|
11. |
Adsorption of phosphate on calcite |
|
Journal of the Chemical Society, Faraday Transactions 1: Physical Chemistry in Condensed Phases,
Volume 82,
Issue 6,
1986,
Page 1733-1743
Toshio Suzuki,
Preview
|
PDF (760KB)
|
|
摘要:
J. Chem. SOC., Faraday Trans. 1, 1986, 82, 1733-1743 Adsorption of Phosphate on Calcite Toshio Suzuki, Shigehiro Inomata and Kiyoshi Sawada* Laboratory of Analytical Chemistry, Faculty of Science, Niigata University, Niigata, Japan 950-21 The distribution of orthophosphate on the calcite surface has been investi- gated at 25.0 "C. The distribution obeys the Freundlich adsorption isotherm. The unprotonated species, PO:-, is adsorbed on the calcite surface at low phosphate concentration and by increasing its concentration the adsorbed species changes to HPOi-, where the phosphates are adsorbed as neutral species by an accompanying calcium ion. At very low phosphate concen- tration, the PO:- ion exchanges with the carbonate ion in the calcite sur- face and is irreversibly incorporated into the calcite lattice.The enthalpy changes of the adsorption of phosphates are comparable to those of ion-pair formation of calcium-phosphate. The study of the crystallization of calcium carbonate is important in a wide variety of fields, e.g. the lime industry, geochemistry and physical chemistry. Nancollas and Reddy have studied the crystal growth of calcite using the seeded growth meth~d.l-~ The effects of crystal growth inhibitors have also been studied in detail. The phosphorus-containing anions show an intense inhibition for the crystallization of calcium ~arbonate.~' 5-8 The effect of polyphosphates on the retardation of crystallization is much larger than that of ort hop hosp ha te6 and the aminopol yp hosp honates such as e t hylenediamine te tra(me t hyl- enephosphonic acid) and hydroxyethylidene- 1,l -diphosphonic acid quench the crystal growth completely at very low concentrati~n.~? 6-8 The magnitude of the retardation effect depends on the polymorph of the calcium carbonate, i.e.calcite, aragonite or vaterite.s As a consequence of these experiments, the inhibition of crystal growth has been interpreted as the selective poisoning of the growth site by the adsorption of the phosphates, rather than the complex formation of the anions in the solution. The adsorption equilibria of the phosphates on calcite have been estimated by the analysis of the retardation effect on the growth rate.5 Because this method is indirect, the adsorption data are less reliable than direct measurements of adsorption isotherms.In addition, the knowledge of the adsorption equilibria of anionic species on the ionic crystals should give fundamental information for applications in analytical and industrial chemistry. In the present paper, the adsorption of the orthophosphate on the calcite surface from aqueous solution has been investigated by using the radioactive isotope 32P as a tracer. The chemical form of phosphate adsorbed on the calcite and the thermodynamic parameters have been determined and mechanisms describing the phosphate adsorption have been proposed and discussed. Experimental Reagents Guaranteed grade calcium chloride and sodium hydrogen phosphate (Wako, Japan) were recrystallized. Orthophosphoric acid labelled with radio isotope 32P was obtained from the Japan Isotope Association. Other reagents used were of guaranteed grade (Wako).Calcite crystals were prepared by the addition of 5 dm3 of 0.2 mol dm-3 calcium chloride solution to 5.5 dm3 of 0.2 mol dm-3 sodium carbonate solution at 25 "C at a rate of 17331734 Adsorption of Phosphate on Calcite 250 cm3 h-l [ref. (S)]. The calcite suspension was allowed to age with stirring for 1 day. The crystals were filtered and resuspended in distilled water for a week. After drying at 110 "C, the crystals were characterized by X-ray diffraction (Rigaku Denki, Japan) and scanning electron microscopy (JEOL-JSM25, Japan). The specific surface area (Ass) was determined by a single-point B.E.T. method (Micrometrics 2205-2, U.S.A.) as A,, = 0.70 m2 g-l. Experimental Procedure The pH of the calcite suspension was adjusted by gassing with CO, or N,, or by the addition of NaOH solution. The activity of the calcium ion in the solution was adjusted by the addition of the calcium chloride.The activity of the calcium ion was measured potentiometrically (Orion pH/mV meter 80 1A) by using a calcium ion-selective electrode (Orion model 93-20). After stirring for 30 min, the suspension was centrifuged and the crystals were discarded. A quantity of the supernatant liquid (20cm3) was transferred into 50 cm3 stoppered centrifuge tube and 1 g of characterized calcite crystals were added. By using the reaction solution pre-equilibrated with calcium carbonate, the change in the surface condition of the working crystals of calcite was minimized. After the solution was equilibrated with calcite for 30 min, a quantity of phosphate solution labelled with 32P, the pH of which was adjusted to that of the sample solution, was added.After shaking for 25 min, the activities of the hydrogen and calcium ions were checked. Then, the suspension was separated by centrifugation. A quantity of the supernatant liquid (1 cm3) was transferred onto a 25 mm diameter /3-counting plate and dried on it. The crystals were dissolved in 3.0 mol dmV3 HC1 up to 10 cm3. A quantity of the solution (1 cm3) was transferred and dried on the plate. The /3-ray activities were measured by G.M. counter (Aloka, model TDC-2, Japan) generally up to 10000 counts to minimize the statistical error. The activities were corrected by the Q-ray standard source prepared by the same procedure and the recovery of the 32P tracer was checked.The zeta potential of the calcite surface was measured at several calcium concentrations and pH by a streaming potential analyser (Shimadzu ZP- 1 OB, Japan). Desorption Experiments After the adsorption of phosphate labelled with 32P was complete, the extra phosphate (radioactivity-free) was added to the suspension. The change in the distribution of radioactive phosphate was measured as a function of the total concentration of phosphate. Results Solution Equilibria To analyse the distribution equilibria of phospate ion, the following protonation9 and ion-pair formationlo equilibria of phosphate ion in the solution have been taken into consideration : K , H++PO:-+HPOi- logK, = 12.15 K2 H+ + HPOi- H,PO, log K2 = 7.05 K3 Ca2+ +PO:- CaPO, log K3 = 6.46 (3) K4 Ca2+ + HPOi- CaHPO, log K4 = 2.74 a (4)T.Suzuki, S. Znomata and K. Sawada 1735 where the equilibrium constants are the thermodynamic constants. As the calcium concentration is relatively high [log a,, = - (5.&3.0)] under these experimental con- ditions, the carbonate concentration is very low, thus the ion-pair formation of calcium ions with carbonate, CaCO, and CaHCO;, is negligibly small. Under the conditions of pH (pH ca. 7.5-10.5) of these experiments, the formation of H$O, is negligible. The activity coefficient of z-valent ions at 25.0 "C,f,, was calculated using the Davies equation : logfi = -0.51 z"[P/(l+[P)-0.31] ( 5 ) where Z is the molar ionic strength. Distribution Ratio The distribution ratio, D, of phosphate between the calcite surface and the solution is defined as D = c p , S I C P (6) where Cp,s is the total concentration of phosphate distributed on the calcite surface (moles per kg of calcite) and C, is the total concentration of phosphate in the aqueous solution (moles per dm3 of solution).By taking into consideration the eqn (1)-(4), C, is written: C, = [HPOi-] +[PO:-] + [H,PO,] + [CaPOJ + [CaHPO,] (7) where [ 3 refers to the equilibrium concentration. Hereafter, subscript s refers to the chemical species adsorbed on the crystal surface or its concentration and the symbols having no subscript are those in aqueous solution. By using the activity of the monoprotonated phosphate ion HPOi-, which is the predominant species of the phosphate ion in the solution under the present experimental conditions, C, is rewritten: = aHPOg % (8) where ( I H p o g is the activity of HPOi- and ap is the side reaction coefficient term of HPOi- corrected for the activity coefficients.Here, paying attention to HPOZ- as a representative species, we define a distribution ratio corrected for the activity coefficients and side reactions as: Do Cp,s/aHP04 = Dap. (9) Here, as the side reactions of phosphate are corrected, the contributions of other species, e.g. H,PO,, PO:-, etc. to the distribution equilibria have been taken into consideration in Do. Effect of Phosphate Concentration The distribution ratios of phosphate, D, at various phosphate concentrations were measured under the conditions that the activities of calcium and hydrogen ions were kept constant.The logarithmic values of the corrected distribution ratio, log Do, calculated from D using eqn (9) are plotted as a function of logarithmic concentration of phosphate on the crystals, log C,, (fig. 1). The value of Do shows a gradual decrease with increase in the phosphate concentration on the calcite. The change in the concentration of crystals in the suspension has little effect on the plot of fig. 1. In order to study the kinetics of phosphate uptake, the phosphate concentration in 58 FAR 11736 Adsorption of Phosphate on Calcite 3 0 9 M c( 2 1 Fig. 1. Plot of log Do as a function of log Cp, at pH 8.5 and loguca = -3.14. The dotted line is the curve calculated using the Langmuir adsorption isotherm with the assumption of maximum adsorption of log Cp, = - 3.0.0 50 100 time/min Fig. 2. Change in phosphate concentration in the solution phase with time. Tp = 1.13 x mol dmP3. pH 8.60. logaca = 3.10. the solution was measured as a function of time. As seen in fig. 2, the distribution curve shows an instantaneous decrease in C,, followed by a gradual decrease and C, is kept constant after 60 min.T. Suzuki, S. Inomata and K . Sawada 1737 Effect of pH and the Calcium Ion Concentration The effect of pH on the distribution of phosphate is shown in fig. 3, where the activity of the calcium ion and the total concentration of phosphate initially added to the solution, T,, are kept constant. At low concentration of phosphate (T, = 1.25 x mol dm-3), the slope of the plot of log Do us.pH is close to unity 5 4 0 ;3 4 2 1 2 3 0 n 0 0 d o 0 0 1 I I 8 9 10 PH Fig. 3. Plot of logD" as a function of pH at logaca = -3.00. Tp: 1, 1.25 x 2.00 x mol dm-3; 2, mol dm-3; 3, lo-* mol dm-3. The slopes of the solid lines 1,2, and 3 are 0.89,0.23 and 0.08 , respectively. (slope = 0.89). The slope decreases with the increase in T,. The change in distribution ratio with pH is very small (slope = 0.08) at higher Tp (T, = mol dm-3), where because of relatively small distribution ratio the data are scattering compared with that at low T,. In order to study the effect of the activity of the calcium ion on the distribution of phosphate, Do was measured as a function of a,, at constant pH and Tp. Log D is plotted as a function of loga,, in fig.4. The slopes of the straight lines 1-4 (low T,) are all ca. 1.5 and the slope of curve 5 (high Tp) is 1.7. Effect of Temperature The temperature dependence of the distribution ratio was studied under the conditions of constant a,,, pH and Tp (fig. 5). Although the temperature range is narrow, the plots of log Do as a function of the reciprocal temperature at Tp = 1.25 x mol dm-3 58-21738 3 - 2 - 0 9 M c( 1 - 0 - 3.0 2 8 - Adsorption of Phosphate on Calcite - Fig. 4. Plot of logD" as a function of logaca. pH: 1, 9.8; 2, 9.3; 3, 8.8; 4 and 5, 8.1. Tp: 1.25 x mol dm-3 for curves 1-4; lo-* mol dm-3 for curve 5. The slope of solid lines 1-4 is 1.5.* The slope of the curve 5 is ca. 1.7. 3.0 3.5 lo3 KIT Fig. 5. Temperature dependence of distribution ratio of phosphate at pH 8.10 and at loguca = -3.10.Tp: 1, 1.25 x mol dm-3; 2, 7.62 x lod5 mol dm-3.T. Suzuki, S. Inomata and K. Sawada 1739 mol dm-3 (curve 2) show straight lines with negative slopes (curve 1) and 7.62 x of -0.74 x lo3 and -0.87 x lo3, respectively. Desorp tion In order to check the reversibility of adsorption of phosphate, the exchange of the radioactive phosphate was studied. The labelled phosphate was equilibrated with calcite at a given concentration. Then the 32P labelled phosphate was desorbed by the successive addition of 32P free phosphate (fig. 6), i.e. the increase in the total phosphate concentration, Tp, causes a decrease in the distribution ratio. The fact that the distribution ratio of radioactive phosphate between solid phase and solution (desorption curves 2-4) does not agree with the adsorption curve (curve 1) indicates that the adsorption of phosphate is partially irreversible.Fig. 6. Adsorption (curve 1) and desorption curves (curves 2-4). Curves 2, 3, 4: percent of radioactive phosphate remaining on calcite by the addition of stable phosphate. Total concentration of phosphate initially loaded: 2, mol dm-3; 3, 1.25 x lo-' mol dm-3; 4, 1.25 x 10+ mol dm-3. pH 8.1. log+& = -3.1. As can be seen from fig. 6, the desorption becomes more irreversible with lower initial loaded concentration of phosphate. More than 80% of phosphate loaded at Tp = mol dm-3 is not exchanged with that in solution and remains irreversibly on the crystals even when a large excess of phosphate is added. When mol dm-3 of phosphate is initially loaded, the irreversible adsorption is < 30%.Discussion Reddy has studied the effect of the orthophosphate ion on the growth rate of calcite and interpreted the retardation effect by a Langmuir-type adsorption of phosphate ion on the calcite ~urface.~ The log Do vs. log Cp, curve calculated using the Langmuir adsorption isotherm is shown in fig. 1 by dotted line, where the maximum adsorption is assumed as log Cp, = - 3.0. The disagreeement of the experimental data with the calculated curve indicates that this distribution system of phosphate cannot be explained by a Langmuir-type adsorption, i.e. monolayer adsorption of phosphate. The plot of log Cp, as a function of log Cp, i.e. Freundlich adsorption isotherm,12 is shown in fig.7. The plot gives a straight line with a slope of 1.72-l. Thus, adsorption1740 -a, Adsorption of Phosphate on Calcite , 1 1 I of an uncharged species is anticipated, i.e. the phosphate anion is adsorbed on the calcite by accompanying the counter cation, hydrogen and/or calcium ion(s). The fact that a plot of Cp, as a function of log Cp does not fall on a straight line supports the adsorption of uncharged species. The surface charge of calcite was measured directly by a zeta potential measurement. The change in the surface charge due to the adsorption of phosphate is negligible compared with the amount of the adsorbed phosphate. By using HPOi- as a representative species, we express the distribution equilibrium as Kad lCa2+ + n(m - 1) H+ + nHPOi- Z(Ca2+)s + n(HmP0,"-3)s (10) where the thermodynamic adsorption constant of eqn (lo), Kad is defined as: As the total concentration of phosphate on the calcite surface, Cp,s, is equal to [HmPOp-3]s, the corrected distribution ratio, Do, is written as: Substitution of eqn (1 1) into (12) leads to: n log Do = log Kad + Z log a,, + n(m - 1) log aH - 1 log [Ca2+Is.(13) By the assumption of the adsorption of a single neutral species we obtain the relationship, [Ca2+Is: [HmP0,"-3]s = 1: n, where 21 = -n(m - 3). Thus, [Ca2+Is = l/n [H,P0,"-3]s = l/n Cp, s. Finally, eqn (13) is written (14) 1 1 1 1 n n n n l0gD" = - log Kad+- logaca-(m- 1)PH-- log- Cp, s.T. Suzuki, S. Inomata and K. Sawada 1741 As can be seen from fig. 1, Do changes with the change in the phosphate concentration on the calcite, Cp, ,, even though a,, and pH are kept constant.This fact means that the adsorption constant, Kad, changes with the change in Cp, s, i.e. Kad is the conditional constant of CP,,. Under the conditions of low Tp, as the distribution ratio is considerably high, the most of phosphate is adsorbed on the calcite surface. When Tp, i.e. the total amount of phosphate added, is kept constant (fig. 3, curve 1 and fig. 4, curves 1-4), the amount of phosphate adsorbed on the crystals is almost constant, i.e. the surface concentration of phosphate on the calcite, CP,,, is constant in a series of experiments. Thus, the conditional constant, Kad, can be assumed to be eventually constant at a given Tp. As the adsorption of a single chemical form of phosphate is assumed, Z/n becomes constant, i.e.Z/n logZ/n Cp, , is also constant. Thus, eqn (14) is reduced to 1 n log Do = - log a,, + (1 - m) pH +constant. The slope of the plot of logD" us. pH at low Tp (fig. 3, curve 1) is close to unity. Thus, 1 -m = 1, i.e. m = 0, indicates the adsorption of unprotonated phosphate ion, (Poi-),. The decrease in slope at low pH is explained by the relatively low distribution ratio of phosphate. The plots of log Do vs. log a,, at low Tp (fig. 4, curve 1-4) fall on straight lines with a slope of 1.5, irrespective of pH. Thus, the ratio l/n is 1.5, i.e. the ratio of calcium : phosphate is obtained as (Ca2+)s : (Poi-), = 3 : 2. Consequently, the dependences of Do on the hydrogen and calcium ion activities reveal the adsorption of 3(Ca2+), - 2(PO:-),, i.e.the negative charge of the adsorbed phosphate anion is neutralized by the accompanying adsorption of the calcium cation. mol dm-3), the slope of the plot of log Do as a function of pH (fig. 3, curve 3) is very small. Thus, the phosphate concentration of the calcite surface is again almost constant, irrespective of pH at constant Tp and thus the adsorption constant Kad is also assumed to be constant. Consequently, eqn (1 5) can be employed for the determination of m even at high Tp, i.e. even at the low distribution ratio. The slope of the plot of log Do us. pH (1 - m = 0, i.e. m = 1) indicates the adsorption of the monoprotonated species, (HPOi-),, at higher phosphate concentration. Analysis of the a,, dependence of the distribution ratio is rather complicated at high Tp.As the distribution ratio is relatively small at high Tp, the surface concentration of phosphate, CP,.,, changes with the change in distribution ratio even at constant Tp. The change in D with a,, causes the change in Kad, thus the slope of the plot of logD" vs. a,, does not indicate the ratio l/n, directly. However, the increase in the distribution ratio with increase in a,, indicates the adsorption of phosphate accompanied with absorption of calcium ions. Thus, under conditions of high Tp, the adsorption of the neutral species (Ca2+), - (HPOi-), is supported. The slope of the plot of log Do vs. pH is less than unity at the intermediate concentration of phosphate (Tp = 2 x mol dm-3). This may indicate the adsorption of both forms of phosphate, PO:- and HPOi-, in the intermediate concentration.The assumption of the adsorption of the ion pair of calcium-phosphate on the calcite surface, [(Ca2+)l (HmPOi-)n]s, gives the distribution of the same chemical species as that of the separate adsorption of calcium ion, i.e. the adsorption of [(Ca2+),(PO,3-),], at low Tp and [(Ca2+)(HPOi-)], at high Tp. Although these two mechanisms are indistinguishable, the difference between them is merely whether the calcium ion is adsorbed at the neighbouring site of the adsorbed phosphate ion or distributed at random over the whole surface. In the above discussion, adsorption equilibria were analysed using the predominant species in the solution, HPOi-, as a representative species. This does not mean that adsorption of other species such as H2POy, PO:-, etc.is ignored, but these species are At high phosphate concentration ( Tp =1742 Adsorption of Phosphate on Calcite taken into consideration by using the side-reaction coefficient % in eqn (8). E.g. the equilibrium constant c d of the adsorption of unprotonated species : K L ZCa2+ + nmH+ + nP0:- l(Ca2+)s + n(HmP0,"-3)s is written as The constant c d is correlated with the adsorption constant of HPO:-, Kad, as: = Kad KF. (18) By assuming that the distribution of phosphate is controlled by the exchange reaction of the carbonate ion in the calcite surface with the phosphate ion in the solution, we obtain the following : Kex r(COi-)s + n(m - 1) H+ + nHPOi- rCOi- + n(HmPOr-3)s where 2r = n(m - 3). The equilibrium constant of anion exchange is written as: By using the solubility product of calcite, Ksp = a,,aCo3, we obtain the following equation for the corrected distribution ratio : n log Do = log Kex - r log Ksp + r log a,, + n( 1 - m) pH + r log [COi-],.(21) When the phosphate concentration on the calcite surface, Cp,s, is kept constant, the values of Kex, Ksp and [CO:-], are assumed to be constant and we obtain the following equation : (22) log Do = - log a,, + (1 - m) pH +constant. Thus, the ion-exchange mechanism again explains the results in fig. 3 and 4 reasonably and indicates the distribution of the same chemical species of phosphate as that of adsorption mechanism, i.e. the exchange of carbonate ion in the surface of calcite with the unprotonated phosphate ion, PO:-, at low Tp and by the monoprotonated phosphate ion, HPO:-, at high Tp.As can be seen from the above discussion, the distribution of phosphate can be interpreted by either the adsorption mechanism or the ion-exchange mechanism, and it is impossible to distinguish between them from the results of the pH and pCa dependences. The distribution curve in fig. 2, however, indicates that the distribution of phosphate proceeds by two steps, the fast step, which is accomplished instantaneously, and the subsequent slow one. The results of desorption experiments (fig. 6) also suggest the two mechanisms : reversible and irreversible distribution. Thus, the irreversible distribution of phosphate at low concentration may suggest the exchange of COi- in the crystal surface and the incorporation of phosphate in the calcite lattice.The proportion of reversible adsorption increases with increase in the phosphate concentration and the chemical form of the distributed phosphate changes to HPOi-, which is weakly adsorbed on the crystal surface. r nT. Suzuki, S. Inomata and K. Sawada 1743 The coprecipitation of phosphate and calcite from a supersaturated solution of calcium carbonate has been studied13 and the coprecipitation of unprotonated species, PO:-, has been observed even when the pH of the solution was relatively low, i.e. HPOi- is the predominant species in the solution. Thus, the PO:- ion has a much higher affinity for the calcite lattice than the HPOi- ion. However, as can be seen from fig. 6, the amount of phosphate irreversibly distributed on the crystals, PO:-, increases with increase in phosphate concentration, Tp and the magnitude of the increase in HPOi- is much larger than the increase in PO:-.Consequently, the chemical form distributed on the crystals has eventually been changed to HP0;- at high Tp. As the dehydration energy of HPOi- is lower than that for the highly charged PO:- the predominant species of the phosphate ion in the solution, HPOi-, will be favourably adsorbed. The amount of phosphate distributed irreversibly on the calcite surface is estimated as at most Cp, = to lo-* mol kg-l, which is much less than that of the carbonate ion on the calcite surface exposed to the solution, where the amount of C0:- is roughly estimated as 7 x From the slope of the plot of fig.5, the enthalpy change of the distribution has been tentatively estimated as - AHex = - 3.4 kcal mol-1 for the exchange of PO:- at low Tp and -AHad = -4.0 kcal mol-1 for the adsorption of HPOi- at high Tp. The value of - AHad of HPOi- is the same magnitude of that of the ion-pair formation (Ca2+ * HPOi-) in the aqueous solution (- AHip = - 3.3 kcal mol-l).lo Thus, the enthalpy change of the distribution of HPOi- indicates very weak adsorption compared to the ion-pair. By taking into consideration the deprotonation of HPOi- in solution ( - AH = - 2.4 kcal mol-l),l5 the net enthalpy change on distribution of PO:- [PO:-(sol- ution) + PO:-(crystal)] is - AHe,(net) = - 1 .O kcal mol-l. Although the value of -AHex(net) of PO:- is relatively larger than that of -AHad of HPOi-, it is still endothermic. This may be interpreted as compensation of the exchange energy with the very large dehydration energy of trivalent ion, PO:-. This may be corroborated by the fact that the enthalpy change of the ion-pair formation of Ca2+.PO:- (- AHip = - 3.1 kcal mol-l) is the same magnitude as that of HPOi- and is endothermic. mol kg-l, using A,, = 0.7 m2 g-l. We are grateful for financial support from the Japanese Ministry of Education and the Saneyoshi Scholarship Foundation. We thank Professor T. Sotobayashi for helpful discussions. References 1 M. M. Reddy and G. H. Nancollas, J. Colloid Interface Sci., 1971, 36, 166. 2 G. H. Nancollas and M. M. Reddy, J. Colloid Interface Sci., 1971, 37, 824. 3 M. M. Reddy and G. H. Nancollas, Desalination, 1973, 12, 61. 4 M. M. Reddy and G. H. Nancollas, J. Crystal Growth, 1976, 35, 33. 5 M. M. Reddy, J. Crystal Growth, 1977, 41, 287. 6 M. Miura, H. Naono and S. Otani, Kogyo Kagaku Zasshi, 1963,66, 597. 7 G. H. Nancollas, T. K. Kazmierczak and E. Schuttringer, Natl Assoc. Corrosion, Chicago, 1980, paper 8 G. H. Nancollas and K. Sawada, J. Petrol. Tech., 1982, 645. 9 P. Salomaa, R. Hakala, S. Vesala and T. Aalto, Acta. Chim. Scand., 1969, 23, 2116. no. 226. 10 A. Chughtai, R. Marshall and G. H. Nancollas, J. Phys. Chem., 1968,72, 208. 1 1 C. W. Davies, Ion Association (Butterworths, London, 1964). 12 H. A. Laitinen and W. E. Harris, Chemical Analysis (McGraw-Hill, New York, 1975). 13 M. Ishikawa and M. Ichikuni, Geochem. J., 1981, 15, 283. 14 R. H. Stokes, J. Am. Chem. Soc., 1964,86, 979. 15 G. Ferroni, Bull. SOC. Chim. Fr., 1974, 2698. Paper 5/1117; Received 2nd July, 1985
ISSN:0300-9599
DOI:10.1039/F19868201733
出版商:RSC
年代:1986
数据来源: RSC
|
12. |
Stability of monochloride complexes of some divalent transition-metal cations inN,N-dimethylformamide |
|
Journal of the Chemical Society, Faraday Transactions 1: Physical Chemistry in Condensed Phases,
Volume 82,
Issue 6,
1986,
Page 1745-1753
Wacław Grzybkowski,
Preview
|
PDF (566KB)
|
|
摘要:
