|
11. |
Statistical thermodynamics of hard spheres in a narrow cylindrical pore |
|
Journal of the Chemical Society, Faraday Transactions,
Volume 90,
Issue 7,
1994,
Page 973-977
Qingsheng Xin,
Preview
|
PDF (467KB)
|
|
摘要:
J. CHEM. SOC. FARADAY TRANS., 1994, 90(7), 973-977 Statistical Thermodynamics of Hard Spheres in a Narrow Cylindrical Pore Qingsheng Xin, lsao Hiyane and Paul Siders" Department of Chemistry, University of Minnesota, Duluth, Duluth MN 55812,USA Statistical thermodynamics of hard spheres in a narrow right-circular cylindrical pore have been studied by the transfer-matrix method. The diameter of the pore is of the order of the particle diameter. The chemical potential and density calculated from the eigenvalues of the transfer operator show that there is no phase transition. This article reports statistical thermodynamics of hard spheres in a narrow, hard, cylindrical pore. The model is intermediate between one-dimensional and bulk conditions. It is well known that no phase transition is possible for hard spheres on a line, which is the limiting narrow pore.'*2 On the other hand, hard spheres in bulk two and three dimen- sions do freeze at high press~re.~ Model pores containing hard sphere^^-^ and Lennard-Jones partic1esg-l8 have been studied but not for pores so narrow as in the present model and not to determine high-pressure phase behaviour.We find the narrow pore essentially one-dimensional : no phase tran- sition occurs. The canonical partition function with a periodic boundary condition along the pore's length is our starting point for thermodynamic calculations. The model pore is so narrow that the particles in the pore cannot pass one another. The particles can, thus, be numbered from one end of the pore to the other, which makes the transfer-matrix applicable. We identify the appropriate transfer operator and calculate its eigenvalues and eigenfunctions.Model Pore We use right-circular cylindrical coordinates to locate the hard spheres in the pore as sketched in Fig. 1. The notation is as follows: 0 is the diameter of a hard sphere; D is the diam- eter of the right-circular cylindrical pore; rl, r2, ..., rNare circular polar coordinates of the hard spheres 1, 2, ..., N, respectively; zl, z2, . . . ,zN are coordinates of the hard spheres 1, 2, . . . , N, respectively, on the z pore axis; and z: = zi+ -zi. The pore length, L,is N N L = C(Zi+1 -Zi) = cz; (1)i-1 i-1 Our model is for the limit L -,co,with NIL constant. We set sphere N + 1 identical to sphere 1, so our system is I I I -z;-+z;--z;-; z2 z4 Fig.1 Spheres in the right-circular cylindrical pore periodic along the z axis. For our numerical calculations, D = (2 + ,/3)0/2 = 1.8660. This is the largest D for which only the nearest neighbours interact. For simplicity, we set 0 to unity. Partition Function The classical partition function, Q, for distinguishable par- ticles in the canonical ensemble (fixed V, N and T)is Q = ZN/A3N,where A is the thermal de Broglie wavelength and 2, is the configuration integral. The cross-sectional area of our model pore is fixed, so the pore's volume is determined by its length, L. We write the configuration integral as a func- tion of L rather than V. We have fixed z1 = 0.Let s be the force along z per cross- sectional area. Take the Laplace transform of the configu- ration integral with respect to L: ZN(s)= exp(-psL)ZJL) dL rrr N 6(rN+l -'1) x dr, dr2 *** drN dz; dz; dzk (3) A typical z;integral is exp[ -Bu(z:, ri+ -r,)]exp( -Dsz:) dzi (4) For hard spheres the pair potential, u, is infinity for z: < (0' -I ri+ -riI2)'l2 and zero elsewhere. We change the lower limit of the integration and evaluate the integral. exp[ -Bu(zf, ri + -ri)]exp[ -Bszf] dz; exp(-Bsz;) dzf J(uz-l*i+l -ril2) 1 = -exp[-BsJ(a’ -I ri+ -ri 12)] (5)Ps There are N such integrals in the configuration integral, so Transfer Operator We define a transfer operator, K(ri+ 1, ri), as 1’ -psK(ri+,,r.)=-expC-BsJ(a2-Iri+l -ri12)I (7) Although the operator is symmetric, it is not positive and definite.We assume the existence of a complete set of ortho- normal eigenfunctions { Y,(r)} of K. K(r, r)Y,(r’) dr’ = A,Y,(r) (8)s In eqn. (8), A, is an eigenvalue. Substituting K into the con- figuration integral yields (9) We choose N to be even so that Z,(S) is positive even if A,,, < 0. The sign of A, is found to be (-1)”’. Suppose one eigenvalue, A,,, , is dominant: I A,= I = max( I Ai I }. Then, 2, -A:,, as N + a.This is the usual transfer-matrix result connecting thermodynamic properties to the eigenvalue equation. Thermodynamic Properties The chemical potential can be calculated from the dominant eigenvalue.pp’ = -in(n,,j~37 (10) The ideal chemical potential is Eqn. (11) for the ideal chemical potential contains both the cross-sectional area, .nD2/4, and the force per area, s, along the z direction because the Laplace transform was taken only with respect to z. The excess chemical potential is pex. The average pore length, (L), is obtained from the deriv- ative of Z&) with respect to Bs. In the limit of large N, the density, p, is simply N/(L). If a phase transition is induced in our model by increasing pressure, it will be evident in three ways: (1) the largest and second-largest eigenvalues will cross at some s; (2) the chemi- J. CHEM. SOC. FARADAY TRANS., 1994, VOL. 90 cal potential will have a discontinuous slope at that s; and (3) the density will be discontinuous at that s.The transfer oper- ator appears to favour a crystal-like structure at high pres- sure. The operator K reaches its largest value [K,,, = (l/Bs)exp(-#ls/2)], when ri+l and ri are maximal and COS(~~+-4i)= -1. The operator favours this point increasingly as Ps -,0, so high pressure forces neighbouring spheres out toward the wall and apart by an angle n. Freezing in two and three bulk dimensions is reviewed by Barker and Henderson3 Hard spheres on a plane freeze at BP/p0= 8.08 from fluid of density 0.761 to solid of density 0.798, where P is pressure. Hard spheres in three dimensions freeze at BP/po= 8.27 from fluid of density 0.667 to solid of density 0.736. Judging from these density jumps, we expect a density increase of a few per cent in the pore if freezing occurs.Eigenfunctions and Eigenvalues The radial vector to particle i from the pore axis is ri. We separate this into distance, ri, from the pore axis and angle, 4i,about the axis. In these coordinates, the transfer operator 1s K(r,r’, 4’ -4) 1 = -exp{ -&/[a2 -r2 -r’2 + 2rr’ cos(4’ -4)]}Bs (14) We first solve the eigenvalue problem, eqn. (8), by iteration on a