- Email: [email protected]
Precipitation of dodecyl amine in KCl–NaCl saturated brine and attachment of amine particles to KCl and NaCl surfaces
Int. J. Miner. Process. 93 (2009) 34–40
Contents lists available at ScienceDirect
Int. J. Miner. Process. j o u r n a l h o m e p a g e : w w w. e l s ev i e r. c o m / l o c a t e / i j m i n p r o
Precipitation of dodecyl amine in KCl–NaCl saturated brine and attachment of amine particles to KCl and NaCl surfaces E. Burdukova a, J.S. Laskowski a,⁎, G.R. Forbes b a b
N.B. Keevil Institute of Mining Engineering, University of British Columbia, Vancouver, B.C., Canada Department of Materials Engineering, University of British Columbia, Vancouver, B.C., Canada
a r t i c l e
i n f o
Article history: Received 17 February 2009 Accepted 13 May 2009 Available online 19 May 2009 Keywords: Dodecyl amine Krafft point Amine precipitation Sylvite Halite Potash ore Potash ore flotation
a b s t r a c t Long-chain amines, used in potash ore flotation as collectors, are insoluble in NaCl–KCl saturated brine. In commercial applications, these amines are melted at 70–90 °C, dispersed in acidic solution of hydrochloric or acetic acids, and such emulsions are then introduced to the flotation pulp. To model the commercial potash ore flotation process, dodecyl amine, used in this study, was melted at 70 °C, dispersed in hydrochloric acid aqueous solution and was added to a KCl–NaCl saturated brine at room temperature. This results in the precipitation of the amine. The present study summarizes the influence of the conditions on the particle size and morphology of the precipitating amine particles. Methyl isobutyl carbinol (MIBC), common frother in flotation processes, was shown to affect amine dispersion when added into a hot amine emulsion prior to mixing with a saturated brine. This study demonstrates that the precipitating amine particles are selectively abstracted by KCl particles, but not by NaCl particles. © 2009 Elsevier B.V. All rights reserved.
1. Introduction Sylvite (KCl) and halite (NaCl), two main components of sylvinite (potash) ores, are commonly separated by flotation. Since both these minerals are soluble in water the process is carried out in NaCl–KCl saturated brine. Long-chain primary amines are utilized as a collector to float selectively sylvite. Since potash ores also contain a few percent of water-insoluble minerals, which in this process appear in the form of slimes, the process critically depends on desliming. At 20 °C, 1.450 kg of the NaCl–KCl saturated solution contains about 0.300 kg of NaCl and 0.150 kg of KCl and 1 kg of water (Gaska et al., 1965). Then, in the saturated NaCl–KCl brine there is about 5.1 kmol of NaCl and 2 kmol of KCl per 55.5 kmol of water. This, assuming complete NaCl and KCl dissociation, gives for each ion 3.9 molecules of water. It is thus obvious that in the saturated brine, all water molecules are bound in hydration shells around hydrated ions. In the potash flotation process the collector molecules are then transferred from an electrolyte to the interface whose composition does not essentially differ from that in the solution (Schubert, 1988). The question thus arises what is a driving force for the transfer of the collector to the interface. Since chemisorption must be ruled out in this system, the conclusion is that the major component of the adsorption energy is the energy of association of hydrocarbon chains.
⁎ Corresponding author. E-mail address: [email protected] (J.S. Laskowski). 0301-7516/$ – see front matter © 2009 Elsevier B.V. All rights reserved. doi:10.1016/j.minpro.2009.05.001
As discussed by Schubert (1967, 1988) this energy differs only slightly from the precipitation energy of the collector molecules within the liquid phase. There is a huge amount of evidence that indicates that collector selective adsorption is a main reason for selectivity of flotation. However, these results come from the research carried out with the low ionic strength systems. But even in such systems a quite different mechanism is also possible. In the emulsion flotation, in which “oily” water-insoluble collectors are utilized, the selectivity of the process depends on the selective attachment of the oil droplets to mineral surfaces (Klassen and Mokrousov, 1963). So, in this case the process is not accompanied by diffusion of collector molecules/ions and selective adsorption onto given mineral surfaces, but by a transport of dispersed oil droplets towards mineral surfaces. The selectivity in such a process critically depends on the initial wetting of the floated minerals since the oil droplets can attach only to hydrophobic surfaces. By the way, this is a reason why the emulsion flotation is utilized in processing coals, molybdenite, graphite, and other inherently hydrophobic minerals. The fact that wettability of different water-soluble minerals in brines can be different was already taken into consideration by Rogers (1956/57) and was further substantiated by Hancer et al. (2001), and Burdukova and Laskowski (2009). As indicated by Rogers (1956/57), when crystals of dodecylammonium chloride (equivalent of 2.26 × 10− 4 mol/L) were added to a saturated brine (either NaCl or KCl) even after one week the solution did not froth and the surface tension had decreased by only a few mN/m.
E. Burdukova et al. / Int. J. Miner. Process. 93 (2009) 34–40
35
(also at 70 °C). The quantity of HCl was calculated to achieve a 1:1 molar ratio of DDA to HCl, as indicated in Eq. (1). þ
RNH2 + HCl Y RNH3 + Cl
−
ð1Þ
For the sake of convenience, it is still referred to as “amine” or “DDA” throughout the text. In some cases (see next section), MIBC was added to the hot solution of distilled water and HCl to obtain 2:1 molar ratio between amine and MIBC. 2.1.2. Saturated brine Saturated brine was prepared by dissolving an excess amount of potash ore (provided by the Saskatchewan Potash Corporation) containing both sylvite and halite minerals in tap water. The ore/water mixture was allowed to equilibrate for 3 days and thereafter filtered to remove any insoluble impurities and un-dissolved salt particles.
Fig. 1. Summary of the types of amine/brine dispersions.
Leja (1983) reported that when amine collector was deposited on the surface of unstirred brine containing KCl particles, no particle bubble attachment took place. However, when the brine was stirred for a few minutes KCl particles were being picked-up by bubbles and a contact angle was developed on KCl discs. This prompted the author to conclude that since long-chain amines are practically insoluble in brine the underlying mechanism at such a high ionic strength is likely to be different from that in conventional flotation systems which involve diffusion of the dissolved collector and selective adsorption. Common observations that the potash ore flotation is only possible when the solubility limit of collectors is exceeded (Schubert, 1967, 1988; Roman et al., 1968; Miller et al., 1997) seems to be entirely in line with such conclusions. In commercial flotation operations, C16–C22 long-chain primary amines are melted by heating up to 70–90 °C and neutralized with hydrochloric or acetic acids. This converts the amine to an ammonium salt. The resulting hot emulsion/dispersion is introduced to the flotation pulp (which in Saskatchewan is at a temperature of 24–32 °C). Once added to the flotation pulp (where the flotation medium is saturated brine) the hot amine dispersion rapidly cools down to a temperature far below the Krafft point (Dai and Laskowski, 1991; Laskowski et al., 2007). All reported observations reveal that a white precipitate immediately appears when the hot amine dispersion is added to the potash flotation pulp. The rapid conversion from a hot emulsion to a cold precipitate is a very severe transformation but nothing is known about the kinetics of these changes. It is interesting to observe that in the summertime, when temperatures are higher, longer chain amines perform better. Apparently, when the difference between the pulp temperature and the Krafft temperature of the utilized amine is not sufficiently large, a longer chain amine, the amine with a higher Krafft point, is more efficient. As the discussed pieces of evidence indicate the nature of the species that render KCl surfaces hydrophobic in the cationic flotation of potash ores is not known at all. The objective of this study is to investigate the dispersion system which long chain amine forms when it is introduced into a KCl–NaCl saturated brine, and the abstraction of the species formed in such systems by KCl and NaCl particles.
