Langmuir 1990,6, 602-611
602
of flocs and using the Stoke's law that the effective density increases as size decreases:
- -
larger than the volume of silica in the flocs. A longrange organization of the small chaplet-like silica aggregates certainly exists, but it is not accessible by SAXS.
p = AL-" with a varying from 1.1to 1.5 depending on the nature Conclusion of the flocculants. The fractal geometry of flocs yields (1)The flocculation of silica colloids by aluminum polya = 3 - D,, where D,was varied from 1.9 to 1.5. In the cations or colloids brings about flocs with a fractal geompresent system (silica/aluminum), the parameter a varetry at the semilocal range order. One observe a linear ies according to the aluminum dosage from -0.70 to 1.20. relation between log I(Q) and log Q on approximately 1 More recently, Klimpel and H ~ g g and , ~ Hogg and Ray40 decade of distance. The slopes vary and depend on the showed from the flocculation of quartz with organic macphysical and chemical parameters of the different sysromolecules that the floc density-size relationships are tems. complex, depending on the size of flocs and the dynam(2) The fractal dimension and the density of the flocs ics in the mixing reactor. The fractal geometry is limdepend on (a) the chemical nature of the flocculant, i.e., ited in an intermediate floc size range from 50 to 500 its affinity for the surface of the silica colloids (0,and p pm. The density-floc size plots exhibit a power law which are the larger in the case of All, flocculant) (b) the pH corresponds to fractal dimension values of 1.7-2.1. The (hydrolysis), D, and p decrease as pH increases, and (c) small flocs are dense, and the large ones are very porous. the flocculant size ( p drastically decreases as the size of In our study, the variations of D,and p according to the flocculant colloids are large compared with that of the chemical nature of the aluminum flocculants (Al13 silica). and WAC) are in a large part due to the higher affinity of the Al,, species for solids as silicas or s i l i ~ a t e s ~ ~ , ~ ~(3) - ~The ~ mechanism of flocculation depends on the pH. At pH 4.5, a charge neutralization can be evoked yieldcompared with monomers or others kinds of aluminum ing dense flocs. At pH 7.5, the mechanism is charge neucolloids. Recently it has been shown that All, adsorbed tralization and interparticle bridging, and the flocs are onto silica is sufficiently chemically modified to exhibit less dense. an 27AlMAS NMR resonance at 54 ~ p m This . ~ could ~ be the origin of the variation of D, and p . In the particAcknowledgment. We thank the staff of LURE (P. ular case of R = 2.6 flocculant, one observed that the Vachette and J. P. Benoit). J.Y.B. thanks B. Montez density of the flocs was very low. It is certainly due to (Chemistry Dept, University of Illinois, Urbana-Chamthe large size of the flocculant particles compared with paign). WAC samples were supplied by ATOCHEM. This silica colloids. The volume occupied by aluminum is much work was supported by a research grant from GS "Traitement chimique des Eaux". (43) Changui, CH. Ph.D. Thesis, Universitb Catholique de Louvain-
-
la-Neuve, Belgique, 1988, p 165. (44) Fripiat, J. J. Personal communication.
Registry No. SO,, 7631-86-9; Al, 7429-90-5.
Protonation of Phosphate on the Surface of Goethite As Studied by CIR-FTIR and Electrophoretic Mobility M. Isabel Tejedor-Tejedor* and Marc A. Anderson Water Chemistry Program, 660 N . Park St., University of Wisconsin, Madison, Wisconsin 53706 Received May 26, 1989. I n Final Form: September 6, 1989 CIR-FTIR "in situ" spectroscopic studies have provided evidence for the formation of three different type of complexes, protonated and nonprotonated bridging bidentate as well as a nonprotonated monodentate, between orthophosphate ions and surface Fe(II1) of o-FeOOH particles in aqueous suspensions. The speciation of these complexes is a function of pH and phosphate surface coverage (I'). Furthermore, the combination of CIR-FTIR, adsorption isotherm, and electrophoretic mobility data allows us to calculate the intrinsic pK value (4.6) for the bridging bidentate iron phosphate surface complex. I. Introduction The scientific literature is replete with studies concerning adsorption, particularly phosphate adsorption, because of its importance in a variety of disciplines including soil
* To whom correspondence should be addressed. 0743-7463/90/2406-0602$02.50/0
fertility, eutrophication, corrosion control, et^.'-^ In addition, many adsorption or surface complexation models have emerged Over the past 2o years* These have (1) Parfitt, R. L.; Russell, J. D. J. Soil. Sci. 1977, 28, 297. (2) Lyklema, L. Enuiron. Sci. Technol. 1980, 14, 537. (3) Stumm, W.; Furrer, G.; Kunz, B. Croat. Chern. Acta 1983,585.
0 1990 American Chemical Society
Protonation of Phosphate on Goethite mainly attempted to interpret potentiometric titration results in oxide systems in terms of a complexation constant which may be perturbed by the electrostatic contributions of surface charge in some manner.495Unfortunately, direct analytical techniques for testing the validity of these hypothetical surface models have not been available until very r e ~ e n t l y . ~ , ~ We have been developing methodology for an "in situ" infrared technique which can be easily applied to aqueous suspensions and which we have used in the qualitative analysis of surface reaction^.^'^ This cylindrical internal reflection-FTIR (CIR-FTIR) technique is applied here to the phosphate-goethite (a-FeOOH) adsorption system in two ways. First, we use this technique to describe the phosphate surface species. Second, with results from adsorption experiments and electrophoretic mobility measurements, we utilize the CIR-FTIR method to describe the protonation of the phosphate species on the surface. In spite of the fact that single-crystal X-ray diffraction studies and Mossbauer spectroscopy led to the unambiguous characterization of some iron orthophophatesl' and provide very valuable descriptions of the phosphate tetrahedra, infrared data, which could otherwise be used to characterize surface species, have unfortuantely not been published for these compounds. The same can be said with respect to numerous investigations on iron(II1) phosphate complexes in aqueous solution, in whose characterization a broad variety of techniques have been used, such as ion exchange, magnetic susceptibility, and UVvis or stopped-flow s p e c t r o s ~ o p i e s , ~but ~ - ~once ~ again IR spectroscopy has not been employed in these investigations. Some authors14 have claimed to have identified by IR spectroscopy bridging bidentate complexes of orthophosphate with adjacent atoms of iron on the surface of wet goethite. However, these claims have been based on spectral data for complexes of Co(II1) whose structures were not unambiguously ~haracteri2ed.l~Even if the structure of cobalt complexes were correct, certain errors in band assignment and some extrapolations have led us to conclusion that, while there is not experimental evidence against the formation of these complexes, they are not the only possible explanation for the spectra of phosphate species a t the surface of goethite. Because of the paucity of IR data for the phosphate/ Fe(II1) species, we wished to establish a thorough spectral basis from which to assign bands belonging to phosphate goethite surface complexes in a more definitive manner. In order to do this, we had to develop good quality (4) Westall, J. C. In Geochemical Processes at Mineral Surfaces. ACS Symp. Ser. 1986,323,54. (5) Stumm, W., Ed.; Aquatic Surface Chemistry; Wiley: New York, 1987. (6)Tejedor-Tejedor, M. I.; Anderson, M. A. Langmuir 1986,2, 203. (7)Hayes, K.F.; Roe, A. L.; Brown, G. E., Jr.; Hodgson, K. 0.;Leckie, J. 0.;Parks, G. A. Science 1987,238,783. (8) Yost, E. C.; Tejedor-Tejedor, M. I.; Anderson, M. In situ study of bonding mechanisms of salicylate on aqueous goethite suspensions using CIR-FTIR. Enuiron. Sci. Technol., in press. (accepted for publication). (9)Tejedor-Tejedor, M. I.; Yost, E. C.; Anderson, M. Characterization of benzoic and phenolic complexes a t the goethite/aqueous solution interface using cylindrical internal reflection Fourier transform infrared spectroscopy. Part I Methodology. Langmuir, in press. (10)Bosman, W. P.; Beurskens, P. T.; Smits, J. M. M.; Behm, H.; Mintjens, J.; Meisel, W.; Fuggle, J. C. Acta Crystallogr. 1986,C42,525. (11) Wilhelmy, R. B.; Ramesh, C. P.; Matijevit, E. Inorg. Chem. 1985, 24,3290. (12)Filatova, L. N.;Galochkina, G. V. Russ. J . Inorg. Chem. 1978, 23,369. (13)Holrovd. A.: Salmon. J. E. J. Chem. SOC.1957.959. (14) Pa&, R. L.; Atkinson, R. J.; Smart, R. St. C. Proc. Soil Sci. SOC.Am. 1975.39.837. (15) Lincoln, S.'F.; Stranks, D. R. Aust. J . Chem. 1968,21, 37.
