5802
Langmuir 2003, 19, 5802-5807
In Situ FTIR-ATR Examination of Poly(acrylic acid) Adsorbed onto Hematite at Low pH Luke J. Kirwan,† Phillip D. Fawell,‡ and Wilhelm van Bronswijk*,† A. J. Parker Cooperative Research Centre for Hydrometallurgy at Department of Applied Chemistry, Curtin University of Technology, G.P.O. Box U1987, Perth, Western Australia 6845, Australia, and CSIRO Minerals, P.O. Box 90, Bentley, Western Australia 6982, Australia Received December 17, 2002. In Final Form: May 1, 2003 The adsorption of water-soluble polymers (e.g., flocculants, coagulants) onto mineral particles is typically characterized by “trains” of adsorbed polymer segments with “loops” and “tails” of unadsorbed polymer segments that extend into the solution. Ex situ spectroscopic studies carried out in the past have been complicated by the way in which the unadsorbed segments of the polymer interact with the surface of the substrate upon drying. In this study, an in situ Fourier transform infrared attenuated total reflection technique was used to examine the interaction of poly(acrylic acid) and hematite at pH 2. A hematite colloid was deposited onto a ZnSe crystal, and the poly(acrylic acid) solution was subsequently pumped across the coated crystal at a known rate. Polymer adsorption was irreversible and could be described by a Langmuir isotherm, the rate of curvature of which suggested that the adsorption was weak. The mode of adsorption was shown to be bidentate chelate complexation by the carboxylate functional group to a surface ferric ion. Pimelic acid, a simple model compound effectively representing a polymer consisting of only two monomer functionalities, adsorbed onto hematite in the same manner at pH 2 and gave no infrared spectral peaks associated with unadsorbed carboxylate groups, thereby supporting the proposed adsorption mechanism. The technique could also discriminate between the adsorbed and the unadsorbed segments of the adsorbed polymer molecule, however; the fraction of adsorbed segments was difficult to quantify. When a number of assumptions were made, it was found that at most 9% of the carboxylate functional groups of the polymer were adsorbed onto the hematite.
1. Introduction The mode of adsorption of carboxylic acids and their corresponding polymers onto mineral surfaces has been studied extensively using many spectroscopic techniques, including X-ray photoelectron,1-3 inelastic electron tunneling,4 and Mo¨ssbauer5 spectroscopies. Infrared spectroscopic techniques such as diffuse reflectance (DRIFT)6-9 and polarized grazing angle (PGA)10 have also been popular choices because the molecular vibrations associated with the carboxylate functionality offer considerable potential for identifying its binding mode. This has been demonstrated by Deacon and Phillips,11 who correlated transmission infrared spectroscopic and X-ray crystallographic data of acetato transition metal complexes to determine the relationship between the carbon-oxygen stretching frequencies of carboxylato * To whom correspondence should be addressed. Telephone: +61 8 9266 7321. E-mail:
[email protected]. † Curtin University of Technology. ‡ CSIRO Minerals. (1) Alexander, M.; Beamson, G.; Blomfield, C.; Leggett, G.; Duc, T. J. Electron Spectrosc. Relat. Phenom. 2001, 121, 19. (2) Leadley, S. R.; Watts, J. F. J. Electron Spectrosc. Relat. Phenom. 1997, 85, 107. (3) Underhill, R.; Timsit, R. S. J. Vac. Sci., Technol. A 1992, 10, 2767. (4) Coast, R.; Pikus, M.; Henriksen, P. N.; Nitowski, G. A. J. Phys. Chem. 1996, 100, 15011. (5) Zhou, N. F.; Chen, L.-W.; Wang, Y.-S.; Jiang, J.-S. Chem. Phys. Lett. 1994, 224, 595. (6) Jones, F.; Farrow, J. B.; van Bronswijk, W. Langmuir 1998, 14, 6512. (7) Lee, D. H.; Condrate, R. A., Sr.; Reed, J. S. J. Mater. Sci. 1996, 31, 471. (8) Parker, R. W.; Frost, R. L. Clays Clay Miner. 1996, 44, 32. (9) Gong, W. Q.; Parentich, A.; Little, L. H.; Warren, L. J. Colloids Surf. 1991, 60, 325. (10) Allara, D. L.; Nuzzo, R. G. Langmuir 1985, 1, 52. (11) Deacon, G. B.; Phillips, R. J. Coord. Chem. Rev. 1980, 33, 227.
