsorption of maleic acid on hematite. No breakage of the ester bond was observed in either the model compounds or Atrac at the experimental conditions used in this work. Ethyl oleate showed very low adsorption on hematite, suggesting that the ester carbonyl does not form strong complexes with the hematite surface. For PEGMO, the adsorption on hematite likely took place through the tail hydroxyl group accompanied by the interaction of the poly(ethylene glycol) chain with the surface. Based on the adsorption behavior of maleic acid, it was concluded that Atrac could not adsorb on hematite through deprotonated carboxylic group at the chosen experimental conditions. The most probable mode of adsorption of Atrac on the hematite surface is through the ester carbonyls and the ethoxy group. Adsorption isotherms for both Atrac and PEGMO were in good agreement with the Freundlich model of adsorption describing adsorption on energetically heterogeneous surfaces. Based on the desorption experiments, it was concluded that the strength of adsorption of Atrac on the surface of hematite varied for different species. Some of them, probably those attached directly to the surface, were strongly adsorbed and remained on the surface even when the sample was flushed with water at increased ph compared to the ph of adsorption. Other Atrac species showed rather weak interaction and could be removed from the surface by flushing with water at ph 8.5 (i.e., the same ph as during the adsorption). For PEGMO, the intensity of the band originating from the symmetric stretching vibration (ν s ) of the C-H bond in PEGMO was reduced by only 5% upon flushing of the cell with water at ph 8.5 for 1 h, suggesting a stronger interaction with hematite as compared with Atrac, probably because of the longer ethoxy chain and the presence of the tail hydroxyl group, which is not likely to be deprotonated at ph 8.5 and is thus not causing electrostatic repulsion of the molecule from the surface. Further, the results of the desorption experiments suggested that a separate washing unit after the flotation step could be beneficial in reducing the contamination of iron ore by the flotation collector and possibly improving green-pellet strength. The method developed here will be used in future works in which the effects of ionic strength, calcium ions, and silicates on the adsorption of Atrac and selected model compounds will be studied in detail. Acknowledgment The Hjalmar Lundbohm Research Center (HLRC) is gratefully acknowledged for financial support of this work. Literature Cited (1) Rao, K. H.; Forssberg, K. S. E. In Froth Flotation: a Century of Innovation; Fuerstenau, M. C., Jameson, G., Yoon, R.-H., Eds.; Society for Mining, Metallurgy, and Exploration, Inc.: Littleton, CO, 2007; p 498. (2) Fuerstenau, M. C.; Gutierrez, G.; Elgilliani, D. A. The influence of sodium silicate in non-metallic flotation systems. Trans. AIME 1968, 241, 319. (3) Gong, W.-Q. Selective flotation of Mt Weld phosphate: Methods and principles. Ph.D. Thesis, University of Western Australia, Perth, Australia, (4) Gustafsson J.-O.; Adolfsson G. Adsorption of carboxylate collectors on magnetite and their influence on the pelletizing process. In Proceedings of the XX International Mineral Processing Congress: September 1997, Aachen, Germany; Hoberg, H., von Blottnitz, H., Eds.; Gesellschaft für Bergbau, Metallurgie, Rohstoff-, und Umwelttechnik (GMDB): Clausthal-Zellerfeld, Germany, 1997, Vol. 3, p 377. (5) Su, F. Dephosphorization of magnetite fines. Ph.D. Thesis, Luleå University of Technology, Luleå, Sweden, (6) Forsmo, S. P. E.; Forsmo, S.-E.; Björkman, B. M. T.; Samskog, P.-O. Studies on the influence of a flotation collector reagent on iron ore green pellet properties. Powder Technol. 2008, 182, 444. (7) Forsmo, S. P. E.; Samskog, P.-O.; Björkman, B. M. T. A study on plasticity and compression strength in wet iron ore green pellets related to real process variations in raw material fineness. Powder Technol. 2008, 181, 321. (8) Rao, K. H.; Samskog, P.-O.; Forssberg, K. S. E. Pulp chemistry of flotation of phosphate gangue from magnetite fines. Trans. Inst. Min. Metall. C 1990, 99, 147.