J. Chem. SOC., Faraday Trans. 1, 1986,82, 1745-1753 Stability of Monochloride Complexes of some Divalent Transition-metal Cations in N,N-Dimethylformamide Waclaw Grzybkowski* and Michal Pilarczyk Department of Physical Chemistry, Institute of Inorganic Chemistry and Technology, Technical University of Gdarisk, 80-952 Gdarisk, Poland Equimolar mixtures of divalent transition-metal perchlorates in N,N- dimethylformamide (DMF) have been used as constant ionic media for studying the formation of monochloride complexes of metal cations. The derived stability constants of the MCl(DMF)$-type complexes (M = Mn, Co, Ni, Cu) disobey the Irving-Williams series forming the sequence Mn > Co > Ni < Cu. It has been shown recently that the monohalide complexes of divalent transition-metal cations in non-aqueous systems disobey the sequence known as the Irving-Williams1 series.According to this series a monotonic increase in stability from manganese@) to copper(I1) is expected and the stability trend is Mn2+ < Fe2+ < Co2+ < Ni2+ < Cu2+ > Zn2+. The results obtained by LibuS et al., for the monochloride complexes in dimethyl sulphoxide and by Solomon3 for monofluoride complexes in methanol indicate that the corresponding sequence appears to be Mn2+ > Co2+ > Ni2+ < Cu2+ > Zn2+. A similar sequence was obtained by Doe and Kitagawa for the monochloride complexes in methan01.~ Unfortunately, their work does not include the chloride complex of manganese(r1). Although the complexes of copper(I1) are clearly the most stable, the monotonic increase from manganese@) to copper(I1) seems to be violated. Moreover, there is good evidence that the same sequences are valid for the monochloride and monobromide complexes in aqueous s~lution.~ The present work was undertaken in order to establish the stability trend for the monochloride complexes in N,N-dimethylformamide solution using the spectrophoto- metric method recently developed in our laboratory., The presented approach consists of studying equimolar mixtures of cobalt(I1) perchlorate with the other metal perchlorates in the presence of a small amount of Et4NCl.In a previous paper6 it has been shown that the pseudotetrahedral CoC1,DMF- and the pseudo-octahedral CoCl(DMF)z chloride complexes are formed in the Co(C10, ),- Et4NC1-DMF system when Cl-/Co2+ -= 2.0.Owing to the high value of the molar absorption coefficient, an intensity of the band due to the presence of the CoC1,DMF- complex is a measured solution property depending on the free chloride anion concentration. Moreover, LibuS et al. have shown that equimolar solutions of the divalent transition- metal perchlorates in dimethyl sulphoxide2 and acetonitrile' solutions behave as effective constant ionic media. This was ascribed to the fact that the metal perchlorates exist exclusively as the MLi+ - 2C10; complex electrolytes, where L denotes the solvent molecule (slightly associated only) and the properties of the hexasolvo-cations depend 17451746 Transition-Metal Chloride Complexes in DMF I wavelength/nm Fig. 1. Absorption spectra of cobalt(I1) in DMF solutions of: (1) Co(ClO,), (0.0758 mol dm-3); (2) Co(ClO,), (0.0754 mol dm-3)-Et,NC1 (0.0142 mol dm-3); (3H7) Co(C10,),-Mn(C10,),- Et4NC1.For the equimolar mixtures the total metal perchlorates (0.0754 mol dm-3) and Et,NCl (0.0142 mol dm-3) are constant while the mole fraction of Mn(C10,), varies from 0.17 [curve (3)] to 0.83 [curve (7)]. on the nature of the central metal ion to a very small extent. The same was found for the metal perchlorates in DMF solution.8 Moreover, the DMF-solvated cations display the lowest ability for outer-sphere association, being practically independent of the nature of the central metal atom. In this way we accomplish the constancy of ionic media, eliminating the difficulties arising from a variation of the activity coefficients. Despite all approximations the method gives useful results, which are especially reliable for studying the differences in the stability constants.Experiment a1 The DMF-solvated metal perchlorates were obtained from the respective hydrates by dissolving them in DMF, followed by removing any excess of the solvent under reduced pressure at 60 "C. The crystalline solids were filtered off and further purified by repeated crystallizations from anhydrous DMF. Reagent grade tetraethylammonium chloride was recrystallized twice from anhydrous acetonitrile and dried in vacuo at elevated temperature. N,N-Dimethylformamide (analytical grade) was dried using 4A molecular sieves and distilled under reduced pressure at 45-50 "C. The specific conductivity of the final product was 5 x The stock solutions of the metal perchlorates in DMF were analysed for the respective metals by standard EDTA titrations.Solutions used in the final measurements were obtained by weight from the respective stock solutions and solvent. Their concentrations R-l cm-l.W. Grzybkowski and M. Pilarczyk 1747 10 a * I Ei - L 6 E E m a 1 4 2 C 600 650 700 w aveleng th/nm Fig. 2. Absorption spectra of cobalt(I1) in DMF solutions of: (0) Co(ClO,), (0.1236 mol dm-3)- Et,NCl (0.01 mol dm-3); (1H3) Co(C10,),-Ni(ClO,),-Et,NCl ; (l'H3') Co(ClO,),-Mn(ClO,),- Et,NCl. For the equimolar solutions the total metal perchlorates and Et,NCl are constant (ca. 0.124 and 0.01 mol dm-3, respectively). The mole fractions of Ni(C104), and Mn(C10,), in mixed-metal perchlorates are indicated on the curves.were calculated using the densities determined independently. Preparations of the solutions and other manipulations were carried out in a dry box. Absorption spectra were measured using a VSU-2-P Zeiss spectrophotometer equipped with a thermostatted cell compartment (25.00 0.05 "C). Results and Discussion Fig. 1 shows the spectra of cobalt(I1) from a series of equimolar mixtures of Co(ClO,), with Mn(ClO,), at constant concentration of Et,NCl along with the spectra of cobalt(r1) for the pure Co(ClO,), solution (curve 1) and for the Co(ClO,), solution containing Et,NCl (curve 2). The absorption band observed at 525 nm is characteristic of the Co(DMF)g+ octahedral comple~,~ while the band with maxima at 610 and 675 nm (curves 2-7) is due to the presence of the pseudotetrahedral CoC1,DMF- complex, known from an earlier study.s Formation of an octahedral complex, e.g.CoCl(DMF)t, having relatively low absorptivity may well remain without a noticeable effect on the measured spectrum. Inspection of fig. 1 shows that a substitution of Mn(C10,), for Co(ClO,), in the equimolar mixtures (curves 3-7) results in a marked decrease in intensity of the latter band, indicating a decrease of the mole fraction of cobalt(I1) existing as the CoC1,DMF- complex anion. The opposite effect is caused by the substitution of Ni(C10,), for an equivalent amount of Co(ClO,),. In fig. 2 is shown the ' tetrahedral'band of the spectra of cobalt(r1) for the series of equimolar mixtures of Co(ClO,), with either Mn(ClO,),1748 Transition-Metal Chloride Complexes in DMF 50 LO 4 I / I I I I I I I I I 100 80 60 LO 20 co (%) Fig.3. The mean molar absorption coefficient of cobalt(1r) at ,661.5 nm us. composition of the Co(Clo,),-M(ClO,),-Et4NCl equimolar mixtures. (a) M = Ni; (6) M = Mn; (c) M = Cu. or Ni(C10, ),, curve 0 represents the spectrum obtained for the Co(ClO,),-Et,NCl solution. The spectra determined for the mixtures were corrected for the absorption due to the presence of the octahedral solvo-cations, Mn(DMF)t+ and Ni(DMF):+. As is seen, the substitution of Ni(ClO,), for Co(ClO,), (curves 1-3) brings about a distinct increase in intensity of the band under consideration. The effect indicates increasing mole fraction of cobalt(I1) present in form of the CoC1,DMF- complex.This may only result from the fact that nickel@) forms chloride complexes, presumably octahedral ones, less stable than those of cobalt(I1). An opposite conclusion follows for the C0(ClO4),-Mn(C10,),-Et4NC1 system. Thus, the chloride complexes of manganese@) have a higher stability than corresponding complexes of cobalt(I1) and nickel@). Fig. 3 presents the molar absorption coefficient of cobalt(I1) at 66 1.5 nm plotted against the composition of the equimolar mixture for all systems studied. As is seen, Cu(ClQ,), is the most effective in depressing the intensity of the band under consideration. The effect is an indication of the fact that the chloride complexes of copper(I1) are more stable than the corresponding complexes of manganese(I1). Assuming that the monochloride complexes of nickel@), manganese@) and copper(I1)W.Grzybkowski and M . Pilarczyk 1749 are formed in the mixtures studied, we take the effects presented in fig. 1 and 2 as evidence of the following trend of the stability constants Mn2+ > Co2+ > Ni2+ < Cu2+ identical to the sequer ses reported for dimethyl sulphoxide2 and methanol3* solutions. Analogous results were obtained for equimolar mixtures differing in composition, irrespective of the total concentration of the metal perchlorates and the concentration of Et,NCl, indicating the validity of the assumption that the monochloride complexes only are formed in the systems studied. For the whole series we determined a concen- tration of the CoC1,DMF- complex, providing a possibility of a quantitative evaluation of the presented results.This concentration was found using the knowns molar absorption coefficients of the CoC1,DMF- and Co(DMF)i+ complexes. Assuming that the complexes formed in the Co(C10,)z-M(C10,)2-Et4NCl-DMF mixture (M = Mn, Ni, Cu) are CoCl;, CoCl+ and MC1+ (apart from their solvation), the respective stability constants are defined as [COCl,] [COZ'] [C1-]3 - [CoZ+][Cl-] cop3 = y3 = '"Q3 y3 & = cop - [CoCl+ ] [MCl+] MP1 = [M2+] [cl-] K = MQl K (3) where y3 and & are quotients of the activity coefficients for the respective complex-forming equilibria and c0Q3, coQ, and are the corresponding 'medium ' stability constants, i.e. the quotients of the equilibrium concentrations. Taking into account the material balance for the chloride anion, Libui et aZ.l0 have derived the relation where C,, denotes the concentration of Et,NCl.The second term on the left-hand side may be neglected as being relatively small and the [M2+]/[Co2+ 3 ratio may be approximated as CM/Cc,, since the total concentrations of the metal perchlorates, CM and Cco, are much higher than the concentration of Et,NCl. Thus, eqn (4) is reduced to In the above, [Co2+] may be taken as Cco - [CoCl;]. Outer-sphere association with the perchlorate anion was omitted as an effect of minor importance.8 Plotting the left-hand side of eqn (5) us. CM/Cco, C0Ql/C0Q;/3 is obtained from the intercept and MQl/c0Qi/3 from the slope, whence the ratio MQl/coQl is obtained. A representative set of data for one of the equimolar series is shown in table 1, while resulting plots are presented in fig.4. It may seem that the plots are linear, irrespective of the nature of the metal perchlorate which makes up the second.component of the mixture. Moreover, the intercepts are practically the same. These facts are an additional support of a validity of the assumptions presented above. It should be pointed out that the derived ratio of the 'medium' stability constants, MQl/coQl, is equivalent to the ratio of 'thermodynamic' stability constants, since the quotients of the activity coefficients are constant within the equimolar series and independent of the nature of the metal cation. The data obtained for four series of the equimolar mixtures of Co(ClO,), with each1750 Transition-Metal Chloride Complexes in DMF Table 1.Representative set of data obtained for the Co(ClO,),- M(ClO,),-Et,NCl equimolar mixtures in DMF at 25 "C concentration/mol dm-3 Co(ClO,), M(ClO,), Et,NCl [CoCl3DMF-la 0.1 10 25 0.099 85 0.088 28 0.077 50 0.067 09 0.1 10 85 0.099 56 0.088 98 0.077 51 0.065 92 0.120 72 0.1 17 35 0.11409 0.110 73 Co(ClO,),-Mn(ClO, ),-Et,NCl 0.013 75 0.009 89 0.000 925 0.025 15 0.008 58 0.000 495 0.036 07 0.009 35 0.000 441 0.046 28 0.009 93 0.000 381 0.057 62 0.009 24 0.000 221 Co(C10, ),-Ni(ClO, ),-Et,NCl 0.013 36 0.009 51 0.001 181 0.024 52 0.009 69 0.001 219 0.035 82 0.008 97 0.001 113 0.046 97 0.009 38 0.001 237 0.058 11 0.009 94 0.001 375 Co(ClO, ),-Cu(C10, ),-Et,NCl 0.003 54 0.009 55 0.000 676 0.006 66 0.009 84 0.000 453 0.010 03 0.009 72 0.000 307 0.013 36 0.009 76 0.000 227 a Concentration of the CoC1,DMF- complex taken as an average of the results obtained for three wavelengths.0.2' I 1 I I I I I I 0.0 0.2 0.L 0.6 0.8 CMICCO Fig. 4. Evaluation of the spectrophotometric data for the Co(C1O,),-M(C10,),-Et4NC1 equimolar mixtures in accordance with eqn (5). The total concentration of the metal perchlorates amounts to 0.124 mol dm-3. (a) M = Ni; (b) M = Mn; (c) M = Cu.Table 2. The least-squares parametersa of eqn (5) derived for the equimolar Co(CIO, ),-M(ClO, ),-Et4NCI-DMF mixtures at 25 "C M=Mn M = Ni M = Cu 0.075 0.0142 0.259 (0.00 5) 0.088 0.0072 0.272 (0.0 1 2) 0.124 0.0100 0.232 (0.002) 0.160 0.0075 0.221 (0.005) 0.890 0.54 0.773 0.45 (0.004) 0.734 0.50 (0.003) 0.703 0.50 (0.005) (0.002) average: log (2) = 0.50 (k0.04) 0.252 (0.008) 0.253 (0.001) 0.236 (0.003) 0.252 (0.003) 0.133 - 0.28 (0.006) 0.127 -0.30 (0.001) 0.099 -0.38 (0.005) (0.003) 0.086 - 0.47 (2) = - 0.36 (& 0.09) 0.203 5.51 1.43 % (0.03 1) (0.25) T 0.312 3.97 1.10 2 (0.022) (0.26) ?x (0.01 3) (0.16) Q % a 3 0.269 3.19 1.07 0.239 4.11 1.24 (0.012) (0.17) a The standard deviations in intercepts and slopes are given in parentheses.1752 Transition-Metal Chloride Complexes in DMF Table 3.Summary of log /I1 values for monohalide complexes in non-aqueous solvents at 25 "C DMF DMSOa MeOHb X=C1 X=C1 X = F ~~ MnX+ 4.00 4.53 3.24 3.48 c o x + 3.5OC 4.03 2.79 3.04 Nix+ 3.14 3.67 2.41 2.9 1 CuX+ 4.71 5.24d 5.01 4.52 a From ref. (2). From ref. (3) for ionic strength of 0.05 mol dm-3.From ref. (6). Estimated using value from ref. (1 1). of the other metal perchlorates were used in the calculations. The resulting parameters of the least-squares linear approximation of the respective plots are listed in table 2 together with the derived logarithms of a ratio of the stability constant. As can be seen, the results obtained for the Co(C10, ),-Cu(ClO,),-Et,NCl-DMF system exhibit relatively high error resulting from difficulties in a compensation of the measured absorbance for the contribution due to copper(I1). The average values of log (MPl/Copl) are shown in the last row of table 2. As can be seen, the numbers reflect the sequence reported above, arising from inspection of the spectral effects in fig. 1-3. Unfortunately, the method cannot be used effectively for the corresponding cobalt(II)-zinc(II) system.The extremely high stability of the tetrahedral chloride complexes of zinc(I1) makes derivation of its stability constants impossible for either dimethyl sulphoxide2 or DMF solutions. Despite this lack, the sequence seems to be well established and, in its part, distinctly opposite to the Irving-Williams series. The next problem for consideration is to estimate the absolute value of the corresponding stability constant. In a previous paper6 we reported a value of 3 . 5 f 0 . 5 for the stability constant of the CoCl(DMF); complex. However, the uncertainty of this value is very high, owing to the method used in the calculation as related to unusual stability of the CoC1,DMF- complex. As a next reference value, that of log = 3.76 f 0.13 reported by Elleb et for a solution of 1 mol dm-, ionic strength may be used.This value leads to a value of 2.55 for log " O Q , . We have tried to estimate the difference due to the ionic medium on the basis of the Debye-Hiickel equation with the ion-size parameter and the corresponding quotient of the activity coefficients calculated as 4 A d I 1 + B d d I log q = where ii is the ion-size parameter estimated from the conductometric measurementsg and corresponding to B6 = 3.3. Thus, it is not surprising that on this basis Elleb's value for log leads to a value of 4.03 for log cop1, markedly higher than the value of 3.50 reported in our paper.6 The resulting values of logarithms of the stability constants are listed in table 3 together with the data for the monochloride complexes in dimethyl sulphoxide2 and the monofluoride complexes in methanol3 solution.Inspection of this table clearly shows that the monohalide complexes of the divalent transition-metal cations in non-aqueous solvents disobey the Irving-Williams series. The exceptions to the sequence were considered as rather atypical and rationalized in terms of entropy controlling the stabilityW. Grzybkowski and M . Pilarczyk 1753 of the complexes. The existing experimental data do not provide a sufficient basis for a profound discussion of this problem. However, the same order of magnitude and the same trends are an indication of the existence of a factor common for the complexes under consideration. Thus, the sequence MnX+ > COX+ > Nix+ -c CuX+ > ZnX+ seems to be characteristic of this class of complexes. This investigation was supported by grant MR-1-11, Poland. References 1 H. Irving and R. Williams, J. Chem. Soc., 1953, 3192. 2 W. Libui, R. Pastewski and T. Sadowska, J. Chem. Soc., Faraday Trans. I , 1982,78, 377. 3 L. R. Solomon, A. M. Bond, J. W. Bixler and D. R. Hallenbeck, Inorg. Chem., 1983, 22, 1644. 4 H:Doe and T. Kitagawa, Znorg. Chem., 1982,21,2272. 5 Z. Libui and H. Tialowska, J. Solution Chem., 1975, 4, 101 1. 6 W. Grzybkowski and M. Pilarczyk, unpublished results. 7 W. Lib& and H. Strzelecki, Electrochim. Acta, 1972, 17, 577. 8 W. Grzybkowski and M. Pilarczyk, J. Chem. SOC., Faraday Trans. I , 1983,79, 2319. 9 W. LibuS, B. Chachulski, W. Grzybkowski, M. Pilarczyk and D. Puchalska, J. Solution Chem., 1981, 10 W. Libui, M. Pilarczyk, R. Pastewski and T. Szuchnicka, Electrochim. Acta, 1982, 27, 573. 11 M. Elleb, J. Muellemeestre, M. J. Schwing and F. Vierling, Inorg. Chem., 1980, 19, 2699. 10, 631. Paper 5/1120; Received 2nd July, 1985
ISSN:0300-9599
DOI:10.1039/F19868201745
出版商:RSC
年代:1986
数据来源: RSC
|
13. |
Interfacial tension minima in oil–water–surfactant systems. Effects of alkane chain length and presence of n-alkanols in systems containing aerosol OT |
|
Journal of the Chemical Society, Faraday Transactions 1: Physical Chemistry in Condensed Phases,
Volume 82,
Issue 6,
1986,
Page 1755-1770
Robert Aveyard,
Preview
|
PDF (925KB)
|
|
摘要:
J. Chem. Soc., Faraday Trans. 1, 1986,82, 1755-1770 Interfacial Tension Minima in Oil-Water-Surfactant Systems Effects of Alkane Chain Length and Presence of n-Alkanols in Systems containing Aerosol OT Robert Aveyard,* Bernard P. Bids and Jeremy Mead Chemistry Department, University of Hull, Hull HU6 7RX In alkane and aqueous NaCl systems containing Aerosol OT (AOT) the alkane-solution interfacial tension becomesconstant at the onset of surfactant aggregation, which can occur in either the aqueous or alkane phase, depending on conditions. The constant tension, yc, can attain ultralow values (ca. 10-3mNm-1), and has been shown previously to pass through a minimum value with respect to salt concentration and temperature. In the present paper we report an investigation into minima in yc brought about by variation of the chain length of the alkane, and by addition of n-alkanols ('cosurfactants ') to the system.Results are discussed in terms of the effective molecular geometry of the surfactant and cosurfactant, and penetration of alkane into the surfactant monolayer. A thermodynamic treatment of the effect of cosurfactant demonstrates that minimum yc results when the ratio of surfactant to cosurfactant is equal at the plane alkane-water interface and in mixed aggregates, as expected from the simple geometrical picture. As discussed elsewhere,'. in alkane-aqueous NaCl systems containing a micelle-forming surfactant such as diethylhexyl sodium sulphosuccinate (AOT), the oil-water interfacial tension y falls as the surfactant concentration increases and then attains a constant value yc at and above the concentration where surfactant aggregation occurs (c.m.c.).The aggregation may occur either in the oil phase or in the aqueous phase, depending on conditions. The tension yc can be made to pass through a minimum by adjustment of the temperature or salt concentration. The significance of these effects has already been explored.'* In this paper we will be concerned with the way in which yc depends upon the chain length of the alkane and upon the concentration of added n-alkanols, termed cosurfactants in this context. It is known2 that minimum yc results for the condition where phase inversion occurs, i.e. where the coarse emulsion formed by agitation of the equilibrium oil and aqueous phases changes (inverts) from an oil-in-water (O/W) to a water-in-oil (W/O) emulsion, or vice versa. In cases where an O/W emulsion is given, surfactant is found to reside mainly in the aqueous phase, which is a swollen micellar solution (or O/W microemulsion). The equilibrium oil phase in a coarse W/O emulsion, where all of the aggregated surfactant is present in the oil, is a W/O microemulsion.The way in which the shape, size and curvature of the surfactant aggregates change as phase-inversion conditions are approached has been discussed in some detail in terms of surfactant molecular geometry by Mitchell and Ni~~ham,~ and in terms of a statistical mechanical model by Mukherjee et aL4 We have sought in this paper to apply these ideas to changes in yc, and to extend our earlier thermodynamic treatment', of the occurrence of tension minima with respect to temperature and salt concentration to include effects due to addition of cosurfactants.17551756 Tension Minima in AlkaneAqueous NaCl-A0 T Experimental The heptane, Aerosol OT, and water samples have all been fully described in ref. (2). The other alkanes, from various sources, had purities close to 99% as determined by g.1.c. in this laboratory. The alkanols, also from various sources, had similar purities and were used without further treatment. All alkanes were passed through chromatographic alumina immediately prior to use to remove small amounts of polar materials. Surfactant concentrations were obtained by the hyamine titration m e t h ~ d . ~ Conduc- tances of emulsions were determined using a Jenway PCM3 digital conductivity meter.Interfacial tensions were measured using a Kruss spinning-drop tensiometer ( 10-4-20 mN m-l) and a Kruss KlO automatic tensiometer (3-72 mN m-l), as described elsewhere. The tension data in fig. 9 (later) for systems containing alkanols requires some comment. In experiments leading to these data a solution of the alkanol in heptane was injected into the aqueous surfactant in the spinning-drop tensiometer; the two phases were not pre-equilibrated. The alkanols octanol and higher distribute strongly in favour of the oil phase in pure oil-water systems and the concentration in the oil phase is not therefore expected to be reduced significantly by loss to water. Some alkanol will, how- ever, be incorporated into surfactant aggregates as discussed in detail later (see fig.13). We have therefore used surfactant concentrations only ca. 10% in excess of the c.m.c. so that all alkanol concentrations used greatly exceed the concentrations of aggregated surfactant, and the alkanol activity is unlikely to be significantly affected by solubil: isation. Loss of hexanol and pentanol to water may be more significant than for the higher alkanols, but since these two lower alkanols do not reduce the tensions (fig. 9, later) this is not a problem in the context. Results and Discussion The way in which surfactant structures in oil-water systems change as phase inversion is approached has been discussed by Mitchell and Ninham3 in terms of a ‘packing ratio’ P = v/aolc, where a, is the cross-sectional area of a surfactant head group, and v and lc are the volume and effective length of the hydrocarbon tail.For P < 1 normal structures (i.e. structures in water in which the head group is on the outer side of a convex curved surface) are predicted, whereas for P > 1 inverse droplets or micelles (in oil) are expected. For P = 1 phase inversion occurs and here the surfactant ‘prefers’ to be at a plane interface, and the tension of the plane oil-water interface (i.e. 7,) assumes its minimum value. Changes in a normal micellar system which serve to reduce a,, (e.g. increase in temperature or salt concentration) or to increase v (e.g. increased penetration of alkane into the surfactant tail region of the aggregate) bring the system closer to phase inversion and lead to a reduction in yc.At phase inversion v/lc = a,, i.e. effective cross-sectional areas of head and tail regions are equal, and P = 1 ; beyond this, inverted structures in the oil are the preferred state and the tension of the plane interface rises. When P is not unity, the difference in the cross-sectional area of the molecule on the two sides of a plane monolayer gives rise to a bending stress in the film. Mukherjee et aL4 take explicit account of the penetration of alkane molecules into the tail region of surfactant in spherically curved aggregates. It is concluded that as a result of the entropy of mixing, the shorter alkanes penetrate into surfactant layers more effectively than do longer chain homologues. In geometrical terms, penetration increases the effective value of v and gives an increase in P.Thus shorter chain alkanes tend to induce inverted surfactant structures, i.e. W/O microemulsions. From this we would expect that if a series of alkanes were used such that phase inversion of an oil- water-surfactant system occurred within this range, a minimum in yc would also result. It is predicted by the theory of Mukherjee et al. that the effect of cosurfactant addition should depend upon the cosurfactant chain length. In terms of the Mitchell and Niqham100 80 n E 2 5 2 40 d 60 a e, .- s 20 0 5 R. Aveyard, B. P . Binks and J. Mead 10 N 15 1757 I00 00 n E % 60 2 Y n 5 40 .- t- 0 4 20 0 Fig. 1. Distribution of AOT between aqueous NaCl(O.0684 mol dm-3) and n-alkanes, chain length N, at 298 K.The phase-volume ratio of oil: aqueous phase is 1 : 1. The dotted line shows surfactant lost to the surfactant-rich phase. The overall concentration of AOT is 17 mmol dm-3. picture, the presence of an alkanol in a surfactant film can affect the operative values of both v and a,, and can, depending on the alkanol chain length, cause an increase or a decrease in P and hence in yc. With the above considerations in mind, we now report and interpret the effects of alkane chain length and cosurfactant addition on values of yc in systems containing AOT. The following nomenclature will be used: Subscripts 1, 2 and 3 denote surfactant, salt (NaCl) and cosurfactant, respectively; m = concentration; m; = salt concentration giving minimum y c ; y, yc = tension of plane alkane-aqueous solution interface for m, < c.m.c.( y ) and rn, >*c.m.c. (yc); c.m.c. = rn, in aqueous phase for onset of surfactant aggregation in either aqueous or alkane phase; f+ = mean ionic activity coefficient. Surfactant is present in swamping excess of NaCl and it is assumed thatf, for salt and surfactant are equal; r = surface excess, which for strong adsorption is efffectively equal to surface concentration; T, = I' of surfactant in saturated monolayer; A, = l/l?,, area per surfactant molecule in saturated film; A: = limiting value of A, for high m,; N = alkane chain length. Effects of Alkane Chain Length Surfactan t Distribution The distribution of AOT above the c.m.c. between aqueous NaCl (m, = 0.0684 mol dm-3) and a range of n-alkanes at 25 "C is represented in fig.1. For N < 9, much of the surfactant is in the alkane and, from previous work using heptane,, we presume the oil phases are dilute W/O microemulsions and that m, in the aqueous phases is approximately equal to the c.m.c. For N 2 10, the surfactant is largely in the aqueous phase above the c.m.c. The trends are in accord with the findings of Mukherjee et aL4 that for lower N surfactant tends to reside in the oil phase as inverted structures. In the region of N = 9, much of the AOT is present in a third, surfactant-rich phase1758 Tension Minima in Alkane-Aqueous NaCl-A0 T Ol I 5 10 1 N I 5 Fig. 2. Conductivity of emulsions formed on stirring alkanes, chain length N , with 0.05 mol dm-3 AOT in 0.0684 mol dm-3 NaCl at 298 K.The phase-volume ratio of alkane: water = 1 : 5. (dotted line in fig. l), which may be a bicontinuous microemulsions in which the oil-water interface has zero net curvature. That phase inversion occurs around N = 9 (for the same m,) can be seen from the conductivities of the coarse emulsions, depicted in fig. 2. For N < 9, the continuous phase is the alkane (low conductivity), i.e. the emulsion is of the W/O type as expected from the distribution data. For N > 9, the conductivities rise sharply, indicating phase inversion. As will be seen later, as well as surfactant transfer and phase inversion, a minimum yc is also observed for N = 9. Eflect of Salt Concentration on Tensions Effects of salt have been discussed in some detail previously.l* It has been shown that for a given alkane yc passes through a minimum as m, is increased (see fig.3) and that phase inversion and surfactant transfer between phases coincides with the occurrence of minimum yc. These effects result from a reduction of lateral electrostatic repulsion and possibly from hydration of surfactant head groups caused by salt addition. This reduces a, and increases P, as discussed earlier. An alternative and presumably equivalent explanation was given by us previously1. in terms of surfactant dissociation. Salt reduces the degree of dissociation, h, of surfactant in micelles in the aqueous phase until it becomes equal to that, %, of surfactant at the plane alkane-aqueous phase interface, when minimum yc is achieved. The equation relating yc and mNa (the total counterion concentration) was given for constant temperature as We have shown2 that where heptane is the oil Tcl = 0 at minimum yc, but not for other values of mNa.If Tcl is set equal to zero for all mNa the predicted (In yc, mNa) curve has the minimum at the correct mNa, but there are deviations between predicted and experimental yc on either side of this salt concentration. Nonetheless, the correct general features are given when Tcl is set equal to zero and we proceed on that basis here. UsingR. Aveyard, B. P . Binks and J. Mead 1759 I I 0 0.0 5 0.10 mNa /mol dm-3 Fig. 3. Variation of y, with total Na+ concentration. (a) Experimental results for AOT in aqueous NaCl against n-alkanes at 298 K. Points 0, a, 0 and 0 are for heptane, nonane, undecane and tetradecane, respectively. (b) Curves generated by eqn (3) using for curves (A)-@), respectively, values for d In c.m.c./d In mNa of 0.8550, 0.8423, 0.8315 and 0.8300.NaCl as the salt, setting TCl = 0 and assuming for the moment rl is constant, eqn (1) becomes :2 yc = -mr,i -6.83 x 10-4(in~,,)3-0.025(inmNa)2 ) ln -.a] + B (3) d 1nc.m.c. d lnm,, where B is an integration constant. We find experimentally that the (ye, m,) curves depend markedly on alkane chain length as seen in fig. 3 (a). From eqn (l), assuming Tcl is zero, we see that the shift in the position of minimum yc is associated with different values of d lnc.m.c./d lnm,, for various N;1760 Tension Minima in Alkane-Aqueous NaCl-A0 T 7 . 