摘要:

J. Chem. Soc., Faraday Trans. I, 1987,83, 1405-1416 Monolayer Adsorption of Non-spherical Molecules on Solid Surfaces Part 1.-Adsorption of Hard Dumb-bells on Flat Surfaces Leszek Zajtar, Jaroslaw Penar and Stefan Sokoiowski" Department of Theoretical Chemistry, Institute of Chemistry UMCS, 20031 Lublin, Nowotki 12, Poland In this paper the RAM-type perturbational theory is adapted to description of monolayer adsorption of linear molecules on flat surfaces. The theory is next applied to the study of a two-dimensional system of hard dumb-bells. The results are compared with computer simulations and with theoretical predictions based upon the scaled-particle theory. We also report the values of the third and fourth virial coefficients obtained using the Monte Carlo technique. The problem of obtaining a full statistical-mechanical description of the physical adsorption of gases on solid surfaces has attracted a considerable amount of interest.' As a natural extension of earlier studies on monoatomic systems, increasing attention is now devoted to more complicated systems containing polyatomic specie^.^-^ Of course, theoretical modelling of such systems is much more difficult, and for this reason the application of some simplified models may be helpful.One of the most widely used approximations in the theory of monolayer adsorption of atomic fluids is the assumption that the adsorbate forms a strictly two-dimensional (2D) layer. According to this approach much of the thermodynamics of the adsorption process is in fact the thermodynamics of this 2D fluid on the s ~ r f a c e .~ Physically, the assumption of two-dimensionality consists of constraining the adsorbed atoms to be on or very near to a plane parallel to the solid surface (the XY-plane), i.e. assuming the amplitude of motion perpendicular to the surface is negligible. Experimental and numerical studies of the adsorption of diatomic molecules on well characterized solid surfaces have that at sufficiently low temperatures the most probably orientation of the adsorbed molecules is with their long axes parallel to the surface. To a first approximation the adsorbed layer can be treatedg9l0 as two- dimensional, with molecules rotating and translating in the plane of the surface. It would thus be useful to start studies of such systems from the simplest 2D model.In this work we consider monolayer adsorption of linear molecules on flat surfaces by using the 2D model. Although many of the equations presented below can be applied in the case of any attractive-repulsive potential, the main part of this work is devoted to the problem of the description of a 2D system of hard dumb-bells. In particular, we compare equations of state resulting from the scaled-particle-type theory, RAM-type theory and from virial expansions. The paper is arranged as follows. In the next section we present the 2D counterpart of the RAM perturbational theory. The scaled-particle equation-of-state for a 2D system of hard dumb-bells is developed in the following section, in which we also present the results of calculations of the first three virial coefficients for the system of hard dumb-bells.The final section contains the results of numerical calculations. 14051406 Monolayer Adsorption of Non-spherical Molecules RAM Perturbational Theory for Two-dimensional Fluids We consider here a 2D system of rigid linear molecules interacting via a pair potential u(r, el, 0,) which is a function of the distance r between two selected points located within molecules and molecular orientations Oi. In the case of the site-interaction fluid model, the total intermolecular potential is the sum of the four interaction site potentials, uSs(r).l1 As in the theory of bulk three-dimensional fluids, any two-particle function can be expressed in terms of orientation variables measured with respect to molecular centres separation r,, or with respect to the interaction site separation rss.Both these options will be considered here, and the corresponding quantities referring to the centre of mass or site-centred frame will be labelled by the subscripts cc and ss, respectively. In the case of equations of general validity we will drop the subscripts. Our theoretical description of the system defined above is based upon the RAM-typell perturbational expansion. The main advantage of such an approach consists of the fact that it expands about simple spherically symmetric reference fluid and yields complete description of the angular dependent pair correlation functions. Moreover, as recent investigations of bulk molecular fluids have indicated, the RAM theory provides in many cases the best method of calculation of many thermodynamic properties of molecular fluids.l29 l3 According to the RAM approach we define the spherically symmetric reference state potential by14 uref(t) = - kT In (exp [ -u(r, el, O,)/kT]),,, B2 where (.. .) denotes an unweighted average over orientations. The free energy, F, and the two-particle background correlation function,15 z(r, B,, O,), which is related to the pair correlation function g(r, 81,62) via the equation g(r, 01782) = exp [ - u(r, 01, 0,)lkT)I z(r, 91, $2) (2) are then evaluated by using a method similar to that outlined in details in ref. (1 1). The final equations are (F- Fref)/kT = -0.5r3 gref(r12)gref(r13)gref(r23) a(r,,, 0,, 0,) s [a(r13,01,83) +a(r,3,0,, 03)I dr1 dr2 dr3(de1/2n) (de,/2n) (d03/2n) (3) and I- + [gref(r23) - l] zref(r23) Aflr23, '1, '2)) dr3 de.3/2n] * (4) In eqn (3) and (4) a(rij, 8,, ej) = 1 - exp [ - u(rij, 0i, 0j)/kT+ uref(rij)/kT] 0i, 0j) = exp [ - u(rij, Oi, Bj)/kT] - exp [ - uref(rij)/kT].The last expansion' r is the two-dimensional density and the superscript 'ref' denotes the reference system. two equations can be simplified by considering the circular harmonic where H is and (4) we and H(r, el, 0,) = X Hll exp (ill 0, + if2 0,) ( 5 ) any two-particle function, such as u, z, g etc. Inserting eqn (5) into eqn (3) (4) (7) get 11 1 2 AF = (F- Fref)/NkT = - 2x2r2 rAho(r) J l ( r ) Zref(r) s l > O %,, 0,,0,> = zref(r12) { 1 - Z Jl(r12) [exp (i0, I ) + exp (i0, l ) ] ) 1L. Eajtar, J. Penar and S . Sokolowski 1407 We note that the number of non-zero terms appearing in both eqn (6) and (7) depends upon the symmetry of the system under considerations.For example, in the case of homonuclear diatomics and the centre-of-mass reference frame, only the terms with even 1 will contribute into eqn (6) and (7). There are several theoretical routes to evaluating the 2D pressure 4 of the system. The general relation between the pressure and the full pair correlation function is 4/TkT = 1 - (T/4kT) g(r, 01, 0,) [au(r, el, 6,)/arCc] dv d6, dO2/4n2. (9) s If the site-site coordinate frame is employed, the equation of state can be expressed in terms of the circular harmonic coefficients gss, lo,gss, o1 and g,,, oo only [cf. ref. (12)]: 4 P T = 1 - ( W 2 k T ) x [du,,WdrI [ r g s , ) , O O W - d, gs,,, l o w + d,%s~, 01(r)l r dr (10) ssr where d, is the distance of the site s from the reference point.In contrast, the application of the centre-of-mass reference frame leads to an infinite sum [cf. ref. (1 5)] : where t,,, lzJ(r) are the circular harmonic coefficients of the function Another way of calculating the equation of state is by means of the expansion of f(rcc, 01,02) = M w r c c , 01, @,)larccl exp [- u(r,,, 81, &)/kTI* the Helmholtz free energy.6 We have 4/TkT = (4/TkT)ref+ r(aAF/aT) (12) where the equation of state of the reference system can be evaluated by using standard methods2 The equation of state can be also computed integrating the compressibility (1 3) equation 03 kT(ar'/a4) = 1 + rs [goo@) - 11 2 ~ d r or a(d/kT)/X = 1 - r coo@) 2nr dr i," where coo(r) is the lowest-order circular harmonic coefficient of the two-particle direct correlation function. The question concerning accuracy of the equations given above can be answered by comparing the resulting pressures with computer simulations.The investigations performed in the case of three-dimensional systems showed that for short-range repulsive fluids eqn (10) yields results which remain in excellent agreement with machine simulations, whereas in the case of Lennard-Jones site-site potentials the best results are obtained via eqn (12).121 l3 Theoretical adsorption isotherms can be derived by evaluating the chemical potential p2D of the 2D fluid. Formally, the isotherm equation is?* l6 In pK,/kT = In r + ApzD(T, T)/RT (15)1408 Monolayer Adsorption of Non-spherical Molecules where A,u2D is the part of the 2D chemical potential arising from interparticle interactions. The last quantity can be evaluated from the relationship by using the equation of state computed according to the formulae given above.computed by using the site-centred reference frame : It is obvious that the average potential energy per particle U is more conveniently r m U = 0.5r E 2n J rgssJ, oo(r) ussf(r) dr ssI 0 because the application of eqn (17) requires only the knowledge of the site-site distributions gssT,oo(r). The average energy U is related to the r-dependence of the experimental isosteric heat qst by qst(r) = qst(o) + u+ r au/ar (18) where the isosteric heat at zero coverage, qst(0), can be evaluated from the temperature dependence of the Henry constant KH.17 Analytical Equations describing Monolayer Adsorption of Hard Dumb-bells The hard dumb-bell fluid is an important model in the statistical mechanics of fluids.As numerous investigations of bulk fluids have indicated, the steep repulsive part of the interaction potential plays a central role in determining the structure and many other properties of these fluids. Moreover, the results obtained for systems interacting uia hard repulsive forces can provide insights into profitable directions for further theoretical studies,ls One of the most promising theories of bulk fluids has been based on the scaled-particle theory (STP), originally derived for spherically symmetric molecules,19* 2o which has been next extended to systems of convex hard molecules.21~22 With an appropriate value of the shape factor, the thermodynamic properties of hard, non- spherical and non-convex molecules can be approximated by those evaluated for hard, convex 22 However, it is also possible to develop the SPT for a 2D system of hard dumb-bells by using the method introduced by Ne~beda,~ for three-dimensional fluids. Because in the case of planar hard dumb-bells with elongations d > 2 b , where a is the hard-sphere diameter, a problem connected with the existence of dynamically forbidden configurations arise^,^ we limit ourselves here to the case when a / d > 24.The method presented below is fully analogous to that described in detail in ref. (23); for this reason we give here only the basic points of the theory.We consider a single scaled particle having the hard-core diameter xa and the elongation xd. The surface area S(x) and the perimeter L(x) of this particle are S(x) = (am2 (n/2 + arcsin ( d / a ) + (d/a) [ 1 - (d/0)~]4)/2 (19) and L(x) = 2ox[n/2 + arcsin (dla)]. (20) The process of adding of the scaled particle into the system can be divided into two steps: the creation of a cavity for inserting a point particle ( x = 0) and then its growth to the required size. The work connected with both these steps is given by [cf. ref. (23)]: Wo = -kTln[l-TS(l)] (21) and where W(X) = kTor dx(1 + x ) Y,,.(x) ss’ s Y,,!(X) = (1 /87t2) d0, d0Ja + d cos 0,) g,,.. ssL. Lajtar, J . Penar and S . Sokofowski 1409 gSsr is the value of gsst, oo(r) at r = B = a( 1 + x)/2 and the s-particle is scaled.Similarly [see ref. (23)] we can write a/TkT = 1 + a n r C ySs/(x = 1). (24) SSI In other words, the state behaviour of the system is fully determined by the function ySsI(x) at x = 1. For an estimation of a functional form of ySs. we use the usual thermodynamic relation where y is the ‘surface’ tension. The crux of the Nezbeda’s modification of the SPT23 lies in the assumption that the dependence of the surface tension upon the scaling parameter x for molecular fluids can be described by the equation identical with that used in the theory of spherically symmetric fluids.lg* 2o Thus, as in ref. (20), we assume here that y does not depend upon x . In this case from eqn (19), (20) and (25) we obtain (26) dW= q5dS+ydL (25) dW(x)/dx = 2xTS(l)+yL(l) = kTnaT C (1 +x) ySSJ(x).where A = L(l)y/2kTy, and y = rS(1) is the packing fraction. The constant A is evaluated comparing the derivatives dW/dx computed at x + O+ and at x + 0-. The derivative at x + O+ is calculated from eqn (26) and (27): (28) However, the evaluation of the derivative dW(x + 0-) is more complicated. For x < 0 the reversible work W(x) is d W(X + O+)/dx = 2kTyA. W(X) = - kT In (1 - Ssz+s.) ss’ where Ssz+s. is the excluded volume of a pair of two dumb-bells when one of them (s) is scaled. In the case of molecular elongations considered here the excluded volume can be approximated with accuracy better than 0.5% by Ssz+st = 0.5a2[n/2 + arcsin (d/2B) + arcsin (dx/2~) + (d/na) {n[1 - (S/2~)2]4 + [n/2 + arcsin (d/2i7) - arcsin (dx/2~)1 x[l - (xd/2~)2]t - [I - (d/2~)2]9 + arcsin (d/2~) arcsin-l (xd/2~) + arcsin (xd/2-6) arcsin-l (d/2~)] x (dx/2D) [x(d/2a)2 + ([l - (d/2a)2] [ 1 - (xd/2q2]1~]. (30) We stress that for x = 1 the last equation reproduces the value of the second planar virial coefficient :9 B, = 0.5 x Ss+s.= a2{n/2 + arcsin (d/2a) + (d/a) [ 1 - (d/20)~]t + ( d / ~ ) ~ / n > . (3 1) Making use of eqn (29) and (30) and comparing the derivatives dW(x = O+)/dx and dW(x = 0-)/dx we get ss’ For d = 0 the last equation reduces to the well known equation of state of hard discs. We also note that the complicated term in the square brackets in eqn (32) can (within 47 FAR 11410 Monolayer Adsorption of Non-spherical Molecules 2% accuracy) be approximated by i(d/a),.With the last approximation, the adsorption isotherm equation becomes ln(pK,/kT) = In r+([3+g(d/o)2]y-[2+9(d/o)2] y2}(1 -y)-2-ln(1 -y). (33) We now consider the virial expansion for equation of state: 4/TkT = 1 + B, r + B, T2 + B, r3 + . . , = 1 + B,* y + 8: y2 + B: y3 + . . . (34) where Ba = Bi/[S(1)li-l. The values of the second planar virial coefficient can be computed from eqn (31), which with a fair agreement can be approximated by B, = n / 2 + 2(d/o) + 0.25(d/0)' B,* = 2+&(d/0)~. ( 3 5 ) The third and fourth virial coefficients are given by and Here the bonds imply Mayer f-functions. For the system under consideration the integrals (36) and (37) correspond to the volume fraction of configurational space for which all appropriate bonds are established, so that the Monte Carlo method24 provides a suitable technique for their evaluation.Ca. 5 x lo5 trial configurations are generated to estimate B, and B4 to within 2% error, and the results of computations are summarized in table 1 . The approximating formulae for B, and B4 are the following : B: = 3.128+(d/0)+0.8(d/o)~ (38) and BZ = 4.262 + (d/o). (39) The numerical values of the third and fourth virial coefficients can be next used to generate an analytical formula for the equation of state. Different theoretical techniques, such as the Padk appr~ximant~~ and Carnahan-Starling26 method as extended by Ne~beda,,~ can be used for this purpose. The simplest method for improving the equation of state consists of adding to eqn (32) terms such that the resulting equation would reproduce correctly at least the first three virial coefficients [cf.ref. (27)]. We obtain Z = 4/TkT = ( 1 +al y+a2y2+a3y3)(l - Y ) - ~ . (40) The corresponding adsorption isotherm equation, obtained from eqn ( 1 5 ) , has the following form : ln(pK,/kT) = l n r + Z + ( - l + a , + a , ) l n ( l - y ) + a f l + (U + a, +a2 + a,) ( 1 --y)-' - ( 1 +a, +a, + a,) (41) where a, = &(d/o)2, a, = 0.125+0.1(d/a)L. Lajtar, J . Penar and S. Sokolowski 141 1 Table 1. The third and fourth two-dimensional virial coefficients for hard dumb-bells 0 0.1 0.2 0.3 0.4 0.5 0.6 0.7 0.8 0.9 1 .o 3.128 3.25 3.38 3.52 3.68 3.85 4.02 4.24 4.50 4.82 5.27 4.262 4.36 4.46 4.56 4.66 4.81 4.89 5.02 5.23 5.44 5.90 and For d = 0 eqn (40) reduces to the equation of state for hard discs developed by Henders~n.~' a3 = 0.8(d/o) - (d/a)2.Numerical Results and Discussion Computational Details Application of the RAM perturbational theory outlined in the second section requires a knowledge of the reference-state distribution. In studies of bulk molecular fluids1l-l3? l8 this function was often evaluated by performing additional reference-system simulations or identifying gref(r) with the simulated lowest-order spherical harmonic coefficient of the full two-particle correlation function. In this work, however, we have decided to use the simplest method of evaluation of the reference state distribution gref(r), which consists of the solution of the Percus-Yevick integral equation : Zref(r12) = I + Jdr3~ef(rl3) Zref(r13) [ ( p f ( r 2 3 ) + 1) Tref(r23) - I].(42) We have solved eqn (42) for both: site-centred and centre-of-mass reference frames by using the method described by Lado.28 In the case of the centre-of-mass reference frame perturbation calculation for first-order RAM theory have been carried out according to eqn (2), (7) and (8), and the circular harmonic coefficients have next been determined by integrating the equation : 2n 2n gcc, lj(r) = (2n)-'J J dB1 dB2 exP [-ucc(r, 01,BdIkTI Tcc(r, 01,02> 0 0 ' x [cos lol cosj0, - sin 10, sinj0J. (43) In the case of the site-centred frame, however, only the zeroth-order perturbational calculations were carried out, so that we set gss, = e s s , I j W Gf(r) e s s , j d r ) = 610 djo +fL, jdr) where denotes the circular harmonic coefficient of the Boltzmann function.We stress that in the case considered here the equation of state (10) becomes 4lTkT = 1 + 2nO2gss, O O ( 4 - dagss, lO(4l = 1 + 2nrb2es,, o o ( 4 -does,, 1o(4lYref(4* (44) 47-21412 Monolayer Adsorption of Non-spherical Molecules 1 2 3 d o Fig. 1. A comparison of the circular harmonic coefficients gll l,(r) obtained for the 2D system of hard dumb-bells with d/a = 0.4 and the packing fraction y = 0.587 (To2 = 0.5) by using the RAM theory (lines) and the Monte Carlo simulations (points). The numbers in the parenthesis denote the values of 1, and 1,. The symbols RO and R1 denote the results of the zeroth- and first-order RAM theory, and the letter M indicates that the perturbational calculations were performed assuming that the reference-state distribution function was identified with the lowest-order Monte Carlo coefficient of the pair distribution function goo(r).Fig. 2. As in fig. 1, but for the system with d/o = 0.4 and at y = 0.352 (To2 = 0.3) and at y = 0.705 (To2 = 0.6).L. Lajtar, J . Penar and S. Sokoiowski 1413 t Fig. 3. As in fig. 1, but for the system with d/a = 0.6 at y = 0.606 (raz = 0.45). In order to compare the results of theoretical predictions we have also performed Monte Carlo simulations by using an algorithm essentially the same as that described in ref. (29). Two model 20 fluids of hard dumb-bells with d/a = 0.4 and d/a = 0.6 were investigated. In the case of the first system our Monte Carlo simulations were performed for the following packing fractions: y = 0.352, y = 0.578 and y = 0.705, whereas in the case of the second system we investigated the following state points: y = 0.269, y = 0.471 and y = 0.606.The basic cell contained 128 particles, and the usual boundary conditions were applied. The parameter determining the rate of acceptance of proposed moves was adjusted during the course of the simulations to maintain, approximately, the 0.5 rate of acceptance. We generated ca. lo6 steps to bring the system into a stable homogeneous and disordered phase. The histogram used to evaluate the required functions [g,,, lj(r) for l,j < 4, gss, oo(r) and gss, lo(r)] was then obtained from the next 1.28 x 10" steps, sampling after every 256 steps. Discussion In fig. 1-3 we present the average centre-of-mass distributions, the average site-site distributions and the selected circular harmonic coefficients of the two-particle correla- tion function.The behaviour of all these functions is similar to that observed in the case of bulk three-dimensional fluids. The contact values of gss, oo(r) decrease with decreasing density. Next, all these functions exhibit slope discontinuities (cusps) at the separation d+a. The existence of these discontinuities can be explained in terms of molecular geometry [cf. ref. (1 1) and (12)]. The density seems to affect the sharpness of the cusps, which become less evident at higher densities. The agreement between theoretical predictions and computer simulations is only fair, and is better in the case of the functions computed using the centre-of-mass reference frame.The observed differences between results of perturbational calculations and computer experiments can be at least partially explained by the method of evaluation of the reference state distribution. In fact, if the simulated centre-of-mass distributions, as well as the site-site distributions, are next used instead of gref(r) in perturbational calculations, the accuracy of the predicted circular harmonic coefficients improves1414 Monolayer Adsorption of Non-spherical Molecules Y 0.8 0.6 0.4 0.2 - 10 Z - 20 I 1 1 I I 1 I I I 0 0.2 0.4 0.6 0.8 Y Fig. 4. Compressibility factors 2 = d/TkT us. the packing fraction y = TS(1) [cf. eqn (19)] for the systems of hard dumb-bells with d/o = 0.4 and d/o = 0.6. 0, Results of Monte Carlo simulations; (-) results of the SPT eqn (32); (--) computed according to the improved SPT eqn (40); (----) calculated integrating eqn (13) with the first-order RAM distribution function gcc, oo(r); (.. . . . - ) determined from the virial expansion (34). slightly ( c - fig. l), but this effect is smaller than one would expect on the basis of similar calculations performed for three-dimensional fluids. 11-13* l8 It is well known that the three-dimensional RAM theory is particularly accurate if one considers its reduced version.ll9 1 3 9 l8 The two-dimensional reduced circular harmonic In the case of three-dimensional hard dumb-bell fluids the centre-of-mass reduced spherical-harmonic coefficients approach exactly and easily calculated values in the limit of the closest approach, rc.These limiting values do not depend upon density and are equal to the contact values of the corresponding reduced spherical harmonic coefficients of the Boltzmann function, 11-139 l8 Consequently, the three-dimensional zeroth-order RAM theory of hard dumb-bells predicts exactly the contact values of the reduced spherical coefficients. For the 2D fluid of hard dumb-bells this is no longer true, because the closest-contact value rc can be reached for several orientations. We have0.6 Y 0.4 0.2 L. tajtar, J . Penar and S. Sokolowski In pK,lkT 10 5 0 -5 I I 1 1415 7 I 10 0.4 Y 0.6 In pKHlkT Fig. 5. Adsorption isotherms y = TS( 1) us. In pK,/kTfor the 2D systems of hard dumb-bells with d/o = 0.4 and d/a = 0.6. The lines have the same meaning as in fig. 4. where z:,, jl(r) = T,,, jl(r)/Tcc, ,,(r) and the contact values the reduced circular harmonic coefficients of the Boltzmann function exp [ -u(r, 0,, O,)/kT] are given by and re = 0.Thus, if the background correlation function z(r, O,, 0,) is independent (or nearly independent) on orientations, the exact eqn (46) reduces to the zeroth-order RAM In fig. 4 we show the compressibility factors 2 = #/TkT, and in fig. 5 we display the theoretical adsorption isotherms computed for two molecular elongations by using different theoretical methods. The integration of the compressibility eqn (1 3) with the centre-of-mass correlation function gee, ,,(r) determined according to the first-order RAM theory leads to the highest values of the 2D pressure. The improved STP eqn (40) gives the pressures higher than those computed according to the ordinary SPT eqn (32), whereas eqn (44) leads to the lowest values of 2.Moreover, we can state that the virial expansion (34) diverges at the packing fractions > 0.2. We stress, that if the contact values of g,",~o,(a) obtained from simulations and the contact values of gss, ,,(a) approximated by g,M?,,(a) e,*,! ,,(a) are substituted into eqn (44), the resulting pressures are close to those obtained directly from simulations. The planar hard dumb-bell fluid can be a useful model in possible two-step perturba- tional scheme of calculations, according to which [cf. ref. (30)] the properties of a Lennard-Jones diatomic fluid are computed by first expanding about a non-spherical reference system and next by using the RAM or SPT theory to determine the properties of this non-spherical reference fluid.Of course, questions concerning the accuracy of the theories outlined above require more complex computer simulation studies. Such investigations and attempts to use the theory reported here in the description of real adsorption systems will be presented in later papers. theory result g;c, jl(rc> = e:c, j l ( r c ) .1416 Monolayer Adsorption of Non-spherical Molecules References 1 0. Nicholson and N. G. Parsonage, Computer Simulation and the Statistical Mechanics of Adsorption 2 S. M. Thompson, K. E. Gubbins, D. E. Sullivan and C. G. Gray, Mol. Phys., 1984, 51, 21. 3 S. Sokolowski and W. A. Steele, Langmuir, 1985, 1, 180. 4 K. M. Wojciechowski, A. Bronka and M.Parrinello, Mol. Phys., 1984, 53, 1541. 5 W. A. Steele, J. Chem. Phys., 1976, 65, 5256. 6 J. Talbot, D. J. Tildesley and W. A. Steele, Mol. Phys., 1984, 51, 1331. 7 J. Talbot, D. J. Tildesley and W. A. Steele, Faraday Discuss. Chem. Soc., 1985, 80, 191. 8 C. Peters and M. L. Klein, Mol. Phys., 1985, 54, 895. 9 J. S. Rowlinson, J. Talbot and D. J. Tildesley, Mol. Phys., 1985, 54, 1065. (Academic Press, New York, 1982). 10 J. M. Philips and M. D. Hammerbacher, Phys. Rev. B, 1984, 29, 5859. 11 W. R. Smith and I. Nezbeda, Adv. Chem. Ser., 1983, 209, 235. 12 I. Nezbeda and W. R. Smith, J. Chem. Phys., 1981,75,4060. 13 I . Nezbeda and W. R. Smith, J. Chem. Phys., 1984,81, 935. 14 W. R. Smith and I. Nezbeda, Faraday Discuss. Chem. SOC., 1978,66, 130. 15 W. A. Steele, Faraday Discuss. Chem. Soc., 1978, 66, 138. 16 S. Sokolowski, Adv. Colloid Interface Sci., 1981, 15, 71. 17 W. A. Steele, Interactions of Gases with Solid Surfaces (Pergamon Press, Oxford, 1974). 18 I. W. Melnyk and W. R. Smith, Mol. Phys., 1980, 40, 317. 19 H. Reiss, H. L. Frisch and J. L. Lebowitz, J. Chem. Phys., 1959, 31, 369. 20 H. Helfand, H. L. Frisch and J. L. Lebowitz, J. Chen;. Phys., 1961,34, 1037. 21 C. G. Gray and K. E. Gubbins, Theory of Molecular Fluids (Oxford University Press, Oxford, 1984), 22 M. Rigby, Mol. Phys., 1976, 32, 575. 23 I. Nezbeda, Mol. Phys., 1977, 33, 1287. 24 I. Nezbeda, Chem. Phys. Lett., 1976,41, 55. 25 G. A. Barker and J. L. Gammel, The Pad& Approximant in Theoretical Physics (Academic Press, 26 N. F. Carnahan and K. E. Starling, J. Chem. Phys., 1969, 51, 635. 27 D. Henderson, Mol. Phys., 1975, 30, 1581. 28 F. Lado, J. Chem. Phys., 1968,49, 3092. 29 W. B. Street and D. J. Tindesley, Proc. R. Soc. London, Ser. A , 1976, 348,485. 30 J. Fischer, J. Chem. Phys., 1980,72, 5371. vol. 1. New York, 1970). Paper 61940; Received 15th May, 1986

ISSN:0300-9599    

DOI:10.1039/F19878301405

出版商:RSC    

年代:1987

数据来源: RSC

摘要:

J. Chem. SOC., Furuduy Truns. 1, 1987,83, 1417-1425 A Study of the Adsorption Sites on Thoria by Scanning Transmission Electron Microscopy and Fourier-transform Infrared Spectroscopy Adsorption and Desorption of Water and Methanol Xavier Montagne, John Lynch and Edouard Freund Institut Frangais du Pitrole, B.P. 31 1 , 92506 Rued Malmaison, France Jean Lamotte and Jean-Claude Lavalley" Laboratoire de Spectrochimie, UA 04 414, I.S.M.Ra, Uniuersitk de Caen, 14032 Caen Ckdex, France A study of the adsorption sites of thoria has been made using scanning transmission electronic microscopy (STEM) and Fourier-transform infrared spectroscopy (FTIR). The STEM study showed that the thoria surface was composed mainly of (1 10) (21 1) and (1 11) faces in about equal proportions. The infrared study of activated Tho, at 873 K indicated the existence of two remaining hydroxyl groups at 3660 and 3510 m-l.With water adsorption, a third one appeared at 3745cm-' which was thermally less stable. Methanol adsorption gave rise to three kinds of species: (i) one reversibly adsorbed at room temperature and linked by hydrogen bonds in which thoria acts as a proton receiver; (ii) two methoxy species characterized by v(C0) bands at 1127 cm-l (species I) and 1060 cm-l (species 11). Species I desorption bands led to methanol whilst species 11, much more thermally stable, gave methanol, dimethyl ether and carbon monoxide. Species I, which corresponds to the hydroxyl species at 3745 cm-l, are of the Th-OMe (Th-OH) type, and are formed on the (1 10) faces. Species 11, which correspond to hydroxyl species at 3660cm-l, are of the '0-Me / Th Th Th 0-H) type, and are formed on the (21 1) faces.The third hydroxyl species, characterized by the v(0H) band at 3510 cm-l, is probably due to type appearing on (1 1 1) faces. The stability species of the and reactivity of the methoxy species are shown to depend on their structure, and therefore on the local arrangement of the surface. \ ( / Th OH Th' i h 'Th It has long been recognized that the surface of a catalyst is not homogeneous, but is composed of various crystal planes and lattice imperfections. Structural differences between exposed crystalline faces involve different active centres, thus explaining that structure sensitivity occurs on oxide surfaces1 as well as on metals.2 The present work deals with the adsorption sites of thoria, an oxide active for alcohol synthesis3 and which develops interesting properties in the process termed isosynthe~is.~ and the adsorption of H, and CO over we report here results on the determination of the most frequently Following our study of the acid-base properties of 14171418 Adsorption Sites on Thoria exposed planes of the microcrystallites of this oxide.Two techniques have been mainly used : scanning transmission electronic microscopy (STEM) and Fourier-transform infrared spectroscopy (FTIR). The first permits the determination of the relative importance of the different crystalline planes, while the second provides useful inform- ation on the nature of the adsorbed species: either hydroxyl groups (their wavenumber depending directly on their surroundings, as has been shown on alumina)', or species given by probe molecules.Among the various possible probes, water and methanol have been chosen, and a study of the adsorption of the former completes those for hydroxyl groups. For methanol, interesting results have already been obtained on various oxides,s the structural arrangement of the species formed depending on the local surface organization. Experimental Materials The Tho, (Rh6ne Poulenc) used had a specific area of 120 m2 g-l, a low pore volume (0.1 cm3 g-l) and a pore diameter of ca.3.5 nm. It contained 0.7% N and 0.2% C, mostly as nitrates and carbonates. As it is a basic ~ x i d e , ~ these species are thermally stable. Methods STEM A VG HB5 STEM equipped with a high-sensitivity TV camera to observe micro- diffraction patternslO was used to study the local structure of the thoria.The surface giving rise to the diffraction pattern may be defined by the area scanned on the specimen, the limit being the size of the electron probe when stationary. The probe size and the angular size of the diffraction discs are determined by the beam convergence,ll which in this case was typically 2 mrad, giving well separated discs whilst maintaining a high spatial resolution (of ca. 1 nm). Diffraction patterns from individual grains attached to the carbon support film were obtained in the stationary-probe mode. These reveal the orientation of the grains with respect to the (flat) support. Our interpretation is based on the hypothesis that the grains, whatever their form, fall with their largest face parallel to the support film and thus perpendicular to the electron beam (this has been confirmed for alumina platelets in a similar study).l2 Indexation of the diffraction patterns thus gives the orientation of preferentially exposed surface planes. Spherical particles would give rise to a random zone-axis distribution, weighted according to the number of symmetry eq~iva1ents.l~ In attributing a surface plane to the corresponding zone axis we do not take into consideration the possibility of surface reconstruction. Weak-beam dark-field images show that the crystallites have smooth surfaces down to a scale of ca. 1 nm, but cannot give more precise details about the state of the surface.FTIR The infrared spectra were recorded at room temperature on a Nicolet MX-1 instrument (Caen) or a Digilab FTS-I5E instrument (IFP). Self-supporting pressed discs (ca. 25 mg) were activated by heating under vacuum at 473 K for 12 h and then at 873 K for 4 h. After such a treatment only v(0H) bands, due to residual hydroxyl groups, and weak bands at 1480 and 1330 cm-l, due to residual carbonate and nitrate species, were apparent .1418 Adsorption Sites on Thoria exposed planes of the microcrystallites of this oxide. Two techniques have been mainly used : scanning transmission electronic microscopy (STEM) and Fourier-transform infrared spectroscopy (FTIR). The first permits the determination of the relative importance of the different crystalline planes, while the second provides useful inform- ation on the nature of the adsorbed species: either hydroxyl groups (their wavenumber depending directly on their surroundings, as has been shown on alumina)', or species given by probe molecules.Among the various possible probes, water and methanol have been chosen, and a study of the adsorption of the former completes those for hydroxyl groups. For methanol, interesting results have already been obtained on various oxides,s the structural arrangement of the species formed depending on the local surface organization. Experimental Materials The Tho, (Rh6ne Poulenc) used had a specific area of 120 m2 g-l, a low pore volume (0.1 cm3 g-l) and a pore diameter of ca.3.5 nm. It contained 0.7% N and 0.2% C, mostly as nitrates and carbonates.As it is a basic ~ x i d e , ~ these species are thermally stable. Methods STEM A VG HB5 STEM equipped with a high-sensitivity TV camera to observe micro- diffraction patternslO was used to study the local structure of the thoria. The surface giving rise to the diffraction pattern may be defined by the area scanned on the specimen, the limit being the size of the electron probe when stationary. The probe size and the angular size of the diffraction discs are determined by the beam convergence,ll which in this case was typically 2 mrad, giving well separated discs whilst maintaining a high spatial resolution (of ca. 1 nm). Diffraction patterns from individual grains attached to the carbon support film were obtained in the stationary-probe mode.These reveal the orientation of the grains with respect to the (flat) support. Our interpretation is based on the hypothesis that the grains, whatever their form, fall with their largest face parallel to the support film and thus perpendicular to the electron beam (this has been confirmed for alumina platelets in a similar study). l2 Indexation of the diffraction patterns thus gives the orientation of preferentially exposed surface planes. Spherical particles would give rise to a random zone-axis distribution, weighted according to the number of symmetry eq~iva1ents.l~ In attributing a surface plane to the corresponding zone axis we do not take into consideration the possibility of surface reconstruction. Weak-beam dark-field images show that the crystallites have smooth surfaces down to a scale of ca.1 nm, but cannot give more precise details about the state of the surface. FTIR The infrared spectra were recorded at room temperature on a Nicolet MX-1 instrument (Caen) or a Digilab FTS-I5E instrument (IFP). Self-supporting pressed discs (ca. 25 mg) were activated by heating under vacuum at 473 K for 12 h and then at 873 K for 4 h. After such a treatment only v(0H) bands, due to residual hydroxyl groups, and weak bands at 1480 and 1330 cm-l, due to residual carbonate and nitrate species, were apparent .J . Chem. SOC. Faraduy Trans. I , Vol. 83, purt 5 Plate 2 Plate 2. Indexed diffraction patterns from individual thoria crystals lying flat on the carbon support film. X. Montagne e t (11.X . Montagne et al.1419 A I I I 1 1 I I 1 1 4000 3500 3000 2500 2000 1500 1000 500 B 4000 3500 3200 2800 2400 wavenum berlcm-' wavenumberlcm - ' Fig. 1. (A) Infrared spectra of species given by water adsowtion after addition of (a) 0.14, (b) 0.28, (c) 0.83 and (d) 1.5 molecule nm-2. (B) Infrared spectra of species given by desorption after evacuation at (e) 293, (f) 373, (g) 473 and (h) 673 K. T.P.D. of Methanol On Tho, samples activated at 923 K under flowing helium, methanol was adsorbed at room temperature until saturation. Desorption was then carried out by heating at a rate of 5 K min-l under flowing helium. The effluents were analysed by infrared spectroscopy using a gas cell, by mass spectrometry (Kratos MS 80 spectrometer) or by gas chromatography, using a 3 m column filled with a 50/50 mixture of chromosorb 103 and chromosorb 104 (Girdel N30 apparatus). Thermogravimetric experiments (with a Mettler MT 1 microbalance) were also carried out using the same process.Results STEM The bright-field image of the specimen [plate 1 (a)] shows agglomerations of small grains of average individual diameter 10 nm. Electron diffraction of the agglomerates [plate 1 (b) and (c)] at different spatial resolutions shows them to be well organized, the grain orientation varying little (k 10" from the average) within a group. This is confirmed by dark-field imaging, in which grains with common directions appear together as bright regions [plate 1 (d) and (e) show dark-field images for two directions]. Diffraction patterns from individual grains could easily be obtained (plate 2), the thoria being well crystallized (no defect was detected either by diffraction or by high-resolution imaging) and constituting a strong scatterer.The observation of a large number of grains (more than fifty) showed roughly equal proportions of (1 lo), (1 1 1) and (21 1) faces, these three accounting for 75% of the patterns recorded, with much lower proportions of (100) and (310) orientations; (123) and higher-index zone axes were very rarely seen.1420 Adsorption Sites on Thoria FTIR After the activation process, spectra show two v(0H) bands at 3660 and 3510 cm-l, with a weak shoulder at 3640 cm-l. Water Adsorption and Desorption The adsorption of a small amount of water (0.12 molecule nrn-,) induces the appearance of two sharp bands at 3745 and 3670 cm-l of equal intensity and another of low intensity at 3695 cm-l.The spectrum [fig. 1 (A)] also presents a very broad band centred at ca. 3500 cm-l and two others at 1635 (weak) and 670 cm-l. Addition of supplementary doses of water increases the intensity of all these bands, except for that at 3695 cm-l, which remains weak, while a new broad band appears at ca. 3000 cm-l. The intensity of the two bands at 3000 and 1635 cm-l d(H0H) greatly decreases by evacuation at room temperature, showing that they correspond to weakly adsorbed water. This water may be bound by a hydrogen bond to the hydroxyl groups giving rise to the two sharp bands at 3745 and 3670 cm-l, since their intensity increases as water is evacuated. Evacuation at increasing temperature (fig.1 B) decreases the intensity of all the broad bands. At 373 K, the 3100 cm-l band totally disappears while the 1635 and 3695 cm-l ones are still detected. At 473 K, the molecularly adsorbed water is completely eliminated since the 1635 cm-1 band is not apparent. There remain only two sharp bands at 3745 and 3670 cm-l and another at 3520 cm-l, superimposed on the broad band centred at ca. 3500 cm-l. The 670 cm-l band is still very intense. At 673 K, the broad band at 3500 cm-1 disappears, bringing the 3525 cm-l band into evidence. The sharp band at 3745cm-l is then much less intense. At 873 K, only the bands observed on the spectrum of the starting activated thoria are present. The adsorption of D,O leads to similar results [v(OD) at 2760, 2706 and 2606 cm-l; 6(DOD) at 1205~m-~]. The substitution of H by D does not produce any band equivalent to that of 670 cm-l in the wavenumber range studied (4000-600 cm-l).These experiments have been completed by temperature-programmed desorption measurements. Activation of thoria by increasing the temperature shows that water desorption takes place discontinuously. An initial large desorption at 423 K is followed by a second at 650 K; a slow desorption is then noted at higher temperatures. These results show the presence of at least three kinds of species. (i) Weakly adsorbed water, disappearing by evacuation up to 473 K and characterized by the 1635 cm-l band. The sharp band at 3695 cm-l is also due to this species and corresponds to the v(0H) vibration of H,O species with a free OH group.(ii) OH species bound by weak hydrogen bonds and characterized by the broad band at ca. 3500 cm-l. They progressively desorb by increasing temperature up to 673 K. (iii) Free OH groups characterized by the sharp bands at 3745,3670,3640 (shoulder) and 3525 cm-l. The band at the higher wavenumber is due to a relatively unstable free hydroxyl, as it is hardly detectable at 673 K. The 670 cm-l band, still very intense after heating at 673 K, may be attributed to the 6(OH) vibration of hydroxyl groups responsible for the 3520 or 3670 cm-l absorption. Methanol Adsorption and Desorption When 0.1 molecule nm-, of MeOH is introduced, a very broad band between 3550 and 2440 cm-l as well as bands at 2920 (sharp), 2900 (weak), 2865 (broad), 2800 (sharp and intense), 1445 (weak), 1127 (very sharp and intense), 1050 (broad), 1015 (very weak), 700 and 670 cm-l (broad) are formed.By increasing the amount of adsorbed CH,OH, an increase in the intensities of all bands is observed, except that of the 1050 cm-l band which becomes a shoulder of a new band, broad and intense at 1060 cm-l (fig. 2). Up to a value of 2.8 methanol molecule nrnp2, no gaseous phase is detectable in the cell.X . Montagne et al. 1421 1.100 P: 0.800 s e 9 -2 0.500 0.200 - 0 . l O O I I , , , I I I I I I I . 1 1 4000 3500 3000 2500 2000 1500 1000 500 wavenumber/cm-' Fig. 2. Infrared spectra of species given by methanol adsorption on thoria (a) after addition of 3.2 molecule nm-2; (b) after evacuation at 650 K. 373 473 573 673 773 TIK Fig.3. Study by mass spectrometry of amounts of methanol (-) and dimethyl ether (---) formed at increasing temperatures from methoxy species on thoria. Addition of further doses of CH,OH increases a little more the intensity of all the bands and induces the appearance of others at 3550 (broad) , 3100 and 2650 (very broad) and 1465 cm-l. They can be assigned to reversibly adsorbed species. The same results have been obtained when adsorbing MeOH on a Tho, pellet exchanged with H,180 as in ref. (14) and then evacuated at 873 K. Outgassing at increasing temperatures up to 673 K causes a decrease in the intensity of all bands between 3000 and 2800 cm-l except that of the 2856 cm-l, which vanishes totally. We also note a significant decrease in the intensity of the 1127 cm-l band, whereas that of the 106Ocm-l one is scarcely altered.The broad band between 3500 and 2400 cm-l has completely disappeared. Between 473 and 573 K some very weak bands at 1565, 1375 and 1363 cm-l appear temporarily, indicating the formation of a small amount of formate species. These results allow us to conclude that at least three types of species are formed. One, characterized by a band at 1127 cm-l, desarbs between 373 and 623 K (species I); the second, characterized by the 1060 cm-l band, is stable at 673 K (species 11). The third1422 Adsorption Sites on Thoria 0 thorium 0 oxygen 111 0 0 0 0 0 Fig. 4. Arrangement of the Th4+ and 02- ions on the preferentially exposed faces of thoria. is made up of the reversibly adsorbed species (species 111).A fourth very similar to species 11, may be distinguished from the latter as its characteristic band is at 1050 cm-l. It only appears when the first dose of methanol is introduced (species 11’). The nature of the desorbed products has been studied by different methods: mass spectrometry, infrared spectroscopy and gas chromatography. They all clearly show (fig. 3) the occurrence of three desorption steps: (i) below 373 K there is a large methanol release, corresponding to species 111, (ii) from 500 to 573 K another desorption of methanol takes place (species I); (iii) around 673 K a more complicated desorption process occurs (species I1 and 11’), giving rise to the desorption of methanol, dimethyl ether and carbon monoxide. Thennogravimetric measurements specify the amounts of the different species.The total amount of methanol adsorbed at room temperature is 4.2 molecule nm-2, the irreversible quantity being 3.75 molecule nm-2. From a plot of the weight variation us. temperature, one can estimate the amount of species I11 (1.6 molecule nm-2), species I (1.0 molecule nmV2) and species I1 and 11’ (1.15 molecule nm-2). Discussion The thoria used in the present study has a surface area (1 20 m2 g-l) very similar to that prepared by Maj et aZ.3 and claimed as one of the largest reported in the literature. X-Ray diffraction has confirmed that it is of the fluorite type, the thorium being positioned according to an f.c.c. lattice (cell parameter 0.56 nm). The STEM study shows that the (1 lo), (21 1) and (1 11) faces are the most frequently exposed.The (loo), (310) and (123) are less exposed and will not be considered in the present discussion. The arrangement of the Th4+ and 02- ions on the three preferentially exposed faces is shown in fig. 4. The (1 10) face contains Th4+ ions with two anionic vacancies in the neighbourhood of the 02- ions, which are coordinatively unsaturated (cus). The (21 1) face is formed by rows of cus Th4+ ions between two rows of cus 02- ions. The (1 1 1) face only contains cus Th4+ ions, the oxygen ions, slightly below, being coordinatively saturated. Let us now discuss the results obtained by infrared spectroscopy, beginning with thoseX . Montagne et al. 1423 relevant to methanol adsorption. Species I and I1 (11’) resulting from methanol chemisorption on thoria are of the methoxy type: they are characterized by a v,(CH,) vibration at 2805 cm-l, i.e.at a low wavenumber, and a v(C0) vibration (species I, 1127 cm-l; species 11, 1060 cm-l; species 11’, 1050 cm-l) the wavenumber of which is higher than that observed with gaseous MeOH. Such features are typical of dissociative adsorption leading to methoxy groups.15 It is important to note that there corresponds to species I a broad v(0H) band due to the dissociative adsorption. Its low wavenumber (between 3550 and 2400 cm-l) favours the following type of structure: Me I Th 0 type 1 0-H. I The reversible species (type 111) is bound to the surface by a hydrogen bond. Considering the high basicity of thoria5, the most likely species are: Me 0 I H type I11 0 This would explain the high wavenumber of the d(OH) vibration (1465 cm-l) and the presence of the pair of broad bands at 3100 and 2650 cm-l [v(OH) in Fermi resonance with the first overtone 2S(OH)].Our infrared results on thoria are similar to those observed by Tench et aZ.ls on MgO. On such an oxide, the authorslG considered that the most stable methoxy species are formed by the dissociation of the C-0 bond: H I Me MeOH \ P ---Mg-O-Mg-O- --Mg-O-Mg-O- In a later study, Tsyganenko et aL8 mentioned this result and explained the difference between the methoxy species I and I1 by the coordination number (I or V) of the surface oxygens. Our result obtained from MeOH adsorption on Tho, exchanged with H,180 does not support such a statement: considering that the surface oxygen has been exchanged with l80 (as confirmed by the previous study on HCHO adsorptionf4, we expect from the breaking of the C-0 bond the formation of Me-180 species characterized by a v(C0) band14 at 1086 cm-l (type I) or 1017 cm-l (type 11).Since no band can be detected at such wavenumbers, we must conclude that type I1 methoxy species are formed in a similar way as type I, i.e. by dissociation of the 0-H bond. Dissociation of methanol needs pairs of sites formed by coordinatively unsaturated ions Th4+ and 02-. Only the (1 10) and (21 1) faces of the three preferentially exposed faces present such pairs of sites. On the (1 10) face the orientation of the free valences of Th4+ ions and the position of the 02- ions allow only the formation of Me0 species bound to a single thorium ion.The released hydrogen adsorbed on an adjacent 0,- ion creates a hydroxyl group able to form a hydrogen bond typical of species I.1424 Adsorption Sites on Thoria allows the formation of bridged methoxy groups: On the other hand, the orientation of the free valences of Th4+ ions on the (21 1) faces Me species I1 Such species would therefore result from the surface ion arrangement on the (21 1) face. Note that the number of species I and I1 obtained by thermogravimetric measurements is in agreement with the STEM results showing the almost equal abundance of faces (1 10) and (21 1). The results obtained from water adsorption complete those relative to the residual hydroxyl groups on the activated samples and show the occurrence of a thermally unstable hydroxyl group at 3745 cm-l.This corresponds to species I given by methanol and therefore can be assigned to the following Th-0-H group: H(D) C ) species I 3745 (2760) cm-l Th We assign to species I1 the hydroxyl group at 3670 cm-l, which is much more stable: H(D) species I1 3670 (2706) cm-l I Th /O\ Th Such an assignment is in agreement with Knozinger’s considerations on a l ~ m i n a : ~ the v(0H) frequency of bridged hydroxyl groups is lower than that of monodentate ones. The third OH group in the spectrum of Tho, has a very low wavenumber (3510 cm-l). It may be considered as due to OH groups slightly perturbed by hydrogen bonding, as is the case for that on ZnO at 3444cm-l.17 However, such an assignment can be discarded, since activation of thoria at high temperature does not perturb the 3510 cm-l band.Another possibility is to consider that these hydroxyls are located in the bulk, as for silica.(18) They would then be unperturbed by surface adsorption processes and not exchangeable with D20. This is not in fact the case, leading us to propose a third possibility, that these hydroxyls would be arranged as follows : H I A\ Th Th Th which is quite possible on the (1 1 1) surface, one of the three preferentially exposed ones. H,O adsorption and desorption at 673 K leads to the 3745 and 3670 cm-l bands and also a broad band near 3550cm-l. These certainly arise from water dissociation on Th-0 couples such as those presented by the (1 10) or (21 1) faces. The OH- formed is adsorbed on a Th4+(cus) atom, leading to OH species I or I1 according to the type ofX .Montagne et al. 1425 face. The proton gives another OH group by adsorption on an adjacent 02-(cus) surface atom. This latter group is of the type I /?\ Th Th Th explaining the appearance of the 3550 cm-l band. The fact that singly bonded or bridging methoxy and OH groups correspond to different frequencies accounts for results obtained in organometallic chemistry. For instance, the OH group of Me,SiOH (an Si-0-H group like species I) corresponds to a band at 3737 cm-l in the gas phase.lg Its wavenumber is higher than those character- istic of species I1 (v(0H) = 3643 cm-' in the case of [Me,GaOH], in CCl, or of species 111 ((v(0H) = 3595 cm-1 in the case of [Me,PtOH],}.21 Moreover, experiments in progress(22) show that the addition of water or ethanol to bridged methoxy species (species 11) transforms them into species I, in agreement with the expected behaviour of bridged species according to ref.(23). It is therefore possible to couple the results obtained by STEM and FTIR spectroscopy and to conclude that the local arrangement of ions on some faces induces the formation of different types of species. Their thermal stability depends on their structure, according to whether they are monodentate (methoxy groups of type I are easily desorbed as methanol) or bridged (desorption of methoxy species of type I1 leads to dimethyl ether and CO). Work is in progress to show that reactivity also depends on the structure of species formed, i.e. on the adsorption faces. References 1 2 3 4 5 6 7 8 9 10 11 12 13 14 15 16 17 18 19 20 21 22 23 J.E. Germain, Stud. Surf. Sci. Catal., 1985, 21, 355. M. Boudard, Adv. Catal., 1969, 20, 153. J. J. Maj, C. Colmenares and G. A. Somerjai, Appl. Catal., 1984, 10, 313; J . Catal., 1985, 95, 385. B. Denise and R. P. A. Sneeden, React. Kinet. Catal. Lett., 1984, 26, 265. J. Lamotte, J.-C. Lavalley, E. Druet and E. Freund, J . Chem. SOC., Faraday Trans. 1, 1983, 79,2219. J. Lamotte, J.-C. Lavalley, V. Lorenzelli and E. Freund, J. Chem. SOC., Faraday Trans. 1, 1985,81,215. H. Knozinger and P. Ratnasamy, Catal. Rev. Sci. Eng., 1978, 17, 31. A. V. Alekseev, Yu. N. Lopatin, A. A. Tsyganenko and V. N. Filimonov, React. Kinet. Catal. Lett., 1974, 1, 443. G. Busca, P. F. Rossi, V. Lorenzelli, M. Benaissa, J. Travert and J.-C. Lavalley, J . Phys. Chem., 1985, 89, 5433. J. P. Lynch, E. Lesage, H. Dexpert and E. Freund. Znst. Phys. Conf. See. no. 61 (Institute of Physics, Bristol, 1981), p. 67. J. M. Cowley and J. C. H. Spence, Ultramicroscopy, 1979, 3, 433. E. Rosenberg, Thise de Docteur Ingenieur IFP no 32330 (Institut Franqais du Petrole, 1984). M. M. J. Treacy, PhB. Thesis (University of Cambridge, 1979). J.-C. Lavalley, J. Lamotte, G. Busca and V. Lorenzelli, J. Chem. SOC., Chem. Commun., 1985, 1006. J. Travert, 0. Saur, M. Benaissa, J. Lamotte and J.-C. Lavalley, in Vibrations at Surfaces, ed. R. Caudano, J. M. Gilles and A. A. Lucas (Plenum Press, New York, 1982), p. 333. A. J. Tench, D. Giles and J. F. J. Kibblewhite, Trans. Faraday SOC., 1971, 67, 854. A. A. Tsyganenko and V. N. Filimonov, J. Mol. Struct., 1973, 19, 579. A. J. Tyler, F. H. Hambleton and J. A. Hockey, J. Catal., 1969, 13, 35. J. Rouviere, V. Tabacik and G. Fleury, Spectrochim. Acta, Part A , 1973, 29, 229. R. S. Tobias, M. J. Sprague and G. E. Glass, Znorg. Chem., 1968, 7 , 1714. P. A. Bulliner and T. G. Spiro, Znorg. Chem., 1969, 8, 1023. J. Lamotte, V. Moravek and J.-C. Lavalley, to be published. V. Moravek and M. Kraus, J. Catal., 1984, 87, 452. Paper 611 136; Received 6th June, 1986

ISSN:0300-9599    

DOI:10.1039/F19878301417

出版商:RSC    

年代:1987

数据来源: RSC

摘要:

J . Chem. SOC., Faraday Trans. 1 , 1987, 83, 1427-1436 Infrared Spectroscopic Study of Catalytic Activity of Rh/TiO, for the CO + H, Reaction Isao Mochida" and Hiroshi Fujitsu Research Institute of Industrial Science, Kyushu University, Kasuga, Fukuoka 81 6, Japan Nobuhide Ikeyama Department of Molecular Technology, Graduate School of Industrial Sciences, Kyushu University, Kasuga, Fukuoka 81 6, Japan The catalytic hydrogenation of CO with hydrogen over Rh/TiO, reduced at 200, 400 and 500 "C (denoted Cat-200, -400 and -500) was studied with in situ F.t.i.r. by monitoring the behaviour of adsorbed CO species to explain their activity orders of Cat-400 > Cat-200 > Cat-500. The very high reactivity of irreversibly adsorbed CO with hydrogen was retarded by reversibly adsorbed CO and hydrocarbon intermediates in the steady state of the catalytic reaction.A competitive adsorption study of H, and CO on the catalyst revealed that their adsorption varied delicately according to the reduction conditions of the catalyst to influence the catalytic activity. Adsorption of CO monitored by F.t.i.r. was fully consistent with the reaction kinetics. Based on these results, the highest catalytic activity of Cat-400 may be ascribed to the following factors: (i) very high reactivity of irreversibly adsorbed CO with H,; (ii) the largest number of adsorption sites under the reaction conditions; (iii) the balanced adsorption of both substrates, which is very sensitive to their pressure, leading to the minimum retardation of H, adsorption by reversibly adsorbed CO and larger reaction orders in both substrates.Since Tauster et al. reported the SMSI (strong metal-support interaction) phenomena,l it has been well documented that catalytic activities and adsorption abilities of metals supported on oxide carriers are strongly influenced by the reduction conditions.2-8 The discovery of better catalysts,2$9-11 as well as the physical description of metals on support^,^^ 6 y 12-14 has been extensively studied from the viewpoint of SMSI. In spite of extensive studies on SMSI, the interaction of substrates with metals, particularly in their working states, has scarcely been in~estigated.~ We have reported kinetic studies of the CO-H, reaction at 250 "C over Rh supported by titania which was reduced under variable conditions (at 200,400 and 500 "C for 2 h) to reveal kinetically how the highest catalytic activity was achieved by a particular reduction condition (at 400 "C for In the present paper we report an in situ infrared spectroscopic study of CO adsorption on the Rh/TiO, catalysts prepared under several reduction conditions.The adsorption of CO alone, its competitive adsorption with H,, its irreversible adsorption and the reactivity of adsorbed CO species with H, under the catalytic conditions were investigated spectroscopically. All measurements were performed at 200 "C because the reactions proceeded too fast at 250 "C for a spectroscopic study. Comparison of these spectroscopic results with the kinetic ones may reveal the behaviour of substrates on the catalyst in the catalytic reaction, providing some clues to solve the reaction mechanism.Influences of the reduction conditions are, thus, described more directly in terms of the catalyst- substrate interaction. 2 h).15? l6 14271428 Catalytic Activity of Rh/TiO, Experimental Rh/TiO, was prepared by the impregnation of RhCl, - 3H,O on TiO, (Aerosil Inc., P-25, 50 m2 g-l) using a methanol solution (Rh: 4.6 wt % ). The catalyst (100 mg) was moulded into a disc (4 = 20 mm) and reduced with 200 Torr? hydrogen in situ in an i.r. cell (CaF, window) connected to a gas-circulating line, by raising the reduction temperature from room temperature to the prescribed one, holding the temperature for 2 h and then cooling to 200 "C where the catalyst was evacuated for 1 h before the introduction of the substrate gas.Three kinds of reduction conditions were examined according to a previous paper:15 200 "C, 2 h (Cat-200); 400 "C, 2 h (Cat-400) and 500 "C, 2 h (Cat-500). 1.r. spectra of CO adsorbed on the catalyst were recorded at 200 "C by an f.t.i.r. spectrometer (FT-IR-O3F, JEOL Inc.) which could provide a spectrum within 0.3 s of the scanning time (400-4000 cm-l). In the single adsorption of CO, i.r. spectra of adsorbed species were measured 10 s and 10 min after the introduction of CO, where the pressure of CO ranged from 1 to 200 Torr. The spectra of irreversibly adsorbed CO were observed by evacuation for 1 h after CO of prescribed pressure was adsorbed on the catalyst for 10 min at 200 "C. Intensities of adsorbed CO in i.r. spectra corresponded well to the adsorption amounts measured volumetrically in the pressure range 0.3-200 Torr at 200 "C. Competitive adsorption of CO and H, was recorded at intervals from 1 s to 1.5 h after well mixed quantities of CO and H, at their prescribed partial pressures were introduced onto the catalyst, products except for methane being trapped in a cooled trap at The reactivity of irreversibly adsorbed CO in the competitive adsorption with H, was also measured spectroscopically by the introduction of H, (400 Torr).The gas phase was analysed by a mass spectrometer (Anelva Inc., TE-600). Because the transmittance of the background decreased in the presence of H, in the gas phase for Cat-200, intensities of the adsorbed CO species were normalized by taking account of this change.- 196 "C. Results Reaction Orders in CO and H, on Cat-200, -400 and -500 Table 1 summarizes reaction orders in both substrates at 200 "C on Cat-200, -400 and -500 with the reaction rates (CO/H, = 200/400 Torr). The reaction orders as well as the activities were very variable on the catalysts as observed at 250 "C in previous papers.l57 l6 Adsorption of Carbon Monoxide on Rh/TiO, Fig. 1 shows the i.r. spectra of CO of variable pressures at 200 "C over Cat-200. At a low pressure of CO (50 Torr), adsorption bands at 2044 and ca. 1900 cm-l were observed, which have been identified as linear and bridging species,179 respectively. Increasing the CO pressure (100 Torr) intensified these bands and provided another linear species at 2100 cm-l. Further increase of CO pressure (200 Torr) intensified these bands still more and produced another band of a physisorbed species at 2 170 cm-l.The catalysts reduced at higher temperatures (400 and 500 "C) exhibited similar profiles, although the intensities of the species were different, depending upon the reduction conditions. Quantities of adsorbed CO species (equilibrium pressure = 200 Torr) on the catalysts are illustrated in fig. 2. Similar amounts of CO were adsorbed on Cat-200 and Cat-400 (CO/Rh = 0.3). Similar amounts of bridging and linear species were present on the former catalyst, whereas linear species were dominant on the latter catalyst. The amount t 1 Torr = 101 325/760 Pa.I. Mochida, H . Fujitsu and N . Ikeyarna Table 1. Catalytic activity and reaction orders in H, and CO orderb catalyst activitya X Y Cat-200 120 0.5 -0.2 Cat-400 200 1.0 0.0 Cat-500 80 1.3 -0.3 a CO/H, = 200/400 Torr, reaction temp: 200 "C.Rate: u = kPg2 PEo. 0*6 0.4 0.2 0 1429 0.6 - 04 0.2 0 - - _/v I I 2200 1900 1600 wavenum ber/cm-' Fig. 1. 1.r. spectra of CO adsorbed on Cat-200 under variable pressures at 200 "C. Equilibrium pressure of CO: (a) 50 Torr, (b) 100 Torr, (c) 200 Torr. of CO adsorbed on Cat-500 was much less (CO/Rh = 0.12) than on the former two catalysts. More linear species than bridging species were adsorbed on this catalyst. A significant contribution of physisorbed species (20 % of adsorption) on Cat-500 should be noted. Fig. 3 shows the spectra of CO adsorbed on Cat-200 at 200 "C after evacuation at the same temperature for 1 h.Adsorption bands of linear and bridging species were observed at 2040 and ca. 1900cm-l. The intensity of these bands increased1430 Catalytic Activity of Rh/TiO, I I (a) ( b ) (0) ( b ) (a) ( b ) Fig. 2. Intensity of adsorbed CO species on Rh/TiO, at 200°C. (A) Cat-200, (B) Cat-400, (C) Cat-500, (a) equilibrium pressure of CO = 200 Torr, (b) irreversibly adsorbed CO after the evacuation of (a) at 200 "C for 1 h. 0.6 0.4 0.2 0 -2 0.6 v 2 0.4 2 0.2 2 0 0.6 0.4 0.2 0 n s e, .$! 8 I ,L, (c 1 b I I 2200 1900 1600 wavenumberlcm-' Fig. 3. 1.r. spectra of irreversibly adsorbed CO on Cat-200 at 200°C. Spectra were measured after the evacuation of adsorbed CO under variable equilibrium pressures (Torr): (a) 50, (b) 100, ( c ) 200.I. Mochida, H. Fujitsu and N .Ikeyama 143 1 L I 3200 2800 2400 2000 1600 wavenumberlcm -' Fig. 4.1.r. spectra of CO adsorbed on Cat-200 in the competitive adsorption with H, at 200 "C. CO-H, = 200/400 (Torr). Reaction time: (a) 3 min, (b) 6 min, (c) 1.5 h, (d) after evacuation of (c) for 1 h at 200 "C, (e) 1 s after the introduction of H, (1 Torr) onto (d), (f) 1 s after the introduction of H, (400 Torr) onto (d), ( g ) 25 s after the introduction of H, (400 Torr) onto (d). significantly with the increasing pressure below 50 Torr and saturated around 100 Torr of CO pressure. Other catalysts exhibited very similar profiles, although relative intensities of the bands were different. Quantities of irreversibly adsorbed CO species on three catalysts are illustrated in fig. 2(b). Both Cat-200 and Cat-400 carried ca.0.2 CO/Rh. Cat-200 lost much more bridging species by the evacuation, leading to the dominant presence of linear CO on this catalyst. In contrast, similar amounts of bridging and linear species were present on the latter catalyst. Cat-500 carried ca. 0.1 CO/Rh. Similar amounts of linear and bridging species were present on this catalyst. Competitive Adsorption of CO and H, Fig. 4 shows a series of i.r. spectra (in transmittance mode) of CO on Cat-200 after the introduction of the mixed gas (CO/H, = 200/200 Torr). 1 s after the mixed gas introduction physically and chemically adsorbed CO species were observable in the spectra. The intensities of the adsorption bands increased gradually with the time until steady state was reached after 20 min, however the intensities were still much smaller than those obtained with CO alone, indicating the competitive adsorption of both substrates at the steady state.The total amount of adsorbed CO was comparable to that of irreversibly adsorbed CO obtained with CO alone, although the former amount included the same amount of physisorbed CO species to that in the single adsorption.1432 Catalytic Activity of Rh/TiO, I A L O3 t 5 n ]R P L C Fig. 5. Intensities of CO species adsorbed on Cat-200 in the competitive adsorption with H, at 200 "C. CO/H, (Torr): (A) 200/400, (B) 200/200, (C) 100/400. Reaction time: (a) 1 s, (b) 25 s, (c) 6 min, ( d ) 20 min, (e) 1.5 h. In addition to the bands ascribed to CO species, bands at 2800-3000cm-1 (C-H stretching region) were also observed in the spectra after 3 min, indicating the production of hydrocarbon intermediates.Their intensity increased gradually with the reaction time to reach steady amounts after 30 min. The conversion of CO in the gas phase was ca. 10% by this stage. The major product found in the gas phase was methane. Small amounts of C,-C, hydrocarbons were collected in a cold trap at - 196 "C. No CO, was found in the product, indicating that no carbon formed on the catalyst. The quantities of chemically and physically adsorbed CO under variable pressures of CO and H, on Cat-200, -400 and -500 are summarized in fig. 5-7. Bridging, linear and physically adsorbed CO species were observed on Cat-200 under competitive conditions (fig. 5A, CO 200 Torr, H, 400 Torr).The amounts of the bridging species increased gradually with time until 90 min, while those of other species remained constant. Note that the same amount of physically adsorbed CO was found for single (CO alone) and competitive adsorption, whereas the amount of chemically adsorbed species in the competitive adsorption was roughly a half of that in the single adsorption. Although the linear species was dominant among the irreversibly adsorbed CO species, similar amounts of linear and bridging species were observed in the competitive adsorption. Decreasing the H, pressure to 200 Torr at the same CO pressure provided very similar profiles of CO adsorption (fig. 5B), indicating that CO adsorption is not influenced by H, pressure on this catalyst in this pressure range.Decreasing the CO pressure to 100 Torr (fig. 5C) made the adsorption of CO slow. However, the same amount of chemically adsorbed species as that under a CO pressure of 200 Torr was observed after 20 min, although the amount of physically adsorbed species decreased to ca. one-half. Cat-400 exhibited a peculiar change of CO adsorption in the competitive adsorption under CO and H, pressures of 200 and 400 Torr, respectively (fig. 6A). Within the first second, the same profiles of CO adsorption were observed as those in for single CO adsorption, however the amount adsorbed decreased gradually to reach that of the stationary state after 6 min remained constant until 90 min, while the hydrogenation reaction proceeded to give a CO conversion of 25 % . The amount of all adsorbed species in the stationary states was ca.one-third of that of the single adsorption. The amount of chemically adsorbed CO was half of that of the irreversibly adsorbed CO [see fig. 2 B(b)], indicating strong competition in the adsorption of the substrates on the catalyst.I . Mochida, H . Fujitsu and N . Ikeyarna A B C 1433 Fig. 6. Intensities of adsorbed CO species on Cat-400 in the competitive adsorption with H, at 200 "C. CO-H, (Torr): (A) 200/400, (B) 200/200, (C) 100/400. Reaction time: (a) 1 s, (b) 3 min, ( c ) 6 min, ( d ) 1.5 h. 2 0.1 8 \ A n C Fig. 7. Intensities of adsorbed CO species on Cat-500 in the competitive adsorption with H, at 200 "C. CO-H, (Torr): (A) 200/400, (B) 200/200, (C) 100/400. Reaction time: (a) 1 s ; (b) 25 s, 1.5 h; (c) 1 s, 1.5 h.The amount of physisorption was ca. 60% of the single adsorption. Decreasing H, or CO pressures drastically changed profiles of competitive adsorption as shown in fig. 6 B and C. Lower pressure of H, (200 Torr) provided the same adsorption as that observed under CO alone at the same pressure, while decreasing CO pressure to 100 Torr decreased CO adsorption to ca. one-half of that observed at a CO pressure of 200 Torr under the same hydrogen pressure (400 Torr). No linear species was observed. In the competitive adsorption of 200/400 Torr of CO-H, Cat-500 exhibited a very small amount of CO adsorption (CO/Rh was only 0.05), which was ca. one-third of that for the former two catalysts (fig. 7A). The amount adsorbed increased slightly with time and levelled off within 25 s.The physisorption occupied roughly half of the whole1434 Catalytic Activity of Rh/TiO, adsorption, which was roughly the same as that for single adsorption. Decreasing the H, pressure to 200 Torr increased the CO adsorption significantly to reach the amount observed with adsorption of CO alone, the chemical adsorption becoming dominant (fig. 7B). The amount of physisorbed CO remained at a similar level. The adsorption was rapid in this case, the same profiles being observed from 1 s to 1.5 h after gas introduction. Decreasing CO pressures to 100 Torr decreased the amount adsorbed to the level of 0.03 in CO/Rh ratio (fig. 7C). Note that only the linear form of chemisorbed CO was observed in addition to the considerable amount of physisorbed CO.Reactivity of Irreversibly Adsorbed CO at the Steady State in the Competitive Adsorption with H, At a very early stage in the competitive adsorption at 200 "C, the irreversibly adsorbed CO species exhibited the same band shapes and positions as those obtained by the single adsorption on all catalysts, although their amounts were less than those in CO alone (40-90%). The reactivities of these CO species with H, were very high and were similar to those of the singly adsorbed ones reported previously,19 their disappearance being completed within 1 s (producing methane). The rates of converison were calculated to be at least (1-6) x lo3 mmol CO(g Rh)-l h-l, and were much larger than those of the catalytic reaction [80-200 mmol CO(g Rh)-l h-l]. The spectrum of irreversibly adsorbed CO on Cat-200 at the steady state after 1.5 h of competitive adsorption (CO-H, = 200/200 Torr) is shown in fig.4(d). Two bands were observed in the CO stretching region at 2000 and 1850 cm-l. These bands may be identified as linear and bridging species (linear/bridged ratio : 5/3), respectively, although they were bathochromically shifted by 20 cm-l compared to those (linear/bridging ratio : 9/1) of the single adsorption shown in fig. 3. According to literature,,O such a bathochromic shift is due to the adsorbed hydrocarbons which had been produced catalytically. Their intensity was 45% [CO/Rh = 0.08, fig. 5B(f)] of that observed in the single adsorption. In addition to these bands, C-H stretching bands around 2900 cm-l remained after evacuation.The irreversibly adsorbed species was converted to the extent of 69 and 95% with 400 Torr of H, after 1 and 25 s, respectively. The conversion rate after 1 s was calculated to be 2 x lo3 mmol CO(g Rh)-l h-l, two-thirds of the corresponding rate observed in the single CO adsorption [3 x lo3 mmol CO(g Rh)-l h-l 19], respectively. The bands around 2900 cm-l also disappeared after the introduction of H,. Thus, the surface of the catalyst was blocked to some extent by hydrocarbon intermediates which retard the reaction, although the rate was still much higher than that of the CO-H, stationary reaction. The irreversively adsorbed CO on Cat-400 at the steady state [CO-H, = 200/400 Torr, fig. 6A(e)] exhibited similar bands at 1850 and 2000 cm-l to those on Cat-200, although their intensity (CO/Rh = 0.09) was 40% of the irreversibly adsorbed ones in the single adsorption.The similar bands in the C-H stretching region to those on Cat-200 were also observed. Both the CO species and hydrocarbon intermediates showed similar reactivities with H, as those on Cat-200, respectively. In contrast to the former two catalysts, no irreversibly adsorbed CO species was observed on Cat-500 in the steady state [CO-H, = 200/400 Torr, fig. 7A(b)]. The bands in the C-H stretching region also disappeared after the introduction of H,. Discussion We have reported that the catalytic activity of Rh supported on titania varied delicately depending upon the conditions of its reduction temperature and time.l59 l6 Such reduction conditions significantly influenced the kinetics of the reaction as shown in the presentI. Mochida, H.Fujitsu and N . Ikeyama 1435 study to define the catalytic activity under particular reaction conditions. The present study revealed that such catalytic parameters of Rh/TiO, catalysts reflect the adsorption of CO under reaction conditions which is easily monitored by in situ F.t.i.r. First of all, the catalytic reaction of Rh/TiO, starts with the hydrogenation of adsorbed CO by H,. The Boudouard reaction leading to the hydrogenation of the carbonaceous intermediate21 is minor in the present case, although the reactivity of the intermediate depends on the nature of the supports.22 The reactivity of irreversibly adsorbed CO present in the steady state of the catalytic reaction decreased to two-thirds of that of the species produced from CO alone;19 however, the rate of disappearance of irreversibly adsorbed CO after the introduction of hydrogen was still much larger (14 times) than that of the catalytic reaction. Such observations suggest that the catalytic reaction may proceed through the reaction of irreversibly adsorbed CO and hydrogen which is adsorbed or in the gas phase, and that CO in the gas phase or adsorbed as other species, physically or reversibly adsorbed CO, retards the reaction.Based on such a consideration, the rate of the catalytic reaction can be described by the following equation : pco PH2 rate = kSKCO(irr)KH2 KCO(irr)PCO -k KCO(rev)PCO -k KH2pH~ where S is the number of adsorption sites and Kco(irr), Kco(rev) and KH2 are adsorption equilibrium constants of the respective species.Kco(rev) Pco defines the amount of reversibly adsorbed CO which retards the reaction between irreversibly adsorbed CO and hydrogen. At present, adsorbed hydrogen species are not identified. Nevertheless, the change in the amount of adsorbed CO observed in the competitive adsorption with H, suggests the importance of its adsorption. In addition, hydrocarbon intermediates which stay on the catalyst to be further hydrogenated into long-chain hydrocarbon products also retard the reaction by disturbing the adsorption of both substrates at the steady state. The catalytic activity and reaction orders in CO and H, on the three Rh/TiO, catalysts shown in table 1 can be discussed based on the eqn (1).Cat-500, which showed the lowest activity and - 0.3 and 1.3 order in CO and H,, respectively, has the smallest ability for irreversible adsorption of CO so that S is smallest, although the amounts of physisorption and reversible adsorption are proportional to CO pressure. The amounts of hydrogen adsorbed on the catalyst in the competitive adsorption with CO are estimated from fig. 7 by assuming that the hydrogen can adsorb only on the vacant sites among the whole sites for CO adsorption, i.e. the difference between the amount of single and competitive adsorption. The values of H/Rh are calculated as 0.06 and 0.07 when the CO/H, pres- sures are 200/400 and 100/400 Torr, respectively, while H/Rh is 0.13 when hydrogen alone of the same pressure is adsorbed, Thus, the adsorption of hydrogen on the catalyst is limited by the adsorption of CO to give a negative order in CO.In turn, higher H, pressure replaces the reversibly adsorbed CO (see fig. 7A and B), giving an order in H, larger than unity. Hence, the adsorption step of H, may also be a slow step. Thus, this catalyst exhibited the lowest activity. Cat-400 has balanced abilities for reversible CO and H, adsorption, hence both adsorption of CO and H, are influenced strongly by their partial pressures. In contrast, the amount of irreversibly absorbed CO, which was largest among the catalysts to indicate largest number of S, was independent of the CO pressure above 100 Torr to give zero order in CO. Although at high pressure H, is substituted for adsorbed CO, CO occupies almost all sites of Rh when the H? pressure is low, as shown in fig.6. The participation of H, adsorption in the slow step IS assumed to explain the reaction being first-order in H,. Thus, the largest activity is provided under these present conditions. The reversible adsorption of CO was proportional to CO pressure on Cat-200,1436 Catalytic Activity of Rh/TiO, although irreversible adsorption was independent of the pressure above 100 Torr. Reversible adsorption of CO may decrease the amount of dissociatively adsorbed hydrogen to give a negative order in CO and a half order in H,. In spite of the similar reactivity and amount of irreversibly adsorbed CO in the CO single adsorption to those of Cat-400, Cat-200 showed a smaller activity under the H, and CO pressures used in the present study because of its smaller amount of irreversibly adsorbed CO species (CO/Rh = 0.08) and adsorbed hydrogen (H/Rh < 0.12, see fig.5A) under the reaction conditions than those (CO/Rh = 0.09 and H/Rh < 0.2 see fig. 6A) on Cat-400. The present study revealed some details in terms of the adsorption ability of the catalysts according to the reduction conditions which strongly influence the catalytic activity and reaction orders. Such an ability certainly reflects the state of the catalyst. The chemistry regarding the origin of such an ability is still unknown, although the strong metal-support interaction is often ascribed as the 5 7 Although this question is beyond the scope of this investigation, the present catalyst-substrate interaction may suggest that Rh is electronically modified by the support to give the respective adsorption abilities.References 1 S. L. Tauster, S. C. Fung and R. L. Garten, J. Am. Chem. SOC., 1978, 100, 170. 2 M. A. Vannice and R. L. Garten, J. Catal., 1979,66, 236; 1979,66, 242. 3 M. A. Vannice, J. Catal., 1982, 74, 199. 4 P. Meriaudeau, 0. H. Ellestad, M. Defaux and C. Naccache, J. Catal., 1982, 75, 243. 5 D. E. Resasco and G. L. Haller, J. Catal., 1983, 82, 179. 6 R. Burch and A. R. Flamford, J. Catal., 1982,78, 389; J. D. Bracey and R. Burch, J. Catal., 1984,86, 7 K. Kunimori, S. Matsui and T. Uchijima, J. Catal., 1984, 85, 253. 8 H. Miessner, S. Naito and K. Tamaru, J. Catal., 1985, 94, 300. 9 M. A. Vannice, S-Y. Wang and S. H. Hoon, J. Catal., 1981,71, 152. 384. 10 M. A. Vannice, Pan-Pasific Synfuels Conference, Abstract, 1982, vol. 1, p. 208. 11 E. Kikuchi, H. Nomura, M. Matsumoto and Y. Morita, Pan-Pasific Synfuels Conference, Abstract, 12 R. T. K. Baker, E. B. Prestridge and R. L. Garten, J. Catal., 1979, 66, 390. 13 A. K. Singh, N. K. Pande and A. T. Bell, J. Catal., 1979, 66, 390. 14 T. C. Chang, J. J. Chem and A. T. Bell, J. Catal., 1985, 94, 422. 15 H. Fujitsu, N. Ikeyama, Y. Shigaki and I. Mochida, Bull. Chem. SOC. Jpn, 1985, 58, 1849. 16 H. Fujitsu, N. Ikeyama and I. Mochida, J. Catal., 1986, 100, 274. 17 J. E. Hulse and M. Morkoirts, Surf. Sci., 1975, 57, 125. 18 A. M. Bradshaw and F. M. Hoffmann, Surf. Sci., 1978, 72, 513. 19 I. Mochida, N. Ikeyama and H. Fujitsu, Bull. Chem. SOC. Jpn, submitted. 20 M. Primet, J. Catal., 1984, 88, 273. 21 A. Erdohelyi and F. Solymosi, J. Catal., 1983, 84, 446. 22 F. Solymosi, I. Tombacz and M. Kocsis, J. Catal., 1982, 75, 78. 1982, vol. 1, p. 216. Paper 611 158; Received 9th June, 1986