77 Ind. Eng. Chem. Res., Vol. 49, No. 4, (9) Gregory, G. R. E. C. The determination of residual anionic surfaceactive reagents in mineral flotation liquors. Analyst 1966, 91, 251. (10) Giesekke, E. W. A review of spectroscopic techniques applied to the study of interactions between minerals and reagents in flotation systems. Int. J. Miner. Process. 1983, 11, 19. (11) Mielczarski, J. A.; Cases, J. M.; Bouquet, E.; Barres, O.; Delon, J. F. Nature and structure of adsorption layer on apatite contacted with oleate solution. 1. Adsorption and fourier transform infrared reflection studies. Langmuir 1993, 9, (12) Gong, W. Q.; Parentich, A.; Little, L. H.; Warren, L. J. Diffuse reflectance infrared Fourier transform spectroscopic study of the adsorption mechanism of oleate on hematite. Colloids Surf. 1991, 60, 325. (13) Kirwan, L. J.; Fawell, P. D.; van Bronswijk, W. In situ FTIR-ATR examination of poly(acrylic acid) adsorbed onto hematite at low ph. Langmuir 2003, 19, (14) Fredriksson, A.; Holmgren, A.; Forsling, W. Kinetics of collector adsorption on mineral surfaces. Miner. Eng. 2006, 19, 784. (15) Norén, K. Coordination chemistry of monocarboxylate and aminocarboxylate complexes at the water/goethite interface. Ph.D. Thesis, Umeå University, Umeå, Sweden, (16) Depalma, S.; Cowen, S.; Hoang, T.; Al-Abadleh, H. A. Adsorption thermodynamics of p-arsanilic acid on iron (oxyhydr)oxides: In-situ ATR- FTIR studies. EnViron. Sci. Technol. 2008, 42, (17) Grahn, M.; Lobanova, A.; Holmgren, A.; Hedlund, J. Orientational Analysis of Adsorbates in Molecular Sieves by FTIR/ATR Spectroscopy. Chem. Mater. 2008, 20 (19), (18) Fredriksson, A.; Larsson, M. L.; Holmgren, A. n-heptyl xanthate adsorption on a ZnS layer synthesized on germanium: An in situ attenuated total reflection IR study. J. Colloid Interface Sci. 2005, 286, 1. (19) Sukhishvili, S. A.; Granick, S. Kinetic regimes of polyelectrolyte exchange between the adsorbed state and free solution. J. Chem. Phys. 1998, 1 (09), (20) Vlasak, R.; Klueppel, I.; Grundmeier, G. Combined EIS and FTIR- ATR study of water uptake and diffusion in polymer films on semiconducting electrodes. Electrochim. Acta 2007, 52, (21) Hallinan, D. T., Jr.; Elabd, Y. A. Diffusion and sorption of methanol and water in nafion using time-resolved fourier transform infrared-attenuated total reflectance spectroscopy. J. Phys. Chem. B 2007, 111, (22) Sammon, C.; Yarwood, J.; Everall, N. A FTIR-ATR study of liquid diffusion processes in PET films: Comparison of water with simple alcohols. Polymer 2000, 41, (23) Hanh, B. D.; Neubert, R. H. H.; Wartewig, S.; Christ, A.; Hentzsch, C. Drug penetration as studied by noninvasive methods: Fourier transform infrared-attenuated total reflection, Fourier transform infrared, and ultraviolet photoacoustic spectroscopy. J. Pharm. Sci. 2000, 89, (24) Fieldson, G. T.; Barbari, T. A. Analysis of diffusion in polymers using evanescent field spectroscopy. AIChE J. 1995, 41, 795. (25) Pintar, A.; Malacea, R.; Pinel, C.; Besson, M. Catalytic three-phase diastereoselective hydrogenation of o-toluic and 2-methyl nicotinic acid derivatives: In situ FTIR/ATR investigation. Vib. Spectrosc. 2007, 45, 18. (26) Hamminga, G. M.; Mul, G.; Moulijn, J. A. Reaction kinetics and intermediate determination of solid acid catalysed liquid-phase hydrolysis reactions: A real-time in situ ATR FT-IR study. Catal. Lett. 2006, 109, 199. (27) Grahn, M.; Wang, Z.; Larsson, M. L.; Holmgren, A.; Hedlund, J.; Sterte, J. Silicalite-1 coated ATR elements as sensitive chemical sensor probes. Microporous Mesoporous Mater. 2005, 81, 357. (28) Aamouche, A.; Goormaghtigh, E. FTIR-ATR biosensor based on self-assembled phospholipids surface: Haemophilia factor VIII diagnosis. Spectroscopy 2008, 22, 223. (29) Lu, Y.; Miller, J. D. Carboxyl stretching vibrations of spontaneously adsorbed and LB-transferred calcium carboxylates as determined by FTIR internal reflection spectroscopy. J. Colloid Interface Sci. 2002, 256, 41. (30) Free, M. L.; Miller, J. D. The significance of collector colloid adsorption phenomena in the fluorite/oleate flotation system as revealed by FTIR/IRS and solution chemistry analysis. Int. J. Miner. Process. 1996, 48, 197. (31) Sperline, R. P.; Muralidharan, S.; Freiser, H. In situ determination of species adsorbed at a solid-liquid interface by quantitative infared attenuated total reflectance spectrophotometry. Langmuir 1987, 3, 198. (32) Jang, W. H.; Miller, J. D. Verification of the internal reflection spectroscopy adsorption density equation by fourier transform infrared spectroscopy analysis of transferred Langmuir-Blodgett films. Langmuir 1993, 9, (33) Kellar, J. J.; Cross, W. M.; Miller, J. D. Adsorption density calculations from in situ FT-IR/IRS data at dilute surfactant concentrations. Appl. Spectrosc. 1989, 43, (34) Grahn, M.; Holmgren, A.; Hedlund, J. Adsorption of n-hexane and p-xylene in thin silicalite-1 films studied by FTIR/ATR spectroscopy. J. Phys. Chem. C 2008, 112, (35) Grahn, M.; Lobanova, A.; Holmgren, A.; Hedlund, J. A novel experimental technique for estimation of molecular orientation in zeolite. Stud. Surf. Sci. Catal. 2007, 170, 724. (36) Wang, Z.; Grahn, M.; Larsson, M. L.; Holmgren, A.; Sterte, J.; Hedlund, J. Zeolite coated ATR crystal probes. Sens. Actuators B: Chem. 2006, 115, 685. (37) Wang, Z.; Larsson, M. L.; Grahn, M.; Holmgren, A.; Hedlund, J. Zeolite coated ATR crystals for new applications in FTIR-ATR spectroscopy. Chem. Commun. 2004, 24, (38) Fredriksson, A.; Holmgren, A. An in situ ATR-FTIR investigation of adsorption and orientation of heptyl xanthate at the lead sulphide/aqueous solution interface. Miner. Eng. 2008, 21, (39) Fredriksson, A.; Holmgren, A. An in situ ATR-FTIR study of the adsorption kinetics of xanthate on germanium. Colloids Surf. A: Physicochem. Eng. Aspects 2007, 302, 96. (40) Fredriksson, A.; Hellström, P.; öberg, S.; Holmgren, A. Comparison between in situ total internal reflection vibrational spectroscopy of an adsorbed collector and spectra calculated by ab initio density functional theory methods. J. Phys. Chem. C 2007, 111, (41) Larsson, M. L.; Fredriksson, A.; Holmgren, A. Direct observation of a self-assembled monolayer of heptyl xanthate at the germanium/water interface: A polarized FTIR study. J. Colloid Interface Sci. 2004, 273, 345. (42) Forsmo, S. P. E. Oxidation of magnetite concentrate powders during storage and drying. Int. J. Miner. Process. 2005, 75, 135. (43) Haneda, K.; Morrish, A. H. Magnetite to maghemite transformation in ultrafine particles. J. Phys. (Paris) 1977, 38 (C1), 321. (44) Gedikoglu, A. Mössbauer study of low-temperature oxidation in natural magnetite. Scr. Metall. 1983, 17, 45. (45) Wesolowski, D. J.; Machesky, M. L.; Palmer, D. A.; Anovitz, L. M. Magnetite surface charge studies to 290 C from in situ ph titrations. Chem. Geol. 2000, 167, 193. (46) Matijević, E. Production of monodispersed colloidal particles. Annu. ReV. Mater. Sci. 1985, 15, 483. (47) Swiatowski, P.; Andersen, A.; Askenbom, A. Process for the froth flotation of oxide and salt type minerals and composition. U.S. Patent 5,130,037, (48) Fuerstenau, M. C.; Elgillani, D. A.; Miller, J. D. Adsorption mechanisms in nonmetallic activation systems. Trans. AIME 1970, 247, 11. (49) Cromieres, L.; Moulin, V.; Fourest, B.; Giffaut, E. Physico-chemical characterization of the colloidal hematite/water interface: Experimentation and modeling. Colloids Surf. A 2002, 202, 101. (50) Su, C.; Suarez, D. L. In situ infrared speciation of adsorbed carbonate on aluminum and iron oxides. Clays Clay Miner. 1997, 45, 814. (51) Zeltner, W. A.; Anderson, M. A. Surface charge development at the goethite/aqueous solution interface: Effects of CO 2 adsorption. Langmuir 1988, 4, 469. (52) Bargar, J. R.; Kubicki, J. D.; Reitmeyer, R.; Davis, J. A. ATR- FTIR spectroscopic characterization of coexisting carbonate surface complexes on hematite. Geochim. Cosmochim. Acta 2005, 69 (6), (53) Schmitt, T. M. Analysis of Surfactants, 2nd ed.; Marcel Dekker, Inc.: New York, (54) Fox, J. J.; Martin, A. E. Investigations of infra-red spectra. Absorption of the CH 2 group in the region of 3μ. Proc. R. Soc. Ser. A 1938, 167, 257. (55) McMurry, H. L.; Thornton, V. Correlation of infrared spectra: Paraffins, olefins, and aromatics with structural groups. Anal. Chem. 1952, 24, 318. (56) Hartwell, E. J.; Richards, R. E.; Thompson, H. W. The vibration frequency of the carbonyl linkage. J. Chem. Soc. London 1948, (57) Cameron, D. G.; Umemura, J.; Wong, P. T. T.; Mantsch, H. H. A Fourier transform infrared study of the coagel to micelle transitions of sodium laurate and sodium oleate. Colloids Surf. 1982, 4, 131. (58) Beentjes, P. C. J.; Van Den Brand, J.; De Wit, J. H. W. Interaction of ester and acid groups containing organic compounds with iron oxide surfaces. J. Adhesion Sci. Technol. 2006, 20, 1. (59) Zeiss, H.; Tsutsui, M. The carbon-oxygen absorption band in the infrared spectra of alcohols. J. Am. Chem. Soc. 1953, 75, 897. (60) Do, D. D. Adsorption Analysis: Equilibria and Kinetics; Imperial College Press: London, (61) Hwang, Y. S.; Lenhart, J. J. Adsorption of C4-dicarboxylic acids at the hematite/water interface. Langmuir 2008, 24, (62) Kulkarni, R. D.; Somasundaran, P. Flotation chemistry of hematite/ oleate system. Colloids Surf. 1980, 1, 387. (63) Fowler, R. G.; Smith, R. M. Similarities in the infrared spectra of homologous compounds. J. Opt. Soc. Am. 1953, 43, 1054.