2.1.3. Amine/brine dispersions • With MIBC added to the 2 wt.% DDA emulsion. The amine/brine dispersions were prepared by adding 10 ml of the 2 wt.% dispersions of DDA to 100 ml of saturated brine. Three types of dispersions were prepared. • With MIBC added to saturated brine. • Without MIBC. In the first case, MIBC was added directly to the hot solution of amine in distilled water, as described in Section 2.1.1. In the second case, the equivalent amount of MIBC was added to the 100 ml of saturated brine prior to mixing with the 2 wt.% amine dispersion. In the last case, no MIBC was added to the system. Two further variations of the above mentioned dispersions were prepared following either fast cooling or slow cooling. In the former case, the hot 2 wt.% amine dispersions were added to saturated brine at room temperature. This resulted in a rapid cooling of the amine dispersions, simulating the industrial flotation process. In the latter case, the saturated brine was heated up to 70 °C prior to mixing with the hot amine dispersion, and the final mixture was allowed to slowly cool overnight. A flowchart summarising all the different dispersion types is given in Fig. 1.
2. Experimental details 2.1. Materials 2.1.1. Dodecyl amine The amine used in this study was 98% pure dodecyl amine (DDA), supplied by Dachat Chemicals Inc. In order to prepare 2 wt.% dispersions of DDA in water, DDA was heated up to 70 °C. Once the solid melted, it was mixed with 100 ml of distilled water containing HCl
Fig. 2. Original image and segmentation of dispersions containing opaque particles.
36
E. Burdukova et al. / Int. J. Miner. Process. 93 (2009) 34–40
2.3.1. Images containing opaque particles As mentioned, visual inspection of the dataset containing images of small, opaque particles suggested that simple thresholding could be used to segment the individual particles in the optical microscopy images. To achieve this, the images where converted from colour to greyscale images. Otsu's method (Otsu, 1979) was used to determine the optimal greyscale threshold level. The greyscale image was then thresholded using this value to create a binary image of particles. A binary filling operation was then performed. Finally an image labelling operation (Ritter and Wilson, 2001) was used to identify the individual particles from the binary image. The results of these operations can be seen in Fig. 2 where it is evident that all of the particles have been correctly identified.
The approach to segmenting these images was as follows: It was assumed that for each of the images, the majority of the image pixels where in fact background pixels of similar colour. So, by analysing the distribution of the colours in the image, it should be possible to identify the background areas of the images, and thereby identify the foreground areas (particles) as well. In order to segment according to the colour of the particles, the RGB images where transformed to the HSV (hue, saturation and value) colour space. For each of the H, S and V layers, histograms were determined, and the maximal location of the histogram identified. The maximum location in the histogram is indicative of the most common (probable) value for a pixel, and is therefore representative of the background colour. A user-specified parameter was used to specify how similar to the most probable background value a pixel should be, to be considered a background pixel. This method was utilized to identify background and foregrounds for each of the H,S, and V layers. A binary OR operation was then used to combine these results into a single binary image. Visual inspection of the results showed that while they were mostly accurate, at times single large particles would be incorrectly over-segmented, and occasionally clusters of tiny particles were under-segmented. Due to the relatively small size of the data set, it was possible to manually improve the segmentation of these over- and under-segmented particles by manipulation of the output binary images in image editing software. The results of this process are shown in Fig. 3. Once again an image labelling operation (Ritter and Wilson, 2001) was then used to identify the individual particles from the binary image. It should be noted, that in the case of both opaque and transparent particles, all the segmented regions containing fewer than five pixels were automatically removed. This was done because it was impossible to tell from a visual inspection whether or not the segmentation represented a real particle or an arbitrary artefact of the image.
2.3.2. Images containing translucent particles The nature of these images was such that they did not lend themselves to be segmented using a simple greyscale thresholding. Visual inspection of the images showed that the particles in the images were typically different from the background colour, but sometimes the particles appeared translucent.
2.3.3. Particle size distribution calculation Once the individual particles have been identified, it is a trivial task to calculate the area of each particle, initially in pixels, and then converted into μm2. The area measurement was then converted to an equivalent particle diameter. Finally, area-weighted cumulative particle size distributions were calculated for each of the data sets.
2.2. Optical microscopy images Optical microscopy images were obtained using the Olympus BX600 microscope, which was fitted with a digital camera (Olympus U PMTVC). Samples of the dispersions were placed on a glass slide using a pipette and positioned in the microscope stage and viewed in transmitted light. To ensure reproducibility, over twenty images were collected per condition. All the images were in the .JPG format, 2560 × 1920 pixel image size. 2.3. Particle size measurement Visual analysis of the optical microscopy images suggested that image analysis methods would be suitable for determining particle size information. The samples fell into two categories, those which could be easily segmented using simple thresholding techniques (see Fig. 2), and those which would require some level of user assistance, due mainly to the presence of translucent particles (see Fig. 3).
2.4. Turbidity measurements The turbidity measurements were performed using HATCH 2100AN turbidimeter. The measurements were performed in triplicate, using 10 ml samples. The dispersions of amine were not stable, as the particles have a tendency to aggregate at the air/solution interface. Due to this, there was a substantial amount of drift within the turbidity measurements with time. For this reason, the turbidity measurement was consistently taken after 60 s have elapsed. 2.5. FTIR spectroscopy
Fig. 3. Original image and segmentation of dispersions containing translucent particles.
FTIR spectra were determined using a PerkinElmer System 2000 FT-IR spectrometer. Polished KCl and NaCl plates, obtained from Xymotech Labs, were used for the measurements, with the plates being placed directly in the beam path. The background spectra were obtained using both KCl and NaCl plates which were first rinsed in saturated brine. After rinsing, the plates were placed edge-on on glass slides and allowed to dry overnight. In order to determine whether or not amine adsorption takes place on the plate surfaces, the plates were first rinsed in brine then immediately brought into contact with the amine dispersions using a pair of tweezers. The plates were swirled around in the dispersions for the duration of 1 min, making sure they came into frequent contact with the air interface. After 1 min, the plates were removed from the amine suspension and gently rinsed in fresh saturated brine. After
E. Burdukova et al. / Int. J. Miner. Process. 93 (2009) 34–40
37
Fig. 4. Images of DDA particles in a slowly cooled suspension (A) with MIBC present in the DDA emulsion, (B) with MIBC present in the brine, (C) in the absence of MIBC.
Fig. 5. Images of DDA particles in a rapidly cooled suspension (A) with MIBC present in the DDA emulsion, (B) with MIBC present in the brine, (C) in the absence of MIBC.
The DDA/brine dispersions, prepared under a variety of conditions, were viewed under an optical microscope in order to examine the size and morphology of the resulting DDA particles. Examples of the obtained images are shown in Figs. 4 and 5. As it is seen in the images, the solid particles are present in these systems. This confirms that DDA precipitates in saturated brine and
forms a colloidal system. Furthermore, it can be seen that the addition of MIBC to the hot DDA emulsion (Figs. 4A and 5A) results in a significant reduction in particle size and increase in dispersion degree of the particles compared to the other conditions (Figs. 4B and C and 5B and C). These results are in good agreement with the rheological tests (Wang et al., 1995; Laskowski et al., 2008) in which the presence of MIBC was shown to result in a significant reduction of the yield stresses of the DDA/brine dispersions. It is also clear from Figs. 4 and 5 that MIBC does not act as a dispersing agent when present in the brine (and not in a hot DDA emulsion). The particle size in these dispersions is roughly equivalent to those where no MIBC was present. This implies that the dispersing action of MIBC (and potentially other alcohol based surfactants) is
Fig. 6. Particle size distribution of the amine particles under a variety of conditions.