rectangular grid of (r, 4) points. The 4 points are equally spaced : the r points are distributed non-uniformly with more points where Y is largest. The eigenfunctions are graphed in Fig. 2-4 for (m,j?s) = (0, l), (0, 10) and (1, 1). The surface meshes are several times coarser than the calculation L.U 0.2\ ; Fig.2 Eigenfunction for rn = 0 and Ps = 1.0 ”.”0.0 Fig. 3 Eigenfunction for rn = 0 and Ps = 10.0 J. CHEM. SOC. FARADAY TRANS., 1994, VOL. 90 2.0 Fig. 4 Eigenfunction for m = 1 and Ps = 1.0 grids. Fig. 2 and 3 illustrate that increasing pressure forces the eigenfunction toward large r. Fig. 4 shows the angle dependence corresponding to non-zero rn. It can be shown that the eigenfunction Ym(r)separates into a radial function and an angle function.24 Fig. 4 illustrates the radial-angular separation. The angle dependence in Fig. 4 is simply -cos(~).This separation occurs because the kernel is symmetric, and its only angle dependence is on 14’ -4 I. The angle function must be a linear combination of cos(rn4) and sin(rn4).We choose cos(m6) and normalize $, as follows: 1 = $i(r)cos2(rn4) d6r dr (15)IJ:, [$:r dr = 1/~(1+am,0)n1 (16) The upper limit of radial integration, R = (D -a)/2 = J3/4 = 0.433. By substituting ylm(r)= @,(r)cos(m4)into the eigenvalue equation, eqn. (8) above, it can easily be shown that the radial function is itself an eigenfunction of the radial kernel KJr, r’). The eigenvalue of K,,, is A,, the same as the eigenvalue of the full kernel K. K,,,(r,r’) = K(r, r’, x)cos(rnx) dx; x = 4’ -4 (17)s:. The eigenfunction expansion of the full kernel, K, is a, K(r, r’, 4’ -4) = 1Am $m(r)$m(r’)COS(rn6)COS(m~’) (18) m=O We multiply both sides by cos[m($’ -4)]and integrate over (4’ -4), obtaining the following expression for K, in terms of $, and A,,,, : Krn(r, r’) = Am @m(r)@rn(r’Xl+ dm,Ob (19) The special case of r = r’ is K,,,(r,T’ = r) = Am $:(r)(l + a,,,, o)n: (20) Once A, is known, eqn.(20) allows us to calculate $,,,(r).To derive a formula for A,,,,we multiply both sides of eqn. (20) by r and integrate over r. We simplify the right-hand side by applying the normalization condition given in eqn. (16). In the left-hand side, we use the defining eqn. (17) for K, and eqn. (14) with r = r’. Eqn. (21) for the eigenvalue results. fR rn exp{ -ps,/[02 -2r2(1 -cos x)])Jo J-, x cos(rnx) dxr dr (21) Eqn. (21) for A,,, implies that A. > [An/ for all values of ps and all n > 0. Numerical results suggest that the A,,, are all distinct, and that I AiI > I Ai+ I for all i.The radial integral in eqn. (21) can be done analytically, leading to eqn. (22). Fig. 5 shows eigenvalues calculated from eqn. (22). + BsJ(3 cos x + 5)] -4 exp(-Bs)(l + Bs) 8(p~)~(1cos X)-x cos(mx) dx (22) For ps > 1/10, the integral is evaluated using Simpson’s rule. Even though the denominator of the integrand is zero at x = 0, x = 0 is not a singularity of the integrand. The inte- grand approaches 3 exp( -ps)/(32Bs) as x 40. We use that value at x = 0 in numerical integration. For small Bs, the two terms in the numerator of the inte- grand are nearly equal. This leads to serious loss of signifi- cance with numerical quadrature. To avoid that problem, we expand the integrand in a Laurent series about Bs = 0 and integrate term by term.We evaluate and use coeficients up to and including n = 5.24 We use the series expansion for < ps d10-l. The eigenfunction corresponding to the dominant eigen- value is $o . The single-particle radial distribution function, p(’)(r),is simply the square of $o. This can be shown as follows: Supposef(r) is any one-particle function of the radial position vector r. The ensemble average offis We expand K(ri+,,ri)in its eigenfunctions and evaluate suc- cessive integrals just as in evaluating the partition function itself. We obtain As in the partition function, the largest eigenvalue, A,, domi-nates (unless the zeroth integral is identically zero: a case we neglect here). <fW4S,nY:n *; as N --* 03 (26) It must be that p(l)’ = $&).-E 5 -10 -C -1 5 -20 -25 Bs Fig. 5 Eigenvalues for rn = 0 (-), 1 (--), 2 (---) and 3 (-* -) J. CHEM. SOC. FARADAY TRANS., 1994, VOL. 90 3.0 I I 2.5 1 2.0 h v L. E 1.5-3- 1.o 0.5 0.0 L--------YI I 0.0 0.1 0.2 0.3 0.4 R r Radial eigenfunctions, $Jr): (-)Fig. 6 m = 0, Bs = 10; (---) m = 1, /?s = 1 m = 0, Bs = 1; (-----) The thermodynamically significant distribution is $i. We define the radial distribution corresponding to other eigen- values by analogy: p:) = Y;(r) = +f(r)cos2(m+). We interpret the $2, rn # 0, as metastable, single-particle distribution functions. The distribution functions p:) differ from the usual three-dimensional singlet distribution. In three dimensions, the iimit of ph') at low pressure is bulk density, NIT/.By con- trast, the low-pressure limits of pi) are independent of pres- sure. In fact, p:) -,(16/3n) (1 + d,,,, o)cos(m&)as s -,0.The constant 3n/16 is nR2,the cross-sectional area available to the spheres' centres. A few radial eigenfunctions are shown in Fig. 6. As pres- sure increases, the particle distribution is forced out toward the pore wall. For all m > 0, $,,,(r = 0) is zero. Angle distribu- tions are not shown in Fig. 6; they are proportional to cos2(rn+). The rn = 0 distribution has no angle dependence, as expected for a fluid state. The m = 1 distribution places the particles preferentially at + = 0, n,as would be expected for a close-packed crystal in the pore.We calculate the density as a function of fls, according to eqn. (13). We differentiate eqn. (22) with respect to fls analyti-cally and evaluate the angle integral in the same way as for Am. Fig. 7 shows the pressure-density relationship for m = 0 and rn = 1: s is pressure along the pore axis, and po is the close-packing density, which equals 16/[3(2 + J3)2].24 The rn = 0 and rn = 1 densities approach one another at high pressure but do not cross. Densities for m = 2 and m = 3 lie to the right of the m = 1 line. Twodimensional Pore We have also examined hard discs in a two-dimensional slit- like pore. The hard discs of diameter 0 = 1 are constrained between two parallel lines distance D = (2 + ,/3)/2 apart.We 50 (I1 40 t I1i1 -30 a -20 -10 0-. ./ .II . 1 . 