78 1502 Ind. Eng. Chem. Res., Vol. 49, No. 4, 2010 (64) Driessen, W. L.; Groeneveld, W. L.; Van der Wey, F. W. Complexes with ligands containing the carbonyl group. Part II. Metal(II) methyl formate, ethyl acetate and diethyl malonate solvates. Recl. TraV. Chim. Pays-Bas 1970, 89, 353. (65) Colthup, N. B.; Daly, L. H.; Wiberley, S. E. Introduction to Infrared and Raman Spectroscopy, 3rd ed.; Academic Press: London, (66) Dobson, K. D.; McQuillan, A. J. In situ infrared spectroscopic analysis of the adsorption of aliphatic carboxylic acids to TiO 2, ZrO 2,Al 2O 3, and Ta 2O 5 from aqueous solutions. Spectrochim. Acta A 1999, 55, (67) Westerstrand, M. Process water geochemistry at the Kiirunavaara iron mine, northern Sweden. Ph.D. Thesis, Luleå University of Technology, Luleå, Sweden, ReceiVed for review August 27, 2009 ReVised manuscript received December 3, 2009 Accepted December 19, 2009 IE901343F

79 PAPER II The effect of calcium ions and sodium silicate on the adsorption of anionic flotation collector on magnetite studied by ATR-FTIR spectroscopy E. Potapova, M. Grahn, A. Holmgren, and J. Hedlund Journal of Colloid and Interface Science 345 (2010)

80

81 Author's personal copy Journal of Colloid and Interface Science 345 (2010) Contents lists available at ScienceDirect Journal of Colloid and Interface Science The effect of calcium ions and sodium silicate on the adsorption of a model anionic flotation collector on magnetite studied by ATR-FTIR spectroscopy E. Potapova *, M. Grahn, A. Holmgren, J. Hedlund Division of Chemical Engineering, Luleå University of Technology, SE Luleå, Sweden article info abstract Article history: Received 25 November 2009 Accepted 14 January 2010 Available online 28 January 2010 Keywords: Magnetite Flotation collector Calcium Silicate Adsorption ATR-FTIR Previous studies have shown that agglomeration of the magnetite concentrate after reverse flotation of apatite is negatively affected by the collector species adsorbed on the surface of magnetite. In this work, the effect of ionic strength, calcium ions and sodium silicate on the unwanted adsorption of a model anionic flotation collector on synthetic magnetite was studied in situ using attenuated total reflectance Fourier transform infrared spectroscopy (ATR-FTIR). The amount of collector adsorbed was found to increase with increasing ionic strength at ph 8.5 providing evidence to the contribution of electrostatic forces to the adsorption of the collector. Adding sodium silicate to the system resulted in a threefold decrease in the amount of collector adsorbed compared to when no sodium silicate was added, confirming the depressing activity of sodium silicate on magnetite. Calcium ions were shown to increase the adsorption of both the collector and sodium silicate on magnetite. The depressing effect of sodium silicate on collector adsorption was completely suppressed in the presence of calcium ions under the conditions studied. Furthermore, the amount of collector adsorbed on magnetite from the silicate-collector solution increased 14 times upon addition of calcium ions suggesting that calcium ions in the process water may increase undesired adsorption of the collector on the iron oxide. Ó 2010 Elsevier Inc. All rights reserved. 1. Introduction Being the most commonly used metal in the world, iron is usually found as oxide minerals in nature and is extracted from iron-containing ores, of which the most common are magnetite (Fe 3 O 4 ), hematite (a-fe 2O 3) and goethite (a-feo(oh)) ores. Due to their magnetic properties, magnetite and hematite can be separated from the gangue minerals using magnetic separators [1]. However, additional upgrading steps can be required in order to reduce the amount of trace elements, such as phosphorous, to an acceptable level for the blast furnace process. Dephosphorization of the iron ore is carried out by reverse flotation with modified fatty acid based collectors [2]. Ideally, the collector should selectively adsorb on the surface of phosphorous-containing mineral rendering it hydrophobic and thus easily floated and not affecting the surface of iron oxide. In order to achieve a high recovery of phosphorous in the flotation of calcareous gangue containing minerals, such as apatite (Ca 5(PO 4) 3(F, Cl, OH)), fluorite (CaF 2) and calcite (CaCO 3), sodium silicate is widely used as a dispersant. Depending on the dosage, sodium silicate can also have a depressing effect on iron oxides. Gong et al. [3] showed that polymeric silicate species formed in concentrated solutions provided a depressing effect on iron oxide due to * Corresponding author. Fax: address: elisaveta.potapova@ltu.se (E. Potapova). the ability of the polymeric silicate species to adsorb strongly on the surface of the iron oxide, thus blocking it from collector adsorption. In another study, it was found that too high amounts of sodium silicate resulted in decreased flotation selectivity as both iron oxide and gangue minerals lost their floatability [4]. At moderate concentrations, silicate species adsorbed on the gangue mineral are replaced by the collector while the iron oxide surface is still protected from the collector adsorption, thus a selective depressing effect is achieved [5]. Low concentrations of sodium silicate (<500 g t 1 ) have not been found to prevent collector adsorption on the iron oxide surface to any greater extent [4,6,7]. In our previous study [8] it was shown that a commercial anionic fatty acid based collector readily adsorbs on hematite from aqueous solutions at ph 8.5. Furthermore, from desorption experiments, it was proven to be difficult to completely eliminate the collector adsorbed on the surface by flushing with water even at ph 10, which was higher than the adsorption ph. Besides the obvious fact that undesired adsorption of the flotation collector on the iron ore increases collector consumption in the process, it may have a negative effect on the subsequent agglomeration process where the iron ore is balled into pellets [9]. Forsmo et al. [10] showed that even small amounts of the collector added to the balling feed caused a significant decrease in the green (before thermal treatment) pellet strength. As suggested by the authors, the collector adsorbed on magnetite decreases its /$ - see front matter Ó 2010 Elsevier Inc. All rights reserved. doi: /j.jcis