Langmuir, Vol. 6, No. 3, 1990 603 spectral data for uncomplexed phosphate species as a function of pH.' In this case, we were helped by the interpretation that Chapman et a1.16 made of the IR transmission data of concentrated aqueous solutions of phosphoric acid and alkaline orthophosphates. We next had to perform adsorption experiments under a variety of pH and phosphate adsorption levels to determine how spectral features would vary as a function of pH and surface coverage. Finally, we synthesized and analyzed by CIRFTIR some aqueous ferric phosphate species described in the literature. The combination of these spectral data enabled us to develop band assignment criteria for the interfacial phosphate complexes that limits the number of structures that can be assigned to a given set of spectral bands if not to clearly determine the exact one. Furthermore, we can show as well a definitive intrinsic pK value for phosphate deprotonation a t the surface and demonstrate how this is affected by the effective potential, which is in turn a function of the potential-determining H+ concentration in solution. 11. Experimental Section A. Goethite Properties. Goethite was prepared by hydrolysis of ACS grade ferric nitrate with NaOH (Fe/OH = 2).17 This solution was aged 2 days at room temperature and then alkalized to pH 12.5 with NaOH. The resulting precipitate was aged for 6 more days a t 60 "C. This suspension was washed with Milli-Q water until constant conductivity was reached in the supernatant. The suspension was then freeze-dried. TEM micrographs show particles having a rod-like morphology with approximate dimensions of 60 X 20 nm, and N, BET analysis indicates a surface area of 81 m2/g. Stock suspensions (30 g/L) were prepared as needed by resuspending the freeze-dried goethite in Milli-Q water and sonicating intermittently for 2 or 3 days to ensure particle hydration. B. Phosphate Adsorption Studies. 1. Sample Preparation. Batch equilibrium adsorption experiments were conducted at 25 "C for 20 h. Prior to adding any phosphate, aliquots of goethite suspensions (30 g/L concentration) were brought to a fixed ionic strength of 0.01 M NaCl and desired pH conditions by using microliter quantitites of 1 M NaCl and 0.01-1 M solutions of NaOH or HC1 and left to reach equilibrium for 24 h in a temperature-controlled shaker (New Brunswick). Next, a given number of microliters of 0.1 M NaH,PO, solution were added to vigorously stirred, preequilibrated goethite suspensions. After 1 h of equilibration, pH was measured and readjusted. When the measured pH values differed by more than a few tenths of a pH unit from the desired one, pH was readjusted again after 4 or 5 h of further equilibration. A few minutes before any kind of analysis was performed in these phosphate goethite samples, pH was measured again. 2. Adsorption Density Measurements. Phosphate adsorption was calculated from the difference between phosphate added and the one measured in the supernatant after 20 h of equilibration. T o separate the particles from bulk solution, samples were first centrifuged and then filtered through a 0.05-~mNucleopore filter. The phosphate concentrations were determined by transforming the orthophosphate into heteropoly molybdophosphoric acid and then reducing this compound with stannous chloride to form molybdenum blue. The intensity of the color is measured with a Varian DMS80 spectrophotometer at a wavelength of 660 nm.18 3. Electrophoretic Mobility. The electrophoretic mobility was measured a t several pH values at a constant ionic strength (0.01 M NaC1) and phosphate content (r = 190, 150, 100, and 50 pm/g) with a PenKem system 3000 electrokinetics analyzer. (16)Chapman, A. C.; Thirlwell, L. E. Spectrochim. Acta 1964,20, 937. (17)Atkinson, R. J.; Posner, A. M.; Quirk, J. P. J. Inorg. Nucl. Chem. 1968,10,2371. (18)Standard Methods for the examination of Water and Wastewater, 13th ed.; American Public Health Association: New York, 1971.