0 -4 -1 2 -1 6 -20 0 4 8 12 16 20 ~-Bs Fig. 8 Eigenvalues for the two-dimensional slit pore: (-) even; (---) odd again choose the z axis to be the pore axis. We use y as the coordinate perpendicular to the pore axis with y = 0 on the pore axis. Fig. 1 applies with y in place or r and with & con-strained to 0 or n. The theoretical development is analogous to the three-dimensional case above, but the eigenfunctions are functions of y alone: there is no angle dependence. The eigenfunctions are either even or odd with respect to y. The eigenvalues corresponding to the even eigenfunctions are positive; those corresponding to the odd function are nega- tive.The two eigenvalues of the largest magnitude are shown in Fig. 8. Conclusions Because the eigenvalues do not cross, the chemical potentials do not cross. The eigenvalues, and therefore the chemical potential, appear to be continuous and differentiable func- tions of pressure. Therefore, there is no exchange of stability between the angle-disordered rn = 0 sphere distribution and the angle-ordered m = 1 distribution. The phase behaviour is essentially one dimensional. The hard spheres do not freeze in the cylindrical or slit-like pore. We thank the Donors of the Petroleum Research Fund administered by the American Chemical Society for support of this work. We also thank the Pittsburgh Supercomputing Center for a starter grant under which some calculations were performed.References 1 L. Tonks, Phys. Rev., 1936,50,955. 2 L. Van Hove, Physica, 1950, 16, 137, reprinted in Mathematical Physics in One Dimension, ed. E. H. Lieb and D. C. Mattis, Aca- demic Press, New York, 1966. 3 J. A. Barker and D. Henderson, Rev. Mod. Phys., 1976,48,587. 4 P. Sloth, J. Chem. Phys., 1990,93,1292. 5 D. Bratko, L. Blum and M. S. Wertheim, J. Chem. Phys., 1989, 90,2752. 6 J. M. D. MacElroy and S-H. Suh, Mol. Phys., 1987,60,475. 7 G.Subramanian and H. T. Davis, Mol. Phys., 1979,38, 1061. 8 A. K. Macpherson, Y. P. Carignan and T. Vladimiroff, J. Chem. Phys., 1987,86,4228. 9 B. K. Peterson, J. P. R. B. Walton and K. E. Gubbins, J. Chem. SOC.,Faraday Trans., 1986,82, 1789.10 B. K. Peterson, K. E. Gubbins, G. S. Heffelfinger, U. M. B. Marconi and F. van Swol, J. Chem. Phys., 1988,88,6487. 11 E. Kozak and S. Sokolowski, J. Chem. SOC., Faraday Trans., 1991,87,3415. 12 J. J. Magda, M. Tirrell and H. T. Davis, J. Chem. Phys., 1985, 83, 1888. 13 I. Bitsanis, J. J. Magda, M. Tirrell and H. T. Davis, J. Chem. Phys., 1987, 87, 1733. 14 I. Bitsanis, S. A. Somers, H. T. Davis and M. Tirrell, J. Chem. Phys., 1990, 93, 3427. J. CHEM. SOC. FARADAY TRANS., 1994, VOL. 90 977 15 16 17 18 19 20 G. S. Heffelfinger, F. van Swol and K. E. Gubbins, J. Chem. Phys., 1989,88,5202. S. Sokolowski, Phys. Rev., A, 1991,44,3732. W. J. Ma, J. R. Banavar and J. Koplik, J. Chem. Phys., 1992, 97, 485. J. R. Macdonald, Phys. Rev. A, 1992,46,2988. M. Kac, Phys. Fluids, 1959,2, 8. M. Kac, G. E. Uhlenbeck and P. C. Hemmer, J. Math. Phys., 1963,4, 216. 21 22 23 24 A. Munster, Statistical Thermodynamics, Springer, Berlin, 1969, vol. 1, ch. 4. L. M. Casey and L. K. Runnels, J. Chem. Phys., 1969,51,5070. J. L. Lebowitz, J. K. Percus and J. Talbot, J. Stat. Phys., 1987, 49, 1221. Quingsheng Xin, M.Sc. Thesis, University of Minnesota, Duluth, 1993. Paper 3/07030A; Received 25th November, 1993
ISSN:0956-5000
DOI:10.1039/FT9949000973
出版商:RSC
年代:1994
数据来源: RSC
|
12. |
Enzymatic reaction in water-in-oil microemulsions. Part 2.—Rate of hydrolysis of a hydrophobic substrate, 2-naphthyl acetate |
|
Journal of the Chemical Society, Faraday Transactions,
Volume 90,
Issue 7,
1994,
Page 979-986
Yoshikazu Miyake,
Preview
|
PDF (889KB)
|
|
摘要:
J. CHEM. SOC. FARADAY TRANS., 1994, 90(7), 979-986 Enzymatic Reaction in Water-in-Oil Microemulsions Part 2.t-Rate of Hydrolysisof a Hydrophobic Substrate, 2-Naphthyl Acetate Yoshikazu Miyake," Takuya Owari, Fumio lshiga and Masaaki Teramoto Department of Chemistry and Materials Technology, Faculty of Design and Engineering, Kyoto Institute of Technology, Sakyo-ku, Kyoto 606, Japan The catalytic hydrolysis rates of a hydrophobic substrate, 2-naphthyl acetate (NA), have been measured both in aqueous solution and in water-in-oil microemulsions (ME) formed by di(2-ethylhexyl)sodium sulfosuccinate (AOT)in heptane. The catalysts used were Iipase, achymotrypsin and imidazole. The dependence of W, = [H,O],,/[AOT],, at a constant [AOT],, and of [AOT],, at a constant W, on the overall rate constant were discussed in terms of a reaction model.The reaction model includes three param-eters, the distribution constants of NA and the catalysts and the rate constant of the local reaction field. The distribution constant of NA was evaluated by measurements of the distribution of NA between the ME in the Winsor II region and the aqueous phase, but that of catalysts was treated as a fitting parameter. It is suggested that the reactions with these catalysts proceeds at the interfacial region of ME. For the imidazole-catalysed reaction, if the imidazole exists preferentially at the interface, the rate constant is indepen-dent of W, , but the value was 0.2 times that in the aqueous phase. Superactivity of lipase was observed, that is the turnover number in ME was greater than that in the aqueous phase. However, as the Michaelis constant was also large, the rate constant (kcnT/km)in ME was smaller than that in the aqueous phase.The rate constant increased as W, increased. The dependence was caused by the conformation change of lipase due to the interaction of AOT molecules. The rate constants for both imidazole and lipase decreased with increase in [AOT] at constant W,, which might be caused by the change in the structure of ME. The turnover number for a-chymotrypsin at the interface was lower than that in the aqueous phase and approached the values in the aqueous phase as W, and [AOT] increased. However, the Michaelis constant decreased with increase in W, and approached a constant value, which was 25 times that in the aqueous phase.Many studies on the activity of enzymes in microemulsions (ME) have been reported.'