0 n 7.5 E I -3 - > 2 ci t: W 7 8 . 0 a . 5 3.5 3.0 2.5 2.0 -In (rnN,/mol dm-3) Fig. 4.Variation of the c.m.c. of AOT at 298 K, in the presence of alkane phase, with total Na+ concentration. Results for systems containing heptane and tetradecane, respectively, are denoted by 0 and 0. for a given system d hc.m.c./d hm,, is known to be constant. A series of curves generated by eqn ( 3 ) using slightly different values of d lnc.m.c./d lnm,, is shown in fig. 3 (b); the series closely resembles that observed experimentally. Previously we found, using heptane, that d lnc.m.c./d lnm,, = -0.855., In the present work we have determined c.m.c. tensiometrically2 in the presence of tetradecane; the plot of In c.m.c. versus In mNa is shown in fig. 4 and d In c.m.c./d In mNa is found in this case to be - 0.832. As expected from eqn (1) and fig.3 (a), this is slightly lower than the value for the system containing heptane. To give the minimum yc at the salt concentration shown in fig. 3(a) (m, = 0.089 mol dm-9, d In c.m.c./d In mNa for the tetradecane system would have to be -0.830. Effect of Salt Concentration and N on Limiting Areas The simple geometrical ideas discussed earlier imply that the area per surfactant molecule, A,, in a saturated film at a plane oil-water interface should be governed (perhaps mainly) by the headgroup area a, at low m,, where strong lateral electrical repulsion operates between head groups. Alkane penetration into the surfactant chain region may also play a part as discussed below. The area A, should fall with increasing m2 and would ultimately assume an effectively constant value, AS, for the salt concentration rn, which corresponds to minimum yc.The value of A: is expected to be determined by the size of the surfactant chains and by the degree of penetration of alkane into the monolayer, and so should be higher in the presence of the shorter chain alkanes. From fig. 3(a) it is expected that the attainment of Al, should occur at higher m, for the longer-chain alkanes.R. Aueyard, B. P . Binks and J . Mead 1761 11 9 7 -In ([AOT]/mol dm-3) 30 20 " I E z E -s 10 0 12 10 8 -In ([ AOT]/mol dm-3) Fig. 5. Variation of y with aqueous phase concentration of AOT at 298 K. (a) Air-aqueous solution interface; points 0, 0, 0, (> are for salt concentrations of 0.0257, 0.0428, 0.0513 and 0.1027 mol dm-3, respectively. (b) Alkane - 0.1027 mol dm-3 NaCl (aq) interface; 0, 8, (>, 0 are for hexadecane, dodecane, nonane and heptane, respectively.To test these ideas we have determined A, as a function of rn, in systems containing heptane and dodecane. Values of A, (at the c.m.c.) were obtained from plots of y against lnm, (examples of which are depicted in fig. 5) using the form of the Gibbs equation:' A = - k T f 1 + [c.m.c./(c.m.c. + m,)]} d In m,/dy. (4)1762 Tension Minima in Alkane-Aqueous NaCI-A0 T 1.2 1.0 N E --.- VJ T 0 . 8 0 . 6 0 0.05 0.10 0.15 [ NaCl]/mol dm-3 T 0.60 ~ 10 15 N Fig. 6. (a) Variation of the limiting area A, with salt concentration, m2, for AOT at the heptane-solution interface (0) and dodecane-solution interface (0) at 298 K. (b) Variation of A: (see text) for AOT at 298 K with alkane chain length N.R. Aveyard, B.P . Binks and J. Mead 1763 In the region of rn, studied, y is, within experimental error, a linear function of lnrn, for all the systems, i.e. A becomes effectively equal to A , well below the c.m.c. We have also determined A, as a function of rn, for the air-aqueous solution interface for comparison; some of the ( y , In m,) data are shown in fig. 5 (a). Values of A, are plotted against m, in fig. 6(a), where the results for the air-solution interface have been omitted for clarity. A, falls with m2 and as predicted becomes effectively constant in the correct range of m,; also A; is greater in the presence of heptane (0.73 nm2) than dodecane (0.63 nm2). We can also see that A; is achieved at a higher m, for dodecane than for heptane.We have determined A: (for m2 = 0.103 mol drn-,) for systems containing other alkanes also and it is seen [fig. 6(b)] that A: falls smoothly with N . The results for the air-solution interface are a little puzzling. For this interface A: is 0.73 nm2, close to that for the heptane-aqueous solution interface, and larger than that for the interfaces involving the larger alkanes. The volume u of the tail region of an AOT molecule, calculated from the molar volume of ethylhexane (noting that a terminal CH, group is replaced by a CH, group in the AOT), is 0.479 nm3. The tail length, obtained from molecular models, ranges from 0.65 to 0.9 nm, depending on conformation, so one might expect a chain cross-sectional area of between 0.74 and 0.53 nm2.It would appear therefore that at the air-solution interface the AOT molecule adopts a conformation such that the tail length is minimised. This effect is presumably dominated by the unfavourable chain/air interactions in comparison with both intra- and inter-chain interactions. At the oil-solution interface favourable alkane-chain interactions serve to stabilise the extended conformations and thus can reduce the cross-sectional area. Eflect of Alkane Chain Length at Constant m, From previous work1? we know that when dyc/d lnrn, is negative, surfactant resides almost entirely in the aqueous phase, and when it is positive all the aggregated surfactant is present in the alkane. For a given m, therefore it can be appreciated from fig. 3(a) that yc will pass through a minimum with respect to N and that (as found in the distribution experiments) aggregated surfactant will be present in the alkane for small N , and in the water for large N .Values of yc are plotted against N in fig. 7 for systems containing AOT and 0.068 mol dm-, NaCl, and a minimum yc is observed for N = 9, in accord with the distribution data shown in fig. 1. Effects of Cosurfactant As already discussed, addition of alkanols as cosurfactants can, depending on the chain length of the alkanol, take an oil-water-surfactant system either closer to or further from phase inversion and by implication cause, respectively, a decrease or an increase in yc. We have investigated here the effects of a range of n-alkanols in systems containing AOT, aqueous NaCl and heptane and we have proceeded as follows.We consider two points on the (log yc, m,) curve [fig. 3 (a)], one (A) at low salt concentration (m, < m,*) and the other (B) at m, > m:. At A, a, is high and greater than v/Zc so that P c 1 in the Mitchell and Ninham treatment. If we add an alkanol with a long chain we expect the tail region of the film to swell relatively more than the head region and hence force the system towards, and possibly through, phase inversion. On the other hand at B, a, < u/Zc ( P > 1) and addition of the same alkanol will increase the disparity between a, and v/Zc and push the system further from phase inversion, hence causing an increase in yc. That these effects can be realised in practice is clearly demonstrated by the results in fig.8. Starting with the system at A (m, = 0.017 mol dm-3), dodecanol was added to the heptane. As the mole fraction x, is increased yc falls and passes through a minimum, as predicted. At the mole fraction corresponding to minimum yc, phase inversion occurs and surfactant transfers to the oil phase (as determined experimentally by surfactant1764 Tension Minima in Alkane-Aqueous NaCl-A0 T 5 10 N 15 Fig. 7. Variation of yc with alkane chain-length N for AOT at the alkane - 0.068 mol NaCl interface at 298 K. 0 n 1 E z E s - I --. W M - 7 2 3 0 5 10 lo3 a3 Fig. 8. Variation of yc with the mole fraction activity a3 of dodecanol in heptane at 303 K. The AOT is above the c.m.c., and the NaCl concentrations are (0) 0.0171 and (0) 0.1027 mol dm-3.R.Aveyard, B. P . Binks and J. Mead 1765 I 0 - 1 E z E s I --. W M c1 ‘ 2 3 L I I I 0 10 20 30 40 Fig. 9. Variation of yc with mole fraction x, of various alkanols in heptane at 303 K. The AOT is above the c.m.c., and the NaCl concentration is 0.0171 mol dm-3. 0, 0, a, 0 and a are for hexadecanol, nonanol, octanol, hexanol and pentanol, respectively. 103 x3 analysis). Starting from the system at B however (where m, = 0.103 mol dm-,), addition of dodecanol causes only an increase in yc. Shorter chain-length alkanols will be expected to be less efficient than dodecanol in promoting phase inversion, as already discussed. In conformity with this we find that the reduction in yc is smaller (for a given x, in heptane) the shorter the alkanol chain (fig. 9). Indeed, hexanol leaves yc unchanged and pentanol gives an increase.Thermodynamic Treatment Changes in y for a system in which m, and rn, may vary (for constant N , T and m,) may be writ ten The activity a, is included rather than m3 since the alkanols are present largely in alkane at concentrations where activity coefficients f3 differ significantly from unity as a result of autoassociation.8 The concentrations in the aqueous phase, m,(aq), at distribution equilibrium will be very small so that the aqueous solutions can be regarded as ideal and d lnm,(aq) = d lna, = d lnm,f,, where m, is the oil-phase concentration. From the Gibbs equation it is readily shown that, in the presence of swamping electrolyte, to a good approximation ( a y / a In ml)m3 = - RTT,1766 Tension Minima in Alkane-Aqueous NaCl-A0 T ‘0 - I E z E + \ 5 0 9 8 7 -In ([AOT]/mol dm-3) Fig.10. Variation of y for the heptane - 0,0171 mol dm-3 NaCl interface with aqueous phase AOT concentration at 303 K for various mole fraction activities, a3, of dodecanol in the heptane. a3 = (0) 0, (0) 5.15 x and (0) 8.79 x At the c.m.c., from eqn (5)--(7) d 1nc.m.c. d lna, Hall9 has shown that for an ionic surfactant in supporting aqueous electrolyte in the presence of a sparingly soluble solubilised additive (say long chain alkanol) d 1nc.m.c. N3 d lnm,(aq) - Nl - -- (9) where N are average numbers of molecules per micelle. In our systems the aggregates can be in either the aqueous or the oil phase, depending on conditions, and we also have alkane present. However, the alkane activity may be assumed to remain constant, so eqn (9) is still valid.From eqn (8) and (9) we see that -- dye - - RT(T, - rl N3/N1). (10) d lna, The quantity r3 is the number of moles of alkanol in an area of surface containing rl mole surfactant; rl N,/Nl is the number of moles of alkanol in micelles containing rl mole surfactant. Thus (r, - rl N3/N1) is an excess of alkanol (associated with rl mole surfactant) in the plane surface over that in micelles. Clearly when r3/r1= N3/N1, yc is minimum. Thus, as in the case of the (ye, rn,) and (ye, T ) curves1* we see that minimumR. Aveyard, B. P. Binks and J. Mead 6 0 5 10 lo3 u3 1767 Fig. 11. Surface pressure, n, of dodecanol at 303 K as a function of mole fraction activity a3 in heptane, for various aqueous-phase concentrations of AOT in 0.0171 mol dm-3 NaCl.0, 8 and 0 are for [AOT] = 5.54 x 1.02 x and 3.72 x lop5 mol dm-3, respectively. yc occurs when there is some kind of equivalence between the plane oil-water interface and the aggregates in equilibrium with the interface, in this case the mole ratio of cosurfactant to surfactant. The equivalence is presumably arrived at by the increase in aggregate size as phase inversion and minimum yc are approached. The third surfactant- rich phase which is often observed when ultralow tensions are produced could be regarded as an 'infinite' aggregate in which the surfactant film is effectively planar. We have determined values of rl, at the c.m.c. from (y, In m,) data (examples of which are shown in fig. lo), obtained in the presence of eight different concentrations of dodecanol.From these same data we have extracted surface pressures (lowerings of y ) of alkanols as a function of a, for various m,, sample results being shown in fig. 1 1. From this we have determined r, and hence r3 at the c.m.c. by extrapolation. Activity coefficients of dodecanol in heptane have been assumed to be the same as those for dodecanol in octane.* Values of rl and r3 at the c.m.c. for eight alkanol activities in heptane at 30 "C are given in table 1. Values of the c.m.c. have also been obtained (tensiometrically) as a function of alkanol activity and these are plotted in fig. 12. As the changes in the c.m.c. caused by alkanol are small, values of N1/N3 cannot be reliably determined by the use of eqn (9).Therefore the values of r3/r1, 1/RTI', and yc were fitted to equations in terms of alkanol activity (giving the full lines shown in fig. 8 and 13), and values of N1/N3 subsequently determined using eqn (10). The ratios N3/N1 and r3/r1 are plotted as a function of alkanol activity in fig. 13. At low a,, r3/r1 > N3/N1; at high a,, N3/N1 > I'3/r1; and at a, corresponding to minimum yc, the two ratios are equal at 0.35, i.e. minimum yc is given for a monolayer in which there 59 FAR 11768 Tension Minima in Alkane-Aqueous NaC1-A0 T Table 1. Interfacial concentrations of AOT and dodecanol at the heptane4.017 mol dm-3 NaCl interface as a function of mole fraction activity, a3, of dodecanol in heptane at 303 K lo3@, r,/10-6 mol m-2 r,/lO-’ mol mP2 r3/r1 0 3.66 5.15 6.13 6.95 8.07 8.79 9.51 10.12 2.1c 2.05 1.99 1.95 1.91 1.87 1.85 1.82 1.78 0 2.9 1 4.09 5.08 5.68 7.03 7.59 8.22 9.02 0 0.14 0.21 0.26 0.30 0.38 0.41 0.45 0.51 6.9 7 .0 n rn I E a CI z > 7.1 0 E 6 c I W - 7.2 7.3 4 6 8 10 103 u3 Fig. 12. Variation of the c.m.c. of AOT at 303 K with dodecanol activity in heptane - 0.0171 mol dm-3 NaCl systems. Points are experimental, the full line is obtained as described in the text. are ca. 3 AOT molecules to one alkanol. The mole fraction of alkanol in the inverted structures rises sharply with increasing a3. Although, as mentioned, N3/N, cannot be reliably calculated from c.m.c. values, we can nevertheless demonstrate that the results in fig. 12 are consistent with the rest of the data. Integration of eqn (8) yields d lna,.d lna,0 R. Aveyard, B. P . Sinks and J. Mead 1769 I I I I I I I I I I I I I I I I I I I I I I I 5 103 u3 10 Fig. 13. Molar ratios of dodecanol to AOT at 303 K (-) at the heptane-solution interface (T3/T,) and (---) in aggregates (N,/N,) as a function of dodecanol activity, u3, in heptane. The integrals have been evaluated using the fitting equations referred to earlier; the numerical analysis is simple but cumbersome. The full line in fig. 12 is obtained after the suitable choice of a value for an integration constant, which only affects the vertical position, but not the shape of the line. As seen, the line follows closely the experimental c.m.c. values. Finally, the consistency of the simple geometrical approach with the preceding data is readily demonstrated.In a mixed surfactant<osurfactant film giving minimum y, we expect the average headgroup area to be equal to the average tail area so that where a, represents the chain cross-sectional area. In the absence of dodecanol = 0.73 nm2 (the value of A: in the heptane system) and a,,, = 0.79 nm2 [the value of A , for rn, = 0.017 mol dm-3, corresponding to the point A mentioned in connection with fig. 3(a) and 81, therefore the addition of alkanol must effectively swell the chain region more than that of the head region to achieve minimum y,. From a simple geometrical approach the alkanol (for which we estimate = 0.25 nm2 and = 0.06 nm2) can pack into the interface to produce this result when r3/r1 = 0.3. However, calculations by Jonsson et a2.l0 demonstrate that the electrostatic repulsions between ionic headgroups are reduced by the incorporation of an alcohol into the interfacial film, effectively reducing the value of a,,,. It is probable, therefore, that the equivalence of head and chain areas arises from a combination of these effects. We thank the British Petroleum Company for Extramural Research Award funding and for the award of a BP Research Studentship to one of us (B.P.B.). 59-21770 Tension Minima in Alkane-Aqueous NaC1-A0 T References 1 R. Aveyard, B. P. Binks and J. Mead, J. Chem. SOC., Faraday Trans. I , 1985, 81, 2169. 2 R. Aveyard, B. P. Binks, S. Clark and J. Mead, J. Chem. SOC., Faraday Trans. I , 1986,82, 125. 3 J. D. Mitchell and B. D. Ninham, J. Chem. SOC., Faraday Trans. 2, 1981,77, 601. 4 S. Mukherjee, C. A. Miller and T. Fort, J. Colloid Interface Sci., 1983, 91, 223. 5 V. W. Reid, G. F. Longman and E. Heinerth, Tenside, 1967, 4, 292. 6 S. Friberg, I. Lapczynska and G. Gillberg, J. Colloid Interface Sci., 1976, 56, 19. 7 E. Matijevic and B. A. Pethica, Trans. Faraday SOC., 1958,54, 1382. 8 R. Aveyard, B. J. Briscoe and J. Chapman, Trans. Faraday SOC., 1973,69, 1772. 9 D. G. Hall, in Aggregation Processes in Solution, ed. E. Wyn-Jones and J. Gormally (Elsevier, 10 B. Jonsson, P.-G. Nilsson, B. Lindman, L. Guldbrand and H. Wennerstrom, in Surfactants in Solution, Amsterdam, 1983), chap. 2. ed. K. L. Mittal and B. Lindman (Plenum Press, New York, 1984), vol. 1, p. 3. Paper 511130; Received 4th July, 1985
ISSN:0300-9599
DOI:10.1039/F19868201755
出版商:RSC
年代:1986
数据来源: RSC
|
14. |
Catalysis by amorphous metal alloys. Part 4.—Structural modification towards metastable states and catalytic activity of amorphous NI62B38ribbon alloy |
|
Journal of the Chemical Society, Faraday Transactions 1: Physical Chemistry in Condensed Phases,
Volume 82,
Issue 6,
1986,
Page 1771-1780
Hiromi Yamashita,
Preview
|
PDF (902KB)
|
|
摘要:
J. Chem. SOC., Faraday Trans. 1, 1986,82, 1771-1780 Catalysis by Amorphous Metal Alloys Part 4.-Structural Modification towards Metastable States and Catalytic Activity of Amorphous Ni,,B,, Ribbon Alloy Hiromi Y amashita," Masahito Yoshikawa, Takuzo Funabiki and Satohiro Yoshida Department of Hydrocarbon Chemistry and Division of Molecular Engineering, Kyoto University, Kyoto, Japan Alloys in the three different states, i.e. precrystallisation, metastable and stable crystallised states have been prepared by the structural modification of amorphous Ni,,B,, ribbon alloy by thermal treatment. The catalytic activities of the alloys in these states for the hydrogenation of ethene and hydrogenolysis of ethane and cyclopropane have been correlated to the surface states which were characterised by XRD, DTA, SEM, ESCA, electrical resistivity and hydrogen chemisorption measurements.The maxi- mum catalytic activity observed in the precrystallisation state is caused by the structural relaxation (rearrangement in short range) involving the change in the strength of the chemical bonding among nickel, boron and oxygen atoms. The aggregation of nickel atoms and the increase in the surface content of boron oxides, which is observed in the metastable state, reduce the number of active sites change the electronic state of surface nickel and bring about a great decrease in the catalytic activity. Further aggre- gation eventually brings about the crystallisation of the bulk alloy, leading to very low catalytic ability. When an amorphous alloy is crystallised by heating, several metastable states appear before the alloy attains the stable crystalline state3 It is known that several kinds of physical properties, such as electrical resistivity, mechanical strength, etc.change before crystallisation over the long In the previous studies6-13 of catalysis by amorphous metals, the catalytic activities of amorphous alloys were compared with those of fully crystallised alloys prepared by heating the amorphous alloys at very high temperatures. We found6-lo that the crystallisation of amorphous Ni-P and Ni-B alloys resulted in a significant decrease in the catalytic activity for the hydrogenation of olefins, and this was correlated with the change in number of the surface active sites and in the strength of the chemical bonding among metal, metalloid and oxygen.In order to deter- mine the catalytic properties of an amorphous alloy, it is very interesting and important to study the surface state and the catalytic ability of the alloy in the metastable states. In this study, a structure-sensitive reaction14-17 (hydrogenolysis of ethane and cyclo- propane) and an insensitive reaction (hydrogenation of ethene) were carried out,lo and the results of these reactions were compared with information on the surface state obtained by measurements of XRD, DTA, ESCA, SEM, hydrogen chemisorption and electrical resistivity. We found an unusual change in catalytic activity which is related to the formation of metastable structures between amorphous and crystalline states. 17711772 Metastable States and Catalysis Experiment a1 Catalyst and Pretreatment The Ni-B (B, 38 at %) amorphous alloy in the form of ribbons ca.2 mm wide and 10-20pm thick was prepared by the rapid quenching method using a single steel roll as reported previ~usly.~ The alloy was pretreated in the following order before the reaction, (1) with 1.5 mol dm-, HNO,, (2) with heat at varying temperatures in the range 573-783 K under 13.3 kPa hydrogen for 2 h, (3) under 6.7 kPa oxygen at 473 K for 1 h and (4) under 13.3 kPa hydrogen at 573 K for 2 h. The conditions of these pretreatments were determined by reference to the results of the hydrogenation of olefins which were reported previo~sly.~-~ Hydrogen was purified by a hydrogen diffusion purifier (Japan Pure Hydrogen Co., LS-09B) and commercial oxygen, olefins and alkanes were purified by vacuum distillation using a liquid-nitrogen trap.Hydrogenation and Hydrogenolysis Reactions The reactions were carried out in a conventional closed circulation system (total volume 300cm3). Purified hydrogen and olefins or alkanes were introduced into the reaction vessel under the following conditions: (1) hydrogenation of ethene (PH2 = 19.3, PE = 7.3 kPa, 373 K); (2) hydrogenolysis of ethane (PH2 = 10.6, PE = 5.3 kPa, 603 K); and (3) hydrogenolysis and hydrogenation of cyclopropane (PH2 = 10.6, Pc = 5.3 kPa; 573 K). The composition of the products was analysed by g.l.c., using a 2m column of Porapack Q and a 3 m column of dimethylsulpholane on C22. Initial rates of the hydrogenation and hydrogenolysis were estimated from the initial changes of pressure and conversion, respectively.The hydrogenolysis reaction of cyclopropane was accom- panied by the hydrogenation reaction to produce propane, and the selectivity of the hydrogenolysis reaction in total cyclopropane conversion14 was represented by S(% ) = 100A/(A + B), where A and B are the numbers of moles of two- and three-carbon compounds, respectively. Measurement of Chemisorption of Hydrogen To study the effect of thermal treatment at different temperatures on the surface state, measurements of chemisorption of hydrogen on the alloys which were pretreated successively, were carried out in a conventional volumetric adsorption apparatus at room temperature in a similar way as reported previo~sly.~*~ The amount of surface nickel was estimated by assuming an adsorption equilibrium according to the Langmuir isotherm.The assumption of the dissociative adsorption of hydrogen was supported by good Langmuir plots. Measurements of DTA and XRD The calorific values which correspond to the heat involved in the change from amorphous or metastable states to the fully crystallised state were measured to examine the states of the pretreated alloys. Differential thermal analyses (DTA) were carried out in flowing N, gas and the alloys were heated at a constant rate of 10 K min-l (Rigaku YGHD, YGSD). The X-ray diffraction analysis (XRD) (Rigaku geigerflex 2013) was carried out to examine the degree of crystallisation.H. Yamashita et al. 1773 I Y 100 0 600 700 800 TIK Fig.1. Effect of thermal pretreatment of amorphous Ni-B alloy on the catalytic activities. (A), Hydrogenation of ethene (PH2 = 19.3, PCzH4 = 7.3 kPa, at 373 K); (O), hydrogenolysis of ethane (PH2 = 10.6, PCzHs = 5.3 kPa, at 603 K); and (O), hydrogenolysis, accompanied by hydrogenation of cyclopropane (PHI = 10.6, PC3Hs = 5.3 kPa, at 573 K). Ordinate scale (A), x lo5; (0, O), x 108. Measurement of SEM The growth of crystallised compounds on the alloys was monitored by scanning electron microscopy (SEM) operated at an accelerating voltage of 20 kV (Hitachi X-650). The surfaces of the sample alloys, which were heated at various temperatures, were etched with 3 mol dm-, HNO, for 5 min before the measurements. Measurement of Electrical Resistivity To study the effect of the thermal treatment on the change in the structure of the alloy, electrical resistivity was measured by means of a four-probe technique. l* An amorphous Ni-B ribbon with 1.9 mm width and 36.4 mm length was placed in the frame made of quartz and heated at a constant rate of 2.5 K min-l. Measurement of ESCA Spectra ESCA spectra were recorded with Shimadzu ESCA-750 using Mg radiation (10 kV, 30 mA) in a similar way as described previ~usly.~~ After the pretreatments, alloys were placed in an ESCA analyser chamber under an Ar atmosphere.All binding energy values were referred to the value of the contaminant carbon [C(ls) = 285.0 eV] for convenience. Results Fig, 1 shows the effect of thermal treatment on the catalytic activity. The catalytic activities increased with temperature up to 623 K and decreased abruptly at higher temperatures. These changes in the catalytic activity were more prominent in the hydrogenolysis reactions than those in the hydrogenation reactions.The alloy exhibited the maximum activity in all the reactions investigated when it was heated at 623 K and very low activity when heated at 723 K. Fig. 2 shows the effect of thermal treatment on the relative catalytic activities for the1774 Metastable States and Catalysis 600 700 T/K 800 Fig. 2. Effect of thermal pretreatment of amorphous Ni-B alloy on the relative initial rate against the initial rate of hydrogenation of ethene. (O), Hydrogenolysis of ethane; (O), hydrogenolysis, accompanied by hydrogenation of cyclopropane; (O), selectivity for hydrogenolysis in the reaction of cyclopropane.0.3 - I 0 n Y o$ 0 \ $ Y $ c3 Q) Y 0.2 .- c Q) G a ccr Y c 0 !i 0 600 700 800 T/K Fig. 3. Effect of thermal pretreatment of amorphous Ni-B alloy on the amount of surface nickel atoms. hydrogenolysis of ethane and hydrogenolysis and hydrogenation of cyclopropane against that for the hydrogenation of ethene. The change of the selectivity for hydrogenolysis in the reaction of cyclopropane is also shown. The activities and selectivity exhibited the greatest change at 623-683 K. The relative activity for hydrogenolysis reactions against that for the hydrogenation of ethene increased at T > 623 K. The selectivity forJ . Cheni. Soc., Furuday Tram. I , 1/01. 82,part 6 PIate 1 Plate 1. SEM (at 20 kV) of Ni-B alloys heated at different temperatures.The surface of the sample alloys which were heated at (a) 623, (b) 683 and (c) 783 K were etched with 3 mol dm-, HNO, for 5 min before the measurements. Magnifications: x 2000 (a) and x 500 (b) and (c). H. Yamashita er al. (Facing p . 1775)H. Yamashita et al. 1775 600 700 800 TI K Fig. 4. Effect of therrnal pretreatment of amorphous Ni-B alloy on the measured calorific value in DTA. hydrogenolysis in the reaction of cyclopropane exhibited the minimum value at 668 K and increased at T > 668 K. The amount of surface nickel was estimated by measurement of hydrogen chemisorption on the Ni-B alloys after thermal treatment. As shown in fig. 3, the amount of surface nickel depended on the temperature of thermal pretreatment and decreased gradually with increasing temperature.The effect of thermal treatment on the calorific values measured by DTA and the XRD patterns of the alloys are shown in fig. 4 and 5, respectively. These figures indicated the following: (1) The alloy treated at T < 623 K, which exhibits the broad halo but no peaks of crystalline compounds in XRD and no change in the calorific value, is regarded as being in the amorphous state. (2) The alloy treated at 623-723 K, which exhibits both broad halo and peaks of crystalline BNi, and B,O, in XRD and the constant decrease in the calorific values, is regarded as being in the metastable state. Local rearrangement of the structure takes place prior to full crystallisation of the bulk alloy. (3) The alloy heated at T > 783 K, which exhibits no broad halo and many intense peaks of crystalline B,O,, BNi,, Ni, etc. in XRD and very small DTA calorific values, is regarded as being in the stable crystalline state.The geometrical surface states of three representative samples were examined by SEM and the images are shown in plate 1. The crystalline structure was hardly observed on the alloy in amorphous state, but the growth of the crystalline compounds of ca. ,urn was observed on the alloy heated at 683 K. These crystalline compounds grew in size and number with increasing temperature and covered the surface of the alloy heated at 783 K. Fig. 6 shows the temperature dependence of the electrical resistivity of amorphous Ni-B alloy. The electrical resistivity of the alloy in the amorphous state was high, but that of the alloy heated at T > 673 K decreased abruptly.Interestingly, the electrical resistivity changed stepwise even in the amorphous state, and several steps of the change were observed at T > 643 K. Fig. 7 shows the ESCA spectra of alloys heated at different temperatures [(a) 573; (b) 623; (c) 683; and ( d ) 783 K] and treated with oxygen and hydrogen, successively. A1776 Metastable States and Catalysis 30 40 50 60 291’ Fig. 5. X-ray diffraction patterns of Ni-B alloys heated at different temperatures. The alloys were heated at (a) 623, (b) 653, (c) 683 and ( d ) 783 K and were pretreated successively with oxygen and hydrogen. significant deposit of boron oxide was observed on the alloys when heated at 683 and 783 K. The binding energy of boron oxide (193.2 eV) and nickel (853.9 eV) of the alloy heated at 623 K were higher than those (192.5 and 853.5 eV) of the alloy heated at 573 K, respectively.Discussion Catalytic Activity It is known that when amorphous alloys prepared by the rapid quenched method are crystallised by heating, the precrystallisation state and several metastable states appear before the alloys attain the stable crystalline state.l, The effect of thermal pretreatment on the catalytic activity (fig. 1) indicated several interesting changes in the catalytic activity: (1) the catalytic activities become maximum when the alloy is heated at 653 K and decrease abruptly by the treatment at T > 653 K. (2) The catalytic activities of the alloys heated at T > 683 K are very small.