Tejedor- Tejedor and Anderson
604 Langmuir, Vol. 6, No. 3, 1990 To be able to perform mobility measurements, samples of phosphate-goethite prepared as described in II.B.l were diluted to a concentration of 1g/L with their own supernatant. The mobility data were converted to { potential by using the calculation method of O’Brien and Whitelg a t 25 “C and 0.01 M NaCl ionic strength. We made the assumption that the particles were spheres of 39-nm radius since this is the value for the hydrodynamic radius.” 4. CIR-FTIRAnalysis. Infrared spectra of both orthophosphate aqueous solutions and phosphate-goethite suspensions were recorded interferometrically with a Nicolet 60 SX Fourier transform infrared (FTIR) spectrometer and a HgCd-Te (MCT) detector. The sampling method was attenuated total reflection with cylindrical optics (CIR) (SpectraTech CIRCLE System) and a ZnSe crystal rod. We used a 47” angle of incidence for solutions and a 39’ angle for suspensions. Spectral resolution for aqueous solutions of phosphate was 4 cm-’ and for goethite suspensions 1 cm-’ unless otherwise indicated. Singlebeam IR spectra were the result of 2000 (phosphate solutions) or 6000 (goethite suspensions) co-added interferograms. The system cutoff is near 800 cm-’ for work in H,O and 700 cm-’ in D,O. All final spectra were the result of subtracting either the spectra of the suspension supernatants or the ionic strength and pH-adjusted Milli-Q water solutions (reference) from spectra of the suspensions or orthophosphate in solutions (sample), respectively. References and samples spectra were ratioed against empty cell spectra. Further description of this method is presented elsewhere (Tejedor-Tejedor and Anderson).‘ Dissolved orthophosphate in H,O or D,O was analyzed by using as reference the spectra of 1 M NaCl a t a specified pH and 0.1 M solutions of phosphate a t the same pH and ionic strength as the sample. Aliquots of phosphated goethite suspensions prepared as described in section II.B.l were used as samples in these CIR studies, and the supernatant obtained by centrifugation of another aliquot was used as reference. In D,O suspensions, goethite was washed once with D,O and centrifuged, and the supernatant was discarded and the goethite resuspended with fresh D,O to eliminate residual H,O (all D,O work was performed under dry N, atmosphere). From this moment, the preparation of the samples was the same as that in H,O. For the spectra of iron-phosphate complexes in solution reported in Figure Sa-e, the compositions of samples and references are as follow: (a) Sample 1.4 X M H,PO,, 8 X lo-, M Fe(ClO,),, 0.1 M HClO,, 0.25 M NaClO,; reference 1.32 x lo-’ M H,PO,, 0.1 M HClO,, 0.25 M NaC10,. (b) Sample 1.4 X lo-’ M H,PO,, 8 X M Fe(NO,),, 0.1 M HNO,, 0.25 M NaNO,; reference 1.32 X lo-’ M H,PO,, 0.1 M HNO,, 0.25 M NaNO,. (c) Sample lo-, M H,PO,, 2 x lo-, M Fe(NO,),, 0.1 M HNO,, 0.5 M NaNO,; reference 2 X lo-’ M Fe(NO,),, 0.1 M HNO,, 0.5 M NaNO,. (d) Sample IO-’ M H,PO,, 5 X lo-’ M FeCL,, 0.12 M HCl; reference 5 X lo-’ M FeCL,, 0.12 M HC1.
111. Results a n d Discussion A. Uncomplexed Phosphate in Aqueous Solution. Figure la-d shows the CIR-IR spectra of phosphate ions in aqueous solution at several pH values. The spectral changes are interpreted in terms of acid-base reaction equilibria:
H,PO, fl: H , P o ~ ~ 4- HPO:- 5 pod3- (1) where pK, = 2.2, pK, = 7.2, and pK, = 12.3.’, The spectrum of Figure ICis attributed to the HPOL’ ion since this phosphate anion is the predominant species at pH 9.2. The vibrational spectrum of this anion can be interpreted on the bases of a tetrahedral pentaatomic group O’PO,” (0’= oxygen bridging between P (19) O’Brien, R. W.; White, L. R. J. Chem. Sac., Faraday Trans. 2 1978, 74, 1607. (20) Hackley, V. A.; Anderson, M. A. Langmuir 1989,5,191. (21) Cotton. F.A.: Wilkinson. G. Aduanced Inoreanic Chemistry, 5th ed.; Wiley-Interscience: New York, 1988.
1.
m 4:
r
I
\ a
WAVENUMBERS(cm-1)
Figure 1. CIR-FTIR spectra of 0.1 M orthophosphate aqueous solutions at different pH values and Z = 1 M NaCl: (a) difference spectrum of (solution a t pH 1.3)-(solution a t p H 4.3), scale h,, (height a t 1006 cm-’) = 0.07 au; (b) pH 4.3, h10,5 = 0.15 au; (c) pH 9.2, h1077= 0.11 au; (d) pH 13.5, h,, = 0.11 au.
and H atoms, 0” = oxygen bonded only to the P atom), which has a C,, symmetry and thus six IR-active fundamental modes of vibration: ,(PO,”) (E), ,(PO,”) (Al), v(P-0’) (Al), &(PO,”)(Al), &(PO,”) (E),and p,(PO,”) (E).22 In addition, the phosphate ion should exhibit other modes due to the existence of the POH group such as ,(OH) and 6(POH).16 According to the frequency value trends for ZXY, tetrahedral molecules reported by Nakarnoto2, and the studies of Chapman et a1.16 on the IR transmission spectra of alkali metal phosphates in aqueous solution, bands observed at 1077, 989, and 847 cm-l in the spectrum of Figure ICshould be assigned to the modes ,(PO,”) (E),,(PO,”) (A,), and v(P-0) (A]), respectively. The OH stretching is masked by the strong absorption band of the water molecules in the region around 3000 cm-l, the s(POH) mode appears as a very weak band near 1250 cm-l, and the rest of the vibrational modes should appear below 800 cm-l, a spectral region which is opaque to our system. The CIR-IR spectrum of an orthophosphate aqueous solution at pH 4.3, Figure lb, exhibits bands a t 1155, 1075,940, and 874 cm-’. Since at this pH value the predominant species is the H,PO,-l ion, which has a Czu symmetry,16 these absorption bands can be assigned to the vibrational modes ,(PO”) (B&, ,(PO”) (A]), ,(PO’) (BJ, and ,(PO’) (B,),22respectively. To isolate the absorption bands produced by the H,PO, molecule, the spectrum of an orthophosphate aqueous solution at pH 4.3 has been subtracted from the spectrum of a solution a t pH 1.3. The spectrum coming from this difference is shown in Figure la. The reason for this procedure is that at the latter pH value a significant amount of (OH)zPOz-l ions are still in equilibrium with the phosphoric acid. Furthermore, working a t a lower pH value, which would admittedly isolate the neutral H,PO, molecule, would result in damage to the ZnSe crystal. The band at 940 cm-’, in the spectrum of Figure Ib, was used to normalize the quantity of (OH),P02-1 at both pH values. The H,PO, is a tetrahedral molecule that, even in aqueous solutions, can be considered as having a
Langmuir, Vol. 6, No. 3, 1990 605
Protonation of Phosphate on Goethite
h
I\
14b0 12'72 11'44 10'18
z 0'88
7'80
WAVENUMBERS(cm")
Figure 2. CIR-FTIR spectra of 0.1 M orthophosphate in D,O a t different pD values and I = 1 M NaCl: (a) difference spectrum of (solution at pD = 1.5)-(solution a t pD = 5), hi013 = 0.02 au; (b) pD = 5.0, h,,,, = 0.10 au; (c) pD = 9.5, h,,,, = 0.09 au.