-'' Most of the various substrates used in these studies have both hydrophilic and hydrophobic groups, i.e. a surface-active substrate and large molecules. Since the enzyme is entrapped in the water pools of ME, it seems that the behaviour of the reaction strongly depends on the locus of the substrate. However, in a few papers6.' '-14 the kinetics and mechanisms of the enzymatic reaction in ME were discussed by focussing on the locus of substrate and enzyme. We have studied in a previous paper the hydrolysis rate of a hydrophilic and surface-inactive substrate in ME formed by di(2-ethylhexy1)sodium sulfosuccinate (AOT) in heptane.In this work, we used the hydrophobic and surface-active sub- strate, 2-naphthyl acetate (NA).16 The catalysts used were lipase from Rhizopus delemar, a-chymotrypsin from bovine pancreas and imidazole as an acid-base catalyst. We mea- sured the hydrolysis rate of NA with these catalysts in both aqueous and ME phases. We determined the effects of W, = [H,O]oV/[AOT]oV at constant [AOT],, and [AOT],, at constant W, on the overall rate constants. Since the water pool size was determined only by W,, in the former condi- tions the effect of water pool size, and in the latter conditions the effect of microemulsion concentration on the overall rate constant can be discussed. These effects of W, and [AOT] on the overall rate constant were analysed using a reaction model which considered the distribution constants of sub-strate and catalysts.We will here discuss the characteristics of the enzyme-catalysed reaction in ME, compared with the imidazole-catalysed reaction. t Part 1: Ref. 15. Experimental The fine grade of lipase from Rhizopus delemar was pur- chased from Seikagaku Kogyo Co. Ltd. (Japan) and used without further treatment. The molar mass of the lipase was ca. 4.4 x lo4. a-Chymotrypsin from bovine pancreas was purchased from Sigma Chemical Co. Ltd., and used without further purification. The a-chymotrypsin (molar mass 24 800) is a hydrophilic and globular (radius 2.2 nm) enzyme.I7 The substrate used was NA and the solubility in the aqueous phase was 0.4 mmol drnw3.The concentration of the substrate in the aqueous phase was determined from the absorbance at 270 nm by use of a spectrophotometer (Hitachi U2000). The molar absorption coefficient at 270 nm was 4.42 x lo3 dm3 mol-' em-'. The concentration of NA in heptane was also determined using the molar absorption coefficient of 2.29 x 10' dm3 mol-' cm-' at 314 nm. The pH of the aqueous feed solution was adjusted to 6.9 or 8.9 by using 1/30 (or 1/10) mol dm-3 KH,PO, and Na,HPO, buffer solutions. Both aqueous solutions of dissolved catalyst and substrate were mixed in a 1 cm quartz cell. Then the change in absorb- ance at 330 nm that was assigned only to naphthol (NT), which is a product, was measured by spectrophotometry.The reaction rate in water-in-oil (w/o) MEs was also mea- sured by a similar method. The ME solutions were prepared by solubilizing the aqueous feed solution of substrate or cata- lyst into the AOT in heptane. The fine grade of AOT pur- chased from Nacalai Tesque Co. Ltd. (Japan) was used without further purification. The initial rate of hydrolysis of NA in ME, which was defined by the use of overall concen- tration (subscript OV), was obtained from the absorbance change at 330 nm. Since the naphthol exists in both organic 980 and aqueous phases,16 the absorbance at 330 nm, A330 in ME can be expressed by A330 = &AQCNTIWP 4WP -k &OCNT1O 40 (1) where 4o and 4wpare the volume fractions for the organic and water pools, respectively.The molar absorption coeff- cients in both phases are e0 = 1.70 x lo3 and E,~= 1.53 x lo3 dm3 mol-' cm-'.The overall concentration defined by the overall volume in the microemulsion system consists of two terms which are the concentrations in the water pools and in the organic phase, as follows: "TI,, = 4WP"TIWP + 40"TIo The overall rate, rOV,in ME can be expressed as rov = d"Tlov/dt = (4WP d"TlWP/dt + 40 d"TlO/dt) (3) It was assumed that the distribution constant of NT between the organic phase (0)and the water pools (WP) is equal to that between heptane and the aqueous phase in the absence of AOT.16 PNT = [NT]o/[NT]wp = 1.6 (4) Then, the overall rate can be evaluated from eqn. (1)-(4) as follows.A linear relationship between the absorbance at 330 nm and time was obtained for about 20 min. The initial rate of hydro- lysis was calculated from this slope by using eqn. (5). Since the value of E~~ was in the range 1600-1670 dm3 mol-' cm-' under the experimental conditions, the effect of this correction was small. The volume fraction of the water pools was determined by measuring the amount of water by the Karl-Fischer method. When [AOT],, > mol dmd3, [H20],, in heptane increased in relation to [AOT], .la-'' The radius of AOT microemulsions, r/nm, in heptane is expressed by r/nm = 0.16[H20]oV/[AOT]oV + 1.2 (6) This means that the concentration of microemulsion droplets at a constant value of W, = [H20]oV/[AOT)ov increases in proportion to [AOT],, ,when r is held constant.In order to obtain the distribution constants between the water pool and interfacial region and between the interfacial region and the continuous organic phase for NA, 10 cm3 of NA dissolved in heptane and AOT were contacted with 200 cm3 of the aqueous solution of constant ionic strength in the flask. The overall distribution constant was determined from measurement of the concentration of NA in the aqueous phase at equilibrium. The ionic strength in the aqueous solu- tion was adjusted to the range 0.25-2.0 mol dm-3 with NaCI, in which the two-phase system is in the Winsor I1 region." All experiments were carried out at 298 K. Results and Discussion Reaction Model for Hydrophobic Substrate in Microemulsions The w/o MEs formed by AOT are heterogeneous media, which disperse globular water pools of nanometre size in the J. CHEM.SOC. FARADAY TRANS., 1994, VOL. 90 organic continuous phase with the aid of surfactant. To study reactions in MEs we consider a model which divides the volume into three subvolumes,22 i.e. organic domains (Vo), water-pool regions (Vwp) and interfacial domains (h).For the enzyme-catalysed reaction of a hydrophobic substrate, the substrate preferentially resides in the organic and/or the interfacial domains. In contrast, the enzyme is present in the water pools and/or the interfacial domains of the ME. We assume that the reaction fields comprise both the inter- face and the water pools of ME in this reaction model.Assuming that the Michaelis-Menten equation for enzymatic reactions in these reaction fields is applied, the following equations, which use the local concentration in the individual regions, can be derived. 