Fig. 7. Turbidity of the DDA/brine dispersions under a variety of conditions.
rinsing, the plates were placed edge-on on glass slides and allowed to dry overnight. 3. Results and discussion 3.1. Size and morphology of amine particles
38
E. Burdukova et al. / Int. J. Miner. Process. 93 (2009) 34–40
Fig. 8. FTIR spectra of DDA placed on NaCl and KCl plates.
only effective when MIBC is brought into contact with liquefied amine (at elevated temperature), and not when the particles already solidified in saturated brine. This agrees with the results reported by Klassen. Tyurnikova and Naumov (1981) in their monograph discussed the effect of the way the oleic acid dispersion is prepared on its flotation activity. They quoted earlier publications by Klassen in which he had showed how the use of surfactants, applied at elevated temperature, can improve the dispersion of the oleic acid–water colloidal system and its flotation performance. Another interesting observation that can be drawn from Figs. 4 and 5 is that when MIBC is absent, or is only present in the brine, DDA tends to form what appear to be large, transparent crystals (Figs. 4B and C and 5B and C), as opposed to amorphous colloidal matter (Figs. 4A and 5A). It was this difference in morphology that necessitated a different approach to the determination of particle size described in Section 3.2. Finally, it can be observed from Figs. 4 and 5 that the rate of cooling of the DDA/brine dispersions does not have a significant effect on the size of the precipitating particles. However, it may be erroneous to attempt to estimate subtle changes in the size of the DDA particles from purely visual observations. For this reason, the images were digitally analysed in order to obtain a quantitative estimate of particle size. 3.2. Particle size distributions In order to obtain a quantitative comparison of the particle sizes in the DDA dispersions, the images were digitally analysed in a manner described in Section 3.3. The resulting particle size distributions are presented in Fig. 6.
The results clearly demonstrate the size range of solidified amine particles. The particles range in size between circa 3 and 800 µm. These numbers should be considered with caution, specifically in the case of the larger crystal-like particles, as they represent two-dimensional rather than 3 dimensional shapes. The tendency of amine to aggregate at the air–water interface results in the formation of thin flat planes, rather than spheroids. The results also confirm the visual observations (Figs. 4 and 5). As Fig. 6 demonstrates, for the dispersions with MIBC present in the DDA emulsion, the resulting particle size is significantly reduced. The p80 of the particle size distributions representing the tests where MIBC was present in the DDA emulsion are approximately equal to 30 µm. The p80 of the remainder of the tests ranges between approximately 170 and 200 µm. Fig. 6 (as well as Figs. 4B and 5B) also shows that MIBC no longer acts as a dispersing agent when present in the brine. In fact, the conditions were MIBC was present in the brine exhibited marginally coarser particle size distributions (compared to those in the absence of MIBC). However, it is difficult to say whether these differences are significant. In addition, Fig. 6 further demonstrates that no clear or consistent differences in particle size distributions exist between the slow cooling and rapidly cooling conditions. This is in line with the visual observations described in the previous section. 3.3. Turbidity measurements In order to further confirm the findings of the measurements obtained using microscopic images of DDA/brine solutions, the same solutions were subjected to turbidity measurements. Turbidity measurements are strongly indicative of the degree of aggregation/dispersion as well as a good gauge of particle size differences. The precipitating amine colloidal particles tend to aggregate at the air/brine interface. It follows that the particles with the lowest volume (and hence the smallest size) would rise to the surface a lot slower and result in higher solution turbidity. The turbidity results for the DDA/ brine dispersions are presented in Fig. 7. Once again, the results clearly confirm the findings described in the previous sections. The turbidity of the dispersions where MIBC was present in the DDA emulsion show turbidity levels more than double those of the remainder of the dispersions. This confirms that the colloidal system formed under this condition exhibits both smaller particle size and a greater degree of dispersion. The results presented in Fig. 6 suggest that the conditions were MIBC was present in the brine resulted in marginally coarser particle sizes than that in the absence of MIBC. Surprisingly, Fig. 7 shows a very similar trend. For both slow cooling and rapid cooling dispersions the condition where MIBC was present in the brine exhibits lower levels of turbidity than in the absence of MIBC, indicative of the greater degree of aggregation/larger particle size. The differences are small, but significant on a 99% confidence interval. The reason for these differences is not clear. Unlike the size analysis results, the turbidity results show substantial differences between slow cooling and rapidly cooling conditions. The
Fig. 9. Images of polished NaCl plates (A) rinsed in brine (B) treated with DDA prepared with MIBC then rinsed in brine, (C) treated with DDA prepared without MIBC then rinsed in brine.
E. Burdukova et al. / Int. J. Miner. Process. 93 (2009) 34–40
Fig. 10. FTIR spectra of NaCl plates (A) rinsed in brine and used as a background scan (B) treated with DDA prepared with MIBC then rinsed in brine, (C) in treated with DDA prepared without MIBC then rinsed in brine.
conditions were dispersions were cooled rapidly exhibit higher levels of turbidity than the slow cooling conditions, indicating a smaller particle size and a greater degree of dispersion. The differences are small, but significant on a 99% confidence interval. 3.4. Attachment of DDA onto KCl and NaCl plates As discussed in the previous sections, the method of preparation of amine/brine dispersion affects the size, dispersion degree and the morphology of the resulting DDA particles. However, the rate at which the dispersions were cooled down played only a small role in determining the size of the particles. Similarly, only a very insignificant effect of MIBC was observed when MIBC was added to brine as compared with the effect of MIBC when added to a hot DDA emulsion. In lieu of these findings, the next section of the paper focuses on only two conditions: absence of MIBC and the presence of MIBC in the DDA emulsion. For all the results that follow, the amine/brine dispersions were prepared by the slow cooling method. This section of the paper aims at determining whether large amine aggregates and crystals are able to selectively attach to the surface of KCl plates. This was achieved by contacting polished NaCl and KCl plates with the amine/brine dispersions and analysing the resulting surfaces using both microscopy and FTIR spectroscopy. In order to establish an FTIR spectrum for DDA (more specifically the ammonium salt of dodecylamine), the concentrated DDA disper-
39
sion was placed directly onto the surfaces of KCl and NaCl particles and allowed to dry. The resulting spectra are presented in Fig. 8. The figure shows spectra in the wavelength number range between 2700 and 3100 cm− 1. The spectra reveal strong absorbance peaks at 2954, 2917, and 2850 cm− 1. These peaks typically correspond to alkyl chains of primary amines and are consistent with those identified as dodecyl amine by other researchers (Bellamy, 1975; Vidyadhar et al., 2002). Once the spectra corresponding to DDA were established, the attachment characteristics and selectivity of solid amine particles could be investigated. First, the attachment behaviour of DDA on sodium chloride was tested. The plates were analysed using FTIR spectroscopy as well as viewed under an optical microscope to examine the visual effect of amine attachment on the plate's surface. Fig. 9A represents the image of the NaCl particle, rinsed in saturated brine. The image shows a relatively clean plate surface that is slightly damaged by polishing. Fig. 9B and C represents the images of polished NaCl plates which have been contacted with DDA/brine dispersions, both in the presence and absence of MIBC. In both cases, no particles are visible other than NaCl/KCl crystals forming on the plates' surfaces. These crystals can be easily identified by their regular square shape and likely occur as a result of drying brine. This result is expected; Schubert (1967) showed that halite floats with amines only when pH is higher than 10. In order to confirm that no attachment of DDA particles occur on the NaCl surface, the FTIR spectrum was obtained for these NaCl plates. Spectrum A (Fig. 10) represents a background scan of a NaCl plate, which was obtained using a plate rinsed with saturated brine. As expected, no peaks are visible. Spectra B and C (Fig. 10) represent the measurements of the NaCl plates brought into contact with DDA/ brine dispersions both in the presence and absence of MIBC. Once again, no peaks are visible in the 2700–3100 cm− 1 range. This confirms that neither DDA particles nor DDA molecules attach/adsorb on the surface of sodium chloride. Similar tests were preformed using sylvite, or KCl plates. Fig. 11A contains the microscopic image of a KCl plate that has been rinsed in saturated brine and allowed to dry. Once again, the image shows a relatively smooth surface, covered by a small number of square crystals, which can be identified as sodium chloride. Fig. 11B shows an image of the KCl plate which has been contacted with a DDA/brine dispersion, prepared with MIBC. Unlike the reference image A, the plate appears to be coated with tiny, finely disseminated solid particles. These particles do not appear similar to NaCl crystal formations seen in Fig. 11A. Fig. 11C demonstrates an image of a KCl plate which has been contacted with a DDA/brine suspension prepared in the absence of MIBC. The image clearly shows a presence of large, transparent crystals on the surface of the plate. The size and shape of these crystals directly corresponds to those identified as DDA particles shown in Figs. 4C and 5C. As the images
Fig. 11. Images of polished KCl plates at pH 7 (A) rinsed in brine (B) treated with DDA prepared with MIBC then rinsed in brine, (C) in treated with DDA prepared without MIBC then rinsed in brine.