ISSN:0300-9599    

DOI:10.1039/F19878301427

出版商:RSC    

年代:1987

数据来源: RSC

摘要:

J. Chem. SOC., Faraday Trans. 1, 1987,83, 1437-1447 Effect of 1,4-Dioxane on the Conductance and Ion-pairing of Hydrogen Chloride in Wet and Dry Methanol Mixtures at 25 O C Mario Goffredi Istituto di Chimica Fisica, Universita di Palermo, Via ArchiraJi 26, 90123 Palermo, Italy The molar conductances of solutions of HCl in wet (74% methanol-26% 1,4-dioxane) and (54% methanol46% 1,4-dioxane) mixtures at 25 "C have been measured. The data have been analysed by the expanded Fuoss-Hsia equation. The pK, and Am in the anhydrous binary solvent systems have been determined. The influence of water on the percentage excess proton mobility with respect to the potassium ion has also been considered. A tentative interpretation of the observed trend is suggested, based on the formation of a hydrogen-bonded associate between methanol and 1,4- dioxane molecules. Our interest in a more detailed interpretation of the observed change of the anomalous proton mobility upon modification of the structural properties of an amphiprotic solvent either by changing its molecular size and shapely2 or by mixing it with an organic co~olvent,~ has led to a systematic investigation of the conductance behaviour of hydrogen chloride in this type of solvent medium. As far as mixture of solvents is concerned, data have been presented so far on the transport properties at 25 "C of HC1 in dry and wet methanol-methylcyclopentane mixtures with a molar ratio of ca.8: 1 and a dielectric constant value E = 24.3.3 Here we report the conductance results for the same ionogen in several dry and wet methanol- 1,4-dioxane mixtures, namely two solvent systems made up with (74% methanol-26 % 1,4-dioxane, E = 24.3) and with (54% methanol-46% 1,4-dioxane, E = 17.5) by weight, respectively (molar ratio ca.8: 1 and 3: l), to which variable amounts of water were added. As is known, the 1,4-dioxane molecule is apolar (p = 0.45 D t ) and aprotic, but the two oxygen atoms in the ring make it a bifunctional hydrogen-bond acceptor. These results, therefore, should yield further information on the proton conductance behaviour resulting from possible structural modifications of the amphiprotic medium. Experimental 1,4-Dioxane, reagent grade, was further purified by refluxing over KOH pellets for at least 48 h and then by fractional distillation.The middle fraction was collected and dried by refluxing over metallic Na and was finally fractionated at the time of use. The specific conductivity of the final material was always 10-lo S cm-l at 25 "C and no water content was detected by Karl Fischer titration. Methanol, reagent grade, was dried over calcium hydride for several days and was subsequently fractionally distilled under dry nitrogen. The specific conductance of the final product varied between 1 x S cm-l at 25 "C and its water content, estimated by Karl Fischer titration, was always < 0.01 % . Stock solutions of HCl were and 2 x t 1 D = 3 . 3 3 5 6 4 ~ C m. 14371438 Ion-pairing of HCl in Methanol-Dioxane Mixtures prepared and analysed as described previously.4 All solvent mixtures for the conductance runs and successive additions of concentrated stock solution of hydrogen chloride in the conductivity cell were made up by weight and all weights were vacuum corrected. Solvent densities were used to compute the concentrations (mol dm-3) of these solutions.Details of the electrical equipment used for measuring the conductance and the other experimental details can be found in a previous p~blication.~ The precision of the conductance measurements, estimated in previous studies, is of the order of 0.1 % . The general techniques and instrumentation for dielectric constant (e), viscosity ( q ) and density ( d ) measurements of the anhydrous and wet solvent mixtures have been described in detail el~ewhere.~ The dependence of these physical properties on water concentration, expressed as weight percent, WHzo, for both solvent systems at 25 "C is described well by the following equations : d8/1 = 0.838 95 + 2.42 X WH20 (1) d3/1 = 0.885 80 + 2.06 x WHzo (2) &8/1 = 24.20+0.563 WH,O (3) 8311 = 17.50+0.610 WHzo (4) q8/1 = 0.005592+2.74 X lop4 WH2,-3.5 X pHz0+2.5 x lo-' wHro ( 5 ) where d is expressed in g and q in P.Results and Discussion Electrolyte concentrations c (mol dm-3) and the observed molar conductances A (Scm2mol-l) in both aqueous solvent systems are presented in tables 1 and 2. The limiting molar conductances Am, the association constant KA and the distance para- meter B were evaluated by means of the equation5 A = Am-Sz/(ca)+Eca Iogca+J,ca+J,(ca)~-K,cay$A (7) and are summarized in tables 3 and 4 together with their standard deviations and the standard deviations of the individual points o,,, The general trend displayed by these tables is that as the water concentration increases both Am and KA decrease.Secondly, by increasing the amount of the aprotic cosolvent in the mixture, the molar limiting conductance decreases while the degree of association increases as a result of lowering the bulk dielectric constant of the solvent medium. The corresponding values of pK, for HCl in the two anhydrous solvent systems are 2.06 and 2.87 for the 8: 1 and 3: 1 mixtures, respectively. In this respect the present results do not differ qualitatively from those obtained in our earlier studies. The addition of water to the anhydrous solvent system results, in fact, in a dramatic change in the Am values.This effect is shown in fig. 1, where the values of Am in the (8: 1 MeOH-1,4-dioxane) and (3: 1 MeOH-l,4-dioxane) solvent mixtures, respectively, are plotted as a function of the water content (wt % ) of the solvent media. The plot shows the same characteristic U-shaped pattern previously observed in other amphiprotic systems. Both curves, in fact, after a marked decrease pass through a minimum above which they rise again. This observation is taken once more as evidence of the controlling influence of the higher proton affinity of water with respect to the amphiprotic cosolvent on the proton transfer process.g By graphical interpolation of the flat portion of these curves the minimum values of Am were estimated. They are (Am)min = 98.5 and (Am)min = 90.0 for the [H,O+(8: 1 MeOH-l,LC-dioxane)] and [H20 + (3 : 1) MeOH-1,4-dioxane)] systems, respectively.Additionally, by linear extrapo-M. Gofredi 1439 Table 1. Molar concentrations and conductances of hydrogen chloride and potassium chloride in water48 : 1 MeOH-1,4-dioxane) mixtures at 24 "C H,O q10-4 A H,O q 1 0 - 4 A H,O q10-4 A (wt x ) mol dm-3 /cm2 S mol-l (wt % ) mol dm-3 /cm2 S mol-1 (wt % ) mol dm-3 /cmz S mol-' 0.010 0.149 0.713 3.020 6.504 13.976 6.0610 8.7313 1 1.355 13.062 15.239 17.370 19.029 3.5262 5.5362 7.9469 10.446 12.700 14.599 16.847 4.2569 7.7334 9.5295 11.805 13.056 14.914 17.729 4.7885 6.9830 10.079 12.664 14.797 16.216 5.3182 6.9830 10.079 12.664 14.797 16.216 20.910 7.7339 9.6628 12.030 13.942 15.542 17.926 19.204 156.52 152.73 149.74 148.00 145.87 144.10 142.83 144.68 141.42 138.49 135.90 133.68 132.15 130.52 1 16.83 1 13.47 112.08 110.46 109.62 108.47 106.85 98.79 92.93 91.37 90.28 89.54 88.94 93.87 92.93 91.37 90.28 89.54 88.94 87.81 95.14 94.40 93.75 93.18 92.75 92.19 91.87 0.016 0.300 0.904 4.100 8.075 0.010 HC1 4.8854 8.2000 10.559 11.526 13.420 15.948 17.974 5.2851 7.469 1 10.483 13.478 15.544 17.545 20.153 10.195 12.815 15.568 17.336 18.994 20.891 6.6160 6.3533 8.3129 10.427 12.212 14.288 17.863 4.0029 6.6078 8.5578 1 1.503 13.319 16.260 18.354 KCl 3.8621 7.8627 12.368 17.768 23.267 31.149 157.36 152.71 149.79 148.74 146.89 144.72 143.09 133.22 130.93 128.22 125.89 124.3 1 123.06 121.46 108.77 106.03 104.32 102.81 101.83 101.04 100.24 95.16 93.96 92.83 92.01 91.10 89.76 93.98 92.48 91.35 90.22 89.42 88.56 88.00 91.85 88.43 85.63 82.98 80.86 78.38 0.069 0.409 2.019 5.970 10.080 10.040 8.5747 10.315 12.700 15.129 17.587 20.50 1 22.321 4.2905 6.6703 8.2642 9.8032 11.701 13.332 6.6428 9.061 1 11.575 12.853 16.349 18.620 20.737 4.7177 8.5305 10.009 11.366 12.602 14.426 15.892 7.1974 9.1743 11.371 13.551 14.994 17.060 8.0419 12.653 17.38 1 22.240 27.359 32.672 141.71 139.92 137.58 135.65 133.86 131.87 130.77 128.33 125.71 124.06 122.76 121.13 119.96 99.94 98.36 96.87 96.17 94.60 93.64 92.82 94.41 92.12 91.71 91.04 90.50 89.79 89.32 92.45 91.56 90.64 90.04 89.62 88.99 75.63 73.99 72.67 71.50 70.46 69.50 lation of these curves in the limited range of the lowest water concentration, a rough estimate of the limiting molar conductances of HCl in the anhydrous solvent mixtures were obtained, (A,),,, = 175 and 154 for the mixed solvent system with a molar ratio between methanol and 1,4-dioxane of ca.8: 1 and 3: 1, respectively. A better estimate of the limiting conductance values was obtained by means of the Thomas-Marum equation : (8) nHzO = -NIK + ALE+ n ~ ~ ~ / [ ( A o o ) m a x - (A,InI1440 Ion-pairing of HC1 in Methanol-Dioxane Mixtures Table 2. Molar concentrations and conductances of hydrogen chloride and potassium chloride in water-(3: 1 MeOH-1,Cdioxane) mixtures at 25 "C HZO c/ 10-4 A H,O c/10-4 A H,O ~110-4 A (wt % ) mol dm-3 /cm2 S mol-l (wt % ) mol drn-3 /cm2 S mol-l (wt % > mol dm-3 /cm2 S mo1-I 0.009 0.085 0.298 0.956 5.178 9.980 0.010 3.1218 4.5675 7.0949 9.6513 13.192 17.285 3.5820 5.3926 7.0705 11.210 14.490 20.222 2.5008 4.3017 7.0 134 10.263 12.725 16.005 2 1.028 1.8509 4.5437 7.9771 10.632 13.656 16.648 19.989 4.1322 6.4684 8.8830 11.106 13.128 15.430 2.9537 4.9441 6.8985 10.600 13.691 16.939 22.918 4.0598 8.5018 13.243 17.967 22.879 28.145 125.60 119.60 111.86 106.02 99.91 94.95 114.92 109.03 104.74 96.90 92.43 86.39 105.63 100.02 93.92 88.54 85.44 82.05 77.94 96.58 89.98 84.51 81.33 78.35 75.99 73.72 84.30 81.83 79.77 78.11 76.8 1 75.49 84.80 83.08 81.40 79.55 78.20 76.95 75.01 75.824 69.399 64.927 61.648 59.026 56.760 0.014 0.154 0.399 2.013 6.240 11.075 9.996 HCl 3.5176 4.9806 7.6493 10.608 12.999 18.507 2.1678 3.6854 8.3970 10.565 13.212 19.884 2.1777 3.5788 5.4927 8.2201 14.658 17.954 21.001 2.2022 5.2030 8.0309 12.473 16.73 1 21.217 3.6659 6.1 130 8.3 196 10.857 13.406 17.579 24.507 3.020 1 5.1321 7.4709 10.182 14.284 17.417 24.168 KCl 4.7 154 10.019 15.690 21.560 27.446 33.486 123.43 1 18.00 110.27 104.25 100.21 93.20 1 15.66 109.91 98.24 94.62 90.98 84.22 104.17 99.78 95.17 90.19 82.26 79.35 77.06 90.52 85.06 81.49 77.28 74.36 71.79 84.52 82.14 80.35 78.62 77.18 75.12 72.44 85.22 83.49 8 1.95 80.55 78.67 77.58 75.57 68.044 64.798 62.327 60.344 58.7 13 57.358 0.019 0.197 0.599 2.991 8.803 15.210 2.7270 4.8473 6.6002 8.7380 10.952 13.778 1 7.349 20.349 2.4003 4.6467 6.7375 8.9469 12.662 16.31 1 2.8035 4.7328 6.5750 10.146 13.296 16.739 24.072 1.9680 4.4045 7.4747 10.219 14.593 17.818 22.969 2.821 1 5.21 59 6.9 180 9.3890 12.686 16.587 23.407 2.8898 4.8838 6.6568 9.9518 12.550 15.342 20.516 125.59 116.76 111.69 106.69 102.33 98.00 93.49 90.37 11 1.70 103.96 98.98 94.83 89.41 85.48 98.12 93.46 89.87 84.78 81.40 78.38 73.47 89.85 85.53 8 1.66 79.19 75.90 73.99 71.41 84.75 82.41 80.99 79.52 77.79 76.12 73.74 88.36 86.93 85.74 84.32 83.35 82.41 80.91M .Gofredi 1441 Table 3. Conductance parameters for hydrogen chloride and potassium chloride in water-(8: 1 MeOH-1,4-dioxane) mixtures at 25 "C. H2O A KA H A (wt%) /cm2 S mol-l /dm3 mol-1 /10-lo m /cm2 S mol-I 0.010 0.016 0.069 0.149 0.300 0.409 0.713 0.904 2.019 3.020 5.970 6.504 8.075 10.080 13.976 0.010 10.0 172.65 f 0.02 171.32 f 0.02 159.04 f 0.02 155.09 f0.02 145.07 k0.02 138.47f0.01 125.94 f 0.02 119.700f0.008 109.332 & 0.007 106.193 f 0.005 100.40 f 0.02 100.10 f 0.02 99.04 f 0.02 98.57 f 0.03 100.777 +0.008 99.379 & 0.005 8 1.795 f 0.004 HCl 11 1.7f0.1 107.0 f 0.2 97.5 f 0.1 102.0+_0.2 85.3 k0.2 85.4 f 0.2 78.1 k0.2 75.82 f 0.08 61.09f0.07 57.19k0.06 38.3 f0.3 37.8k0.3 35.4f0.2 26.9 f 0.3 19.44 & 0.08 KCl 81.91 k0.06 34.07 f 0.03 4.6 4.7 4.9 4.8 5.3 5.3 5.5 5.6 6.0 5.9 6.5 6.4 6.0 6.6 6.6 5.6 5.5 0.06 0.08 0.05 0.09 0.1 0.07 0.07 0.03 0.03 0.03 0.09 0.1 0.09 0.1 0.03 0.03 0.02 where AAZ+ represents the excess of mobility, i.e.the difference between the AZ' value in the anhydrous solvent mixture and lHsO+ value at the minimum, (A,),,, and (A,)n are the limiting molar conductances in the anhydrous binary solvent system and in the wet system containing n mol dm-3 of water, respectively, N (mol dm-3) is the stoichio- metric concentration of methanol in the anhydrous solvent mixture, nHzO (mol dm-3) is the stoichiometric concentration of water in the two-component solvent system and K, the equilibrium constant of the proton distribution for the exchange reaction.(1 4-dioxane + xMeOH;) + H,O + (1 4-dioxane + xMeOH) + H30+ (9) where x represents the number of moles of the amphiprotic cosolvent in the binary solvent mixture. In this equation the whole system has been considered as a pseudo- binary solvent system. Accordingly, by the least-squares procedure described in a previous paper,4 the following values for (A,),,,, AAZ' and K, were obtained: (A,),,, = 174.7k0.5; AAZ' = 77_+2; K, = 112+9 for the solvent mixture (8: 1 MeOH-1,4-dioxane) and (A,),,, = 153.0k0.2; AAZ' = 60.6f0.4; Kr = 114+4 for the solvent mixture (3 : 1 MeOH- 1,4-dioxane).The percentage of abnormal conductance of the proton in both anhydrous mixed solvents and in the corresponding aqueous mixtures where the minimum in A, occurs, were evaluated using the equations 48 FAR 11442 Ion-pairing of HCl in Methanol-Dioxane Mixtures Table 4. Conductance parameters for hydrogen chloride and potassium chloride in water-(3: 1 MeOH-l,.Q-dioxane) mixtures at 25 "C H2O A K A H A (wt%) /cm2 S mol-1 /dm3 mol-1 /10-lo m /cm2 S mol-l 0.009 0.014 0.019 0.085 0.154 0.197 0.298 0.399 0.599 0.956 2.013 2.991 5.178 6.240 8.803 9.980 1 1.075 15.210 0.010 9.996 151.04k 0.01 150.19 f 0.01 148.07 & 0.02 138.68+0.01 131.87 f 0.01 128.07 & 0.02 121.21 kO.01 1 17.577 & 0.004 1 12.037 & 0.009 106.460 k 0.004 99.417k0.006 97.3 14 k 0.006 93.626 & 0.004 92.399 & 0.005 90.31 kO.01 90.03 f: 0.02 90.363 f 0.008 92.44 f 0.02 90.173 & 0.004 74.642 f 0.004 HC1 719.8 f0.2 70 1.7 & 0.2 690.8 k 0.4 635.0 f 0.2 579.7 f0.3 563.9 & 0.4 529.6 & 0.3 482.93 & 0.08 425.3 & 0.2 378.60 & 0.08 278.8 f 0.1 232.8 & 0.1 164.59k0.08 132.43 f 0.08 96.0 & 0.2 78.8 f 0.3 71.0f0.1 47.6 & 0.2 KCl 429.17 f 0.09 106.26f0.06 4.6 4.6 5.6 4.7 4.9 4.9 5.0 5.1 5.3 5.4 5.8 5.9 6.0 6.3 6.4 6.7 6.6 6.5 5.5 5.4 0.1 0.1 0.1 0.09 0.1 0.1 0.1 0.03 0.07 0.03 0.05 0.05 0.03 0.03 0.08 0.1 0.05 0.09 0.04 0.02 Table 5.Percentage of abnormal conductance of the proton in some anhydrous methanol solvent mixtures and in the wet mixtures where the minimum in Am occurs MeOH 52.4 49.5 146.4 55.8 94.0 6.3 64 11 17.3 8:l MeOH-1,Cdioxane 49.7 40.9 125.0 57.6 75.3 16.7 60 29 18.3 3:l MeOH-1,Cdioxane 45.1 37.3 107.9 52.7 62.8 15.4 58 29 20.8 6:l MeOH-MCP 48.4 47.1 122.2 ca.52.9 73.9 ca. 5.7 60 ca. 12 ca. 5.1 a Mole fraction of water in the mixture in which the minimum value of AZcl occurs. which involves the following assumptions : (i) the hydrodynamic mobility of the proton is equal to that of K+; (ii) the transference numbers at infinite dilution of K+ and C1- are considered to be equal. The KCI experimental conductance data and the corres- ponding parameters obtained using eqn (7) are given at the end of the above tables. The excess percentage proton conductances in both mixed-solvent systems are listed in table 5 where, for comparison, the same data in the isodielectric solvent mixture (8 : 1 MeOH-MCP)3 and in pure methanol1* are reported. As far as the influence of the molecular structure of the cosolvent molecules on the excess proton mobility is concerned, some qualitative points can be made from the experimental information presented in table 5.First, the percentage of abnormal protonM. Goflredi 1443 8C 16C - 140 E M 0 m N E 28 \ c) 120 100 00 4\ 0 I I 0 L 8 12 16 H,O (wt %) Fig. 1. Variation of in wet methanol-l,4-dioxane mixtures as a function of the water content of the mixture expressed as wt % . 0 , H,O-(8: 1 MeOH-1,4-dioxane); 0, H,O-(3: 1 MeOH-1,4-dioxane). conductance in the two anhydrous isodielectric solvent mixtures, % Imax, is the same within the limits of experimental error.It seems, therefore, as a first approximation that the proton-transfer process in these two different isodielectric solvent mixtures is affected, with respect to pure methanol, in the same way by both cosolvents. Similarly, the dependence of [(AT;f')E%]m,, on the 1,4-dioxane content seems to be in the direction intuition should suggest. Secondly, the resulting values of [(AH,')E %Imin are larger for the methanol-l,4-dioxane mixtures than those for methanol and for the methanol-methylcyclopentane mixture. This trend suggests that the interactions occurring between the proton and the solvent system, when water is present, are qualitatively and quantitatively different depending on the nature and the structural features of the cosolvent.In other words, the change in [(AH,')E %Imin seems to be an indication of the presence in the methanol-l,4-dioxane-water mixtures not only of direct hydrogen-bonding interactions between the proton and the free water molecules but also of specific intermolecular interactions between methanol, 1,4-dioxane and water itself. Moreover, as can be seen, the two oxygen atoms in the 1,4-dioxane molecule have a distinct influence on the magnitude of the water content corresponding to the minimum value of the limiting molar conductance of hydrogen chloride. The converse is true for the MeOH-MCP mixture since here MCP, beyond a partial modification of the degree of alcohol aggregation, seems to reflect a strong increase of 48-21444 Ion-pairing of HC1 in Methanol-Dioxane Mixtures 01 0.2 0.3 2 2 0.4 0.5 W c 85.6 85.2 - - I 84.8 Eo --.. 84.4 L~ 84.0 "0 Q2 0.4 x2 Q6 Q8 1.0 Fig.2. Excess volumetric properties of the methanol-l,4-dioxane system, of the H,O-(8: 1 MeOH-l,6dioxane) (c) and of the H,O-(3: 1 MeOH-l,$-dioxane) ( d ) mixtures, as a function of the mole fraction of the mixture component considered as solute; namely, (a) and (b) l,Cdioxane, (c) and ( d ) water, (e) methanol. the proton affinity of water, probably through a screening effect on the hydroxy group of the alcohol molecules. As is well known, when a cosolvent is added to an electrolyte solution we observe a modification of the macroscopic properties of the solvent system as well as a change of the transport properties of the solute. Generally, in fact, the cosolvent can either produce structural changes in the solvent system or perturb the electric field of the ions.Consequently, owing to the large number of different interactive situations involved, only a qualitative explanation can be presented. We will now attempt to rationalize the possible intermolecular interactions occurring in the solvent by analysing its volumetric properties. Results for the molar excess volumes VE and the excess apparent molar volumes VT of the MeOH-I, 4-dioxane, the H,O-(8 : 1M. Gofredi 1445 MeOH-l,4-dioxane) and the H,O-(3 : 1 MeOH-l,4-dioxane) mixtures are summarized in fig. 2, where the volumetric properties are plotted as a function of the mole fraction of the mixture component considered as solute. Molar excess volumes VE were evaluated from the density data according to where the volume of a solution containing 1 mol of the component, is the molar volume of the pure component, xi its mole fraction and Vs is Vs = Z Mixi/d (12) where Mi is the molecular weight of each component and d is the density of the solution.Available literature density data for the MeOH-l,4-dioxane system were used.lo-12 The apparent molar volumes V, where obtained from the equation where the subscripts 1 and 2 refer to solvent and solute, respectively. As can be noted from fig. 2A the excess volume V E of the methanol-l,4-dioxane system is negative over the whole concentration range, with a minimum located around X , = 0.3. This volume contraction, a phenomenon which is quite common for mixtures of hydrogen-bonding liquids, indicates an increase of structure that can be understood as an enhanced self-association of the amphiprotic component or as a consequence of the formation of aggregates between the two components.Since the 1,4-dioxane molecule via its polar group is, to a certain degree, a hydrogen-bond acceptor, new hydrogen bonds with the methanol molecules can be expected. Obviously, this may lead to significant competition with the strong hydrogen-bonding effect causing self-association in the methanol. Since apparent and partial molar quantities are very sensitive to the local average environment of a molecule in solution we have also considered the concentration dependence of the excess apparent molar volumes V f = V$- Vg. The partial molar volumes Vo, were obtained by simple extrapolation of V, to infinite dilution, as can be seen from t g .2B, while the concentration dependence of V r are reported in fig. 2C. As can be seen, the mixing process occurs for both components with a reduction in volume, but the values of V t are smaller for methanol than those for 1,4-dioxane. We observe, moreover, that in the methanol-1 ,4-dioxane mixtures with mole fraction 1,4-dioxane > 0.5 the V,f of methanol is essentially constant. On the contrary, that of 1,4-dioxane shows an appreciable change which, however, is much smaller than that observed in the methanol-rich mixtures. Such behaviour can be qualitatively interpreted in terms of strong intermolecular interactions between methanol and 1,4-dioxane molecules in the concentration region below x, = 0.5. Let us consider now the concentration dependence of the viscosity excess calculated as where vS is the solution viscosity, log vid = C xi logqi and qi is the viscosity of each component i.Fig. 2A(b) exhibits this dependence. At first glance, it is interesting to note that bgvE is negative over the whole composition range. This trend, which points to a fluidity increase of the solvent mixtures with respect to the ideal behaviour, seems to disagree with the estimated VE values. These two excess parameters, in fact, generally show an opposite trend. The only significant difference in these two curves is that in terms of percent variation the higher deviation is for the viscosity excess than for the volume excess. For instance, in the region of maximum deviation the percentage variation of (14) log V E = log V S - log Vid1446 Ion-pairing of HCl in Methanol-Dioxane Mixtures log qE with respect to VE is 4: 1.It is interesting to note, moreover, that both curves show a minimum located at approximately the same composition value, x, = 0.3, which corresponds to a molar ratio between methanol and 1,4-dioxane of 2: 1. This may be interpreted on the assumption that intermolecular hydrogen-bonding interactions occur via the two other oxygens of 1,4-dioxane and the hydroxy group of the methanol molecules so that a 2: 1 associate is formed. Of course, there may exist in solution other types of 1,4-dioxane-methanol aggregates; in fact, these thermodynamic excess properties are essentially related to the most stable entity that intermolecular interactions can promote.It may be concluded that the increase of packing density and fluidity stems from the specific 1,4-dioxane-methanol interaction, which in turn seems to induce a decrease of the self-association of methanol molecule (increase of fluidity). We consider next the dependence of the excess volume properties of the water- methanol-l,4-dioxane mixtures analysed here on the water concentration. The density values used in the calculations were evaluated from eqn (1) and (2) only in the range of low water composition, in which the filled equations reproduce all the experimental data to within the experimental precision. Moreover, we have considered these three-component systems as binary systems in which water is the cosolvent.The volume excess data are shown in the three sections of fig. 2, where curves (c) and ( d ) refer to the H,O-(8: 1 MeOH-1,4-dioxane) and to the H,O-(3 : 1 MeOH-l,4-dioxane) mixtures, respectively. The reduction of volume on mixing water with the methanol-l,4-dioxane mixtures can be rationalized in terms of a structural increment of the medium as a consequence of hydrophilic interactions. This in turn offers a simple explanation of the peculiar dependence of (Am)min on the water concentration in these methanol-1 ,4-dioxane mixtures. The conspicuous difference of the water content of the solvent mixture in which the minimum value of (Am)HC1 occurs (see table 5), between the methanol-methylcyclopentane and the methanol- 1,4-dioxane system can, in fact, be related essentially to the acceptor-donor abilities of the solvent system.Since in the methanol-l,4-dioxane mixtures the water-dependence of (Am)min increases as the 1,4-dioxane content of the mixture increases it is reasonable to assume an increase in the donor ability of the methanol molecules involved in the 2: 1 associate formation (structure 1, possible average conformation of a methanol-1,4- dioxane associate) with respect to those isolated or self-associated. Thus, change transfer might occur on the hydroxy group of methanol when the hydrogen-bonded associate is formed. This picture could be invoked, moreover, to account for the higher value of [(AH,')E %' Imin for both methanol-l,4-dioxane mixtures with respect to methanol- methylcyclopentane and to pure methanol. In fact, we can assume that the existence in the solution of this molecular entity might also be responsible for the excess proton mobility found in these systems. The research described here has been supported by the Ministry of Public Education (M .P.I.).M. Gofredi 1447 References 1 R. De Lisi, M. Goffredi and V. Turco Liveri, J. Chem. SOC., Faraday Trans. 1, 1978,74, 1096. 2 R. De Lisi, M. Goffredi and V. Turco Liveri, J. Chem. SOC., Faraahy Trans. I , 1979,75, 1667. 3 N. Anno, R. De Lisi, M. Goffredi and V. Turco Liveri, J. Chem. Soc., Faraday Trans. I, 1982,78,3101. 4 R. De Lisi and M. Goffredi, Electrochim. Ada, 1972, 17, 2001. 5 R. M. Fuoss and K. L. Hsia, Proc. Nut1 Acad. Sci. USA, 1967,58, 1808. 6 B. E. Conway, J. O’M. Bockris and H. Linton, J. Chem. Phys., 1956,24, 834. 7 L. Thomas and E. Marum, 2. Phys. Chem., Teil A, 1926, 121, 153. 8 D. A. Lown and H. L. Thirsk, Trans. Faraday Sac., 1971,67, 132. 9 E. S. Amis, J. Phys. Chem., 1956,60,428. 10 A. M. Skodni, N. K. Levitskaya and V. A. Lozhnikov, Electrokhimiya, 1967,3, 1045. I 1 A. M. Skodni and N. K. Levitskaya, Ukr. Khim. Zh., 1968,34, 330. 12 E. S. Amis, A. R. Choppin and F. L. Padgett, J. Am. Chem. Soc., 1942,64, 1207. Paper 6/ 1 199; Received 13th June, 1986