'WP = (kCAT)WP[slWPCEIWP/((Krn)WP + CsIWP) (7) IS = (kCAT)ScslSIEIS/((KdS+ cslS) (8) The overall rate, rov, can be expressed by considering both contributions of the reactions as rov = 4WP FWP + 4s rs (9) where 4wpand 4sare the volume fractions of the water pools and interface of ME, respectively. Furthermore, we assume that the overall rate can be expressed by using the concentration based on the total volume as follows : To obtain the relationships between the rate parameters defined by eqn. (10) and those of eqn. (7) and (S), we use the mass balance equations of substrate and enzyme as, CElov = 4WPCEIWP + 4sCEls The concentration was defined by the use of the volume in each region.The volume fractions can be expressed as 4WP =fWPCH2OlOV =fWP WOCAOTIO" = 0.018W0[AOT]0v (13) 4s =fs[AOT]ov = 0.361[AOT]ov (14) 40 = 1 -4WP -4s (15) where the unit of [AOTlov is mol dm-3. The value off,, is nearly equal to the water molar volume, 0.018 dm' mol-'. The value offs in eqn. (14) can be estimated by assuming the constant surface area occupied by an AOT molecule, SAoT,as follows: fs/dm3 mol-' = S,oTSNA = 0.361 (16) where S is the length of the interfacial region, NA is Avo- gadro's number. Using the values of SAoT= 0.6 nm2 molecule-1 23 and 6 = 1.0 nm,24 we obtainf, = 0.361.Note that the volume fraction of the interfacial region is indepen- dent of the water content, i.e. Wo. The distribution constants for substrate and enzyme between each region are defined as follows: pow = CSlO/CSlWP J. CHEM. SOC. FARADAY TRANS., 1994, VOL. 90 Inserting eqn. (17)-(20) into eqn. (11) and (12), the local con- centrations of substrate and enzyme in each region are given by: CSlO = CSIOV POW(4WP + 4s Psw + 40 Pow)-(21) CSlWP = CSlOV(4WP + 4s Psw + 40 Pow)-' = ~SlOV(fS~WP (22) CSlS = [Slav PSW(4WP + 4s Psw + 40 Pow)-= CSlOV(fs)S (23) CEIwp = CEloV(4wp + #sQsw)-' = CElov(f~h~ (24) ['IS = CEIOV QSW(~WP + 4s Qsw)-' = [EIOV(fE)S (25) On inserting eqn. (21)-(25) into eqn. (7) and (8), the following equations are obtained For the second-order reaction, the following relations can be obtained for both reactions in the water pools and at the interfacial region : The volume fractions of the interface and water pools depend on both W, and [AOT],, ,as expressed by eqn.(13) and (14). The effects of W, and [AOT],, on the overall rate constants can be interpreted from these equations by using the distribu- tion constants. Distribution Constants of Substrate in Microemulsions The overall concentration of NA in the ME phase at equi- librium, [S],,, can be derived from the mass balance equa- tion as where [S]; and Vb are the initial concentration of NA in the organic phase and the initial organic volume (10 m3),and [s]AQ and VAQ are the concentration of NA in the aqueous phase and the aqueous volume at equilibrium.The volume of the ME phase at equilibrium, V,,, was estimated from mea- surement of the water concentration in the organic phase by Karl-Fischer titration and using the molar volume of water, 0.018 dm3 mol-'. The overall distribution constant, Pov, of substrate between the microemulsion and aqueous phases is define as We assume that the concentration of substrate in the water pools is equal to that in the aqueous phase at equilibrium. The overall distribution constant was obtained from the mea- surement of the concentration of substrate in the aqueous phase at equilibrium. From eqn. (ll), (21)-(23) and (31), the following relationship can be derived Po, = pow + C(1 -~0,)fWP~O-(PSW -P0w)fslCAOTIov (32) In the Winsor I1 region, the water pool size, i.e.W,, was *0° t-----600 1 , II I I200 ' 0 0.1 0.2 0.3 0.4 0.5 [AOT] o,/mol dm -Fig. 1 Relationships between the overall distribution constant of NA and AOT concentration in the Winsor I1 region. [NaCl]/mol dmP3,W, : (0) 1.0, 13.0; (A) 2.0,9.6.0.25, 25.6; (a)0.50, 16.9; (0) determined from the concentration of NaCl in the aqueous phase. Fig. 1 shows the relationship between Po, and [AOTlov, which was obtained under the conditions of con-stant NaCl concentration in the aqueous phase. The experi- mental data can be correlated by a straight line, which is expected from eqn. (32). The values of Pow and Psw were calculated from the intercept and slope of the straight line for each W, value corresponding to NaCl concentration. Fig.2(a), (b) and (c) show the effect of W, on values of Psw, Powand Pso. All of the distribution constants decreased with increase in W,. Pow decreased with W, and approached a constant value, which is nearly equal to 350, i.e. the distribu- tion constant of NA between the aqueous and heptane phases in the absence of AOT. Pso is in the range 1.7-3.0. This means that the hydrophobic NA is present in the interface in preference to the organic phase owing to the surface activity of NA. However, the value of W, at equilibrium was almost independent of the concentration of substrate in the organic phase; W, = 16.6 for [NA] = 0.0, W, = 15.9 for [NA] = 24 mmol dmV3 and W, = 15.8 for [NA] = 48 mmol dm-3 at [AOT] = 100 mmol dm-3 and [NaCl] = 0.5 mol drnp3.From this result, it is deduced that the MEs were formed only by AOT molecules and that the substrate resides between the hydrophobic groups of AOT. In order to esti- mate the intrinsic rate parameters in ME, the correlation curves (as shown by the solid line in Fig. 2) are used below. The distribution constants for the imidazole and enzymes could not be obtained because the absorption of these species overlapped with that of AOT. Hydrolysis of NA with Imidazole Fig. 3 shows the relationship between the overall hydrolysis rate of NA and the overall concentration of substrate in an ME of W, = 14.5 and in the aqueous phase.Since the rate in the aqueous phase is proportional to the concentrations of both substrate and imidazole (Im), the following equation is valid : 'AQ = (km)AQ [Sl AQ[IrnlAQ (33) The rate constant (kIm)AQ in the aqueous phase was obtained as (k1m)AQ = 8.0 x dm3 mol-' s-'. The overall rate constant in ME, (kIm)OV, can be obtained by dividing the overall reaction rate by both overall concen- trations of substrate and imidazole. The effects of W, and [AOTlov on the overall rate constant are shown in Fig. 4 and 5, respectively. The value of (k,)ov was independent of W, in the W, = 5-25 region at constant [AOT], = 0.44 mol J. CHEM. SOC. FARADAY TRANS., 1994, VOL. 90 2000 1500 3 &m 1000 500 I 0 10 20 30 WO 800 600 2 400 200 0 10 20 30 wo 5//4 2 312 1 0 10 20 30 wo Fig.2 Effect of W, on the distribution constants of NA: (a) Psw= 9C~lS/~SlWP (4p, = ~~ls/~~lo9 (b)pow = C~Io/~~IWP = PSW/POW~ r:, lo-' m E W 1o4 1 ~,,~~I,1 >'.,,I I 1o-~ 10-* [NA]ov/mol d~n-~ 0.7 0.6 0.5 0.4 0.3 0.2 0.1 I 0 5 10 15 20 25 3b0 2 WO Fig. 4 Effect of W, on the overall rate constant (0)for the hydro- lysis of NA with imidazole. [NA],, = 5 x lop3 mol dm-3, [Im],, = 0.05 mol dmp3, [AOT],, = 0.44mol dmp3, (pH), 6.9. (m) (kJwp 4WP(fim)WP ;(0)('ds 4&firn)s. dm-3. On the other hand, the value of (k,,J0, decreased monotonically as [AOT],, increased at constant W, = 15.5. The overall rate constants for the hydrolysis of NA with imidazole can be expressed by eqn.