40
E. Burdukova et al. / Int. J. Miner. Process. 93 (2009) 34–40
brine. The presence of MIBC, when added to saturated brine at room temperature had no significant effect on either size or morphology of DDA particles (compared to the conditions without MIBC). • The rate of cooling of DDA/brine dispersions had a marginal effect on the size and aggregation of DDA particles. Rapidly cooling dispersion produced slightly finer particles with a greater degree of dispersion. However this effect could only be observed using turbidity measurements. Neither visual observations nor particle size analysis with image processing techniques could detect such effects. 4.2. Attachment of DDA to KCl and NaCl surfaces The selectivity of attachment of DDA particles was tested using both microscopy and FTIR spectroscopy methods. The following conclusions can be drawn:
Fig. 12. FTIR spectra of KCl plates at (A) rinsed in brine and used as a background scan (B) treated with DDA prepared with MIBC then rinsed in brine, (C) treated with DDA prepared without MIBC then rinsed in brine.
shown in Fig. 11 indicate, DDA particles attach to the surfaces of KCl plates. In order to confirm that the attached particles, seen in Fig. 11, correspond to DDA, the plates were tested using FTIR spectroscopy. Spectrum A (Fig. 12) represents a background scan of the KCl plate, which has been rinsed in saturated brine and allowed to dry (corresponding to the plate shown Fig. 11A). As expected, no peaks are visible. Spectra B and C represent the measurements on the KCl plates brought into contact with DDA/brine dispersions both in the presence and absence of MIBC (corresponding to Fig. 11B and C respectively). These spectra clearly exhibit strong peaks at 2850, 2917 and 2954 cm− 1 which correspond directly to those shown in Fig. 8, and can be identified as DDA. 4. Conclusions 4.1. Precipitation of dodecyl amine in saturated brine A number of experimental techniques were used to study the factors that may affect the precipitation of dodecyl amine in saturated brine. Several conclusions are evident from the presented data: • Solidified particles are formed when aqueous dispersion of hot dodecyl amine is added to brine at room temperature. These particles can vary significantly in size, between circa 30 and 800 µm. The morphology of these particles can also vary between small, amorphous colloidal particles and large, transparent crystals — depending on the conditions of the suspension preparation process. • Methyl isobutyl carbinol (MIBC) acts as a strong dispersing agent, affecting both the size of the amine particles and their aggregation. However, MIBC only acts efficiently as a dispersant when added to a hot DDA/distilled water emulsion prior to mixing with saturated
• DDA particles attach to the surfaces of polished KCl plates as evidenced by images obtained via an optical microscope. This finding is confirmed by analysing the KCl plates using FTIR spectra which exhibit peaks characteristic for DDA. • The attachment of DDA particles to the surface of KCl and NaCl is selective as no attachment took place on the surfaces of NaCl plates. • The results indicate that precipitating amine particles when hot amine dispersions are added to cold brine are selectively abstracted by KCl particles in the potash flotation process. References Bellamy, L.J., 1975. The Infrared Spectra of Complex Molecules. Wiley, New York. Burdukova, E., Laskowski, J.S., 2009. Effect of insoluble amine on bubble surfaces on particle-bubble attachment in potash flotation. Canadian J. Chem. Eng. 87, 441–447. Dai, Q., Laskowski, J.S., 1991. The Krafft point of dodecylammonium chloride: pH effect. Langmuir 7, 1361–1364. Gaska, R.A., Goodenough, R.D., Stuart, G.A., 1965. Ammonia as a solvent. Chem. Eng. Progress, Vol. 61, 139–144. Hancer, M., Celik, M.S., Miller, J.D., 2001. The significance of interfacial water structure in soluble salt flotation systems. J. Coll. Interf. Sci., Vol. 235, 150–161. Klassen, V.I., Mokrousov, V.A., 1963. An Introduction to the Theory of Flotation. Butterworths. Laskowski, J.S., Pawlik, M., Ansari, A., 2007. Effect of brine concentration on the Krafft point of long-chain primary amines. Can. Metall. Quarterly 46, 295–300. Laskowski, J.S., Yuan, X.M., Alonso, E.A., 2008. Potash ore flotation — how does it work? In: Wang, D.D., et al. (Ed.), Proc.24th Int. Minerals Processing Congress, pp. 1270–1276. Beijing. Leja, J., 1983. On the action of long chain amines in potash flotation. In: McKercher, R.M. (Ed.), Potash Technology. Pergamon Press, Toronto, pp. 623–629. Miller, J.D., Veeramasuneni, S., Yalamanchili, M.R., 1997. Recent contributions to the analysis of soluble salt flotation systems. Int. J. Min. Proc. 51, 111–123. Otsu, N., 1979. A treshold selection method from gray-level histograms. IEEE Trans. Syst. Man Cybern. 9, 62–66. Ritter, G.X., Wilson, J.N., 2001. Handbook of Computer Vision Algorithms in Image Algebra. CRC Press. Rogers, J., 1956/57. Flotation of soluble salts. Trans. IMM 66, 439–452. Roman, R.J., Fuerstenau, M.C., Seidel, D.C., 1968. Mechanism of soluble salt flotation. Trans. AIME 241, 56–64. Schubert, H., 1967. What goes on during potash flotation. Eng. Min. J. 168, 94–97. Schubert, H., 1988. The mechanisms of collector adsorption on salt-type minerals from solutions containing high electrolyte concentrations. Aufbereit. – Technik 29, 427–435 i. Tyurnikova, V.I., Naumov, M.E., 1981. Improving the Effectiveness of Flotation. Technicopy Limited. Vidyadhar, A., Rao, K.H., Chernyshova, I.V., Pradip, Forssberg, K.S.E., 2002. Mechanisms of amine–quartz interaction in the absence and presence of alcohols studied by spectroscopic methods. J. Colloid Interface Sci. 256, 59–72. Wang, Q., Alonso, E.A., Laskowski, J.S., 1995. The effect of frothers on potash ore flotation. Proc. 19th Int. Mineral Processing Congress, Littleton, pp. 49–53.