Figure 1. Modes of carboxylate-metal complexation: monodentate (I), bidentate chelating (II), and bidentate bridging (III).
complexes and the type of carboxylate coordination. By comparing the separation of the symmetric and antisymmetric stretching frequencies of the carboxylate ion (∆ν) bound to transition metals with the separation measured for the corresponding sodium salt, they were able to propose a set of rules for identifying the bonding mechanism. These rules are summarized as follows and relate to the schematic views of bonding shown in Figure 1: (i) If there is CdO character in the spectrum and ∆νadsorbed is greater than ∆νsalt then the adsorbed structure is monodentate I. (ii) If there is no CdO character in the spectrum and ∆νadsorbed is smaller than ∆νsalt then the adsorbed structure is bidentate chelating II. (iii) If there is no CdO character in the spectrum and ∆νadsorbed is similar to ∆νsalt then the adsorbed structure is bidentate bridging III. In addition, an asymmetric bidentate bridging structure was suggested by Jones et al.6 from DRIFT studies of a high-molecular-weight polyacrylate adsorbed on hematite in caustic media. The structure was based on the fact that ∆νadsorbed was much greater than ∆νsalt but had a lack of substantial CdO character. The evidence was thought to indicate that one CsOsM bond was longer and weaker than the other. Allara and Nuzzo10 proposed a similar
10.1021/la027012d CCC: $25.00 © 2003 American Chemical Society Published on Web 05/29/2003
Examination of Poly(acrylic acid) Adsorbed on Fe2O3
structure, based on PGA infrared spectroscopy, in their study of the adsorption of n-alkanoic acids on oxidized aluminum substrates. The disadvantage of many Fourier tranform infrared (FTIR) studies, in particular those utilizing DRIFT or PGA, is that the analysis is carried out ex situ. When we consider polymer adsorption, only a fraction of the polymer segments are assumed to be adsorbed, with the remainder of the chain length forming loops and tails that extend into solution. Hence, polymer adsorption characteristically reaches a limiting value, corresponding to a large excess over that expected for monolayer coverage of polymer adsorbed flat on the solid surface.12,13 Therefore, it is highly likely that any ex situ examination of such samples may be hindered by potential binding changes during the washing and drying processes. In particular, the polymer loops and tails that were unadsorbed while in solution may collapse onto and subsequently interact with the substrate surface, with their contribution to the resultant spectra either masking or distorting the true binding behavior. In situ surface studies are much more challenging because information pertaining to surface adsorption is commonly swamped by that of the bulk solution. FTIR attenuated total reflectance (FTIR-ATR) spectroscopy, in which infrared radiation is focused at the surface of a polished crystal of high refractive index (the internally reflecting element, or IRE) in contact with a solution, offers the possibility of in situ examinations of the solid-liquid interface. There are several different cell geometries and measurement strategies that can be employed, with the latter including adsorption directly onto the IRE; adsorption onto particulate matter, which is then contacted with the IRE; and adsorption onto a coated IRE. When the substrate of interest cannot be constructed to be an IRE, the adsorption to a coated IRE strategy offers the greatest scope because adsorption can give quantitative surface excess results equivalent to those obtained by solution depletion.14 This strategy has been the choice of a number of researchers.15-18 Whereas internal reflection occurs at the solid-crystal interface, radiation can penetrate into the sample layer and be attenuated, allowing an absorption spectrum to be obtained. The advantage of this strategy is that the solid layer can be in contact with an aqueous phase, giving interfacial information without any sample treatment that may change the surface characteristics. However, it is essential that the particulate substrate is relatively insoluble and can be applied to the IRE as a very thin film with uniform contact. This is represented schematically in Figure 2. The mode of carboxylate adsorption onto a variety of mineral surfaces in situ has been deduced successfully using FTIR-ATR. Dobson and McQuillan17 used a TiO2or Al2O3-coated ZnSe ATR element to examine the adsorption profile of a range of aliphatic mono- and dicarboxylic acids from aqueous solutions. Monocarboxylic acids (formic and acetic) were found not to adsorb, but dicarboxylic acids did. The spectrum of succinic acid adsorbed onto TiO2 and Al2O3 was clearly different to that of dissolved sodium succinate. For dissolved sodium (12) Eirich, F. R. J. Colloid Interface Sci. 1977, 58, 423. (13) Misra, D. N. J. Colloid Interface Sci. 1996, 181, 289. (14) Sperline, R.; Freiser, H. In The Handbook of Surface Imaging and Visualization; Hubbard, A. T., Ed.; CRC Press: Boca Raton, Florida, 1995; pp 245-263. (15) Connor, P.; McQuillan, A. Langmuir 1999, 15, 2916. (16) Hug, S. J. Colloid Interface Sci. 1997, 188, 415. (17) Dobson, K. D.; McQuillan, A. J. Spectrochim. Acta, Part A 1999, 55, 1395. (18) Kuys, K.; Roberts, N. Colloids Surf. 1987, 24, 1.