(28) for the reaction in water pools or by eqn. (29) for the interfacial reaction. As the values of (jJWpand (j& can be evaluated from the dis- tribution constants shown in Fig. 2, the values of (k1m)wp 4wp(hm)wp and ('ds 4s(fIrn)s were calculated and are shown in Fig. 4. If the reaction proceeds in the water pools of the ME, the rate constant, (klrn)WP,becomes larger than 5 x dm3 mol-' s-' because 4WP(firn)WP< 1. This value is larger than that in the aqueous phase and the rate constant decreases as W, increases, because 4wp(fim)wpis an increasing function of W,. Since this result cannot reasonably be inter- preted, we assume that the reaction proceeds only at the interface of ME.As shown in Table 1, the value of (kIrn)scan be estimated by assuming the value of Qsw = [Im]s/[ImlwP. For a large value of QSW,the rate constant at the interface was kept at an almost constant value. Supposing the rate constant is independent of W,, the effect of W, on (K,,),, can be reasonably interpreted by the large value of Qsw. Since imidazole has a nitrogen atom, which is positively charged due to protonation, the concentration at the inter- face is higher than that in the water pool owing to electro- static interactions with the AOT molecules. Assuming a value of Qsw = 5, because (kIrn)sdoes not change for large Qsw, the rate constant is almost independent of W, and is smaller than that in the aqueous phase by a factor of 0.2.I' 1 ---v---I I I 0.2 0.4 0.6 0.8 1 0 [AOT]ov/rnol for theFig. 3 Effect of overall concentration of substrate on rate of hydro- Fig. 5 Effect of [AOT],, on the overall rate constant (0) lysis of NA with imidazole. (0)In aqueous phase; [ImIAQ = 0.05 hydrolysis of NA with imidazole. [NA],, = 5 x mol dm-3, mol dm-3, pH 6.9. (0)In ME with W, = 14.5; [Im],, = 0.05 mol [Im],, = 0.05 mol dm-3, W, = 15.5, (pH), 6.9. (0)For (k& calcu-dm-j, [AOT],, = 0.44mol dm-3, (pH), 6.9. lated by assuming Qsw = 5.0. J. CHEM. SOC. FARADAY TRANS., 1994, VOL. 90 Table 1 Effect of W, on the interfacial rate constant for several values of Qsw (kI,)s/10-4 dm3 mol-' s-l WO Qsw = 0.1 1.o 5.0 10.0 5.3 5.74 2.14 1.82 1.78 8.5 7.30 2.13 1.67 1.61 14.4 12.56 2.81 1.95 1.84 18.6 17.27 3.44 2.21 2.05 22.1 18.77 3.41 2.12 1.95 24.9 21.71 3.82 2.23 2.03 If the reaction proceeds in the immediate neighbourhood of the AOT molecules, the value of (kIm)sshould be indepen- dent of the concentration of AOT.However, the rate con- stant decreased with increase in [AOT] as shown in Fig. 5. This result means that the characteristics of the reaction fields depend on the concentration of ME. This behaviour cannot be reasonably interpreted by this reaction model, but may be caused by the change in the structure of ME with increase in [AOT].25 Hydrolysis of NA with Lipase Fig. 6 shows the relationship between the initial hydrolysis rate and the overall concentration of NA in the aqueous phase and in the MEs of W, = 5.9 and 15.2.The rate parameters in the aqueous phase were obtained in a prevous studyi6 as (kCAT)AQ= 1.04 x mol g-' s-' and (KdAQ = 0.74 x mol dm-3. The solid line in Fig. 6 shows the hydrolysis rate in the aqueous phase calculated from the Michaelis-Menten equation using these parameters. As shown in Fig. 6, the overall reaction rate in ME was proportional to the overall substrate concentration over the range of experimental conditions employed. Since the sub- strate was dissolved in the heptane phase, the rates were obtained over a wide range of substrate concentration com- pared with that in the aqueous phase. Above [NA],, = 10 mmol dm-3, the overall reaction rate becomes greater than that in the aqueous phase, which means that (kc&, is larger than that in the aqueous phase, i.e.the lipase in ME is super- active.",26 Since the rates in ME were not affected by the buffer concentration of the aqueous feed solution (as shown by 0and 0in Fig. 6), the change of pH in the water pool with W, due to the acidic impurities in the AOT is negligible under the experimental conditions employed.27 r IVJ m I 983 Since the overall rate constant was proportional to the overall substrate concentration over the entire concentration range investigated, the values of kCATand K, could not be obtained separately. The overall rate constant (kcAT/K,)ov = (k,Ip)ov in ME can be evaluated from dividing the overall rate by the overall concentrations of substrate and lipase.Fig. 7 and 8 show the effects of W, and [AOT],, on the overall rate constant, respectively. The overall rate constant increased with W, , i.e. the size of the water pools. The behaviour differs from that for imidazole, as shown in Fig. 4. The value is lower than that in the aqueous phase, (kCAT/Km)AQ = 1.41 x dm3 g-' s-', whereas, the overall rate constant decreased with [AOT],, . The overall rate constants can be expressed by the same equations for the imidazole-catalysed reaction, i.e. eqn. (28) for the reaction in water pools and eqn. (29) for the interfacial reaction. From evaluating (fs)wp and (f&, values of (kL1p)Wp 4dfLIp)wp and (kLw)s 4s(fLxp)s were calculated and are shown in Fig.7. If the reaction proceeds in the water pools of ME, the rate constant (kLXp),, is greater than 5 x dm3 g-' s-l, which is larger than that in the aqueous phase. Supposing the rate constant at the local reac- tion field is smaller than that in the aqueous phase, it is sug- gested that the lipase-catalysed reaction in ME proceeds at the interface of ME. ,-Iv) 73 0.15 m m EU d I 0 0.1082 h n--I '", a 0.05 30 5 10 15 20-s WO Fig. 7 Effect of W, on overall rate constant (open symbols) for the hydrolysis of NA with lipase. [AOTIov = 0.45 mol dm-3, [lip],, = 0.25 g dm-3, (pH), 6.9. ~A]ov/10-3 rnol dm-3: 10.0 (V), 12.0 (A), 16.3 (a),24.9 (Oh 49.5 (0). (hP)WP+WP(fLIP)WP, (0)(.I (kL*P)S +S(fLIP)S * F "b\0 F--.h n 2r-41 [NA]ov/mol dm-3 Fig. 6 Effect of overall concentration of substrate on rate of hydro- lysis of NA with lipase. (A) In aqueous phase nip]Ap = 0.25 g dm-3. (0)In ME with Wo= 5.9, [AOT],, = 0.47 mol dm-3, [lip],, = 0.25 g dm-3, (pH), 6.9, 1/30 mol dmP3 buffer solution; (0)1/10 mol dm-3 buffer solution (m) In ME with W, = 15.2, [AOTI,, = 0.44 mol dmV3, @iplOv = 0.25 g dmP3, (pH), 6.9, 1/30 mol dm-j (buffer solution; (O)l/lO mol dmP3 buffer solution. Since the value of #&fLIp)s = 4s Q~~/(+wP+ 6s Qsw) decreased with increase in W, for low QSW, the W, depen-dence of (k,,,), becomes greater than the W, dependence of the overall rate constant. As the lipase is a hydrophobic enzyme, assuming a value of Qsw = 5, the rate constant was calculated to be 0.40 x dm3 8-l s-' for W, = 5.6 and 1.3 x lop4 dm3 8-l s-' for W, = 18.9.The rate con- stant is smaller than that in the aqueous phase. As the turn- over number, (kc,,), , for the