C,, symmetry; thus the band a t 1174 cm-' should be assigned to the P-0 stretching mode, the one a t 1006 cm-' to ,(PO,') (E), the very weak band a t 890 cm-' to ,(PO,') (Al), and the shoulder near 1250 cm-l to the inplane POH bending. Chapman1' reported lower frequency values for the P=O stretching (1160 cm-') and a band a t 1070 cm-' that is not present in the spectrum of Figure la. We believe that the presence of this band is due to some H 2 P 0 c 1 in the aqueous solution of HsPO,, and not to a reduction of symmetry due to hydrogen bonding as has been proposed. This is not surprising, since in the studies of Chapman no control over the pH values of the phosphate solutions has been reported. It may well be that in some cases these solutions are not far away from the pK values as to totally isolate the ion under study. When the phosphate solution pH is 13.5, the spectrum exhibits only one band a t 1006 cm-' (see Figure Id), as expected from POL3 ion which has a Th symmetry." Because the hydrogen bond effect is equal for all these anions, and since we work in relatively dilute solutions (0.1 M), the band frequencies for the ,(PO,") groups are one indication of the P-0 bond order. The IR spectra show that the bond order decreases with increasing value of n. The CIR spectra of orthophosphate in D,O solutions a t pD values of 5 and 9.5 are represented by Figure 2b and 2c, respectively. The main differences with respect to the spectra of orthophosphate in H,O solutions at equivalent pH values (Figure l b and IC)are related to the decrease in the frequency of the in-plane bending mode from 1200 cm-' to 850 cm-' on deuterating the hydroxyl group. Because this deformation mode can couple with the PO," degenerated stretching and the PO' stretching m o d e s of t h e HP04'- ion (1077 and 847 cm-' in Figure 1c),16on deuteration the highest frequency mode will be free from coupling and its frequency will increase (1089 cm-' of Figure 2c), while the lowest frequency mode will couple with 6(POD) giving a doublet (861-833 cm-').16 The same type of argument can be applied to explain (22) Nakamoto, K. Infrared and Raman Spectra of Inorganic and
Coordination Compounds; Wiley: New York, 1986.
3
h
31%
2580
1970
1360
750
WAVENWBERS( Un-')
Figure 3. General features of CIR-FTIR spectra of phosphated a-FeOOH (goethite) in H,O/D,O suspensions (spectrum of suspension - spectrum of supernatant): (a) in H,O, pH 5, 190 Mmol of phosphate/g of goethite (I?), h,,, = 0.08 au; (b) in H,O, pH 7 , r = 165 pmol/g, hag,= 0.13 au; (c) in D,O, pD 8.5, r = 140 kmol/g, h,,, = 0.30 au; (d) in D,O, pD 5, r = 165 pmol/g, h,,, = 0.05 au. The IRE is ZnSe, 6 = 39O, I = lo-' M NaC1, and goethite concentrations are approximately 60 g/L.
the differences between the spectra of Figures l b and 2b. However, some of the increase in the frequency of the bands when passing from H20 to D20 solutions may be due to the fact that hydrogen bonding is weaker in the last ~olvent.'~The spectrum of Figure 2a has been obtained in the same way as the one in Figure la. The deuterated phosphate exhibits the 6(POD) mode in the region of 900 cm-', and the two strongest bands (1201 and 1013 cm-') are at higher frequency values than the normal phosphoric (1174 and 1006 cm-I), although the position of the P=O vibration cannot be accurately determined in D,O since it is very close to the strong absorption band produced by the bending mode of this solvent. B. Characterization of Orthophosphate/a-FeOOH Surface Complexes. 1. Spectra of Phosphate-Goet h i t s Aqueous Suspensions. a. General Description. The spectra of the chemical species a t the particle/ aqueous solution interface can be best observed by subtracting the bulk solution (supernatant) absorption from the spectrum of the corresponding suspension.6 The difference spectra resulting from this subtraction contain absorption bands due to both particle bulk and interfacial species. Figure 3a-d illustrates these IR spectra for suspensions of goethite in H,O and D,O solutions, to which orthophosphate ions have been added. Spectra shown in Figure 3b and 3c are representative of systems with highly charged particles, whereas spectra in Figure 3a and 3d are characteristic of goethite suspensions a t pH values close to the isoelectric pH (pHiep), where the particles have little or no charge. Bands at 3125, 3156 (only in Figure 3c and 3d), 894, and 800 cm-' are assigned to subsurface lattice goethite The bands at 3113-2970 and 1642-1592 cm-' in Figure 3a and 3b and 2180-2500 and 1209-1184 cm-' in Figure 3c and (23) Pinchas, S.; Laulight, I. Infrared Spectra of Labeled Compounds; Academic Press: London, 1971. (24) Cambier, P.Clay Miner. 1986,21, 191.
606 Langmuir, Vol. 6, No. 3, 1990
Tejedor-Tejedor and Anderson
z200F-F
?------
Z150 e 1 -
0
2,000
4,000
6,000
umol/L 1200
11'12 id24
936
k 8
WAVENLIMBERS(m-l)
Figure 4. Effect of phosphate surface coverage on the absorption bands of interfacial phosphate at pH 4.5: (a) r = 190 pmol/ g, h,, = 0.007 au; (b) r = 150 pmol/g, h, = 0.002 au; (c) r = 100 pmol/g, h,,,, = 0.004 au. I = &aCl, and goethite concentrations are approximately 60 g/L.
d C
b
a
1200
11'12
10'24
636
848
WAVENWB€RS(cm-l)
Figure 5. Influence of pH on the CIR-FTIR spectra of interfacial phosphate: (a) r = 190 pmol of phosphate/g of goethite, pH 4.0, hlr .= 0.006 au; (b) r = 190 pmol of phosphate/g of goethite, p 5 0, h,, = 0.005 au; (c) r = 190 pmol of phosphate/ g of goethite, pH 6.0, h,,, = 0.005 au; (d) r = 150 pmol of phosphate/g of goethite, pH 5.0, h, = 0.003 au; (e) F = 150 pmol of phosphate/g of goethite, PI-! 8.4, h,,,, = 0.014 au. I = M NaC1, and goethite concentrations are approximately 60 g/L.
3d are assigned to OH/OD surface groups and interfacial H,0/D,0.6 The small bands observed in the region 1200-900 cm-l correspond to the interfacial orthophosphate. In Figures 4 and 5 , we examine the absorption bands of interfacial phosphate in the spectral region from 1200 to 900 cm-'. Figure 4 shows spectra for suspensions of goethite a t the same pH but having different amounts of adsorbed phosphate. The spectra in Figure 5 correspond to samples having the same amount of phosphate but differing in pH value. Differences in position and number of absorption bands between the spectra of orthophosphate in aqueous solution and of orthophosphate a t the goethite/aqueous solution interface for any given pH value (compare spectra of Figure la-d with the spectra of Figures 4 and 5) indicate phosphate coordination with
Figure 6. Adsorption isotherms of orthophosphate on goethite at different pH values: ( 0 )pH 8.4; (A)pH 7; (+) pH 6; ( X ) pH 5; (+) pH 4. Z = M NaCl, and goethite concentrations are 30 g/L.
