Interaction forces in bitumen extraction from oil sands

Interaction forces in bitumen extraction from oil sands

Journal of Colloid and Interface Science 287 (2005) 507–520 www.elsevier.com/locate/jcis Interaction forces in bitumen extraction from oil sands Jian...

475KB Sizes 0 Downloads 100 Views

Journal of Colloid and Interface Science 287 (2005) 507–520 www.elsevier.com/locate/jcis

Interaction forces in bitumen extraction from oil sands Jianjun Liu a , Zhenghe Xu b , Jacob Masliyah b,∗ a Mining Chemicals Research, Cytec Industries Inc., Stamford, CT 06904, USA b Department of Chemical and Materials Engineering, University of Alberta, Edmonton, AB T6G 2G6, Canada

Received 18 June 2004; accepted 15 February 2005 Available online 24 March 2005

Abstract Water-based extraction process (WBEP) has been successfully applied to bitumen recovery from Athabasca oil sand ore deposits in Alberta. In this process, two essential steps are involved. The bitumen first needs to be “liberated” from sand grains, followed by “aeration” with air bubbles. Bitumen “liberation” from the sand grains is controlled by the interaction between the bitumen and sand grains. Bitumen “aeration” is dependent, among other mechanical and hydrodynamic variables, on the hydrophobicity of the bitumen surface, which is controlled by water chemistry and interactions between bitumen and fine solids. In this paper, the interaction force measured with an atomic force microscope (AFM) between bitumen–bitumen, bitumen–silica, bitumen–clays and bitumen–fines is summarized. The measured interaction force barrier coupled with the contacted adhesion force allows us to predict the coagulative state of colloidal systems. Zeta potential distribution measurements, in terms of heterocoagulation, confirmed the prediction of the measured force profiles using AFM. The results show that solution pH and calcium addition can significantly affect the colloidal interactions of various components in oil sand extraction systems. The strong attachment of fines from a poor processing ore on bitumen is responsible for the corresponding low bitumen flotation recovery. The identification of the dominant non-contact forces by fitting with the classical DLVO or extended DLVO theory provides guidance for controlling the interaction behavior of the oil sand components through monitoring the factors that could affect the non-contact forces. The findings provide insights into megascale industrial operations of oil sand extraction.  2005 Elsevier Inc. All rights reserved. Keywords: Bitumen extraction; Oil sands; Fines; Slime coating; Clays; Colloidal interactions; Zeta potential distribution; Surface force; AFM

1. Introduction Bitumen extraction from oil sand ores in Alberta represents a megascale operation, which produces an oil supply for more than 30% of the overall Canadian oil demand at a speed of mining and processing more than 1000 tons of ore per minute currently. Considerable research efforts over the last several decades have led to the commercial application of water-based extraction process (WBEP) for recovering bitumen from oil sands [1,2]. The classical commercial method for bitumen extraction is the hot water extraction process (HWEP), which was developed by Clark in the 1920s [1,2]. Recently, a low or warm water extraction * Corresponding author.

E-mail address: [email protected] (J. Masliyah). 0021-9797/$ – see front matter  2005 Elsevier Inc. All rights reserved. doi:10.1016/j.jcis.2005.02.037

process (LWEP or WWEP) was proposed to reduce overall energy consumption [3,4]. In these processes, the following fundamental steps are involved: liberation (separation) of bitumen from sand grains; aeration (attachment or engulfment) of the liberated bitumen to air bubbles; and flotation of bitumen–air bubble aggregates to the top of slurry to form a bitumen-rich froth. Bitumen extraction from a good processing ore with HWEP is successful in a number of commercial operations and the bitumen recovery can easily exceed 93%. However, there are many technical challenges in bitumen extraction from poor processing ores. For example, low bitumen recovery or/and poor froth quality is/are often experienced when processing poor processing ores. The significant impact of divalent metal ions and mineral fines present in the slurry on oil sands ore processability has been recognized both in industrial operations and in laboratory tests [5–9]. Therefore, fundamental understanding of the colloidal inter-

508

J. Liu et al. / Journal of Colloid and Interface Science 287 (2005) 507–520

actions between bitumen–sands (silica and clays), bitumen– bitumen, and bitumen–fines, which are dominated by their interfacial properties, is of great interest to researchers and of great importance to optimizing bitumen extraction processes in industry. In earlier studies, the effects of the slurry conditioning environment such as slurry pH and temperature, and various additives such as NaOH and surfactants on bitumen extraction were extensively studied. In early 1930s, Clark and Pasternack [10] reported the necessity of sufficient amount of alkaline additives to maximize the bitumen recovery. In 1979, Sanford and Seyer [6] developed a laboratory-scale batch extraction unit (BEU) to standardize bitumen flotation test. They found that the role of NaOH is to neutralize the organic acid in bitumen and generate natural surfactants that are believed to facilitate the flotation of bitumen. Sanford [5] indicated that the maximum NaOH required was a function of fine solids. Smith and Schramm [8] showed that only a small fraction of NaOH is required to produce the needed natural surfactants, and the bulk of NaOH reacts with multivalent metal carbonate, sulfate ions and clays. Dai and Chung [11] also used model oil sands to study the effect of NaOH addition on the liberation of bitumen from sand grains and on bitumen emulsification. It was identified that a critical NaOH amount was needed to achieve an optimal effect, and overdose of NaOH would cause bitumen to emulsify, thereby resulting in small size bitumen droplets with low flotation recovery. In 1944, Clark [12] noticed the importance of surfactants that displace the oil from sand grains in the HWEP. Later, it was shown that surfactants were not always beneficial, and their presence may make the associated mineral solids floatable [13]. An optimal free surfactant concentration for bitumen recovery was observed [5,6,14–16], which corresponded well to the establishment of surface charges [17]. Kasongo et al. [18] conducted doping tests by adding a given amount of calcium and/or clays to a rich estuarine oil sands ore with a modified BEU. It was found that a sharp depression of bitumen recovery was observed only when montmorillonite clays at about 1 wt% of oil sands ore were co-added with greater than 30 ppm calcium ions in solution. Such a depression in bitumen recovery was not observed with calcium ions and kaolinite or illite. Since late 1980s, with the development of advanced optical microscope and digital video system, visualization methods were developed to observe directly the interaction nature between bitumen/sands/bubble in model systems. The most often used methods include impinging jet apparatus, induction time device, contact angle measurement, and coagulation test. These methods allowed us to study how various individual factors impacted bitumen/sands/bubble interactions. For example, Buckley et al. [19] carried out some adhesion tests and found that the adhesion of crude oil on a glass surface was dependent on pH and ionic strength. The results were explained by double-layer calculations in combination with the surface ionization model. Dai and Chung [20] performed visual pick-up test, and found that the bitu-