lipase at the interface is larger than that in the aqueous phase, the Michaelis constant (K& is larger than that in the aqueous phase, i.e. the substrate- enzyme complex is very unstable. As mentioned in the previous paper," the intrinsic rate constant in water pools for a hydrophilic substrate in ME increased as W, decreased. However, the interfacial rate con- stant increased with W,, i.e. with the increase in water pool size. Possible reasons for this effect include, (i) a change in the conformation of the lipase in ME, (ii) a change in the physi- cochemical properties in local reaction fields, and (iii) an interaction between the AOT molecules and the enzyme. Since the interfacial rate constant for the imidazole-catalysed hydrolysis was almost independent of W,, it is suggested that the physicochemical properties at the interfacial reaction field are independent of W, .Consequently, the decrease in the rate constant with decrease in W, may be caused by the inter- actions with AOT molecules or a change in lipase conforma- tion. To investigate further the effect of AOT we measured the hydrolysis rate of NA in the aqueous phase in the presence of AOT. Fig. 9 shows the effect of AOT on the rate constant in the aqueous phase.The rate constant was enhanced by the presence of AOT. The rate constant is affected by the inter- action with AOT, but the dependence of W, on the interfacial rate constant cannot be interpreted. This behaviour appears to be due to the specificity of the activity of lipase for the hydrophobic substrate in ME, i.e. the interfacial reaction. The [AOT],, dependence of the rate constant was similar to that of the imidazole-catalysed hydrolysis. The behaviour cannot be reasonably interpreted by the reaction model. But it may be caused by the change in the structure of ME due to the increase in the concentration of AOT.*' Hydrolysis of NA with a-Chymotrypsin Fig. 10 shows the relationship between the overall hydrolysis rate and the overall concentration of NA in the aqueous phase and in the MEs of W, = 9.4 and 15.3.The rate param- eters of the Michaelis-Menten equation in the aqueous phase were obtained as (kCAT)AQ = 3.71 x mol g-' s-' and (K,JAQ = 1.42 x mol dm-3. J. CHEM. SOC. FARADAY TRANS., 1994, VOL. 90 ;L 10-~ EU 0 L. lo4 lo-' loo [NA],,,/rnol dm-3 Fig. 10 Effect of the overall concentration of NA on the hyrolysis rate with a-chymotrypsin. (A) In aqueous phase [chy],~ = 0.50 g dme3, (pH),, 8.9.(0)In ME with W, = 9.4, [chylov = 0.50 g dm-3, [AOT],, = 0.46 mol dmP3, (pH), 8.9, 1/30 mol dmP3 buffer solu- tion. (0)1/10 mol dm-j buffer solution. (m) In ME with W, = 15.3, [chy],, = 0.50 g drnp3, [AOT],, = 0.44 mol dm-3, (pH), 8.9.1/30 mol dmP3 buffer solution. (0)1/10 rnol dm-3 buffer solution. As shown in Fig. 10, the overall hydrolysis rate in ME approached a constant value, which was lower than that in the aqueous phase. The hydrolysis rates of NA in ME were also independent of the buffer concentration. The solid lines were calculated from eqn. (10) and the overall rate param- eters in eqn. (10) were determined to fit the data. Fig. ll(a) and (b) show the effects of W, and [AOT],, on the overall m I E -E" 45! U 226/: st 1 " 5 10 15 20 2 5" wo 5 I I I I I I 8 mI E W mI-E mI 0 (DI5 n 0 $ 1 *?5*I---------$nE ?5 I l l I I I 0 1,I I,1I 0 0.2 0.4 0.6 0.8 1.0 [AOT] na/l 0-rnol dm -Fig. 9 Effect of AOT concentration on the rate constant in the aqueous phase.(k,,,), is the rate constant in the absence of AOT. [lip],, = 1.0 g dm-3, [NA],, = 0.26 x mol drn-j, (pH),, 6.9. J. CHEM. SOC. FARADAY TRANS., 1994, VOL. 90 Michaelis constant, (KJOv, respectively. Fig. 12(a) and (b) show the effects of W, and [AOTlOv on the overall turnover constant, (kCAT)OV, respectively. The value of (Km)ov decreased with increases in W, and [AOT],,, but the values were larger than that in the aqueous phase. The value of (kcAT)ov, which is lower than that in the aqueous phase as shown by the broken line, increased as both W, and [AOT],, increased. The value of (kCAT)OVbecomes zero at about W, =5, which corresponds to the water pool radius of 2 nm and is nearly equal to the radius of a-chymotrypsin.If the reaction proceeds only in the water pools of the MEs, the rate parameters can be evaluated from the follow-ing equations, (~A,)ov =(kcAThP 4wdf&)wP (34) (KrnlOv =(Km)WP(fS)WP (35) Whereas, if the reaction proceeds only in the interfacial regions of ME, the rate parameters can be evaluated by (kAT)OV =(kAT)s 4S(fCHY)S (36) (KrnbV =(Km)s(fs)s (37) The Michaelis constants for the local reaction fields were evaluated from eqn. (35) and (37). These values are plotted against [AOT],, and W, in Fig. ll(a) and (b),respectively. The Michaelis constant for the water pool of ME increases with W, and is smaller than that in the aqueous phase. In the value for the interfacial reaction decreased and 4, I I I __________---------3 (a) 1 2-5 10 15 20 25 wcl I I I, I 1 0 0.2 0.4 0.6 0.8 [AOT] ov/moldm-Fig.12 (a) Effect of W, on the overall turnover number (0)of the hydrolysis of NA with a-chymotrypsin. [AOT],, =0.44 mol dmP3, WAlov =0.05 mol dmW3,[chy],, =0.5 g dm-3 (pH), 8.9. (b) Effect of [AOT],, on the overall turnover number (0)of the hydrolysis of NA with a-chymotrypsin. W, = 16.8, [chy],, =0.5 g dm-3, I_NAIOv=0.05 mol dmP3, (pH), 8.9. (0)(k,,,), calculated by assuming Qsw= 1.0. (---) Turnover number in the aqueous phase. 1.2 1.o 0 n> 0.8 IV Y--. 0.6 h >IV 0.4Y 0.2 0.0 0.2 0.4 0.6 0.8 1.0 [AOT]A$10-3 mol dm-3 Fig. 13 Effect of AOT concentration on the rate constant in the aqueous phase.(k,,,), is the rate constant in the absence of AOT. [chy],, =0.25 g drn-’, [NA],, =0.10 x loP3 mol dm-3, (pH),, 8.9. is larger than that in the aqueous phase. It is reported for several reaction system^^.'^.