the surface iron atoms of the goethite particles. It should be noted that the spectra shown in these two figures are the prototypes of the entire set of investigated conditions for all levels of phosphate adsorption and changes in pH. Figure 6 describes the functional dependence of phosphate adsorption on its equilibrium solution concentration and pH. We see the well-established fact that phosphate adsorption decreases with increasing pH. Furthermore, the maximum adsorption for this system seems to be approximately 200 pmol/g. From the visual observation of CIR-IR spectra of aqueous suspensions of goethite particles loaded with different quantities of phosphate, varying from a quarter of a monolayer ( ~ 5 pmol/g 0 of goethite) to a complete monolayer ( ~ 2 0 0pmol/g of goethite), and pH values ranging from 3.5 to 8.4, one notes the existence of two different phosphate surface complexes. Lower pH values and high phosphate coverages favor the formation a phosphate surface complex with adsorption bands a t an average frequency (see Tables I111) of 1123 and 1006 cm-'; meanwhile, the species responsible for absorption bands at approximately 1096 and 1044 cm-' is the predominant one at higher pH values and/or lower phosphate surface coverage. The spectra in Figure 4a-c and Figure 5a-e illustrate these observations. In the case of a-FeOOH suspended in D,O solutions, the first set of bands appears a t approximately 1120 (overlapped with the bending mode of D20) and 1005 cm-' and the second set a t 1090 and 1028 cm-l. These adsorption bands have almost the same frequency values in both H 2 0 and D,O, indicating they are not related with P-OH(D) bending modes of vibration. b. Mathematical Analysis of the Phosphate Bands. A more detailed analysis of the phosphate adsorption bands in the spectral region 1200-900 cm-I has been carried out by using the program CAPR (Nicolet). This program allows us to define up to five Gaussian/Lorentzian peaks which can then be fitted to a sample spectrum. Tables I, 11, and I11 contain the information about the number of curves and the position and the relative intensity that resulted in a good match for each interfacial phosphate spectrum, a t phosphate levels of 190,150, and 100 pmol/g and pH values ranging from 3.5 to 8.3. The integrated absorption intensity in the spectral region 1180-940 cm-', A,, normalized with respect to integrated absorption intensity between 940 and 835 cm-' (absorption band due to goethite), A,, and with respect to phosphate content, I'/lOO, has a value of 0.16 f 0.04 with randomly distributed deviations (see fifth column of Tables 1-111). Since A , has the same value for all the
Langmuir, Vol. 6, No. 3, 1990 607
Protonation of Phosphate on Goethite Table I. Results from Mathematical Analysis of Absorption Bands of Interfacial Phosphate. computed peaks
r, pmol/g 190
pH 3.6
Y,,
cm-' AJAT 1125 ~ _ 1090 1038 1007 975
_
X
38.2 . .. ~
100 (AT X 100)/(A,,
0.20
4.5
1126 1095 1045 1005 978
33.6 9.1 12.3 38.5 9.6
0.20
1127 1106 1043 1006
27.3 18.6 18.8 35.3
0.18
1126 1098 1043 1007 988
22.5 22.9 21.7 24.0 6.6
0.17
1128 1099 1045 1030 1004
18.9 28.9 27.4 1.6 23.1
0.17
1125 1097 1046 1023 1002
12.1 37.3 30.3 4.2 16.1
0.17
6.5
I')
a At different pH values and r = 190 pmol of phosphate/g of goethite. Table 11. Results from Mathematical Analysis of Absorption Bands of Interfacial Phosphate" computed peaks r, pmol/g pH vmSx, cm-' AJAT X 100 (AT X 100)/(A,, X I')
150
3.6
1123 1096 1043 1006
22.3 12.7 13.3 51.7
0.19
4.2
1123 1095 1042 1004
30.9 12.4 17.2 39.4
0.26
4.7
1120 1096 1043 1005
25.9 14.4 26.2 33.4
0.12
5.0
1123 1098 1043 1006
18.1 29.3 23.3 27.7
0.12
1127 1099 1044 1023 1000
17.6 26.9 27.6 25.4 22.5
0.17
1093 1048 1026 1001
37.8 43.3 12.1 6.7
0.19
6.0
140
8.0
r, rmol/g 100
PH urn-, cm-' A ~ / A TX 100 (AT X 100)/(A,, 3.7
1116 1094 1040 1004 980
10.9 22.5 29.6 33.0 4.2
0.15
4
1121 1098 1041 1002
16.6 24.5 31.3 27.6
0.19
4.3
1120 1097 1042 1005
13.8 27.3 35.3 23.7
0.16
4.7
1122 1096 1043 1006
7.7 42.7 36.1 13.5
0.18
5.0
1122 1095 1042 1006
6.6 33.0 51.0 9.4
0.16
6
1097 1045 1025 996
45.6 38.1 8.5 8.0
0.15
7
1092 1046 1026 1001
42.2 36.19 14.3 7.0
0.19
8
1088 1045 1023 1001
37.0 42.5 16.8 5.8
0.22
4.0 4.3 42.3 11.4 34.6 8.8 8.4 28.8 19.4
6
X
~~
1125 1093 1044 1010 984
5.5
computed peaks
0.13
4.2
5.2
Table 111. Results from Mathematical Analysis of Absorption Bands of Interfacial Phosphate.
a At different pH values and goethite.
r)
phosphate/g of
0 0
13'00
a A t different pH values and I' = 150 pmol phosphate/g of goethite.
r = 100 pmol
X
12'00
11'00 io00
goo
aoo
WAVENUMBERNcm-')
Figure 7. CIR-FTIR spectrum of phosphated goethite showing the base lines and the spectral region used in the calculations of the integrated intensities of both interfacial phosphate and goethite absorption bands.
studied samples, independent of which is the predominant phosphate surface species, the integrated absorption coefficients for all these species should be the same. Figure 7 shows the type of base line used to deconvolute and to calculate the integrated absorption for the phosphate absorption bands in the region 1180-900 cm-', as well as the base line used to determine the area under the 900-cm-' goethite band. Although it would be pref-
608 Langmuir, Vol. 6 , No. 3, 1990
erable to analyze the entire region between 1180 and 830 cm-', using the line joining these two points as the base line, this is not possible with our CAPR program. This is due to the fact that the CAPR program does not allow us to define more than five curves, and the number of processed spectral data points cannot be larger than 512. Since our interest is to measure only the relative intensity of the absorption bands of phosphate in this spectral region, the approach is valid. Deconvolution of the original spectra into the individual components not only makes it possible to measure the percentage of each type of phosphate surface complex for different values of the parameter r and pH, but it, has revealed the existence of a third phosphate surface species at pH values above 6, which absorbs near 1025 and 998 cm-'. Furthermore, this species appears to increase in concentration with increasing pH (see Tables 1-111). 2. Band Assignment. a. Spectral Region. From an analysis of the spectra of ionic phosphate in aqueous solution, we find that the spectral region for structural diagnosis of metal-orthophosphate complexes in aqueous solution is from 1200 to 800 cm-'. However, in the case of phosphate complexes at the surface of a-FeOOH, the spectral features appearing in the region 1200 cm-' to approximately 940 cm-' will be the only ones used in structural studies, due to the presence in the spectra of two strong a-FeOOH absorption bands at 900 and 800 cm-' (see Figure 3). b. Vibrational Modes. Both the data from Chapman" and the spectra in Figure 1 for ionic phosphates show that the P-0-H bending vibration (1200 cm-') appears as a very broad and weak band. Although this band would yield the most direct information on protonation/ deprotonation of orthophosphate complexes, for small concentrations of phosphate complexes in the sample, this absorption band would be undifferentiated from the spectral background. Hence, structural information on phosphate surface complexes is limited to that provided by the assignment of the spectral bands to some of the following groups of vibrations: PO, and P-OX (X = H or M) stretching modes of the XOPO, ions (C,, symmetry), the two PO, stretching and the P(OX), u(BJ modes of the ions (XO),PO, with C,, symmetry, and P=O stretching and P(OX), u(E) modes of the (XO),PO groups. c. Assignment Criteria. PO, Group. The IR sp c tra of monodentate complexes of the type MOP0,'3-3' (where r = charge of the metal) in the 1180-940-cm-' region should be similar to the one of Figure ICbut with frequency values for the ,(PO,) vibrations lower than 1077 and 989 cm-' and for u(P0M) higher than 847 cm-l, since, in general, metal ions are not so strongly coordinated to oxygen ions as protons. In this regard, the Co(NH,),PO4 complex, which is described in the literature as a monodentate, mononucleate specie^,'^ has phosphate absorption bands a t 1030, 980, and 934 cm-'. PO, Group. Orthophosphate-metal complexes with PO, groups, (MO),PO, (C2usymmetry) or (MO)(OH)PO, (C, symmetry), should have an infrared spectra similar to that of (OH),PO, in the sense that two sets of doublets are present. The higher frequency one, due to the stretching modes of PO,, will fall between the frequency values of the vibrations u ( ~ ) ( P O of J monodentate complexes and u(P=O) of the (MO),(OH),,PO type; the other, due to the stretching modes of P-(OX),, will appear a t lower energies than either one of the vibrations (25) Kwak, W.; Pope, M. T.Inorg. Chem. 1976, 15, 1732.