82 Author's personal copy E. Potapova et al. / Journal of Colloid and Interface Science 345 (2010) wettability, resulting in the formation of stable air bubbles inside the pellet, which in turn compromises green pellet strength. Low green pellet strength, in turn, causes increased recirculation in the pelletizing plant and breakage of the pellets during drying and induration [9,11]. Apart from the flotation collector and dispersant, another parameter that may affect the flotation performance is the process water chemistry. Thepresenceofinorganicspeciesmayactivateor depress the flotation of a certain mineral, change collector solubility and the zeta-potential of the mineral surface [12]. Monovalent ions with an opposite charge compared to the charge of the surface have been shown to reduce the zeta-potential of a mineral while polyvalent ions are even capable of reversing the surface charge [13]. Previous studies indicate that high concentrations of Ca ions in the process water may affect both the flotation step [14,15] and the strength of the iron ore pellet [16,17]. During flotation, Ca ions in the process water may form precipitates with the collector and/ or adsorb on the iron oxide surface reversing its charge and thus making it more favourable for collector adsorption [14,15]. Infrared spectroscopy has been applied widely for studies of interactions in flotation systems as revealed by several comprehensive reviews [18 20]. The mechanism of fatty acid adsorption on calcareous and iron oxide minerals has been studied extensively [21 26] since 1965, when Peck and Wadsworth published their results on the adsorption of oleate on fluorite and barite [27]. The development of infrared external and internal reflection techniques provided the possibility to carry out in situ studies of collector/mineral systems, which are important since such studies provide real-time information about the complexes formed at the solid liquid interface. Our research group has been utilizing in situ ATR-FTIR spectroscopy for studying interactions of flotation collectors [28 30], sodium silicate [31,32], bentonite [33] and inorganic ions [34,35] with iron oxides as well as adsorption of hydrocarbons in zeolite films [36 39]. In our previous work on the interactions between fatty acid based collectors and iron oxides, the application of ATR-FTIR spectroscopy provided the possibility to elucidate the mechanism by which different collectors and model compounds were adsorbed on the hematite surface in the presence of water, to follow adsorption and desorption kinetics in situ and to make a conclusion about the stability of the complexes formed [8]. The objective of the present work is to study the effect of ionic strength, calcium ions and sodium silicate on the adsorption of a model flotation collector reagent on magnetite. A better understanding on how sodium silicate and the process water chemistry affect the unwanted adsorption of the collector on magnetite may lead to insights on how to minimize the adsorption of the collector on magnetite. 2. Materials and methods 2.1. Materials Magnetite nanoparticles were synthesized by co-precipitation of Fe(II) and Fe(III) according to the method described by Massart and Cabuil [40]. In short, 50 ml of an aqueous solution containing 0.33 M FeCl 24H 2O (pro analysi, KEBO) and 0.66 M FeCl 36H 2O (pro analysi, Riedel-de Haën) was added dropwise under continuous stirring to 450 ml of a 1 M solution of NH 3 (25%, Suprapur, Merck) in degassed MilliQ water. The black precipitate formed was purified by repeated sedimentation and redispersion in degassed MilliQ water until the supernatant remained turbid. The suspension was further dialyzed until the conductivity of the water outside the dialysis tube reached 1.6 lscm 1 and remained constant upon replacing it with the fresh water. The obtained suspension of fine magnetite crystals was then diluted with methanol resulting in a water-to-methanol ratio of 3:1 by volume. The dry solid content of the suspension was estimated to ca 1.2 mg ml 1 by weighing a known volume of the suspension after drying in an oven at 100 C. The suspension was stored in a refrigerator in order to minimize oxidation of magnetite. Dodecyloxyethoxyethoxyethoxyethyl maleate (Sigma Aldrich), for the sake of clarity and simplicity hereafter referred to as collector was chosen as a model flotation collector reagent. This collector was chosen based on the information given in a patent [41] describing a similar collector for froth flotation of oxide and salt type minerals, which consists of a long aliphatic hydrocarbon group connected by an alkylene oxide group to a dicarboxylic acid. The chemical structure of the collector used in this work is shown in Fig. 1. Stock solutions of the collector, sodium silicate (Na 2SiO 39H 2O, 98%, Sigma) and calcium chloride (CaCl 22H 2O, 95%, Riedel-de Haën) were prepared by dissolving appropriate amounts of the corresponding chemicals in 0.01 M aqueous solutions of NaCl (pro analysi, Riedel-de Haën), which was used as a background electrolyte. Further, working solutions were prepared by diluting appropriate amounts of the stock solutions with the 0.01 M aqueous solution of NaCl to give the required final concentrations. All aqueous solutions were prepared using distilled water degassed by applying vacuum. The distilled water used for the spectroscopic measurements was, after degassing, saturated with argon in order to minimize the amount of dissolved carbon dioxide Film deposition and general characterization of the film Prior to film deposition, the trapezoidal ZnSe crystals (Crystran Ltd.), with the dimensions mm and 45 cut edges, were first rinsed in ethanol (99.7%, Solveco chemicals AB) and distilled water. Thereafter, 0.3 ml of the magnetite suspension described above was spread evenly over one side of the ZnSe crystals and dried producing a thin film. The other side of the crystal was left uncoated in order to reduce the attenuation of the IR radiation by magnetite and thus obtain a higher throughput to the detector in the spectroscopic measurements. An X-ray diffraction pattern of the magnetite film was recorded with a Siemens D5000 powder diffractometer operating in Bragg Brentano geometry and utilizing Cu Ka radiation. In order to analyse the film, a magnetite-coated silicon wafer was mounted in a custom-made aluminium holder. The pattern of the assembly without magnetite film was also recorded in order to identify the peaks belonging to the substrate and the holder. The magnetite film on a ZnSe crystal was investigated with an FEI Magellan 400 field emission high resolution scanning electron microscope (HR-SEM) using an accelerating voltage of 1 kv. No gold coating or similar was deposited on the film to render it electrically conductive. The magnetite coated ZnSe crystal was cut in the middle and mounted vertically in the sample holder in order to measure the film thickness and investigate the morphology of the magnetite particles ATR-FTIR spectroscopy Spectral data were collected using a Bruker IFS 66v/S spectrometer equipped with a liquid nitrogen cooled mercury cadmium Fig. 1. Chemical structure of the collector. R represents the linear alkyl chain CH 3(CH 2) 11.

83 Author's personal copy 98 E. Potapova et al. / Journal of Colloid and Interface Science 345 (2010) telluride (MCT) detector. A ZnSe crystal coated with magnetite was placed in a stainless steel flow cell which was further mounted on the ATR accessory in the spectrometer (see Fig. 2). The IR beam, guided by the mirrors, enters the ZnSe through the cut edge at an angle of incidence of 45 and passes through the crystal via a number (ca 25) of total reflections. At each point of reflection, an evanescent wave of IR radiation is formed, which penetrates into the sample perpendicular to the surface of reflection and interacts with the sample material, simultaneously losing a part of its intensity. Adsorption measurements were performed in situ with a flow of the solution pumped continuously through the cell on one side of the ATR crystal at a rate of 10 ml min 1 with recirculation of the solution. The other half of the cell was evacuated. Both single beam background and sample spectra were obtained by averaging 500 scans at a resolution of 4 cm 1. Data evaluation was performed using the Bruker Opus 4.2 software. A spectrum of the pure collector was recorded in argon atmosphere by spreading a droplet of the collector over a bare ZnSe crystal; as a background, a spectrum of the bare crystal in argon was used. All adsorption and desorption experiments were performed at room temperature and ph 8.5. The ph was controlled by a Mettler Toledo T70 titrator using a 0.05 M aqueous solution of sodium hydroxide (pro analysi, Merck). The concentration of the background electrolyte was 0.01 M NaCl in all the experiments if not stated otherwise. Prior to adsorption, the magnetite film was equilibrated with 75 ml of a 0.01 M aqueous solution of NaCl at ph 8.5 for 30 min and at this point a background spectrum of the solution in contact with the magnetite coated ZnSe substrate was recorded. In the experiments where calcium ions or sodium silicate were pre-adsorbed on magnetite, a 4 mm solution of CaCl 2 or a 0.4 mm solution of Na 2SiO 3, both at ph 8.5, was pumped through the cell for 1 h. Infrared spectra were recorded every 5 min and a new background spectrum was recorded when the pre-adsorption was completed. After that, an appropriate amount of the collector was added to the solution to give a final concentration of the collector of 25 mg L 1 whereas the concentration of calcium or silicate ions was kept constant. Adsorption of the collector on magnetite was followed by recording infrared spectra every 5, 10 or 30 min. In the experiment where the influence of calcium ions on the depressing effect of sodium silicate was studied, calcium ions were first pre-adsorbed, thereafter a new background spectrum was recorded and then sodium silicate was added to the solution and the adsorption was monitored with time. After silicate adsorption, another background spectrum was recorded and finally the adsorption of the collector from a 25 mg L 1 solution was performed. Desorption experiments were carried out by flushing the cell with a 0.01 M NaCl solution at ph 8.5 without recirculation of the solution. 3. Results and discussion 3.1. Characterization of the synthetic magnetite Fig. 3 shows an HR-SEM image of a cross-section of a magnetite film on a ZnSe crystal. The image illustrates that the film thickness is about nm and the size of the individual particles vary between about 5 15 nm. Fig. 4 shows an X-ray diffraction pattern of magnetite crystals deposited on a silicon wafer. Except for the narrow diffraction peaks emanating from the silicon wafer, the XRD pattern is characteristic for randomly oriented magnetite crystals. The diffraction peaks from magnetite are quite broad, which shows that the magnetite crystals are quite small in accordance with the HR-SEM observations Effect of calcium ions on the adsorption of sodium silicate Prior to studying the combined effect of calcium ions and sodium silicate on the collector adsorption on magnetite, the influence of calcium ions on the adsorption of sodium silicate on magnetite was investigated in order to assess if calcium ions increased the polymerization of silicate species as proposed by Gong et al. [5]. Fig. 3. Side view SEM image of a magnetite film on a ZnSe crystal. Fig. 2. Schematic image of the ATR-FTIR setup. Fig. 4. XRD pattern of synthetic magnetite crystals deposited on a silicon wafer. The reflections originating from magnetite are indexed with the appropriate Miller indices. The peaks labelled with () emanate from the silicon wafer.