Langmuir, Vol. 19, No. 14, 2003 5803
Figure 2. Schematic representation of the ATR cell, showing a deposited solid substrate and polymer adsorbed from a dilute solution (applied at a constant flow rate).
succinate, the symmetric and antisymmetric stretches of the carboxylate ion were at 1393 and 1550 cm-1, respectively (∆ν ) 157 cm-1). The adsorbed species had equivalent modes at 1539 and 1413 cm-1 (∆ν ) 126 cm-1) on TiO2 and at 1553 and 1420 cm-1 (∆ν ) 133 cm-1) on Al2O3. As a result of the absence of a carbonyl stretch and the molecular flexibility of the adsorbate, it was proposed that coordination was through both carboxylate groups of the molecule. The authors suggested that a bridging bidentate complex was formed, but because of the reduction in the separation of the carboxylate stretches, use of the rules outlined by Deacon and Phillips11 would actually indicate bidentate chelation coordination. Adipic acid behaved in the same manner as succinic acid and, hence, should be considered to also adsorb via bidentate chelation coordination to TiO2 and Al2O3. Specht and Frimmel19 characterized the adsorption of oxalic, malonic, and succinic acids onto kaolinite at pH values below 6. In the spectrum of adsorbed succinic acid, the authors could not distinguish bidentate mononuclear chelation and a pseudochelation adsorption mode from monodentate binding to two surface metal ions. In contrast, for adsorbed oxalic and malonic acids, the presence of a weak carbonyl stretch, which was characteristic of the free protonated acid group, indicated that a small fraction of the molecules were adsorbed through only one carboxylate group. A strong carbonyl stretching peak that has shifted to a lower energy was considered to be indicative of a weakened double bond due to bonding of the other carboxylate oxygen atom to the kaolinite surface. The authors proposed that adsorption was dominated by monodentate or monodentate pseudochelation complexation of these acids to the kaolinite surface. This conclusion is in agreement with the rules outlined by Deacon and Phillips11 and is consistent with the monodentate oxalate binding structures proposed by Hug and Sulzberger20 for oxalic acid adsorption on TiO2. It is also favored on steric grounds because bidentate binding from only one carboxylate group would force the second carboxylate to be nonbinding and, hence, lead to nearer equal intensities of free and bound carboxylate stretches. In this study, FTIR-ATR has been used for the first time to examine in situ the adsorption of a long-chain poly(acrylic acid) and pimelic acid onto a hematite-coated ZnSe crystal at pH 2 for the purpose of obtaining an insight into the behavior at the interface without interference from the bulk solution and understanding the adsorption mechanism for poly(acrylic acid) at low pH. 2. Experimental Section 2.1. Flocculant Solutions. Adsorbates used in this study included pimelic acid (C7H12O4; Aldrich) and poly(acrylic acid) (19) Specht, C. H.; Frimmel, F. H. Phys. Chem. Chem. Phys. 2001, 3, 5444. (20) Hug, S. J.; Sulzberger, B. Langmuir 1994, 10, 3587.