Tejedor- Tejedor and Anderson u
(PO,) of monodentate or the highest u(P-OX) in the
c~%),Po complexes.
P=O Group. Complexes of the type OP(OM),, OP(OM),(OH), or OP(OM)(OH), should produce IR spectra in which an absorption band (assigned to the u(P=O) mode) appears at much higher frequency than the rest, similar to what happened in the spectrum of &PO4 (see Figure la). However, the separation between the highest frequency band and the second highest should become smaller with an increasing number of (OM) groups and in all cases smaller than in the H,PO, spectrum. This criterion is substantiated by the reported IR data for P,Mo,O,,~-,~~ in which phosphate exhibits absorption bands at 1114, 1038, and 985 cm-' and, according X-ray crystal analysis, is coordinated to three different Mo atoms.26 Since OP(OM),(OH) or OP(OM)(OH), complexes have a C,, symmetry, they should exhibit four stretching bands in place of three and, in theory, should be distinguishable from the OP(OM), ones. d. Surface Iron-Phosphate Complexes, Since, in the presence of a-FeOOH, it it very difficult to experimentally observe any phosphate absorption bands below 950 cm-', we are constrained in making structural interpretations to the criteria exposed above, based upon the relative position of only the two higher frequency bands in the spectra. Tables I, 11, and I11 show the existence of three independent sets of bands for the different experimental conditions with the following average frequency values: 1123, 1006, and 982 cm-l; 1096 and 1044 cm-'; 1025 and 1001 cm-l. The first set of bands suggests the presence of a phosphate complex of the type (XO),PO. The possibility of all three X groups being H is evidently discarded because it does not correspond to the case of phosphoric acid (Figure la). We can also discard the possibility of a tridentate complex. This latter option would require the participation of two crystallites of goethite in forming the complex. This would result in a drastic decrease in surface area of the samples, which is not experimentally observed. We are therefore left with the possibility that this set of bands corresponds to either (FeO),(OH)PO or to (FeO)(OH),PO. Although the position of the bands would be different for the two species, it is very difficult without further information to safely deduce which one corresponds to this set of bands. In spite of this uncertainty, we favor the bidentate complex because 1123 cm-l seems somewhat too low to be assigned to a P=O stretching in a phosphate group with two OH ligands (see Figure 1) and also because of the information from infrared data of ferric orthophosphate complexes in aqueous solution, which we will discuss below. The absorption bands at 1096 and 1044 cm-l should be assigned to a PO, group since it is a doublet with frequency values between u(P=O) (1123 cm-') and u(P-OFe) (1004 cm-') mentioned above. Therefore, these bands should be attributed to either the (FeO),PO, or to the (FeO)(OH)PO, complexes. Finally, based on the criterions stated above, the bands at 1025 and 998 cm-' should be due to the MOPO, complex. 3. IR Spectra of Complexes of Iron(II1) and Orthophosphate in Aqueous Solution. In order to differentiate by IR spectroscopy a monodentate diprotonated phosphate complex from a bidentate monoprotonated complex or a monodentate monoprotonated from a bidentate complex, we have prepared aqueous solutions of ferric (26) Strandberg, R.Acta Chem. Scand. 1973, 27, 1004.
Langmuir, Vol. 6, No. 3, 1990 609
Protonation of Phosphate on Goethite
Table IV. Band Assignment for Identified Phosphate Complexes in the Surface of Goethite surface complexn =Fe,HPO,*"
symmetry
c,
=Fe,PO,-Y
C,,
=FePO,-*
C,,
observed frequencies, cm-' assignment 1120-1128 (avg 1123) u(P=O) 1010-1004 (avg 1006) u,(P-OFe) 988-975 (avg 982) u,(P-OFe) or u(P-OH) 1106-1088 (avg 1096) u(P-0) (A,) 1048-1038 (avg 1044) u(P-0) (B,) 1026-1023 (avg 1025) u(P-0) (E) 1002-996 (avg 1001) u(P-0) (A,) and/or v(P-OFe) (A,)
= x , y , and t are functions of residual charge in surface iron
atoms.
1&
11'85
1.070
d55
$40
WAVENUMBER (all-')
Figure 8. CIR-FTIR spectra of Fe(II1)-phosphate complexes in solution: (a) 1.4 X lo-' M H3P0,, 8 X M Fe(ClO,),, 0.1 M HClO,, 0.25 M NaClO , h, = 7 X lo-, au; (b) 1.4 X lo-' M H,PO,, 8 x IO4 M Fe(hOJ,, 0.1 M HNO,, 0.25 M NaNO,, Hlo = 5 X au; (c) lo-* M H,PO,, 2 X lo-' M Fe(NO,),, 0.1 HNO,, 0.5 M NaNO,, h,, = 0.010 au; (d) lo-' M H,PO,, 5 x lo-' M FeCL,, 0.12 HCl, hllZ6 = 0.012 au; (e) same as part c at pH 1.9.