(3) The changes in the catalytic activities owing to the thermal treatment are more prominent in the hydrogenolysis reaction than those in the hydrogenation reaction, as shown in fig. 2. It is reasonable to assume that these changes in the catalytic activities are related to the changes in the surface state which is brought about by the thermal treatment.H. Yamashita et al. 1777 T/K 500 700 2.28 2.24 c: .2 e 1 .C( c .c, ..-I 2.20 2.16 2.5 c: .? 1.5 -2 e 2.0 ‘h + .C( .c) 1.0 400 600 800 T/K Fig. 6. Temperature dependence of the electrical resistivity of amorphous Ni-B alloy. Yokoyama et aZ.l1? l2 reported that the activity of amorphous Fe- and Ni-based alloys for the hydrogenation of CO were several to several hundred times higher than those of the stable crystallised alloy of the same composition.They proposed that the active sites of amorphous and crystalline alloys were similar in nature but different in numbers, because both the alloys exhibited the same activation energy and kinetics. However, it seems unreasonable to believe that the high catalytic activity of the amorphous alloy can be explained only by the amount of active surface sites. We observed in our previous study8 that the catalytic activity of Ni-P amorphous alloy was higher than that of the crystalline alloy for the hydrogenation of olefins and CO and that not only the number of active sites, but also the turnover frequencies of the amorphous alloy were greater than those of the stable crystalline alloy. The results of the hydrogen chemisorption measurements shown in fig.3 indicate that the amount of surface nickel decreases gradually with increasing temperatures of pretreatment. The aggregation of nickel atoms and the increase in the surface content of boron oxides, which was observed in the ESCA spectra, must be responsible for the reduction of the amount of surface nickel atoms. However, the observations of the maximum activity and the abrupt decrease in the activity on the alloy heated at 623 K and at T > 623 K, respectively, cannot be explained only by the changes in the amount of surface nickel atoms. This result indicates that the catalytic activity per surface nickel atom is different, depending on the state of the alloy. DTA and XRD analysis show that there are at least three distinguishable states in alloys heated at various temperatures : the precrystallisation state (T < 623 K), the metastable state (623 -= T < 723 K) and the stable crystalline state (T > 723 K).The relation between the activity and the structure can be summarised as follows: (1) the1778 Metastable States and Catalysis N@P,,d B( 1s) 8 5 3 . 5 8 5 3 . 6 A 8 5 6 . 6 8 5 3 . 5 A 192.5 188.3 8 6 0 8 5 0 190 binding energy/eV Fig. 7. ESCA spectra of the surface of Ni-B alloys heated at (a) 573, (b) 623, (c) 683 and (d) 783 K and were pretreated successively with oxygen and hydrogen. Xcps: Ni(2p,,,) = (a)-(d) 1000, B( 1s) = (a, b) 200, (c, d ) 500. catalytic activity decreases abruptly on the alloy whose surface state is shifted from amorphous state to the metastable state.(2) The catalytic activity of the alloy in amorphous state is much higher than that of the alloy in the stable crystallised state. (3) The maximum activity is observed on the alloy in the precrystallisation state. Metastable and Stable Crystalline States The results of the structure-sensitive reactions (fig. 2) and the observations of SEM (plate 1) are very useful in clarifying these interesting changes in the catalytic activity. As shown in fig. 2, the change in the relative activity of hydrogenolysis to hydrogenation of ethene reflects the surface structure of the alloy heated at various temperatures. Since the structure-sensitive reaction involves multiplet ad~orption,l~-~~ the increase in the relative activity of hydrogenolysis to hydrogenation at T > 623 K indicates that the change of the surface structure involving the aggregation of nickel atoms is promoted by treatment at the higher temperatures.SEM images shown in plate 1 also indicate that large crystalline compounds grow to cover the surface of the alloys heated at T > 683 K. Since these crystalline compounds appear heterogeneously in the amorphous matrix [plate 1 (b)], the aggregation of nickel atoms is promoted by the growth of these crystalline compounds. The aggregation of nickel atoms and the increase in the surface content of boron oxide reduce the number of active sites and the excess aggregation so as to bring about the crystallisation of the bulk alloy which leads to the low catalytic activity. The rearrangement of the surfaceH. Yamashita et al.1779 structure over such a long range brings about change in the electronic states of the surface nickel species, causing the low catalytic activities of the alloys in metastable and stable crystalline state. However, this rearrangement of the surface structure over long range and the aggregation of surface nickel species cannot explain the maximum catalytic activity of the alloy heated at 623 K. It is important to characterise the surface state of the alloy in precrystallisation state. Precry st allisation St ate XRD and calorific values do not give information on the states of the precrystallisation state of alloys heated at T < 623 K, but the electrical resistivity is useful. As shown in fig. 6, the electrical resistivity of the alloy changes sensitively with the change in the structure of the alloy by heating. The electrical resistivity decreases abruptly with temperature at T > 673 K, when the metastable state is observed by DTA and XRD.However, even alloys in the amorphous state or in the precrystallisation state exhibited the stepwise electrical resistivity changes. It is known that the changes in some kinds of physical properties occur before crystallisation, without the evident change of structure over the long ~ange.l-~ These changes are considered to be caused by the structural relaxation which is accompanied by the topological short-range ordering or by the evolution of fine In this case, the stepwise and small change of the electrical resistivity in the precrystallisation state caused by heating at 623 K indicates that rearrangement of the structure in the short range occurs.This rearrangement seems to be accompanied by a change in the chemical bonding between the metal and metalloid atoms. As reported previo~sly,~ the binding energies of nickel and boron of the amorphous Ni-B alloy are shifted from those of pure nickel and boron, indicating the interaction between nickel and boron. The greater binding energies of nickel and boron oxide (853.9 and 193.2 eV) on the alloy heated at 623 K [fig. 7(b)] than those (853.5 and 192.5 eV) on the alloy heated at 573 K [fig. 7(a)], respectively, indicate that the oxidation of boron oxide is significant and the interaction among nickel, boron and oxygen become stronger after treatment with oxygen and hydrogen over the alloy heated at 623 K.This change in the surface state brings about electron transfer from nickel to boron and from boron to oxygen and produces the electron-deficient nickel which is effective in the activation of hydrogen. These changes in the interaction among atoms in the precrystallisation state are due to the rearrangement of the structure in the short range. Conclusion The increase in the catalytic activity of the alloy in the precrystallisation state, which is recognised to be the amorphous state by XRD and DTA measurements, is caused by the generation of electron-deficient nickel. This electron-deficient nickel is formed by structural relaxation in the short range accompanied by a change in the strength of the chemical bonding among metal, metalloid and oxygen.The abrupt decrease in the catalytic activity with crystallisation is caused by the decrease in the number of the surface active sites due to the aggregation of nickel atoms, the increase in the surface content of boron oxide and the change in the electronic state of surface nickel caused by inhomogeneities of the surface structure of the alloys which are in metastable and stable crystallised states. We thank Prof. J. Takamura, Prof. H. Shingu, Dr F. Nakamura and Dr R. Suzuki of the Department of Metallurgy of Kyoto University for their help in the measurements of electrical resistivity and SEM.I780 Metastable States and Catalysis References 1 T. Masumoto, A. Inoue and H. Kimura, J. Met SOC. Jpn, 1977,41, 730. 2 A. Inoue, T. Masumoto, M. Kikuchi and T. Mineyama, J. Met. SOC. Jpn, 1978,42, 294. 3 T. Masumoto, Y. Waseda, H. Kimura and A. Inoue, Sci. Rep. RITU, 1976, A29, 21. 4 H. H. Liebermann, C. D. Graham Jr and P. J. Flanders, ZEEE Truns. Magn., 1977, 13, 1541. 5 T. Egami, Muter. Res. Bull., 1978, 13, 557. 6 S. Yoshida, H. Yamashita, T. Funabiki and T. Yonezawa, J. Chem. SOC., Chem. Commun., 1982, 964. 7 S. Yoshida, H. Yamashida, T. Funabiki and T. Yonezawa, J. Chem. SOC., Faruday Trans. I , 1984,80, 8 H. Yamashita, M. Yoshikawa, T. Funabiki and S. Yoshida, J. Chem. SOC., Furuday Truns. I , 1985,81, 9 H. Yamashita, T. Kaminade, T. Funabiki and S. Yoshida, J. Muter. Sci. Lett., 1985, 4, 1241. 1435. 248 5. 10 H. Yamashita, T. Funabiki and S. Yoshida, J. Chem. SOC., Chem. Commun., 1984, 868. 11 A. Yokoyama, H. Komiyama, H. Inoue, T. Masumoto and H. M. Kimura, J. Cutul., 1981,68, 355. 12 A. Yokoyama, H. Komiyama, H. Inoue, T. Masumoto and H. M. Kimura, J. Chem. SOC. Jpn, 1982, 13 G. V. Smith, 0. Zahraa, A. Molnar, M. M. Khan, B. Rihterand W. E. Brower, J. Catul., 1983,83,238. 14 J. M. Beelen, V. Ponec and W. M. H. Sachtler, J. Cutal., 1973,28, 376. 15 V. Ponec and W. M. H. Sachtler in Proceedings of the 5th International Congress on Catalysis, ed. J. W. Hightower (North-Holland, Amsterdam, 1973), vol. I, p. 645. 16 J. H. Sinfelt, J. L. Carter and D. J. C. Yates, J. Cutul., 1972, 24, 283. 17 J. A. Dalmon and D. A. Marin, J. Catal., 1980,66, 214. 18 K. Osamura and F. Nakamura, Keikinzoku, 1983, 33, 55. 19 T. Masumoto and R. Maddin, Muter. Sci. Eng., 1975, 19, 1. 2, 199. Paper 511135; Received 5th July, 1985
ISSN:0300-9599
DOI:10.1039/F19868201771
出版商:RSC
年代:1986
数据来源: RSC
|
15. |
Electrode kinetics of the CdII/Cd–Hg system in ethylene glycol–water mixtures |
|
Journal of the Chemical Society, Faraday Transactions 1: Physical Chemistry in Condensed Phases,
Volume 82,
Issue 6,
1986,
Page 1781-1787
José Antonio Garrido,
Preview
|
PDF (542KB)
|
|
摘要:
J. Chem. SOC., Faraday Trans. 1, 1986,82, 1781-1787 Electrode Kinetics of the CdII/Cd-Hg System in Ethylene Glycol-Water Mixtures Jos6 Antonio Garrido,* Rosa Maria Rodriguez and Enrique Brillas Departament de Quimica Fisica, Universitat de Barcelona, Avda. Diagonal 647, 08028 Barcelona, Spain Javier DomCnech Departament de Quimica Fisica, Universitat Autdnoma de Barcelona, Bellaterra (Barcelona), Spain The kinetics of the CdII/Cd-Hg system in ethylene glycol-water mixtures containing different concentrations of LiC10, has been studied by cyclic voltammetry. At scan rates > 0.2 V s-l the system shows quasireversible behaviour, whereas at lower scan rates it behaves reversibly. The diffusion coefficient for Cd in mercury, the reversible half-wave potentials of the CdII ion, the transfer coefficient for the electroreduction of CdI' and the apparent standard rate constants of the system have been determined.The change in the kinetics with solvent composition at each LiClO, concentration tested is discussed in terms of existing models. Recent studies of the kinetics of the electroreduction of several cations with amalgam formation in different organic-solvent-water mixtures have shown that the parameters of these electrode reactions depend basically on the solvation energy, which represents the main contribution to the activation energy, as well as on the adsorption of the organic solvent, which modifies the structure of the double layer at the electrode surface. In particular, the change in rate constant with solvent composition has been explained by two different models.'* Thus, Behr et a1.l consider that if the reactant ion can enter the surface layer without kinetic limitations, a large part of the kinetic effect is due to the equilibrium distribution of the reactant between the bulk (a) and surface (a) phases, which is described by the coefficient P: P = exp (- aAoGt/RT) (1) aAuGt being the change in the free energy of transfer of 1 mol of ions between both phases.These authors propose the rate constant, k, in a given aqueous mixture to be determined by means of the expression where k, is the rate constant in the pure aqueous solution. This model seems to describe well the electrode processes in which plots of k against solvent composition show either a maximum or a minimums1. On the other hand, Broda and Galus2 suggest that for aqueous mixtures with solvents more basic than water, some of the reactant ions approaching the surface layer react with the surface organic solvent molecules, whereas the others react with the surface water molecules.Since two parallel reaction pathways can occur, the authors propose to calculate the value of k from k = k , P (2) where k, and ksolv are the rate constants in pure solvents and Osolv is the surface coverage by organic solvent molecules. This model seems to describe well the electrode processes 17811782 Electrode Kinetics of the CdII/Cd-Hg System in which a gradual decrease in k with increasing organic solvent composition is However, it cannot interpret the maxima or minima in the plots of k against solvent composition obsorved for the electroreduction of some cations, such as CdII, in different aqueous mixtures with solvents more basic than In a previous stud Y we investigated the polarographic reduction of the CdI1 ion in ethylene glycol(EG)-water mixtures, determining the standard free energy of transfer of 1 mol of CdII ions from water to each mixture, AG,".From these data we concluded that the ion always presents a greater stability in EG-water mixtures than in pure water, denoting a higher basicity of the mixtures. To provide a better understanding of the electrochemical behaviour of the CdII/Cd-Hg system in EG-water mixtures we now report the results of a kinetic study of this system by cyclic voltammetry in mixtures containing different concentrations of LiClO, as supporting electrolyte.The reaction media were chosen in order to investigate the changes in the kinetics under the influence of the EG content and supporting electrolyte concentration, especially in the water-rich region, where anomalous behaviour has been found in other organic solvent-water mixture^.^^ 5 9 5 9 Experiment a1 Ethylene glycol (Fluka, A.R. grade) was dried over 3 A molecular sieves. The water content determined by the Karl Fisher method was < 0.05 % . Lithium perchlorate (Fluka, A.R. grade) was dried at 130 "C under reduced pressure and kept dry afterwards. Cadmium sulphate (Merck, A. R. grade) was used without further purification. All solutions were prepared by weight with water obtained using a Millipore Milli-Q system.The cyclic voltammetry measurements were performed with a P.A.R. model 175 universal programmer, connected to an Amel model 55 1 potentiostat. The cyclic voltammograms carried out at scan rates, v: values of v < 0.200 V s-l were displayed directly on a Philips model 8043 X-Y recorder, whereas the measurements made at higher scan rates were previously recorded on a Nicolet model 309 1 digital storage oscilloscope. The ohmic-drop compensation of all voltammetric measurements was achieved with a positive feedback network using the same instruments. All voltammetric experiments were carried out in a three-electrode cell under nitrogen. The temperature was kept at 25.0 "C. A saturated calomel electrode (SCE), with an aqueous solution of NaC1, was used as the reference electrode and a Pt wire as the counter-electrode. The working electrode was a hanging mercury drop electrode with an area, A , of 0.0222 cm2.Solutions of CdII (5 x lo-, mol dm-3) in water, and EG-water mixtures of EG mole fraction between 0.10 and 0.70, containing concentrations 0.025, 0.050 and 0.100 mol dm-3 of LiClO, as supporting electrolyte were studied. The cyclic voltammo- grams of each solution were recorded in a scan rate range between 0.200 and 50 V s-l. Results and Discussion The CdII/Cd-Hg system exhibits a redox couple under all the experimental conditions studied, and shows different voltammetric behaviour depending on the EG content of the mixture, the LiClO, concentration and the scan rate. Values of the cathodic peak potential, ECp, and cathodic peak current, ZCp, obtained at a scan rate of 0.020 V s-l are listed in table 1.As can be seen, Pp is gradually shifted to less negative values when either the EG content in the mixture or the LiClO, concentration increases. On the other hand, I; gradually decreases with rising mole fraction of EG, in accordance with the progressive increase in viscosity of the reaction medium,' although it remains practically constant for a given mixture at different concentrations of LiClO,. From these latter data, the cathodic peak-current functions, y: = Z:/nFAc(D,, nFv/RT)1'2, were calculated em-J. A . Garrido, R. M . Rodriguez, E. Brillas and J. Domtnech 1783 Table 1. Cyclic voltammetric results of the CdI'/Cd-Hg system in water and ethylene glycol-water mixtures, containing different concentrations of LiClO,, at a scan rate of 0.020 V s-l and at 25.0 "C 0.00 0.10 0.20 0.30 0.40 0.50 0.60 0.70 2.5 5.0 10.0 2.5 5.0 10.0 2.5 5.0 10.0 2.5 5.0 10.0 2.5 5.0 10.0 2.5 5.0 10.0 2.5 5.0 10.0 2.5 5.0 10.0 -0.607 - 0.603 - 0.602 - 0.594 - 0.590 - 0.588 -0.583 -0.582 - 0.578 - 0.570 - 0.569 - 0.568 - 0.566 -0.562 -0.560 - 0.560 -0.552 -0.554 -0.552 - 0.546 - 0.544 - 0.546 - 0.544 -0.541 3.04 3.11 3.04 2.25 2.30 2.20 1.85 1.90 1.85 1.44 1.45 1.48 1.20 1.22 1.18 1.15 1.05 1.05 0.95 0.95 0.95 0.85 0.90 0.85 1.60 1.53 1.34 1.43 1.37 1.57 1.40 1.36 1.35 1.60 1.44 1.54 1.43 1.54 1.59 1.56 1.60 1.58 1.44 1.59 1.35 1.34 1.36 1.41 -0.589 - 0.585 - 0.584 -0.577 - 0.573 -0.571 -0.566 - 0.565 -0.561 -0.554 -0.553 - 0.552 -0.550 - 0.546 -0.544 -0.544 - 0.536 - 0.538 - 0.536 -0.530 -0.528 -0.530 - 0.528 - 0.525 a E/V us.SCE (aqueous solution of NaC1). ploying a number of transferred electrons per reactant ion, n, of 2 and the diffusion coefficients for the CdII ion, Dox, previously determined by p~larography.~ In all media the value of yi is found to be close to 0.45, as expected for a two-electron reversible reduction process controlled by diffusion.*, For a given solution and at scan rates lower than 0.2 V s-l the ratio of anodic to cathodic peak currents, c/Zi, is higher than unity, its value depending on the applied reversal potential, En. This is due to the effect of concentration of the Cd amalgam formed in the hanging mercury drop electrode because of spherical diffusion and the limited volume of the electrode, Recently, Tokuda et al.9 have studied theoretically this kind of process by cyclic voltammetry for a simple reversible system such as Mn+ + n e- = M-Hg (4) establishing the following equations to calculate the diffusion coefficient for M in mercury, Dred, and the reversible half-wave potential, Ef12, corresponding to the Mn+ electroreduction : ( 5 ) Dred = {[(G/Zi) - 1]/4.130 Dg$72r;1.273(RT/nFv)o.B37[(nF/RT) (E; - E1)]0*719 I- OSg6' elZ = Ecp + (RT/nF) [ 1.109 + 5.047 D~;r344D~~~8r;1~024(RT/nFv)o~512] (6) where ro is the electrode radius.Because the electroreduction of CdII is always reversible at 0.020 V s-l, as stated above,1784 Electrode Kinetics of the CdII/Cd-Hg System Table 2. Kinetic parameters of the CdII/Cd-Hg system in water and ethylene glycol-water mixtures, containing different concentrations of LiCIO,, at 25.0 "C cLiclo,/10-2 mol dm-3 k,/10-3 cm s-l a ki/10-3 cm s-I XEG 0.00 0.10 0.20 0.30 0.40 0.50 0.60 0.70 2.5 5.0 10.0 2.5 5.0 10.0 2.5 5.0 10.0 2.5 5.0 10.0 2.5 5.0 10.0 2.5 5.0 10.0 2.5 5.0 10.0 2.5 5.0 10.0 18.4 21.7 32.4 10.0 11.2 17.1 6.7 8.0 11.4 6.4 6.8 9.4 5.4 6.3 7.9 4.6 5.1 6.8 4.5 4.7 6.2 4.5 4.6 5.8 0.26 0.28 0.24 0.21 0.21 0.2 1 0.2 1 0.20 0.24 0.20 0.21 0.20 0.22 0.22 0.21 - 0.19 0.21 0.20 0.20 0.20 0.19 0.21 0.20 0.19 25.3 33.2 58.0 14.8 19.2 34.5 10.5 14.9 24.6 10.6 13.5 23.2 9.4 13.4 21.4 8.8 12.0 21.0 8.9 12.0 21.3 9.3 12.6 21.9 it seems plausible to consider that the CdII/Cd-Hg system behaves as a simple reversible one at low u values, and hence Dred for Cd in mercury and for CdII electroreduction can be determined by means of eqn (5) and (6), respectively.To corroborate all these suppositions, several cyclic voltammograms were recorded for each solution studied at 0.020 and 0.050 V s-l, with values of (Ec, -En) between 0.120 and 0.200 v . Values of Dred and Ell2 obtained at a scan rate of 0.020 V s-l by using the equations stated above are given in table 1. The diffusion coefficient for Cd in mercury remains practically constant under these experimental conditions, taking a value of (1.47 0.13) x cm2 s-l which agrees with those reported in the literat~re.~, lo The Ell2 values calculated by eqn (6) are in accordance with those previously determined by p~larography.~ In fact, they are shifted to less negative potentials when either the EG mole fraction or the LiC10, concentration rises.This behaviour can mainly be ascribed to the change in solvation energy for the CdII These results indicate that in EG-water mixtures the CdII/Cd-Hg system is always reversible at low scan rates. For each solution studied and at scan rates ~ 0 . 2 V s-l, the Zg/Zi ratio remains close to unity, although it decreases slightly as the scan rate rises. Increasing u, is always accompanied by a progressive decrease in y;. Simultaneously, EP is shifted to more negative values and the anodic peak potential, E i , to less negative ones. All these results are indicative of quasireversible behaviour for the CdII/Cd-Hg system under these conditions. The apparent standard rate constant, k,, of a simple quasireversible system can be calculated according to the expressionll k, = D~ktD~~-a'/2~~nFu/RT)112y (7)J.A . Garrido, R. M . Rodriguez, E. Brillas and J. Domtnech 1785 I I ' I XEG 0 0.20 0.40 0.60 Fig. 1. Surface coverage of the electrode by EG molecules at the potential of CdII electroreduction in various EG-water mixtures. where y/ is a function of (G-EC,)n and of the transfer coefficient for the reduction process, a. For this kind of processes, a linear plot of Eg against -log v, with a slope of O.O296/m V per decade at 25.0 "C, is expected from a given scan rate value;12 under such conditions the transfer coefficient can then be determined. For the cathodic peak of the CdII/Cd-Hg system under quasireversible conditions, good linear correlations between Ecp and -log v were always obtained for v > 5 V s-l.Assuming that the slope of these plots was O.O296/m V per decade and n = 2, the transfer coefficient for CdII electroreduction is always found to be ca. 0.22, as can be seen in table 2. Similar a values have also been reported for this process in other reaction media.3* l1 In the scan-rate range between 0.3 and 3 V s-l the difference (E;-Eg) gradually increases with rising v from ca. 0.045 to 0.110 V. Under these conditions the value of k, for each redox couple was calculated by means of the eqn (7), using the y/-parameter determined from the plot of y/ against (E; - Ecp)n theoretically established by Nicholson? The mean value of k, obtained for each solution, with an accuracy of 9%, is given in table 2. A progressive decrease in k, with rising EG mole fraction can be observed, whereas, for a given mixture, k, gradually increases when the LiClO, concentration rises.The change in the apparent standard rate constant with the reaction medium can be better understood by taking into account that k, depends on the activity coefficient for the CdII ion, yox, according to the expression12 where k,O is the apparent standard rate constant at yox = 1. To determine ki for each solution tested, the corresponding value of yox was estimated by means of the equation13 log10 Yox = (&,/&)"/" ~ ~ ~ 1 0 ~ Y o x ~ w (9) where E, and E are the dielectric constants of water and the mixed solvent, respectively, and (yOx), is the activity coefficient in an aqueous solution with an equal LiC10, concentration.Values of dielectric constants and activity coefficients in aqueous solutions were taken from data reported in the literat~re.l**~~ The ki values thus obtained are compiled in the last column of table 2. For all mixtures a progressive increase in ki with rising LiClO, concentration is found. This kind of variation has also been reported for the CdII/Cd-Hg system in aqueous1786 Electrode Kinetics of the CdII/Cd-Hg System 6o b 40 c( I v) 5 2 3 9 20 0 0 0 0 m 2 1 0 . a 0 0 0 0 3 0 0.20 0.40 0.60 xEC Fig. 2. Dependence of experimental values of kz for the Cd"/Cd-Hg system on EG mole fraction, at LiClO, concentrations of 0, 0.025; e, 0.050 and a, 0.100 mol dm-3. Solid lines (l), (2) and (3) are the corresponding theoretical curves calculated from the model of Broda and Galus.* mixtures with several aliphatic alcohols,6 being qualitatively ascribed to a change in the double-layer structure.Unfortunately there are not sufficient experimental data in the literature to carry out this analysis in EG-water mixtures. On the other hand, for each LiClO, concentration tested, kz decreases monotonically as the EG mole fraction increases. This behaviour has not been observed for the CdII/Cd-Hg system in aqueous mixtures with dimethyl sulphoxide3 and aliphatic alcohols,6 because in such media the corresponding plots of k against solvent composition show maxima or minima in the water-rich region. The change in the kinetics with solvent composition for EG-water mixtures could then be ascribed to the greater adsorption of EG on the mercury electrode, particularly in the water-rich region.l6V l7 This suggests that the change in kinetics with solvent composition found for EG-water mixtures could be described by eqn (3) proposed in the model of Broda and Galus.2 In fact, EG-water mixtures are more basic than water, as is required for the application of this model.To corroborate whether eqn (3) adequately describes the change in kinetics stated above, the theoretical value of k; for each solution was calculated. For this, kz,w values were taken from table 2, whereas values of kz,EG were obtained by extrapolation at xEG = 1. The surface coverage of the electrode by EG molecules, &G, at -0.600 V (i.e. at the potential of Cd'I electroreduction), in the EG-water mixtures studied containing a concentration 0.1 mol dm-3 of LiClO,, was estimated from data reported in the literature.l6* l7 The same &G values were employed for the other LiClO, concentrations tested.Fig. 1 shows the variation of O,, with EG mole fraction. The dependence of experimental ki values on EG mole fraction at LiClO, concentrations of 0.025, 0.050 and 0.100 mol dm-3, as well as the corresponding theoretical curves calculated by using eqn (3) of Broda and Galus, are presented in fig. 2. Good agreement between the experimental and calculated values is found, indicating that the model of Broda and Galus2 allows one to explain the change in the kinetics of the Cd'I/Cd-Hg system with solvent composition for EG-water mixtures. Finally, according to the abovementioned authors a linear dependence of activation energy on surface coverage must be expected.This is difficult to interpret if the variqtionJ. A . Garrido, R. M. Rodriguez, E. Brillas and J. Domtnech 1787 in the solvation energy for the CdT* ion (AG,") is the main contribution to the change in activation energy, as is usually considered to be the case, because AG," increases linearly with the EG mole fraction,' whereas most of the increase in activation energy should take place up to xEG = 0.4, as can be deduced from fig. 1. The increase in activation energy could then be ascribed5 to the activation energy needed for the reorganization of the solvation shell of the reactant on entering the surface layer. References 1 B. Behr, J. Taraszewska and J. Stroda, J. Electroanal. Chem., 1975,58, 71. 2 J. Broda and 2. Galus, J. Electroanal. Chem., 1981, 130, 229. 3 J. Taraszewska and A. Walega, J. Electroanal. Chem., 1984, 171, 243. 4 A. Broda, J. Stroka and Z. Galus, Electrochim. Acta, 1983, 28, 817. 5 W. Jaenicke and P. H. Schweitzer, 2. Phys. Chem. N.F., 1967, 52, 104. 6 J. Lipkowski and Z. Galus, J. Electroanal. Chem., 1975,58, 51. 7 E. Brillas, J. A. Garrido, R. M. Rodriguez and J. DomCnech, Chem. Scr., in press. 8 Z. Galus, Fundamentals of Electrochemical Analysis (Ellis Harwood, Chichester, 1976), chap. 7 and 17. 9 K. Tokuda, N. Enomoto, H. Matsuda and N. Koizumi, J. Electroanal. Chem., 1983, 159, 23. 10 2. Galus, C. R. C. Crit. Rev. Anal. Chem., 1975, 6, 359. 1 1 R. S. Nicholson, Anal. Chem., 1965, 37, 1351. 12 H. Matsuda and Y. Ayabe, 2. Elektrochem., 1955, 59,494. 13 A. Arevalo, A. Vivo and E. Tejera, An. Quim., 1974, 70, 318. 14 F. Accascina, S. Petrucci and S . Schiavo, Sci. Tee., 1959, 3, 242. 15 J. Kielland, J. Am. Chem. Soc., 1937, 59, 1675. 16 S. Trasatti, J. Electroanal. Chem., 1970, 28, 257. 17 B. V. Apparao and M. V. Ramanamurti, J. Chem. Soc., Faraday Trans. I, 1979,75,2576. Paper 5/1167; Received 10th July, 1985