men and silica interactions were highly dependent on pH, particle size, temperature and solvent addition. Basu and Sharma [21] investigated with a model system of bitumen on a glass slide the effect of pH and temperature on the bitumen film pinning and recession, film rupture process and contact angle. They suggested that a pH cycle might be desirable for bitumen liberation from sand grains. In 1998, Basu et al. visually observed in a model system the effect of additives such as NaCl and MIBC/kerosene surfactants [22], and clay [23] on bitumen displacement on a glass surface by water at different pH. A thin coating of bitumen on a glass surface was found to recess spontaneously in the inward radial direction upon exposure to an aqueous environment containing additives. Zhou et al. [9] carried out settling tests using a model system to investigate the coagulation of bitumen with fine silica as a function of pH with the addition of surfactants and calcium ions. They observed a synergetic effect of various surfactants and calcium ions. In 2000, Zhou et al. [24] performed air holdup tests to study the effect of natural surfactant released from the oil sands on air holdup in a water column. Their study showed that the higher holdup and corresponding poor processability of poor processing ores could be due to the release of a larger amount of surface active species during conditioning and the presence of more fines in the ores. Ng et al. [25,26] used this method to compare two different kinds of oil sand ores by changing conditioning parameters. Moran et al. [27] examined the factors affecting the aeration of small bitumen droplets from both a surface energetic perspective and direct observations. From their study, they found that a positive spreading coefficient does not always guarantee the aeration of bitumen droplets, and suggested that such a process may best be described from a statistical stand point. Bitumen–bubble attachment in the presence of montmorillonite clays and 1 mM Ca (∼40 ppm) was also studied with an impinging jet apparatus [28,29]. The flux of gas bubbles attaching to a bitumen surface was found to decrease significantly only with a combination of montmorillonite clay and calcium. In contrast, the presence of clay and calcium alone or the combination of calcium with kaolinite showed little effect on the gas bubble flux attaching to the bitumen surface. Induction time measurement [30] was also applied to the study of interactions between bitumen and air bubbles. A similar conclusion for bitumen and air bubble attachment in the presence of montmorillonite and calcium was obtained. Although bitumen– sand grain–bubble interactions and the effect of clay addition on their interactions were extensively visualized, as noticed from the studies summarized above, there was no direct visualization for bitumen–clays and bitumen–fines interactions. Recently with the development of advanced quantification instruments such as atomic force apparatus, AFM, and surface force apparatus, SFA, quantitatively investigating the colloidal particle interactions in oil sands extraction systems becomes possible. The direct measurement of interaction forces allows us to explore the essence of interactions between the various components in a bitumen extraction sys-

J. Liu et al. / Journal of Colloid and Interface Science 287 (2005) 507–520

tem: the magnitude and nature of interaction forces. The study of interactions between a solid and model oil droplet with AFM in different systems has been reported extensively [31–35]. However, only a few studies have been published on quantitative investigation of surface forces in bitumenrelated systems. Yoon et al. [36,37], for example, used SFA, AFM and L–B (Langmuir–Bloggett) deposition techniques to investigate the effect of solution pH and temperature on the interaction forces between bitumen surfaces. In their study, a strong repulsive force considered as steric force resulting from protruding tails of asphaltene molecules was detected. The asphaltene “tails” on bitumen surface were considered to stabilize bitumen emulsions. Wu et al. [38], on the other hand, determined the colloidal forces between bitumen droplets by analyzing the droplet–droplet collision trajectories [39] and the hydrodynamic forces required to break up a bitumen doublet [40]. Later, they applied these methods to study the effect of solution pH and asphaltene content on the dynamic and static forces between bitumen droplets [41]. They found that in addition to an electrostatic double-layer repulsive force, heterogeneous protrusion of asphaltene on bitumen surface further increased the repulsive force. In such systems, the repulsion between bitumen droplets under static conditions was much weaker than under a dynamic state. These investigations provided some basis for better understanding of bitumen–bitumen interactions. Using AFM and oil dip-coating technique, Basu and Sharma [42] investigated interactions between oil (octadecane and crude oil) and mineral (glass and mica) surfaces, with the purpose of better understanding the wettability alternation in oil reservoir. To our knowledge, few studies have been reported on the quantitative investigation of the interactions between bitumen–bitumen, bitumen–silica, bitumen–clay, and bitumen–fines with the purpose of understanding poor processibility of low grade, high fine content oil sands ores. In summary, there is an unquestionable conclusion that oil sand ores containing high mineral fines (less than 44 µm) processed in a slurry containing high concentrations of divalent ions give a low bitumen recovery. Conversely, unweathered oil sand ores containing low mineral fines processed in a slurry containing low concentrations of divalent ions, would lead to a high bitumen recovery. The hypothesis is that the presence of high mineral fines coupled with high concentration of divalent ions leads to the coating of the bitumen droplets by the mineral fines. As these mineral fines are in general hydrophilic, fines-coated bitumen droplets become less hydrophobic and would not efficiently attach to air bubbles. As bitumen has a similar density as water under most industrial extraction conditions, the bitumen–air attachment process is vital to reduce the bitumen density. The poor bitumen–air attachment would lead to a low bitumen recovery. In this communication, the colloidal force and zeta potential distribution measurements will be used to show that undesirable mineral fines when coupled with high divalent ion concentration would lead to surface coating of solid fines on bitumen.

509

In this paper, the interaction colloidal force between bitumen–bitumen, bitumen–silica, bitumen–clay, and bitumen–fines measured with AFM technique is summarized. The interpreted coagulation behavior was further confirmed with zeta potential distribution measurements, a novel method developed in our research group [43–45]. We will demonstrate that the microscale fundamental studies using simple model systems provide insights into megascale industrial processes of oil sands extraction.

2. Theory for studying colloidal interaction It is well known that the adhesion force corresponds to how strong the two surfaces are attached to each other, while non-contact forces (or so-called long range forces in the literature) indicate how difficult two surfaces approach each other. For a non-sheared colloidal system, the adhesion forces between particles are not of concern since the shearing force to separate particles is negligible. The classical DLVO (Deryaguin–Landau–Verwey–Overbeek) theory was often used to predict colloidal particle interactions. It considers the summation of two force components: van der Waals force and electrostatic double-layer force. Evaluation of the electrostatic double layer and van der Waals interactions is relatively straightforward for non-sheared colloidal systems [46]. One would normally need the system parameters such as surface potentials, size and Hamaker constant of the interacting colloidal particles, as well as electrolyte composition and concentration of the solution medium. However, in some cases, the classical DLVO theory can over- or underestimate the colloidal interactions since some other forces might be present. In this case, the extended DLVO theory has to be used, which may include repulsive hydration force for hydrophilic surfaces [47–50], attractive hydrophobic force for hydrophobic surfaces [51–56], repulsive steric force and attractive bridge force for polymer bearing surfaces [57–65]. The extended DLVO theory can be expressed by Ftotal = FE + FV + FHB + FS + FHD ,

(1)

where FE is the electrostatic double-layer force, FV is van der Waals forces, FHB is the hydrophobic force, FS is the steric force, and FHD is the hydration force. The theories for describing these additional non-DLVO forces (FHB , FS , and FHD ) are less well developed. In practice, these forces are inferred from the deviation of the experimentally measured colloidal forces from those predicted by the classical DLVO theory. Some empirical equations were often used. For example, one formula describing the hydrophobic force between a sphere and a plate is given by [53] K FHB =− 2, (2) R 6D where R is the mean radius of the probe particle, D is the separation of two surfaces, K is an empirical constant, which depends on the hydrophobicity of the surfaces.

510

J. Liu et al. / Journal of Colloid and Interface Science 287 (2005) 507–520

the experiment techniques. In this regard, the colloidal force measurement is the most direct method. Although the particles of clays and fines are not regular, the trend of the measured force profiles with a number of bitumen–clays or bitumen–fines pairs is clear. The conclusions derived from the direct colloidal force measurement are further confirmed with the zeta potential distribution measurement.

3. Experimental technique 3.1. Materials

Fig. 1. Schematics of interpreted coagulation behavior between bitumen and a silica probe from the normalized repulsive barrier and adhesion force measured with AFM.