ISSN:0300-9599    

DOI:10.1039/F19878301437

出版商:RSC    

年代:1987

数据来源: RSC

摘要:

J . Chem. SOC., Faraday Trans. 1 , 1987,83, 1449-1467 Emulsion Polymerization of Butyl Acrylate Kinetics of Particle Growth Ian. A. Maxwell, Donald H. Napper and Robert G. Gilbert* School of Chemistry, University of Sydney, New South Wales 2006, Australia The kinetics of the seeded emulsion polymerization of butyl acrylate have been studied dilatometrically at 50 "C, using both persulphate and y- radiolysis for initiation. Kinetic data obtained for steady state polymeriza- tion and for the approach to (and, for y-radiolysis, relaxation from) a steady state, gave rate coefficients for entry, exit, termination and propagation, using data-fitting techniques which have been shown to yield unambiguous rate parameters from such results. The system is found to have a high average number of free radicals per article (ranging from 3 to 9).The rate coefficient for exit (desorption) of free radicals from the particles is in accord with transfer/diffusion theory ; exited free radicals tend to undergo subsequent re-entry. The entry rate shows that the free-radical capture efficiency is high but significantly less than 100% at the higher initiator concentrations studied; this is quantitatively interpreted by considering competing aqueous- phase events. Values for the propagation and termination rate coefficients at 60% weight fraction polymer are reported, the latter being an order of magnitude less than predicted by current theory. The determination of accurate and unambiguous values for the rate coefficients involved in different types of free-radical polymerization is of importance both scientifically and technically, There are very few reported rate coefficients deduced from emulsion polymerization experiments in such a way as to be free from elaborate model-based assumptions. In previous publications, we have shown1-* that seeded emulsion poly- merization systems offer especial advantages in the determination of particle growth kinetics: (1) one can use well characterized monodisperse seed particles; (2) data interpretation is free of all but the most general model assumptions; (3) use of y-radiolysis to initiate polymerization allows one to follow the kinetics after removal from the source of initiating free radicals and to combine these relaxation kinetics with the results from chemical initiation, so that the rate parameters are over-determined; and (4) the volume fraction of polymer can be varied as desired over any value greater than (say) 30%.Although the heterogeneous nature of an emulsion polymerization system, and the concomitant increase in the kinetic events which must be taken into account, implies that the data analysis is complicated, nevertheless it has been shown1* that one may obtain unique (and frequently overdetermined) values for all relevant rate parameters, by a combination of sufficient data; earlier proofs1*2 that such unique information can be obtained from appropriate emulsion polymerization kinetic data are extended in this paper. One notes that the study of seeded particle growth kinetics is of interest in its own right (for example, particle growth from seed is an important industrial procedure) and, moreover, an understanding of particle growth is an essential prelude to elucidating particle nucleation kinetics.In the present study we apply the methodology for determining rate parameters from seeded growth kinetics to butyl acrylate. The seeded emulsion polymerization of this monomer will be seen to be characterized by an average number of free radicals per latex particle (R) which is typically greater than one-half. Under such circumstances, it is 14491450 Emulsion Polymerization of Butyl Acrylate possiblelv to determine rate parameters for the following microscopic events: free- radical entry into particles @), free-radical exit (desorption) from particles (k), bimolecu- lar termination (k,) and propagation (k,).In addition, one can find the value of the 'fate parameter'l? 5 9 6 y a, which expresses the relative importance of heterotermination and re-entry for desorbed free radicals in the aqueous phase. Experimental data on these various parameters can first be compared with the predictions of basic theoretical models, and second (in cases where reliable models do not yet exist) be compared with equivalent results for a number of other monomers, with the aim of understanding general trends (a necessary prelude to reliable theory). Our present studies are carried out at 50 "C, for comparison with results of other systems at this temperat~re.'-~ There have been few previously reported studies of the emulsion polymerization of butyl acrylate.l*12 The most complete have been reported by Capek et al.1° These authors report values for k,/k,, H and particle number for an unseeded system at 70 "C.Comparison with our value of k,/k, (a quantity which should not be strongly temperature dependent) will be made at a later stage. Experimental Materials Butyl acrylate (Fluka Analytical Grade) was distilled under reduced pressure at 53 "C. The surfactant employed for seed latex preparation and for seeded growth was Aerosol MA (sodium dimethylpentyl sulphosuccinate, Cyanmid Australia Industrial Grade). All other reagents were analytical grade. Preparation of Seed Latex A prerequisite for the methods used here is the absence of secondary nucleation during particle growth. The following recipe yielded a seed which, under all conditions reported below, did not show secondary nucleation: butyl acrylate (375 g), Aerosol MA (5.36 g), NaHCO, (1.27 g) and water (1000 g) were heated to 75 "C under a nitrogen atmosphere, with mechanical agitation until emulsification was effected. Initiator (K,S,O,, 1.24 g) was dissolved in water (104 g), heated to 75 "C and added to the emulsion in the reaction vessel.Polymerization was allowed to continue at 75 "C for 15 h, and the resulting latex dialysed for 3 days, with frequent changes of water. Electron microscopy was used to characterize the latex and to check for the occurrence of secondary nucleation during seeded kinetic runs. It was necessary to harden the latex prior to electron microscopy to avoid melting under the beam.This was achieved by U.V. crosslinking: the latex was diluted to 10-50 ppm and exposed, with constant agitation, for 24 h (in a quartz cell of 2 mm radius) to U.V. radiation using a mercury lamp. To obtain a uniform coating on the electron microscope grid the copper grids were prepared with a coating of Parlodion followed by a thin coating of carbon. The mean particle radius was found using this means to be 67 nm, with a standard deviation of 5 nm. Ultracentrifugation yielded a particle radius of 70 f 4 nm, in good agreement with the value obtained from electron microscopy. A mean value of 70 & 5 nm was adopted for the radius. The following densities of monomer and polymer at 50 "C were determined using an Anton Paar densitometer; polymer: 1.026 g (determined from the densities of seed latex of varying dilution) and monomer: 0.869 g ~ m - ~ .Kinetic data were obtained dilatometrically, using standard techniques for both chemical initiation3 and y-radiolytic initiati~n.~ The contraction factor (0.157 cm3 g-l) calculated from ideal mixing using the above densities of monomer and polymer is in excellent accord with that determined gravimetrically (0.156 cm3 g-l). The solubility ofI. A . Maxwell, D. H. Napper and R. G. Gilbert 1451 monomer in water at 50 "C (6 x mol dm-3)13 is sufficiently low for it to be unnecessary to take it into account7 when calculating fraction conversion from observed meniscus height. All seeded kinetic runs were checked (using electron microscopy as above) to ensure that no secondary nucleation had occurred.Theory The theory used to interpret the data in seeded emulsion polymerization systems is an extended', 5 9 g 9 l4 Smith-Ewart15 mechanism which appears to be sufficiently flexible to be applicable to any heterogeneous polymerization system. Details have been given elsewherely 5 9 7 9 9916 and are briefly summarized below; however, the method of data reduction presented in the subsequent section represents a significant advance on those published hitherto.'? The time dependence of fractional conversion of monomer to polymer (x) is given by dx/dt = (kp N, C, Mo/Ngm) A Aii (1) where N, is the number density of latex particles, C, is the monomer concentration in the particles, M, is the molecular weight of monomer, N is Avogadro's constant and gm is the initial mass of monomer per unit volume of reacting mixture. In the present study, all runs were carried out in Interval I1 (i.e.in the presence of monomer droplets), so that C, is essentially constant. A is found from the number of latex particles containing i free radicals, Ni : which in turn is determined by the generalized Smith-Ewart equations: a = C iN& Ni (2) dN,/dt = P[N,-~ -Nil + k[(i+ 1) N,+' - iNi] + c[(i+ 2) (i+ 1) Ni+J - i(i- 1) Ni. (3) Here p, k and c are, respectively, the (pseudo-) first-order rate coefficients for free-radical entry, exit and termination; the last is related to k, by c = k,/NY,, where Y , is the swollen volume of the latex particle. The effect of aqueous phase events arising from exit is to make p depend on A: Here a is the fate parameter, whose value (- 1 < a < + 1) is determined by the rate coefficients for the various aqueous-phase mechanisms involving the desorbed free radicals (viz.heterotermination and re-entry) and pA is the component of p arising solely from the initiator. The value of p A depends on the initiator concentration [I] (if chemical) or radiation type and intensity (if initiation is by radiolysis). Moreover, a depends on how initiation is induced: we use the subscript c to denote the value of a with a given chemical initiator, and the subscripts i and r, respectively, for y-radiolysis in the 'insertion' mode (i.e. when inserted into the radiation source) and for the 'relaxation' mode (i.e. when removed from the y source): a,, oq and 4. Eqn ( l j ( 4 ) [with the infinite set of eqn (3) truncated at an appropriately large value by the ' instantaneous termination ' approximation given el~ewhere~~] are readily solved numerically, using the Gear method.This method inter alia enables one to take into account any variation of both c and k with x during Interval I1 (owing to the change in y S ; one has c = K,/NV, and k cc Vi2j3, for reasons given below). Because all runs are in Interval 11, and thus involve no change in volume fraction of polymer, k, is independent of x. We now prove that appropriate emulsion polymerization kinetic data will at worst uniquely determine, and at best overdetermine, all rate-determining quantities. This proof extends those given earlier.lq Proof: Consider first 'zero-one' systems: those where c % p, k.It has been shown3 that then A 5 & and Ni = 0, i >, 2, and that c is not rate-determining. Under such circum- p = PA+&R. (4)1452 Emulsion Polymerization of Butyl Acrylate stances one needs to determine pA([I]), k and a,. (There is extensive experimental and theoretical evidencel9 5 9 that or, = + 1 in all systems, and the data analysis does not require a separate determination of 4.) The dependence of pA on [I] (excluding any background thermal component3 which can be determined separately) can be empirically represented by two parameters: e.g. as pA([I]) = One thus has five rate parameters to be found: IC, d, k,, k and a,. Now, in a zero-one system, the observation of a plateau in the steady-state rate as a function of [I] implies that this plateau corresponds to H = 4, which enables kp to be directly determined.3 There are thus four remaining parameters.Each chemically initiated run yields a dependence of x on time which can be represented3 as a long-time slope (a) and intercept (b) : x(t --+ co) = at + b. The experimental dependence of these quantities on [I] would yield at worst three independent pieces of information. Next, consider the amount of information available in a y-radiolysis run in the relaxation mode. For a zero-one system this yields essentially single-exponential decay,5 i.e. a slope and intercept. Thus one has a total of five independent pieces of experimental information, which overdetermines the remaining four rate parameters. Next consider systems (such as the present) where A may exceed 4 termination is now rate-determining, and thus we have six independent rate parameters to be found from the data: K , c, d, k,, k and ac.Moreover, in systems which are not zero-one, one will not in general observe a plateau in the [I] dependence of the steady-state rate,' and thus k, cannot be determined separately from the others. Now, in favourable circumstances k, is available either from bulk measurements or, for preference, from emulsion systems through direct e x . measuremenf.l7 However, such will not always be the case, and we now show that nevertheless all rate parameters can be uniquely determined from seeded emulsion-kinetic data alone. For chemically initiated data with non-zero-one systems, one has the same information available as for the zero-one case considered above, uiz.at worst three separate pieces of independent information. However, for a non-zero-one system in the y-radiolysis relaxation mode, the higher ~i implies that (provided both k and c are rate-determining) multiple exponential decay must be seen,2 thus yielding at least three separate pieces of information (effectively the decay time constants for two exponentials and the ratio of their contribution). Thus the data yield at worst six independent pieces of information, which are sufficient to determine unique values of all rate-determining parameters. This completes the required proof. The detailed method of deducing the actual values of these rate parameters from the data will be given in a later section. Results Fig.1 shows typical dependences of conversion and rate (dxldt) on time for chemically initiated seeded emulsion polymerizations in Interval 11. These data show a significant steady-state region (constant dxldt), with a suitably long approach to this steady state. As shown in the preceding section, such an approach to steady state is an important part of the arsenal for deducing unique rate coefficients. Table I shows the results of all chemically initiated runs. Included herein are the results for a run with zero initiator concentration (to provide data on the background thermal rate). Runs that were duplicates except for varying amounts of added surfactant (BA27, BA29) show insignificant variation in slope and intercept. As part of the data analysis, one needs to know C,, as in eqn (1).This can be obtained by a number of rneam29 39 The optimal method for this particular system was found to be the kinetic technique: from the value of fractional conversion at which the dxldt curve falls below its steady state value, which is expected to correspond to the transition from Interval I1 to Interval I11 (i.e. when the monomer droplets are exhausted). The values of C , so obtained for all kinetic runs are given in table 1. The value of C , so deduced shows no systematic trend with [I] or N, and lies within k0.1 mol dm-3 ofI. A . Maxwell, D . H. Napper and R. G. Gilbert 1453 ‘0 time/min Fig. 1. Conversion (x) and rate (dxldt) as functions of time for run BA32. Scatter for &/dt is because of numerical differentiation of data.3.2 mol dm-3; we thus take C , to be 3.2k0.1 mol dm-3 (corresponding to a weight fraction of polymer equal to 0.58). It is essential to establish that the observed approach to steady state is not an artefact due to the presence of inhibitor (i.e. that it is governed by pA, k and c and not by spurious free-radical trapping species present in the aqueous phase). This can be established as follows. A simple test is to check that the kinetics of approach to steady state do not depend on the degree of oxygenation of the ~ystem.~ A better test is through the use of y-radiolysis, as first suggested and carried out by Ballard et al.’ One carries out a radiolytically initiated run at a y-flux such that the observed rate of approach to steady state is comparable to that seen in chemically initiated studies.The reactor is then removed from the source, allowed to relax to the background thermal rate, and then reinserted into the source. Fig. 2 shows the results of the two insertions (both for Interval 11) superposed (i.e. with a shift of origin for the second insertion). The two x(t) curves are virtually identical, in particular in regard to the approach to steady state. Now, any effects due to inhibitor in the aqueous phase would be at a maximum during the initial part of the first insertion; the subsequent exposure to y-radiolysis would significantly reduce the amounts of inhibitors present in the aqueous phase. The second insertion should thus give the inhibitor-free approach to steady state in the presence of greatly reduced inhibitor concentrations.The essential identity of the x(t) for the first and second insertions in these studies shows that inhibitor effects during the initial insertion can be safely neglected. Since the rate of approach to the steady state observed in this radiolysis experiment is similar to those for the chemically initiated runs, inhibitor effects with the latter can also be safely neglected. Fig. 3 shows a typical y-radiolysis run including data in the ‘relaxation’ mode (after removal from the source). It is apparent that the relaxation rate is sufficiently slow to allow accurate measurement. These data will be used, in conjunction with table 1, to deduce the various rate parameters as described in the next section.Table 1. Results of chemically initiated studiesa run [1]/10-4 mol dm-3 s-' b/s-l C,/mol dm-3 R N,/ lOI5 dm-3 [S]/mol dmP3 BA19 BA22 BA23 BA24 BA25 BA26 BA27 BA28 BA29 BA30 BA3 1 BA32 BA35 BA36 BA37 BA38 1.57 9.46 6.9 1 1.67 0.248 0.191 14.5 0 14.3 14.0 8.92 1.81 1.12 1.27 12.3 0 4.46 7.34 6.36 4.53 3.47 3.10 8.05 3.18 8.59 8.54 7.18 - 0.06 - 0.07 - 0.09 - 0.06 - 0.06 - 0.03 - 0.09 - 0.07 -0.10 -0.08 -0.09 3.17 3.26 3.24 3.25 3.16 3.25 3.19 3.18 3.21 3.18 3.20 3.14 3.14 3.19 3.19 3.1 1 4.6 8.0 7.2 4.9 3.7 3.3 9.0 3.4 9.2 8.7 7.9 9.7 9.4 9.9 9.6 9.8 9.7 9.2 9.5 9.6 9.9 9.7 23 17 18 17 18 2.1 x 10-3 2.3 x 10-3 2.1 x 10-3 2.1 x 10-3 1.5 x 10-3 2.1 x 10-3 6 .4 ~ 10-4 2.1 x 10-3 2.5 x 0 0 0 0 0 0 0 a a and b are the long-time slope and intercept, respectively. A values were calculated using k, = 450 dm3 mol-' s-' (final fitted value).Runs with [I] = 0 are background thermal runs. Conversion x is a fraction (units of intercept are thus s-l). Runs where no slope and intercept reported were for purpose of determining C,. [S] is mol of added surfactant per dm3 of solution.I. A . Maxwell, D. H . Napper and R. G. Gilbert 0. 0. X 0% 1455 3.1 n 7 1.05 Z. 7 G 3 2 4 6 t ime/m in Fig. 2. Conversion/time curves for y-radiolytically initiated run, for first (1) and second (2) insertion into radiation source, with the origin of the second superimposed upon that of the first, to show absence of artefactual inhibition effects on approach to steady state. (1) is in Interval I1 (plotted as x); (2) is in Interval I11 [plotted as -ln(l -x)]. 1 I I I ‘0 10 20 30 time/min t Fig.3. Conversion as function of time for y-radiolysis run; arrow denotes times of removal and reinsertion from source.1456 Emulsion Polymerization of Butyl Acrylate Discussion Deduction of Rate Parameters from Data It was shown above that, if one has steady-state rates and the approach to steady state for both chemically initiated runs and relaxation in y-radiolytically initiated runs, one can deduce unique values for all rate-determining parameters. We now consider the actual implementation of this procedure. Note that the technique given here is a considerable refinement of that given previously.2 The data reduction could of course be carried out by some brute-force fitting of the whole corpus of data. However, it is better to devise a systematic procedure for this.Our systematic technique relies on the fact that k and c must be independent of initiator concentration and initiator type. In summary, the method is as follows. The fitting of chemically initiated data requires values of k, and a,. Initially assume sets of values for these two quantities which cover the likely range. Then with each set of k, and a,, fit the chemically initiated data. One does this in an iterative fashion, by (i) assuming that for the high-[I] (i.e. high-a) data one can at first put k = 0; each slope and intercept then yield pA and c (the value of p A depending on [I]; (ii) taking the resulting value of c, and using this value on the low-[I] data, fit to k and p A (for each [I]); (iii) with these first estimates of k and c, now fit the data over all [I], iterating until the values of PA@]), k and c converge; (iv) then fit the y-radiolysis relaxation data, using the k and c obtained from the chemically initiated data with each assumed set of values of k, and a,; it follows from the discussion given above that only a unique set of k, and a, can fit the whole corpus of data, and thus one obtains unique values of all the rate-determining parameters.The details of the technique are as follows. (1) Note first that non-linear least-squares fitting of data is carried out in a standard way, by minimization of the sum of the squares of the differences between experimental and calculated quantities; in the present case these quantities will be the observed and calculated slopes and intercepts for each run.The requisite minimization is carried out numerically, using an appropriate algorithm such as the Simplex method. (2) Assume a set of values of k, and a, which cover the physically reasonable range: for example, one might take 12 values of k, between 100 and 1200 dm3 mol-1 s-l, and three values of a, (- 1, 0 and + 1). This would give 36 different sets of k, and a, (k, = 100 dm3 mol-l s-l, a, = - 1 ; k, = 100 dm3 mo1-1 s-l, a, = 0; k, = 100 dm3 mol-l s-l, a, = + 1 etc.). Each assumed value of k, then yields a value of the constant A of eqn (1). (3) For each chemically initiated run, calculate the value of the long-time slope (a) and intercept (b), for each assumed value of k, (i.e. of A). N.B. : if the slopes approach a limiting value at high [I] that is independent of [I], the system is zero-one, and the methods given for this case1 are to be used instead. (4) Consider first the a and b values for high [I], i.e.for the highest A. As a first approximation, one can assume that this system follows the pseudo-bulk equations with ( 5 ) This approximation is always applicable if A is sufficiently high, and in any event will always provide a suitable starting point (except for a zero-one system, but that can be much more easily handled using techniques given elsewherel). Now, in a chemically initiated system, the initial value of ~i is 0. From eqn (5) and (l), assuming that all parameters are independent of R and that we are in Interval I1 (i.e. CM is also constant; Interval I11 is handled by a trivial extension1), and taking the limit of long time, one therefore has the following expressions for pA and c, for each slope and intercept :’ k = ():I6 diildt = PA - 2cii2.c = (A/2b)ln2; pA = 2 ~ ( a / A ) ~ . (6) Each run will thus yield a first estimate of c and of pA[I], for each assumed k,.I. A . Maxwell, D. H . Napper and R. G. Gilbert 1457 400 600 800 kp/dm3 rno1-I s-' Fig. 4. Values of k as a function of assumed k, values, as fitted from chemically initiated data: 0, a = 1 ; 0, a = 0; A, a = - 1 . Lines are drawn to indicate trends. ( 5 ) Using the mean value of c obtained from step (4), the low-[I] (i.e. low-il) chemically initiated data are then fitted as follows. Each observed slope and intercept is non-linear least-squares fitted, following the method of (1) above, to the values of p A and k, given this mean value of c and the corresponding assumed values of k, and a,, for each (k,, a,) set.The calculated values of the slope and intercept are found by exact numerical solutions of the generalized Smith-Ewart equations, eqn (1)-(4). Each chemically initiated run will thus yield values of k and of pA[I] for each set of assumed k, and a,. Then take the mean value of k from this low-[I] fitting. (6) Use this mean value of k from step ( 5 ) to re-fit the high-[I] data to yield new values of pA and c ; now the fitting is through exact numerical solution of the generalized Smith-Ewart equations, as in step (5). (7) Now take this mean value of c to find new values of pA and k, for all the data, i.e. for all values of [I] studied; this fitting is carried out as in step (5).(8) Take the mean value of k and again use all the data, as in step (6), to find pA and c. Continue iterating steps (6)-(8) until no significant changes in the mean values of k and c are found. This yields values of pA[I], k and c for each (k,, a,) set. It is useful at this point to reduce the overall scatter in these values from plots of k and c values against assumed k, and a,, as shown in fig. 4 and 5 , and select the smoothed k and c values for each k, and a, value from such plots for the next steps. (9) The next steps will yield the correct set of k, and a, from fitting the y-radiolysis data. An important part of this fitting is to have an estimate of the background thermal entry rate coefficient pT.l' 3 9 A first estimate will be obtained from a run in the absence of chemical initiator [calculated, for each (k,, a,; set, by using the final values of c from steps (6)-(8) and fitting the thermal run as in step (7)].Now, it is not uncommon for the value of the intercept b from a thermally initiated run to have a fairly high uncertainty, which can introduce larger uncertainties in the rates from the y results. It is therefore best to use an extrapolated value of the zero-[I] intercept, obtained by plotting the observed intercepts against [I], and using a linear least-squares fit to give the best1458 Emulsion Polymerization of Butyl Acrylate 0 Fig. 5. Values of c as a function of assumed k, values, as fitted from chemically initiated data. 0, a = 1 ; 0, a = 0; A, a = - 1 .Lines are drawn to indicate trends. 1 2 [I]/ 1 0-3 mol dmd3 Fig. 6. Points: intercepts of chemically initiated runs for different [I]. Line: line of best fit (for determining accurate intercept for [I] = 0). estimate of the zero-[I] value of b. This yields a zero-[I] value of the intercept that has a much lower uncertainty than (but in the range of) that obtained from the runs with [I] = 0. The procedure is illustrated in fig. 6. At this point we consider the y-radiolysis data, which should include both theI. A . Maxwell, D. H. Napper and R. G. Gilbert 1459 steady-state slope in the source, and the time dependence of x(t) in the relaxation mode. To fit these data it is necessary to have a value of a in the absence of initiator: %. There is considerable evidence2* 5 7 that a, is always + 1.In the next step we show how to calculate the initial conditions for the system in the relaxation mode, following removal from the source. The initial conditions are the values of the populations N,(t) of the Smith-Ewart equations, eqn (3), given that the system at t = 0 is in the steady state achieved within the radiation cavity. (10) We are given the steady-state rate in the source; this furnishes the initial conditions for the relaxation results, as follows. For each assumed k , use the in-source steady-state rate to calculate the corresponding value of A, denoted Ti: , from eqn (3) with the left-hand side replaced by 0; methods for this have been given elsewhere.16 Next, for each value of k and c arising from each (k,, a,) set, calculate the value of the radiative (in-source) entry rate coefficient, pi.This is achieved by solving the following implicit equation * @El, (pi, k , C) = R?’ where is obtained from the steady-state so1ution16 of the Smith-Ewart equations, eqn (3). The numerical solution of eqn (7) to yield pi can be carried out by any number of methods for solving non-linear equations, given a suitable means for evaluating the left-hand side, i.e. of calculating iiSs given pi, k and c. Note that in solving these equations it is not necessary for this particular step to know the value of a, since we only wish to find the steady-state Ni: thus the procedure of Ballard et d . 1 6 for the a = 0 case can be used. (1 1) Given the values of pi, k and c from step (lo), for each value of k, and a,, the initial values Ni(t = 0) are calculated by the method of Ballard et Again, because we know p = pi, the simpler procedure for a = 0 is employed.(12) For each set of k, and a,, with the resulting values of k, c and p A = pT, and using the values of Ni(t = 0) from step (10) as initial conditions, a full numerical solution of the generalized Smith-Ewart equations [eqn (1)-(4)] is carried out with the value of a, (one expects q = + 1 ; see above). (1 3) It will be found from step (1 1) that only a single set of values (or rather a small range of values) of the various rate parameters arising from k, and a, will fit the observed time dependence of x ( t ) in the relaxation mode; this is illustrated in fig. 7. (4) Finally, variations of pA, k and c with particle size must be taken into account with Interval I1 runs (i.e.with the conditions used in the present study), since significant particle growth occurs during the portion of the kinetic run used for data analysis. One notes that important features of the emulsion polymerization kinetics of some monomers can only be properly understood by taking such size variations into account.’ These variations can lead to a change in dx/dt even after initial transients have relaxed: a pseudo-steady state rather than a true steady state. Such effects can be quantified in terms of an ‘acceleration’ These variations are as follows. Exit: there are excellent theoreticall* and experimentall, grounds for assuming that k is inversely proportional to swollen particle surface area.Termination : the relation c = k,/NT/, implies that c is inversely proportional to swollen particle volume. Entry: the dependence of pI on particle size has not been properly established; one can put p I c;c r; for some exponent g which can be justified a posteriori. The final step in the data fitting is thus to re-fit using the above exact numerical solution method, with now these variations of k, c and pA with Vs (and thus with conversion x) incorporated specifically. Preparatory to this, one notes that the present butyl acrylate system shows what appears to be a true steady state, ciz. a significant time period where dx/dt assumes a nearly constant value (see fig. 1). I.e. one has a z 0 for the present system. Now, since there is a significant increase in swollen volume over this regime, there must a significant decrease in both k and c.If pr were also to be independent of particle size (or to change i (7) which is zero if dx/dt is constant.1460 Emulsion Polymerization of Butyl Acrylate 0.01 1 "0 5 timelmin Fig. 7. y-Radiolysis relaxation : experimental (broken line) and various calculated (full lines) conversion/time curves (after removal from source, with x = 0 set at instant of removal) for various values of k, and a,: (1) a, = - 1, k, = 320 dm3 mol-l s-l; (2) a, = 1, k, = 440 dm3 mol-1 s-l; (3) a, = - 1, k, = 300 dm3 mo1-1 s-l; (4) a, = 1, k, = 400 dm3 mol-l s-l. in the opposite way to k and c), one could not then obtain the observed negligible value of ZL. A negligible 7i could only be obtained if pI is assumed to be a decreasing function of rs.Since a dependence of ps on the reciprocal of the swollen radius (pS K rs2) has been reportedlo as consistent with data on styrene systems, and in addition such a dependence can be justified by a 'colloidal entry' theory for p,20 such a dependence is not unreasonable, and is adopted here. The final data fitting therefore is in terms of initial values of pI, k, c and of rs (denoted p9, ko, co and rg) with the following size dependences: pI = pi (rg/rs)2; k = P(rg/rs)2; c = ~O(r!/r,>~. For notational simplicity we report the initial values (ko etc.) only in the following, and (since the variation with rs is not very great) drop the zero superscript. By these means, unique values of the various rate parameters are obtained.Deduced values of k, c, k,, kp, pT and a, for the present butyl acrylate system, as well as earlier results on butyl methacrylate,2 are given in table 2. Note that the rate parameters and various deduced quantities for the latter monomer have been recalculated from the original data, using various improved techniques for data fitting as discussed here and Values of p A for each [I] studied are given in table 3, along with pI. The dependence of pI on [I] in table 2 is represented empirically by a power law: pI = ~ 4 1 1 ~ . The value of the exponent d gives an indication of whether or not the capture efficiency is 100% (from any deviations of d from unity). Fitting of the data to a model-based form for p,[I] will be given in the next section; the quantities p, y etc.given in table 2 apply to this interpretation. From these values of k,, the steady-state values of A can be computed for each initiator concentration employed; these are shown in table 1. It is of interest that butyl acrylate is indeed a high-z system, with 3 d A,, d 9 under the conditions studied.I. A . Maxwell, D. H. Napper and R. G. Gilbert 1461 Table 2. Rate parameters deduced for butyl methacrylate [recalculated from data of ref. (2)] and butyl acrylate (present paper) at 50 "Ca _ _ _ _ _ ~ quantity butyl methacrylate butyl acrylate ~ _ _ _ _ ~ ~ particle radius nm range of [I] studied mol dm-3 range of iFs k,/dm3 mol-l s-l d k/s-l C,/mol dm-3 wt fraction of polymer CIS-' k,/dm3 mol-1 s-l a, PTIS-' B/mol dm-3 y/dm3 mol-1 s-l k,/dm3 mo1-1 s-' kI/s-l [M,,]/mol dm-3 r ([I] = lo-, mol dm-3) ([I] = mol dm-3) K at which study carried out 39.5 0.3-1.7 2 x 10-5-10-3 600 2.46 0.74 3.8 0.6 7 x 10-3 3 x 10-3 1.0 x 103 2 x 10-3 2.0 x 104 0.6 x 10-5 3 x 10-3 I x 10-3 0.5-1 1.6 x 2 x 106 14 70 3-9 450 21 0.83 3.2 0.6 10-5-1.5 x 10-3 3 x 10-3 5 x 10-4 0.75 x 103 3 x 10-3 1.1 x 105 1.6 x 10-5 4 x 10-3 0.5-1 0.7 x 4 x 106 6 x 4 a Chemical initiator is K,S,O,.pA = pT+pI, where pT is the background thermal term and pI is the component arising solely from aqueous phase chemical initiator. The dependence of pI on [I] are represented by (i) the empirical equation pI = where pI is in s-l and [I] is in mol dmP3, and (ii) eqn (1 1). Table 3. Component of entry rate coefficient arising from chemical initiator plus background thermal entry bI) ; component arising from initiator decomposition alone (PI) ; capture efficiency for PI ~~ ~~~~~~ radical capture efficiency [1]/10-4 mol dm-3 pA/10-, s-l ~ ~ / 1 0 - ~ s-' (% 1 14.5 14.3 14.0 9.46 8.92 6.9 1 1.67 1.57 0.248 0.191 0 7.9 7.6 7.0 6.7 5.6 4.2 2.2 2.0 1.3 1.0 0.34 7.6 7.3 6.7 6.3 5.3 3.8 1.9 1.7 0.96 0.65 0 32 33 29 40 35 34 71 66 (246) (204)1462 Emulsion Polymerization of Butyl Acrylate Capture efficiencies f for each [I] are also given in table 3, calculated from the relation f = pI/(2kd[I]/N,), where the term in parenthesis is the hypothetical value of pI assuming that every free radical from initiator decomposition is captured by a particle; here k, is the initiator decomposition rate coefficient, taken to be k, = 1.3 x s-l at 50 O C Z 1 The values listed in table 3 for the lowest [I] studied are > 100% (a phenomenon also seen with butyl methacrylate2). This apparent anomaly is no great cause for concern, since the rate of decomposition of persulphate (kd) is not accurately known in the systems under study. It is, however, important to note that the capture efficiencies given in table 3 are significantly less than unity for higher initiator concentrations and, as expected, systematically decrease with increasing [I].This is exactly as has been observed with other monomer^,^-^* 5-7 the lower capture efficiencies being ascribed to aqueous phase homotermination of entering species. Quantitative interpretation will be given in the next section. It is of interest to note that none of our data show a Tromssdorff effect (a rapid rate ascribed to a reduction in kJ.However, the results of Capek et aZ.l0 (at 70 "C, not the 50 "C used here) do shown an extensive period of rapid polymerization for their highest initiator concentrations, which at first sight might appear to be a Trommsdorff effect. Note, however, that their highest initiator concentrations are considerably greater than ours. We find that the rate parameters of table 2 (with c assumed constant), when used to calculate x(t) at the [I] values of Capek et al., in fact predict a period of very rapid growth qualitatively similar to that reported by these authors. We therefore ascribe their period of rapid growth simply to a high entry rate coefficient coupled with a small (but essentially constant) termination rate coefficient, rather than to a true Trommsdorff effect.Mechanistic Interpret ation We now consider the deduced values of the various rate coefficients, with a view to mechanistic interpretation. Entry As stated above, the data show clearly that capture efficiency can be less than 100%. It has been shown6? that a suitable model for taking non-unit capture efficiencies into account is through the following set of reactions: (8) (9) k1 I + R' 2R' inert products kt &q ke R' +particle 3 entry (10) where eqn (8) represents both initiator decomposition and aqueous phase propagation to form a free radical (R') which is able to undergo entry, eqn (9) is aqueous phase homotermination, and the k , in eqn (10) is the second-order rate coefficient for free-radical entry of a free-radical entity derived directly from the initiator.One then finds:6y9 (1 1) where y = k2,/2kt, aq and B = kt,aq kI/k2,. Fitting of the dependence of p r on [I] given in table 3 then yields the p and y values given in table 2. As has been found in a number of other systems,l? 2 * 6, 7 9 eqn (1 1) provides an excellent fit to the observed [I] dependence of pI, as shown in fig. 8 (although, as noted el~ewhere,~ there is a large uncertainty in the final values of and y so obtained). PI = ~ { ( w + W l ) * - N )I. A . Maxwell, D. H. Napper and R. G. Gilbert 1463 0 5 15 [I]/ l 0-3 mol dm-3 Fig. 8. Line: fit of eqn (1 1) to p A for different initiator concentrations, with values of p and y for BA as given in table 3; points: experimental values.The values ofpand y themselves do not lead to immediate physical interpretation (since they are ratios of rate coefficients), but they can be used to determine k, = 2#ly and (given a value of kt, as) k, = (2 ykt,aq)k The values of k, for both monomers are similar: 0 . 6 ~ and 1.6 x dm3 mol-l s-l for butyl methacrylate (BMA) and butyl acrylate (BA), re- spectively. Both values are significantly greater than those found for styrene systems with similar characteristicsB (0.5 x lomg dm3 mol-1 s-l). The greater value for the BA and BMA systems may be ascribed (at least in part) to the greater k, for these monomers (see table 2) compared with that of styrene at the same temperature3 (260 dm3 mol-1 s-l). This is because reaction (8) represents a combination of initiator decomposition (the initiator is the same in all cases), aqueous phase propagation and aqueous phase termination.The greater value of k, for BA and BMA compared with styrene suggests that the former will attain the critical degree of polymerization required to enter a particle more quickly than can styrene (note that the solubilities of all three are comparable), thereby rationalizing the greater k,. The greater value of k, for BA compared with that for BMA may or may not be significant (in view of the large uncertainty in this quantityg). Consider now k,. To calculate this quantity, we assume here a value of 7 x lo7 dm3 mol-l s-l for kt,aq (as applicable to styrene) for both BA and BMA. One sees immediately that the values of k, for both BA and BMA (4x lo6 and 2 x lo6 dm3 mol-l s-l, respectively) are similar (it is not possible to assess the significance of the differences in k, between these monomers because of the moderately large uncertainty in this quantity9).The similarity (within a factor of 4) of the k, values deduced for BA, BMA and styrene suggests that these quantities are comparable for these three systems. The differences in the quoted values may be real (although less than the uncertainty), and can be readily ascribed to either or both of the colloidal effects discussed above. Further measurements on colloidal size and surface charge densities are needed to clarify this point.1464 Emulsion Polymerization of Butyl Acrylate The quantitative value of the capture efficiency is of course determined by the value of pA.However, it is clearly useful to consider also a more qualitative criterion for the entry rate coefficient: is capture efficiency high or low? This question has been considered by Hawkett et aZ.,22 who showed that an indication of capture efficiency could be obtained by considering the various competing effects involved in reactions (8) and (9), leading to a quantity r which is the ratio of the rates, for an aqueous-phase free radical, of termination and propagation to a sufficient degree of polymerization to enter a particle: Here, z is the critical degree for polymerization for entry and [Ma,] is the aqueous phase concentration of monomer. If l7 % 1, then capture efficiency is low; the capture efficiency is high if < 1.With the values of [Ma,] given in table 2, and assuming for BA and BMA that z = 3 (as for styrene, which is of comparable solubility), one obtains the values of F given in table 2. These predict that capture efficiency should be high at the lowest initiator concentrations studied, and should be significantly less than unity at the highest [I]. This is in accord with the results given in table 3 . As has been found with styrene22 and methyl metha~ylate,~ the qualitative model used to develop eqn (12) is successfully able to predict whether capture efficiency is high or low in the present system. Terminat ion The value of k, obtained for butyl acrylate at the weight fraction polymer studied (wp = 0.58, corresponding to a volume fraction 4, = 0.53) is 8 x lo2 dm3 mo1-I s-l.This is five orders of magnitude less than typical values at zero conversion (zero weight fraction). This dramatic decrease can be ascribed to the consequences of chain entanglements. The value of k, at wp = 0.58 can be compared with current theories for this quantity. Of these theories, recent direct mea~urements~~ 23 of kt(wp) for methyl methacrylate over a very wide range of conditions have shown that the theory of Soh and S ~ n d b e r g ~ ~ is (with certain reinterpretation^^^) able to give semi-quantitative agreement with experi- ment for conditions where ‘residual termination’ is the main contributer to k,. Residual termination occurs at weight fractions of polymer so high that chain entanglements severely hamper centre-of-mass motion; as a result the primary mechanism for bimol- ecular termination becomes the encounters that occur between growing macroradicals as a consequence of propagation.The Soh-Sundberg theory yields (13) (14) where aRMS is the root mean-square end-to-end distance per square root of number of monomer units (reported as 0.64 nm for BMA25 and adopted here for BA), j c = xcO/24,, x,, being the critical degree of polymerization for entanglement of pure polymer as x,, = 200 for this monomer) and& being an efficiency factor. Using the value off, = 0.2 found for methyl metha~rylate,~~ one finds that the predicted value of k,(residual) is 7 x lo3 dm3 mol-l s-l. This is an order of magnitude greater than experimental value. This discrepancy cannot be assigned to experimental uncertainty arising from the insensitivity of our data to the value of k,: although it is hard to assign precise uncertainties to our results because of the complex nature of the data-fitting procedure, we found that (for example) it was impossible to provide an adequate fit to our data using a value of k, of twice that given in table 2, all other rate parameters being allowed to vary freely.The order-of-magnitude discrepancy between the Soh-Sundberg theory and our results could perhaps be ascribed to uncertainties in the various parameters (e.g. the theoretical expression is quite sensitive to the values of x,, andf,) or to inadequacies in the theory. This point clearly requires further investigation. k,(residual) = f,no2aRMs Nk, C,/jk CJ = r-l [In (r3 &/n7r3/2)]4; r2 = 3/(2j, akMs)I.A. Maxwell, D. H. Napper and R. G. Gilbert 1465 Propagation The value of kp found here (450 dm3 mol-l s-l) is typical for monomers of this type at this temperature. 26 Literature values of k,/k, for B A at 30 "C are ca. 10-3.26 These values refer to zero added polymer, and so would be expected to be considerably smaller than our value (0.6), notwithstanding the difference in temperatures. We note also that our value for kp/k, is much greater than the range given by Capek et al. :lo 10-1-10-2 at 70 "C, although the latter value is for a wp similar to that used here. The difference could be ascribed to a larger value of k, at the higher temperature due to decreased viscosity. However, a number of factors imply that the results of Capek et al.may not be completely reliable, for the following reasons. (1) No attempt was made to obtain unique rate parameters, as in the methodology used in the present paper; instead, a procedure due to G a r d ~ n ~ ~ with many model-based assumptions was employed. (2) Exit was ignored; as is seen from table 2, k and c are comparable, and thus ignoring a first-order loss process (k) can lead to serious errors in the value of second-order loss rate parameter (c) so deduced. (3) Apparently it was assumed that capture efficiency was 100% in their data reduction; this is unlikely to be correct at the high initiator concentrations used by Capek et al. Exit The value determined for the exit rate coefficient for the latex used here is k = 3 x s-l (table 2). This can be compared with that predicted from theory.An extensive series of studies1? 2* 28 has shown that the transfer/diffusion model for exit, as quantified by Nomura and Harada,l* gives excellent quantitative agreement with a wide range of data. We therefore compare the prediction of this theory with the present results. The transfer/diffusion theory states that exit takes place when a monomeric free-radical unit (arising from transfer) diffuses through and away from the latex particle. In the present system it is easily shown that the rate-determining steps are transfer and diffusion away from the particle, in which case theory yields: where D, is the diffusion coefficient of the exiting free radical of degree of polymerization z) in the aqueous phase, and k,, is the rate coefficient for transfer to monomer.Assum- ing the same values of z and D, as for styrene1 (zD, = 10-lo m2 s-l) and using k,,/k, = 5 x s-l. This is in excellent agreement with the value of 3 x s-l deduced from our data. Use of the values of k,,/kp for either methyl or ethyl acrylate would not alter the predicted value of k substantially. (as applicable to butyl methacrylate26), one predicts k = 4 x Fate Parameter The butyl acrylate system shows a value of a in the range 0.5-1, indicating that desorbed (exited) free radicals re-enter a particle rather than undergoing heterotermination with a free radical arising directly from the initiator. This is a characteristic that this monomer shares with BMA2 and methyl metha~rylate,~ and contrasts with styrene5 (where - 1 < a f 0). The simplest explanation for this behaviour lies in the difference in the respective propagation rate coefficients: all monomers in the first group have a relatively high k, (2 450 dm3 mol-1 s-l), whereas that for styrene is lower (250 dm3 mol-l s-l 3).This suggests that the condition governing whether a desorbed free radical will re-enter or heteroterminate in the aqueous phase is determined in part by how quickly it propagates in the aqueous phase. If it propagates rapidly to a higher degree of polymerization, it is more likely to re-enter because of increased hydrophobicity.1466 Emulsion Polymerization of Butyl Acrylate Conclusions In the present paper we have shown how an established methodology1v2 (with various technical improvements given in detail above) is able to yield precise and unambiguous values of the various rate parameters describing particle growth in an emulsion polymerization, free of most model-based assumptions.For the butyl acrylate system studied here, these kinetic parameters are the rate coefficients for (i) the entry of free radicals into latex particles, (ii) the desorption of free radicals from the particles, (iii) bimolecular termination in the particles, (iv) propagation and (v) the ‘ fate parameter’ describing the fate of desorbed free radicals in the aqueous phase. The various rate parameters are given in table 2. From these results we conclude that for the butyl acrylate system (i) free-radical capture efficiency is high, but significantly < 100% at high initiator concentrations; this capture efficiency is consistent with a theory based on the entering free radicals being colloidal in nature; (ii) the desorption rate coefficient is quantitatively in accord with the prediction of transfer/diffusion mechanism ; (iii) the termination rate coefficient at the weight fraction of polymer in the system (wp = 0.6) indicates that the termination mechanism is ‘ residual ’ (high chain entanglement would imply that growing macroradicals can only encounter each other through propagation), although the experimental value of k, is an order of magnitude less than predicted by current theory; (iv) k, is in the range of that of similar monomers at this temperature; (v) the dominant fate of desorbed free radicals is to re-enter latex particles rather than to heteroterminate with free radicals arising directly from initiator.The present studies suggest that various theoretical approaches (as described above) can be reliably used to predict many of the rate parameters required to model particle growth in emulsion polymerization systems. The financial support of the Australian Research Grants Scheme and of the Australian Institute for Nuclear Science and Engineering are gratefully acknowledged, as is the generous provision of facilities by the University of Sydney Electron Microsccope Unit. Ms M. Adams is thanked for assistance with the kinetic runs using the y-radiolysis facility, and Mr G. Russell for helpful discussion. References 1 2 3 4 5 6 7 8 9 10 11 12 13 14 15 16 17 R. G. Gilbert and D.H. Napper, J. Macromol. Sci. Rev. Macromol. Chem. Phys., 1983, C23, 127. L. F. Halnan, D. H. Napper and R. G. Gilbert, J . Chem. SOC., Faraday Trans. I , 1984, 80, 2851. B. S. Hawkett, D. H. Napper and R. G. Gilbert, J. Chem. SOC., Faraday Trans. 1, 1980,76, 1323. S. W. Lansdowne, R. G. Gilbert, D. H. Napper and D. F. Sangster, J. Chem. SOC., Faraday Trans. I , 1980,76, 1344. B. C. Y. Whang, D. H. Napper, G. Lichti, M. J. Ballard and R. G. Gilbert, J. Chem. SOC., Faraday Trans. 1, 1982, 78, 1 1 17. I. A. Penboss, D. H. Napper and R. G. Gilbert, J. Chem. SOC., Faraday Trans. I , 1983,79, 1257. M. J. Ballard, D. H. Napper and R. G. Gilbert, J. Polym. Sci., Polym. Chem. Ed., 1984, 22, 3225. S. J. McCarthy, E. E. Elbing, I. R. Wilson, R. G. Gilbert, D. H. Napper and D. F. Sangster, Macro- molecules, 1986, 19, 2247. I. A. Penboss, R. G. Gilbert and D. H. Napper, J. Chem. Sac., Faraday Trans. 1 , in press. I. Capek, J. Barton and E. Orolinova, Chem. Zvesti, 1984, 38, 802. T. O’Neill and J. Hoigne, J . Polym. Sci., Part A I , 1972, 10, 581. J. Snuparek, J. Appl. Polym. Sci., 1979, 24, 909. M. K. Lindemann, in Vinyl Polymerization, ed. G. E. Ham (Marcel Dekker, New York, 19671, vol. 1, part 1. J. Ugelstad and F. K. Hansen, Rubber Chem. Technol., 1976,49, 536. W. V. Smith and R. H. Ewart, J. Chem. Phys., 1948,16, 592. M. J. Ballard, R. G. Gilbert and D. H. Napper, J. Polym. Sci., Polym. Lett. Ed., 1981, 19, 533. M. J. Ballard, R. G. Gilbert, D. H. Napper, P. J. Pomery and J. H. O’Donnell, Macromolecules, 1986, 19. 1303.I . A . Maxwell, D. H . Napper and R. G . Gilbert 1467 18 M. Nomura, in Emulsion Polymerization, ed. I . Piirma (Academic Press, New York, 1982);-M. Nomura 19 G. Lichti, R. G. Gilbert and D. H. Napper, J. Polym. Sci., Polym. Chem. Ed., 1983,21, 269. 20 P. J. Feeney, R. G. Gilbert and D. H. Napper, to be published. 21 I. M. Kolthoff and I. K. Miller, J. Am. Chem. SOC., 1951, 73, 3055. 22 B. S. Hawkett, D. H. Napper and R. G. Gilbert, J. Poiym. Sci., Poiym. Chem. Ed., 1981, 19, 3173. 23 M. J. Ballard, D. H. Napper, R. G. Gilbert and D. F. Sangster, J. Poiym. Sci., Poiym. Chem. Ed., 24 S. K . Soh and D. C. Sundberg, J. Polym. Sci., Polym. Chem. Ed., 1982,20, 1299; 1315; 1331; 1345. 25 J. D. Ferry, Viscoelastic Properties of Polymers (Wiley, New York, 3rd edn, 1980). 26 Polymer Handbook, ed. J . Brandrup and E. H. Immergut (Wiley, New York, 2nd edn, 1975). 27 J. Gardon, J. Polym. Sci., Part A I , 1968, 6, 623; 643; 665. 28 G. Lichti, D. F. Sangster, B. C. Y. Whang, D. H. Napper and R. G. Gilbert, J. Chem. SOC., Faraday Trans. I , 1982,78, 2129; M. Adams, D. H. Napper, R. G. Gilbert and D. F. Sangster, J. Chem. SOC., Faraday Trans. 1, 1986, 82, 1979. and M. Harada, J. Appl. Polym. Sci., 1981, 26, 17. 1986, 24, 1027. Paper 6/1235; Received 18th June, 1986