ISSN:0300-9599
DOI:10.1039/F19868201781
出版商:RSC
年代:1986
数据来源: RSC
|
16. |
Application of the competitive preferential solvation theory to ion–molecule interactions |
|
Journal of the Chemical Society, Faraday Transactions 1: Physical Chemistry in Condensed Phases,
Volume 82,
Issue 6,
1986,
Page 1789-1793
Bhukandas Parbhoo,
Preview
|
PDF (338KB)
|
|
摘要:
J. Chem. SOC., Furuduy Trans. I, 1986,82, 1789-1793 Application of the Competitive Preferential Solvation Theory to Ion-Molecule Interactions Bhukandas Parbhoo and Otto B.Nagy* Laboratoire de Chirnie Organique Physique (CHOP), Unitk CICO, Universitk Catholique de Louvain, Bdtirnent Lavoisier, 1 Place L. Pasteur, 1348 Louvain-la-Neuve, Belgium The n.m.r. spectroscopic behaviour of first group monovalent ions, 'Li+, 23Na+, *'Rb+, 133Cs+ and of 19F- has been re-examined in the light of the competitive preferential solvation theory. It appears that this theory can successfully account for the relatively strong ion-molecule interactions in the whole concentration range. It is shown that classical treatment will yield the same results provided it considers the solvent explicitly. The competitive preferential solvation (COPS) theory can be successfully Epplied to the spectroscopic study (u.v., n.m.r., relaxation) of weak molecular interactions1 and to reaction rate measurements in mixed solvents., This theory has been put forward3 to generalize the solvating-complexing interactions in the whole concentration range (&loo%) with the hope that it would also be able to account for certain spectroscopic anomalies without ad hoc arguments.' Since this goal has been successfully achieved by the theory for weakly interacting neutral organic molecules the question arose whether COPS theory would also be valid in the case of the stronger interactions of charged species as well.For this purpose the n.m.r. spectroscopic behaviour of 7Li+, 23Na+, 87Rb+, 133Cs+ and 19F-(A) has been studied in H202(Z)-H20(S) solvent mixtures. All the chemical-shift data are taken from the work of Covington et aL4 and their paper should be consulted for experimental details.Our choice was guided by the fact that Covington's work represents one of the most sound thermodynamic analyses ever published on the behaviour of ions in mixed protic solvent^.^-^ The examples chosen are interesting not only for the fact that they imply strong ion-molecule interactions, but also because of the likely intervention of a stoichiometric problem through the solvation number. All the data used are collected in table 1.6 represents the chemical shift of the various ions in mixed solvents while designates the shift measured in pure water.Similarly, 6A(Z) represents the chemical shift in pure H,02. Results and Discussion The variation of 6-6,,,, as a function of H202 concentration (Cz) is monotonic in the whole concentration range. For 7Li+ it shows an upward bending curve, for 23Na+ and 19F- it gives straight lines with positive slope and for 87Rb+ and 133Cs+ we have downward-bending curves with saturation tenden~y.~ In order to account for these observations Covington et al. supposed that initially each ion is solvated by a definite number of water molecules (n), which are stripped off stepwise by incoming H,02 molecules in a series of successive equilibria (equilibrium constants Kl, K2, . . . , Kn). Introducing reasonable assumptions, these authors obtain the following rather complicated relationship between chemical shift and solvent composition when 17891790 Competitive Preferential Solvat ion Theory Table 1.Chemical shifts of various ions (A) in H,O,(Z)-H,O(S) solvent mixtures as a function of H,O, concentration, relative to H,O solution (6 - 6A(s))a HzO, 6-6A(S) @pm) CZ XZ /mol dm-3 7Li+ 19F- 0.1 0.2 0.3 0.4 0.5 0.6 0.7 0.76 5.386 10.458 15.243 19.765 24.044 28.100 3 1.950 34.168 0.022 0.05 1 0.084 0.123 0.165 0.215 0.266 0.3 1 1.43 2.76 4.21 5.62 7.10 8.52 9.96 10.79 10.15 18.46 25.77 32.15 37.23 41.85 46.15 48.77 19.58 34.03 45.26 53.68 60.70 66.67 71.58 74.03 5.35 9.6 12.75 15.7 18.0 20.0 21.5 22.3 a Ref. (4). Table 2. Comparison of classical and COPS theory treatment of ionic solvation in mixed solvents exptla classicala COPS ~~~~ 'Li+ 0.5 0.50 0.488 0.58 0.383 23Na+ 14.34 14.20 0.975 14.45 0.952 87Rb+ 56.0 56.18 1.96 56.28 1.971 82.63 87.75 2.813 82.76 2.767 25.62 25.62 2.384 25.50 2.364 133Cs+ 19F- a Ref.(4). where Y = az/as, ai being the activity of species i. It can be seen that even for a very limited solvation number (n = 4) this expression is intractable. However, if no cooperativity of any kind is present the various equilibrium constants Ki are related entirely by statistical requirements and eqn (1) is simplified to : s-6A(S) - KY -- 6A(Z) -BA(S) I + KY where K = (Kl K , K3 K$, or more generally K = (Kl K,, . . . , K , ) l k The linear form of (3) 1 this equation : was used to obtain the various constants K (table 2) by a double reciprocal plot. Let us see now how COPS theory handles this same problem.It should be recalled that COPS theory is based on five simple postulates implying microscopic partitioning,l which allow the simultaneous consideration of both com- plexation and solvation. Applied to n.m.r. spectroscopy COPS theory leads to the following Scatchard equation : (l +a - - 1 - BA(S) 6A(Z) - BA(S)B. Parbhoo and 0. B.Nagy 1791 64A(S) I 1 1 1 5 10 15 (e) 20 10 20 30 40 I I I 25 50 75 6-6,,,, upper scale (a, c) lower scale (b, d ) 100 Fig. 1. Linearization of the variation of chemical shift with solvent composition: (a) 7Li+ (the scales are multiplied by lo2); (b) 23Na+ (the scales are multiplied by 10); (c) 87Rb+; ( d ) 133Cs+; (e) leF-. The affinity constant ratio KA(Z)/KA(S) accounts for the overall solute (Ahsolvent (S, Z) interaction in a competitive manner, K ~ ~ ) designating the interaction of species i with solvent component j.vs and vz are the molar volumes of H20 and H202, respectively, while Cz is the molar concentration of H,02. The application of eqn (4) to the data of table 1 leads to excellent linear plots (fig. 1) in the whole concentration range. It appears also that the positive slope obtained for 23Na+ (and for ‘Li+) cannot be interpreted by the classical form of the Scatchard equation : where SA and SAz designate the chemical shift of free and fully complexed A, respectively, while K is the overall association constant for species AZ. It should be remembered that by definition dA = SA(S). Eqn (4) shows that COPS theory allows for zero and positive slopes together with the ‘normal’ negative s1ope.l Using the molar volume values vs = 0.01801 dm3 mol-1 and vz = 0.02358 dm3 mol-l this equation leads to different affinity constant ratios KA(Z)/ICA(S) (table 2).It can be seen that these latter values are in excellent agreemen$ with the K values found by Covington et aL4 The relatively large deviation observed for1792 Competitive Preferential Solvat ion Theory 7Li+ is simply due to the fact that, in this particular case, the chemical shifts are not independent of 'Li+ concentration, i.e. they have not been extrapolated to zero 7Lis concentration . The agreement between classical and COPS results seems rather surprising since the two models are quite different. The reason for this can be found in the following way. According to COPS theory the chemical shift at any solvent composition in the fast-exchange limit is given by [eqn (1 9) of ref.(l)] : Rearranging we obtain the expression : If the ideality of solvent mixtures is assumed in the traditional sense (i.e. ai z Ci) it can be seen that eqn (7) is identical to eqn (2) with K = ICA(Z)/ICA(S). It is therefore not surprising that the Scatchard or any other linear form of these equations will yield the same results. On the other hand, this formal identity has a far-reaching physico-chemical consequence. It stems from the fact that, in COPS theory, the affinity constant ratio KA(Z)/~A(S) represents the partitioning of A between Z and S on the mole-fraction scale provided the solutions are diluted with respect to A.1-3 Therefore we can writel where Ciu, represents the concentration (mol dm-3) of solute i in interaction with solvent componentj.It should also be remembered that owing to some unavoidable assumptions, the rather complicated multi-equilibrium case of Covington et al. reduces to a simple overall equilibrium : (9) A( S) + Z =i= A( Z) + S with equilibrium constant equal to : K=------ CA(Z) cS CAW c z ' Comparing eqn (8) and (10) one can see that the classical equilibrium constant represents in fact a microscopic partitioning coefficient when the solvent S (the traditionally neglected ' inert ' component) is explicitly included in the corresponding thermodynamic treatment. In those cases classical and COPS theories will lead to the same results. Note that the Scatchard form of eqn (7): is very useful since it does not require the explicit knowledge of molar volumes, which are not easily accessible for solid Z components.Furthermore a perusal of data published in ref. (1) shows that this equation always has the correct negative slope even in cases where eqn (4) shows anorna1ies.l The above considerations would suggest that classical and COPS treatments can be used with equal efficiencies in analysing molecular interactions. However, as shown by the fundamental work of Covington et U Z . , ~ - ~ classical treatment has to go through many complicated steps before arriving at the simple situation, which is reached directly byB. Parbhoo and 0. B.Nagy I793 COPS theory based on microscopical partitioning in homogeneous media. For this practical convenience and also for its internal consistency and general ap~licability,l-~ COPS theory should be preferred in the treatment of molecular interactions. In conclusion, we can say that even in the case of relatively strong ion-molecule interactions COPS theory leads to medium independent over-all affinity constant ratios which are also insensitive to the number and the stoichiometry of the interactions present in the solvent mixture.Classical treatment will also lead to the same results provided it takes the solvent explicitly into account. We thank Prof. A. K. Covington (University of Newcastle) for helpful discussions. References 1 0. B.Nagy, Mukana wa Muanda and J. B.Nagy, J. Chem. SOC., Faraday Trans. I , 1978,74,2210. 2 0. B.Nagy, Mukana wa Muanda and J. B.Nagy, J. Phys. Chem., 1979,83, 1961. 3 Mukana wa Muanda, J. B.Nagy and 0. B.Nagy, Tetrahedron Lett., 1974,38, 3421. 4 A. K. Covington, T. H. Lilley, K. E. Newman and G. A. Porthouse, J. Chem. SOC., Faraday Trans. 1, 5 A. K. Covington, K. E. Newman and T. H. Lilley, J . Chem. SOC., Faraday Trans. I , 1973,69,973. 6 A. K. Covington, 1. R. Lantzke and J. M. Thain, J. Chem. SOC., Faraday Trans. I , 1974,70, 1869. 7 A. K. Covington and J. M. Thain, J. Chem. SOC., Faraday Trans. I , 1974,70, 1879. 8 A. D. Covington and A. K. Covington, J. Chem. SOC., Faraday Trans. 1, 1975,71, 831. 9 A. K. Covington, personal communication. 1973, 69, 963. Paper 511176; Received 11th July, 1985
ISSN:0300-9599
DOI:10.1039/F19868201789
出版商:RSC
年代:1986
数据来源: RSC
|
17. |
Paramagnetic metal and oxygen species observed with Rh/γ-Al2O3and Rh/ZrO2. Dependence on the decarbonylation temperature of [Rh4(CO)12] on alumina and zirconia supports |
|
Journal of the Chemical Society, Faraday Transactions 1: Physical Chemistry in Condensed Phases,
Volume 82,
Issue 6,
1986,
Page 1795-1804
Antonella Gervasini,
Preview
|
PDF (797KB)
|
|
摘要:
J. Chem. Soc., Faraday Trans. I , 1986,82, 1795-1804 Paramagnetic Metal and Oxygen Species observed with Rh/y-Al,O, and Rh/ZrO, Dependence on the Decarbonylation Temperature of [Rh,(CO),,] on Alumina and Zirconia Supports Antonella Gervasini, Franca Morazzoni" and Donatella Strum010 Dipartimento di Chimica Inorganica e Metallorganica, Universita di Milano, Via Venezian 21, 20133 Milano, Italy Francesco Pinna and Giorgio Strukul Dipartimento di Chimica, Universita di Venezia, Venezia, Italy Lucian0 Zanderighi Dipartimento di Chimica Fisica ed Elettrochimica, Universita di Milano, Milano, Italy Paramagnetic metal and oxygen species observed when alumina- (or zir- conia-supported rhodium, Rh/y-Al,O,(ZrO,), is obtained by the decar- bonylation of [Rh,(CO),,]-Al,O,(ZrO,) are discussed as a function of the decarbonylation temperature (250-600 "C) and the method of decarbonyl- ation (pyrolysis in vacuo or reduction in an H, stream). On vacuum- pyrolysed samples, Rhp(T,/y-Al,03, RhII formal centres are stable only after decarbonylation at 250 "C; contact with 0, produces [Rh~ll-O,]'- (n = 1,2) and AP+-O;, with an increasing amount of AP+-O; as the decarbonyl- ation temperature increases.Rh,(,,/ZrO, does not show paramagnetic species before contact with 0,: the treatment with 0, stabilizes both [RhI,II-O,]'- and Z1"'+-0; centres at a decarbonylation temperature of 250 "C, while essentially Zr4+-O; alone is formed at higher temperature. On H, reduced samples, RhH(,,/y-Al,03, formal Rho centres formed in the decarbonylation temperature range 250-500 "C, while no paramagnetic species were observed on RhH(,,/Zr0,.Contact with 0, shows behaviour of Rh,,T,/y-A1,03(Zr0,) very similar to that of Rhp(T,/y-A1,03(Zr0,). The stability of the Rh-0, bond is a function of the metal positive charge and it is related to the decarbonylation temperature. The strength of the interaction between 0; and the support acidic centres, A13+ and Zr*+, is always high in y-Al,O,-supported samples, and dramatically increases in ZrO,-supported samples on increasing the decarbonylation temperature. Different pathways of electron transfer can explain the observed behaviour of Rh/ZrO,. An understanding of working catalysts is a major goal in the physicochemical charac- terization of dispersed metal systems. A direct investigation of the electronic properties of the metals in these systems is needed to ensure the possibility of either building new catalysts or modifying their properties as a function of the reaction one wishes to control.We recently reported a spectromagnetic characterization of the metal and oxygen paramagnetic species observed on [ Rh,(CO),,] pyrolysed on y-alumina and zirconia supports.la The aim of the work was to draw correlations between the electronic structure of the supported metal systems and their behaviour as catalysts in the reaction between CO and H,. Previous investigations left several problems unresolved. (i) Could a pyrolysis tem- perature higher than 250 "C affect the nuclearity of the paramagnetic metal particles? If so, is the metal-support electron transfer, which produces the paramagnetic centres, 17951796 Paramagnetic Species on Rh/ y-A1203, Rh/ZrO, modified? (ii) Are the paramagnetic species on the surface dependent on the thermal decarbonylation carried out in vacuo or in a flowing H, stream? (iii) Do the paramagnetic species observed by interaction with a n*-acceptor molecule (O,, CO) depend on the temperature and on the manner of thermal decarbonylation? The purpose of the present paper is to answer these questions, and so to compare the metal-metal and metal-support interactions in Rh/y-Al,O,(ZrO,) systems.Those paramagnetic mononuclear Rh complexes with the metal in a low oxidation state and those Rh carbonyl clusters which are paramagnetic will be taken as models in order to identify the paramagnetic species observed on the catalysts; in addition, the chemistry of these model systems will be referred to in explaining the modification of the supported metal arising under the various experimental conditions reported.Experimental The dispersion of [Rh,(CO),,] on y-Al,O, and ZrO, was performed as in ref. (1 6). Rh/y-Al,O,(ZrO,) was obtained by pyrolysis of [Rh,(CO),,]y-Al,O,(ZrO,) in vacuo ( Pa) or by thermal reduction in an H, stream (1 66 cm3 min-l) at a given temperature. The sample treatment was carried out in appropriate apparatus, where pyrolysed or reduced samples could be transferred to an e.s.r. tube (internal diameter 3 mm) without contact with air. In the same apparatus contact with 0, and CO (high-purity gas) was made at room temperature, under a pressure of 10 Pa.Treatments analogous to those mentioned above were performed on alumina and zirconia supports to compare the signals observed on the metal systems with those observed in the absence of metal. E.s.r. spectra were recorded on a Varian E-109 spectrometer, equipped with automatic Varian temperature control. The computational methods for the spin concentration are those described in ref. 1 (a). Results and Discussion Vacuum-pyrolysed Samples [RhpCr,/y-A1,03(Zr0,)] Rhp(T)/y-Al,O, was obtained by pyrolysis of [Rh,(CO),,]y-Al,03 at 250, 350, 400, 500 and 600 "C (lo-, Pa). The behaviour of Rh,,,,,,/y-Al,O, was described in our previous paper. As the temperature increases to 350 "C the resonance lines due to formal RhII centres disappear. Two different interpretations can be given to this.The first is that the higher temperature promotes coupling between the Rh paramagnetic species to form diamagnetic aggregates. Alternatively, the higher temperature induces the growth of paramagnetic metal particles; thus the interaction between the Rh centres and the y-Al,O, surface decreases and the electron transfer from Rh to y-Al,O,, suggested in Rhp(250)/y- Al,O,, is not efficient enough to stabilize oxidized RhII centres. We cannot discriminate between the two hypotheses at this stage; in fact the two effects could coexist. A third interpretation of the absence of RhII resonance lines involves the oxidation of Rhl* surface centres to RhIII, promoted by the hydroxy surface groups of y-A1203; however, the experiments with gaseous 0, allow one to exclude this possibility (see later).Contact with 0, carried out on Rh/y-Al,O, pyrolysed at different temperatures gives the resonance lines in fig. 1. Based on the same arguments discussed in previous papers,' a, 2 v the complex features of the spectrum of Rhp(350)/y-A1203 can be attributed to the presence of three almost distinguishable 0; species. Two can be assigned to 0; bonded to the Rh centres and are distinguishable by their g,, g,, g,, gi, gi and gi values (table 1). The assignment of the third species to 0; interacting with A13+ is supported by the results from samples pyrolysed at a temperature >350 "C. In fact, above 400 "C the resonance lines due to the Rh-interacting 0; species disappear, and the well known4 resonance lines of 0; interacting with A13+ become more and more dominant.A .Gervasini et al. 1 91 1797 - 4 0 G J 350 400 500 500 600 DPPH g3a 2 50 4 00 (a) Fig. 1. (a) E.s.r. spectra at - 150 "C of Rhp(T,/y-Al,O,, recorded under an 0, (10 Pa) atmosphere. Doubly starred line refers to a spectrum under an 0, (10 Pa)-CO (1 0 Pa) atmosphere. The pyrolysis temperature (in "C) is indicated. (b) E.s.r. spectra at - 150 "C of RhH(T,/y-Al,03, recorded under an 0, (10 Pa) atmosphere. The resonances indicated by gll and g , refer to the A13+-O; species. To define the two different Rh interacting 0; species better, we have considered, in detail, their g values. Both species have g values characteristic of a [RhlI1-O,]'- and the A species has the same g, and g , values observedla for the [Rhlll-O,]'- species on Rhpcz5,,,/y-Al,0,; as for the g , component, it is much more obvious than it was in the spectrum of Rh,,,5,,/A1,0, so that the value reportedla for this species must be corrected: g , = 1.95.The g value components of the B species differ only slightly from the A species,Table 1. E.s.r. data for paramagnetic metal and oxygen speciesa 2 sample 0, pressure /Pa g1 g2 g3 10 2.09 ~ 2.03 1.95 ([Rh(en)2C11201-02))3+ [Rh(en),Cl(O,)]+ (trans form) RhH(350)/Y-A1203 2.06 2.034 2.097 2.076 0 2.17 2.09 10 2.06 2.0324 10 2.03 10 2.09 2.03 2.02 2.00 2.030 2.022 not visible 2.03 2.02 2.00 2.00 not visible 2.00 1.97 2.00 1.988 1.999 1.95 not visible 2.00 2 6 : stable in vacuo 2. monomeric Rh species 3 3 no. of paramagnetic paramagnetic centres species /spin g-l notes $.5 dimeric Rh species 0 x v u stable in vacuo w stable in vacuo \ RhI'I-0, (A) Ref. 2 Ref. 2 Rho 0.3 x 1017 hydrido species k dimeric Rh species "0 ~ 1 3 + - 0 , J w RhI'I-0; (A) monomeric Rh species ' RhIII-0, (B)i 0.55 x l0ls ( v zr4+-0, 1017 stable in vacuo \ zr"'+-o, treatment 0 RhIII-0, (A) ] 0.6 removed by vacuum IY Pa, 24 h) a The reported parameters refer to those samples where the paramagnetic species are easily observable. For the evolution of the spectrum for different decarbonylation temperatures, see text.A . Gervasini et al. 1799 probably owing to the different electronic properties of the superoxo-rhodium species. Although the paramagnetic species observed on Rh/y-Al,O, could be formalized as [RhIII-O,]'-, the unpaired electron is partly lying on the Rh centre, owing to the overlap between the n*O, orbitals and the d, Rh orbitals.The theoretical energy-level scheme proposed by Raynor et aL2 for the paramagnetic dioxygen Rh complexes led us to expect a difference in energy between the monomeric and the dimeric species. If the paramagnetic superoxide species is dimeric and if 0; interacts with two RhIII centres, the energy difference between the molecular orbital containing the unpaired electron (the HOMO made by overlap of d,,o,,, Rh orbitals with the n*O, orbitals) and the low-energy unoccupied molecular orbital (LUMO obtained from the overlap of dyz or ,. Rh orbitals with z; or z; 0, orbitals) is lower in the dimeric species than in the monomeric one; thus a larger deviation from the free-electron g value should be found in the dimeric superoxo species than is found in the monomeric case.The A species was thus attributed to dimeric Rh superoxo species, the B to a monomeric one. For comparison the g values of the Rh complexes to which we refer are reported in table 1. The overlap between the A13+-O; lines and the lines of the [Rh,-O,]'- (n = 1,2) species is the cause of the anomalous relative intensity of the gI1 component against the gl of A13+-O;, in samples pyrolysed at 500 and 600 "C (fig. 1). The true spectrum of A13+-O; was that obtained after contact with CO and then 0, (see doubly-starred line in fig. 1). In practice 0; could not be removed from Rh or from A13+ by vacuum treatment. y-Al,O, pyrolysed under the same conditions as the Rh/y-Al,O, samples and analogously contacted with 0, did not give any resonance lines other than those of Fe3+.The results of the interaction of Rh/y-Al,O, with 0, suggest the following interpretation. The paramagnetic oxygen adducts involving Rh have a low metal nuclearity, probably lower than the diamagnetic aggregates from which they are derived; it seems reasonable that the Rh-0, bonding interaction successfully competes with that of Rh-Rh bonding. As the 0; fixed on formal RhJII centres is present on Rh,/y-Al,O,, formal RhII centres should be present prior to any contact with oxygen; in the absence of O,, however, they have no e.s.r. signal owing to the large spin coupling between the paramagnetic species. As the value of the RhIII-0;- spin concentration in Rhp(350)/y-A1203 differs little from the value of the RhII spin concentration in Rhp(,,,,/y-Al,O,, the absence of RhlI signals in Rhp(250)/y-A1203 cannot be attributed to the oxidation of RhII to RhIII by interaction with the hydroxy groups of y-Al,O,.By increasing the pyrolysis temperature, 0; spills from RhIII to A13+ centres. It is important to point out that the presence of A13+-O; centres has never been observed on Rhp(250)/y-A1203. It was suggested in ref. (1 a) that the strength of the Rh-0, bond depends mainly on the a-acceptor ability of the metal. It seems probable that at a pyrolysis temperature >250 "C the acidic character of Rh, which is related to the a-acceptor ability of the metal, decreases with respect to A13+; consequently the 0; shifts to the A13+ centres of the support.The decrease of the Rh acidic character is probably due to the increase in pyrolysis temperature, which results in an increase of the dimensions of the paramagnetic metal particle (an increase in the metal-metal interaction) and in the consequential reduction in the electron transfer from Rh to Al,O, (a decrease of metal-support interaction). Interaction of Rh/y-Al,O, pyrolysed at temperatures >250 "C with CO does not produce paramagnetic metal species. We now turn to Rh,,,,/ZrO;. Like Rhp(250)/Zr02, the samples obtained by pyrolysis in vacuo at 350, 400 and 600 "C continue to give no new e.s.r. lines, apart from those mentioned for the ZrO, support.la Contact with 0, gives the resonance lines (fig. 2) of the Zr4+-O; species., 0; fixed on Zr4+ is stable in vucuo (lo-, Pa).The absence of Rh-0, adducts on Rh/ZrO, pyrolysed at temperatures >250 "C is exactly what we expect, the ability of Rh/ZrO, to fix 0, on Rh being lower than that of Rhp/y-Al,O,. We have already observedla that the fixation of 0; on Rh is a reversible process in Rhp(250)/Zr0, and that a vacuum destabilizes the Rh-0, bond. What is singular is the 60 FAR 11800 Paramagnetic Species on Rh/ y-Al,03, Rh/ZrO, 2 50 350 400 600 600 (b) Fig. 2. (a) E.s.r. spectra Of Rh,(,)/ZrO,, recorded under an 0, (10 Pa) atmosphere. (b) E.s.r. spectra of Rh,,,)/ZrO,, recorded under an 0, (10 Pa) atmosphere. *ZrO, radical signal. The resonances indicated by gll and g, refer to the Zr4+-O; species. high stability of 0; fixed on Zr4+ centres in the samples pyrolysed at temperatures >250 "C.For Rhp(250)/Zr02 we observed that 0; is removed both from Rh and from Zr4+ by vacuum treatment. Another peculiarity of these systems is that while ZrO, pyrolysed at 250 "C did not give any other resonance lines besides those present on the support before the thermal treatment, the support pyrolysed at 400 and 600 "C gives new resonance lines which are assignable to the Zr4+-0, species. These resonances are stable in vacuo (24 h, Pa). A significant amount of this adduct was formed even when ZrO, was not in contact with 0,. Thus it seems that above 250 "C neither Rh nor contact with 0, was needed to reduce the oxygen and fix it as 0;. It was recently suggested6 that ZrO,, by vacuum and thermal treatment, partially assumes tetragonal and monocline crystalline forms, where the electrons trapped in the oxygen vacancies or those lying on Zr3+ ions are .able to be transferred to the bulk oxygen and to generate paramagnetic oxygen species.This could be our case in ZrO, treated in vacuo at 400 and 600 "C. The amount of 0; fixed on Zr4+ increases when we contact the sample with