Contents lists available at ScienceDirect
Int. J. Miner. Process. j o u r n a l h o m e p a g e : w w w. e l s ev i e r. c o m / l o c a t e / i j m i n p r o
Precipitation of dodecyl amine in KCl–NaCl saturated brine and attachment of amine particles to KCl and NaCl surfaces E. Burdukova a, J.S. Laskowski a,⁎, G.R. Forbes b a b
N.B. Keevil Institute of Mining Engineering, University of British Columbia, Vancouver, B.C., Canada Department of Materials Engineering, University of British Columbia, Vancouver, B.C., Canada
a r t i c l e
i n f o
Article history: Received 17 February 2009 Accepted 13 May 2009 Available online 19 May 2009 Keywords: Dodecyl amine Krafft point Amine precipitation Sylvite Halite Potash ore Potash ore flotation
a b s t r a c t Long-chain amines, used in potash ore flotation as collectors, are insoluble in NaCl–KCl saturated brine. In commercial applications, these amines are melted at 70–90 °C, dispersed in acidic solution of hydrochloric or acetic acids, and such emulsions are then introduced to the flotation pulp. To model the commercial potash ore flotation process, dodecyl amine, used in this study, was melted at 70 °C, dispersed in hydrochloric acid aqueous solution and was added to a KCl–NaCl saturated brine at room temperature. This results in the precipitation of the amine. The present study summarizes the influence of the conditions on the particle size and morphology of the precipitating amine particles. Methyl isobutyl carbinol (MIBC), common frother in flotation processes, was shown to affect amine dispersion when added into a hot amine emulsion prior to mixing with a saturated brine. This study demonstrates that the precipitating amine particles are selectively abstracted by KCl particles, but not by NaCl particles. © 2009 Elsevier B.V. All rights reserved.
1. Introduction Sylvite (KCl) and halite (NaCl), two main components of sylvinite (potash) ores, are commonly separated by flotation. Since both these minerals are soluble in water the process is carried out in NaCl–KCl saturated brine. Long-chain primary amines are utilized as a collector to float selectively sylvite. Since potash ores also contain a few percent of water-insoluble minerals, which in this process appear in the form of slimes, the process critically depends on desliming. At 20 °C, 1.450 kg of the NaCl–KCl saturated solution contains about 0.300 kg of NaCl and 0.150 kg of KCl and 1 kg of water (Gaska et al., 1965). Then, in the saturated NaCl–KCl brine there is about 5.1 kmol of NaCl and 2 kmol of KCl per 55.5 kmol of water. This, assuming complete NaCl and KCl dissociation, gives for each ion 3.9 molecules of water. It is thus obvious that in the saturated brine, all water molecules are bound in hydration shells around hydrated ions. In the potash flotation process the collector molecules are then transferred from an electrolyte to the interface whose composition does not essentially differ from that in the solution (Schubert, 1988). The question thus arises what is a driving force for the transfer of the collector to the interface. Since chemisorption must be ruled out in this system, the conclusion is that the major component of the adsorption energy is the energy of association of hydrocarbon chains.
⁎ Corresponding author. E-mail address: [email protected] (J.S. Laskowski). 0301-7516/$ – see front matter © 2009 Elsevier B.V. All rights reserved. doi:10.1016/j.minpro.2009.05.001
As discussed by Schubert (1967, 1988) this energy differs only slightly from the precipitation energy of the collector molecules within the liquid phase. There is a huge amount of evidence that indicates that collector selective adsorption is a main reason for selectivity of flotation. However, these results come from the research carried out with the low ionic strength systems. But even in such systems a quite different mechanism is also possible. In the emulsion flotation, in which “oily” water-insoluble collectors are utilized, the selectivity of the process depends on the selective attachment of the oil droplets to mineral surfaces (Klassen and Mokrousov, 1963). So, in this case the process is not accompanied by diffusion of collector molecules/ions and selective adsorption onto given mineral surfaces, but by a transport of dispersed oil droplets towards mineral surfaces. The selectivity in such a process critically depends on the initial wetting of the floated minerals since the oil droplets can attach only to hydrophobic surfaces. By the way, this is a reason why the emulsion flotation is utilized in processing coals, molybdenite, graphite, and other inherently hydrophobic minerals. The fact that wettability of different water-soluble minerals in brines can be different was already taken into consideration by Rogers (1956/57) and was further substantiated by Hancer et al. (2001), and Burdukova and Laskowski (2009). As indicated by Rogers (1956/57), when crystals of dodecylammonium chloride (equivalent of 2.26 × 10− 4 mol/L) were added to a saturated brine (either NaCl or KCl) even after one week the solution did not froth and the surface tension had decreased by only a few mN/m.
E. Burdukova et al. / Int. J. Miner. Process. 93 (2009) 34–40
35
(also at 70 °C). The quantity of HCl was calculated to achieve a 1:1 molar ratio of DDA to HCl, as indicated in Eq. (1). þ
RNH2 + HCl Y RNH3 + Cl
−
ð1Þ
For the sake of convenience, it is still referred to as “amine” or “DDA” throughout the text. In some cases (see next section), MIBC was added to the hot solution of distilled water and HCl to obtain 2:1 molar ratio between amine and MIBC. 2.1.2. Saturated brine Saturated brine was prepared by dissolving an excess amount of potash ore (provided by the Saskatchewan Potash Corporation) containing both sylvite and halite minerals in tap water. The ore/water mixture was allowed to equilibrate for 3 days and thereafter filtered to remove any insoluble impurities and un-dissolved salt particles.
Fig. 1. Summary of the types of amine/brine dispersions.
Leja (1983) reported that when amine collector was deposited on the surface of unstirred brine containing KCl particles, no particle bubble attachment took place. However, when the brine was stirred for a few minutes KCl particles were being picked-up by bubbles and a contact angle was developed on KCl discs. This prompted the author to conclude that since long-chain amines are practically insoluble in brine the underlying mechanism at such a high ionic strength is likely to be different from that in conventional flotation systems which involve diffusion of the dissolved collector and selective adsorption. Common observations that the potash ore flotation is only possible when the solubility limit of collectors is exceeded (Schubert, 1967, 1988; Roman et al., 1968; Miller et al., 1997) seems to be entirely in line with such conclusions. In commercial flotation operations, C16–C22 long-chain primary amines are melted by heating up to 70–90 °C and neutralized with hydrochloric or acetic acids. This converts the amine to an ammonium salt. The resulting hot emulsion/dispersion is introduced to the flotation pulp (which in Saskatchewan is at a temperature of 24–32 °C). Once added to the flotation pulp (where the flotation medium is saturated brine) the hot amine dispersion rapidly cools down to a temperature far below the Krafft point (Dai and Laskowski, 1991; Laskowski et al., 2007). All reported observations reveal that a white precipitate immediately appears when the hot amine dispersion is added to the potash flotation pulp. The rapid conversion from a hot emulsion to a cold precipitate is a very severe transformation but nothing is known about the kinetics of these changes. It is interesting to observe that in the summertime, when temperatures are higher, longer chain amines perform better. Apparently, when the difference between the pulp temperature and the Krafft temperature of the utilized amine is not sufficiently large, a longer chain amine, the amine with a higher Krafft point, is more efficient. As the discussed pieces of evidence indicate the nature of the species that render KCl surfaces hydrophobic in the cationic flotation of potash ores is not known at all. The objective of this study is to investigate the dispersion system which long chain amine forms when it is introduced into a KCl–NaCl saturated brine, and the abstraction of the species formed in such systems by KCl and NaCl particles.