84 Author's personal copy E. Potapova et al. / Journal of Colloid and Interface Science 345 (2010) Silicate species adsorbed on magnetite at ph 8.5 (see spectrum (a) in Fig. 5) are characterized by a broad combination of bands between 800 and 1300 cm 1. At this ph, the most pronounced band is located at 1014 cm 1 with two shoulders at ca 1120 cm 1 and ca 950 cm 1. The band at 950 cm 1 has been assigned to the monomeric surface bidentate complex while the band at 1014 cm 1 has been assigned to the oligomeric silicate species on the surface [31]. Further, the same authors attributed the band at 1120 cm 1 to the 3-dimensional silica framework structure. With the increase of surface polymerization of the silicate species the band at 1014 cm 1 is expected to shift to higher wavenumbers [31]. Spectra recorded when calcium ions were pre-adsorbed on magnetite showed rather intense bands at 1490 and 1350 cm 1 assigned to the asymmetric and symmetric stretching vibrations, respectively, of the carbonate species [42] (see spectrum (b) in Fig. 5). This result indicates that some carbonate species were present in the water despite degassing, and further, these carbonate species adsorbed on the surface of magnetite when calcium ions were added to the solution. As sodium silicate was added to the system after pre-adsorption of calcium ions, negative bands associated with the carbonate species were observed in the spectra simultaneously as the positive bands originating from adsorbed silicate species appeared, suggesting that silicate species were replacing the carbonates on the magnetite surface (see spectrum (c) in Fig. 5). The shape of the absorption bands emanating from the silicate species was not particularly affected by the presence of calcium ions; however, the band originating from the oligomeric silicate species was slightly shifted (2cm 1 ) to higher wavenumbers indicating that the silicate species adsorbed on the magnetite surface in the presence of calcium ions were possibly polymerized to a greater extent as compared with no calcium ions present [5]. Moreover, from the intensities of the bands it may be concluded that the amount of sodium silicate adsorbed on magnetite increased by ca 35% when calcium was pre-adsorbed suggesting that electrostatic forces contributed to the interaction between silicates and magnetite. Similar results were reported by Roonasi et al. for the system calcium sulfate magnetite [34]. Calcium ions were found to increase the adsorption of sulfate on magnetite at ph 8.5 while no effect on the adsorption was observed at ph 4 indicating the importance of electrostatic interaction for the calcium/sulfate/ magnetite system and a similar behaviour could be expected for the calcium/silicate/magnetite system Collector adsorption on magnetite The collector studied in this work contains a carboxylic head group (see Fig. 1), which, depending on ph, can be deprotonated when dissolved in water. Fig. 6 shows a spectrum of the collector in pure form (non-dissolved). The strong absorption bands at 2922 and 2854 cm 1 in the spectrum are characteristic for the molecules containing long alkyl chain as these bands emanate from symmetric and asymmetric stretching vibrations (m s and m as) of the CH 2 group [22]. Stretching vibrations of the CAOAC group in the polyethylene glycol chain give rise to a characteristic intense band at 1105 cm 1 [43]. The band at 1728 cm 1 is associated with the stretching vibrations of the C@O bond [43]. As the collector molecule contains two carboxylic groups, one free and one esterified, two separate bands in the carbonyl stretching region are expected to be observed. However, when an ester carbonyl is conjugated with a C@C group (like in maleate), the band from the ester carbonyl, typically observed around 1740 cm 1, is shifted to lower wavenumbers viz. around 1725 cm 1 [44]. At the same time, intramolecular hydrogen bonding between the carboxylic groups in maleic acid may cause a shift of the C@O stretching vibration to higher wavenumbers ( cm 1 ) [44] resulting in possible overlapping with the band originating from the ester carbonyl. In our previous work, it was shown that the adsorption of anionic flotation collectors on iron oxides is not fully governed by electrostatic forces. Anionic collectors can adsorb on iron oxides despite repulsion between the negatively charged head group and the surface bearing the same net charge. The influence of electrostatic forces on the adsorption can be revealed by studying the effect of ionic strength on adsorption. Fig. 7 illustrates the effect of increased ionic strength on the intensity of one of the bands originating from the collector adsorbed on magnetite. Assuming that the band intensity is proportional to the amount of the collector adsorbed on magnetite, the data presented in Fig. 7 suggests that higher ionic strength results in a higher adsorption of the collector on magnetite due to the fact that the increased amount of sodium ions reduces the negative effective surface charge of the magnetite surface and thereby facilitating a greater collector adsorption. These results indicate that electrostatic forces contribute to the adsorption of the collector on magnetite. Fig. 5. Infrared spectrum of silicate species adsorbed on magnetite without any calcium ions present (a), spectrum of the magnetite film recorded after the preadsorption of calcium ions on magnetite (b), and spectrum of silicate species adsorbed on magnetite after pre-adsorption of calcium ions (c). Fig. 6. Infrared spectrum of a droplet of the pure collector on a ZnSe crystal in argon atmosphere.