O,, confirming that electron transfer from the support to the 0, is also efficient in the absence of Rh. Whatever the pyrolysis temperature, if Rh is present, contact with 0, is needed to obtain 0, fixed on Zr4+ centres. At present we have noA . Gervasini et al. 1801 hypotheses about the role played by rhodium in hindering electron transfer to the bulk oxygen; the goals of the present work do not include an investigation of this kind.What is important for our purposes is to point out that more than one electron-transfer pathway to 0, take place above 250 "C. While in Rhp(250)/Zr02 we suggested that 0, is reduced by Rh and then partly shifted as 0; to Zr4+, in samples pyrolysed at higher temperatures electron transfer to 0, can surely take place via ZrO,, and the electrons could be either those of Rh or those of the support. Although we have no means of deciding how many electrons come from Rh and how many come from the support centres, we can conclude that the transfer mechanism in samples pyrolysed at 400 and 600 "C is different from that in samples pyrolysed at 250 "C and that the electrons go through the support. This could be related to the different stabilities of 0; on Rh,/ZrO, pyrolysed at different temperatures.In Rh/ZrO, pyrolysed at 400 and 600 "C, 0; is a 'bulk' species, more similar to that originating in a ZrO, support6 than to that observed by 0, contact in Rhp,R(250)/Zr02. H,-reduced Samples (RhH(T)/y-A1,O,(ZrO,)] In the case of RhH(T)/y-Al,O, samples were obtained by the reduction of [Rh,(CO),,]y- A1,0, in an H, stream (166 cm3 min-l) at 250,400, 500 and 600 "C for 2 h. RhHfzso)/y- Al,O, shows strong resonance lines in an argon atmosphere, both at room temperature and at - 150 "C (fig. 3); as these signals were not observed on y-Al,O, reduced under the same conditions as the Rh samples, their presence can be assigned to metal-derived resonances. The shape of the resonances is characteristic of an axial symmetry for the g tensor; however, only the perpendicular component can be distinguished clearly.No hyperfine interaction with Rh nuclei is visible. The signal is slightly modified in Rh,,,,o~/y-Al,O, and RhH(500)/y-A1203 : another low-intensity paramagnetic species become visible. It is probable that the parallel component of the main paramagnetic species is superimposed on the lines of the subsidiary species. An increase of the reduction temperature leads also to an increase of a narrow signal (starred in fig. 3) at g = 2.00; however the intensity of this signal is so low that we decided not to investigate its nature further. The main paramagnetic species disappears on samples reduced at 600 "C. The comparison of the signal of the main species in fig.3 with that observed on Rhp/y-Al,O, suggests that the paramagnetic species obtained by the two different pathways of decarbonylation are completely different : (i) the resonance lines observed in RhH/y-Al,O, are also visible at room temperature, while those observed in Rhp/y-Al,O, become evident only at low temperature; (ii) the trend of g values deduced from the shape of the lines in Rh,/y-Al,O,(g, > gll) is quite different from that of Rh,/y-Al,O,(gll > 8 3 . It can be concluded that the Rh paramagnetic centres observed in RhH/y-Al,O, have a lower spin-orbit coupling contribution and a different ground-state configuration from the RhII centres observed in Rhp/y-Al,O,. The electronic ground-state configuration proposed for the RhII centres in Rhp/y-Al,O3l a has the unpaired electron in a molecular orbital comprised mainly of dz2--y2 atomic orbitals; in Rh,/y-Al,O, we suggest that the molecular orbital is comprised mainly of the 4 2 metal orbital.' The properties of the paramagnetic species observed in RhH/y-Al,O, suggest that this species is different from that of Rh,/y-Al,O, because of the Rh oxidation state and/or because of the symmetry of the ligand field around Rh.The ligand field around Rh is unknown to us, because in both kinds of decarbonylation we do not know the molecular structure of the paramagnetic unities obtained ; however, the decrease in the spin-orbit coupling contribution should correspond to a decrease in the positive charge of Rh, and this prompts us to suggest a different Rh oxidation state in Rh,/y-Al,O,.The most probable oxidation state, Rho d9, should be stable in a H,-reducing atmosphere and should have resonance lines more obvious at room temperature than those of RhII. The presence of e.s.r.-active Rho centres in Rh,/y-Al,O, leads us to ask why the Rho 60-21802 Paramagnetic Species on Rh/y-Al,O,, Rh/ZrO, D PPH Fig. 3. E.s.r. spectra at - 150 "C of RhH(T,/y-Al,03, recorded under an argon (26.6 kPa) atmosphere. Besides the spectrum of RhH,,,,,/y-Al,03 (b), the spectrum of Rhp(250)/y-A1203 is reported (a). centres are not visible on Rh,/y-Al,O,. In our opinion Rho centres are present; however, they are not sufficiently magnetically dilute to be e.s.r.-active. In other words, in the former case H, induces a magnetic dilution of the Rho.The only way to obtain magnetically diluted Rho centres by H, reduction is to form rhodium hydride paramagnetic aggregates. Rhodium hydride clusters, even if not paramagnetic, are known to occur in the coordination chemistry.* No suggestion can be given about the nuclearity of these aggregates. The resonance lines due to Rho centres disappear after contacting Rh,/y-Al,O, with O,, and the formation of the paramagnetic derivatives where 0; is fixed either to RhIII or to A13+ is observed (fig. 1). Although the trend of distribution of 0; between RhIII and A13+ shows that on increasing the reduction temperature, 0, shifts towards A13+ as in the pyrolysed samples, the relative distribution is not the same at the same temperature. The reduced samples show more 0, remaining bonded with Rh than do the pyrolysed ones;A .Gervasini et al. 1803 it seems that in the H,-reduced samples, A13+ is less competitive with RhIII for 0; fixation. This could be related to a decrease in the average positive charge of A13+ surface centres following the surface reduction of Al,O, by H, treatment.g RhH/y-Al,03 showed the presence of mainly the dimeric [Rh,(III)-O,]*- species. When Rh,/y-Al,O, is contacted with CO, the Rho resonance lines disappear; however, no new paramagnetic species were observed. This can be considered an additional proof that the paramagnetic Rh centres seen on RhH(,,,,/y-Al,03 have a different oxidation state from those observed on Rhp(,,,,/y-Al,03. It could also mean that the Rh-CO bond strength is different for Rh centres having different electronic charges.RhH/ZrO, was obtained by the reduction of [Rh,(CO),,]ZrO, in an H, stream at 250, 400 and 600 "C for 2 h. No paramagnetic species were observed after reduction at 250 and 400 "C other than those of the radical species present on the support. On reduction at 600 "C, strong new resonance lines appear at g values lower than the free-electron value; the lines show axial symmetry (gl = 1.95, gI1 = 1.90) and are indicative of a dl electronic configuration. We assign the signal to surface Zr3+ centres following the partial reduction of the ZrO, surface by the high-temperature H,. After Rh/ZrO, was reduced at 250 "C, it was contacted with O,, and the resonance lines (fig. 2) due to the 0; fixed on RhIII and/or on Zr4+ were observed.Vacuum treatment (24 h Pa) removes 0; from either RhIII or Zr4+ centres. Temperature changes induce the same 0; distribution trend as in Rh,/y-Al,O,; indeed the amount of 0, transferred to Zr4+ at a given temperature is lower than in the case of Rh,/ZrO,. This could be related to a decrease in the average positive charge of Zr4+ surface centres following surface reduction by H,. ZrO, treated in an H, stream under the same conditions as the Rh samples shows the same behaviour as the pyrolysed support. Concluding Remarks Pyrolysis at temperatures >250 "C affects the nuclearity of the paramagnetic metal aggregates and the strength of the metal-support interaction. In the case of Rhp/Al,03 this conclusion derives from the disappearance of the RhI1 paramagnetic entities and from the progressive shift of 0; from the RhIII centres to the more acidic A13+ centres; as the pyrolysis temperatures increase the acidic character of Rh becomes too low to fix 0;.In the case of Rh,/ZrO,, the fixation of 0; only on the Zr4+ centres proves that the ability of Rh to fix 0; decreases as the pyrolysis temperatures increase. If decarbonylation is carried out on Rh/y-Al,O, in an H, stream, no formal RhII species are observed; however, formal Rho paramagnetic entities are stabilized over a wide range of temperatures. Although we did not directly observe any paramagnetic Rh species on Rh,/ZrO,, we suggest that the situation concerning the interaction of Rh with H, is little different from that observed on RhH/y-Al,03.Although spin concentration values of the paramagnetic species reported in the present paper are a small percentage of the total supported rhodium, and in principle the conclusions which can be drawn from the behaviour of the e.s.r.-active centres cannot be extended to the total dispersed metal, we think that a comparison between that suggested from e.s.r. results and that from the catalytic behaviour of the same samples should help to test the ability of e.s.r. spectroscopy to characterize dispersed metal systems. In this light we make the following observations. In agreement with the essentially a-character of the Rh-0, bond, proposed earlier, the interaction with 0, depends on the positive charge on Rh. If the Rh centres do not interact with and do not transfer electrons to the support, as happens in samples pyrolysed at high temperatures, then the 0; binds to the support centres.The irreversible bond observed between 0; and Zr4+ could have a role in the rationalization of methods for the preparation of the Rh/ZrO, catalysts, which are active1804 Paramagnetic Species on Rhly-Al,O,, Rh/ZrO, in the selective hydrogenation of carbon monoxide to give oxygenated products.lb In fact, if the strength of the interaction between 0, and Rh is a probe of the strength of the bonding interaction between Rh and a n*-acceptor molecule, and if the reversibility of this bond plays a role in the selectivity of the catalysts (as proposed previously), it could be suggested that a temperatur- not exceeding 250 "C should be maintained to avoid irreversible fixation of n*-acceptor molecules to the zirconia centres.The presence of the Rh-H interaction does not seem to affect the interaction between Rh and O,, provided that 0; is loosely bound to the support centres. The absence of RhII-CO paramagnetic species in the H, reduced samples proves that the selectivity towards oxygenated products is probably not related to the RhII centres. This conclusion disagrees with that proposed by Poels et aZ.l0 The role played by the oxidized metal centres in the catalytic mechanisms is still an open question. . This work was supported by grants from M.P.I. and C.N.R., Progetto Finalizzato Chimica Fine e Secondaria. References 1 (a) T. Beringhelli, A. Gervasini, F. Morazzoni, D. Strumolo, S. Martinengo and L. Zanderighi, J. Chem. SOC., Faraday Trans. I , 1984,80, 1479; (b) A. Ceriotti, S. Martinengo, L. Zanderighi, C. Tonelli, A. Iannibello and A. Girelli, J. Chem. SOC., Faraday Trans. I , 1984, 80, 1605. 2 J. B. Raynor, R. D. Gillard and J. D. Pedrosa de Jesus, J. Chem. SOC., Dalton Trans., 1982, 1165. 3 H. Caldararu, K. De Armond and K. Hanck, Znorg. Chem., 1978, 17,2030. 4 R. B. Clarkson and A. C. Cirillo Jr, J. Catal., 1974, 33, 392. 5 M. Setaka, S. Fukuzawa, Y. Kirino and T. Kwan, Chem. Pharm. Bull., 1968, 16, 1240. 6 M. J. Torralvo, M. A. Alario and J. Soria, J. Catal., 1984, 86, 473. 7 M. D. Sastry, K. Savitri and B. D. Joshi, J. Chem. Phys., 1980, 73, 5568. 8 (a) V. G. Albano, A. Ceriotti, P. Chini, S. Martinengo and W. M. Anker, J. Chem. SOC., Chem. Commun., 1975, 859; (b) V. G. Albano, G. Ciani, S. Martinengo and A. Sironi, J. Chem. SOC., Dalton Trans., 1979, 978; (c) J. L. Vidal, R. C. Schoening and J. M. Troup, Znorg. Chem., 1981, 20, 227; (d) G. Ciani, A. Sironi and S . Martinengo, J. Chem. Soc., Dalton Trans., 1981,519; (e) G. Ciani, A. Sironi and S. Martinengo, J. Organomet. Chem., 1980, 192, C 42. 9 S. W. Weller and A. A. Montagna, J. Catal., 1971, 21, 303. 10 E. K. Poels, P. J. Mangnus, J. V. Welzen and V. Ponec, Proc. 8th Zntl Congr. Catal., Berlin, 1984, vol. 11, p. 59. Paper 511200; Received 15th July, 1985
ISSN:0300-9599
DOI:10.1039/F19868201795
出版商:RSC
年代:1986
数据来源: RSC
|
18. |
Mixed adsorption of a non-ionic and an anionic surfactant at the carbon–aqueous solution interface |
|
Journal of the Chemical Society, Faraday Transactions 1: Physical Chemistry in Condensed Phases,
Volume 82,
Issue 6,
1986,
Page 1805-1811
Michael J. Hey,
Preview
|
PDF (562KB)
|
|
摘要:
J. Chem. SOC., Faraday Trans. I, 1986,82, 1805-1811 Mixed Adsorption of a Non-ionic and an Anionic Surfactant at the Carbon-Aqueous Solution Interface Michael J. Hey and John W. Mactaggart? Department of Chemistry, The University, Nottingham NG7 2RD Colin H. Rochester* Department of Chemistry, The University, Dundee DDl 4HN Adsorption from mixed aqueous solutions of sodium dodecyl sulphate and 0-n-octyltetraethylene glycol on to graphitised carbon has been determined by solution depletion. A decrease in total adsorption was found in comparison with the amount expected from isotherms for separately adsorbed surfactants. An extension of the' Rubingh theory for mixed-micelle formation is proposed to allow deviations from non-ideal mixing in the adsorbed layer to be specified. Application of the theory indicates positive deviations from ideal mixing for the system studied.In a previous investigation1 we studied the adsorption of sodium dodecyl sulphate (SDS) and 0-n-octyltetraethylene glycol (C,E,) separately from aqueous solution on to graphitised carbon. Here we report results obtained when both SDS and C,E, were adsorbed on to the same adsorbent from binary surfactant mixtures. The extent to which two surfactants adsorb from mixed solutions on to a given surface cannot generally be predicted from the individual isotherms because unlike molecular interactions in the adsorbed layer (e.g. SDS-C,E,) will usually differ from like molecular interactions (e.g. SDS-SDS, C,E,-C,E,). An analogous situation exists for mixed-micelle formation when critical micelle concentrations (c.m.c.) for binary surfactant solutions depend on intermolecular interactions within the micelle.Rubingh2 has shown in this case that the variation of c.m.c. over the composition range for a particular system may be described adequately by a theory involving a regular solution approximation for the surfactant activity coefficients in the mixed micelles. An interaction parameter reflecting the difference between like and unlike molecular interactions is determined by fitting the data to a theoretical expression. This approach has been successfully applied to the SDS-C,E, systems.3, The regular solution approximation has also been further extended to non-ideal mixing at the aqueous solution-gas interfa~e.~-~ In the present study we have tested the applicability of this idea to the co-adsorption of SDS and C,E, at the graphitised carbon-aqueous solution interface.Experimental The adsorption of SDS (B.D.H., specially purified for biochemical work) and C,E, (supplied by the Port Sunlight Laboratory of Unilever Research) from mixed aqueous solutions on to Sterling FTD-4 graphitised carbon (Cabot Corporation, Boston) was determined from the depletion of initial solution concentrations after equilibration at 25.00 0.05 "C. Full details of the materials have been published previous1y.l The experimental procedure was the same as beforel except that the tubes containing the t Present address : Unilever Research, Port Sunlight Laboratory, Quarry Road East, Bebington, Wirral, Merseyside L63 3JW.18051806 Mixed Adsorption at the Carbon-Aqueous Solution Interface dispersions were rotated in an end-over-end stirrer for at least 66 h instead of 16 h. After removal of the carbon by membrane filtration, 1 cm3 of the equilibrated solution was taken for SDS analysis by the method of Karush and Sonenberg.’ The remaining solution was used to determine the C,E, concentration by proton n.rn.r., In calculating surface concentrations a specific surface area of 90 m2 g-l (nitrogen, B.E.T.) was taken for the graphitised carbon. Results and Discussion Table 1 shows the amounts of SDS and C,E, adsorbed from solutions which were sufficiently dilute for micelles to be absent at equilibrium. For a given initial total surfactant concentration, the total number of moles of adsorbed surfactant was approximately constant, i.e.independent of the composition of the adsorbed layer. Initial bulk concentrations of (4, 5, 6 and 8) x 10-3mol dm-3 produced total surface concen- trations of [1.2*0.3, 1.52kO.25, 1.8kO.15 and 2.6k0.31 x mol m-2, respectively. Table 2 contains the data obtained for more concentrated solutions in which micelles were present at equilibrium. The total adsorbed surfactant for these solutions was (2.95 f0.5) x mol m-2. In addition to the observed bulk concentrations at equilib- rium, the table includes estimated concentrations for the non-micellised or monomer fraction. The sum of the SDS and C,E4 monomer concentrations at a particular overall composition was taken to be equal to the c.m.c. at that composition as determined in previous Monomer concentrations of SDS and C,E, were assumed to be in the same ratio as in the overall composition.In order to display the results given in tables 1 and 2, computer-generated plots of ‘adsorption surfaces’ are presented in fig. 1-3. These show how the amount of adsorbed surfactant depends on both the total surfactant concentration at equilibrium and the SDS/C,E, ratio. Adsorption surfaces are shown for SDS, C,E, and SDS-C,E, in fig. 1, 2 and 3, respectively. A general feature of the data is that the total amount of surfactant adsorbed is less than the adsorption which would result if the surfactants adsorbed in accordance with the single surfactant isotherms. If it is assumed that micelles are not surface active, the monomer concentrations shown in table 2 must be used in making the comparison for the more concentrated solutions.When this is done, similar behaviour is observed. A reduction in adsorption may be explained by the reduction in the number of surface sites available for occupation by molecules of one type, which is caused by the adsorption of molecules of the second type. Interactions, both between like and unlike types, however, must also be considered as they may modify the simple competition phenomenon. For example Kurzendorfer et aL9 have suggested that for SDS-alkylphenol polyethylene glycol adsorption on to active carbon the non-ionic molecules have the effect of reducing the electrical repulsions between the negatively charged groups of the anionic surfactant. In the present work we observe a preferential adsorption of SDS at low total surfactant concentrations which results from a drop in the amount of C,E, adsorbed, while the SDS adsorption remains at a level similar to that in the absence of the non-ionic surfactant.At higher concentrations, on the other hand, although both non-ionic and ionic adsorption is reduced there is preference for C,E, adsorption. These results could be regarded as arising first from the greater affinity of the SDS molecules with their longer alkyl chains for the surface at low coverages and secondly from the electrostatic shielding effect of C,E4 molecules at higher coverages, when electrostatic repulsions become more important. Clearly, adsorption from solutions containing more than one surfactant cannot be understood unless information regarding both molecular interactions and configurations of the adsorbates is available. Nevertheless, even without such information, some formal description of the properties of the adsorbates may be possible in thermodynamic terms,M.J. Hey, J . W. Mactaggart and C. H . Rochester 1807 Table 1. Adsorption from sub-micellar SDS-C,E, solutions at 298 K bulk concentration surface concentration mol dm-3 / 1 0-6 mol m-2 SDS C8E4 SDS C*E, initial total concentration = 4 x lop3 mol dm-3 0.99 0.3 1 1.45 0.05 0.98 0.63 1.23 0.09 0.83 0.85 1.09 0.19 0.78 1.27 0.90 0.18 0.65 1.42 0.75 0.32 0.54 1.57 0.59 0.46 0.29 1.74 0.39 0.70 0.07 1.22 0.18 1.32 1.66 0.1 1 1.74 0.04 1.62 0.30 1.60 0.11 1.57 0.64 1.35 0.20 1.16 1.24 1.02 0.42 0.95 1.57 0.86 0.52 0.74 1.74 0.70 0.70 0.53 1.53 0.54 1.09 0.48 2.08 0.32 1.07 0.30 2.29 0.1 1 1.23 0.01 1.98 0.05 1.57 2.12 0.37 1.80 0.13 2.05 0.66 1.51 0.30 1.91 1.07 1.26 0.40 1.21 1.43 0.98 0.87 1.01 1.63 0.76 1.09 0.78 1.89 0.56 1.28 0.60 2.27 0.33 1.39 0.33 2.40 0.15 1.66 3.50 0.48 2.08 0.18 2.30 1.17 1.86 0.69 1.93 1.41 1.61 1 .oo 1.61 1.58 1.34 1.35 1.27 1.75 1.07 1.70 1.19 2.26 0.68 1.86 0.75 2.7 1 0.48 2.06 0.43 2.92 0.2 1 2.39 initial total concentration = 5 x mol dm-3 initial total concentration = 6 x mol dm-3 initial total concentration = 8 x mol dm-3 e.g.by the introduction of surface activity coefficients which specify the deviations of adsorbate chemical potentials from values expected on the basis of an arbitrarily chosen standard state.We adopt this approach in what follows. For the dilute solutions used in this work the number of moles of a surfactant i adsorbed per unit area of adsorbent may be equated to rdN), which represents the surface excess concentration based on a comparison of bulk and interfacial regions containing the same total number of We shall simply write this quantity as Ti.1808 Mixed Adsorption at the Carbon-Aqueous Solution Interface Table 2. Adsorption from micellar SDS-C,E, solutions at 298 K bulk conc. monomer conc. surface conc. / lop3 mol dmP3 / 1 0-6 mol rnP2 / 1 0-3 mol dm-3 SDS SDS SDS C8E4 6.42 6.26 5.06 6.20 4.05 3.70 4.79 5.95 5.49 4.32 3.40 2.85 4.01 3.83 3.05 2.29 2.22 2.93 1.56 1.54 1.89 1.03 0.51 0.95 0.75 0.60 0.68 1.11 1.49 1.03 1.39 1.85 2.34 3.1 1 2.57 2.05 2.41 3.51 4.83 3.91 2.78 4.60 6.14 2.88 5.26 6.80 3.57 3.60 7.15 6.24 4.76 4.42 3.16 3.10 3.08 2.80 2.78 2.76 2.46 2.41 2.40 2.09 2.05 1.70 1.69 1.26 1.25 1.24 0.98 0.87 0.83 0.63 0.5 1 0.49 0.47 0.44 0.48 0.69 0.75 0.78 1.05 1.07 1.09 1.39 1.44 1.45 1.76 1 .so 2.15 2.16 2.59 2.60 2.6 1 2.87 2.98 2.97 3.22 3.60 3.71 3.93 2.99 3.50 2.52 2.76 2.21 1.84 2.01 2.12 1.61 1.60 1.46 1.21 1.11 0.97 0.69 0.96 0.77 0.76 0.81 0.48 0.55 0.52 0.27 0.25 0.25 3.0 1 0.33 0.04 0.72 0.74 0.54 0.90 0.97 1.03 1.37 1.24 1.09 1.44 1.38 1.97 1.83 1.80 2.1 1 1.96 2.30 2.41 2.37 2.47 3.02 3.01 3.09 Fig.1. SDS adsorption from mixed SDS-C,E, solutions as a function of bulk (SDS-C,E,) concentration at equilibrium and the mole fraction of SDS in (SDS-C,E,) in bulk solution.1809 M. J.Hey, J. W. Mactaggart and C. H . Rochester Fig. 2. C8E4 adsorption from mixed SDS-C8E4 solutions as a function of bulk (SDS-C,E,) concentration at equilibrium and the mole fraction of SDS in (SDS-C,E,) in bulk solution. Fig. 3. Total surfactant adsorption from mixed SDS-C8E4 solutions as a function of bulk (SDS-C8E4) concentration at equilibrium and the mole fraction of SDS in (SDS-C8E4) in bulk solution. A hypothetical standard state which is a function of total adsorbed surfactant is defined for each surfactant in which it is present at a surface concentration of rl+T2 and at zero surface pressure. The chemical potential of surfactant 1 in a mixed adsorption layer can then be written as (1) where x, = rl/(T1+T2), nl is a surface pressure equal to the difference in interfacial tension between solvent covered surface and surface with adsorbed surfactant 1, a, is py = mp9°(rl + Tz, T ) + RT In xlfl - xl a,1810 Mixed Adsorption at the Carbon-Aqueous Solution Interface the partial molar area of surfactant 1 and f, is a surface activity coefficient which allows for the dissimilarity between 1-1 and 1-2 interactions: f approaches unity as x, approaches unity.For the surfactant present in the dilute solution phase at a concentration c, ideal behaviour is assumed so that the chemical potential is given by Therefore, at equilibrium when py = p l The corresponding equation for surfactant 2 is Subtracting eqn (4) from (3) leaves where (AGP-AGP) is the difference in the standard partial free energies of desorption of the two surfactants. At this point we introduce the regular solution approximation for the surface activity coefficients : Eqn (5) then becomes At a given total adsorption (I?, + r,) the parameters n,, n,, a, and a, might be expected to be independent of the composition of the adsorbed layer, since the mode of adsorption of each surfactant will be most strongly influenced by the degree of saturation of the surface.A plot of In [x, c2/(c, x,)] against x, should then be linear if the regular solution approximation holds. The four sets of data points corresponding to initial solution concentrations of (4,5,6 and 8) x mol dm-3 and one set corresponding to the more concentrated solutions which were above the c.m.c. at equilibrium are plotted in fig. 4. Each set of points has been used to derive a 'least squares' straight line, which is shown.It can be seen that, in view of the fluctuations in rl + T2 for each set, a reasonably linear correlation is found as predicted. The slopes can be used to find p, which is a measure of the non-ideality of mixing of the two surfactants on the surface. For increasing surface coverage the values of p obtained are 0.8, 0.8, 0.6 and 0.2. In contrast to these positive values p is found to be - 3.0 for mixing of SDS and C8E4 within mi~elles.~ The difference may be accounted for by considering the likely configurations of the adsorbed molecules. At low coverages, adsorption of C8E4 probably ocurs with most of the molecule lying flat so that interactions exist between the hydrophilic ethylene oxide chains and alkyl groups.On the other hand, when micelles form the segregation of hydrophobic groups into the interior region leads to lower free energies. As surface coverage is increased it is believed that both SDS and C8E4 are adsorbed with progressively less of the molecule attached to the surface,, leaving predominantly hydrophobic groups in close proximity. Thus the adsorbed layer will tend to form a structure more akin to a micellar arrangement causing the value of the surface interaction parameter to move towards the value of the micellar interaction parameter.M . J. Hey, J . W. Mactaggart and C. H. Rochester 181 1 2 n - 1 Y, 3 .-. N 0 r( c CI I -1 Fig. 4. The relationship between surfactant mole fractions on the surface (xl, x,) and equilibrium bulk concentrations (c, c,) plotted according to eqn (8) for various total surface concentrations: (0) 1.2 x (0) 1.52 x 10-6; (A) 1.8 x (T7) 2.6 x and (A) 2.95 x mol m-2.A more detailed discussion of the significance of surface p values is unwarranted at this stage. It is sufficient to conclude that for the SDS-C,E,<arbon system the regular solution approximation appears to offer a convenient means of characterising the non-ideality of mixed adsorbate layers at different total coverages. Because SDS and C,E, have similar affinities for the carbon surface, it is easy to achieve approximately constant total coverage by taking constant initial total bulk concentrations. For other systems, however, this may not be possible, and an interpolation procedure may be necessary before p can be calculated. We wish to thank the S.E.R.C. for the award of a Studentship to J. W. M. and Dr M. Hull of Unilever Research for helpful discussions. References 1 M. J. Hey, J. W. Mactaggart and C. H. Rochester, J. Chem. SOC., Faraday Trans. I , 1984, 80, 699. 2 D. N. Rubingh, in Solution Chemistry of Surfactants, ed. K. L. Mittal (Plenum Press, New York, 1979), 3 M. J. Hey, J. W. Mactaggart and C. H. Rochester, J. Chem. SOC., Faraday Trans. I , 1985, 81, 207. 4 G. G. Jayson, G. Thompson, M. Hull and A. L. Smith, in Adsorptionfrom Solution, ed. R. H. Ottewill, 5 B. T. Ingram, Colloid Polym. Sci., 1980, 258, 191. 6 M. J. Rosen and X. Y. Hua, J. Colloid Interface Sci., 1982, 86, 164. 7 F. Karush and M. Sonenberg, Anal. Chem., 1950, 22, 175. vol. 1, p. 337. C . H. Rochester and A. L. Smith (Academic Press, London, 1983), p. 129. 8 M. J. Hey, J. W. Mactaggart, R. B. Thomas and C. H. Rochester, J. Colloid Interface Sci., 1983, 93, 574. 9 C. P. Kurzendorfer, M. J. Schwuger and H. Lange, Ber. Bunsenges. Phys. Chem., 1978, 82, 962. 10 R. Aveyard and D. A. Haydon, An Introduction to the Principles of Surface Chemistry (Cambridge Paper 5/ 1202; Received 15th July, 1985 University Press, Cambridge, 1973), chap. 6.