For a sheared colloidal system such as in bitumen extraction process, the DLVO or even the extended DLVO theory can still under- or over-estimate the coagulation behavior. In this case, both non-contact and adhesion forces are critical since the particles possess hydrodynamic (kinetic) energy. Comparison of non-contact, adhesion and hydrodynamic forces can predict the correlation between the measured force profiles and coagulation behavior for a dynamic colloidal system. As an example, shown in Fig. 1 are the normalized non-contact repulsive barrier and the contact adhesion force between bitumen and a silica probe as a function of solution pH. This comparison provides a theoretical region or boundary between coagulation and dispersion of a given system. In general, particles can easily coagulate when the applied force on the particles is greater than the repulsive force barrier but less than the adhesion force. On the other hand, particles would remain dispersed should the applied force be less than the repulsive force barrier but greater than the adhesion force. As discussed above, the additional non-contact interaction forces including hydrophobic force, hydration force and steric force, and the adhesion force may exist between components in an oil sands extraction system. However, the corresponding theories describing these forces have not been well developed. For this reason, one should not simply use the classical DLVO theory to evaluate interaction forces for a complex system such as in bitumen extraction. Detecting these forces and predicting the coagulation behavior for the oil sand extraction system depend, to a large extent, on

Solvent extracted bitumen sample and two types of oil sand ore samples (good processing ore and poor processing ore) were provided by Syncrude Canada Ltd. The main composition of the two oil sand samples is shown in Table 1. Silica microspheres (∼8 µm) purchased from Duke Scientific Co. (USA) were used for the preparation of silica probe and bitumen probe for the AFM measurement. Two types of clays, kaolinite and montmorillonite, were purchased from Wards’s Minerals and used without further purification. Silicon wafers of 1-0-0 crystal planes were purchased from MEMC Electronic Materials (Italy) and used as the substrate for preparation of bitumen surfaces by spin-coating method. Reagent grade HCl and NaOH (Fisher) were used as pH modifiers. Ultrahigh purity KCl (>99.999%, Aldrich) was used as the supporting electrolyte while reagent grade CaCl2 (99.9965% Fisher) was used as the source of calcium ions. Reagent grade toluene (Fisher) and in-house distilled absolute ethanol were used as the dilution and cleaning solvents, respectively. De-ionized water with a resistivity of 18.2 M cm, prepared with an Elix 5 followed by a Millipore–UV plus ultra water purification system was used throughout this study, unless otherwise specified. 3.2. Preparation of bitumen, fines, froth and process water To obtain fines, froth and process water samples for zeta potential distribution and colloidal force measurements, bitumen flotation tests were performed. Although poor processing ores are processed in practice with NaOH addition and a combination of fresh river water and recycled process water, flotation tests in this study were carried out in de-ionized water without NaOH addition for both good and poor processing ores using a modified batch extraction unit (BEU) procedure described in literature [18]. The tests performed as such, although different from industry operations, serves purpose of our study to understand the difference in

Table 1 Main composition (wt%) of oil sand samples used in this study Oil sand samples

Bitumen, %

Water, %

Solids, %

Fines (−44 µm), %

Good processing ore Poor processing ore

14.3–14.5 7.3–7.5

4–4.5 7.3–7.8

81.3–81.5 85.3–86

6 40

J. Liu et al. / Journal of Colloid and Interface Science 287 (2005) 507–520

Table 2 pH, calcium and magnesium content (A.A. analysis), and equivalent monovalent electrolyte concentration in the process water

pH Ca2+ , ppm Mg2+ , ppm Monovalent, mM

Good processing ore

Poor processing ore

7.65 0.16 0.51 3.72

7.89 8.8 4.8 4.33

processability between good and poor processing ores by studying colloidal interactions. The flotation test provides two products, a froth and a tailings sample. The froth sample is rich in bitumen and contains fine solids and water. The tailings sample is rich with coarse silica sand grains and contains fine mineral solids with some suspended bitumen droplets. The obtained froth was directly emulsified to prepare suspensions for zeta potential distribution measurements. The obtained tailings sample was allowed to settle for 30 min to remove the coarse sand grains. The supernatant collected after 30 min of settling was transferred to a centrifuge tube and centrifuged at 15,000 rpm for 30 min. The sediment in the centrifuge tube, referred to as fines, was collected for chemical analysis, zeta potential distribution measurement, and surface force measurement. To estimate the concentration of fines in the sediment, to prepare the probe particle in surface force measurement, and to characterize the fines, part of the sediment was dried in a desiccator under vacuum. The upper clear water in the centrifuge tube, on the other hand, was further filtered. The filtrate, referred to as the process water, was kept for chemical analysis and used as the probing medium in zeta potential distribution and surface force measurements. The pH of the process water obtained varied between 7.5 and 8. The calcium and magnesium concentration in the process water from the good processing ore was much lower than that from the poor processing ore, as shown in Table 2. X-ray microanalysis indicated that the fines collected from both good and poor processing ores are mainly aluminosilicate minerals. 3.3. Surface force measurement (AFM technique) Surface force measurement was conducted using a Nanoscope E AFM (Digital Instrument, Santa Barbara, CA, USA) with a vendor-supplied fluid cell. A detailed description of using an AFM to measure colloidal forces can be found elsewhere [49,66–71]. In our experiment, bitumen substrate surface was prepared by coating bitumen on a silica wafer with a P6700 spincoater (Specialty Coating Systems Inc.) [67]. The probe particles of spherical silica and pseudo-spherical clay or fines about 5–10 µm in diameter were glued with a two-component epoxy (EP2LV, Master Bound, Hackensack, NJ, USA) onto the tip of a short, wide beam AFM cantilever under an optical microscope for colloidal force measurements [67–69]. The individual clays and the fine particles were characterized by X-ray microanalysis with an energy

511

dispersive X-ray analyzer (EDX) in a scanning electron microscope (SEM) after colloidal force measurements. Bitumen probe surface was prepared by coating bitumen on a probe silica sphere with a dip-coating technique [70,71]. For the less well-defined clay or fines particles, the measurement was repeated with at least 8 bitumen–particle pairs and plotted together to show the general trend. To have a better comparison from test to test, both the measured interaction force and adhesion force (pull-off force) were normalized with the radius R of the probe particle. The normalized force profile was fitted with the classical DLVO or extended DLVO theory by assuming a constant surface charge density for the bitumen surface and a constant surface potential for mineral (silica, clays, fines) surface. A visual basic program running on EXCEL spreadsheet was developed in our laboratory for a general DLVO theory of asymmetric surfaces in symmetrical/asymmetrical electrolyte solutions [70]. The program was checked with known systems. It should be noted that the sensitivity study with current system showed that the fitted surface potentials and decay length are not a strong function of the Hamaker constant used. All the experiments were conducted at the room temperature (22 ± 0.1 ◦ C). 3.4. Zeta potential distribution measurement Zeta potential measurement was carried out with a Zetaphoremeter III (SEPHY/CAD). The bitumen was emulsified in supporting electrolyte solutions or process water to 0.01 wt% using an ultrasonic method. The mineral (clay or fines) suspension was prepared using the similar procedure. In the case of a binary bitumen–mineral mixture suspension, the emulsified bitumen was mixed with the mineral suspension at a specified ratio, and the mixture was conditioned in an ultrasonic bath (Fisher) for a few minutes before zeta potential measurements. About 40 ml of the prepared suspension was used to fill the electrophoresis cell. Through the laser-illuminating and video-viewing system, the movement of particles in the stationary layer was traced. The captured images were then analyzed by a built-in imaging processing software. The data were then converted to zeta potential as desired. The great advantage of this instrument is that it traces 50–100 particles simultaneously and provides a distribution histogram of the corresponding zeta potential. The detailed zeta potential distribution provides a good avenue to investigate the heterocoagulation phenomena in aqueous media for a two-component system. By comparing the different values of the zeta potential distributions for each component, before and after mixing the two components together, whether the two coagulate or not can be assessed. If there is no heterocoagulation, two distinct zeta potential distribution peaks at zeta potential values corresponding to the each of the individual zeta potential distribution will be detected after mixing the two components together. However, if there is heterocoagulation, only one zeta potential distribution peak is detected after mixing the two compo-

512

J. Liu et al. / Journal of Colloid and Interface Science 287 (2005) 507–520

nents. A more detailed description of this methodology can be found elsewhere [43–45,68,69].