85 Author's personal copy 100 E. Potapova et al. / Journal of Colloid and Interface Science 345 (2010) Fig. 7. Intensity of the ester C@O stretching vibrations band as a function of time during in situ adsorption of the collector on magnetite at ph 8.5 from a 25 mg L 1 solution containing 0.01 M NaCl (D) and 0.1 M NaCl (h). The ionic strength in the working solution was increased after 13 h of adsorption by adding the appropriate amount of NaCl. Moreover, as the ionic strength was changed from 0.01 M NaCl to 0.1 M NaCl it was also observed (not shown) that the band originating from the asymmetric stretching vibration of the CH 2 group in the spectra shifted from 2924 to 2922 cm 1 indicating that the interaction of the alkyl chains in the collector increased, probably due to a higher packing density of the molecules in the adsorbed layer [45]. Fig. 8 shows spectra of the collector adsorbed on the magnetite surface from aqueous solutions at ph 8.5. Compared to the spectrum of the pure collector, two new bands are observed in the spectra of the collector (spectra (a d)), viz. at ca 1570 and 1400 cm 1. These bands are assigned to the asymmetric and symmetric stretching vibrations of the carboxylate ion respectively [46] indicating the deprotonation of the head carboxylic group. However, the band originating from the stretching vibrations of the C@O bond is still present in the spectra at cm 1 suggesting that adsorption likely takes place without breaking the ester bond. When pre-adsorption of calcium ions was performed (see spectra (b) and (d) in Fig. 8), an additional band at 1425 cm 1 appeared in the spectra of the collector adsorbed on magnetite suggesting that calcium ions affected the adsorption mode of the collector on magnetite. When the effect of sodium silicate on collector adsorption was studied, the magnetite film was pre-treated with sodium silicate for 1 h before the collector was added to the system. The spectra recorded during collector adsorption (spectrum (c) in Fig. 8), in addition to the bands originating from the collector, contained an intense band at 1030 cm 1 with a shoulder at 1120 cm 1 emanating from the silicate species adsorbed on the surface of magnetite. The shoulder became evident when subtracting a spectrum of the collector adsorbed on magnetite without sodium silicate (spectrum (a) in Fig. 8) from the spectrum (c) in Fig. 8. The position of the band at 1030 cm 1 indicates higher degree of polymerization of the silicate species adsorbed on magnetite as compared to those after 1 h of adsorption prior to addition of the collector (see spectrum (a), the band at 1014 cm 1 ). These results suggest that the silicate species continued to adsorb on magnetite even in the presence of the collector implying competitive adsorption between the collector and sodium silicate for the magnetite surface sites. In the case when the magnetite film was pre-treated with both calcium ions and sodium silicate, which would be the most realistic conditions in a flotation process, no bands associated with silicate species were observed in the spectrum of the collector Fig. 8. Infrared spectra of the collector adsorbed on magnetite at ph 8.5 from a 25 mg L 1 aqueous solution containing no added Ca 2+ or Na2SiO3 (a), 4 mm Ca 2+ (b), 0.4 mm Na 2SiO 3 (c), and 4 mm Ca 2+ and 0.4 mm Na 2SiO 3 (d). All spectra were recorded after 13 h of collector adsorption. For the sake of clarity, the absorbance of the spectra (a) and (c) was multiplied with a factor 5. adsorbed on magnetite suggesting that the adsorption of sodium silicate was terminated by the collector when calcium ions were also present. It should be noted that a new background spectrum was recorded after the pre-adsorption of silicate and calcium ions. However, silicate species already adsorbed on magnetite were not substituted with the collector since no negative bands associated with the silicate species were observed in the spectra. Thereby, it may be concluded that adsorption and desorption of the silicate species were equilibrated by the addition of the collector to the system. After 13 h of adsorption from a solution of 4 mm calcium chloride and 25 mg L 1 collector at ph 8.5, an in situ desorption experiment was performed. Fig. 9 shows spectra recorded during flushing the cell with water containing 0.01 M NaCl at ph 8.5 for 2 h. As has been mentioned above, an additional band at 1425 cm 1 appeared in the spectra of the collector adsorbed on magnetite when pre-adsorption of calcium ions was performed. Upon desorption, the intensity of this band decreased much faster than the band at 1402 cm 1 suggesting the presence of two kinds of carboxylate complexes on the surface of magnetite: an innersphere complex characterized by the band at 1402 cm 1 [47] and an outer-sphere complex characterized by the band at 1425 cm 1 [47] as was previously reported by Hwang and Lenhart for maleate adsorption on hematite [47]. After 13 h of adsorption from a solution of 0.4 mm sodium silicate and 25 mg L 1 collector at ph 8.5, an in situ desorption experiment was performed. Fig. 10 shows spectra recorded during flushing the cell with water containing 0.01 M NaCl at ph 8.5 for 2 h.

86 Author's personal copy E. Potapova et al. / Journal of Colloid and Interface Science 345 (2010) Fig. 9. Spectra recorded during flushing the cell with water at ph 8.5 after collector adsorption in the presence of calcium ions. The time interval between the spectra is 10 min. The intensity of the bands emanating from the silicate species was decreasing upon flushing with water at ph 8.5. After the first 10 min of flushing, the intensity of the band at 1030 cm 1 emanating from the oligomeric silicate species was reduced as much as 40%. The intensity of this band was decreasing more slowly as the flushing continued. Furthermore, the peaks associated with the collector adsorbed on magnetite were slightly increasing in intensity at the beginning of desorption of silicate species likely due to the increasing amount of vacant surface sites available for adsorption of the collector. However, after 1.5 h of flushing with water, the bands originating from the collector also started to decrease slowly, indicating that no more empty surface sites were available for collector adsorption. Thereby, the in situ desorption experiment revealed that the collector was adsorbed on the magnetite surface much stronger than the silicate species since the absorption bands originating from the collector decreased very slowly or even increased upon flushing with water. Fig. 11 illustrates the effect of calcium ions and sodium silicate on the intensity of the band originating from the collector adsorbed on magnetite as a function of time. The data presented in Fig. 11 shows that the intensity of the ester C@O stretching vibrations band increased slowly during in situ adsorption of the collector on magnetite at ph 8.5 when no calcium or silicate ions were present in the solution (open triangles). Assuming that the band intensity is proportional to the amount of the collector adsorbed on magnetite, the data presented in Fig. 11 indicates an almost fivefold increase in the amount of the collector adsorbed on magnetite in the presence of calcium ions Fig. 10. Spectra recorded during flushing the cell with distilled water at ph 8.5 and 0.01 M NaCl after collector adsorption in the presence of silicates. The time interval between the spectra is 10 min. Fig. 11. Intensity of the ester C@O stretching vibrations band as a function of time during in situ adsorption of the collector on magnetite at ph 8.5 from a 25 mg L 1 aqueous solution containing no Ca 2+ and Na 2SiO 3 added (D), 4 mm Ca 2+ (s), 0.4 mm Na 2SiO 3 (N), 4 mm Ca 2+ and 0.4 mm Na 2SiO 3 (d). (open circles) as compared to when no calcium ions were added (open triangles), supporting the results reported earlier [14,15] and suggesting that calcium ions interact with the magnetite surface reducing the negative net charge of the surface thus making it more favourable for collector adsorption. Similarly to collector adsorption on semisoluble calcium-containing minerals like apatite, fluorite, and calcite [48], collector species can then interact specifically with calcium ions adsorbed on magnetite forming magnetite calcium-collector complexes. Additionally, calcium ions can facilitate the formation of calcium-collector precipitate, which may subsequently adsorb on the magnetite surface [9,15]. A threefold decrease in the intensity of the band originating from the collector adsorbed on magnetite was observed in the presence of sodium silicate (filled triangles) as compared to when no silicate was present (open triangles). The observed decrease is an effect of competitive adsorption between the silicate species and the collector on the magnetite surface sites as was also shown in Fig. 8 by comparing spectra (a) and (c). The results confirm that sodium silicate depresses the adsorption of the collector on magnetite in concert with previous findings [3,5]. In the experiment where both silicate and calcium ions were pre-adsorbed, the intensity of the bands originating from the collector adsorbed on magnetite was reduced by less than 8% (filled circles) as compared to the case when only calcium ions were present in the system (open circles) and was increased almost 12 times as compared to when only sodium silicate was present (filled triangles). Thereby, the depressing activity of sodium silicate was less pronounced in the presence of calcium ions, in contradiction to the results previously reported by Gong et al. [5], who observed much stronger depressing activity of the silicate calcium ion mixtures on hematite in the flotation of apatite with tall oil fatty acid as compared to that of pure sodium silicate. The reason for that was probably much higher calcium/si ratio used in the present work (10) as compared to those in the study by Gong et al. ( ). In the present study, calcium ions were found to only slightly increase the amount and polymerization degree of the silicate species adsorbed on magnetite (see Fig. 5), while they made much more significant contribution to the increase of the collector adsorption on magnetite (see Fig. 11). The results presented in this work indicate that calcium ions present in the water significantly increase the adsorption, and possibly precipitation, of the collector on synthetic magnetite. A similar effect may be expected in the industrial flotation process: contamination of the iron ore with the flotation collector may be enhanced upon high concentrations of calcium in the process water increasing collector consumption and affecting the apatite