5804
Langmuir, Vol. 19, No. 14, 2003
(Aldrich; molecular weight ∼450 000, referred to subsequently as PAA450K). Each concentrated stock solution (1000-30 000 ppm) was prepared by the addition of powdered adsorbate to an appropriate aqueous solution followed by steady mixing on a stirrer table for several days to ensure optimal dispersion. The solution was made up to the required pH, then stirred for a further day before recording the final pH. Dilutions were made from these stock solutions as required. 2.2. Hematite Colloid Preparation and Deposition. An aqueous hematite colloid was prepared by mixing equal volumes of FeCl3 (0.01 M) and HCl (0.004 M) solutions, followed by refluxing for 48 h.21 Upon cooling, the colloid was dialyzed against ultrapure water for 7 days with the water changed daily. The resultant solution was analyzed for iron by inductively coupled plasma-atomic emission spectroscopy (using the Fe emission line at 259.94 nm), indicating a colloid concentration of 232 ppm. The colloid was found to have a very narrow particle size distribution, with d10, d50, and d90 measured by low-angle laser light scattering (Malvern Mastersizer) as 0.056, 0.096, and 0.152 µm, respectively. The colloid solids were shown by X-ray diffraction to be pure hematite, while transmission electron microscopy images confirmed the particle-size information and that the particles were monoclinic in shape. A thin layer of hematite was formed on the ZnSe crystal of the ATR accessory by carefully dispensing 30 µL (7 µg) of the colloid to a diameter of 5 mm, which was greater than the diameter of the analysis area of the infrared beam (3 mm), and allowing the slurry to dry in a desiccator. Optical microscopy (Nikon Optiphot microscope equipped with a Pulnix scan video camera and interfaced to Optimas 6.2 software) showed that a continuous uniform film was formed. Grey-scale analysis (ImageJ software) of the optical microscopy image, coupled with the height correlation of the gray scale with the film thickness measured by atomic force microscopy (Digital Instruments, Dimension 3000; scan area 80 × 80 µm; scan rate 2.289 Hz; setpoint 0 V), gave an average film thickness of 950 nm over the analysis area of the infrared beam. 2.3. Spectral Measurements. FTIR-ATR spectra were obtained using a Bruker IFS 66 instrument, mercury cadmium telluride detector, and Harrick “Seagull” variable-angle, singlebounce ATR accessory. Test solutions were pumped through the flow cell of the Seagull accessory to the waste by a peristaltic pump (Masterflex model 7550-92, pump head model 7518-10). The pump was calibrated before each use to accurately deliver 1.0 ( 0.02 mL min-1 by measuring the mass of water delivered per unit time. The total volume from the test solution to the waste was 3.05 mL (tubing: Masterflex Tygon tubing number 6409-13), and the volume of the flow cell was 0.75 mL. The ATR element in all the experiments was a hemispherical ZnSe crystal (Harrick Scientific Corporation). A spectral resolution of 4 cm-1 was used, and the number of scans accumulated ranged from 56 to 256 depending on the intervals between acquiring the spectra. The incident angle of the infrared beam was 45°. The background measurement for all the spectra was that of the clean, dry ZnSe crystal with air at the measured interface. A hematite film was then deposited on the ZnSe crystal, as was described in section 2.2, and its infrared spectrum was recorded. Bands attributable to carbonate were not observed, indicating that CO2 adsorption onto the film was not significant. The solution containing no polymer was introduced, and once equilibrium was established, a solvent spectrum was recorded. The solutions containing polymer were then introduced, and the sample spectra were recorded as a function of time. Spectra attributable to only the polymeric species were obtained by subtraction of the solvent spectrum from the sample spectra. Spectra representative of the concentrated flocculant solution species were acquired in a similar manner in the absence of the hematite colloid film. When the peak area rather than the peak intensity of the spectral bands is used, it was determined by a valley-valley peak area calculation using Bruker Opus version 2.0 software.
3. Results and Discussion 3.1. Solution Spectra. It was anticipated that the mode of adsorption of poly(acrylic acid) onto hematite could be
Kirwan et al.
Figure 3. Spectra of concentrated poly(acrylic acid) (PAA450K; 10 000 ppm) in solutions at pH 2 and 13, acquired by ATR with an uncoated ZnSe element.