4. Colloidal force measurements in model systems 4.1. Bitumen–bitumen interaction in the absence of fines [70,71] The “aeration” of bitumen is affected by the size of bitumen droplets as collision efficiency of bitumen droplets with air bubbles decreases with decreasing bitumen droplet size. Among the other parameters, the size of bitumen droplet in a bitumen extraction system is highly dependent on interactions between bitumen droplets, which is related to the colloidal forces between the bitumen droplets. An attractive colloidal force would facilitate coagulation of fine droplets and hence coalescence, resulting in the formation of relatively large bitumen droplets and therefore more efficient aeration. The solution pH is recognized as an important operating parameter in bitumen recovery. The effect of solution pH on the bitumen–bitumen interaction force is shown in Fig. 2. In a 1 mM KCl solution, the solution pH has a significant impact on both the non-contact and adhesion force. The non-contact interaction force changes from attractive at pH 3.5 to repulsive as the solution pH increases. At a separation of ca. 5–8 nm, the force profiles exhibit a no-

Fig. 2. Normalized interaction forces (F /R) between bitumen surfaces as a function of separation distance in 1 mM KCl solution at different solution pH. Solid lines represent the extended DLVO fitting using A131 = 2.8 × 10−21 J with the best-fitted decay length, Stern potential, hydrophobic force constant K being: pH 3.5 (square), κ −1 = 9.4 nm, ψB = −22 mV, K = 1.8 × 10−20 J; pH 5.7 (circle), κ −1 = 9.4 nm, ψB = −60 mV, K = 10 × 10−20 J; pH 8.2 (up triangle), κ −1 = 9.4 nm, ψB = −74 mV, K = 10 × 10−20 J; and pH 10.5 (down triangle), κ −1 = 9.4 nm, ψB = −80 mV, K = 5 × 10−20 J. Insert: The adhesion force and interaction force barrier as a function of pH, at a loading force of 8–10 mN/m.

ticeable jump-in motion in solutions of pH 3.5–8.2, but no such jump-in is observed in the solution of pH 10.5. With an increase in solution pH, the normalized adhesion force decreases significantly from 12 mN/m at pH 3.5 to nearly 0 mN/m at pH 10.5. The insert of Fig. 2 clearly shows that lower pH corresponded to a weaker repulsive force and a stronger adhesion force of bitumen–bitumen interaction, which is favorable for bitumen coalescence. However, it will be demonstrated in the following sections that the condition of a low pH is not favorable for bitumen liberation from a sand grain as well as a low pH condition will cause heterocoagulation between bitumen and fines. The results suggest the need for choosing a solution pH for optimal bitumen liberation from sand grains. The findings were consistent with the other reports [5,11] stating that a critical NaOH amount was needed to achieve an optimal effect of caustic addition. Overdosage of NaOH would cause bitumen to emulsify and result in the formation of small size bitumen droplets, which are unfavorable for bitumen aeration. Calcium ions are normally present as divalent ions in oil sands processing. Their presence in a processing stream would compress the electric double layer of bitumen droplets. Calcium ions may also specifically adsorb on the bitumen surface, and thereby change the surface electric properties of the bitumen. Therefore, a greater effect of calcium on the colloidal force between bitumen droplets is expected. At pH 8.2, for example, the addition of calcium ions alters both the interaction force and adhesion force between bitumen droplets in a 1 mM KCl solution at different calcium concentration. As shown in Fig. 3, the interaction force is monotonically repulsive with a jump-in at a separation of ∼5 nm. As would be expected, the addition of calcium ion depressed significantly the non-contact repulsive force. The bitumen–bitumen adhesion force was also decreased from 5.8 to 2 mN/m upon the addition of 1 mM calcium. The results indicate that calcium adsorption on the bitumen surface decreases the repulsive force barrier and the bitumen–bitumen adhesion force. The reduction in the repulsive force barrier in the presence of calcium leads us to conclude that in the absence of fines the presence of calcium ions would be desirable for bitumen droplets coalescence. However, in a commercial system, the fines are always present. In the presence of calcium, they alter the bitumen surface characteristics and hence the bitumen–bitumen interaction. The impact of fines in the presence of calcium on bitumen surface characteristics will be illustrated in the later section. 4.2. Bitumen–silica interaction [67] In a water-based extraction process (WBEP) of bitumen from oil sands, “liberation” of bitumen from sand grains and its subsequent stabilization against “heterocoagulation” with sand grains or mineral fines are prerequisite for bitumen extraction using flotation. “Liberation” and “heterocoagulation” are essentially related to the interaction forces between

J. Liu et al. / Journal of Colloid and Interface Science 287 (2005) 507–520

513

Fig. 3. Normalized interaction force (F /R) between bitumen surfaces as a function of separation distance in solution containing 1 mM KCl and different concentrations of calcium ions at pH 8.2. Solid lines represent the extended DLVO fitting using A131 = 2.8 × 10−21 J with a best-fitted decay length, Stern potential, hydrophobic force constant K being: 0 mM CaCl2 (up triangles), κ −1 = 9.4 nm, ψB = −74 mV, K = 10 × 10−20 J; 0.1 mM CaCl2 (circles), κ −1 = 8.3 nm, ψB = −41 mV, K = 6 × 10−20 J; 1 mM CaCl2 (squares), κ −1 = 4.8 nm, ψB = −31 mV, K = 3.5×10−20 J. Insert: The adhesion force and interaction force barrier as a function of calcium concentration, at a loading force of 8–10 mN/m. 1 mM calcium = 40 ppm.

Fig. 4. Interaction forces (F /R) between bitumen and silica as a function of separation distance in 1 mM KCl solution at different solution pH. Solid lines represent the extended DLVO fitting using A132 = 3.3 × 10−21 J with the best-fitted decay length, Stern potential being: pH 3.5 (square), κ −1 = 9.6 nm, ψB = −15 mV, ψS = −20 mV; pH 5.7 (circle), κ −1 = 9.1 nm, ψB = −51 mV, ψS = −41 mV; pH 8.2 (up triangle), κ −1 = 8.9 nm, ψB = −72 mV, ψS = −56 mV; and pH 10.5 (down triangle), κ −1 = 9.2 nm, ψB = −75 mV, ψS = −62 mV. Insert: The adhesion force and interaction force barrier as a function of pH, at a loading force of 8–10 mN/m.

bitumen and sand grains. Since silica sands constitute the main minerals in an oil sand ore, successful separation of bitumen from silica would greatly enhance the bitumen recovery and lead to a high quality froth. In a 1 mM KCl solution, the effect of pH on the interaction force between the bitumen and a silica particle is shown in Fig. 4. Over the pH range tested, the measured force profiles are monotonically repulsive, and the repulsion force barrier increases with increasing pH. Even at a pH of 3.5, there was still a weak repulsive force. The measured adhesion force between the bitumen and silica is shown in the insert of Fig. 4. A strong adhesion of 8 mN/m was observed at pH of 3.5. A dramatic decrease of adhesion force between the bitumen and silica occurred from pH 5.7 to 8.2, and the adhesion force disappeared eventually at pH 10.5. The pHdependent dissociation of cationic/anionic surfactants at the bitumen/water interface could be a reason for the observed variations of adhesion force with pH. Comparison of the repulsive force barrier and the adhesion force is shown in the insert of Fig. 4. The results in this figure suggest that bitumen can be easily liberated from silica sand at a pH greater than 7–8, which is consistent with other experimental observations [9,11,20]. A significant impact of calcium ions on the colloidal and adhesion forces between the bitumen and silica was expected, as calcium could compress the electric double layer and specifically adsorb on the bitumen and silica surfaces. At pH 8.2, as shown in Fig. 5, adding calcium not only

Fig. 5. Interaction forces (F /R) between bitumen and silica as a function of separation distance in solution containing 1 mM KCl and different concentrations of calcium ions at pH 8.2. Solid lines represent the extended DLVO fitting using A132 = 3.3 × 10−21 J with the best-fitted decay length, Stern potential being: 0 mM CaCl2 (square), κ −1 = 8.9 nm, ψB = −72 mV, ψS = −56 mV; 0.1 mM CaCl2 (circle), κ −1 = 8.5 nm, ψB = −45 mV, ψS = −48 mV; and 1 mM CaCl2 (up triangle), κ −1 = 4.6 nm, ψB = −35 mV, ψS = −25 mV. Insert: The adhesion force (Fad /R) as a function of calcium ion concentration, at a loading force of 8–10 mN/m.