87 Author's personal copy 102 E. Potapova et al. / Journal of Colloid and Interface Science 345 (2010) flotation as has been previously reported by Rao et al. [15]. Further, hydrophobic collector coating on the iron ore concentrate has been shown to reduce the strength of iron ore green pellets [9,11] suggesting that high concentration of calcium ions in the process water may have a negative impact on the green pellet strength by increasing adsorption and/or precipitation of the collector on the iron oxide surface. 4. Conclusions The effect of ionic strength, calcium ions and sodium silicate on the adsorption of a model flotation collector on magnetite was investigated by in situ ATR-FTIR spectroscopy. Monovalent cations were found to slightly increase the adsorption of the collector on magnetite at the studied ph by partly compensating the negative charge of the magnetite surface and thus reducing the electrostatic repulsion between the surface and carboxylate ions suggesting that electrostatic forces contribute to the adsorption of the anionic flotation collector on magnetite. Divalent calcium ions were found to have a significant effect on the adsorption of the flotation collector on magnetite. An almost fivefold increase in the amount of the collector adsorbed on magnetite was observed when calcium ions were pre-adsorbed, in concert with previous findings for anionic fatty acid based collectors. From the desorption experiments it became evident that in the presence of calcium ions, collector adsorption took place via both inner-sphere and outer-sphere complexes, the latter could be rather easily removed by flushing with water. When the magnetite film was pre-treated with sodium silicate, a competitive adsorption of the collector and sodium silicate took place on the surface of magnetite resulting in a threefold decrease in the amount of the collector adsorbed on magnetite as compared to the case when no pre-adsorption of silicates was performed confirming that sodium silicate can depress collector adsorption on magnetite. However, the stability of the magnetite-collector complexes was greater as compared to the magnetite silicate complexes. Furthermore, the depressing activity of sodium silicate on the collector adsorption was almost completely suppressed in the presence of calcium ions. Moreover, silicate adsorption on magnetite was terminated when the collector was added to the system and even though the amount and the polymerization degree of the silicate species adsorbed on magnetite in the presence of calcium ions were higher than without calcium, it was not sufficient to prevent the collector adsorption to any greater extent. The results presented in this work suggest that high concentrations of calcium in the process water may enhance collector adsorption and precipitation on iron oxides, resulting in increased collector consumption and a more hydrophobic surface. The latter has been previously shown to decrease the green pellet strength. Acknowledgments The authors acknowledge the financial support from the Hjalmar Lundbohm Research Centre (HLRC) and the Knut and Alice Wallenberg Foundation. References [1] B.A. Wills, Mineral Processing Technology. An Introduction to the Practical Aspects of Ore Treatment and Mineral Recovery, fifth ed., Pergamon Press, Oxford, [2] K.H. Rao, K.S.E. Forssberg, in: M.C. Fuerstenau, G. Jameson, R.-H. Yoon (Eds.), Froth Flotation: a Century of Innovation, Society for Mining, Metallurgy, and Exploration, Inc., Littleton, 2007, p [3] W.Q. Gong, C. Klauber, L.J. Warren, Int. J. Miner. Process. 39 (1993) 251. [4] F. Su, K. Hanumantha Rao, K.S.E. Forssberg, P.-O. Samskog, Trans. Inst. Min. Metall. C 107 (1998) 103. [5] W.Q. Gong, A. Parentich, L.H. Little, L.J. Warren, Int. J. Miner. Process. 34 (1992) 83. [6] M.C. Fuerstenau, G. Gutierrez, D.A. Elgilliani, Trans. AIME 241 (1968) 319. [7] W.Q. Gong, Selective flotation of Mt Weld phosphate: methods and principles, PhD thesis, University of Western Australia, Perth, WA, [8] E. Potapova, I. Carabante, M. Grahn, A. Holmgren, J. Hedlund, Accepted for publication in Ind. Eng. Chem. Res. 3 (in press) 5, doi: /ie901343f. [9] J.-O. Gustafsson, G. Adolfsson, in: H. Hoberg, H. von Blottnitz (Eds.), GDMB Gessellschaft für Bergbau, Metallurgie, Rohstoff- und Umwelttechnik, Clausthal-Zellerfeld, 1997, p [10] S.P.E. Forsmo, S.-E. Forsmo, B.M.T. Björkman, P.-O. Samskog, Powder Technol. 182 (2008) 444. [11] S.P.E. Forsmo, P.-O. Samskog, B.M.T. Björkman, Powder Technol. 181 (2008) 321. [12] J.A. Rajala, R.W. Smith, Trans. AIME 274 (1983) [13] H.J. Modi, D.W. Fuerstenau, J. Phys. Chem. 61 (1957) 640. [14] F. Su, Dephosphorization of magnetite fines, PhD Thesis, Luleå University of Technology, Luleå, Sweden, [15] K.H. Rao, P.-O. Samskog, K.S.E. Forssberg, Trans. Inst. Min. Metall. C 99 (1990) 147. [16] J. Engesser, Miner. Metall. Process. 20 (2003) 125. [17] S.K. Kawatra, S.J. Ripke, Int. J. Miner. Process. 72 (2003) 429. [18] E.W. Giesekke, Int. J. Miner. Process. 11 (1983) 19. [19] A.R. Hind, S.K. Bhargava, A. McKinnon, Adv. Colloid Interface Sci. 93 (2001) 91. [20] R.S.C. Smart, W.M. Skinner, A.R. Gerson, J. Mielczarski, S. Chryssoulis, A.R. Pratt, R. Lastra, G.A. Hope, X. Wang, K. Fa, J.D. Miller, in: M.C. Fuerstenau, G. Jameson, R.-H. Yoon (Eds.), Froth Flotation: a Century of Innovation, Society for Mining, Metallurgy, and Exploration, Inc., Littleton, 2007, p [21] J.A. Mielczarski, J.M. Cases, E. Bouquet, O. Barres, J.F. Delon, Langmuir 9 (1993) [22] Y. Lu, J.D. Miller, J. Colloid Interface Sci. 256 (2002) 41. [23] M.L. Free, J.D. Miller, Int. J. Miner. Process. 48 (1996) 197. [24] J.J. Kellar, W.M. Cross, J.D. Miller, Appl. Spectrosc. 43 (1989) [25] R.P. Sperline, S. Muralidharan, H. Freiser, Langmuir 3 (1987) 198. [26] W.Q. Gong, A. Parentich, L.H. Little, L.J. Warren, Colloids Surf. C 60 (1991) 325. [27] A.S. Peck, M.E. Wadsworth, in: N. Arbiter (Ed.), Seventh International Mineral Processing Congress, New York, 1965, p [28] A. Fredriksson, A. Holmgren, W. Forsling, Miner. Eng. 19 (2006) 784. [29] A. Fredriksson, A. Holmgren, Miner. Eng. 21 (2008) [30] P. Roonasi, A. Holmgren, Appl. Surf. Sci. 255 (2009) [31] X. Yang, P. Roonasi, R. Jolsterå, A. Holmgren, Colloids Surf. A 343 (2009) 24. [32] X. Yang, P. Roonasi, A. Holmgren, J. Colloid Interface Sci. 328 (2008) 41. [33] A. Holmgren, X. Yang, J. Phys. Chem. C 112 (2008) [34] P. Roonasi, A. Holmgren, J. Colloid Interface Sci. 333 (2009) 27. [35] I. Carabante, M. Grahn, A. Holmgren, J. Kumpiene, J. Hedlund, Colloids Surf. A 346 (2009) 106. [36] M. Grahn, A. Lobanova, A. Holmgren, J. Hedlund, Chem. Mater. 20 (2008) [37] M. Grahn, Z. Wang, M.L. Larsson, A. Holmgren, J. Hedlund, J. Sterte, Micropor. Mesopor. Mater. 81 (2005) 357. [38] M. Grahn, A. Holmgren, J. Hedlund, J. Phys. Chem. C 112 (2008) [39] M. Grahn, A. Lobanova, A. Holmgren, J. Hedlund, Stud. Surf. Sci. Catal. A 170 (2007) 724. [40] R. Massart, V. Cabuil, J. Chim. Phys. 84 (1987) 967. [41] P. Swiatowski, A. Andersen, A. Askenbom, US Patent 5130,037, [42] J.R. Bargar, J.D. Kubicki, R. Reitmeyer, J.A. Davis, Geochim. Cosmochim. Acta 69 (2005) [43] P.C.J. Beentjes, J. Van Den Brand, J.H.W. De Wit, J. Adhesion Sci. Technol. 20 (2006) 1. [44] N.B. Colthup, L.H. Daly, S.E. Wiberley, Introduction to Infrared and Raman Spectroscopy, third ed., Academic Press, London, [45] A. Gericke, H. Huehnerfuss, J. Phys. Chem. 97 (1993) [46] K.D. Dobson, A.J. McQuillan, Spectrochim. Acta A 55 (1999) [47] Y.S. Hwang, J.J. Lenhart, Langmuir 24 (2008) [48] Y. Lu, J. Drelich, J.D. Miller, J. Colloid Interface Sci. 202 (1998) 462.