ISSN:0300-9599    

DOI:10.1039/F19878301449

出版商:RSC    

年代:1987

数据来源: RSC

摘要:

J. Chem. SOC., Faraday Trans. 1, 1987,83, 1469-1476 Adsorption of Benzene on Acidified Alumina Edouard Garbowski" and Michel Primet Institut de Recherches sur la Catalyse, Laboratoire Propre du C.N.R.S. conventionnt a r Universitt Claude Bernard, Lyon I, 2, avenue Albert Einstein, 69626 Villeurbanne Ckdex, France U.v.-visible reflectance spectroscopy of adsorbed benzene has been used to study the surface modification of alumina containing platinum and chloride. On unchlorinated alumina there are a few sites strongly adsorbing benzene. After chlorination by CCl,, the Lewis acidity of alumina becomes so strong that some adsorbed benzene molecules are ionized and cracked : further condensation occurs leading to coke precursors. Chlorination of platinum- containing alumina leads to a support whose ionizing power depends upon the metal loading : a platinum-aluminum-chloride complex is formed by CCl, interaction with support and metal.This complex is adsorbed on the strongest Lewis acidic centre, preventing benzene ionization. Addition of HC1 to a chlorided platinum alumina leads to a superacid solid-state catalyst. This support reacts immediately with adsorbed C,H,, giving the benzenium C,Ht cation that slowly transforms into the bulky arenium cation. Platinum alumina reforming and isomerization catalysts are always acidic : the acidity, which is due to the presence of Lewis sites on alumina, may be modified by incorporation of a halogen, i.e. chlorine or fluorine. These types of catalyst are always bifunctional because they contain an acidic function (support) and a metallic one (noble metal).Sometimes the acidity is so strong that ionization occurs with electron-rich hydrocarbons. For example, benzene adsorbed on silica-alumina gives rise to carbonium ions,l whereas on H-ZSM-5 and mordenite zeolites cation radicals are formed.2 On the other hand, benzene adsorption has been thoroughly studied for it is a well known molecule from a spectroscopic point of In that sense adsorption of C,H, on several aluminas containing platinum and/or chlorine would be worthwhile in order to obtain information on the adsorptive capacity and reactivity of aluminas towards aromatic molecules. In addition, information about the nature of acidic sites may be obtained. Experimental GFS 400 y-alumina from RhGne-Poulenc has been used in this study.The characteristics of the support are reported in a previous paper,5 as well as experimental conditions for chlorine incorporation by CCl, and reduction of platinum in order to obtain well dispersed metal particles. Thus the following samples have been studied : pure activated A1,0,; unchlorinated Pt/A1,0, catalysts containing 1 wt % Pt ; Pt/Al,O, catalysts chlorinated by CCI, and containing 0.1, 0.5 and 1 wt % platinium; 1 wt % Pt/AI,O, sample chlorinated by HCl at 573 K. Pure benzene was obtained by using the freeze-pumpthaw technique. The liquid was then stocked in a small flask containing activated 4A molecular sieves. Adsorption of C6H6 onto the solids was performed at room temperature by connecting the U.V. cell to the flask via the vacuum line under the vapour pressure of benzene at 298 K, i.e.10 kPa. After allowing a few minutes for equilibrium the cell was removed out of the line. U.v.-visible diffuse reflectance spectra obtained with an Optica Milano CF 4DR 14691470 Adsorption of Benzene on Acidified Alumina r I . . . . . . . . . . . . 5a o 4 90 4 00 31 0 220 wavelength1 nm Fig. 1. A, Reflectance spectra of benzene adsorbed on y-alumina: (a) A1,0, background, (b) C,H, adsorption at room temperature, (c) and (d) desorption at 298 and 372 K, respectively, for 1 h. B, Reflectance spectra of benzene adsorbed on Pt-loaded alumina: (a) background due to metal-particle diffusion, (b) C,H, adsorption at room temperature, (c) desorption at 373 K for 1 h. spectrophotometer using MgO as reference.Spectra were recorded on log R-l scale (apparent aborbance).6 Results y- Alumina After activation under 0, and vacuum at 773K, the solid shows a small charge transfer band in the U.V. range at 260 nm. That absorption has been previously observed7 and assigned to coordinatively unsaturated surface A13+ or 02- ions. After adsorption of C6H6 at room temperature the spectrum shows the fine-structure band of benzene at 256 nm and erosion of the alumina charge transfer (fig. 1). The peaks are very slightly displaced with respect to that of C6H6 in dilute solution. The excess of benzene was removed by desorption at room temperature for 1 h. The intensity of benzene bands strongly decreased, and the fine structure was no longer observed. Moreover, the maximum was slightly shifted towards lower wavelength.After desorption up to 573 K the charge transfer band of alumina recovered its initial intensity and no benzene was detected spectroscopically. Unchlorinated Pt-Alumina Samples Adsorption of benzene was performed on the catalyst containing 1 wt % of metal. After H, reduction at 773 K, the solid strongly scattered light because of the presence of veryE. Garbowski and M . Primet 1471 I . . . . . . . ' . . . . 580 490 40 0 310 22 1 wavelength/nm Fig. 2. Reflectance spectra of benzene adsorbed on chlorinated alumina: (a) A1,0, background, (b) and ( c ) C,H, adsorption after 5 min and overnight, respectively, at room temperature. small metal particles :8 these particles are responsible for the general absorption reported in fig.1 (B). After benzene adsorption the spectrum was strongly modified: the presence of C,H, was easily observed by the fine-structure band, whereas the metal was less opaque to light because the background decreased substantially. The intensity of the benzene band reached the same value as that for alumina. With our spectrometer it was relatively difficult to measure with accuracy the difference of absorption at so high an optical density. After desorption at room temperature a lot of benzene was eliminated, whereas between 373 and 473 K adsorbed benzene was removed from the surface and the background absorption caused by metal diffusion recovered its initial intensity. y-Alumina Chlorinated by CCI, After chlorination by CC1, for 1 h at 573 K, the U.V.spectrum showed that the A1-0 charge transfer band is slightly decreased. When benzene was adsorbed on such a solid, it turned yellow immediately. The spectrum then showed pronounced modification of the surface. The fine-structure band observed at 256 nm is still present, but with half the intensity of the same band obtained with pure alumina. The other band responsible for the yellow colour is rather broad and has its maximum at 450 nm (fig. 2). After desorption at room temperature the amount of adsorbed benzene was very weak, whereas the visible band shifted to 400-430 nm and a small band appeared at 330 nm. With time there was also a modification of the spectrum in that both bands at 430 and 330 nm increased. However, it should be noted that the intensity ratio 1430/1330 was continuously decreasing, suggesting that these bands are not due to the same species.Pt/AI,O, Samples Chlorinated by CCI, at 573 K We have previously studied the interaction of CC1, with platinum on al~mina.~ It has been shown that chlorination of platinum-supported alumina leads to oxidation of metal particles and gives formation of [PtC1J2--like species, as revealed by u.v.-visible spectroscopy. We also proposed that some Pt(AlCl,), species may exist when both platinum and the alumina support were chlorided. Samples containing 0.1, 0.5 and 11472 Adsorption of Benzene on AcidiJied Alumina I . . . . . . . . . . . . 580 4 90 400 310 2: wavelengthfnm Fig. 3. Reflectance spectra of benzene adsorbed on chlorinated Pt-Al,O, samples : (A) 0.1, (B) 0.5 and (C) 1 wt% .(a) Solid background after chlorination by CCl,, (b) C,H, adsorption at room temperature, (c) desorption at room temperature. wt% of platinum were activated, reduced, then chlorinated by CCl, according to the procedure described previ~usly.~ Differences in behaviour were revealed when benzene was adsorbed on the three solids. With the 0.1 wt % metal catalyst, there was almost no difference in comparison with chlorinated alumina, Two bands at 430 and 330 nm were observed as well as the fine-structure band at 256 nm. By desorbing at room temperature almost all the adsorbed benzene disappeared, because the 256 nm band vanished, but visible bands were still present and even increased slightly: moreover, a third band appeared at 550 nm (fig.3). With the second sample (0.5 wt% metal) the behaviour was different. Although the adsorbed benzene produced the fine-stucture band, the bands at 330 and 400-430 nm were very weak (fig. 3). Desorption at room temperature increased the intensity slightly, but the fine structure at 256 nm was no longer observed. Absorption of C&6 on the 1 wt % Pt/Al,O, sample did not produce any band in the visible part of the spectrum. Only the fine structure at 256 nm due to C,H, adsorbed in molecular form was observed. After desorption at 298 K, there was removal of almost all adsorbed benzene. Pt/Al,O, Samples chlorinated by HCl One sample of the 1 % platinum alumina, chlorided by CCl, at 573 K was contacted with 20 Torr t dry HCl. The solid was then heated at 473 K for 2 h and evacuated at room temperature for 5 min.Benzene was then adsorbed at 298 K on the resulting solid. Bands developed immediately in the U.V. range at 256 (fine structure), 330 and 370 nm. With t 1 Torr = 101 325/760 Pa.E. Garbowski and M . Primet 1473 580 490 400 310 220 wavelength/nm Fig. 4. Reflectance spectra of C,H, adsorbed on 1 wt % Pt-Al,O, sample chlorinated by CCl, and contacted with HCl: (a) initial background, (b)-(d) adsorption of C6H6 after 1 min, overnight and 4 days, respectively. time there was also an evolution of the spectrum and after 4 days a strong band was observed at 430nm. In contrast, the bands at 550 or 460nm were never observed Discussion All the results show that alumina is a strong acidic support and that in some cases there is reactivity towards benzene.On pure GFS 400 alumina a lot of benzene is adsorbed at room temperature, but the molecules are physically adsorbed because most are eliminated by pumping off at 298 K. For those molecules which are irreversibly adsorbed at room temperature one cannot see fine structure in the U.V. spectrum. The position of the maximum is shifted to longer wavelength from 256 to 260 nm. The excited state of non-polar molecule like benzene is more polarizable or may have a permanent moment dipole when adsorbed on a polar surface, leading to bathochromic shift. This small shift, previously reported by Leermakers et aZ.,9 indicates that London dispersion forces are important in the adsorp- tion process).1o This phenomenon explains the bathochromic effect and also the fact that the fine structure is buried.Note, however, that the last traces of benzene require a (fig- 4)- 49 FAR 11474 E. Garbowski and M . Primet 573 K desorption temperature for removal. A few sites adsorb the electron-rich molecule very strongly: the sites have electron-deficient character and are Lewis acidic centres. This suggests that some sites are relatively strong, as we noted previously from i.r. studies of pyridine ad~orption.~ The optical properties of hydrogen-reduced solids loaded with platinum are quite different. The increasing background with frequency is due to light scattering by small Pt particles. The broad maximum at ca. 280 nm is due to plasmon resonance absorption : when benzene was adsorbed on such a solid, the absorption of metal decreased, revealing a strong electronic interaction.The amount of C6H6 adsorbed, which was deduced by subtracting the F(R) values, was estimated to be the same as for the alumina sample. Besides, when desorbing at 298 K, then at 373,423 and 473 K, the amount of adsorbed C6H6 was the same as for alumina. One can thus conclude that the metal particles are not blocking access to Lewis centres adsorbing benzene, at least with low metal concentration. After chlorination by CCl,, the alumina is a strong oxidizer. The bands appearing at 330, 400-450 and 550nm cannot be assigned to C6H6, for benzene is completely transparent in the visible range. The positions of the bands suggest more readily that ions are formed, i.e. some aromatic molecules have been altered by adsorption with stronger Lewis centres.The stronger maximum which is a little flattened around 400-430 nm is merely due to triphenyl carbonium ion Ph,C+, which is very stable and has been observed when adsorbing triphenylmethanell or triphenylcarbinol12 on various supports. The second maximum at 550 nm has also been observed for benzene dissolved in chlorinated solvents after y-irradiation.13 The monomer radical cation C6Hi or dimer (CeH6); have bands at 560 nm, and at 460 and 920 nm, respectively. The observed band at 460 nm may well correspond to (c~H6):; moreover, in the near-i.r. a second band has been observed at 1100 nm. Although it is not at the same frequency, one can reasonably deduce that on the strongest Lewis acidic centres some benzene molecules are ionized despite having such a high ionization potential.Moreover, during desorption the band at 460 nm disappeared and a band at 550 nm was sometimes observed (due to radical cation monomer). The small peak observed at 330nm is not due to radical cation. The position corresponds well to (C6H6-H)+ formed by protonation of benzene by a strong acid.', In that sense it could signify that during adsorption of C,H6 onto Lewis sites there may be some ring opening, leading to production of protons: alumina thus turns into a Brsnsted acid. For Pt/Al,O, samples chlorinated by CCl, the situation is rather complicated. For very low metal complex contents, the acidic support behaves exactly the same way as for metal-free chlorinated alumina. This means that minute amounts of chloro complex do not inhibit benzene ionization.On the other hand, with the 0.5 wt% and a fortiori for the 1 wt% Pt/Al,O, sample, there is no reactivity of C6H6 towards Lewis-acidic centres: the sites appear to be blocked by the chloro complex. In fact it is likely to think that the strongest Lewis acidic centres that ionize C& are involved in the formation of chloro complex. It has been shown5 that chlorination by CCl, leads to nascent AlCl, (Lewis centres) and oxidation of platinum particles into PtCl,, giving the following species : c1 cl\pt / cl\Al / c1 / \c/ \Cl/ \Cl Subsequent C6H6 adsorption cannot give rise to carbonium ion. In that sense the disappearance of benzenium cation as a function of metal loading is a measure of the number of strongest Lewis sites.In our case 1 wt% Pt is sufficient to inhibit the strongest Lewis acidity. A simple calculation, taking into account the surface area ofAdsorption of Benzene on AcidiJied Alumina 1475 alumina, i.e. 200 m2 g-l leads to a value of ca. 0.15 sites nm-2, i.e very few. However, this number corresponds to the number of strongest sites determined by pyridine adsorption, i.e. 0.6 sites nm-2. The factor of four is not to be considered strictly since U.V. reflectance spectroscopy determines the number of oxidizing sites only, whereas i.r. spectrometry gives the total number of stronger sites adsorbing pyridine at 350 *C.* More interesting is the result of C,H, adsorption on the platinum alumina containing 1 wt % metal, and contacted with HCl after CCl, chlorination.The major peak detected immediately after adsorption is the 330 nm band, i.e. that corresponding to the C,HT cation: the benzene molecule has been protonated by a strong Brransted acid. Thus the Lewis-acidic centres may react with HCl giving a superacid catalyst : -Al-OAlCl, + HCl + Al-OAlC1,- + H'. The proton obtained is very labile and reacts with the base C,H,, giving C,H:. The cation turns slowly and cracks itself giving a linear charged species that condenses with other C,H, molecules. Finally species like Ph,C+ are formed, giving an absorbance near 400-430 nm. Coke precursors cannot be avoided. It should also be noted that the hydrocarbon condensation is relatively slow at room temperature, for after 4 days the intensity of both bands at 420 and 330 nm is very high.On the other hand, no band is detected at 460 or at 550 nm. It seems that no oxidizing sites are present, they have all been transformed into Brransted sites by HCl addition. Thus, it is then proved that Lewis sites can easily be converted into Brransted sites on the alumina surface. The proton obtained is nearly free, i.e. a superacid is obtained. The support which had only Lewis sites before any adsorption turns into a Brlernsted solid-state acid, either by HCl addition or by carbocation evolution (uide supra). It is then possible that the whole activity of that type of catalyst is solely due to the proton, which is the actual catalyst. That proton can exchange with gaseous hydrocarbon, leading to a carbocation that can isomerize.One can then understand the role of HCl, which is added in the feed for isomerization or alkylation. Further works concerning HCl addition onto the nC, isomerization activity are in progress. Conclusion The interaction of benzene with alumina containing platinum and/or chlorine has shown several facts. (1) The amount of C,H, irreversibly adsorbed at room temperature is limited, and is connected to the number of sufficiently strong acidic sites. (2) The presence of platinum has little effect on the adsorptive capacity. Metal particles do not seem to inhibit the adsorption: i.e. the metal does not interact directly with Lewis sites (at least at low metal concentration). (3) When chlorinated, alumina turns into a very strong solid-state Lewis acid able to ionize benzene, giving coke precursors.(4) Platinized alumina chlorinated by CCl, gives species on interaction with Lewis centres that prevent benzene ionization. Knowing the amount of platinum deposited onto the Al,O, surface, it is possible to deduce a number of strong acidic centres. (5) Addition of HCl to the platinum-aluminium chloro complex gives a solid-state superacid : the proton obtained is able to protonate benzene, giving the C,HT carbenium ion. It is possible that in the presence of saturated hydrocarbons the proton can give a carbonium ion: this could explain the role of added HCl in the feed during the isomerization of light alkanes on acidified platinum alumina. References 1 V. A. Barachevsky and A. N. Terrenin, Opt. Spektrosk., 1964, 17, 304. 2 P. Wierzchowski, E. Garbowski and J. C. Vedrine, J. Chim. Phys., 1981, 72, 41. 3 L. Doub and J. M. Vandenbelt, J . Am. Chem. SOC., 1947, 69, 2714. 4 L. Doub and J. M. Vandenbelt, J . Am. Chem. SOC., 1949, 71, 2414. 49-21476 E. Garbowski and M . Primet 5 A. Melchor, E. Garbowski, M. Primet and M. V. Mathieu, J. Chem. SOC., Faraday Trans. 1, 1986,82, 6 G. Kortum, Reflectance Spectroscopy (Springer-Verlag, Berlin, 1969), p. 59. 7 E. Garbowski, M. Primet and J. P. Candy, J. Chem. SOC., Faraday Trans, 1, 1983,79, 835. 8 A. Melchor, E. Garbowski, M. Primet and M. V. Mathieu, J. Chem. SOC., Faraday Trans. I , 1986, 9 P. A. Leermakers, H. T. Thomas, L. D. Wels and F. C. James, J. Am. Chem. SOC., 1966, 88, 5075. 1893. 82, 3667. 10 N. S. Bayliss and E. G. Mac Rae, J. Phys. Chem., 1954,58, 1006. 1 1 H. P. Leftin and W. K. Hall, Proc. Int. Congr. Catal., Paris, 1960, (Editions Technip, Park, 1961), 12 L. N. Izmailova and E. I. Kotov, Kinet. Katal., 1977, 18, 488. 13 B. Badger and B. Brocklehurst, Trans. Faraday SOC., 1969,65, 2582. 14 H. Luther and G. Pockels, Z. Electrochem., 1962, 59, 169. vol. I, p. 1307. Paper 611275; Received 24th June, 1986