514

J. Liu et al. / Journal of Colloid and Interface Science 287 (2005) 507–520

Fig. 6. Interaction forces (F /R) between bitumen and silica as a function of separation distance in solution containing 1 mM KCl and different concentrations of calcium ions at pH 10.5. Solid lines represent the DLVO fitting using A132 = 3.3 × 10−21 J with the best-fitted decay length and Stern potential being: 0 mM CaCl2 (square), κ −1 = 9.2 nm, ψB = −75 mV, ψS = −62 mV; 0.1 mM CaCl2 (circle), κ −1 = 7.9 nm, ψB = −45 mV, ψS = −35 mV; and 1 mM CaCl2 (up triangle), κ −1 = 4.8 nm, ψB = −35 mV, ψS = −8 mV. Insert: The adhesion force and interaction force barrier as a function of calcium ion concentration, at a loading force of 8–10 mN/m.

depressed the non-contact repulsive force, but also substantially increased the adhesion force. Comparison of the repulsive force barrier and the adhesion force shown in the insert of Fig. 5 clearly indicates that bitumen can coagulate with silica sand grains in a solution containing greater than 0.1 mM calcium ions at a solution pH 8.2. Using the DLVO theory, Takamura and Chow [73] arrived at a similar conclusion. A much more dramatic effect of calcium addition on both the colloidal and adhesion forces was observed at a pH 10.5, as shown in Fig. 6. The non-contact colloidal force changed progressively from repulsive to attractive with increasing calcium ion addition to 1 mM, while the adhesion force increased from zero to 8 mN/m. The results in the insert of Fig. 6 suggest that the bitumen can easily coagulate with silica sand grains in a solution containing greater than 0.03 mM calcium ions at a solution pH 10.5. The implication from these results is that excess caustic addition to increase pulp pH may not be always beneficial for bitumen digestion when calcium ions are present in the system. This fundamental finding provides an important justification as to why industrial scale bitumen extraction operates at a pulp pH around 8.5 as a compromise. 4.3. Bitumen–clays interaction [68] It is generally recognized that bitumen recovery reduces with increasing fines content in an oil sand ore and metal ion concentration in the process water, as the interaction

Fig. 7. Normalized interaction forces (F /R) between bitumen and clays as a function of separation distance at pH 8.2 in a 1 mM KCl solution. Solid circle: bitumen–kaolinite; open circle: bitumen–montmorillonite. Solid line represents the classical DLVO fitting using A132 = 6.5 × 10−21 J with the best-fitted decay length and Stern potential being κ −1 = 9.4 nm, ψB = −76 mV, ψC = −25 mV.

between bitumen and fines may lead to the change of the surface hydrophobicity and thereby affect its “aeration.” Most of fines in oil sands ores are clay minerals with kaolinite clays being the most representative. Although only little montmorillonite clays are present, the behavior of montmorillonite clays with their unique surface properties cannot be overlooked. As montmorillonite clays have very high charge and surface area, they become useful model compounds for the purpose of accentuating possible effects of clays on bitumen extraction processes. In a 1 mM KCl solution at pH 8.2, non-contact repulsive forces between the bitumen and kaolinite or montmorillonite were similar, as shown in Fig. 7. With the addition of 1 mM calcium, again, almost identical repulsive force profiles between the bitumen and kaolinite or montmorillonite were clearly observed. This is probably due to the similar electrokinetics for kaolinite and montmorillonite [44]. Although the non-contact repulsive force profiles between bitumen and kaolinite or montmorillonite clays were similar, a stronger adhesion force was observed between bitumen and montmorillonite than between bitumen and kaolinite, in particular when calcium ions were present, as shown in Fig. 8. This finding suggests a stronger interaction of bitumen with montmorillonite clay than with kaolinite clay. The difference becomes more pronounced when calcium ions were present. The strong attachment of montmorillonite clays on bitumen sets a barrier for bitumen droplets to coalesce with each other and to attach to air bubbles. The general observations derived from the colloidal and adhesion force measurements further confirm the hypothesis for the observed depression of bitumen flotation by montmorillonite, but not with kaolinite clay addition in the presence

J. Liu et al. / Journal of Colloid and Interface Science 287 (2005) 507–520

Fig. 8. Distribution of normalized adhesion forces (Fad /R) between bitumen and clays at pH 8.2 in (a) 1 mM KCl solution and (b) 1 mM KCl solution containing 1 mM CaCl2 at a loading force of 8–10 mN/m.

of calcium ions. The strong adhesion of montmorillonite clays with bitumen in the presence of calcium ions accounts for the observed detrimental synergetic effect of montmorillonite clays and calcium ions on bitumen recovery by flotation.

515

Fig. 9. Normalized interaction forces (F /R) between bitumen and fines collected from oil sand ore as a function of separation distance in the corresponding process water. Solid circle: bitumen–fines from good processing ore; open circle: bitumen–fines from poor processing ore. Solid line represents the classical DLVO fitting using A132 = 6.5 × 10−21 J with the best-fitted decay length and Stern potential being κ −1 = 4.6 nm, ψB = −60 mV, ψF = −30 mV. Dotted line represents the extended DLVO fitting using A132 = 6.5 × 10−21 J with the best-fitted decay length, Stern potential, hydrophobic force constant K being κ −1 = 4.3 nm, ψB = −42 mV, ψF = −28 mV, K = 10 × 10−20 J.

5. Colloidal force measurements in bitumen extraction systems [69] In earlier studies, model systems were often used to investigate the role of mineral fines and water chemistry in bitumen flotation system. In such systems, mineral clays such as montmorillonite, kaolinite and illite were used as a model sample to represent fines so that the effect of individual operating parameters on oil sand ore processibility, bitumen–bubble and bitumen–clay interactions can be studied. Although the earlier studies identified the role of model fine clays in bitumen extraction, the question remains whether these clays are representative of fines in oil sand ore. Since the real mineral fines in an oil sand ore are much more complex than single mineral clays, the role of real fines in oil sand processing needs to be validated. Therefore, it becomes necessary to study interactions between bitumen and fines derived from real oil sand ores. The mineral fines extracted from the oil sand ore were found to be aluminosilicates. Film flotation showed that the mineral fines from good processing ores are hydrophilic, whereas the mineral fines from poor processing ores are, to some extent, hydrophobic [69]. This observation leads us to conclude that there is a hydrophobic force between bitumen and fines from the poor processing ore, while such hydrophobic force between bitumen and fines from the good processing ore can be considered marginal, if not absent. The surface force profiles measured with bitumen–fines pairs in their corresponding process water are shown in

Fig. 10. Distribution of normalized adhesion forces (Fad /R) between bitumen and fines in the corresponding process water. (a) Bitumen–fines collected from good processing ore. (b) Bitumen–fines collected from a poor processing ore at a loading force of 8–10 mN/m.

Fig. 9. The interaction forces between bitumen and fines from a good processing ore are strongly repulsive. However, the interaction forces between the bitumen and fines from a poor processing ore are attractive at a separation less than 10 nm. The adhesion forces between bitumen–fines pairs in the corresponding process water are shown in Fig. 10.