k
d
salts and phosphoric acid, under the conditions reported in the literature to form either FeH,PO,,+ (ref 11) or Fe2P0,3+ (ref 13 and 27). Monodentate complexes have been identified in aqueous solutions when the iron/phosphate ratio is smaller than unity, and their concentrations are low enough so as to avoid dimerization,. The spectra of Figure 8a and 8b correspond to solutions in which the total concentration of Fe(II1) is 8 X M, the concentration of orthophosphate is 17 times higher, and the pH = 1 in HC10, and HNO,, respectively. The S / N ratio in these spectra is very small, but the presence of absorption bands a t 1150 and 1085-1050 cm-' is very consistent for both perchloric and nitric solutions. Although the best characterization of species has been done in perchloric medium," due to the existence of an absorption band of perchlorate a t 1095 cm-l we have repeated the experiment, substituting this ion for NO3-. In this way, we can evaluate the correctness of the spectral subtraction of this perchlorate band from the spectrum of the sample. Without any information concerning the structure of the predominant ferrimrthophosphate complex in these solutions other than the infrared spectra, one could easily postulate the formation of a FeHPO,l+ complex, since this spectrum is very similar to the spectrum of H2P0,'(Figure lb) and assign the 1150- and 1085-1050-~m-~ bands to the PO, stretching modes. Nevertheless, previous results from UV, thermodynamic, and kinetic studies provide some proof for the FeH,PO,'+ complex. In this case, the higher frequency band should be assigned to u(P=O) and the lower frequency one to v(P-OFe). Since the coordination of Fe(II1) to H,PO,'- does not have any noticeable effect on its IR spectrum, the bond should have a highly ionic character. The spectrum shown in Figure 8c corresponds to aqueous solutions (2 X lo-' M) of ferric nitrate, a Fe/ (27) Filatova, L.N.; Chepelevetskii, M. L.Ross. J.Inorg. Chem. 1966,
II,88.
phosphate mole ratio of 2, and pH values between 0.6 and 1.0. Solutions 5 times more concentrated in ferric chloride (5 x lo-, M) than in phosphoric acid and pH values close to 1.0 have an IR spectrum identical with the one in nitrate medium a t the same pH value (see Figure 8c and 8d). Salmon and c o - ~ o r k e r s 'suggested ~ the formation of Fe,P0,3+ in the experiment with ferric chloride solutions at pH values below 1.0. Meanwhile, Filatova et alez7reported the formation of a bridging bidentate complex of the type FezHP0:+ for nitrate solutions. Although the presence of a bridging bidentate complex is solidly supported by experimental data in both ferric nitrate and ferric chloride solutions, there is not any valid evidence on whether or not this complex is protonated. The presence of two adsorption bands a t 1122 and 1042 cm-' in the IR spectra of these solutions (Figure 8c and 8d) strongly suggests a deprotonated bridging bidentate complex, and the absorption bands should be associated with the stretching modes of the PO, group. Increasing the pH of the ferric nitrate solutions to 1.9 causes the profile of the spectrum to change to the one in Figure 8e, where phosphate absorption is represented by a unique very broad band. This band seems to be the envelope of a t least three bands centered near 1120, 1080, and 1045 cm-l and can be described as a combination of the spectrum of Figure 8c and the spectrum of the interfacial deprotonated phosphate complex of Figure 4c. Filatova et al." detected a conversion of the bridging bidentate complex into a hydroxophosphate of the type Fe,(OH)PO," a t pH 21.7. The substitution of one H,O molecule by an OH in the coordination sphere of the Fe(II1) reduces the residual charge of this cation and makes these complexes more similar to the one formed with the surface iron of goethite. We believe that the studies of iron phosphate in aqueous solution allow as to do a more definitive identification of the phosphate surface species since they clearly establish that a set of two bands between 1120 and 1040 cm-' belongs to bridging bidentate complexes, and the absorption bands for monodentate protonated ones appear a t higher wavenumbers. Table IV summarizes our findings and lists the observed frequencies and band assignment for the three identified phosphate complexes on the surface of goethite. C. Surface Protonation. As mentioned before, the Fe,HPO, complex (1123- and 1004-cm-' absorption bands) is the predominant species from pH values of 3.6 to about 5.5-6.0 in the case of surface coverages of 190 and 150 pmol/g. Meanwhile, the Fe2PO, complex responsible for the 1096- and 1044-cm-' bands becomes the main species between pH values 6-8.3. For r = 100 pmol/g, both
Tejedor- Tejedor and Anderson
610 Langmuir, Vol. 6, No. 3, 1990
Table V. Protonation Parameters of Phosphate Complexes on the Surface of Goethite mobility,
r, pmol/g 190
4
3
5
6
7
8
Figure 9. Electrophoretic mobility data for the phosphategoethite system. Each curve represents a constant r value (pmol of phosphate/g of goethite): ( 0 )r = 50; (+) r = 100; ( X ) r = 150; (A) r = 190. Z = lo-’ M NaCI, and goethite concentrations are 1 g/L.
type of complexes are present on the goethite surface and additionally are equal in concentration a t a pH value of 3.7. At higher pH values, the Fe,PO, complex becomes the predominant phosphate surface species. This complex is the only other phosphate species present a t the surface for these r values and pH values below 6.0. Meanwhile, the Fe,PO, complex is the predominant species at lower surface coverages (I? = 50 pmol/g) over the entire pH range investigated. This species increases in concentration with increasing pH. Therefore, changes in speciation of the phosphate surface complexes as a function of pH can be interpreted in terms of the following acid-base reaction: (FeO),(OH)PO 5 (Fe0),P02 + Ht
(2)
The apparent pK (the pH value a t which the concentration of the surface species in equilibrium is equal) for this reaction is a function of the amount of phosphate on the surface. Tables 1-111 show pK 6 for r = 190 pmol/g, pK 5.5 a t r = 150 pmol/g, pK 3.7 at I’ = 100 pmol/g, and a pK value below 3.7 for r = 50 pmol/g. The monodentate nonprotonated complex, FeOPO,, starts being detected a t pH 6 for r = 190, 150, and 100 pmol/g, although the percentage of this species at this pH value increases as r decreases. It seems that, at r = 50 pmol/g, FeOPO, is already present at pH values as low as 4. Unfortunately, the spectra for this surface coverage are too noisy to be able to make a definitive interpretation of the results coming from the deconvolution of the spectral bands. The electrophoretic mobility qesults plotted against pH are shown in Figure 9. This figure contains four curves, each of which represents a constant value for r (50, 100, 150, and 190 pmol/g). With increasing coverage, the curves shift to the left and have differing isoelectric pH values. The isoelectric pH (pHiee)for goethite, in the absence of phosphate (r = 0) and in the presence of atmospheric CO,, is around pH 9.7 as illustrated in a recent paper by Zeltner and Anderson.28 With increasing phosphate coverages, the pHiePshifts from 9.7 at r = 0, to 7.0 a t r = 50, to 5.3 a t r = 100, to 4.4 a t r = 150, and finally to 3.9 a t r = 190 pmol/g. Surface charge is both a function of pH and phosphate a d ~ o r p t i o n .The ~ ~ isoelectric pH pro-
-
--
(28) Zeltner, w. A.; Anderson, M. A. Langrnuir 1988,4,469. (29) Anderson, M. A.; Malotky, D. T. J. Colloid Interface Sci. 1979, 72,413.