determined by analyzing spectral changes in the adsorbate before and after adsorption. To achieve this, the chemical environment of the polymer in a solution was first characterized by recording the infrared spectra of concentrated polymer solutions using a clean, uncoated ZnSe crystal, onto which chemisorption was found not to occur. The carboxylate functionality of poly(acrylic acid) has characteristic absorption bands in the 1000-2000 cm-1 region of the infrared spectrum. Figure 3 shows the absorption bands of PAA450K in this region at pH 2 and 13, after subtraction of the bands associated with the solution media. Very high concentrations (>1000 ppm) were required to achieve detectable peaks using the uncoated crystal, and at 100 ppm of polymer or less, it was not possible to distinguish any polymer peaks in the spectra. Figure 3 shows that at pH 2 the characteristic stretching frequencies include the carbonyl stretch (CdO) at 1717 cm-1, CH2 stretching at 1455 cm-1, and the CsO stretch at 1265 cm-1. At pH 13, the characteristic stretching frequencies include the symmetric and antisymmetric stretching frequencies of the carboxylate ion (COO-) at 1408 and 1562 cm-1, respectively, and once again the CH2 stretching frequency at 1453 cm-1. Therefore, at pH 2, the carboxylate functional groups of the polymer are completely protonated, whereas at pH 13 the groups are fully ionized. These results complement those of Boisvert et al.,22 who found that poly(acrylic acid) neutralized through a pH range of 3-9. The positions of the spectral bands for a concentrated polyacrylate solution spectrum at pH 13 were not affected by the addition of an electrolyte (NaCl) to the polymer (21) Kan, S.; Yu, S.; Peng, X.; Zhang, X.; Li, D.; Xiao, L.; Zou, G.; Li, T. J. Colloid Interface Sci. 1996, 178, 673. (22) Boisvert, J.-P.; Malgat, A.; Pochard, I.; Daneault, C. Polymer 2002, 43, 141.
Examination of Poly(acrylic acid) Adsorbed on Fe2O3
Langmuir, Vol. 19, No. 14, 2003 5805
Figure 5. Schematic representation of the proposed mode of adsorption of poly(acrylic acid) onto hematite at pH 2.
Figure 4. Spectra of adsorbed PAA450K (50 ppm) on hematite at pH 2 for different contact times. Table 1. Comparison of FTIR-ATR Peak Assignments for Polyacrylic Acid PAA450K in Solution (10 000 ppm) at pH 2 and 13 and after Adsorption from a 50-ppm Solution onto Colloidal Hematite at pH 2 peak positions (cm-1) solution (pH 2)
solution (pH 13)
1717 1455
1562 1453 1408
1265 154
adsorbed (pH 2)
peak assignment
1716 1543 1455 1410 1275 133
sCdO (free COOH) sCOO- (antisymmetric) sCH2 scissor sCOO- (symmetric) sCsO ∆ν
solution over a range of concentrations (0-1 M). The spectrum shown for PAA450K at pH 13 in Figure 3 was, therefore, taken to be that of the polyacrylate sodium salt. 3.2. Adsorbed Species Spectra. Figure 4 shows the spectra obtained upon flowing a 50-ppm PAA450K solution over a deposited hematite film at pH 2 as a function of time. This concentration is well below the infrared detection limit for contact with an uncoated ZnSe ATR element (∼1000 ppm). As the polymer adsorbed and concentrated onto the surface of the hematite film, a signal attributable to the adsorbed polymer was detected, unbiased by the polymer in the bulk solution. A similar signal enhancement was described by Hug16 for the in situ FTIR-ATR study of the adsorption of sulfate onto hematite. The principle features of these spectra are summarized in Table 1, together with those for the corresponding solution spectra at 10 000 ppm. In the adsorbed species spectra, the occurrence of spectral bands at 1410 and 1543 cm-1 is characteristic of the symmetric and antisymmetric stretching frequencies of the carboxylate ion. These peaks are absent from the solution spectrum at pH 2 and, hence, indicate that they are primarily associated with the adsorption process. For the sodium salt of poly(acrylic acid) in solution, ∆νsalt was 154 cm-1, whereas ∆νadsorbed at pH 2 was significantly lower at 133 cm-1. On this basis, as was argued by Deacon and Phillips,11 an adsorbed species involving bidentate chelation (Figure 1, II) is proposed. Only a fraction of the polymer functionalities will actually adsorb onto the surface, with the remainder free to form loops and tails that extend into the solution. Other peaks present in the adsorbed species spectra that are also present in the solution spectra are characteristic of the carboxylic acid functionality (1717 and 1275 cm-1) and are due to the unadsorbed loops and tails. A schematic