ISSN:0300-9599
DOI:10.1039/F19868201805
出版商:RSC
年代:1986
数据来源: RSC
|
19. |
Determination of micelle size and polydispersity by fluorescence quenching. Theory and numerical results |
|
Journal of the Chemical Society, Faraday Transactions 1: Physical Chemistry in Condensed Phases,
Volume 82,
Issue 6,
1986,
Page 1813-1828
Gregory G. Warr,
Preview
|
PDF (941KB)
|
|
摘要:
J. Chem. SOC., Faraday Trans. 1, 1986,82, 1813-1828 Determination of Micelle Size and Polydispersity by Fluorescence Quenching Theory and Numerical Results Gregory G. Warr and Franz Grieser" Department of Physical Chemistry, The University of Melbourne, Parkville, Victoria 3052, Australia A new interpretation of both steady-state and time-resolved fluorescence quenching of a probe molecule in micellar solutions is presented. It is sensitive to weight-average aggregation number as well as to the shape of the micelle size distribution. Numerical results and experimental data are analysed and the results are compared with other techniques and other interpretations of the results. The advantages and drawbacks of the fluore- scence quenching technique are discussed. The determination of micelle aggregation numbers by the quenching (or excimer formation) of a solubilized fluorescence probe molecule is currently a quite common practice. Much work has been done on all aspects of the theory of this technique since Turro and Yekta originally published their steady-state fluorescence quenching analysis on measurements in micellar sodium dodecyl sulphate solutions. 1-6 In virtually all cases reported the analysis, whether from steady-state or transient fluorescence experiments, has been done using a single micelle aggregation number under the assumption that the aggregates are monodisperse.l? ' 9 * In the Turro and Yekta methodl the fluorescence intensity, Z(Q), at quencher concentration, Q, is related to aggregation number, i, via where X is the total surfactant concentration at Z(0) is the fluorescence intensity in the absence of added quencher.The denominator of eqn (1) has been corrected for the concentration of monomeric amphiphile, assumed equal to the critical micelle concen- tration (c.m.c.). For time-resolved measurements, the analogue of eqn (1) is where F(0) is the fluorescence intensity at time zero, and &(O) is the fluorescence intensity from micelles containing no quencher molecules, at t = 0. (In practice this is determined by extrapolation of the unquenched, exponential, fluorescence decay curve back to t = 0.)9 The micelle aggregation number is extracted from a plot of the left-hand side of either of these relations vs. Q / ( X - c.m.c.). Under the conditions of monodispersity this should yield a straight line of slope i and intercept zero.In fact, as we shall show, such a plot is quite insensitive to deviations from linearity and can lead to spurious results, as well as disguising information which is present in the results. Almgren and Lofroths recently developed an alternative treatment of fluorescence quenching data which does not rely on the micellar solution being monodisperse. They 18131814 Fluorescence Quenching in Micelles- Theory showed that the quenching behaviour was actually related to the entire distribution of micelle sizes, Xi. The treatment of fluorescence quenching experiments proposed by Almgren and Lofroth for a polydisperse solution of micelles is a sum of all the contributions in the system. The experimental sample of the solution consists of all micelles Xi, z, containing more than one amphiphile, i, and containing z probe molecules (z 2 1) and x quencher molecules (x 2 0).The fraction of fluorescence at t = 0 from micelles containing no quencher is thus:9 x zxi,o,z i > 1 z=o z=1 Results derived beyond eqn (3) require several assumption^:^ (i) the mean number of probes and quenchers in a micelle is proportional to the aggregation number, i, of the micelle; (ii) the number of quencher molecules in a micelle is independent of the presence of a probe molecule (and vice versa); (iii) quenching of fluorescence occurs in the static limit, i.e. there is no significant exchange of amphiphiles in or out of micelles during the lifetime of the fluorescent probe. These assumptions lead to a much more straightforward representation of the t = 0 fluorescence.Namely, x iXi exp (- ri) F,(O)/F(O) = i > l Z iXi (4) i > l where 51 is the ratio Q/(X-c.m.c.) = Q/xi > iXi. average aggregation number, ( i)&g given by By analogy with the Turro and Yekta result it is convenient to define the quencher 1 4' i>l i > l = - In [ iXi/ iXi exp (-qi)]. (The prime is included to denote summation over all species except i = 1.) In a monodisperse micellar solution (i)& will be independent of 51 (or of Q) and equal to the micelle aggregation number, i. However, when Xi is polydisperse, (i)b is a function of quencher concentration, and of the size distribution function Xi. Theory Time-resolved Fluorescence Experiments The quencher average aggregation number of a micellar solution derived from transient fluorescence results can be related to the shape of the micelle size distribution in a straightforward manner.Expanding the exponential function in eqn (5) gives a series representation of (i)& : 51 12-1 n! i, where in is the nth moment of the micelle size distribution, Xi, , inxi. The sum inside parentheses can be transformed into a direct power series for ( i ) b in 7.G. G. Warr and F. Grieser 1815 I I I 80 - ( b ) (i>b 40 - - I 1 I 0.01 0.02 0.03 7) Fig. 1. Schematic diagram of the quencher-average aggregation number, (i)& versus 7 for various distributions of micelle size, with (i)(w fixed at 80. (a) Monodisperse micelles; (b) symmetrical distribution with standard deviation 0.5 ; (c) highly skewed distribution with standard deviation 1.0.Note that (b) will exhibit curvature at higher 7 due to higher order, even moments (7zn"-1) of the size distribution, and that it will not intercept (i)& = 0. where the coefficients of yn-l are the nth central moments of the micelle size distribution function iXi/(X- c.m.c.), i.e. ( i ) b = (i)&- -((i-(i)&)n)qn-l. (- 12-2 n! This equation generates Almgren and Lofroth's resultg that limv-,o ( i ) b = (i)&, the weighted-average aggregation number. Note that the coefficients of the polynomial, up to third order in q are the mean-square deviation, the raw skewness and the raw kurtosis, or peakedness, of the micelle size distribution, respectively. It follows immediately from eqn (8) that the results of such a transient fluorescence quenching experiment can be used to determine not only whether Xi is a monodisperse solution ((i);d constant versus r), but also whether the distribution is symmetrical (linear ( i ) b versus 7) or skewed (non-linear (i);).This representation of ( i ) b is, of course, independent of any model distribution which could be fitted to the results. Experimentally, only a small fraction of 7-space is accessible in order that solubilization of quencher molecules does not perturb the host micelle. Although ( i ) b must be bounded above the (i);N and below by zero as a function of 7, it is unlikely that curvature due to effects of higher order than kurtosis (q3) will be apparent in the experimentally available regime.(Typically 7 varies between 0.0 and 0.03 in quenching experiments.) This is shown schematically in fig. 1. With micellar solutions which are nearly monodisperse, such as ionic micelles at low salt,l? 8 * lo the series represented in eqn (8) may be curtailed at the linear 7 term. This gives a very useful approximation to ( i ) b valid over a considerable region of the experimental range : ( i ) b = (i);N -+((i- (i)&)2)q = (i);N - io2q (9) where 0 is the root-mean-square deviation of iXi.1816 Fluorescence Quenching in Micelles-Theory Steady State Fluorescence Experiments The fluorescence intensity detected in a static measurement is the integral of the time-dependent fluorescent decay curve. In a polydisperse solution, this must be a summation over all micelles in the system.The normalised time-dependent fluorescent intensity, according to the present model, is9 I: iXi exp [ - k, t + qi(e-K2/i - l ) ] F(t) - i > l -- F(0) I: iXi i > l where k , is the rate constant for the unquenched fluorescence decay of the excited probe molecule and Kq is a bimolecular collision coefficient for the rate of quenching. The quenching rate was assumed to be proportional to the number of quenchers, and inversely proportional to the micelle size. The integrated intensity is obtained by integration of each term in eqn (10) and is normalised to the unquenched fluorescence intensity, I(0). This gives Similarly to the time-resolved results, this expression lends itself to the definition of an intensity average aggregation number ( i ) ; : (12) 1 ( 0 ; = In [I(O)/I(v)l.Clearly this quantity may be expressed as a power series in q, just as ( i ) & i.e. where 1 ( i ) ; = --In C Tn(q)qn [n:o 1 I: in+lXi exp(-qi) Tn(q) = i’’ n! I: iXi i > l Note that the first term in this series, &(q), is simply exp (-q(i)b), which leads to a considerable simplification for ( i ) ; e-q(i)h Tn(q)qn). (15) Again, as for (i)& the power series in eqn (1 5 ) may be transformed into a direct power series in q for ( i ) ; , recalling that ( i ) b is also a function of q: 00 ( i ) ; = (i>b+ I: tn(q) vn n-o where tn(q) =fi[q(q); j = 1, 2, . . . , n], eq(i)b} [ref. (1 l ) ] .G. G. Warr and F. Grieser 1817 (i); is now expressed as the sum of two power series. The first of these is simply dependent on the shape of the micelle size distribution, iX6, and the second may be viewed as a correction term due to the finite rate of quenching in a dynamic process.As the quenching rate, Kq/k,, becomes large, in eqn (ll), I(q)/I(O) tends to F,(O)/F(O), corresponding to a static quenching process. (Time-resolved fluorescence techniques are inapplicable in the case of static quenching as the fluorescence decay proceeds as a single exponential function.) Expansion of eqn (1 6) to first order in q gives an expression for integrated intensity measurements which is useful for comparison with the expansion of (i)b: Rather than use the unwieldly eqn (17), it will be convenient to compare the behaviour of (i); with (i)b for a monodispersion of micelles. In this situation, with a unique aggregation number, i = (i)b, the steady state behaviour is given by (i); will not differ significantly from (i)b as long as Kq/ko i is of an order of magnitude greater than 1, i.e.as long as the quenching is relatively fast. The finite quenching correction to is, to first order in q (19) Again in a similar way to (i); the linear form of (i); is sufficiently accurate over the usual experimental range (q < 0.03). However, finite quenching affects even the limiting value of (i); as q approaches zero and can make is significantly less than i. i2 2 (i); = i[ 1 - (1 + Kq/k, i)-l] -- [( 1 + 2Kq/k, i)-l- (1 + Kq/k, i)-2] q. Results Steady State Fluorescence The effect of the finite quenching rate, which is expressed in terms of Kq/iko, is illustrated in fig.2 for micelles of aggregation number 100. The apparent aggregation number and standard deviation are determined solely from results generated for fig. 2, and are given It is clear that steady state results can be grossly in error in both aggregation number and polydispersity, while giving every indication of being reliable data. Time-resolved fluorescence results are to be preferred if the relative quenching rate, R = Kq/ik,, is not known, or is known to be of order < 10. The magnitude of the error in (i)blapp may1818 Fluorescence Quenching in Micelles- Theory 100 1 - (i); SO 8or== 0.01 0.02 7) Fig. 2. Integrated intensity-average aggregation number, ( i ) ; versus q for monodisperse micelles at various K,/ko i values and with i = 100. KQ/koi B (i)& (a) 100 0.07 99 (b) 25 0.18 98 (c) 10 0.26 91 (d) 5 0.33 84 (e) 1 0.56 50 cf) 0.5 0.67 33 Fig.3. Aggregation number correction factor versus relative quenching rate, R = KQ/ko i. Note that increasing i decreases R, and that (i)blaPp is more in error. ( x = numerical results for exponential distributions; see text.)G. G. Warr and I;. Grieser 1819 'I 0 0 A A A A A A A AAA P A R S Q 1 I 1 I 0.2 0.4 0.6 [ NaCl] /mol dmT3 Fig. 4. Weight-average aggregation numbers for SDS in NaCl solutions: (A) uncorrected fluorescence results ; (A) corrected fluorscence results ; (0) time-resolved fluorescence results [ref. (S)]; (0) other results. be seen in fig. 3, where the aggregation number correction factor ( i)&/(i);Nlapp is plotted against R. It has beem assumed throughout that the rate of quenching of the fluorescent probe in a micelle is inversely proportional to the micelle size i.Intuitively the frequency of encounters between probe and quencher must decrease with increasing i, but the form of this i-dependent decrease is not clear. Since (i)& is not dependent upon quenching rate, time-resolved fluorescence experiments are unaffected by this assumption. However, the steady state result would be significantly changed by a surface-area-determined quenching rate (i-2/3 for spherical micelles) in all the rate-dependent terms of eqn (llh(l9): thus (i)&lapp would not be as shown in eqn (19). A consequence of the i-dependence of the relative quenching rate (and hence of (.i)blapp) is that a given probe-quencher combination which is reliable for moderately sized micelles could be quite inappropriate for larger micelles. The system of Ruthenium tris(bipyridy1) and 9-methylanthracene in sodium dodecyl sulphate (SDS) micelles was used in steady state experiments by Turro and Yekta.l Subsequent workE* lo has shown that while the results were reliable at low salt concentrations, the marked increase in aggregation number occurring at ca.0.4 mol dm-3 NaCl was not reproduced. Almgren and Lofroth's kinetic study of this system produced aggregation numbers in much better accord with other lo, l2 They also reported rate constants for quenched and unquenched decay of the excited state of the probe. These values have been used to correct the steady state results of Turro and Yektal to time-resolved (i)b values according to the curve in fig.3. They are shown in fig. 4 together with the steady state results,l time-resolved results* and several reference data from various techniques.l0V l2, l3 Discrepancies are largely due to the original analysis of the aggregation numbers and to errors in the rate constants. The steady state results were determined1820 Fluorescence Quenching in Micelles-Theory at a single quencher concentration,l while the transient fluorescence data were analysed according to eqn (1) :* this presumes monodispersity at all NaCl concentrations. The value of i derived from eqn (1) is equivalent to averaging ( i ) ; over all selected experimental rl values. Corrected aggregation numbers and raw experimental data cannot, therefore, ever quite coincide unless the static fluorescence data are analysed according to eqn (19).The difficulties with steady state fluorescence determinations of aggregation numbers are essentially the same for polydisperse micellar solutions. Kq/iko must be greater than 10 to keep ( i)klapp within reasonable experimental accuracy. This polydisperse systems this precaution also ensures the reliability of other moments of the micelle size distribution. Moments of Model Distributions Several model distribution functions, Xi, were used in this study to generate ( i ) b and ( i ) ; data. These are: i - ( i ) N ) 2 (i) Gaussian : Xi = exp - (( 2s2 ) where ( i ) N is the number-average of Xi over the range [ - co, co] and s is the root-mean-square deviation of i about ( i ) N .(for i < io (ii) Exponential: 1 Xi = exp (i0;i) - I \when i < io, Xi = 0) I ' where io is some starting value for the exponential distribution and CT is the decay length. for i, < i < iu Xi = - I , - il when " - i i < il; i > iu,Xi = 0 I (iii) Triangular I (peak at low i): where il and iu are the lower and upper limits of the finite distribution, respectively. for il < i < iu (iv) Triangular I1 (peak at high i): when i c il; i > iu, Xi = 0 (v) Zeroth-order logarithmic normal distribution (ZOLD) : ( (In i --sin i)2 Xi = (2~112 $1 exp - where i and s define the peak and decay length of the distribution:l* All model distributions are normalized to peak at Xi = 1. The distributions selected have no basis in thermodynamics, save the exponential form which has been put forward15 as a description of the size distribution of cylindrical micelles, in the limit of large i.They are used here only as examples of functions with variable central moments.G. G. Warr and F. Grieser 1821 800 600 (i>b 400 200 I I I 0.01 ' 0.02 rl Fig. 5. Quencher-average (-) and intensity-average (---) aggregation numbers for exponential distributions with i, = 60. CT: (a) 50; (b) 100; (c) 200; ( d ) 500. Results for each distribution were calculated by taking up to 300 terms in i over the full range of i and summing eqn ( 5 ) or (11). For the nominally infinitely ranged distributions [eqn (22), (23) and (26)], sums were performed from i = 1 up to i > ( i ) & where Xi < 0.01. Gaussian distributions yielded ( i ) h which were well fitted by the linear order moment expansion of eqn (1 8) in all cases where s < 1 /2( i);V and the function was symmetrical for i > 1.Results for exponential size distributions are shown in fig. 5 and 6. In fig. 5 both ( i ) b and ( i ) ; are shown for i, = 60 with a range of decay lengths. ( i ) ; was determined using K,/k, = lo3, which is of the order of the best experimental system available.'. l6 Even for this system, large (i)& leads to static results, ( i)&lapp, which are enormously in error. The ratio (i)&/(i)&lapp for these systems is also shown in fig. 3 with the curve for the monodisperse case. Agreement between the monodisperse and highly asymmetric, polydisperse cases gives a deal of confidence in applying the correction factor to polydisperse systems.Fig. 7 and 8 show (i)& values for triangular and ZOLD size distributions of various polydispersity. A polynominal fit up to order 3 in q to these model distributions yields the following moments . I a2q Aq2 Eq3 ( i ) b = ( z ) ~ - - - + - - ~ 2 6 where A is the raw skewness of iXi and E is the raw excess, or kurtosis. The even moments (a2 and E ) are always positive, and A can take either sign.1822 Fluorescence Quenching in Micelles- Theory 800 600 (i>h 400 200 100 8 0 1 I 0.01 0.02 I I 1 0.01 0.02 .r7 Fig. 6. Quencher-average aggregation numbers for exponential distributions with i, = 20. (a)-(d) as for fig. 5. Inset: expanded scale plot of 0 = 50 distribution. For convenience the skewness and kurtosis are best standardised with respect to a Gaussian distribution, viz.1A A = - - 2 0 3 & +3) 8 o4 II > 0 denotes positive skewness (mean > mode) and E > 0 means that iXi is more peaked about (i)& than a Gaussian distribution. The positively skewed functions [eqn (23), (24) and (26)] all show significant deviations from linearity in plots of (i);d vs. 7. On an expanded scale even moderately broad, skewed distributions are quite non-linear as is shown in fig. 9. However, the negatively skewed triangular distributions yield far more linear (i);d curves. This is largely due to the lower bound on (i)b, which forces the function to be convex down. Negative skewness would imply a concave down function: thus the higher moments at least compensate for the effect of skewness in such a system.Cubic fits" to eight data points for several of the data are shown in fig. 5-9. These are sufficiently good that higher order polynominals were not attempted and it is not anticipated that experimental data will require higher orders of 7 to be fitted. Derived distribution moments are listed in table 1 with independently determined (i)& values. All results for (i)& are in excellent agreement with the correct value; 0 is also in good agreement with theory; and both II and E are of the expected sign and order of magnitude for all but one system. The negatively skewed, triangular distribution function [eqn (25)] has an apparentG. G. Warr and I;. Grieser I 1 I I 8ool 1 1823 0.01 0.0 2 B Fig. 7. Quencher-average aggregation numbers for triangular distributions with il = 60, i, = 135, 210, 360 and 810 (for curves bottom to top of diagram).(---) Distributions with peak at low i; (-) distributions with peak at large i. a o o 600 (i)h 400 2001824 Fluorescence Quenching in Micelles- Theory L 1 I 0.01 0.02 9 Fig. 9. Expanded scale plots of (i)& for exponential (-) and ZOLD (---) distributions with various (i,,, a) or (i, s) as shown. Table 1. Reduced micelle size distribution moments from cubic fits to model fluorescence decay data cubic feet input true equation parameters (i)& (i)& a I I E (i>;’+Y 60 500 60 200 60 500 20 100 60 810 60 810 20 3.0 20 1.0 856 3 59 143 182 41 1 636 749 145 816 410 0.213 363 197 0.317 135 90 0.910 180 112 0.897 414 188 0.236 631 183 0.011 757 342 0.189 145 100 0.462 - 0.364 -0.348 -0.162 -0.31 1 -0.373 - 0.392 - 0.366 - 0.305 41 5 245 115 135 258 462 402 108 positive skewness.This result is clearly unreliable as E derived from the polynomial fit is negative and this is not possible. Also note that A is an order of magnitude smaller than the other model systems. Inspection of the fits to (i)& data is perhaps the best way to decide upon the gross acceptability of the polynomial representation. That is, o2 and E must be positive and the curve must not exhibit a major inflection or accommodate experimental scatter, as we will show below (Appendix 1). Based on the polynomial fits to model data, the valuesG . G. Warr and F. Grieser 1825 000 750 500 250 a- 1 I I 0.01 0.0 2 r) Fig.10. Best polynomial fits (see table 2) to (i)b data for C,,E,. (---) Exponential distribution best fit to data at 44.1 "C [ref. (18)]. Inset: comparison of linear and cubic fits to data at T = 17.4 "C. T/OC: (V) 17.4; (a) 24.7; (V) 34.4; (A) 44.1; (0) 46.2; (0); 48.7; (A) 50.6. of (i)& and o are able to be derived quite reliably from fluorescence quenching data. The skewness is a less reliable result, but values of il significantly greater than zero reflect true positive skewness. However, the reliability of E is less certain; its sign ( E > 0) is a useful check on the fitted curve. Growth of Micelles of Hexaethylene Glycol Mono-n-dodecyl EtherlS The data of Almgren and Lofroth for the quenching of pyrene fluorescence by di-N,N-butylaniline in micelles of hexaethylene glycol mono-n-dodecyl ether (C,,E, ) shows a growth of the micelles as the temperature is increased up to the cloud point.This is shown in fig. 10. Polynomial fits to these data reproduce the gradually increasing curvature in (i)& with temperature, but are of limited usefulness as there are only five data points, which are quite scattered in some curves. The advantage of fitting the data to model distributions is that the fitting function cannot accommodate experimental scatter. Table 2 lists parameters derived from polynomial fits to the data of fig. 10, together with the results of fitting Gaussian and exponential functions of Almgren and Lofroth.l8 For the three temperatures; 17.4,24.7 and 34.4 "C the linear fits are preferred because the experimental scatter is a significant fraction of (i)&.This means that, although (i)& from the quadratic fits are in good agreement with the linear result, a minimum occurs in the fitted (i)& which suggests that both o and A are unreliable. At 34.4 "C this is not so and is reflected in the very small value of A. (Cubic fits to the low- temperature data are inappropriate with only five data points; see inset, fig. 10.) At higher temperatures the data are well fitted by higher-order polynomials where1826 Fluorescence Quenching in Micelles-Theory Table 2. Polynomial fits to fluorescence decay data for C,,E, micelles, compared with results for model distributions Gaussian fita exponential fita polynomial fit 77°C W T + Y (i>& I7 (i>& order* (i>& I7 I.& ~~ ~ I* 17.4 137 156 55 2 1* 24.7 203 245 83 - 2 1* 34.4 310 372 102 - 2* 44.1 477 719 204 1200 2* 3* - 46.2 504 801 225 1600 2 48.7 559 908 244 2000 2 50.6 593 995 260 2600 3 3* 3* 152 48 163 89 242 80 274 145 373 102 374 103 819 287 869 345 937 328 1018 402 1110 382 1177 440 1622 666 0 0.462 0 0.275 0 0.004 0.101 0.180 0.094 0.166 0.090 0.133 0.138 0 0 0 - 0.37 0 -0.37 0 - 0.37 -0.37 a From ref. (17). * Asterisk denotes best fit, see fig. 10. experimental errors are less significant. It is only at 50.6 "C that the fitted curve has an inflection point and a fourth- or higher-order fit would be more appropriate. The other temperatures yield (i)& much lower than for exponential fits, but the agreement between second- and third-order fits is such that the cubic fits are taken as being reliable.The cubic fit to the data at 44.1 "C is considerably better than an exponential fit,18 as would be expected with two additional parameters. This is also shown in fig. 10. Independent of any model for Xi, Almgren and Lofroth's results for C,,E, show a distinct increase in both size and polydispersity as temperature is increased towards the cloud point. Also the onset of marked, positive skewness is apparent at and above 44.1 "C, and it remains almost constant in its reduced form, A. The negative kurtosis is also approximately constant, and is a consequence of the positive skewness : mean % mode leads to iXi not being peaked about the mean. Discussion The moment expansion [eqn (S)] for ( i ) b was derived following several assumptions about the distribution of probe and quencher molecules in micelles.The key assumption leading to integral order moments is that the mean number of probes, z, in a micelle of size i, ( z ) ~ , is proportional to i. Mass balance requires that the ratio of probe to amphiphile concentrations is = C. ( z ) ~ Xi/C iXi. j7j- z i x i i i ( ( z ) ~ ) is a set of undetermined coefficients of Xi, and the assumption of linearity is the simplest form to adopt. It would be expected that rod-shaped micelles would follow a linear ( z ) ~ relationship as both surface area and volume are linear with i, so any solubilization description willG. G. Warr and F. Grieser 1827 be suitable. Geometrical argumentsl5 suggest that spherical micelles cannot be grossly polydisperse, so the exact form of ( z ) ~ should not affect derived ( i ) b data seriously.Eqn (8) is equally applicable to pyrene excimer formation.2* 9 9 l3 The formalism leading to F(0)/Fm(O) is somewhat different, but the result identical. It is only the steady state experiment which cannot be performed using pyrene excimers because the probe concentration, and hence I(O), is not constant. The classical, Turro and Yekta method for interpreting fluorescence quenching data1 has been used in pyrene excimer studies of SDS micelles in NaCl Time-resolved results for this system were vastly improved over static measurements with Ru(bipy)i+ and 9-methy1anthracene.l However they still yielded results at 0.5 mol dm-3 NaCl which were lower than literature values.This was attributed to the rate of diffusion of pyrene molecules to form dimers, which should be reflected in the fast decay portion of the time resolved fluorescence. It is equally well described by eqn (8) which shows how ( i ) b may be strongly dependent upon q in a polydisperse system. Aggregation numbers, ( i)’;c+y, derived from a classical analysis1 of both computer generated and experimental18 data are included in tables 1 and 2. In all cases ( i);’+y is much less than (i)& determined by other techniques. Not only is (i);’+y incorrect for these systems, the plot of eqn (1) [or eqn (2)] is insensitive to the effects of 0, A, E, . . . which is highlighted by an ( i ) b plot. Conclusions Based on the principles of analysis of fluorescence quenching data in micelles, a polynomial representation of experimental results has been .developed which is sensitive to the moments of the micelle size distribution, iXi.Analysis of both simulated and experimental time-resolved fluorescence data yields reliable weight average aggregation numbers and mean-square deviations, as well as indicating the skewness and kurtosis. The limits of this technique with respect to experimental scatter of data and its applicability as a static fluorescence technique were also investigated. Several drawbacks of the fluorescence quenching technique as it is currently used were also discussed. G. G. W. acknowledges the receipt of a Commonwealth Postgraduate Research Award. The authors wish to thank Dr J-E. Lofroth for providing preprints and experimental information during the course of this work.This work was supported by Grants from the Australian Research Grants Committee. Appendix It is a straightforward piece of algebra to show that ( i ) b versus q cannot exhibit any stationary points. Beginning with the first q-derivative of eqn (9, we have, (A 1) Both terms on the right of eqn (A 1) are negative for all finite q > 0. Hence there are no horizontal turning points on (i)&(q), q > 0. The second derivative of eqn (5) is1828 Fluorescence Quenching in Micelles- Theory The first term on the right-hand side of eqn (A 2) is > 0 for all q > 0. Performing the differential on the term in parentheses yields where = iXi e-qi is a positive function for positive q and i. Eqn (A 3) is the negative of the mean-square deviation of & (which is always positive), and becomes a positive quantity when multiplied through by - l/q. (i32(i)b)/(aq2) is therefore a sum of two positive contributions so is > 0 for all q > 0, and ( i ) & can exhibit no inflection points. References 1 N. J. Turro and A. Yekta, J. Am. Chem. SOC., 1978, 100, 5951. 2 S. S. Atik, M. Nam and L. A. Singer, Chem. Phys. Lett., 1979, 67, 75. 3 S. S. Atik and L. A. Singer, Chem. Phys. Lett., 1979,66, 234. 4 P. P. Infelta, Chem. Phys. Lett., 1979, 61, 88. 5 M. Tachiya, J. Chem. Phys., 1982, 76, 340. 6 M. Almgren, F. Grieser and J. K. Thomas, J. Am. Chem. SOC., 1979,101, 279. 7 M. A. J. Rodgers and J. H. Baxendale, Chem. Phys. Lett., 1981, 81, 347. 8 M. Almgren and J-E. Lofroth, J. Colloid Interface Sci., 1981, 81, 486. 9 M. Almgren and J-E. Lofroth, J. Chem. Phys., 1982,76,2734. 10 J. P. Kratohvil, J. Colloid Interface Sci., 1980, 75, 271. 11 M. Abramowitz and L. A. Stegun, Handbook of Mathematical Functions (US. Natl. Bur. Stand., 12 H. F. Huisman, Proc. K. Ned. Akad. Wet., 1964, 1367; 367; 376; 388; 407. 13 P. Lianos and R. Zana, J. Phys. Chem., 1980,84, 3339. 14 R. L. Rowell, J. P. Kratohvil and M. Kerker, J. Colloid Interface Sci., 1968, 27, 501. 15 J. N. Israelachvili, D. J. Mitchell and B. W. Ninham, J. Chem. SOC., Faraday Trans. 2, 1976,72, 1525. 16 F. Grieser and R. Tausch-Treml, J. Am. Chem. SOC., 1980,102, 7258. 17 K. J. Johnson, Numerical Methods in Chemistry (Marcel Dekker, New York, 1980). 18 J-E. Lofroth and M. Almgren, Surfactants in Solutions, K. L. Mittal and B. Lindman (Plenum Press, Washington, D.C., 1970). New York, 1984), vol. 1, p. 736. Paper 5/1204; Received 15th July, 1985