2.1.3. Amine/brine dispersions • With MIBC added to the 2 wt.% DDA emulsion. The amine/brine dispersions were prepared by adding 10 ml of the 2 wt.% dispersions of DDA to 100 ml of saturated brine. Three types of dispersions were prepared. • With MIBC added to saturated brine. • Without MIBC. In the first case, MIBC was added directly to the hot solution of amine in distilled water, as described in Section 2.1.1. In the second case, the equivalent amount of MIBC was added to the 100 ml of saturated brine prior to mixing with the 2 wt.% amine dispersion. In the last case, no MIBC was added to the system. Two further variations of the above mentioned dispersions were prepared following either fast cooling or slow cooling. In the former case, the hot 2 wt.% amine dispersions were added to saturated brine at room temperature. This resulted in a rapid cooling of the amine dispersions, simulating the industrial flotation process. In the latter case, the saturated brine was heated up to 70 °C prior to mixing with the hot amine dispersion, and the final mixture was allowed to slowly cool overnight. A flowchart summarising all the different dispersion types is given in Fig. 1.
2. Experimental details 2.1. Materials 2.1.1. Dodecyl amine The amine used in this study was 98% pure dodecyl amine (DDA), supplied by Dachat Chemicals Inc. In order to prepare 2 wt.% dispersions of DDA in water, DDA was heated up to 70 °C. Once the solid melted, it was mixed with 100 ml of distilled water containing HCl
Fig. 2. Original image and segmentation of dispersions containing opaque particles.
36
E. Burdukova et al. / Int. J. Miner. Process. 93 (2009) 34–40
2.3.1. Images containing opaque particles As mentioned, visual inspection of the dataset containing images of small, opaque particles suggested that simple thresholding could be used to segment the individual particles in the optical microscopy images. To achieve this, the images where converted from colour to greyscale images. Otsu's method (Otsu, 1979) was used to determine the optimal greyscale threshold level. The greyscale image was then thresholded using this value to create a binary image of particles. A binary filling operation was then performed. Finally an image labelling operation (Ritter and Wilson, 2001) was used to identify the individual particles from the binary image. The results of these operations can be seen in Fig. 2 where it is evident that all of the particles have been correctly identified.
The approach to segmenting these images was as follows: It was assumed that for each of the images, the majority of the image pixels where in fact background pixels of similar colour. So, by analysing the distribution of the colours in the image, it should be possible to identify the background areas of the images, and thereby identify the foreground areas (particles) as well. In order to segment according to the colour of the particles, the RGB images where transformed to the HSV (hue, saturation and value) colour space. For each of the H, S and V layers, histograms were determined, and the maximal location of the histogram identified. The maximum location in the histogram is indicative of the most common (probable) value for a pixel, and is therefore representative of the background colour. A user-specified parameter was used to specify how similar to the most probable background value a pixel should be, to be considered a background pixel. This method was utilized to identify background and foregrounds for each of the H,S, and V layers. A binary OR operation was then used to combine these results into a single binary image. Visual inspection of the results showed that while they were mostly accurate, at times single large particles would be incorrectly over-segmented, and occasionally clusters of tiny particles were under-segmented. Due to the relatively small size of the data set, it was possible to manually improve the segmentation of these over- and under-segmented particles by manipulation of the output binary images in image editing software. The results of this process are shown in Fig. 3. Once again an image labelling operation (Ritter and Wilson, 2001) was then used to identify the individual particles from the binary image. It should be noted, that in the case of both opaque and transparent particles, all the segmented regions containing fewer than five pixels were automatically removed. This was done because it was impossible to tell from a visual inspection whether or not the segmentation represented a real particle or an arbitrary artefact of the image.
2.3.2. Images containing translucent particles The nature of these images was such that they did not lend themselves to be segmented using a simple greyscale thresholding. Visual inspection of the images showed that the particles in the images were typically different from the background colour, but sometimes the particles appeared translucent.
2.3.3. Particle size distribution calculation Once the individual particles have been identified, it is a trivial task to calculate the area of each particle, initially in pixels, and then converted into μm2. The area measurement was then converted to an equivalent particle diameter. Finally, area-weighted cumulative particle size distributions were calculated for each of the data sets.
2.2. Optical microscopy images Optical microscopy images were obtained using the Olympus BX600 microscope, which was fitted with a digital camera (Olympus U PMTVC). Samples of the dispersions were placed on a glass slide using a pipette and positioned in the microscope stage and viewed in transmitted light. To ensure reproducibility, over twenty images were collected per condition. All the images were in the .JPG format, 2560 × 1920 pixel image size. 2.3. Particle size measurement Visual analysis of the optical microscopy images suggested that image analysis methods would be suitable for determining particle size information. The samples fell into two categories, those which could be easily segmented using simple thresholding techniques (see Fig. 2), and those which would require some level of user assistance, due mainly to the presence of translucent particles (see Fig. 3).
2.4. Turbidity measurements The turbidity measurements were performed using HATCH 2100AN turbidimeter. The measurements were performed in triplicate, using 10 ml samples. The dispersions of amine were not stable, as the particles have a tendency to aggregate at the air/solution interface. Due to this, there was a substantial amount of drift within the turbidity measurements with time. For this reason, the turbidity measurement was consistently taken after 60 s have elapsed. 2.5. FTIR spectroscopy
Fig. 3. Original image and segmentation of dispersions containing translucent particles.
FTIR spectra were determined using a PerkinElmer System 2000 FT-IR spectrometer. Polished KCl and NaCl plates, obtained from Xymotech Labs, were used for the measurements, with the plates being placed directly in the beam path. The background spectra were obtained using both KCl and NaCl plates which were first rinsed in saturated brine. After rinsing, the plates were placed edge-on on glass slides and allowed to dry overnight. In order to determine whether or not amine adsorption takes place on the plate surfaces, the plates were first rinsed in brine then immediately brought into contact with the amine dispersions using a pair of tweezers. The plates were swirled around in the dispersions for the duration of 1 min, making sure they came into frequent contact with the air interface. After 1 min, the plates were removed from the amine suspension and gently rinsed in fresh saturated brine. After
E. Burdukova et al. / Int. J. Miner. Process. 93 (2009) 34–40
37
Fig. 4. Images of DDA particles in a slowly cooled suspension (A) with MIBC present in the DDA emulsion, (B) with MIBC present in the brine, (C) in the absence of MIBC.
Fig. 5. Images of DDA particles in a rapidly cooled suspension (A) with MIBC present in the DDA emulsion, (B) with MIBC present in the brine, (C) in the absence of MIBC.
The DDA/brine dispersions, prepared under a variety of conditions, were viewed under an optical microscope in order to examine the size and morphology of the resulting DDA particles. Examples of the obtained images are shown in Figs. 4 and 5. As it is seen in the images, the solid particles are present in these systems. This confirms that DDA precipitates in saturated brine and
forms a colloidal system. Furthermore, it can be seen that the addition of MIBC to the hot DDA emulsion (Figs. 4A and 5A) results in a significant reduction in particle size and increase in dispersion degree of the particles compared to the other conditions (Figs. 4B and C and 5B and C). These results are in good agreement with the rheological tests (Wang et al., 1995; Laskowski et al., 2008) in which the presence of MIBC was shown to result in a significant reduction of the yield stresses of the DDA/brine dispersions. It is also clear from Figs. 4 and 5 that MIBC does not act as a dispersing agent when present in the brine (and not in a hot DDA emulsion). The particle size in these dispersions is roughly equivalent to those where no MIBC was present. This implies that the dispersing action of MIBC (and potentially other alcohol based surfactants) is
Fig. 6. Particle size distribution of the amine particles under a variety of conditions.