representation of the adsorption of poly(acrylic acid) onto hematite at pH 2 is given in Figure 5. It is also possible to estimate the relative amounts of unadsorbed (loops and tails) and adsorbed (trains) polymer segments. The intensity of the antisymmetric stretch at 1562 cm-1 (0.012) in Figure 3 is due solely to polyacrylate in solution because its concentration dependence obeys Beer’s law. However, the intensity of the carbonyl stretch at 1717 cm-1 (0.006) is due to poly(acrylic acid) that is in solution (0.0036) and is weakly and reversibly physisorbed to the ZnSe crystal (0.0024) because its Beer’s law plot, though linear, does not pass through 0. The ratio of the molar coefficients of absorptivity of the 1562 and 1717 cm-1 peaks is thus 3.33:1. In Figure 4, it is clear that the carbonyl stretch at 1716 cm-1 (0.015) is three times the intensity of the antisymmetric stretch at 1543 cm-1. If it is assumed that the coefficient of molar absorptivity of the antisymmetric stretch is the same for both sodium polyacrylate in solution and the hematite-polyacrylate complex, then the ratio of bound to free is 1:10, suggesting that ∼9% of the polymer segments are adsorbed. Intuitively, this value seems high because a coiled, highmolecular-weight polymer would have the majority of its functional groups within the coils and, hence, not available for adsorption, unless it uncoiled significantly. Thus, the molar absorptivity assumptions made for the calculation are probably not valid. The shallow penetration of the ATR technique and exponential decay of the evanescent wave may also be contributing factors to the overestimation. The attenuation of the infrared radiation is biased toward the components closest to the hematite film. As a result of the size of the polymer molecule in a solution, it would be expected to only adsorb onto the outer surface of the cast hematite film, resulting in a signal biased toward the adsorbed parts of the polymer because the unadsorbed loops and tails of the polymer reside further from the hematite film and, hence, in a less-intense field of the evanescent wave. Also apparent in the adsorbed species spectra in Figure 4 is the increasingly negative peak at approximately 3300 cm-1. The bidentate chelate surface-adsorbed structure proposed requires that a carboxylic acid functional group becomes deprotonated in the process of binding directly to a surface ferric ion. A consequence of this process must be the removal of hydroxyl groups from the hematite surface. This may be achieved by protonation of surface hydroxyl groups that can then leave as water molecules, and such a loss would be expected to generate a negative signal in the OsH stretching region at 3300 cm-1. In addition, as the polymer adsorbs onto the surface, some of its nonbonded segments may displace the solvent from the surface, which may contribute a further signal loss. These findings are similar to those of Connor and McQuillan,15 who in a study of phosphate adsorption onto TiO2 attributed a similar negative peak (3150 cm-1) to the displacement of terminal hydroxyl groups from the
5806
Langmuir, Vol. 19, No. 14, 2003
Figure 6. Spectra of adsorbed of pimelic acid (100 ppm) onto hematite at pH 2 for different contact times.
TiO2 surface with a contribution also from the loss of adsorbed water molecules. To clarify both the origin of the negative hydroxyl stretching intensity and the adsorption mode proposed for poly(acrylic acid), pimelic acid [HOOC-(CH2)5-COOH] was adsorbed onto hematite at pH 2. It was considered that this molecule would mimic a small segment of poly(acrylic acid), with a chain length long enough to allow the molecule to bind through both carboxylic acid groups, if that proved most favorable. Again, there was no evidence of adsorption onto the uncoated ZnSe crystal, with pimelic acid concentrations in excess of 1000 ppm being required to give any detectable peaks in the solution spectra. The spectra of pimelic acid (100 ppm) adsorbed onto hematite at pH 2 (Figure 6) clearly show the appearance and growth of adsorbed carboxylate peaks. The intensity of these peaks was much lower than those seen in Figure 4, despite the concentration of pimelic acid being double that used for PAA450K. This most likely reflects the greater affinity of the longchain polymer for the hematite surface as a result of the potential for multiple binding to occur along the chain. Interestingly, the adsorption of pimelic acid onto hematite did not lead to the observation of a negative peak at approximately 3300 cm-1, as was the case for the much-higher-molecular-weight PAA450K. The implication from this result is that the negative peak in the adsorbed polymer specta was due to displaced solvent from the surface rather than displaced surface hydroxyl groups. This is likely because the adsorbed poly(acrylic acid) occupies a much greater volume (at the expense of solvent) within the interaction zone of the evanescent wave compared to pimelic acid, as is shown schematically in Figure 7. The principle features of the adsorbed species spectra are summarized in Table 2, together with those for the corresponding concentrated solution spectra (30 000 ppm) of pimelic acid at pH 2 and 13. As was the case for poly(acrylic acid), the spectra for pimelic acid exhibited peaks that are attributable to the symmetric and antisymmetric stretches of the carboxylate ion. The value of ∆νsalt (137 cm-1) for pimelic acid was significantly larger than the corresponding ∆νadsorbed (120 cm-1). This again indicates bidentate chelate complexation, supporting the proposed mode of adsorption shown in Figure 5. Furthermore, for adsorbed pimelic acid there was no signal due to a carbonyl stretch, suggesting that pimelic acid was bound to the hematite surface through both carboxylate groups. This supports the assertion that in the adsorbed polymer spectra the presence of the carbonyl stretching frequency is attributable to the unadsorbed loops and tails of the adsorbed polymer molecules.