88

89 PAPER III The effect of calcium ions, sodium silicate and surfactant on charge and wettability of magnetite E. Potapova, X. Yang, M. Grahn, A. Holmgren, S. P. E. Forsmo, A. Fredriksson, and J. Hedlund Colloids and Surfaces A: Physicochemical and Engineering Aspects 386 (2011) 79-86

90

91 Author's personal copy Colloids and Surfaces A: Physicochem. Eng. Aspects 386 (2011) Contents lists available at ScienceDirect Colloids and Surfaces A: Physicochemical and Engineering Aspects jo ur nal homep a ge: The effect of calcium ions, sodium silicate and surfactant on charge and wettability of magnetite E. Potapova a,, X. Yang a,b, M. Grahn a, A. Holmgren a, S.P.E. Forsmo c, A. Fredriksson d, J. Hedlund a a Division of Sustainable Process Engineering, Luleå University of Technology, SE Luleå, Sweden b Research Centre for Eco-Environmental Sciences, Chinese Academy of Science, Beijing , China c LKAB, SE Malmberget, Sweden d LKAB, SE Kiruna, Sweden a r t i c l e i n f o Article history: Received 31 March 2011 Received in revised form 23 June 2011 Accepted 30 June 2011 Available online 7 July 2011 Keywords: Adsorption ATR-FTIR Contact angle Silicate Surfactant Zeta-potential a b s t r a c t Anionic carboxylate surfactants and sodium silicate are used in the reverse flotation of iron ore to separate magnetite from apatite. In this work, consecutive adsorption of sodium silicate and an anionic surfactant on synthetic magnetite modified with calcium ions was studied in the ph range using in situ ATR- FTIR spectroscopy. The effect of these chemicals on the zeta-potential and wetting properties of magnetite was also investigated. While adsorption of silicate increased with increasing ph, subsequent surfactant adsorption went through a maximum at ph 8.5. Surfactant adsorption in the presence of calcium ions was not affected by the amount of silicate adsorbed on magnetite. Calcium ions were found to render the magnetite surface positive in the ph range 3 10 and could reduce the dispersing effect of silicate in flotation of apatite from magnetite. While treatment with calcium chloride and sodium silicate made magnetite more hydrophilic, subsequent adsorption of the anionic surfactant increased the water contact angle on the magnetite surface from about 10 to Although the latter values are not high enough to make magnetite float, the hydrophobic areas on the magnetite surface could result in the incorporation of air bubbles inside the iron ore pellets produced by wet agglomeration, lowering the pellet strength Elsevier B.V. All rights reserved. 1. Introduction Surfactant adsorption on mineral surfaces is important for many industrial applications including detergency, dispersion of pigments, stabilization of colloidal suspensions in cosmetics and pharmaceuticals, flocculation of fine mineral particles, and ore flotation. Separation of minerals by flotation can occur if the minerals have different affinities for air and water. A mineral can be flotated only if the work of adhesion between a mineral particle and an air bubble is high enough to prevent the disruption of the particle bubble interface. The work of adhesion between a particle and an air bubble increases with increasing contact angle at the surface air interface implying that the floatability of a mineral improves as the mineral surface becomes more hydrophobic. Most mineral surfaces are highly polar and have a high Gibbs energy, which makes them hydrophilic. To make the mineral float with the air bubbles, the surface of such minerals has to be modified by a suitable surfactant in order to reduce the Gibbs energy. Providing that surfactant Corresponding author. Tel.: ; fax: address: elisaveta.potapova@ltu.se (E. Potapova). adsorption in a flotation system occurs selectively, good separation of the minerals can be achieved. Interactions between ionic surfactants and minerals are to a large extent controlled by the charge density of the mineral surface [1]. Adsorption of different species as well as ph of the process water can alter the charge density enhancing or reducing the interactions between the mineral and the surfactant and therefore affect flotation performance. It is thus very important to know the effects that various species in the process water may have on the charge density of a mineral surface. The effect of adsorption on the charge density of mineral particles can be determined from electrophoretic measurements and then expressed in terms of changes in the zeta-potential. Another phenomenon that is highly dependent on the charge density of the mineral surface is the dispersion of the mineral particles. High dispersion is required to maximize mineral recovery and flotation selectivity [2]. Dispersion is facilitated by increasing the net surface charge density of the mineral particles, which results in increased electrostatic repulsion. Increased surface charge can be achieved by the adsorption of compounds forming charged complexes on the mineral surface. One of the important parameters in flotation that affects both the surface properties of the minerals and the speciation of the flotation chemicals as well as species naturally occurring in the pro /$ see front matter 2011 Elsevier B.V. All rights reserved. doi: /j.colsurfa