516

J. Liu et al. / Journal of Colloid and Interface Science 287 (2005) 507–520

Table 3 Comparison of fitted Stern potentials from the measured force profile with the DLVO or extended DLVO theory with measured zeta potentials Bitumen–particles, medium at pH 8

Sample

Fitted Stern potentials, mV

Measured zeta potentials, mV

Bitumen–bitumen, 1 mM KCl

Bitumen

−74

−78

Bitumen–silica, 1 mM KCl

Bitumen Silica

−72 −56

−78 −60

Bitumen–kaolinite, 1 mM KCl

Bitumen Kaolinite

−76 −25

−78 −27

Bitumen–montmorillonite, 1 mM KCl

Bitumen Montmorillonite

−76 −25

−78 −26

Bitumen–fines, process water (good processing ore)

Bitumen Fines

−60 −30

−69 −42

Bitumen–fines, process water (poor processing ore)

Bitumen Fines

−42 −28

−49 −29

A stronger adhesion force between bitumen and fines from poor processing ore than between bitumen and fines from good processing ore is observed, as expected from the difference in surface wettability of the fines [69]. The findings from the measured force profiles clearly imply that the fines from poor processing ore can attach to bitumen surface more strongly than those from a good processing ore. The strong attachment of the fines on a bitumen surface in poor processing ore is responsible for bitumen coating by the fines, and consequently for the deterioration of bitumen flotation from a poor processing ore.

6. Data fitting and the nature of colloidal forces To further understand the nature of interactions measured in various systems above, the measured force profiles were compared with theoretical predictions based on either the classical or extended DLVO theory. As shown by the solid lines in Figs. 2 and 3, the measured force profiles between bitumen droplets can be described only with the extended DLVO theory by including a hydrophobic force. The best fitted Stern potentials are compared with the measured zeta potential values in Table 3. The fitted Stern potentials are slightly smaller than the measured zeta potentials, which could be well accounted for by surface roughness of the fine solids used in the force measurement. The close match between the measured zeta potential and the fitted Stern potential with the extended DLVO theory implies that the noncontact forces indeed arise from van der Waals force, electrostatic double-layer force and hydrophobic force. Since van der Waals force is not sensitive to the environmental condition, electrostatic double-layer force and hydrophobic force are collectively the dominant forces. Considering a quite irregular shape with an enormous surface roughness of fines/clays, only semi-quantitative trend was considered significant in this study. Through monitoring the factors that can affect the dominant forces, we can control the bitumen–

bitumen coagulation behavior in an aqueous solution. The presence of a hydrophobic force between bitumen droplets exhibiting a water contact angle value of 70◦ [71] is not unexpected. The K value at contact angle of about 70◦ is 10 × 10−20 J, which is much greater than the Hamaker constant for van der Waals force (see figure caption). Yoon et al. [72] obtained a similar value (7–20 × 10−20 J) for a pair of silanated silica surfaces with a contact angle of 81◦ . The fitted hydrophobic force constant K decreases with increasing solution pH or with calcium ion addition, suggesting that increasing solution pH or calcium ion addition reduces the hydrophobicity of bitumen surfaces. The reduction in hydrophobicity of bitumen surfaces is anticipated to reduce bitumen aeration efficiency. For the systems of bitumen interacting with silica, the measured force profiles at separations greater than about 2–4 nm can be reasonably fitted with the classical DLVO theory, as shown by the solid lines in Figs. 4 and 5. More importantly, the fitted Stern potentials agree exceptionally well with the measured zeta potential values for both silica and bitumen with the fitted Debye lengths being in the range of the values calculated from the actual electrolyte concentrations in the medium. The good fit and excellent agreement suggest that the colloidal force between the bitumen and silica is predominantly due to the overlap of the electric double layers. Recognizing the electrostatic nature of non-contact colloidal forces guides us to control the coagulation behavior of bitumen with silica through monitoring factors such as pH, salinity and divalent ions that can affect surface charge and electric double-layer structures. In the case of montmorillonite and kaolinite clays, the measured force profiles between the bitumen and clay particles are highly scattered due to the variations in surface roughness of the clay particles and uncertainty in measuring clay particle sizes. However, a general trend is clear for each condition, as shown in Fig. 7. The fitting procedures using the classical DLVO theory were applied to the average force profiles for each condition. A good fit between the

J. Liu et al. / Journal of Colloid and Interface Science 287 (2005) 507–520

experimental results and the theoretical predications is seen by the fitted curves running through the middle of respective scattered force profiles, as shown in Fig. 7. The good agreement between the fitted Stern potential and the measured zeta potential as shown in Table 3 implies that the electric double-layer force dominates the interaction forces between the bitumen and clay particles. It is anticipated that any changes that cause an increase in the electrostatic repulsion between the bitumen and clays or a decrease in adhesion forces by a dispersant addition, such as sodium silicates or phosphates, would mitigate the depression of bitumen recovery in flotation. For the measurements performed with fines from a good processing ore, the obtained average force profiles can be well described with the classical DLVO theory, as shown by the solid curve in Fig. 9. In contrast, the measured average force profiles between the bitumen and fines from the poor processing ores can only be fitted with the extended DLVO theory by considering the presence of a hydrophobic force. Considering a hydrophobic nature of fines from poor processing ores [69], the presence of the hydrophobic force between the hydrophobic bitumen and fines from the poor processing ores is not unexpected. The excellent agreement between the measured zeta potentials and the fitted Stern potentials with the DLVO theory for the fines from the good processing ore and the extended DLVO theory for the fines from the poor processing ore indicates that the interaction forces between the bitumen and fines from the good processing ores are predominantly controlled by the electric doublelayer force while the interaction forces between the bitumen and fines from the poor processing ore are controlled by both the electric double-layer force and hydrophobic force. More importantly, the current work provides the directions to resolve the difficulties of handling poor processing ores. The key issue is to reduce the hydrophobicity of fines or to significantly increase the electrostatic repulsion in order to prevent the fines from heterocoagulating with bitumen. A few potential approaches to achieve this objective are: (a) to decrease the hydrophobicity of fines by masking the surfactants on the fines surface with polymers, or stripping of adsorbed surfactants on the fines surface with NaOH addition; (b) to increase electrostatic repulsive force between bitumen and fines by dispersing chemical, such as sodium silicate and polyphosphates; and (c) to eliminate the effect of calcium ions with chelating or precipitating chemicals such as modified dextrin or bicarbonates. This is exactly what the addition of NaOH in the commercial processing accomplishes.

7. Zeta potential distribution measurement The current study clearly shows that direct colloidal force measurement can identify the coagulative nature of oil sand components. More importantly, it provides insights regarding the nature of interaction forces. Since the measurement with AFM was performed largely using single particles, even

517

Fig. 11. Zeta potential distribution of emulsified bitumen and kaolinite suspension in 1 mM KCl solution containing 1 mM CaCl2 at pH 8.2, measured (a) separately and (b) as a mixture.

though a number of particles were used in the measurement, its generalization to a real oil sand extraction system remains a concern. To further confirm the conclusions derived from the force profiles and adhesion forces measured with AFM, the zeta potential distributions of oil sand components, individually and as a mixture, were measured and the results were interpreted in terms of colloidal interactions. For illustrative and confirmative purposes, only the results with bitumen–clay [44] and bitumen–fines [69] are presented here. 7.1. Bitumen–clay model system [44] The zeta potential distributions measured with bitumen droplets or kaolinite particles alone in a 1 mM KCl solution containing 1 mM calcium ions, were centered at −38 and −10 mV, respectively, as shown in the overlaid histogram of Fig. 11a. A similar bimodal zeta potential distribution histogram was obtained for the mixture of the bitumen and kaolinite clay particles, as shown in Fig. 11b. The presence of two distinct distribution peaks at −32 and −10 mV indicates that bitumen droplets and kaolinite particles in the mixture are non-coagulative, i.e., they are present separately as individual particles, as schematically illustrated in the insert of Fig. 11b. This observation is anticipated for a system of a weak adhesion force, as shown in the AFM force measurement (Fig. 8). For the bitumen droplets and montmorillonite particles, the zeta potential distributions measured individually in a 1 mM KCl solution containing 1 mM calcium ions at pH 8.2 are shown in Fig. 12a. Bitumen droplets and montmorillonite clay particles each exhibited a zeta potential distribution peak centered at −38 and −9 mV, respectively. However, when zeta potential distribution was measured with the mixture of the bitumen and montmorillonite clays, the results

518

J. Liu et al. / Journal of Colloid and Interface Science 287 (2005) 507–520

Fig. 12. Zeta potential distribution of emulsified bitumen and montmorillonite suspension in 1 mM KCl solution containing 1 mM CaCl2 at pH 8.2, measured (a) separately and (b) as a mixture.