Fig. 7. Turbidity of the DDA/brine dispersions under a variety of conditions.
rinsing, the plates were placed edge-on on glass slides and allowed to dry overnight. 3. Results and discussion 3.1. Size and morphology of amine particles
38
E. Burdukova et al. / Int. J. Miner. Process. 93 (2009) 34–40
Fig. 8. FTIR spectra of DDA placed on NaCl and KCl plates.
only effective when MIBC is brought into contact with liquefied amine (at elevated temperature), and not when the particles already solidified in saturated brine. This agrees with the results reported by Klassen. Tyurnikova and Naumov (1981) in their monograph discussed the effect of the way the oleic acid dispersion is prepared on its flotation activity. They quoted earlier publications by Klassen in which he had showed how the use of surfactants, applied at elevated temperature, can improve the dispersion of the oleic acid–water colloidal system and its flotation performance. Another interesting observation that can be drawn from Figs. 4 and 5 is that when MIBC is absent, or is only present in the brine, DDA tends to form what appear to be large, transparent crystals (Figs. 4B and C and 5B and C), as opposed to amorphous colloidal matter (Figs. 4A and 5A). It was this difference in morphology that necessitated a different approach to the determination of particle size described in Section 3.2. Finally, it can be observed from Figs. 4 and 5 that the rate of cooling of the DDA/brine dispersions does not have a significant effect on the size of the precipitating particles. However, it may be erroneous to attempt to estimate subtle changes in the size of the DDA particles from purely visual observations. For this reason, the images were digitally analysed in order to obtain a quantitative estimate of particle size. 3.2. Particle size distributions In order to obtain a quantitative comparison of the particle sizes in the DDA dispersions, the images were digitally analysed in a manner described in Section 3.3. The resulting particle size distributions are presented in Fig. 6.
The results clearly demonstrate the size range of solidified amine particles. The particles range in size between circa 3 and 800 µm. These numbers should be considered with caution, specifically in the case of the larger crystal-like particles, as they represent two-dimensional rather than 3 dimensional shapes. The tendency of amine to aggregate at the air–water interface results in the formation of thin flat planes, rather than spheroids. The results also confirm the visual observations (Figs. 4 and 5). As Fig. 6 demonstrates, for the dispersions with MIBC present in the DDA emulsion, the resulting particle size is significantly reduced. The p80 of the particle size distributions representing the tests where MIBC was present in the DDA emulsion are approximately equal to 30 µm. The p80 of the remainder of the tests ranges between approximately 170 and 200 µm. Fig. 6 (as well as Figs. 4B and 5B) also shows that MIBC no longer acts as a dispersing agent when present in the brine. In fact, the conditions were MIBC was present in the brine exhibited marginally coarser particle size distributions (compared to those in the absence of MIBC). However, it is difficult to say whether these differences are significant. In addition, Fig. 6 further demonstrates that no clear or consistent differences in particle size distributions exist between the slow cooling and rapidly cooling conditions. This is in line with the visual observations described in the previous section. 3.3. Turbidity measurements In order to further confirm the findings of the measurements obtained using microscopic images of DDA/brine solutions, the same solutions were subjected to turbidity measurements. Turbidity measurements are strongly indicative of the degree of aggregation/dispersion as well as a good gauge of particle size differences. The precipitating amine colloidal particles tend to aggregate at the air/brine interface. It follows that the particles with the lowest volume (and hence the smallest size) would rise to the surface a lot slower and result in higher solution turbidity. The turbidity results for the DDA/ brine dispersions are presented in Fig. 7. Once again, the results clearly confirm the findings described in the previous sections. The turbidity of the dispersions where MIBC was present in the DDA emulsion show turbidity levels more than double those of the remainder of the dispersions. This confirms that the colloidal system formed under this condition exhibits both smaller particle size and a greater degree of dispersion. The results presented in Fig. 6 suggest that the conditions were MIBC was present in the brine resulted in marginally coarser particle sizes than that in the absence of MIBC. Surprisingly, Fig. 7 shows a very similar trend. For both slow cooling and rapid cooling dispersions the condition where MIBC was present in the brine exhibits lower levels of turbidity than in the absence of MIBC, indicative of the greater degree of aggregation/larger particle size. The differences are small, but significant on a 99% confidence interval. The reason for these differences is not clear. Unlike the size analysis results, the turbidity results show substantial differences between slow cooling and rapidly cooling conditions. The
Fig. 9. Images of polished NaCl plates (A) rinsed in brine (B) treated with DDA prepared with MIBC then rinsed in brine, (C) treated with DDA prepared without MIBC then rinsed in brine.
E. Burdukova et al. / Int. J. Miner. Process. 93 (2009) 34–40
Fig. 10. FTIR spectra of NaCl plates (A) rinsed in brine and used as a background scan (B) treated with DDA prepared with MIBC then rinsed in brine, (C) in treated with DDA prepared without MIBC then rinsed in brine.
conditions were dispersions were cooled rapidly exhibit higher levels of turbidity than the slow cooling conditions, indicating a smaller particle size and a greater degree of dispersion. The differences are small, but significant on a 99% confidence interval. 3.4. Attachment of DDA onto KCl and NaCl plates As discussed in the previous sections, the method of preparation of amine/brine dispersion affects the size, dispersion degree and the morphology of the resulting DDA particles. However, the rate at which the dispersions were cooled down played only a small role in determining the size of the particles. Similarly, only a very insignificant effect of MIBC was observed when MIBC was added to brine as compared with the effect of MIBC when added to a hot DDA emulsion. In lieu of these findings, the next section of the paper focuses on only two conditions: absence of MIBC and the presence of MIBC in the DDA emulsion. For all the results that follow, the amine/brine dispersions were prepared by the slow cooling method. This section of the paper aims at determining whether large amine aggregates and crystals are able to selectively attach to the surface of KCl plates. This was achieved by contacting polished NaCl and KCl plates with the amine/brine dispersions and analysing the resulting surfaces using both microscopy and FTIR spectroscopy. In order to establish an FTIR spectrum for DDA (more specifically the ammonium salt of dodecylamine), the concentrated DDA disper-
39
sion was placed directly onto the surfaces of KCl and NaCl particles and allowed to dry. The resulting spectra are presented in Fig. 8. The figure shows spectra in the wavelength number range between 2700 and 3100 cm− 1. The spectra reveal strong absorbance peaks at 2954, 2917, and 2850 cm− 1. These peaks typically correspond to alkyl chains of primary amines and are consistent with those identified as dodecyl amine by other researchers (Bellamy, 1975; Vidyadhar et al., 2002). Once the spectra corresponding to DDA were established, the attachment characteristics and selectivity of solid amine particles could be investigated. First, the attachment behaviour of DDA on sodium chloride was tested. The plates were analysed using FTIR spectroscopy as well as viewed under an optical microscope to examine the visual effect of amine attachment on the plate's surface. Fig. 9A represents the image of the NaCl particle, rinsed in saturated brine. The image shows a relatively clean plate surface that is slightly damaged by polishing. Fig. 9B and C represents the images of polished NaCl plates which have been contacted with DDA/brine dispersions, both in the presence and absence of MIBC. In both cases, no particles are visible other than NaCl/KCl crystals forming on the plates' surfaces. These crystals can be easily identified by their regular square shape and likely occur as a result of drying brine. This result is expected; Schubert (1967) showed that halite floats with amines only when pH is higher than 10. In order to confirm that no attachment of DDA particles occur on the NaCl surface, the FTIR spectrum was obtained for these NaCl plates. Spectrum A (Fig. 10) represents a background scan of a NaCl plate, which was obtained using a plate rinsed with saturated brine. As expected, no peaks are visible. Spectra B and C (Fig. 10) represent the measurements of the NaCl plates brought into contact with DDA/ brine dispersions both in the presence and absence of MIBC. Once again, no peaks are visible in the 2700–3100 cm− 1 range. This confirms that neither DDA particles nor DDA molecules attach/adsorb on the surface of sodium chloride. Similar tests were preformed using sylvite, or KCl plates. Fig. 11A contains the microscopic image of a KCl plate that has been rinsed in saturated brine and allowed to dry. Once again, the image shows a relatively smooth surface, covered by a small number of square crystals, which can be identified as sodium chloride. Fig. 11B shows an image of the KCl plate which has been contacted with a DDA/brine dispersion, prepared with MIBC. Unlike the reference image A, the plate appears to be coated with tiny, finely disseminated solid particles. These particles do not appear similar to NaCl crystal formations seen in Fig. 11A. Fig. 11C demonstrates an image of a KCl plate which has been contacted with a DDA/brine suspension prepared in the absence of MIBC. The image clearly shows a presence of large, transparent crystals on the surface of the plate. The size and shape of these crystals directly corresponds to those identified as DDA particles shown in Figs. 4C and 5C. As the images
Fig. 11. Images of polished KCl plates at pH 7 (A) rinsed in brine (B) treated with DDA prepared with MIBC then rinsed in brine, (C) in treated with DDA prepared without MIBC then rinsed in brine.