ISSN:0300-9599    

DOI:10.1039/F19878301469

出版商:RSC    

年代:1987

数据来源: RSC

摘要:

J. Chem. SOC., Faraday Trans. I , 1987, 83, 1477-1484 Effect of Diphosphonates on the Precipitation of Calcium Carbonate in Aqueous Solutions Aglaia G. Xyla and Petros G. Koutsoukos" Department of Chemistry, Physical Chemistry Laboratory and Research Institute of Chemical Engineering and High Temperature Processes, University of Patras, Patras, Greece The effect of two diphosphonates, ethane- 1 -hydroxy- 1,l -diphosphonic acid (EHDP) and 1,2-dihydroxy-l,2-bis(dihydroxyphosphonyl)ethane (DDPE) on the precipitation of calcium carbonate has been investigated. EHDP and DDPE were found to reduce markedly the rate of crystallization at solution concentrations as low as lo-' mol dm-3, both in cases where precipitation was spontaneous and when it took place on well characterized calcite crystals introduced in the supersaturated solutions.In spontaneously pre- cipitating supersaturated solutions, small concentrations of the additives caused a marked increase in the induction period preceding precipitation. Inhibition was mainly at the crystal-growth step by adsorption at the active-growth sites. The adsorption isotherms of EHDP and DDPE showed that these compounds adsorb extensively on to calcite surfaces. Interpreta- tion of the kinetics and adsorption data in terms of the Langmuir isotherm gave good agreement for the 'affinity' parameter obtained by these two met hods. The formation of calcium carbonate in aquatic systems attracts the interest of various disciplines ranging from water and waste-water treatment to oil and gas production.The various processes engaging machinery in contact with water present the problem that the parts involved have a limited time of function owing to the formation of insoluble scale, of which calcium carbonate is an important constituent. The problem is more pronounced at elevated temperatures, since the solubility of calcium carbonate decreases with temperature. The problems associated with the undesirable formation of calcium carbonate scale have spurred research in the last decades towards inhibition or prevention of the formation of this mineral. The use of acids in order to maintain an unfavourable pH for calcium carbonate formation, or for the removal of the scale already formed, may accelerate corrosion of the involved metallic parts and is thus not desirable.Pretreatment of the water with ion-exchange resins, on the other hand, is very costly and requires frequent maintenance. Attention has been focused on the use of water-soluble additives, added in water at very low concentration levels, which were capable of either preventing or drastically retarding the formation of insoluble calcium carbonate. Since the use of phosphate compounds was first suggested,l a large number of scientific papers and patents concerning the effect of various compounds which are inhibitors of calcium carbonate formation have been published.2* Organophosphorus compounds have been tested for the prevention of calcium carbonate scale, and were found to be effective under conditions prevalent in geothermal energy production wells.4 The most effective use of inhibitors requires a thorough investigation of the mechanism of action of those ions or molecules at the calcite-water interface.The present work reports the influence of two diphosphonates on calcium carbonate precipitation from supersaturated solutions. Thus a commercially available, 14771478 Precipitation of Calcium Carbonate well known water additive, ethane- 1 -hydroxy- 1,l -diphosphonic acid (henceforth EHDP) and 1,2-dihydroxy- 1,2-bis(dihydroxyphosphonyl)ethane (henceforth DDPE) have been tested. The latter compound is resistant to hydrolysis at high temperatures5 and shows no toxicity.6 The experimental conditions in the present work were such that spontaneous precipitation occurred. These conditions are more relevant to conditions encountered in cases where undesirable scale is encountered.Moreover, the mechanism of action of the two diphosphonates tested in the present work is investigated by looking at the adsorption properties of these compounds on calcite surfaces. An additional answer sought in this work is whether the inhibitors tested inhibit nucleation or the crystal growth of the solid phase. Experimental The experiments were done in a 0.6 dm3 thermostatted, double-walled Pyrex glass vessel at 25.0 f 0.1 "C. Triply distilled, C0,-free water was used throughout the experiments. Calcium nitrate stock solutions were prepared from analytical-grade tetrahydrate solid salt (Fluka AG) and they were subsequently standardized by ion exchange and by spectrophotometric titrations at 504 nm with murexide indicator using a Pye Unicam SP-460 u.v.-visible spectrophotometer.For each experiment, fresh sodium bicarbonate solutions were prepared from the solid salt (Fluka AG) dried at 105 "C overnight. The inhibitors were added to the working solutions, from stock solutions prepared from pure samples of sodium salts of EHDP and DDPE. Standard potassium hydroxide solutions, 0.1 mol dm-3, were prepared (Merck, Titrisol) and standardized against potassium hydrogen phthalate dried overnight at 105 OC, using phenolphthalein as indicator. The pH of the solutions was measured by the cell: GE 1 working solution I KCl (sat) I Hg,Cl, I Hg. A pair of glass (Radiometer G202C) and saturated calomel (Radiometer K502) electrodes was used. The electrodes were standardized before and after each experiment using NBS primary standard buffer solutions at pH 6.865 (0.025 mol dm-3 KH,P04+0.025 mol dmP3 Na,HPO,) and pH 9.180 (0.01 mol dm-3 borax).* The working solutions for nucleation and crystal growth of calcium carbonate were prepared by mixing equal volumes (0.25 dm3 each) of calcium nitrate and sodium bicarbonate solutions.pH was immediately adjusted at 8.50 by the addition of potassium hydroxide, and remained constant until the formation of calcium carbonate began. The release of protons, accompanying the formation of the solid phase, triggered the addition of potassium hydroxide from the burette (Radiometer, ABU 12) of a pH-stat (Radiometer TT2b). The piston of the burette was mechanically coupled to the recorder (Radiometer REC 51) pen, and the consumption of base as a function of time was followed.At the same time, the calcium-ion activity in solution was monitored by using a calcium ion-specific electrode (Radiometer F-2 1 12) connected to a recorder (GOERZ RE 541) via a constant potential source and an electrometer operational amplifier (MPI-1000), enabling us to read changes as small as 0.05 mV. The calcium ion-specific electrode was calibrated before and after each experiment. During precipitation, at different time intervals (larger as the reaction proceeded to completion) samples were withdrawn, filtered through 0.22 pm membrane filters (Millipore, Bedford, Mass.) and the filtrate was analysed for calcium by spectrophotometric titrations. The solid material on the filters was analysed by scanning electron microscopy (SEM) (IS1 model super 11, 40" tilt angle), infrared spectroscopy (Perkin-Elmer 177 grating spectrophotometer) and, when the quantity was sufficient, by X-ray diffraction spectrometry and analysis for N,, B.E.T.specific surface area (Perkin-Elmer 212D sorptometer with a GOERZ RE 541 recorder and an HP 3370b integrator). The crystallization experiments were repeated by initiating the reaction by inoculationA . G. Xyla and P. G. Kuutsoukos 1479 of the unstable supersaturated solutions with well characterized synthetic calcite crystals (specific surface area 3.3 m2 g-l) prepared according to the method of Reddy and Nanc~llas.~ Addition of the seed crystals took place immediately following pH adjust- ment and the calcium carbonate formation reaction started immediately.In order to understand the interaction between the diphosphonates and the calcite substrate, the adsorption isotherms of both EHDP and DDPE on calcite were studied: 10 cm3 of different concentrations of the diphosphonates were equilibrated with ca. 100 mg of calcite. The ionic strength was adjusted by adding the appropriate amount of potassium nitrate stock solution. The pH of the suspensions was initially 9.2 0.2. The tubes were sealed and were rotated end-over-end at 22k2 "C for 21 h. After the equilibration period the pH dropped in all cases to 7.1. To suspensions were centrifuged and the liquid phase analysed for diphosphonate as follows. The 8 cm3 of the liquid, 0.15 g potassium persulphate were added and the solution was irradiated by a U.V.lamp for 30 min. After the irradiation the aqueous solution was analysed for inorganic orthophosphate.l0* l1 Results and Discussion The solution speciation at various stages of the precipitation process was calculated from pH, the appropriate equilibria equations' and the mass-balance equations for total carbonate, [C],: [C], = [H2C0,*] + [HCO;] + [COi-] + [CaCOt] + [CaHCOz] [Ca], = [Ca2+] + [CaCO;] + [CaHCO,+] + [CaOH+] (1) (2) Zm+z, = Zm-z-. (3) total calcium, [Ca],: and from the charge-balance equation: In eqn (1) and (2) square brackets denote concentrations of the species enclosed and in eqn (3), m+ and m- are the concentrations of the positively and negatively charged species, respectively, and z+, z- their respective charges.The partial pressure of carbon dioxide was assumed to be constant and equal to that of the atmosphere (10-3.5 atm,? corrected for atmospheric pressure changes when necessary). This assumption has proved to be valid, since the volume of the supersaturated solutions was large and the air space above them minimal.12 The system of equations (1)-(3), together with their appropriate equilibrium stability constants, was solved for the solution species, making successive approximations for the ionic strength.ll The Davies equationll was used for the calculation of activity coefficients, yz, of the z-valent ionic species: (4) -logy, = A [ I / ( 1 + I ) - 0.311. In eqn (4) A is taken as 0.5 11 5 for 25 "C and I is the ionic strength of the supersaturated solution. The driving force for the formation of calcium carbonate is the change in Gibbs free energy, AG, for the transition from supersaturated to saturated solutions, and is given by eqn ( 5 ) : RT 2 AG = --lnR.In eqn (5), R is the gas constant, T the absolute temperature and R the solution supersaturation, defined by (Ca2+) (COi-) K: Q = t 1 atrn = 101 325 Pa.1480 Precipitation of Calcium Carbonate Table 1. Precipitation of calcium carbonate from supersaturated aqueous solutions, initiated by the addition of 200 mgdm-3 calcite seed crystals (pH 8.50, 25 "C, [Ca], = [C], = 3.227 x mol dm-3) expt [inhibitor] 'Gcalcite rate no. inhibitor mol dm-3 /kJ mo1-I mol dm-3 min-' 33 78 68 69 71 71A 79 79A - 5.03 DDPE 3.0 - 5.03 DDPE 4.0 - 5.03 DDPE 4.5 - 5.03 DDPE 5.0 - 5.03 DDPE 5.5 - 5.03 EHDP 3.0 - 5.03 EHDP 5.0 - 5.03 - - 85.6 21.4 9.7 2.1 1.7 0.1 21.2 t/min Fig.1. Relative effect of EHDP and DDPE inhibitors on the precipitation of calcium carbonate in supersaturated aqueous solutions, seeded with calcite, 200 mg 1-l; [Ca], = [C], = 3.227 x mol dm-3 pH 8.50, 25 "C. A, No additive; 0, 3 x lo-' mol dm-3 DDPE; a, 5 x mol dmV3 DDPE; 0 , 3 x mol dm-3 EHDP; a, 5 x lo-' mol dm-3 EHDP. where parentheses denote activities of the included ionic species, and K: the thermo- dynamic solubility product of the phase precipitated. In our calculations, the solubility product of calcite was considered, since this phase was the only one formed under our experimental conditions. This finding is in agreement with reports of other workers in the 1iterat~re.l~. l4 Scanning electron micrographs (SEM) of the precipitate obtained in the absence of any additive are shown in plate 1.The rhombohedra1 calcite particles formed had a size distribution from 1 to 50 pm, with a maximum of the distribution curve at a diameter of 20 pm. Calcite particles of a similar size have been reported to precipitate spontaneously from supersaturated calcium carbonate solutions at elevated temperat~res.~~ The experimental conditions for the precipitation of calcium carbonate both in the absence and in the presence of EHDP and DDPE inhibitors are summarized in table 1. In all cases ion-pair formation between Ca2+ and the organophosphorus compounds was taken into account. The rates of precipitation reported therein were obtained by numerical differentiation of the initial part of the curve representing theJ .Chem. SOC., Faraday Trans. 1, Vol. 83, part 5 Plate 1 Plate 1. Scanning electron micrograph of calcite crystals spontaneously precipitated at constant pH (8.50) and temperature (25 "C). A. G. Xyla and P. G. Koutsoukos (Facing p . 1480)J . Chem. Soc., Faraday Trans. I , Vol. 83,part 5 Plate 2 Plate 2. Calcite crystals precipitated in the presence of diphosphonate inhibitors. [Ca], = [C], = 3.227 x lop3 mol dm-3, pH 8.50, 25 "C. (a) In the presence of 3 x mol dm-3 DDPE. (6) In the presence of 3 x mol dm-3 EHDP. A. G. Xyla and P. G. KoutsoukosA . G. Xyla and P. G. Koutsoukos 1481 1.00' d I I 1.0 2.0 3.0 4.0 CT1/ lo-' dm3 mol-' Fig. 2. Langmuir-type kinetics for the precipitation of calcium carbonate in aqueous solutions in the presence of DDPE inhibitor.[Ca], = [C], = 3.227 x mol dm-3, pH 8.50, 25 "C. Table 2. Spontaneous precipitation of calcium carbonate in supersaturated solutions in the presence of DDPE inhibitor ([Ca], = [C], = 3.227 x mol dm-3, pH 8.50, 25 "C) expt [DDPE] AGcalcite rate no. z/minu mol dmV3 /kJ mol-l mol dm-3 64 6 - - 5.03 8.6 80 10 2 - 5.03 8.0 82 25 5 - 5.03 4.2 83 42 8 - 5.03 2.7 a Induction period. variation of total calcium, [Ca],, in solution with time, using Newton's method.ls Both DDPE and EHDP seemed to have a similar effect towards inhibition of calcium carbonate precipitation, as can be seen in fig. 1. Neither inhibitor had any effect with respect to the calcium carbonate phase precipitating, which was in all cases identified as calcite [plate 2(a)].EHDP, however, had a striking effect on the morphology of the precipitated calcite, as can be seen in plate 2(b). It is possible that the spherulitic formations of calcite result from aggregation of the crystals formed in the supersaturated solutions. X-ray diffraction analysis of the surface of the samples, shown in plate 2(b), revealed the presence of the organophosphorus compound. Similar changes in the morphology of the precipitates formed in the presence of EHDP have been reported recently for gypsum and barium ~u1phate.l~~ l8 In experiments where calcium carbonate precipitation was allowed to proceed to ca.1482 0.3C 0.25 N E - 0.20 a L3 -.- 0. t5 0.10 ( Precipitation of Calcium Carbonate 0 0 I I I 5 10 15 Ces/ 1 0-6 mol dm-3 Fig.3. Adsorption of DDPE on calcite. Initial pH 9.2, final pH 7.1 , 25 "C, total volume 0.010 dm3, 100 mg calcite. 50 % , in inhibitor-free supersaturated solutions, additions of EHDP caused an immediate reduction or even stopped the precipitation process, thus indicating that inhibitors act at the crystal-growth stage rather than in the nucleation step.19 As was suggested earlier, inhibitors of crystal growth act by blocking the active growth sites by 21 Assuming a simple Langmuir isotherm for the adsorption of DDPE on calcite, a plot of R,/(R, - Ri), where R, and Ri are the rates of precipitation in the absence and in the presence of DDPE, respectively, as a function of the inverse of DDPE concentration in solution, should yield a linear relationship.Such a plot is shown in fig. 2. The inverse of the slope of this line (2.3 x 10') is a measure of the 'affinity' of the inhibitor for the particular substrate and may be useful for comparing the effect of various inhibitors on a variety of substrates.21 Moreover, the presence of both DDPE and EHDP in solutions where calcium carbonate precipitates spontaneously not only reduced calcite formation markedly but also prolonged the induction period, z, preceding the onset of precipitation. The induction period is defined as the time over which no change in pH or free calcium activity has been observed in this work.22 Initial conditions of the experiments in the presence of increasing amounts of DDPE, as well as the respective induction periods, are summarized in table 2.It can be seen that the induction period is an extremely sensitive function of the inhibitor concentration in the working solution. The kinetics results of the experiments summarized in table 2 gave an excellent fit to the kinetic Langmuir model as well, yielding a slope of 3.3 x lo6 for the resulting straight line. This value compares well with that obtained when the kinetics were studied following inoculation with calcite seed crystals. A study of the absorption of EHDP and DDPE revealed that both compounds readilyA . G. Xyla and P. G. Koutsoukos 1483 60 - "E --.. 40- iz' 1 e .e, 20 - 0 0 5 10 15 Ce,/ 1 0-6 mol dm-3 Fig. 4. Langmuir isotherm for the adsorption of inorganic DDPE on to calcite initial pH 9.2, 22 "C. adsorb on to calcite surfaces in aqueous solutions.The adsorption isotherm of DDPE on calcite is shown in fig. 3. The plateaux obtained at equilibrium concentrations of ca. 5 x mol dm-3 correspond to a surface coverage of 0.030 m2, which is ca. 9% of the available surface area, assuming an area of 100 A2 per DDPE molecule (corresponding approximately to two phosphonate groups).23 Note that in the presence of this high level of DDPE concentration the precipitation of calcium carbonate, both seeded and spontaneous, was completely inhibited. Similar behaviour was shown by EHDP which, however, did not show a plateau in the adsorption isotherm. Instead, a linear relationship between EHDP adsorbed on to calcite and the equilibrium solution concentration was found. Adsorption of DDPE on to calcite can be described by the commonly used, rearranged form of the Langmuir equation: where Ce, is the equilibrium phosphate concentration (mol dm-3), r the amount of phosphate adsorbed on to calcite (mol m-2), Trn the maximum amount of phosphate that can be adsorbed on calcite and b a constant, proportional to bonding energy.Adsorption data for DDPE gave an excellent fit according to eqn (3) shown in fig. 4. The slope of the straight line obtained was found to be 3 . 8 6 ~ log, which is in good agreement with the value of 2.34 x lo6 obtained from the kinetics data applied to a simple Langmuir kinetic model, as shown in fig. 2. From EHDP adsorption data a value of 1.34 x log was obtained from the Langmuir isotherm data. The magnitude of the slopes of such modified Langmuir isotherms, obtained either from adsorption studies or from kinetic data in the absence and in the presence of inhibitors, may be useful in evaluating the relative affinity of a compound for a certain substrate.211484 Precipitation of Calcium Carbonate Conclusion In the present work it was found that two diphosphonates, EHDP and DDPE, cause significant retardation in the rates of calcite formation, which takes place either spontaneously or on synthetic calcite seed crystals.Kinetic data were found to fit a simple Langmuir model, from which a measure of the ‘affinity’ between additive and substrate could be obtained. The affinity thus determined was close to that determined from the adsorption isotherms of DDPE and EHDP on to calcite, thus confirming the close relationship between adsorption, presumably at the active crystal growth sites, and kinetic inhibition.Moreover, additions of EHDP, following the onset of precipitation, resulted in retardation of the rates of precipitation, indicating that the inhibitors act on the crystal growth rather than the nucleation part of the process of formation of the solid phase. We thank the Directorate for Science and Technology of the Ministry of Energy, Industry and Natural Resources for a grant (EPE 624/84) in support of this work. We are also indebted to Dr J. Mikroyannidis (this department) for providing us with pure DDPE. References 1 L. Rosenstein, US. Patent 2038416 (1936). 2 J. C. Cowan and D. J. Weintritt, Water-formed Scale Deposits (Gulf, Houston, Texas, 1976). 3 J.Glater, J. L. York and K. S. Campbell, Principles ofDesalination (Academic Press, New York, 2nd 4 0. J. Vetter and D. A. Campbell, Report (Lawrence Berkeley Laboratory, Berkeley, California, 1979). 5 J. Mikroyannidis, A. K. Tsolis and D. J. Gourghiotis, Phosphorus Sulfur, 1972, 13, 279. 6 S. Arhimandritis, Ph.D. Thesis (University of Patras, Greece, 1983). 7 P. G. Koutsoukos and C. G. Kontoyannis, J. Chem. SOC., Faraday Trans. 1, 1984,80, 1181. 8 R. G. Bates, pH Determination (John Wiley, New York, 2nd edn, 1973). 9 M. M. Reddy and G. H. Nancollas, J . Colloid Sci., 1971, 36, 166. edn, 1980), part B, p. 627. 10 A. Gee and V. R. Deitz, Anal. Chem., 1953, 25, 1320. 1 1 G. H. Nancollas, Interactions in Electrolyte Solutions (Elsevier, Amsterdam, 1966). 12 T. F. Kazmierczak, M. B. Tomson and G. H. Nancollas, J. Phys. Chem., 1982, 86, 103. 13 V. M. Kharin, Russ. J. Phys. Chem. (Engl. Transl.), 1974,48, 1018. 14 Y. Kitano, Bull. Chem. Soc. Jpn, 1962,35, 1981. 15 G. Wilken, Desalination, 1980, 33, 1201. 16 J. H. Noggle, Physical Chemistry on a Microcomputer (Little Brown, Boston, 1985), p- 49. 17 M. P. C. Weijnen and G. M. Rosmalen, in Industrial Crystallization 84, ed. S . J. Jancic and E. J. 18 M. C. van der Leeden and G. M. Rosmalen, ref. (17), p. 325. 19 A. E. Nielsen, Faraday Discuss. Chem. Soc., 1976, 61, 153. 20 J. Christoffersen, M. R. Christoffersen, S. B. Christensen and G. H. Nancollas, J . Cryst. Growth, 21 P. G. Koutsoukos, Z. Amjad and G. H. Nancollas, J. Colloid Interface Sci., 1981, 83, 599. 22 J. W. Mullin, Faraday Discuss. Chem. SOC., 1976, 61, 238. 23 J. S. Gill and G. H. Nancollas, Corrosion, 1981, 27, 120. DeJong (Elsevier, Amsterdam, 1984), p. 61. 1983, 62, 254. Paper 6/ 1290; Received 26th June, 1986

ISSN:0300-9599    

DOI:10.1039/F19878301477

出版商:RSC    

年代:1987

数据来源: RSC

摘要:

J. Chem. SOC., Faraday Trans. I , 1987,83, 1485-1491 Electrical and Catalytic Properties of some Oxides with the Fluorite or Pyrochlore Structure CO Oxidation on some Compounds derived from Gd,Zr,O, Stefan J. Korf, Harry J. A. Koopmans, Bernard C. Lippens Jr, Anthonie J. Burggraaf and Paul J. Gellings* Laboratory for Inorganic Chemistry, Materials Science and Catalysis, Twente University of Technology, P.O. Box 21 7 , 7500 AE Enschede, The Netherlands The catalytic properties of some mixed zirconates with the pyrochlore or fluorite structure have been investigated using CO oxidation as the test reaction. The presence of terbium ions, leading to mixed conductivity, and the extent of pyrochlore ordering affect the kinetic behaviour and the catalytic activity of the investigated materials.Bismuth-containing com- pounds show an increased rate of reoxidation. In some previous papers1v2 the electrical and catalytic behaviour of some oxides with the fluorite or pyrochlore structure was reported. These were devoted to some pyrochlore titanates and cerium-substituted neodymium zirconates. It was shown that the presence of ions with a variable valence state and the extent of pyrochlore ordering affect the kinetic behaviour and the activity of the investigated materials. In this paper the study of some substituted compounds derived from Gd,Zr,O, or Tb,Zr,O, is presented. [Note, for the compounds (Tb,Gd,-,),Zr,O, the shorthand notation TGZ (1OOx) is used, thus TGZ40 is a compound in which x = 0.4.1 The electrical properties of these compounds have been studied by van Dijk et al.37 The transport number measurements show that TGZO is a purely ionic conductor, while with increasing terbium content an increasing electronic conduction is observed.The ionic transport number decreases with decreasing temperature, which is mainly caused by the observed increase in charge carriers for electronic conduction (Tb4+) with decreasing temperature, and partly by the differences in activation energy for ionic and electronic cond~ction.~ All measurements have shown that in these compounds there is p-type electronic conduction. Arrhenius plots for the total conduction are non-linear, showing that, in agreement with the transport number measurements, at least two conduction mechanisms with different activation energies play a role. Van Dijk3 has also shown that increasing pyrochlore ordering is accompanied by a decrease of the activation energy for ionic conduction.This may increase the catalytic oxidation and reduction activity when these processes involve the participation of lattice oxygen ions, because the lower activation energy for ionic conduction might be connected with the ease of oxygen exchange with the gas phase. Experimental Synthesis of Compounds The compounds were synthesized using the so-called ' citrate method ', for a detailed description of which we refer to the literature.6 In principle this method consists in making a solution of the nitrates of the different metals in the desired proportion, complexing with an excess of citric acid, neutralizing with ammonia, followed by evaporation and pyrolysis of the resulting viscous solution.14851486 CO Oxidation on Gd,Zr,O, Table 1. Surface areas of catalysts calcined at 1073 K catalyst SBET/rn2 g-' Gd2Zr207 9.8 f0.8 (Tb0.2Gd0.8)2Zr207 1.5f0.6 (Tb0.,Gd0.6 )2Zr207 4.0 f 1 .O (Tb0.6Gd0.4 )22r207 5.8 f0.6 Tb2Zr207 2.5k0.7 (Tb0.95Bi0.05)2zr207 22.8 f 0.8 tTb0.9Bi0.1)2Zr2O, 20.4 f 0.9 Table 2. Surface areas of heat-treated catalysts (Tb0.4Gd0. 612- T / K Gd2Zr207 Zr207 Tb2Zr207 1073 9.8f0.8 14.0 & 1 .O 12.5 _+ 0.7 1423 3.6k0.2 3.3 f0.4 4.1 f 0.3 1523 3.0f0.4 2.3 k0.4 3.6 f 0.3 1623 1.3f0.6 1.4k0.6 2.4 f 0.4 All oxides were calcined at 1073 K in air to remove carbon formed by incomplete combustion of the citrate. In order to obtain compounds with different amounts of pyrochlore ordering, three heat treatments were carried out at 1423, 1523 and 1623 K for 24 h on isostatically pressed samples of the powders.Analysis and Characterization The composition of the oxides was checked by X-ray fluorescence analysis and was found to deviate < ca. 1% from the desired composition in all cases. Only the Bi-containing compounds showed some bismuth deficiency. X-Ray powder diffraction was carried out on a Philips PW1050 diffractometer using Cu Ka radiation. Surface areas were determined by the B.E.T. method using Ar adsorption. Cat a1 y tic Measurements Catalytic measurements were performed in a tubular microreactor with a fixed catalyst bed, consisting of 300 mg of catalyst grains of 0.3-0.6 mm diameter, mixed with the same amount of quartz grains of the same diameter to avoid the formation of hot spots.The gas flow was ca. 1 cm3 s-l with a composition of 610% CO, 612% O,, balance N,, always with an excess of 0,. The CO conversion was determined by titration of the CO, formed and the conversions were between 0.5 and 40% (T = 600-750 K).S. J . Korf et al. 1487 Table 3. CO oxidation on calcined catalysts (qalcd = 1073 K) n m W,) EA(44) r723a catalyst ( f 0.1) (4 0.1) (f 0.6) /kJ mol-l mol mP2 s-l ~ ~~ ~~ TGZO TGZ20 TGZ40 TGZ60 TGZlOO 0.6 0.25 16.3 113 5.6 0.5 0.4 12.5 90 3.0 0.45 0.3 11.5 90 8.7 0.35 0.35 11.5 91 17 0.25 0.4 11.6 92 32 (Tb0.!35Bi0.05 )2zr207 0.4 0.2 10.2 86 36 (Tb0.BBi0.1)2zr207 0.5 0.05 9.6 82 70 a Gas composition: 2.5% CO, 6% 0,. Table 4. CO oxidation on heat-treated catalysts n m W O ) E*(+4) r723a catalyst (k 0.1) ( 0.1) ( 0.6) /kJ mol-1 /lod7 mol m-, s-l TGZO 1073 TGZO 1423 TGZO 1523 TGZO 1623 TGZ40 1073 TGZ40 1423 TGZ40 1523 TGZ40 1623 TGZlOO 1073 TGZlOO 1423 TGZl00 1523 TGZl00 1623 0.6 - 0.05 -0.25 0.2 0.45 0.35 0.25 0.75 0.25 0.15 0.05 0.35 - 0.25 16.3 0.9 17.0 0.2 8.9 0.1 8.0 0.3 11.5 0.25 11.1 0.3 9.7 0.35 12.7 0.4 11.6 0.25 7.3 0.4 6.2 .0.05 7.3 113 109 92 79 90 86 83 83 92 79 82 75 5.6 39 40000 920 97 81 32 400 74 44 8.7 0.66 a Gas composition: 2.5% CO, 6% 0,.Results The surface areas of catalysts are given. in tables 1 and 2. equation : r = k, exp ( -EA/RT) [COIn [O,]" where r is the specific reaction rate (mol mY2 s-l), k, is the pre-exponential constant, EA is the activation energy (kJ mol-l) and n and m are the reaction orders in CO and O,, respectively. Yao7 uses a similar equation to describe the kinetics of several catalytic oxidation reactions for which the exact mechanism is still unknown.As remarked in a previous paper2 the reaction orders vary appreciably from one material to another, making it impossible to characterize the catalysts with k, and EA only, as is usually done. Therefore the specific reaction rates are compared at an arbitrarily chosen temperature and gas composition, namely 723 K, 2.5% CO and 6% In table 3 the results obtained in this way for the calcined catalysts are given. The orders n and m, the pre-exponential factor k, and the activation energy EA have been calculated The catalytic behaviour of the different compounds is characterized using the kinetic 0 2 .1488 CO Oxidation on Gd,Zr,O, from the experimental data using a computer program based on the Nelder-Mead optimization method.In table 4 the results obtained on the heat-treated catalysts are collected. Discussion Introduction In most studies concerning the oxidation of CO on oxidic catalysts reduction of the surface by formation of CO+ is supposed to be the first step of the reaction mechanism.8* The CO+ ions then react with lattice oxygen ions to form adsorbed carboxylate (CO;) ions.87 lo, l1 After the formation of gaseous CO, the surface is reoxidized by oxygen from the gas phase. The following mechanism can be used in the description of the oxidation of CO on the catalysts studied here:, cog + h* co&-& (1) Note that in all reactions h* can be replaced by e' on the other side of the reaction equation, depending on the type of electronic defect present, It is clear that on most catalysts investigated neither reaction (1) nor reaction ( 5 ) can be rate-determining, since all orders in CO and 0, differ from 1.An exception may be TGZO heat-treated at 1423 K, where the order in oxygen is nearly 1 and that in CO nearly 0, suggesting in that case that reaction (5) is rate-determining. Calcined Catalysts In the system (Tb,Gd,-,),Zr,O,, increasing x gives a decrease in the order of CO, while that in 0, remains practically the same. This indicates a shift of the rate-determining step from reduction to reoxidation of the surface. This indicates increasing ease of reduction, which can be explained by the possibility of valence change of terbium: Tb4+ + e' g Tb3+ which is correlated with reaction (1) in the mechanism.Moreover, the presence of Tb increases the catalytic activity, probably for the same reason. In the system (TbzBil-z),Zr207 the order in CO increases and that in 0, decreases with increasing bismuth content, i.e. the rate-determining step shifts from reoxidation to reduction of the surface. This is in agreement with the literature,l2V l3 where it is shown that bismuth accelerates adsorption and/or surface diffusion of oxygen. Heat-treated Catalysts The heat treatment of the catalysts at 1423, 1523 and 1623 K in principle leads to an increase in the degree of pyrochlore ordering, and in this way the catalytic behaviourS. J . Korf et al.1489 may be influenced. As shown in table 4, the orders in CO and 0, are indeed strongly influenced in this way. The increase in pyrochlore ordering first leads to a decrease in the order in CO and an increase in that in 0,, indicating increasing ease of reduction so that reoxidation becomes more rate-determining. This is probably caused by the formation of stable carboxylate complexes.2 As a consequence the occupation of a CO adsorption site hinders 0, chemisorption. The negative orders in CO indicate that the reactants adsorb either on the same sites or on closely neighbouring sites. With a further increase in ordering the order in CO increases again while that in 0, decreases. This may mean that the adsorption sites on the more completely ordered material are farther apart.One possibility is that in particular the boundaries between the ordered and unordered parts are the most active parts on the surface. This is similar to the proposal of Andersson14 that in the oxidation of aromatic hydrocarbons in particular the boundaries between phases are most active. With increasing Tb content these differences become smaller and for Tb,Zr,O, (TGZl00) they are reversed. In the latter compound only limited, very localized ordering (not related to the pyrochlore structure) is observed by transmission electron micro~copy,~~ which is probably the cause of the different behaviour. As well as the effect on the orders there is also a strong influence on the catalytic activity. Here again partial ordering leads to an increase in activity, which decreases again with increased ordering.When pyrochlore ordering is only present as microdomains distributed randomly in a fluorite matrix the activation energy for conduction is not changed. Only when the ordering becomes so great that microdomains come into contact does the activation energy start to decrease. For Gd2Zr20, (TGZO) this results in an increase in total conduction. As suggested in the introduction, this might have a favourable influence on the catalytic activity. It is striking that the maximum in catalytic activity of TGZO is observed after a heat treatment which is insufficient for pyrochlore superstructure reflections to become visible in the X-ray diffractogram, so that ordering can only be present in the form of very small microdomains.A possible explanation for this effect is that the increased activity is due to an enhanced activity of the boundary between the ordered pyrochlore domains and the fluorite matrix, as also suggested above. If no ordered domains are present the activity is low. Correspondingly, when the domains become so large that they come into contact with each other the activity drops again. Because the conductivity is mainly determined by the bulk component (the fluorite matrix when only separated microdomains are present) this does not change much in the beginning of the ordering process, so that the expected correlation with conductivity is not observed. A problem in the description of the observations is the extremely large influence the reaction orders have on the reaction rates and on the reaction rate constants.For example, comparing TGZO, 1423 with TGZO, 1523 we have at 723 K: kTGZo, 1423 = 3.3 x 10-1 mol m-2 s-1 (mol cm-3)-0.85 which is much greater than kTGZO, 1523 = 1.7 x lop3 mol m-, s-l (mol C ~ - ~ ) O - O ~ while for the reaction rates as given in table 4 we have: ‘TGZO, 1423 ‘TGZO, 1523 at 723 K. This illustrates that, as indicated earlier, the description using the kinetic equation r = k[CO]n[O,]m makes it impossible to compare the k and ko values directly, owing to the fact that these have changing dimensions depending on the values of n and m. As a better description might be obtained by using a kinetic equation based on the Mars-van Krevelen reduction-oxidation mechanism, as used for example by Agarwal et aZ.16 for the oxidation of methanol, several equations based on this were tested using1490 CO Oxidation on Gd,Zr,O, Table 5.Kinetic parameters calculated for the oxidation-reduction mechanism El E2 105k1 105k, catalyst /kJ mol-1 /kJ mol-1 k,, k,, (723 K) (723 K) TGZ20 1073 87+3 81 + 3 22 3.8 1 . 1 1 . 1 TGZ18 1073 81_+28 96+21 97 370 14 8.6 TGZO 1423 228+40 93+2 4x101* 94 1300 3.6 TGZO 1523 174+84 77+16 6 x lo1, 8 . 8 ~ lo3 160000 4800 the multi-response regression computer program RKPES.17 As an example the results obtained for some catalysts using the equation: 1 1 1 +--- r = klPC, k,Po2 are given in table 5. Comparing these results with those based on the orders in CO and 0,, and the rates given in tables 3 and 4, it can be seen that they lead to the same conclusions.For example, comparing TGZlOOO 1073 with TGZ20 1073 it is seen that both reduction and reoxidation of the catalyst are faster for the former. Furthermore, the reduction is less rate-determining than the oxidation. This is reflected by the increase in order for CO and the decrease in that for 0, (see table 4). For TGZO 1423 reoxidation of the catalyst has the largest influence (k, 9 i t 2 ) , which gives r = k,[O,], in agreement with the orders -0.05+0.1 and 0.9kO.l given in table 4. For TGZO 1523 the deviations, e.g. in El and E,, are so large that no meaningful conclusions can be drawn in this case, probably owing to the negative order in CO, which cannot be reproduced by an equation of the simple oxidation-reduction model. Of course it might be possible to apply more sophisticated models,16 but then more experimental data would be needed than are available.Because the qualitative conclusions remain the same it was decided to use the original kinetic equation. Conclusions In the compounds (Tb,Gd,-,),Zr,O, the catalytic activity in CO oxidation increases with increasing terbium content. Simultaneously the order in CO decreases while that in 0, remains nearly unchanged. This can clearly be explained by the increasing concentration of charge carriers making reduction easier, while reoxidation is not much influenced. Increasing bismuth content, in the compounds (TbzBil-z),Zr207, causes the order in CO to increase while that in 0, decreases: bismuth facilitates oxygen uptake so that reoxidation becomes easier.Heat treatment of the TGZ series of compounds was applied at temperatures of 1423, 1523 and 1623 K and has a great influence on the orders and the catalytic activity. At the lower temperatures of heat treatment and at the lower terbium contents the order in CO decreases: the reduction is facilitated. The negative orders in CO indicate that the reactants adsorb on the same or on closely neighbouring sites. With increasing ordering the order in CO increases. In the absence of terbium the order in 0, first increases and then decreases with increasing annealing temperature, while at TGZ 100 with a very high terbium content the reverse is observed. At an intermediate terbium content the order in oxygen remains nearly constant. The catalytic activity shows a maximum as a function of heat treatment and this isS.J . Korf et al. 1491 highest for TGZO. TGZ40 and TGZlOO behave in this respect more like catalysts with a fluorite structure. This is clearly connected with the obser~ationl~ that pyrochlore formation is decreased with increasing terbium content. The authors thank Dr G. M. H. van de Velde and Ir H. J. Fontein for their assistance with the computer calculations. References 1 M. P. van Dijk, J. H. H. ter Maat, G. Roelofs, H. Bosch, G. M. H. van de Velde, P. J. Gellings and 2 J. H. H. ter Maat, M. P. van Dijk, G. Roelofs, H. Bosch, G. M. H. van de Velde, P. J. Gellings and 3 M. P. van Dijk, Thesis (Twente University of Technology, 1985). 4 M. P. van Dijk, K. J. de Vries and A. J. Burggraaf, Solid State Ionics, 1985, 16, 21 1 . 5 M. P. van Dijk and A. J. Burggraaf, Solid State Ionics, 1985,17, 159. 6 M. A. C. G. van de Graaf and A. J. Burggraaf, Sci. Tech. Zirconia 2, 1983, 12, 744. 7 Y. Yao, J. Catal., 1975, 39, 104. 8 P. Meriaudeau, M. Breysse and B. Claudel, J. Catal., 1974, 35, 184, 9 J. H. Lunsford and J. P. Jayne, J. Chem. Phys., 1966, 44, 1492. A. J. Burggraaf, Mater. Res. Bull., 1984, 19, 1149. A. J. Burggraaf, Mater. Res. Bull., 1984, 19, 1271. 10 W. S. Brey, R. B. Gammage and Y. P. Virmani, J. Phys. Chem., 1971, 75, 895. 1 1 K. H. Kim, S. H. Lee, Y. R. Kim and J. S. Choi, J. Catal., 1984,88, 283. 12 M. J. Verkerk and A. J. Burggraaf, J. Electrochem. SOC., 1983, 130, 70. 13 B. C. H. Steele, J. A. Kilner, P. F. Dennis, A. E. McHale, M. van Hemert and A. J. Burggraaf, Solid 14 A. Anderson, J. Catal., 1981, 69, 465; 1982, 76, 144. 15 M. P. van Dijk, F. C. Mijlhoff and A. J. Burggraaf, J. Solid State Chem., accepted for publication. 16 D. C. Agarwal, P. C. Nigam and R. D. Srivastava, J. Catal., 1978, 55, 1. 17 Y. Bard, Non-linear Parameter Estimation (Academic Press, New York, 1974), p. 61. State lonics, 1986, 18/19, 1078. Paper 611292; Received 26th June, 1986

ISSN:0300-9599    

DOI:10.1039/F19878301485

出版商:RSC    

年代:1987

数据来源: RSC

摘要:

f. Chem. SOC., Faraday Trans. I , 1987, 83, 1493-1506 Characterisation of Aerosol OT-stabilised Oil-in-water Microemulsions using a Time-resolved Fluorescence Method Paul D. I. Fletcher Chemistry Department, University of Hull, Hull HU6 7RX Oil-in-water microemulsion phases stabilised by sodium bis(2-ethylhexyl) sul- phosuccinate (AOT) form in equilibrium with conjugate phases containing virtually pure oil when the concentration of added aqueous electrolyte (NaC1) is low. Using turbidity titrations, the molar ratio of a1kane:AOT (= Ralkane) in the equilibrium microemulsion phase has been determined as a function of the NaCl and AOT concentrations, the chain length of the alkane component and the temperature. Increasing [NaCl], increasing [AOT], decreasing alkane chain length and decreasing temperature all cause the Ralkane value to increase.The aggregation numbers (Nagg) of the microemulsion particles have been determined using time-resolved fluores- cence measurements of the intensity decays of solubilised pyrene. Nagg is found to depend on Italkane, but is independent of the particle concentration. The variation of Nagg is consistent with the microemulsion droplets being spherical and monodisperse. The AOT monolayer is not penetrated by the solubilised alkane. The rate measurements indicate that the local viscosity of the dispersed alkane is tenfold higher in small droplets than its bulk value, but decreases to its bulk value as the droplet size increases. Much of the recent interest in microemulsions arises from their potential application in enhanced oil recovery.' The ultra-low interfacial tensions between microemulsion phases and phases of excess water or oil offer the possibility of using these systems to remove oil trapped by capillary forces in the porous rocks of oil reservoirs.For a particular surfactant, ultra-low interfacial tensions are observed normally only over a narrow range of'the variables of interest such as NaCl concentration, oil component composition or concentration of added co-surfactant. Much work is aimed at understanding and predicting the values of these variables (in relation to the molecular structure of the surfactant) necessary to produce ultra-low tensions. The molecular geometry of the surfactant monolayer film is important. The preferred curvature of the film (and hence the microemulsion droplet size) is determined mainly by the effective areas of the surfactant headgroup and tailgroup which, in turn, depend on the variables mentioned above.Two surfactant monolayer films are present in the two-phase systems. First, there is the planar interface separating the bulk microemulsion and excess oil or water phases and, secondly, the curved surfactant interfacial film between the dispersed and continuous components in the microemulsion colloidal particles (water droplets or oil droplets). The interfacial tension is a measure of the energy difference between these two films. Ultra-low tensions are observed when the interfacial surfactant film compositions in the planar and the droplet interfaces are This condition corresponds to large droplet sizes (i.e.the preferred film curvature is close to zero). Data concerning the molecular interfacial areas at the planar interface may be gained from surface tension measurements, and extensive studies have been made for systems containing the surfactant sodium bis(2-ethylhexyl)sulphosuccinate (AOT).4 Information concerning the area per surfactant molecule at the curved droplet interface can be derived 14931494 A 0 T Oil-in-water Microemulsions from a knowledge of the sizes of the microemulsion particles in two-phase systems, where the microemulsion phase is in equilibrium with a phase of the excess dispersed component [e.g. water-in-oil (W/O) microemulsion phase in equilibrium with an excess water phase]. Data are available for the sizes of AOT-stabilized W/O microemulsion particles in equilibrium with an excess of water phase as a function of the variables of relevance in enhanced oil recovery (e.g.salt concentration, alkane variation, etc.).* The purpose of this paper is to report the sizes of O/W microemulsion particles stabilised by AOT as a function of the variables mentioned above. This allows a detailed and complete picture of the effects of salt concentration, alkane chain length and temperature on the curved and planar monolayers of AOT. Conventional techniques used to determine colloidal particle sizes such as radiation scattering and ultracentrifuge methods generally require extrapolation of the results to infinite dilution in order to separate the effects of particle size and inter-particle interactions. This is particularly important in the case of electrically charged (i.e.strongly interacting) particles. O/W microemulsion particles fall into this category. Unfortunately, dilution of these particles is seldom possible since the particles are fluid, dynamic aggregates whose composition can (and often does) change upon dilution. For this reason, the sizes of AOT-stabilised O/W microemulsion particles have been determined using time-resolved fluorescence intensity measurements of the emission of pyrene probe molecules solubilised in the microemulsion droplets. Analysis of the decay curves allows the determination of the aggregation number of the particles at finite volume fractions of the dispersed component^.^ Experimental Materials AOT (Sigma) is a di-ester and some samples contain mono-ester and 2-ethylhexanol as impurities, arising from partial hydrolysis or incomplete esterification.s Sigma samples, however, showed no minimum in plots of surface tension versus concentration and hence are thought to be pure.? Pyrene (Aldrich) was extensively zone-refined before use.n-Heptane (Fisons HPLC grade) was passed over an alumina column prior to use. n-Nonane (Fluka purum), n-pentane (BDH), n-undecane (Fluka purum) and NaCl (BDH AnalaR) were used as received. Microemulsion Phase-boundary Determinations The boundaries of the one-phase O/W microemulsion region with respect to NaCl concentration, alkane chain length, AOT concentration and temperature were determ- ined by turbidity titrations.' AOT was weighed into a small bottle with a tightly fitting cap.The required volumes of water and alkane were added. The mixture was stirred rapidly (using a magnetic stirrer) and thermostatted at the desired temperature. The initially turbid mixture was then titrated using small volumes of a concentrated aqueous NaCl solution until a transparent O/W microemulsion was obtained. The mixture composition after the addition of the minimum quantity of NaCl required to produce clear, stable microemulsions was taken as being the microemulsion phase composition at the phase boundary at which separation of an excess oil phase occurs. Titrations using alkane as titrant into mixtures containing salt produced identical results to those obtained using the salt solution as titrant, but were observed to give end-points which took longer to develop.The O/W microemulsions were found to be stable over a narrow temperature range only (2 or 3 "C typically), unlike AOT-stabilised W/O t Interfacial tensions were measured by Dr J. Mead, University of Hull.P . D. I. Fletcher 1495 microemulsions, which are found to be stable over tens of degrees for some composi tions.8 Time-resolved Fluorescence Intensity Measurements 'Time-resolved fluorescence intensity measurements were made using an Edinburgh Instruments 199 fluorescence spectrometer equipped with a sample cuvette holder thermostatted to & 0.2 "C with a water circulator. One-phase microemulsion samples were prepared by weighing in the required amount of AOT and addition of the required volumes of alkane and NaCl solution, Pyrene was added as a solution in the alkane.The mixtures were stirred for 30 min. The samples were de-gassed by six freeze-pump-thaw cycles prior to the measurements in order to ensure the absence of oxygen, which is an efficient quencher of pyrene fluorescence. An excitation wavelength of 337 nm was used. The pyrene monomer emission was monitored at 380 nm (using an Ealing Corporation interference filter, f.w.h.m. 11 nm) with an additional cut-off filter (Schott type WG360) to further prevent scattered light detection. Analysis of Time-resolved Fluorescence Intensity Decay The analysis of the fluorescence intensity decay curve from a microemulsion or micellar solution containing fluorescent probe molecules and quencher molecules to yield the aggregation number of the particles is established as an accurate techniq~e.~-l~ For the case of pyrene emission quenching by excimer formation, the monomer pyrene emission intensity at time t following an instantaneous excitation pulse is given by5 (1) where I(0) is the intensity at t = 0.When the pyrene does not partition significantly into the continuous water component (i.e. is totally bound to the droplets) and the rate of inter-droplet exchange of pyrene between droplets is negligible on the experimental I(t) = I(0) exp { - A , t - A,[ 1 - exp ( - A , I)]} (2) timescale,13 then A, = k, A , = A (3) A , = kform (4) k , is the unquenched monomer pyrene emission decay rate, A is the average number of pyrene molecules per microemulsion aggregate and kform is the first-order rate constant for excimer formation in a droplet containing two pyrene molecules.The aggregation number (Nagg) of the microemulsion droplet may be calculated from the experimentally determined value of A using the equation ( 5 ) Nagg = (A([AOT] - c.m.c.))/[pyrene]. The c.m.c. was a minor correction and was calculated using the empirical relationship [eqn (6)] determined for this system:3 c.m.c./mol dm-, = 0.00202 exp (- 49.52 [NaCl]) + 0.00023. (6) The derivation of eqn (1) relies on two assumptions. First, that the distribution of pyrene molecules amongst the droplets is random. This is likely to be true when the solubility of pyrene in the microemulsion is high (i.e. each droplet can solubilise a large number of pyrene molecules). Experimentally, this assumption may be tested since A should depend linearly on [pyrene].This linear dependence is observed in the present work. Secondly, it is assumed that the quenching of pyrene monomer fluorescence by excimer formation is irreversible (i.e. that the excimer decay rate k , is very much faster than the rate of excimer dissociation kdiss),5 This assumption is not valid. The ratio1496 A 0 T Oil-in-water Microemulsions 0 ' I 1 0 20 40 [ NaCl]/mmol dm-3 Fig. 1. Plots of Rheptane us. NaCl concentration (expressed per total volume of microemulsion) at 25.0 "C. Concentration of AOT are 12.5 (@), 25 (O), 48 (a), and 93 (0) mmol dm-3. k E / k d i s s ranges from 1 to 3 in various so1vents,l4 cetyltrioxyethylene sulphate micelles15 and various non-ionic surfactant micelles adsorbed on to solid surfaces.l6 The invalidity of this assumption means that A , cannot be equated with kform, but is approximately equal to zkform, where z = k E / ( k E + k d i s s ) . In the calculation of the parameters A,, A , and A , from the experimental results the finite width of the exciting pulse (ca. 3 ns f.w.h.m.) was neglected since it is small compared with the experimental timescale (2500 ns). The experimental intensity decay curves can only be analysed to yield the parameters A , - A , when A , is greater than A,. Since A , is essentially determined by the oil droplet size, this has the effect of limiting the particle aggregation number that can be determined by this method. This limit corresponds to a molar ratio of alkane to AOT of ca.6 (see later discussion). Results and Discussion Microemulsion Phase Boundaries Fig. 1 shows plots of the NaCl concentration required to produce stable n-heptane- in-water microemulsions at 25 "C for various concentrations of AOT. Increasing the salt concentration or the AOT concentration increases the number of moles of heptane per mole of AOT (mole ratio n-heptane: AOT = Rheptane) which may be solubilised. The point at which Rheptane for the lowest AOT concentration rises sharply corresponds well with the NaCl concentration at which the interfacial tension for this system achieves an ultra-low, minimum value (0.05 mol dm-, NaC1).3 Above 0.05 mol dm-3 NaCl, theP. D. I . Fletcher 1497 0 6 a 2 4 2 0 0 0 -1 0 *2 0 -3 [ AOTIfmol dm-3 Fig.2. Plots of Rheptane us. AOT concentration [expressed per total volume of microemulsion (0) and per volume of water (*)I at 25.0 "C in the absence of added NaCl. AOT-heptane system undergoes phase inversion and W/O microemulsion phases in equilibrium with excess water are observed. The effect of salt on this system may be rationalised as follows. At low salt concentrations, the effective area per AOT headgroup is large, owing to electrostatic repulsion, and AOT monolayers show a tendency to curve such that the AOT headgroups form the exterior surface and the alkyl chain tailgroups form the interior surface (defined here as positive curvature). Hence, the preferred microemulsion aggregate structure is one of small, normal micelles. As the salt concentration is increased the headgroup repulsion is decreased and larger O/W microemulsion droplets (Le.higher Rheptane values) are stable. Above 0.05 mol dm-3 salt, negative curvature is preferred and W/O microemulsions and reversed micelles are observed. Increasing the AOT concentration beyond the values of fig. 1 leads to stable microemulsions in the absence of added NaCl. This is shown in fig. 2 as a plot of Rheptane us. AOT concentration (expressed both as per unit total microemulsion volume and per unit volume of water). The [NaCl] required for the stability of a microemulsion of a particular Ralkane value was found to depend linearly on the AOT concentration: where [NaCl],, ([AOT] -+ 0) is the salt concentration in the limit of zero AOT concen- tration and a IS the proportionality constant (dependent on Ralkane).The subscript aq indicates that the concentration is expressed per unit volume of water in microemulsion as opposed to the total volume. a was calculated by comparing the salt concentrations required to yield the same Rheptane value for different AOT concentrations using1498 A 0 T Oil-in-water Microemulsions 0.2 (Y 0.15 0.1 0 0 0 0 0 O O. 0 0 I 1 0 5 10 15 oil core radius/nm Fig. 3. Plots of a us. droplet core radius for heptane (0) and nonane (a) at 25.0 "C. The radii were calculated from the measured Ralkane values using eqn (1 1). interpolations of the data of fig. 1 and 2. The values of a for heptane and nonane as dispersed components (shown in fig. 3) decrease with increasing oil-droplet radius. Using the a values calculated, the data for all salt and AOT concentrations shown in fig.1 and 2 lie on a common curve as shown in fig. 4. The phase behaviour of AOT-stabilised O/W microemulsions differs from that of the W/O microemulsions found at higher salt concentrations in two respects. First, the O/W microemulsions are stable over only a narrow range of composition and temperature, whereas the W/O microemulsions of comparable droplet sizes are stable over a wider range. Secondly, for W/O microemulsion phases in equilibrium with an excess salt- containing water phase the Rwater values (and hence droplet sizes) are independent of the AOT concentration in the range 1-500 mmol dm-3.39 l7 For the O/W case, however, the phase boundary is very dependent on the AOT concentration.One explanation of this dependence may be that the degree of dissociation of the AOT-stabilised droplets could be large. Increasing the droplet concentration would then lead to an increased sodium-ion activity in the solution causing the phase boundary to shift to lower added salt concentration, It is tempting, therefore, to equate a with the particle degree of dissociation. However, surface-tension measurements show that the degree of dissocia- tion of the droplet and the planar interfaces are much lower than the values of a, being of the order of a few per cent and decreasing to zero at the NaCl concentration (0.05 mol drny3), corresponding to inversion.29 This is further supported by electro- phoretic measurements of AOT-stabilised Nujol-in-water emulsions, which again indi- cate a near-zero degree of dissociation.18 Hence a may not be equated with the degree of dissociation but is a parameter reflecting the inter-particle interactions which cause theP. D .I. Fletcher 1499 10 % 5 0 0 20 40 60 INaCIl, + OL [AOT],,/mmol dm-3 Fig. 4. Plots of Ralkane us. ([NaCl],,+@AOT],,) for heptane (0) and nonane (0) at 25.0 "C. shifts in phase boundary as the AOT concentration is changed. The observed decrease in a with increasing Ralkane is consistent with the reported decreasing degree of dissociation (and hence decreasing inter-particle interactions) as inversion is Note that shifts are observed for the O/W case where the electrostatic interactions are repulsive and long-range, but are not present in the W/O case for which light-3* l9 and neutron-scattering20 results indicate that the interactions are short-range and only very weakly attractive (i.e. close to ' hard-sphere') at the phase boundary where an excess water phase separates. The effect of changing the alkane chain length N is shown in fig.5. Increasing N increases the salt concentration required for stability. This is consistent with the observation that long-chain-length alkanes swell the tail region of AOT monolayers less than short-chain-length alkanes. Hence the headgroup region has to be compressed further by the addition of more salt to achieve the same monolayer curvature. The solubilisation capacity for n-undecane is limited, as shown in the figure. This correlates with a relatively high value of the minimum in interfacial tension observed for this alkane.Increasing temperature causes the salt concentration required for stability to rise (fig. 6). Counter-ion dissociation of the charged microemulsion droplet surface is an endothermic process, and therefore increases with increasing temperature. Hence increased salt must be added to achieve the same monolayer curvature. Additionally, increased thermal motions of the alkyl tails may swell the tail region. Limited solubilisation is noted at 10 "C for this system.1500 AOT Oil-in-water Microemulsions 5 n R - 4 9 3 .- s 3 2 $ ? * h Y Q * - 1 0 0 20 40 [ NaCl]/mmoI dm-3 Fig. 5. Plots of volume fraction of alkane us. NaCl concentration (expressed per total volume of microemulsion) at 25.0 "C for n-pentane (O), n-heptane (O), n-nonane (a) and n-undecane (0).The AOT concentration is 25 mmol dm-3. The horizontal bar indicates the limit of solubilisation of n-undecane. Time-resolved Fluorescence Intensity Results Using the titration results, single-phase, 'made-up ' microemulsions whose composition corresponds to the phase boundary at which an excess oil phase separates (at the selected temperature) could be prepared. These O/W microemulsions are stable over a range of only 1 or 2 "C. At higher temperatures, the solution becomes turbid, and creaming of a less dense (excess oil) phase is observed. At lower temperatures the solution again becomes turbid, but sedimentation of a more dense phase is observed. This phase (rich in AOT) is likely to be a liquid-crystalline phase.The phase boundaries were not significantly shifted in the presence of solubilised pyrene or as a result of successive freeze-thaw cycles. Changes in the temperature range of stability of a few degrees over several hours observed in some cases (particularly at high temperatures) are thought to be due to slight hydrolysis of the AOT, which may be catalysed by the micellar aggregates. This was not observed to affect the results significantly. The values of Nagg, k , and zkform for n-heptane in water microemulsions are shown in table 1. Table 2 shows the values obtained for microemulsions containing n-nonane. The measured values of k, show no systematic variation with pyrene concentration. [The values are close to that measured for pyrene in n-heptane at 25 "C (2.35 x lo6 s-l).] This is important, since it confirms that the pyrene does not partition out of the droplets or exchange between droplets on the experimental timescale,13 hence eqn (2) and (3) are valid for this system.P .D . I. Fletcher 1501 Fig. 6. Plots of Rheptane us. NaCl concentration (expressed per total volume of microemulsion) at 10.8 "C (e), 25.0 "C (0) and 39.4 "C (a). The concentration of AOT is 25 mmol dm-3. The horizontal bar indicates the limit of solubilisation at 10.8 "C. The values of Nagg are dependent on the mole ratio Ralkane, but are independent of the AOT concentration and variations in [NaCl] and temperature conditions required for microemulsion stability. If it is assumed that the microemulsion droplets are spherical and monodisperse then N [droplets] (4n/3)r: = iAoT1 &lkane Klkane (8) (9) N [droplets] 4nrE = [AOT] NAAoT where N is Avogadro's number, rc is the radius of the droplet alkane core, I/alkane is the molar volume of the alkane and AAOT is the area occupied by one AOT molecule at the surface of the alkane droplets.Substituting Nagg = [AOT]/[droplets] leads to the equation (10) Hence plots of Nagg us. Ralkane: (shown in fig. 7) are expected to be linear if the microemulsion droplets are spherical and monodisperse, with the additional assumption that is independent of Ralkane. From the slopes, the values of AAOT for n-heptane and n-nonane microemulsions are 0.54 & 0.03 nm2 and 0.50 f 0.05 nm2, respectively. Using an average value of 0.52nm2 for AAOT, the radii of the alkane cores of the microemulsion droplets are given by the equation Nagg = {(36n v&kane)/(N 2AkOT)) &lkane. rc/nm = 9.58 x Ralkane ( Klkane/cm3 mol-l) (1 1)1502 A 0 T Oil-in-water Microemulsions Table 1.Fluorescence decay parameters for AOT-stabilised microemulsions containing dispersed n-heptane (concentrations in mol dm-3 and rate constants in s-l) [pyrenel [AOT] Rheptane [NaCl] T/"C 0.05 0.05 0.05 0.05 0.05 0.05 0.05 0.05 0.05 0.05 0.05 0.05 0.05 0.01 0.0 1 0.02 0.02 0.10 0.10 0.20 0.20 0.30 0.30 0.05 0.05 1.37 1.37 2.74 2.74 2.74 4.10 4.10 4.10 5.47 5.47 6.60 6.60 6.60 2.74 2.74 2.74 2.74 2.74 2.74 2.74 2.74 2.74 2.74 2.74 2.74 0.0 10 0.010 0.020 0.020 0.020 0.0225 0.0225 0.0225 0.025 0.025 0.025 0.025 0.025 0.020 0.020 0.030 0.030 0.003 0.003 0 0 0 0 0.036 0.01 1 25.5 25.5 27.5 30.0 25.9 22.0 23.8 22.0 23.8 22.0 19.0 19.0 19.0 18.5 18.5 34.4 34.4 13.7 13.7 38.6 38.6 72.0 72.0 41.3 12.0 627 3 14 125 230 3 14 31.3 62.7 125 47.0 94.1 30.0 40.0 63.0 25.1 62.7 25.1 50.1 376 752 627 1254 94 1 1881 176 176 k, Zkforrn A /lo6 /lo6 Nagg 1.12 2.36 63.2 89 0.62 2.49 59.4 99 0.83 2.52 19 332 1.32 2.47 21 286 1.69 2.26 19 269 0.45 2.27 8.7 722 1.05 2.29 6.4 834 2.04 2.25 6.9 813 1.91 2.45 3.0 2000 2.21 2.60 4.7 1200 0.94 a 3.4 1600 1.43 a 2.9 1800 1.96 a 3.2 1600 0.94 2.58 19 374 1.43 2.69 28 229 0.44 2.42 17 3.51 0.85 2.99 18 340 1.37 2.25 16 363 2.38 2.16 19 316 1.17 2.55 14 372 2.07 2.75 14 330 1.02 3.07 5.2 327 1.87 3.15 5.7 298 1.08 2.51 20 307 1.08 2.78 20 307 a A value of k, equal to that in bulk heptane was assumed in the data analysis.Table 2. Fluorescence decay parameters for AOT-stabilised microemulsions containing dispersed n-nonane (concentrations in mol dm-3 and rate constants in s-l) 0.05 1.12 0.020 28.6 388 1.22 2.29 42.0 157 0.05 2.24 0.030 27.7 155 1.26 2.52 11.5 406 0.05 3.36 0.036 24.3 116 1.92 2.38 4.38 825 0.05 4.48 0.042 25.1 38.8 1.35 2.24 2.18 1700 0.05 6.72 0.048 25.2 19.4 1.06 2.26 1.1 2700 The overall radius of the microemulsion droplets is given by the sum of rc and the thickness of the AOT monolayer (ca. 1.5 nm). It is interesting to compare the magnitude of AAOT in the microemulsion droplets with its value at the planar alkane/water interface. The area per AOT molecule in a saturatedP . D. I . Fletcher 1503 Fig. 7. Plots of Nagg us. Rilkane for n-heptane (0) and n-nonane (a).film at the planar heptane/water interface at 25 "C is 1.14 nm2 at zero [NaCl] and decreases as PaCl] is increased and reaches a plateau value of 0.73 nm2 at salt concentrations > 0.05 mol dm-3. At low salt concentrations the area is determined by the effective AOT headgroup area (larger than the geometrical area owing to electrostatic repulsion). At high salt concentrations the area is determined by the tailgroup area and hence becomes independent of added electrolyte. The value of the area in the plateau region is a measure of the area per tailgroup in the planar interface. This area is found to depend on the alkane present, being larger for short-chain-length alkanes, which presumably penetrate the tail region and hence swell the area.Data for the tailgroup area (i.e. saturated areas at high [NaCl]) of AOT as a function of the alkane chain-length N (for N values of 7-16) obey the equation4 (12) tail area/nm2 = 0.50 exp (- 0.146N) + 0.56. The value of 0.52 nm (independent of alkane chain length) observed in O/W micro- emulsions agrees well with the high N limiting value of 0.56 nm2, corresponding to no penetration of the AOT monolayer by the alkane. The area per AOT tail estimated using molecular models and assuming a fully extended conformation of the alkyl tails is 0.53 nm2. It appears then that the AOT monolayer around the oil droplets is not penetrated by the dispersed alkane to any significant extent. In addition to the results shown in tables 1 and 2, qualitative observations were made of the effect of decreasing the sample temperature such that the sample became turbid and started to sediment (sometimes very slowly). This caused A to increase, the increase becoming larger with time.Thus holding the temperature below that required for microemulsion stability causes the growth of large particles which are probably1504 4 ' A 0 T 0 il- in- wa t er Microem ulsions 0 0 0 - 0 8 8 - 0 5 10 alkane core radius/nm Fig. 8. Plots of the local viscosity within the dispersed oil droplets us. oil core radius for n-heptane (0) and n-nonane (0). The dashed horizontal lines show the viscosities of bulk heptane (lower) and nonane (upper) at 25 "C. fragments of a liquid-crystalline separating phase. Increasing the temperature produced a slight apparent decrease in n (and hence Nagg).However, pyrene is likely to cream together with the oil out of the excitation beam, thereby making Nagg indeterminate. These changes were fully reversible, and measurements at the temperature at which the sample was transparent were time-independent. Excimer formation is a diffusion-controlled process. The second-order rate constant for excimer formation in bulk heptane at 25 "C was measured as 1.23 x 1 O 1 O dm3 mol-l s-l and is close to the calculated diffusion-controlled value k,, for spherical molecules of equal radius [eqn (13)] of 1.7 x 1O1O dm3 mol-1 s-l: k,, = 8RT/3000~ (13) where rl is the solvent viscosity. Hence the second-order rate constant for excimer formation within the oil droplet [kform (2nd order)] provides a measure of the local viscosity experienced by the pyrene within the oil droplets.Assuming that the pyrene is located only in oil volume of the microemulsion droplets, the concentration of pyrene corresponding to one pyrene molecule within an oil droplet is equal to (Nagg Rheptane beptane)-l. The second-order rate constant kform (2nd order) may then be calculated from the measured values of the first-order rate constants zkform: kform (2nd Order) = kforrn (Nagg Rheptane heptsne )- (14) Values of kform (2nd order) calculated in this way (assuming z = 0.7) are in the rangeP . D. I . Fletcher 1505 (1.5-7) x lo9 dm3 mot1 s-l for small and large droplets, respectively. Values of the viscosity inside the oil droplets may be estimated by comparing these values with the value in heptane (q = 0.386 cP).It should be noted that this calculation is approximate, since z can only be estimated and bulk solvent kinetics [assumed in eqn (14)] are unlikely to be completely valid in the small droplets where the pyrene diffusion is confined within a droplet. This approximation is likely to underestimate the viscosities. Lastly, it has been assumed that the pyrene is located in the oil core and does not penetrate the AOT monolayer, since pyrene does not dissolve in AOT solutions that contain no solubilised oil. The dependence of the local viscosities on droplet size for heptane and nonane is shown in fig. 8. The solubilised alkane viscosity is increased approximately tenfold in the smallest droplets, but the bulk solvent values are approached in the larger droplets.It appears that the interaction between the AOT curved monolayer (with apparently little or no alkane penetration) with the solubilised alkane reduces its mobility. An alternative explanation may be that the monolayer proximity hinders the mutual orientation of the pyrene excited-state and ground-state species necessary for excimer formation. The value of zkform depends only on the droplet size and internal viscosity for spherical droplets. However, distortion of the droplets to an elongated shape would be expected to decrease zkform since the pyrene molecules would have to diffuse further to achieve excimer formation. The results (table 1) indicate that zkform is constant for different AOT concentrations in the range 0.01-0.1 mol dmP3 at a constant Rheptane.Under these conditions the Nagg (and hence size) of the droplets is constant and only the droplet concentration changes. However, for [AOT] > 0.1 mol dm-3 zkform decreases, although Nagg remains constant. Assuming z is unaffected, this indicates that the particles distort from a spherical shape at high-volume fractions of droplets. For 0.3 mol dm-3 AOT and Rheptane = 2.74 (corresponding to a droplet volume fraction of 0.24), zkform is reduced by a factor of four. Conclusions The Italkane values of AOT-stabilised O/W microemulsions in equilibrium with an excess alkane phase as functions of [NaCl], alkane chain length and temperature may be rationalised in terms of the preferred curvature of the AOT monolayer. The dependence of Ralkane (and hence droplet size) of the equilibrium microemulsion phases on the AOT concentration is thought to be a consequence of the relatively large inter-droplet interactions between the charged oil droplets, since no such dependence is observed for the corresponding system of weakly interacting W/O microemulsion particles.These measurements of the droplet sizes in microemulsions in equilibrium with an excess dispersed component phase, together with measurements of the interfacial tension between concentrated microemulsions and excess phases, will allow the testing of theories of microemu1sions.21-24 Such measurements are currently in progress. The variation of Nagg with Ralkane is consistent with the microemulsion oil droplets being spherical (for dispersed volume fractions < ca.0.15) and monodisperse. The oil-droplet radii depend linearly on the mole ratio Ralkane and the molar volume of the dispersed alkane, but are independent of the droplet concentration. The area occupied by one AOT molecule at the oil droplet surface (0.52 nm2) is close to the geometrical area of the AOT tail group in a fully extended conformation. This implies that there is little or no penetration of the AOT monolayer by the dispersed alkane. A low degree of polydispersity in the particle size is further indicated by the lack of systematic variation of Nagg with pyrene concentration according to the analysis given in reference.25 (The differences in Nagg shown in table 1 are estimated to be of the order of the experimental errors.) It should be noted, however, that a non-random distribution of pyrene molecules amongst the particles would also produce variations in Nagg.50 FAR 11506 A 0 T Oil-in-water Microemulsions Finally, the rate measurements indicate that the viscosity of the dispersed alkane is increased from its bulk value. The bulk value is approached for the largest size droplets. I thank the S.E.R.C. for support of this work by the provision of an equipment grant. References 1 K. S. Chan and D. 0. Shah, J. Disp. Sci. Tech., 1980, 1 55. 2 R. Aveyard, B. P. Binks and J. Mead, J. Chem. SOC., Faraday Trans. 1, 1985,81, 2169. 3 R. Aveyard, B. P. Binks, S. Clark and J. Mead, J. Chem. SOC., Faraday Trans. 1, 1986, 82, 125. 4 R. Aveyard, B. P. Binks and J. Mead, J. Chem. SOC., Faraday Trans. I , 1986,82, 1755. 5 S. Atik, M. Nam and L. Singer, Chem. Phys. Lett., 1979, 67, 75. 6 P. D. I. Fletcher, N. M. Perrins, B. H. Robinson and C. Toprakcioglu, in Reverse Micelles, ed. 7 J. E. Bowcott and J. H. Schulman, 2. Elektrochem., 1955,59, 283. 8 P. D. I. Fletcher, A. M. Howe and B. H. Robinson, J. Chem. SOC., Faraday Trans. I , 1987, 83, 985. 9 P. P. Infelta, M. Gratzel and J. K. Thomas, J. Phys. Chem., 1974, 78, 190. P. L. Luisi and B. E. Straub (Plenum Press, New York, 1984), p. 69. 10 M. Tachiya, Chem. Phys. Lett., 1975, 33, 289. 1 1 N. J. Bridge and P. D. I. Fletcher, J. Chem. SOC., Faraday Trans. I , 1983, 79, 2161. 12 A. Malliaris, J. Lang and R. Zana, J. Chem. SOC., Faraday Trans. I , 1986, 82, 109. 13 J. C. Dederen, M. Van der Auweraer and F. C. De Schryver, Chem. Phys. Lett., 1979, 68, 451. 14 J. B. Birks, Photophysics of Aromatic Molecules (Wiley-Interscience, Chichester, 1970). 15 P. P. Infelta and M. Gratzel, J. Chem. Phys., 1979, 70, 179. 16 P. Levitz and H. Van Damme, J. Phys. Chem., 1986,90, 1302. 17 P. D. I. Fletcher, J. Chem. SOC., Faraday Trans. 1, 1986,82, 2651. 18 B. D. Powell and A. E. Alexander, Can. J. Chem., 1952, 30, 1044. 19 J. D. Nicholson and J. H. Clarke, in Surfactants in Solution, ed. K. Mittal and B. Lindman (Plenum 20 M. Kotlarchyk, S. H. Chen, J. S. Huang and M. W. Kim, Phys. Rev. A, 1984, 29,2054. 21 E. Ruckenstein, Chem. Phys. Lett., 1985, 118, 435. 22 S. Mukherjee, C. A. Miller and T. Fort Jr, J. Colloid Interface Sci., 1983, 91, 223. 23 C. A. Miller, J. Disp. Sci. Tech., 1985, 6, 159. 24 M. J. Grimson and F. Honary, Phys. Lett., 1984, 102A, 241. 25 G. G. Warr and F. Grieser, J. Chem. SOC., Faraday Trans. 1, 1986, 82, 1813. Press, New York), vol. 111. Paper 611297; Received 27th June, 1986