Fig. 13. Zeta potential distributions of (a) individual emulsified bitumen and fine suspension, (b) their mixture, and (c) the corresponding froth collected from a good processing ore in the corresponding process water.

in Fig. 12b showed only a single distribution peak with the zeta potential value close to that for the montmorillonite clays. The disappearance of the original distribution peak of bitumen suggests that all the bitumen droplets were fully covered by montmorillonite particles to form composite aggregates, as schematically shown in the insert of Fig. 12b. This behavior indicates a strong attraction between the bitumen and montmorillonite clays, as anticipated for a system having a strong adhesion force as measured with the AFM (Fig. 8). 7.2. Bitumen–fines in a bitumen extraction system [69] 7.2.1. Fines from good processing ore The zeta potential distributions of solvent-extracted bitumen and mineral fines from good processing ore were measured separately in the corresponding process water. The results in Fig. 13a show two distinct distribution peaks at −71 and −42 mV, corresponding to bitumen and fines, respectively. For the mixture of the bitumen and fines, a bimodal distribution histogram in Fig. 13b is observed. The peak values of the zeta potential distribution histogram correspond to those for the bitumen and fines, illustrating a negligible attraction between the two components in a real flotation system. Measurement with emulsified froth in the process water also shows two distinct distribution peaks located at zeta potential values corresponding to these for the bitumen and fines, as shown in Fig. 13c. The results confirm the negligible attraction between the fines and bitumen. The observation suggests that the fine solids in the froth of good processing ores are likely the result of carry-over by mechanical entrainment/entrapment and not due to bitumen– fines attachment. The measured strong repulsive force and weak adhesion force (Figs. 9 and 10) between the bitumen and fines collected from good processing ores accounts for

Fig. 14. Zeta potential distributions of (a) individual emulsified bitumen and fine suspension, (b) their mixture, and (c) the corresponding froth collected from a poor processing ore in the corresponding process water.

the negligible slime coating of the fines on bitumen during flotation. The general findings here would suggest a high bitumen flotation rate from good processing ores as often is the case in practice. 7.2.2. Fines from poor processing ore The zeta potential distributions of bitumen droplets and mineral fines from poor processing ores were also measured separately in the corresponding process water. The results in Fig. 14a show that the distribution for individual emulsified bitumen and fines suspension was present by two peaks at −50 and −29 mV, respectively. Compared with Fig. 13a, a reduction of zeta potential values for both bitumen and fines is observed as anticipated from the compression of electric double layer by the presence of calcium and mag-

J. Liu et al. / Journal of Colloid and Interface Science 287 (2005) 507–520

nesium ions in the process water from a poor processing ore (see Table 2). The zeta potential distribution of Fig. 14b for the mixture of the bitumen and fines shows a distribution peak at a zeta potential value corresponding to that for the fines, with a small tail spreading towards the bitumen distribution peak. Comparing the results for the fines from the good processing ore as shown in Fig. 12b, the results here suggest a stronger attachment between the bitumen and mineral fines derived from the poor processing ore in the corresponding process water. Measurement with the emulsified froth shown in Fig. 14c gave only one narrow distribution peak at the same zeta potential value as that for the fines, indicating a strong attachment of fines on bitumen surface. The measured non-contact attractive force and strong adhesion force shown in Figs. 9 and 10, respectively, explains the observed strong attachment. The results imply a stronger coagulation of fines from a poor processing ore with bitumen in a real flotation system. Such a heterocoagulation leads to a harmful slime coating. Therefore, a low bitumen flotation rate and poor froth quality are anticipated for poor processing ores. Indeed, this is a situation as experienced in industrial operations.

8. Conclusions • The microscale colloidal force measurement starting with a model system and gradually progressing to a real system using AFM provided insights into the megascale industrial operations of oil sand extraction. The measured force profiles can be used to predict the colloidal interaction behavior between the oil sand components. • The interactions between oil sand components are significantly impacted by solution pH and calcium ion addition. The increase of solution pH can lead to a weak attraction between bitumen–bitumen and bitumen–silica. While addition of calcium ions can increase the attraction between bitumen–bitumen, bitumen–silica, bitumen–clay, and bitumen–fines to a different degree. The predication of colloidal interactions from the measured force profiles is consistent with the interpretation of zeta potential distribution measurement. • The non-contact interaction force profiles can be reasonably described with the classical or extended DLVO theory. The electric double-layer force dominates the noncontact force for systems of bitumen–silica, bitumen– clays, and bitumen–fines from good processing ores, while both the electric double-layer force and hydrophobic force dominate the non-contact force for systems of bitumen–bitumen and bitumen–fines from poor processing ores. This finding should be used to guide the control of the particle interaction behavior through monitoring factors, such as solution pH and divalent ions in the process water, which can affect the dominating noncontact forces.

519

• The results support the hypothesis regarding the processability of oil sand ores and demonstrate the important role of colloidal forces in bitumen recovery from oil sand ores. For a good processing ore, low content mineral fines with lower concentration of divalent ions in the ore can only weakly attach on the bitumen surface. This leads to a higher bitumen flotation recovery. For a poor processing ore, the higher content mineral fines with higher concentration of divalent ions in the ore can cause the fines to strongly attach to the bitumen surface. This accounts for a low bitumen recovery.

Acknowledgments The authors acknowledge the financial support from COURSE with Syncrude Canada Ltd. and Albian Sands Inc. as industrial partners, Natural Sciences and Engineering Research Council of Canada (NSERC), and NSERC Industrial Research Chair in Oil Sands Engineering. The provision of oil sand samples by Syncrude Canada Ltd. is also acknowledged.

References [1] L.G. Hepler, C. Hsi, AOSTRA Technical Handbook on Oil Sands, Bitumen and Heavy Oils, AOSTRA Technical Publication Series, vol. 6, Alberta Oil Sands Technology and Research Authority, Edmonton, AB, 1989. [2] L.G. Hepler, R.G. Smith, The Alberta Oil Sands: Industrial Procedures for Extraction and Some Recent Fundamental Research, AOSTRA Technical Publication Series, vol. 14, Alberta Oil Sands Technology and Research Authority, Edmonton, AB, 1994. [3] K.N. Sury, U.S. Patent 4,946,597, 1990. [4] J. Spence, R. Siy, K. Sury, K. McDowell S. Ng, Aurora Extraction Process: 1995 OSLO Cold Water Extraction Research Pilot Test Program, Syncrude Research Report 96-3, June 1996. [5] E.C. Sanford, Can. J. Chem. Eng. 61 (1983) 554. [6] E.C. Sanford, F.A. Seyer, CIM Bull. 72 (1979) 164. [7] K. Takamura, D. Wallace, J. Can. Pet. Technol. 27 (1988) 98. [8] R.G. Smith, L.L. Schramm, Fuel Proc. Technol. 30 (1992) 1. [9] Z.A. Zhou, Z. Xu, J.H. Masliyah, J. Czarnecki, Colloids Surf. A 148 (1999) 199. [10] K.A. Clark, D.S. Pasternack, Ind. Eng. Chem. 24 (1932) 1410. [11] Q. Dai, K.H. Chung, Fuel 75 (1996) 220. [12] K.A. Clark, Trans. Can. Inst. Min. Met. 47 (1944) 257. [13] M.V. Baptista, C.W. Bowman, The flotation mechanism of solids from the Athabasca oil sand, in: 19th Can. Chem. Eng. Conf., Edmonton, AB, 1969, preprint. [14] L.L. Schramm, D.S. Smith, J.A. Stone, AOSTRA J. Res. 1 (1984) 5. [15] L.L. Schramm, D.S. Smith, J.A. Stone, Colloids Surf. 11 (1984) 247. [16] L.L. Schramm, R.G. Smith, Can. J. Chem. Eng. 65 (1987) 799. [17] L.L. Schramm, R.G. Smith, Colloids Surf. 14 (1985) 67. [18] T. Kasongo, Z. Zhou, Z. Xu, J. Masliyah, Can. J. Chem. Eng. 78 (2000) 674. [19] J.S. Buckley, K. Takamura, N.R. Morrow, SPE Reservoir Eng. (August 1989) 332–340. [20] Q. Dai, K.H. Chung, Fuel 74 (1995) 1858. [21] S. Basu, M.M. Sharma, J. Colloid Interface Sci. 181 (1996) 443. [22] S. Basu, K. Nandakumar, J. Masliyah, Colloids Surf. A 136 (1998) 71.