Kirwan et al.
Figure 7. Diagrammatic representation of the large volume occupied by adsorbed poly(acrylic acid) in the interaction zone of the evanescent wave resulting in the detection of the solvent displacement. In comparison, adsorbed pimelic acid occupies a small volume and no solvent displacement is detected. Table 2. Comparison of FTIR-ATR Peak Assignments for Pimelic Acid in Solutions (30 000 ppm) at pH 2 and 13 and after Adsorption from a 100-ppm Solution onto Colloidal Hematite at pH 2 peak positions (cm-1) solution (pH 2)
solution (pH 13)
adsorbed (pH 2)
1544 1460 1407
1531 ∼1450 1411
137
120
1707 1460 1272
peak assignment sCdO sCOO- (antisymmetric) CH2 scissor sCOO- (symmetric) sCsO ∆ν
The mode of adsorption proposed for polyacrylic and pimelic acid adsorption onto hematite is similar to that described by Dobson and McQuillan17 in their in situ FTIRATR study of succinic and adipic acid adsorption onto TiO2 and Al2O3 from aqueous solutions. However, the results do differ from those of Specht and Frimmel19 for oxalic, malonic, and succinic acids adsorbed on kaolinite, which, at low pH, favored monodentate or monodentate pseudochelation complexation of oxalic and malonic acids, respectively, but could not distinguish between bidentate chelation (Figure 1, II) and monodentate complexation (Figure 1, I) for succinic acid. The monodentate oxalate binding structures proposed by Specht and Frimmel19 are consistent with those suggested by Hug and Sulzberger20 for oxalate adsorption on TiO2 and illustrate the importance of chain length to the adsorption mechanism. Increasing the carbon chain by one atom, that is, malonic acid, allows the second carboxylate to also bind, but to a different surface metal ion, to produce a pseudochelate and gain a stable sevenmembered-ring formation. The similar five-membered chelate complex with oxalic acid appears to form only at higher pH (>6.5). Increasing the chain by a further carbon, that is, succinic acid, leads to bidentate mononuclear chelation, stabilized by its four-membered ring, or pseudochelation from monodentate binding to two surface metal ion centers, which generates a less-rigid eight-membered ring. Specht and Frimmel19 could not distinguish between these two modes, but in view of the bidentate chelation found for adipic acid by Dobson and McQuillan17 and in this work for pimelic and poly(acrylic acid)s, it is likely that succinic acid also binds via bidentate chelation. 3.3. Adsorption Isotherm for PAA450K on Hematite. Obtaining data for the adsorption of poly(acrylic acid) onto hematite as a function of concentration is complicated by the irreversibility of polymer adsorption and the need for reproducible surfaces. To overcome the need to reproduce the surface a number of times, the adsorption data was obtained by flowing known concentrations of
Examination of Poly(acrylic acid) Adsorbed on Fe2O3
Langmuir, Vol. 19, No. 14, 2003 5807
Figure 8. PAA450K adsorption at pH 2 (monitored by the 1716 cm-1 peak area), demonstrating the equivalent adsorption achieved either by the direct application of a 50-ppm polymer solution or by sequential addition of an increasing polymer solution concentration up to 50 ppm.