~Hbuik
+1.10
0.45 0.37 0.16 -0.03
+0.50 -0.75 -1.45 -2.45 -2.60 +1.9 +0.50 -0.50 -0.90
-0.03 -0.10 -0.33 -0.57 -0.72
+2.6 +2.3 +2.0 +1.5 +0.9
+6.20 +5.30 +4.60 +3.40 +2.10
3.6 4.2 4.5 5.2 5.5
0.9 0.69 0.57 0.22 0.07
150
3.6 4.2 4.7 5.0
100
3.7 4.0 4.3 4.7 5.0
9
PH
m2 V-‘s-l x $9 [Sl/[S-l lo-* V X lo-’ P
log
-1.17 -3.4 -5.80 -6.20 +4.35 +1.10 -1.10 -2.10
H PK~ 3.78 4.00 3.93 4.22 4.45 4.33 4.38 4.51 4.64 4.74 4.90 5.08 5.27 5.35
4.7 4.7 4.50 4.44 4,5 4.8 4.7 4.6 4.6 4.7 4.8 4.7 4.65 4.6
vides a convenient pH to test the chemical energies of adsorptionz9 and also is a reference point for modeling the electrostatic component of adsorption. We now apply our results from the mathematical analysis of the absorption bands for phosphate surface species (see Tables 1-111) to a standard “pK” model for the surface protonation of phosphate. We apply a Boltzmann correction to relate bulk pH to the surface proton concentration where the surface is charged. In the first case, we test the validity of eq 2 at the isoelectric pH. At pHiep,no electrostatic correction terms need be applied, and eq 2 can be expressed pK,, = log S/S- + pH
(3)
where S- refers to the bidentate deprotonated Fe,PO, and S refers to the protonated complex Fe,HPO,. Recall that for 190 pmol/g of phosphate adsorption the pHiep = 3.9. For this value of adsorption and pH, the ratio of nonprotonated to protonated species on the surface is 0.72 (from interpolation-see Table V). Substituting this p H and ratio of surface species into eq 3, we obtain p P n t = 4.6. Similarly, if we do the same for 150 and 100 pmol/g of surface coverage, we obtain pHie values of 4.4 and 5.3 and calculate values of p P n t of 4.8 and 4.3, respectively. In the last case, the ratio is obtained by extrapolation and places us in a region where the protonated species approaches zero in its concentration and therefore where the error in this ratio is very large. Therefore, is seems that the intrinsic pK for surface protonation is around pH 4.6. By converting electrophoretic mobility data to l potential values (using the program of O’Brien and White),” we can correct for the effect of electrostatics on these protonation constants. In this analysis, we presume that the { protential can be equated with the potential term in the Boltzmann expression. Since the exact location of the shear plane is certainly a subject of debate, there is a degree of uncertainty introduced by this assumption. We use the following equation: pK,, = log S/S- + pH
+ 2.3erC.JkT
(4)
In Table V, we see that the “pK” values vary between 4.8 and 4.5 for the total range of adsorption levels and
pH values studied. These corrected pK values, representing intrinsic pK values, illustrate once again that the surface protonation constant is around pH 4.6. Furthermore, the assumption that the { potential can be used
~
~
Langmuir 1990,6, 611-620 as the effective potential affecting surface protonation seems to be largely valid.
Acknowledgment. This work was funded in part by a contract from the Ecological Research Division, Office of Health and Environmental Research, U.S. Depart-
611
ment of Energy (DE-FG02-87ER60508) and in part from a NSF grant (CESA504276). We gratefully acknowledge all support received. Registry No. FeO(OH), 20344-49-4; phosphate, 14265-442;goethite, 1310-14-1.
Structure and Phase Behavior in Five-Component Microemulsions John F. Billman? and Eric W. Kaler*?' Department of Chemical Engineering BF-10, University of Washington, Seattle, Washington 98195, and Department of Chemical Engineering, University of Delaware, Newark, Delaware 19716 Received June 16, 1989. I n Final Form: October 2, 1989 Droplet-to-bicontinuous structure transitions in a family of five-component microemulsions formed with sodium 441'-heptylnonyl)benzenesulfonate, isobutyl alcohol, D,O, sodium chloride, and alkanes with even carbon numbers from octane to hexadecane are probed by using small-angle neutron scattering, electrical conductivity, and NMR self-diffusion measurements. The phase behavior and structure of these microemulsions are intimately linked and depend on salinity and the chain length of the alkane. Both the range of salt concentration in which the three-phase region is observed and the range of microemulsion water volume fraction within the three-phase region decrease with decreasing alkane chain length. Further, the appearance of the three-phase region is preceded by droplet-to-bicontinuous transitions. Microemulsions not exhibiting three-phase regions become bicontinuous only when they contain equal amounts of oil and water. The coincidence of the so-called percolation thresholds as determined by using electrical conductivity and self-diffusion measurements shows that electrical conduction in a dispersion of water droplets occurs with the exchange of material between the droplets. The scattering of dilute microemulsions is interpreted by using a variety of models in which the microemulsion is treated as a dispersion of hard or attractive spheres or as a dispersion of charged ellipsoids. The effect of alkane chain length on the droplet-to-bicontinuous transitions is interpreted in terms of the droplet interaction potentials.
Introduction Microemulsions are isotropic, thermodynamically stable dispersions of oil, water, surfactant, and often salt and cosurfactant. The oleic and aqueous components of microemulsions reside in distinct domains with length scales on the order of 100 A. These domains are separated by an interfacial sheet rich in surfactant and cosurfactant. The diversity of structure found in microemulsions and the ability of microemulsions to solubilize both polar and nonpolar substrates have generated great interest in microemulsion-forming systems.' Nevertheless, the relationship between the phase behavior of these systems and the structure of the microemulsions is unclear. The purpose of this paper is to delineate that relation for a model five-component microemulsion of a type often used commercially. The phase behavior of microemulsion-forming systems has been rationalized by Kahlweit and co-workers2-8 in terms of a simple phenomenological model. Studies
* Author to whom correspondence
should be addressed. University of Washington. University of Delaware. (1)Langevin, D. Acc. Chem. Res. 1988,21, 255.
*
0743-7463/90/2406-0611$02.50/0
of homologous series of oils and surfactants2-*have shown the phase behavior of three-component nonionic mixtures to be a consequence of their proximity to a tricritical point. By considering the effect of added electrolyte and ionic surfactant, Kahlweit et aLk8 argue that the phase behavior of microemulsions containing both ionic and nonionic surfactants and salt evolves continuously from a line of tricritical points. Further, these systems may be driven toward their tricritical point with systematic variation of the nature of their components. This model provides a basis for understanding the phase behavior in microemulsion-forming mixtures, but no account of structure in the microemulsion phase($ is made. The composition of many microemulsions can be varied continuously from oil-rich to water-rich within a sin(2) Herrmann, C. U.; Klar, G.; Kahlweit, M. J. Colloid Interface Sci. 1981,82,6. (3) Kahlweit, M. J. Colloid Interface Sci. 1982, 90, 197. (4) Kahlweit, M.; Lessner, E.; Strey, R. J . Phys. Chem. 1983,88,1937. (5) Kahlweit, M.; Strey, R.; Hasse, D. J . Phys. Chem. 1985,89, 163. (6) Kahlweit, M.; Strey, R.; Firman, P.; Hasse, D. Langmuir 1985, I , 281. (7) Kahlweit, M.; Strey, R. J. Phys. Chem. 1986, 90, 5239. (8) Kahlweit, M.; Strey, R.; Firman, P.; Hasse, D.; Jen, J.; Schomiicker, R. Langmuir 1988, 4,499.
0 1990 American Chemical Society