520

J. Liu et al. / Journal of Colloid and Interface Science 287 (2005) 507–520

[23] S. Basu, W.C. Kanda, K. Nandakumar, J. Masliyah, Ind. Eng. Chem. Res. 37 (1998) 253. [24] Z. Zhou, Z. Xu, J. Masliyah, Can. J. Chem. Eng. 78 (2000) 617. [25] S. Ng, P. Warszynski, M. Zembala, K. Malysa, Miner. Eng. 13 (2000) 1505. [26] S. Ng, P. Warszynski, M. Zembala, K. Malysa, Miner. Eng. 13 (2000) 1519. [27] K. Moran, A. Yeung, J. Czarnecki, J. Masliyah, Colloids Surf. A 174 (2000) 147. [28] C. Yang, T. Dabros, D.Q. Li, J. Czarnecki, J.H. Masliyah, Ind. Eng. Chem. Res. 39 (2000) 4949. [29] C. Yang, Ph.D. thesis, Univ. of Alberta, Edmonton, AB, 2000. [30] G. Gu, Z. Xu, K. Nandakumar, J.H. Masliyah, Int. J. Miner. Process. 69 (2003) 235–250. [31] P. Mulvaney, J.M. Perera, S. Biggs, F. Grieser, G.W. Stevens, J. Colloid Interface Sci. 183 (1996) 614. [32] B.A. Synder, D.E. Aston, J.C. Berg, Langmuir 13 (1997) 590. [33] D.E. Aston, J.C. Berg, J. Pulp Paper Sci. 24 (1998) 121. [34] P.G. Hartley, F. Grieser, P. Mulvaney, G.W. Stevens, Langmuir 15 (1999) 7282. [35] Y.I. Rabinovich, M.S. Esayanur, K.D. Johanson, J.J. Adler, B.M. Moudgil, J. Adhes. Sci. Technol. 16 (2002) 887. [36] R.H. Yoon, D. Guzonas, B.S. Aksoy, J. Czarnecki, A. Leung, in: Processing of Hydrophobic Minerals and Fine Coal: Proceeding of the 1st UBC–McGill Bi-Annual International Symposium, 1995, pp. 277– 289. [37] R.H. Yoon, Y.I. Rabinovich, in: Polymers in Mineral Processing: Proceeding of the 3rd UBC–McGill Bi-Annual International Symposium, CIM, Montreal, PQ, 1999. [38] X. Wu, J. Czarnecki, N. Hamza, J.H. Masliyah, Langmuir 15 (1999) 5244. [39] X. Wu, T. Dabros, J. Czarnecki, Langmuir 15 (1999) 7806. [40] X. Wu, I. Laroche, J.H. Masliyah, J. Czarnecki, T. Dabros, Colloids Surf. A 174 (2000) 133. [41] I. Laroche, X. Wu, J.H. Masliyah, J. Czarnecki, T. Dabros, J. Colloid Interface Sci. 250 (2002) 316. [42] S. Basu, M.M. Sharma, J. Colloid Interface Sci. 181 (1996) 443. [43] J. Liu, Z. Zhou, Z. Xu, Ind. Eng. Chem. Res. 41 (2002) 52. [44] J. Liu, Z. Zhou, Z. Xu, J. Masliyah, J. Colloid Interface Sci. 252 (2002) 409.

[45] Z. Xu, J. Liu, J. Choung, Z. Zhou, Int. J. Miner. Proc. 68 (2003) 183. [46] J. Masliyah, Intensive Short Course: Extraction of Oilsands Bitumen, Univ. of Alberta, Edmonton, AB, 2003. [47] L. Meagher, J. Colloid Interface Sci. 152 (1992) 293. [48] A. Grabbe, R.G. Horn, J. Colloid Interface Sci. 157 (1993) 375. [49] W.A. Ducker, T.J. Senden, R.M. Pashley, Nature 353 (1991) 239. [50] S. Nishimura, M. Kodama, H. Noma, K. Inoue, H. Tateyama, Colloids Surf. A 143 (1998) 1. [51] J.N. Israelachvili, M. Tirrel, J. Klein, Macromolecules 17 (1984) 204. [52] P.M. Claesson, C.E. Blom, P.C. Herder, B.W. Ninham, J. Colloid Interface Sci. 114 (1986) 234. [53] Y.I. Rabinovich, R.H. Yoon, Langmuir 10 (1994) 1903. [54] S.A. Ravishankar, R.H. Yoon, Miner. Metall. Proc. 14 (1997) 10. [55] R.H. Yoon, D.H. Flinn, Y.I. Rabinovich, J. Colloid Interface Sci. 185 (1997) 363. [56] V.S.J. Graig, B.W. Ninham, R.M. Pashley, Langmuir 15 (1999) 1562. [57] J. Klein, Nature 288 (1980) 248. [58] J. Klein, P.E. Luckham, Macromolecules 17 (1984) 1041. [59] J.N. Israelachvili, R.M. Pashley, J. Colloid Interface Sci. 98 (1984) 500. [60] Y. Almog, J. Klein, J. Colloid Interface Sci. 106 (1985) 33. [61] J. Marra, M.L. Hair, Macromolecules 21 (1988) 2356. [62] P.G. de Gennes, Adv. Colloid Interface 27 (1987) 189. [63] P. Pincus, Macromolecules 24 (1991) 2912. [64] S.J. O’Shea, M.E. Welland, T. Rayment, Langmuir 9 (1993) 1826. [65] H.J. Butt, M. Kappl, H. Mueller, R. Raiteri, Langmuir 15 (1999) 2559. [66] S. Veeramasuneni, M.R. Yalamanchili, J.D. Miller, J. Colloid Interface Sci. 184 (1996) 594. [67] J. Liu, Z. Xu, J. Masliyah, Langmuir 19 (2003) 3911. [68] J. Liu, Z. Xu, J. Masliyah, AIChE J. 50 (2004) 1917. [69] J. Liu, Z. Xu, J. Masliyah, Can. J. Chem. Eng. 82 (2004) 655–666. Note: the entire volume is devoted to oil sands. [70] J. Liu, Ph.D. thesis, Univ. of Alberta, Edmonton, AB, 2004. [71] J. Liu, Z. Xu, J. Masliyah, Colloids Surf. A (2005), in press. [72] R.H. Yoon, D.H. Flinn, Y.I. Rabinovich, J. Colloid Interface Sci. 185 (1997) 363. [73] K. Takamura, R.S. Chow, J. Can. Pet. Technol. 22 (1983) 22.