40
E. Burdukova et al. / Int. J. Miner. Process. 93 (2009) 34–40
brine. The presence of MIBC, when added to saturated brine at room temperature had no significant effect on either size or morphology of DDA particles (compared to the conditions without MIBC). • The rate of cooling of DDA/brine dispersions had a marginal effect on the size and aggregation of DDA particles. Rapidly cooling dispersion produced slightly finer particles with a greater degree of dispersion. However this effect could only be observed using turbidity measurements. Neither visual observations nor particle size analysis with image processing techniques could detect such effects. 4.2. Attachment of DDA to KCl and NaCl surfaces The selectivity of attachment of DDA particles was tested using both microscopy and FTIR spectroscopy methods. The following conclusions can be drawn:
Fig. 12. FTIR spectra of KCl plates at (A) rinsed in brine and used as a background scan (B) treated with DDA prepared with MIBC then rinsed in brine, (C) treated with DDA prepared without MIBC then rinsed in brine.
shown in Fig. 11 indicate, DDA particles attach to the surfaces of KCl plates. In order to confirm that the attached particles, seen in Fig. 11, correspond to DDA, the plates were tested using FTIR spectroscopy. Spectrum A (Fig. 12) represents a background scan of the KCl plate, which has been rinsed in saturated brine and allowed to dry (corresponding to the plate shown Fig. 11A). As expected, no peaks are visible. Spectra B and C represent the measurements on the KCl plates brought into contact with DDA/brine dispersions both in the presence and absence of MIBC (corresponding to Fig. 11B and C respectively). These spectra clearly exhibit strong peaks at 2850, 2917 and 2954 cm− 1 which correspond directly to those shown in Fig. 8, and can be identified as DDA. 4. Conclusions 4.1. Precipitation of dodecyl amine in saturated brine A number of experimental techniques were used to study the factors that may affect the precipitation of dodecyl amine in saturated brine. Several conclusions are evident from the presented data: • Solidified particles are formed when aqueous dispersion of hot dodecyl amine is added to brine at room temperature. These particles can vary significantly in size, between circa 30 and 800 µm. The morphology of these particles can also vary between small, amorphous colloidal particles and large, transparent crystals — depending on the conditions of the suspension preparation process. • Methyl isobutyl carbinol (MIBC) acts as a strong dispersing agent, affecting both the size of the amine particles and their aggregation. However, MIBC only acts efficiently as a dispersant when added to a hot DDA/distilled water emulsion prior to mixing with saturated
• DDA particles attach to the surfaces of polished KCl plates as evidenced by images obtained via an optical microscope. This finding is confirmed by analysing the KCl plates using FTIR spectra which exhibit peaks characteristic for DDA. • The attachment of DDA particles to the surface of KCl and NaCl is selective as no attachment took place on the surfaces of NaCl plates. • The results indicate that precipitating amine particles when hot amine dispersions are added to cold brine are selectively abstracted by KCl particles in the potash flotation process. References Bellamy, L.J., 1975. The Infrared Spectra of Complex Molecules. Wiley, New York. Burdukova, E., Laskowski, J.S., 2009. Effect of insoluble amine on bubble surfaces on particle-bubble attachment in potash flotation. Canadian J. Chem. Eng. 87, 441–447. Dai, Q., Laskowski, J.S., 1991. The Krafft point of dodecylammonium chloride: pH effect. Langmuir 7, 1361–1364. Gaska, R.A., Goodenough, R.D., Stuart, G.A., 1965. Ammonia as a solvent. Chem. Eng. Progress, Vol. 61, 139–144. Hancer, M., Celik, M.S., Miller, J.D., 2001. The significance of interfacial water structure in soluble salt flotation systems. J. Coll. Interf. Sci., Vol. 235, 150–161. Klassen, V.I., Mokrousov, V.A., 1963. An Introduction to the Theory of Flotation. Butterworths. Laskowski, J.S., Pawlik, M., Ansari, A., 2007. Effect of brine concentration on the Krafft point of long-chain primary amines. Can. Metall. Quarterly 46, 295–300. Laskowski, J.S., Yuan, X.M., Alonso, E.A., 2008. Potash ore flotation — how does it work? In: Wang, D.D., et al. (Ed.), Proc.24th Int. Minerals Processing Congress, pp. 1270–1276. Beijing. Leja, J., 1983. On the action of long chain amines in potash flotation. In: McKercher, R.M. (Ed.), Potash Technology. Pergamon Press, Toronto, pp. 623–629. Miller, J.D., Veeramasuneni, S., Yalamanchili, M.R., 1997. Recent contributions to the analysis of soluble salt flotation systems. Int. J. Min. Proc. 51, 111–123. Otsu, N., 1979. A treshold selection method from gray-level histograms. IEEE Trans. Syst. Man Cybern. 9, 62–66. Ritter, G.X., Wilson, J.N., 2001. Handbook of Computer Vision Algorithms in Image Algebra. CRC Press. Rogers, J., 1956/57. Flotation of soluble salts. Trans. IMM 66, 439–452. Roman, R.J., Fuerstenau, M.C., Seidel, D.C., 1968. Mechanism of soluble salt flotation. Trans. AIME 241, 56–64. Schubert, H., 1967. What goes on during potash flotation. Eng. Min. J. 168, 94–97. Schubert, H., 1988. The mechanisms of collector adsorption on salt-type minerals from solutions containing high electrolyte concentrations. Aufbereit. – Technik 29, 427–435 i. Tyurnikova, V.I., Naumov, M.E., 1981. Improving the Effectiveness of Flotation. Technicopy Limited. Vidyadhar, A., Rao, K.H., Chernyshova, I.V., Pradip, Forssberg, K.S.E., 2002. Mechanisms of amine–quartz interaction in the absence and presence of alcohols studied by spectroscopic methods. J. Colloid Interface Sci. 256, 59–72. Wang, Q., Alonso, E.A., Laskowski, J.S., 1995. The effect of frothers on potash ore flotation. Proc. 19th Int. Mineral Processing Congress, Littleton, pp. 49–53.