ISSN:0300-9599
DOI:10.1039/F19868201813
出版商:RSC
年代:1986
数据来源: RSC
|
20. |
Determination of micelle size and polydispersity by fluorescence quenching. Experimental results |
|
Journal of the Chemical Society, Faraday Transactions 1: Physical Chemistry in Condensed Phases,
Volume 82,
Issue 6,
1986,
Page 1829-1838
Gregory G. Warr,
Preview
|
PDF (721KB)
|
|
摘要:
J. Chem. SOC., Faraday Trans. I, 1986, 82, 1829-1838 Determination of Micelle Size and Polydispersity by Fluorescence Quenching Experimental Results Gregory G. Warr and Franz Grieser* Department of Physical Chemistry, University of Melbourne, Parkville, 3052, Victoria, Australia D. Fennel1 Evans Department of Chemical Engineering and Materials Science, University of Minnesota, Minneapolis, 55455, U.S.A. Micelle aggregation numbers of a variety of surfactants have been determined experimentally using the fluorescence quenching method. Cationic, anionic and non-ionic surfactants were investigated using as a fluorophore a dialkyl derivative of the ruthenium@) tris(bipyridy1) complex cation, with 9- methylanthracene as a quencher. Results have been interpreted in terms of the theory of fluorescence quenching in a polydisperse micellar solution outlined in the preceding paper.Excellent agreement with literature values for weight-average aggregation numbers was obtained, as were reliable values of higher moments of the micelle size distribution. In the preceding paper we detailed a fluorescence technique for studying the entire size distribution function of a micellar solution. Experimental access to such information has been very limited or has been largely qualitative until Indeed polydispersity in micellar solutions has often been inferred from t h e ~ r y ~ - ~ or from the interpretation of results213 rather than being observed directly. The major advantage of the ( i ) b interpretation of fluorescence quenching results is that it does not require the sol- ution to be polydisperseg~10 and collapses to a constant in the case of monodisperse micelles.It is also immediately apparent from the experimental results whether or not there is significant polydispersity, skewness etc. in the size distribution function. The assumptions underlying the application of the ( i ) b interpretation can only be tested experimentally. While the case of hexaethylene glycol mono-n-dodecylether with pyrene and dibutylanilinelO9 l1 appears to work extremely well, it is highly desirable to examine a more widely studied and better characterized surfactant. To that end we have applied this technique and interpretation to studying micelles of sodium dodecyl sulphate (SDS) in the presence of added salt. Micellar solutions of several other surfactants have also been studied here by time-resolved fluorescence of the ruthenium tris(bipyridy1) cation and its di-n-octyl derivativel29 l3 to evaluate the usefulness of this technique.To pre-empt our results, we have found that experimental quenching rates as well as aggregation number parameters for anionic, cationic and non-ionic systems can all be reliably determined and are reported below. Experimental The surfactants used in this study were: sodium dodecyl sulphate, SDS (BDH specially pure, recrystallized) ; dodecyl dimethylamine oxide, DDMAO (Fluka, purum) ; dodecyl trimethylammonium bromide, DTABr (Tokyo Kasei, recrystallized). Dodecyl tri- 18291830 Fluorescence Quenching in Micelles-Experiment methylammonium hydroxide, DTAOH, was prepared by ion exchanging the Br anion as described previo~s1y.l~ Zinc dodecyl sulphate, Zn(DS),, was prepared by the usual method from SDS and ZnS0,.15 Ruthenium tris(bipyridy1) chloride, Ru(bpy):+, was obtained from ICN Pharmaceuticals and was used as received.Its di-n-octyl derivative, ruthenium(I1) di-4,4’-(n-octyl)-2,2’-bipyridyl bis(2,2’-bipyridyl) perchlorate (RuL;+),l2 was a gift from the CSIRO Division of Applied Organic Chemistry. 9-Methylanthracene (9-mA) was obtained from Tokyo Kasei. Fluorophore and quencher stock solutions were prepared as previously described.13 All solutions of amphiphile were made up in Milli-Q filtered conductivity water, and all other reagents used were analytical grade or better. Quencher concentrations were always sufficiently low that a Poisson distribution could reliably be used to describe the population in micelles.161 l7 Experiments were performed at 21 “C.Transient fluorescence experiments were carried out using a single photon counting apparatus under excitation at 580 nm by a mode locked, cavity dumped dye-laser system. The excitation pulse had a half-width of ca. 100 ps. Emission from the ruthenium probe was monitored at 650 nm, which is close to the emission maximum. The time-dependent emission was collected in a 1000 channel multichannel analyser and the decay curve fitted to a double exponential function with a variable background count and a known lamp profile. The unquenched lifetime of the fluorophore was determined in the absence of quencher and was fixed when fitting the double exponential. The lamp pulse was only one channel wide and the zero time channel was taken to be the lamp channel plus one.This yielded an acceptable result for SDSl89 l9 and changing this by _+ 1 channel did not alter the derived aggregation number parameters significantly. Results SDS with Added Salt In fig. 1 are shown quencher-average aggregation numbers for the much studied surfactant SDS in the presence of salt at various concentrations. With no added salt the SDS micelles are monodisperse with an aggregation number of 55 1. This is in agreement with light scatteringlg and other measurementsz0 on this system. Weight-average aggregation numbers, as well as root-mean-square deviation (0) and other size distribution parameters are listed in table 1 for SDS in the presence of salt at various concentrations.These results all show good agreement with the best data available in the literature.lO9 18-,0 Note the excellent agreement between these results and literature values with no added salt and with 0.189 mol dm-3 NaCl.lg? 2o At higher salt concentrations the literature results are based on fluorescence quenching experiments18 interpreted according to the transient analogue of the Turro and Yekta analysis.l? lo Inspection of fig. 1 immediately reveals why our results for (i)b are higher than those obtained elsewhere:18 the Turro and Yekta analysis essentially corresponds to averaging ( i ) b for the experiment,1° and not to the number average aggregation number of the system.18 Our results are actually in better agreement with aggregation numbers for SDS determined by quasielastic light scattering, although some doubt appears to remain as to their reliability [see discussion in ref.(1 8)]. However this and other studiesg-ll suggest that the fluorescence quenching technique is as good as other available methods for determining aggregation numbers.20 These results also represent the first direct experimental determination of the theoreti- cally predicted, and generally accepted, transition from monodisperse (spherical) to polydisperse micelles67 21 with the addition of salt. A previous attempt to study the SDS systemg showed no discernible trends, and it was concluded that size fluctuations were such that the static limit approximation was not valid.1° However, several studies on the dynamics of micelle structure would suggest that significant fluctuations in aggregation number occur over timescales longer qhanG.G. Warr, F. Grieser and D . F. Evans 1831 2 5 0 200 100 0.01 0.02 77 Fig. 1. Quencher-average aggregation numbers, (i)& versus scaled quencher concentration, 7, for the fluorophore-quencher-amphiphile system Ru(bpy):+-9-mA-SDS at 0.05 mol dm-3 SDS with added salt. Solid line is the best polynominal fitlo to the data (see table 1). Salt concentration/mol dm-3: (a) 0; (b) NaC1, 0.189; (c) NaC1, 0.424; (d) NABr, 0.475. Table 1. Micelle size distribution parameters derived from fluorescence measurements amphip hile conditions (i>;Y 0 3, (i)b (lit.). SDS SDS SDS SDS DTABr DTAOH DDMAO DDMAO DDMAO DDMAO DDMAO Zn(DS), + Zn(DS), + Zn(DS), + Zn(DS), SDS SDS SDS no salt 0.189 mol dmP3 NaCl 0.424 mol dm-3 NaCl 0.475 mol dm-3 NaBr no salt no salt 0.1 18 mol dm-3 NaCl p = 0.0 0.1 16 mol dm-3 NaCl p = 0.5 0.1 14 mol dm-3 NaCl 0.208 mol dm-3 NaCl p = 0.5 1.06 rnol dm-3 NaCl p= 1.0 p = 0.0 a,+ = 0.419 a,+ = 0.623 a,+ = 0.805 no salt 55 98.6 f 0.4 162+ 1 241 & 1 79+3 30f5 123+20 234k 1 114f7 301 +3 125 f 3 110 99 92 74 0 29+3 99+7 139f3 0 0 68 f 20 132f2 32+ 10 156+20 57+ 10 < 17 < 28 < 27 0 0 0 - 0.1 0 0 0 0.39 + 0.06 0.21 k0.02 0 0.09 & 0.02 - 0 0 0 0 0 55 (20) 1 OO( 20)b 140( 1 8)b 180( 1 8)b 92(24)c 1 87(24)c 97(24)" 593(24)d a Reference numbers in parentheses. Interpolated result for added NaC1.In 0.10 mol dmb3 NaCl. In 0.20 mol dmP3 NaC1.61 FAR 11832 Fluorescence Quenching in Micelles-Experiment 500-1000 n ~ . l - ~ All that is required in actuality is that fluctuations in size are slower than the quenching rate, which must necessarily be < 200 ns for resolution of the two decay curves. Our results bear out the static limit model of a polydisperse micellar solution. Such effects as fast exchange of amphiphiles between micelles, or a significant deviation from direct proportionality in the mean quencher concentration per micelle,1° would alter the form of ( i ) b and a polynomial expansion would be meaningless. Hence ( i ) b at q = 0 would not be equal to (i&. Examination of the equations and assumptions used to derive ( i ) b [ref. (9) and (lo)] reveal that if (i)’w is indeed the q = 0 limit, as we find, then all higher moments of the size distribution function are also correctly described.Our experiment is not sensitive to micelle morphology, however, and we can only draw indirect conclusions in this respect. The increase in both o and skewness, A, with added salt is inconsistent with spherical micelles, but rather it is suggestive of a sphere-to-rod transition. The theoretical arguments for such a shape and size transition have been recurrent in the literature on micelles.6y 22 Put simply, addition of salt screens the electrostatic head-group-repulsion between amphiphiles within a micelle, allowing the surface area per molecule to be reduced. This gives rise to a new packing condition for the micelle with a lower surface to volume ratio than is acceptable for spherical micelles.Rods are generally accepted to be the preferred geometry6 and we shall proceed as if this were the case. However, we note that any micelle shape other than spherical allows one dimension (at least) of the micelle to be greater than the alkyl chain length of the amphiphile. The energy barrier to addition of more amphiphiles to a micelle of any given size is hence small and a polydispersion of micelles is to be expected. In fact, spherical micelles are unique in that they have a deep energy minimum corresponding to optimal micelle size.’ In the case of rod-like micelles the free energy of micellization may be viewed as the free energy of transfer of an amphiphile from aqueous solution into an infinite cylinder. Such a description gives rise to an exponential distribution6 for the micelle .size distribution function, Xi, the decay length of the function being due to end effects of the cylinder. With increased salt (whence screening), cylinder ends would become more difficult to sustain (due to their higher area per molecule) and the system’s total free energy is lowered by a lengthening of the rods.Simultaneous with this is an increase in polydispersity and the onset and subsequent increase in skewness of Xi, as is observed. The exponential decay function is a much simplified model of Xi for rods, particularly in the region of the sphere-to-rod transition. However, it adequately describes the trends observed in our experiments. As previously suggested,8 it is anticipated that the determination of size distribution parameters will lead to a fuller understanding of the energetic contributions to micelle formation and the factors controlling micelle geometry. Micelles of DTABr and DTAOH are also observed to be monodisperse in the absence of added salt (fig.2; table 1) as expected, although their aggregation numbers are strongly counterion dependent. Counterion effects on (i)& will be addressed in some detail in a forthcoming cornm~nication.~~ Micelles of Dodecyldimethylamine Oxide (DDMAO) The anomalous behaviour of micelles of the zwitterionic amphiphile DDMAO in aqueous solution have been reported several times. Addition of acid to DDMAO solutions gives rise to partial ionization of the amphiphile with the formation of DDMAOH+. Ionization of DDMAO may conveniently be described by the degree of neutralizati~n,~~ /3 = [acid]/[amphiphile]. In the fully ionized (DDMAOH+) or neutral (DDMAO) forms the micelles have aggregation numbers near 100 [ref.(24)] and behave much like isolated hard At half neutralization (/? = 0.5) their solution characteristics are chronically modified and exceedingly large micelles are formed ; farG. G. Warr, F. Grieser and D . F. Evans 1833 0.01 0.02 0.03 8 Fig. 2. Quencher-average aggregation numbers uersus 7 for (a) RuLi+-9-mA-DTABr and (b) RuL;+-g-mA-DTAOH. more than would be accounted for entropically by an ideally mixed micelle, although their c.m.c. behaviour, treated as a mixture of DDMAO and DDMAOH+, is nearly Again it is expected that gross increases in micelle aggregation number would be accompanied by a concomitant onset of polydispersity.Fig. 3 shows that this is so for DDMAO in 0.1 and 0.2 mol dm-3 NaCl solutions. The results obtained here for (i);V are in agreement with light-scattering results on this system, also shown in table 1, within experimental error. ElectrophoreticmobilityZ6 and Langmuir trough2' investigations of alkyl dimethylamine oxides suggest a weak binding of anions to the neutral amine oxide head group, However, near pH 7 the surface potential in NaCl is -25 < 'y,/mV < 25, depending upon salt concentration, whereas at full ionization the surface potential should be ca. 100 mV in NaC1.26 Anion binding should not be a sufficiently strong effect to account for the observed head group repulsion in DDMAO micelles causing small micelles. A head-group area of ca.20 A2v 2 8 1 2 9 is theoretically far too small for spherical or near spherical micelles,6 yet this is experimentally observed. Although the source is by no means clear, it may be that some attractive interaction between charged and neutral head groups (compared with interactions between like amphiphiles) causes the micelles to go over to large, polydisperse species at one-half neutralization. Mixed Micelles of Na+ and Zn2+ Dodecyl Sulphate Several studies on the micelle formation of divalent metal-ion dodecyl sulphate salts have been undertaken in the past decade. So extensive is this work that these paraffin-chain salts are all exceedingly well characterised by their c.m.c., Krafft temperatures and phase diagrams.l5? 29-32 Mixtures of counterions have also been investigated at some length.15 As the amphiphile used was DS-, the electrostatically bound fluorophore Ru(bpy)g+ was used in the quenching study of Na+/Zn2+ mixed counterion effects.Na+ and Zn2+ were selected for study because the Krafft temperature of each is sufficiently low [Krafft T of Zn(DS), is 11 OCI3l that experiments could be performed at ambient temperature (21.0 "C). 61-21834 Fluorescence Quenching in Micelles-Experiment 300 200 100 I I I I 0.01 0.02 0.03 7) Fig. 3. Quencher-average aggregation numbers for RuLi+-9-mA-DDMAO (0.05 mol dm-3) with added NaCl and HC1. symbol /3 [NaCl]/mol dm-3 T7 0.0 0.1 18 A 0.5 0.116 v 1.0 0.1 14 0.0 1.06 0 0.5 0.208 The results for (i)lW versus mole fraction of the univalent counterion Na+ are shown in fig.4. It is evident from this that some extremely ‘non-ideal’ behaviour is occurring with these mixed counter ion^.^^ Maxima in such curves for mixed amphiphiles are purported to reflect demixing of the micelle~.~~ Some onset of polydispersity is indeed observed in the mixed system of the order of 0 = 22 5, which is not present in the single counterion micelles (table 1). We are not able, however, to resolve whether this is due to two semi-discrete micelle populations or to a continuum of aggregation numbers. Discussion The potential of the fluorescence quenching technique for studying polydisperse micellar solutions remains to be ascertained fully. Clearly the transient fluorescence experiment is preferable in that there are no hidden details which may be manifested as artifacts: the resolution (or otherwise) of the two components of the fluorescence decay is unambiguous,l09 35 whereas it is always unclear in a static e~periment.~~ As discussed by Lianos and Zana,35 the fast decay rate depends upon the micelle size.We have found this also to be the case. Fig. 5 and 6 show the fast decay lifetimesG. G. Warr, F. Grieser and D . F. Evans 100 7 2 50 1835 - - 120 100 60 Fig. 4. Weight-average aggregation number, (i);V, versus mole fraction of Na+ in the Na+/Zn2+ dodecyl sulphate mixed counterion surfactant system. 0 0 d A 0.01 0.02 0.0 3 rl Fig. 5. Fast-decay lifetime of Ru(bpy):+-g-mA-SDS + salt systems, z2 (ns), versus q, as determined from a biexponential fit to the data.(A) SDS; (A) SDS+0.189 mol dm-3 NaCl; (0) SDS + 0.424 mol dm-3 NaCl; (0) SDS + 0.475 mol dm-3 NaBr. determined from a biexponential fit to the transient fluorescence from both SDS and DDMAO micellar solutions. At a given value of q the quenching is much slower in larger micelles, just as one would expect from a simple consideration of diffusion-controlled collision between solubilizates. This imposes an experimental restriction on the size of structures which can be examined. Reliable resolution of two components of a biexponential decay from single photon counting requires that they be different by at least a factor of 1 .8.37 This limit will be exceeded at large aggregation numbers for most1836 Flu0 rescen ce Quenching in Micelles- Expe r imen t 150 - 7 2 100 - 50 - 0 A 0 A O A v v .V o m 0 V 0 1 1 I J 0.01 0.02 0.03 77 Fig. 6. Fast-decay lifetime of RuL:+-9-mA-DDMAO + NaCl systems, zz (ns), uersus q, as determined from a biexponential fit to the data. fractional symbol [NaCl]/mol dm-3 neutralization v 0.1 18 1 a 0.1 16 1/2 ‘I 0.1 14 0 0 0.208 1 /2 0 1.06 1 micelles, although we note that the resolution limit will depend on the nature of the amphiphile [see ref. (36)]. Owing to the strong dependence of the fast decay rate on quencher concentration it will not be possible to determine the point at which the resolution limit is exceeded by any simple extrapolative technique. Beyond this point, at even larger aggregation numbers, collisions between probe and quencher will be so rare that no quenching will be observed at al135j38 (i.e.at such quencher concentrations as are normally used for micellar studies). In that region of micelle sizes where fast decay times are resolvable from unquenched fluorescence, the microenvironment of the micelle interface will determine the rate of quenching. In the same way as intram~lecular~~ and intermolec~lar~~ excimer formation have been correlated with ‘microviscosity ’, the rate of quenching should reflect the resistance of the micelle-water interfacial region to diffusion.18 However, intermolecular quenching is complicated by both (i) the statistical distribution of quenchers and (ii) the micelle size or confinement of the probe and quencher. The first of these effects is eliminated by extrapolation to = 0 and, as expected, the quenching rate is much lower in larger micelles (table 2).The trends observed in ks are not explained solely by confinement volume, however, and it is possible to assign some order to the relative fluidity of DTABr, DDMAO,G . G . Warr, F. Grieser and D. F. Evans 123 1.67 428 125 1.67 417 114 1.30 417 234 0.45 406 301 0.35 397 , 1837 RuLi+ Table 2. Decay rates of ruthenium probe with 9-mA quenching in various micelle systems To probe (i);y /lo7 kcl s-l /ns surfactant SDS SDS + 0.189 mol dm-3 NaCl SDS + 0.424 mol dmV3 NaCL SDS + 0.475 mol dmP3 NaBr DDMAO, /3 = 0 + 0.12 mol dm-3 NaCl DDMAO, /3 = 0 + 1.06 mol dmP3 NaCl DDMAO, /3 = 1 + 0.1 1 mol dmP3 NaCl DDMAO, /3 = 0.5 + 0.12 mol dm-3 NaCl DDMAO, /3 = 0.5 + 0.21 mol dm-3 NaCl DTABr 241 0.37 507 80 2.22 441 RuLit DDMAOH+ and, tentatively, SDS micelles of comparable size to quencher and probe diffusion, i.e.DTABr > SDS E DDMAO > DDMAOH+. (The position into which SDS fits is somewhat masked by the different fluorophore used.) All derived k, rates are comparable, however, which is not the case for micelles of significantly different amphiphile~,~~ and little significance should be attached to the differences in view of the complexities of intermolecular quenching in micelles. The probe RuL:+ is of far more general use than Ru(bpy);+ as it associates with non-ionic and cationic as well as anionic amphiphiles. This has been the major advantage of pyrene over Ru(bpy):+ until Its long unquenched lifetime of 400-500 ns (table 1) and insensitivity to 0, quenching lends itself to static fluorescence 2 3 9 3 5 7 38 on a variety of systems, and also permits a larger range of aggregation numbers to be studied than does pyrene.Further, the adequate fluorescence quantum yield and broad absorbance in the visible region of Ru(bpy);+ derivatives make them ideal fluorophores for such studies as this, in which accurate temporal decay profiles are required. The authors thank Prof. B. W. Ninham for his enthusiasm and encouragement for this project. G. G. W. acknowledges receipt of a Commonwealth Postgraduate Research Award from the Australian Government. This work was undertaken while G. G. W. was on study leave at the University of Minnesota, and thanks are due to the Erich Heymann Fund for financial assistance with travel expenses.We also gratefully acknowledge the assistance of Prof. D. D. Thomas, Dr Nagarajan and Dr S. Flom with maintenance and tuning of the single photon counting facility at the University of Minnesota.1838 Fluorescence Quenching in Micelles-Experiment References 1 F. Grieser, T. W. Healy, W. P. Hsu, J. P. Kratohvil, G. G. Warr and L. R. White, J. Colloid Interface 2 E. A. G. Aniasson, S. N. Wall, M. Almgren, H. Hoffman, J. Kielman, W. Ulbricht, R. Zana, J. Lang 3 J. Lang, C. Tondre, R. Zana, R. Bauer, H. Hoffman and W. Ulbricht, J. Phys. Chem., 1975,79, 276. 4 N. Muller, J. Phys. Chem., 1972, 76, 3017. 5 T. Nakagawa, Colloid Polym. Sci., 1974, 252, 56. 6 J. N. Israelachvili, D. J. Mitchell and B. W. Ninham, J. Chem. Soc., Faraday Trans. 2, 1976, 72, 1525.7 G. G. Warr and L. R. White, J . Chem. Soc., Faraday Trans. 2, 1985,81, 549. 8 G. G. Warr and L. R. White, J. Colloid Interface Sci., submitted for publication, 9 M. Almgren and J. E. Lofroth, J. Chem. Phys., 1982, 76, 2734. Sci., submitted for publication. and C. Tondre, J. Phys. Chem., 1976, 80, 905. 10 G. G. Warr and F. Grieser, J. Chem. Soc., Faraday Trans. 1, 1986, 82, 000. 11 J-E. Lofroth, and M. Almgren, Surfactants in Solution, K. L. Mittal and B. Lindman (Plenum Press, 12 0. Johansen, C. Kowala, A. W-H. Mau and W. H. F. Sasse, Aust. J. Chem., 1979,32, 1435. 13 G. G. Warr and F. Grieser, Chem. Phys. Lett., 1985, 116, 505. 14 Y. Talmon, D. F. Evans and B. W. Ninham, Science, 1983, 221, 1047. 15 Y. Moroi, S. Motomura, R. Matuura, J. Colloid Interface Sci., 1974, 46, 11 1.16 M. Almgren, F. Grieser and J. K. Thomas, J. Am. Chem. SOC., 1979, 101, 279. 17 M. Tachiya, J. Chem. Phys., 1982, 76, 340. 18 M. Almgren and J-E. Lofroth, J. Colloid Interface Sci., 1981, 81, 486. 19 H. F. Huisman, Proc. K. Ned. Akad. Wet., 1964, 1367, 367; 376; 388; 407. 20 J. P. Kratohvil, J. Colloid Interface Sci., 1980, 75, 271. 21 Papers in Solution Chemistry of Surfactants, ed. K. L. Mittal (Plenum Press, New York, 1979), vol. 1. 22 S. Ikeda in Surfactants in Solution, ed. K. L. Mittal and B. Lindman (Plenum Press, New York, 1984), 23 J. E. Brady, D. F. Evans, F. Grieser and G. G. Warr, unpublished results. 24 S. Ikeda, M. Tsunoda and H. Maeda, J. Colloid Interface Sci., 1979, 70, 448. 25 D. L. Chang and H. L. Rosano, in Structure Performance Relationships in Surfactants ACS Symposium 26 R. Buscall, R. B. Donaldson, R. H. Ottewill and D. Segal, Foams, Proc. Symp., 1975 ed. R. J. Akers 27 R. Buscall and R. H. Ottewill, Adv. Chem., 1975, 144, 83. 28 E. D. Goddard and H. C. Kung, J. Colloid Interface Sci., 1973, 43, 51 1. 29 I. Satake, I. Iwamatsu, S. Hosokawa and R. Matura, Bull. Chem. SOC. Jpn, 1963, 36, 204. 30 S. Miyamoto, Bull. Chem. Soc. Jpn, 1960, 33, 371. 31 Y. Moroi, T. Oyama and R. Matuura, J. Colloid Interface Sci., 1977, 60, 103. 32 A. Khan, F. Fontell and B. Lindman, Colloid Surf., 1984, 11, 401. 33 R. Nagarajan, Langmuir, 1985, 1, 331. 34 D. J. Mitchell and B. W. Ninham, J. Chem. Soc., Faraday Trans. 2, 1981, 77, 601. 35 P. Lianos and R. Zana, J. Phys. Chem., 1980,84, 3339. 36 C. J. Drummond, G. G. Warr, F. Grieser, B. W. Ninham and D. F. Evans, unpublished results. 37 A. Gafri, R. L. Molin and L. Brand, Biophys. J., 1975, 15, 263. 38 J. E. Brady, D. F. Evans, F. Grieser, G. G. Warr and B. W. Ninham, J. Am. Chem. Soc., submitted for 39 J. Emert, C. Behrens and M. Goldenberg, J. Am. Chem. Soc., 1979, 101, 771. 40 N. Turro, M. Aikawa and A. Yekta, J. Am. Chem. Soc., 1979, 101, 772. New York, 1984), vol. 1, p. 736. vol. 2. Series 253 ed. M. J. Rosen (Am. Chem. SOC., Washington D.C., 1984). (Academic Press, London, 1976), p. 73. publication. Paper 5/ 1205 ; Received 15th July, 1985
ISSN:0300-9599
DOI:10.1039/F19868201829
出版商:RSC
年代:1986
数据来源: RSC
|
|