ISSN:0300-9599    

DOI:10.1039/F19878301493

出版商:RSC    

年代:1987

数据来源: RSC

摘要:

J. Chern. Soc., Furuduy Trans. I , 1987, 83, 1507-1514 Kinetics of Pyrrole Polymerisation in Aqueous Iron Chloride Solution Robert B. Bjorklund Laboratory of Applied Physics, Linkoping University, S-581 83 Linkoping, Sweden The chemical oxidation of pyrrole by iron(in) chloride in aqueous solutions of methylcellulose has been investigated. The kinetics of the reaction were studied by following the absorption of the soluble (colloidal), oxidised poly(pyrro1e) product at 800 nm. In the pH interval 0-0.5 the reaction was first-order in Fe"I and pyrrole concentrations and zero-order in proton concentration. For pH > 1.3, the kinetic curves exhibited behaviour char- acteristic of autocatalytic reactions. A mechanism is proposed in which the rate-limiting step is the oxidation of pyrrole to a radical cation by an outer-sphere activatedcomplex mechanism. The observed pH dependence on the rate is attributed to the hydrolysis equilibrium for FelI1 in solution, in which lower pH decreases the concentration of OH-, which can hinder pyrrole from approaching the oxidant.The observed autocatalytic behaviour at higher pH is caused by the fact that the polymerisation produces protons. A brief comparison with oxidation by ammonium persulphate is given. Electrically conducting polymers made from heterocyclic monomers have been the subject of much research in recent years. Poly(pyrro1e) has received considerable interest, partly because it can be prepared both electrochemically1 and by chemical oxidation2 and partly because of its relatively good stability in the conducting, oxidised form.How- ever, despite its long history, the mechanism of pyrrole polymerisation under oxidising conditions is not well understood. Results from several studies of anodic oxidation of pyrrole have led to a mechanism involving n-radical cations as intermediates in the p~lymerisation.~-~ Pyrrole monomer is believed to be oxidised at the electrode surface to an unstable n-radical cation which then reacts with neighbouring adsorbed species. The mechanism for electropolymerisa- tion is complicated, involving a series of oxidation and deprotonation steps as the product polymer precipitates on to the electrode surface. It has been reported that an average of 2.2-2.4 electrons are removed from each monomer during polymerisation.The extra fractional electron has been attributed to oxidation of the polymer during growth, since the oxidation potential of the polymer (E"' = - 0.2 V) is considerably lower than that of the monomer (E"' = 1.2 V).3 Chemical oxidation of pyrrole in acid yields both soluble oligomer and insoluble polymer products.2 It is well known that pyrrole treated with acid forms an unconjugated trimer.6 The trimerisation is believed to occur by Mannich-like reactions involving electrophilic pyrrolium cations. It is uncertain if oligomers are intermediates for polymer formation.' Lamb and Kovacic have reported that the trimer resulting from the reaction of pyrrole with thrichloroacetic acid does not produce polymers because of deactivation by the formation of a stable quaternary salt.8 Rapoport et al.have reported that a methanolic solution of terpyrrole (made by dehydrogenating the unconjugated trimer) polymerised after 24 h on exposure to air and light and polymerised instantaneously when added to H202 in acetic acid.g A mechanism for chemical, oxidative polymerisation of pyrrole by FeCl,, in a non-aqueous medium, has recently been proposed by Myers which involves a so1vent-FeCl3-pyrrole intermediate complex. lo 1507 50-21508 Oxidation of Pyrrole by Iron(II1) Chloride Kinetic studies of pyrrole polymerisation are difficult because the product, poly(pyr- role), is insoluble in common solvents. We have recently reported that chemical oxidation of pyrrole by FeC1, - 6H20 in aqueous solutions of water-soluble polymers such as methylcellulose, poly(viny1 alcohol), poly(vinylpyrrolidone), or poly(acry1amide) yields colloidal solutions of poly(pyrrole).ll The present work reports a kinetic study of the formation of colloidal poly(pyrro1e) in aqueous solutions of methylcellulose using visible spectroscopy to monitor the reaction.Experimental All reactions were carried out in distilled water in Erlenmeyer flasks sealed with Parafilm or glass stoppers. Pyrrole was purchased from Alfa and was distilled under vacuum and stored at 243 K in sealed bottles prior to use. Reagent-quality FeCl,.6H20 (Alfa) was obtained sealed under argon and used as received. Concentrated HCl(37 wt % in H20) was supplied by Merck. The soluble polymer, methylcellulose, was a high substitution form from BDH Chemicals Ltd (the viscosity of a 2% aqueous solution at 293 K was ca.4 poise). The average molecular weight was 95000 dalton from the relationship [q] = 2.8 x lo-, where [q] is the intrinsic viscosity and M the weight-average molecular weight.12 The degree of methyl substitution was determined by elemental analysis to be ca. 2. Analysis: calcd for (C8H1405)n: C, 50.53; H, 7.37; 0,42.11; found: C, 49.45; H, 7.39; 0, 43.16. All solutions were prepared fresh before each run. FeCl,.6H20 and pyrrole were dissolved in either an aqueous solution containing methylcellulose or water. Certain solutions were acidified with concentrated HCl. Both types of solution were equilibrated in a constant-temperature bath (295 K) and then mixed to start the reaction. The normal reaction volume was 50 cm3.Between 8 and 12 0.5 cm3 aliquots were removed at intervals and injected in a cuvette (2 mm pathlength) in order to monitor the formation of oxidised poly(pyrro1e) at 800 nm (see fig. 1). Results and Discussion The kinetics of oxidised poly(pyrro1e) formation were measured in the pH interval 0-0.5. Preliminary experiments had indicated that pH < 0.5 was necessary in order to obtain first-order kinetics for runs with FelI1 or pyrrole concentration in tenfold excess. Plots of In ( [ A (final) - A (time)]/A (final)}, where A is absorption, against time were linear for at least three half-lives (correlation coefficients > 0.997). We had previously observed the reaction stoichiometry to be 2.25 mol FeIII reacted per mole of pyrrole polymerised,ll and this fact was used in calculating the concentrations required to satisfy pseudo- first-order conditions for the kinetic runs.It was also observed in our previous work that when pyrrole monomer was present in excess concentration, a linear relationship existed between the starting concentration of FelI1 and the quantity of oxidised poly(pyrro1e) formed.ll It was thus possible to confirm that the Lambert-Beer law was obeyed over the concentration range studied in the kinetic experiments by measuring the final absorption at 800 nm for several calibration solutions prepared from a tenfold excess of pyrrole monomer and different initial FeIII concentrations. In addition, a linear relationship between final absorption and initial pyrrole monomer concentration was also observed for calibration solutions prepared with the FeIII concentration in tenfold excess.These results support the conclusion that the increase in absorption at 800 nm is proportional to the quantity of oxidised poly(pyrro1e) formed in the kinetic runs. Analysis by scanning electron microscopy of films made from the calibration solutions indicated that the poly(pyrro1e) product had a spherical shape, 50-100 nm in diameter. We observed no anomalous effects on the absorption caused by light scattering for the concentration range studied in the kinetic experiments.R . B. Bjurklund 1509 0.75 0 ii f 0.50 0, 2 0.25 300 800 1300 wavelengthhm Fig. 1. Absorbance spectra of reaction mixtures after polymerisation. Initial concentrations (mol dm-3) were: (a) 5.55 x (C8H1405)n; (b) 5.55 x pyrrole, 7.9 x 10-1 Hf and 6.58 x loF4 (C8H140Jn; ( c ) 2.22 x Fe"', 5.97 x pyrrole, 1.6 x H+ and 1.05 x (C8H1405)n. Pathlength 1 mm.FeIII, 2.98 x pyrrole, 7.9 x 10-l H+ and 1.05 x FelIJ, 2.98 x The rate of reaction for polymerising n moles of pyrrole to oxidised poly(pyrrole), was defined as 1 d[FelI1] - 1 d[pyrrole] dt 2.2% dt n dt ' -_-______- - dbolY(pYrrole)nI The observed pseudo-first-order rate constants are shown in table 1. It was also observed that a doubling and tripling of the pyrrole concentration when pyrrole was in excess resulted in rates two and three times, respectively, those shown in table 1 B. However, a saturation in the rate was observed when FeIII was in excess (table 1A). The reaction rate was dependent on methylcellulose concentration.The average rate increase which occurred upon doubling the methylcellulose concentration was ca. 1.17. This is best seen in table 1 B, where the methylcellulose concentration is varied over an order of magnitude. The highest final product absorptions were observed at the higher methylcellulose concentrations when pyrrole concentration was in excess, and the values decreased as the methyl cellulose concentration decreased because of formation of insoluble poly(pyrro1e) on the reaction vessel walls (see fig. 1). Methylcellulose plays an important role in the reaction since its presence is necessary to prevent the precipitation of the poly(pyrro1e) product. One can speculate that the polar groups on methylcellulose1510 Oxidation of Pyrrole by Iron(rIr) Chloride Table 1.Observed rate constants for polymerisation of pyrrole H+a (C8H1405)~ FeI1Ia pyrrolea Ab kobsdC 7.9 x 10-1 7.9x 10-1 1.6 x 10-3 A 5.26 x 5.55 x 2.49 x 0.75 1.04 x loF4 1.05 x 2.78 x 2.49 x lop3 0.75 6.10 x 1.05 x 5.55 x 2.49 x low3 0.75 1.22 x 1.05 x 1.11 x 10-l 2.49 x 0.75 1.62 x 2.10 x 10-2 1.05 x B 1.0 7 . 5 4 ~ 10-4 1.0 6.55 x 10-4 0.83 4.10 x 0.79 3.47 x 10-4 C 0.19 3 . 6 2 ~ 0.19 1.10 x 10-5 2.63 x 10-3 1.31 x 10-3 6.58 x 10-4 1.05 x 5.26 x 10-3 ] 2.22 x 10-3 5.97 x 10-2 2.63 x 10-3 a Initial concentration, mol dmA3. * Absorbance at 800 nm of reaction mixture after com- pletion of reaction normalised to the highest absorbances observed, group B. s-l. (and other water-soluble polymers) interact with the oxidised poly(pyrro1e) and influence its morphology and solubility.We assume in the analysis of the kinetic results that methylcellulose is not consumed in the reaction. However, as shown in table lB, methylcellulose appears not to have an unlimited capacity with regard to interaction with the poly(pyrro1e) product. The rate law in the pH range 0 4 . 5 is dboly(pyrrole)l = kobs [FeIII] [pyrrole] [methylcellulose]0~2 dt and the reaction rate was independent of proton concentration. No activation parameters are reported, since the temperature interval over which the reaction could be studied was too narrow. (Methylcellulose forms a gel when heated.12) For pH > 1.3, the kinetics of poly(pyrro1e) formation were observed to depend on the H+ concentration. Fig. 2 shows the typical S-shaped curves observed for the reaction in the pH region 1.3-2.8.This behaviour is characteristic of second-order autocatalytic reactions13 and is related to the fact that protons are a product of the polymerisation. Analysis of curves such as those shown in fig. 2 was complicated and a simpler measurement of reaction kinetics at constant H+ concentration was made at pH 2.8. Preliminary experiments had shown that for an initial FeIII concentration of 2.22 x mol dm-3 (pH 2.8) and excess pyrrole concentration, the proton concentra- tion was held essentially constant throughout the reaction because the increase from polymerisation was balanced by the corresponding decrease when FeIII was converted into FeI1.l4 It was thus possible to measure the reaction rate under pseudo-first-order conditions with FelI1 as the limiting reactant without adding an excess of HCI as was done previously.Table 1 C lists first-order rate constants for the self-buffering reaction at pH 2.8 when pyrrole concentration was in excess. Under these conditions, a doubling of the methylcellulose concentration resulted in an increase in reaction rate of ca. 1.8. Absorbances at 4% and 950 nm in the final products for Group C reactions were not of the same ratio as observed for reactions at pH < 0.5 (fig. 1). Therefore the normalisedR. B. Bjorklund 151 1 0.75 * a * 0 a 0 0 0 0 0 * o 0 0 0 0 2 4 6 tlh Fig. 2. Relative absorbance at 800 nm for reaction mixtures of 5.55 x lov2 mol dm-3 Fel*I, 2.49 x mol dm-3 (0) and mol dm-3 pyrrole and (C,H,,O,), concentrations of 1.05 x 2.63 x mol dm-3 (0).Initial pH was 1.94 and final pH was 1.79. absorbance at 800 nm, reported in table 1 C , was 0.19 instead of the expected 0.4 for complete reaction of the initial FelI1. It is not possible to describe a complete mechanism for the formation of oxidised poly(pyrrole), since the important intermediates [pyrrole oligomers and neutral poly(pyrrole)] were not observed. However, the results can be used to extend previously reported mechanisms describing pyrrole oxidation and to confirm a few of the postulates proposed for both the electrochemical and the chemical polymerisation of pyrrole. The initial step in the electropolymerisation of pyrrole is thought to be formation of unstable pyrrole radical cations at the anode ~urface.~ Polymerisation in solution can also proceed by oxidation of pyrrole (P) by FelI1 to form radical cations (Pa+) which can dimerise with the expulsion of H+ in the manner proposed for electrochemical synthesis.15 Thus P + FeIII -+ Pa+ + FeII P-P + FeIII -+ P-P'+ + FeII P-P'+ + P'+ -+ P-P-P + 2H+.Polymer chains continue to grow as long as pyrrole and FeIII are available. Unfortuna- tely, the observed first-order dependence of the rate on FeIII and pyrrole concentrations places the rate-determining step early in the mechanism [reaction (l)]. Several steps follow the formation of radical cations, including the oxidation of oligomers and the removal of 7t electrons from the growing poly(pyrro1e) chains, the event observed experimentally: P-P-P ... +FelI1 -+ P-P-P+ ...+Ferl. ( 5 ) (1) 2P*+ -+ P-P+2H+ (2) (3) (4)1512 Oxidation of Pyrrole by Iron(@ Chloride These several steps are assumed to be relatively fast as compared to reaction (1). It has previously been reported that the anodic peak potentials for the oxidation reaction of dipyrrole (0.55 V) and terpyrrole (0.26 V) lie intermediate between the monomer and The oligomers are more easily oxidised than is the pyrrole monomer and they have been used as starting materials to form poly(pyrro1e) by anodic oxidation.'' Thus it seems reasonable to assume that reactions (3) and ( 5 ) are relatively fast compared with the oxidation of the monomer. The dependence of the poly(pyrro1e) formation rate on the pH is related to the hydrolysis of FelI1 in solution. It is well known that FelI1 is solvated by six water molecules in aqueous solution and rapidly forms protons by hydrolysis, the first step being [Fe(H20)6]3+ + [Fe(H,O),OHI2+ + €3'.If we assume that the electron-transfer step described in reaction (1) occurs by an outer-sphere, activated-complex mechanism,18 then the reaction rate will depend on how effectively pyrrole can approach the solvated FelIr. Since pyrrole is a weak base, it can better approach FerI1 which are solvated by water than those which have one or two hydroxide ions in the solvation sphere. Thus the reaction rate increases with decreasing pH as the hydrolysis equilibrium is shifted to the left. Myers has described the importance of the solvent-FeC1,-pyrrole interaction in the polymerisation of pyrrole in non-aqueous solvents.He observed a qualitative agreement between the heat evolved when pyrrole was added to anhydrous FeC1, dissolved in various solvents.lo The magnitude of the exotherm was directly related to the yield of poly(pyrro1e). Strong donor solvents such as DMSO and pyridine interacted so strongly with FeC1, that pyrrole monomer had no chance to approach the oxidant and no polymer was formed. We have apparently observed the same phenomenon in aqueous solution, in which increasing the pH shifts the hydrolysis equilibrium to the right, with the resulting inability of pyrrole monomer to approach FeIrl and subsequent decrease in polymerisation rate. For pH > 1.3, the protons produced by the polymerisation affect the FeIII hydrolysis and thus the poly(pyrro1e) formation rate (fig.2). Two reactions compete with eqn (1) during the polymerisation. Pyrrole is known to form non-conjugated oligomers in acid media?, For reactions where pyrrole was in excess, the band at 450 nm continued to increase for ca. 20 h after the band at 950 nm had reached a constant value (which marked the end of the polymerisation reaction), Thus oligomers continued to form long after chemical oxidation was complete but their rate of formation was slow. In addition, only a small fraction of pyrrole was consumed by the oligomerisation when pyrrole was the limiting reactant. Oxidation of neutral poly(pyrro1e) by FeIrl to yield the final product [eqn (5)] competes with eqn (1) for FelI1 in the proposed mechanism. Reaction (5) is part of the overall synthesis reaction and is thus different from the oxidation of already prepared poly- (pyrrole) made neutral by electrochemical1s or chemica120 reduction upon contact with anhydrous FeCl,.It is known that alkyl substituents increase the basicity of pyrrole,6 so that the polymer chain may be better able to approach FeIII, eqn (5), than the monomer, eqn (1). Further comparison of the activated complexes for monomer and polymer oxidation is not possible, but Myers has reported that the nature of the counteranion present in poly(pyrro1e) is dependent on the solvent used in the synthesis.lo This may indicate that polymer oxidation occurs via a bridged activated complex mechanism18 in those solvents (acetonitrile and water) where only Cl- is present as counterion. Ammonium persulphate can also chemically oxidise pyrrole in an acidic medium to yield poly(pyrrole).21 In methylcellulose solutions the polymerisation was much faster than observed for Fe1I1 oxidation and, under pseudo-first-order conditions, the kinetics depended on which reactant was in excess.First-order dependence on persulphate andR. B. Bjurklund 1513 1.0 a 3 2 s1 2 0.5 5 10 tlmin Fig. 3. Absorbance as a function of time at 800 nm for reaction mixtures of 1.05 x mol dm-3 (C8H1405)n, 2.78 x mol dmW3 S,Oi-, and 2.98 x lo-, mol dm-3 pyrrole for two initial H+ concentrations. For curve (a) the pH was 5.6 at the start, 3.2 after 5 min and 2.6 at the end of the reaction. For curve (b) the initial pH was 0.1 because of added H,SO, and was constant throughout the reaction.Pathlength 2 mm. pyrrole concentrations was observed when persulphate was present in a large excess. However, when pyrrole was in excess the polymerisation, upon mixing the reactants, was instantaneous at low pH. It was characterised by an induction period (see fig. 3) at higher pH, as has been observed for the electrochemical polymerisation of pyrrole.22 The induction period shown in fig. 3 may be caused by the shift to the left in the equilibrium NH; 3t NH, + H+ as protons are produced from the polymerisation, thus removing competition for pyrrole’s interaction with the SVI1 of the persulphate. However, it should be emphasised that the mechanism described in this paper is specific for the FeIII oxidant and may not be applicable to other oxidants, such as persulphate.The work reported here was supported by a grant from the National Swedish Board for Technical Development (STU) and ASEA AB. References 1 A. F. Diaz, K. K. Kanazawa and G. P. Gardini, J . Chem. Soc., Chem. Commun., 1979,635. 2 G. P. Gardini, in Advances in Heterocyclic Chemistry, ed. A. R. Katrizky and A. J. Boulton (Academic, 3 A. F. Diaz, Chem. Scr., 1981, 17, 145. 4 A. F. Diaz, in Extended Linear Chain Compoundrs, ed. J. S . Miller (Plenum, New York, 1983), vol. 3, 5 B. L. Funt and S. W. Lowen, Synth. Met., 1985, 11, 129. 6 R. A. Jones and G. P. Bean, The Chemistry of Pyrroles (Academic Press, London, 1977), chap. 4. New York, 1973), p. 67. p. 1417.1514 Oxidation of Pyrrole by Iron@) Chloride 7 8 9 10 1 1 12 13 14 15 16 17 18 19 20 21 22 J. A. Joule and G. F. Smith, Heterocyclic Chemistry (Van Nostrand Rheinhold, London, 1979), p. 201. B. S. Lamb and P. Kovacic, J. Polym. Sci., Polym. Chem. Ed., 1980, 18, 1759. H. Rapoport, N. Castagnoli and K. G. Holden, J. Am. Chem. Soc., 1964,29, 883. R. E. Myers, J. Electron. Muter., 1986, 15, 61. R. B. Bjorklund and B. Liedberg, J. Chem. SOC., Chem. Commun., 1986, 1293. G. K. Greminger and A. B. Savage, in Industrial Gums, ed. R. L. Whistler (Academic Press, London, 1973), p. 633. A. A. Frost and R. G. Pearson, Kinetics and Mechanism (Wiley, New York, 1961), p. 19. D. A. Skoog and D. M. West, Fundamentals of Analytical Chemistry (Holt, Rinehart and Winston, New York, 1969), p. 361. E. M. Genies, G. Bidan and A. F. Diaz, J. Electroanal. Chem., 1983, 149, 101. A. F. Diaz, J. Crowley, J. Bargon, G. P. Gardini and J. B. Torrance, J. Electroanal. Chem., 1981, 121, 355. G. B. Street, T. C. Clarke, R. H. Geiss, V. Y. Lee, A. Nazzal, P. Huger and J. C. Scott, J. Phys. (Paris) Colloq., 1983, C3, 599. F. Basolo and R. C. Johnson, Coordination Chemistry (W. A. Benjamin, Menlo Park, 1964), chap. 6. P. Pffuger, M. Krounbi, G. B. Street and G. Weiser, J. Chem. Phys., 1983, 78, 3212. A. Pron, Z. Kucharski, C. Budrowski, M. Zagorska, S. Krichene, J. Suwalski, G. Dehe and S. Lefrant, J. Chem. Phys., 1985, 83, 5923. F. Hautiere-Cristofini, D. Kuffer and L. T. Yu, C.R. Acad. Sci., Ser. C , 1973, 277, 1323. J. Prejza, I. Lundstrom and T. Skotheim, J. Electrochem. SOC., 1982, 129, 1685. Paper 6/1303; Received 30th June, 1986

ISSN:0300-9599    

DOI:10.1039/F19878301507

出版商:RSC    

年代:1987

数据来源: RSC