Figure 9. Plot of plateau absorbance values from Figure 8 as a function of the polymer concentration for PAA450K adsorption at pH 2 and the subsequent Langmuir adsorption isotherm fit to the data.
polymer solution over the same film until equilibrium was obtained at each increased concentration. The validity of this approach was confirmed by the fact that sequential increases in the concentration gave a final plateau value almost the same as that of a single high-concentration solution onto a second, and similar, film (Figure 8). The plateau integrated absorbances obtained as a function of the polymer solution concentration for each adsorption-time profile are well-described by the Langmuir function (eq 1, Figure 9)
θ ) Ac/Amax ) Kc/(1 + Kc)
(1)
where K ) 0.236 and the fraction of coverage (θ) is the ratio of the integrated peak area (at concentration c) relative to that of the maximum integrated peak area observed. The Langmuir isotherm assumes that the sites where adsorption occurs are all thermodynamically equivalent and that the adsorption is independent of whether nearby sites are occupied. When the adsorption has reached equilibrium for a specific concentration, this normally indicates that the rate at which surface sites are filled due to adsorption is equal to the rate at which they are vacated by desorption. This interpretation of the equilibrium constant (K) cannot be applied to polymer adsorption because it is irreversible as a result of its many attachments to the substrate surface. Generally, adsorption isotherms for polymers that are described by the Langmuir function show almost no concentration dependence, which is characteristic of a high affinity between
the adsorbate and the substrate.23 For the case of poly(acrylic acid) adsorbed on hematite at pH 2, there is a significant concentration dependence, which, coupled with K < 1, suggests that the affinity is low. The isotherm plateau value may be used to estimate the maximum adsorption density if it is assumed that the hematite film acts as an extension of the ZnSe ATR crystal (which is reasonable because the evanescent wave penetrates it) and that all functional groups directly adsorbed to it will be detected. However, obtaining such an estimate is a far from trivial exercise, with significant errors potentially introduced from additional assumptions regarding the surface area, adsorbed layer thickness, depth of penetration of the IR radiation, and molar absorbance coefficients of the solution and adsorbed polymer. We are currently investigating a more rigorous model, which takes into account the refractive indices of the film and polymer solutions, and independently obtained coefficients of molar absorptivity. Adsorption-density measurements are even more difficult for pimelic acid because a small molecule would be expected to penetrate into the colloidal film, creating still greater uncertainty in estimates of its surface area. 4. Conclusion As an in situ technique, FTIR-ATR avoids the inevitable binding changes associated with ex situ sample preparation and, therefore, yields more reliable information specifically pertaining to the solid-liquid interface. Its application to poly(acrylic acid) adsorption on hematite has provided evidence of how such polymers adsorb at low pH. Poly(acrylic acid) is chemisorbed to the hematite surface at pH 2 via bidentate chelate complexation. The adsorption is irreversible and is driven predominantly by the large number of points of attachment generated by a single polymer molecule, as the Langmuir isotherm obtained suggests that poly(acrylic acid) has a low affinity for hematite. The technique was able to distinguish between the adsorbed and the unadsorbed segments of the adsorbed polymer molecules. Quantifying the fraction of adsorbed segments is problematic because a number of assumptions have to be made; the value of ∼9% obtained is likely to be an overestimate. It was also found that solvent was displaced from the surface as the amount of poly(acrylic acid) adsorbed onto hematite increased. This is attributed to the volume of the nonadsorbing segments of the polymer displacing solvent from within the measured interaction zone of the evanescent wave. Pimelic acid was found to complex to hematite in the same manner as poly(acrylic acid) (bidentate chelation) with adsorption to the hematite surface occurring through both carboxylic acid groups. There is no solvent displacement associated with this adsorption process as a result of the relatively small volume of the pimelic acid unadsorbed segment in comparison to that of adsorbed polymers. The results also indicated that the mode of attachment for polycarboxylate molecules at low pH was not dependent on the chain length. Acknowledgment. This research has been supported by the Australian Government’s Cooperative Research Centre (CRC) program, through the A. J. Parker CRC for Hydrometallurgy. This support is gratefully acknowledged. L.J.K. is grateful for the support of an Australian Research Council Postgraduate Award. LA027012D (23) La Mer, V. K.; Healy, T. W. Rev. Pure Appl. Chem. 1963, 13, 112.