^^ that the Michaelis constant in ME is larger than that in the aqueous phase. For this reac-tion system, it is suggested that the reaction proceeds at the interfaces in the ME. However, the W, and [AOT] depen-dence cannot be interpreted by the reaction model. To estimate the turnover number, (k,,,), , we assume Qsw = 1.0 for a-chymotrypsin, because of the hydrophilic nature of the enzyme. The turnover number calculated for the interfacial reaction is shown in Fig. 12(a)and (b).The value is smaller than that in the aqueous phase and approaches that value with increase in W, and [AOT].This result means that the turnover number approaches that in the aqueous phase with increase in the water content of ME. The activity of a-chymotrypsin depends on the water content and the activ-ity may be restored at high water contents. However, the value of (kCAT/Km) =(kCHY)is lower than that in the aqueous phase by a factor of ca. 0.02, because of the large Michaelis constant in the ME. Fig. 13 shows the effect of AOT on the rate of hydrolysis of NA in the aqueous phase. The overall rate constant, k,,, ,decreased with increase in AOT concen-tration, the value at [AOT] =0.9 x mol dm-3 is 0.27 times that without AOT. This behaviour does not directly correspond to the effects in the ME, but the decrease in the activity of a-chymotrypsin is probably caused by the increase in Michaelis constant, i.e.instability of the enzyme-substrate complex due to the adsorption of AOT molecules on the enzyme. Furthermore, the value of the Michaelis constant in the ME becomes large because the complex is formed at the ME interface. The authors acknowledge financial support in the form of a Grant-in-Aid for scientific research from the Ministry of Edu-cation, Science and Culture of Japan (No. 02650707). References 1 F. M. Menger and K. Yamada, J. Am. Chem. SOC., 1979, 101, 6731. 2 S. Barbaric and P. L. Luisi, J. Am. Chem. SOC.,1981,103,4239. 3 K. Martinek, A. V. Levashov, N. L. Klyachko, V.I. Pantin and I. V. Berezin, Biochim. Biophys. Acta, 1981,657, 277. 4 R. Schomaecker, B. H. Robinson and P. D. I. Fletcher, J. Chem. SOC.,Faraday Trans. I, 1988,84,4203. 5 P. D. I. Fletcher and D. Parrott, J. Chem. SOC.,Faraday Trans. I, 1988,84, 1131. 6 D. Han and J. S. Rhee, Biotech. Bioeng., 1986,28, 1250. 7 M. Gonnelli and G. B. Stramlini, J. Chem. Phys., 1988,92,2854. 8 R. Bru, A. Sachez-Ferrer and F. Garcia-Carmona, Biotech. Bioeng., 1989,34,304. 986 J. CHEM. SOC. FARADAY TRANS., 1994, VOL. 90 9 Yu. L. Khmelnitsky, A. V. Kabanov, N. L. Klyachko, A. V. Levashov and K. Martinek, Structure and Reactivity in Reverse 19 M. Adachi, A. Shioi and M. Harada, J. Phys. Chem., 1991, 95, 7925. Micelles, ed. M. P. Pileni, Elsevier, Amsterdam, 1989, p.230. 20 R. A. Day, B. H. Robinson, J. H. R. Clarke and J. V. Doherty, J. 10 R. Bru, A. Sanchez-Ferrer and F. Garcia-Carmona, Biochem. J., Chem. SOC.,Faraday Trans. 1, 1979,75,132. 1989,259,355. 21 M. Kotlarchyk, S-H. Chen and J. S. Huand, J. Phys. Chem., 11 P. D. I. Fletcher and B. H. Robinson, J. Chem. Soc., Faraday Trans. 1, 1985,81,2667. 22 1982,86,3273. R. Schomacker, K. Stickdorn and W. Knoche, J. Chem. SOC., 12 13 K. M. Larsson, P. Adlercreutz and B. Mattiasson, J. Chem. SOC., Faraday Trans., 1991,87,465. Q. Mao and P. Walde, Biochem. Biophys. Res. Commun., 1991, 178, 1105. 23 24 Faraday Trans., 1991,87,847. P. L. Luisi, M. Giomini, M. P. Pileni and B. H. Robinson, Biochim. Biophys. Acta, 1988,947,209. M. Harada, M. Kishida and Y. Miyake, Solvent Extraction, Else- 14 15 H. Stamatis, A. Xenakis, M. Provelegioin and F. N. Kolisis, Biotech. Bioeng., 1993,42, 103. Y. Miyake, T. Owari, K. Matsuura and M. Teramoto, J. Chem. Soc., Faraday Trans., 1993,89,1993. 25 26 vier, Amsterdam, 1990, p. 777. A. M. Howe, J. A. McDonald and B. H. Robinson, J. Chem. SOC.,Faraday Trans. 1, 1987,83,1007. P. Karpe and E. Ruckenstein, J. Colloid Znterface Sci., 1991, 141, 16 Y. Miyake, M. Ohkubo and M. Teramoto, Biotech. Bioeng., 1991,38,30. 27 534. P. D. I. Fletcher, R. B. Freedman, J. Mead, C. Oldfield and B. H. 17 M. Caselli, P. L. Luisi, M. Maestro and R. Roselli, J. Phys. Robinson, Colloids Su$, 1984, 10, 193. 18 Chem., 1988,92,3899. M. Harada, M. Adachi and Y.Miyake, J. Chem. Eng. Jpn., 1990, 23, 50. Paper 3/04562E; Received 30th July, 1993
ISSN:0956-5000
DOI:10.1039/FT9949000979
出版商:RSC
年代:1994
数据来源: RSC
|
13. |
Enzyme catalysis at hydrogel-modified electrodes with soluble redox mediator |
|
Journal of the Chemical Society, Faraday Transactions,
Volume 90,
Issue 7,
1994,
Page 987-995
Fernando Battaglini,
Preview
|
PDF (1043KB)
|
|
摘要:
J. CHEM. SOC. FARADAY TRANS., 1994, 90(7), 987-995 Enzyme Catalysis at Hydrogel-modified Electrodes with Soluble Redox Mediator Fernando Battaglinit and Ernest0 J. Calvo*S Departamento de Quimica lnorganica ,Analitica y Quimica Fisica, Facultad de Ciencias Exactas y Naturales, Universidad de Buenos Aires, Pabellon 11, Ciudad Universitaria, 1420 Buenos Aires, Argentina The kinetics of amperometric enzyme electrodes are reported for glwose oxidase immobilized in three different bovine serum albumin (BSA) hydrogel modified electrodes using as soluble electron mediator either ferrocene sulfonate or a pentammine(pyrazine)ruthenium complex. The effect of hydrogel layer thickness, enzyme loading, substrate and redox mediator concentrations on the amperometric response are reported.The kinetic data are analysed in terms of a physico-chemical model that takes into account diffusion of sub-strate, redox mediator and non-linear kinetics in the hydrogel layer and external mass transport in the electro- lyte outside the enzyme layer. Analytical solutions for selected conditions are presented and tested with experimental results. Simulated concentration profiles give an insight into the detailed diffusion-kinetic process and allow the conditions under which the analytical equations are valid to be tested. No difference in the enzyme kinetic pattern (ping-pong mechanism for glucose oxidase) is found for soluble enzyme and for immobilized glucose oxidase (GOx). Rate coefficients are reported for the two redox mediators. Amperometric enzyme electrodes have become of increasing interest in recent years, particularly because of their potential applications in molecular recognition with chemical sensors and in the understanding of electron transfer in biological systems.In vivo, flavoenzyme oxidases catalyse the oxidation of a biologically active substrate with oxygen as cosubstrate. In vitro, however, it has been found more convenient to develop anaerobic redox enzyme electrodes with molecules other than oxygen acting as electron shuttles between the electrode and the redox active centre (FAD/FADH,) of the enzyme. In these systems, the electrochemical and enzymatic reaction sites are physically separated and therefore diffusion of the electron shuttle or diffusion of electrons should be con-sidered.s By combination of a redox enzyme and a suitable electrode reaction, transduction of the enzymatic reaction into an elec- trical current can be achieved with very high selectivity and ~ensitivity.~.~However, several aspects of the physico-chemical mechanisms which are relevant to the electrode per- formance in terms of selectivity, enhanced linearity and lifetime are not yet adequately understood.Electrode surfaces modified by immobilized enzymes via co-reticulation with inactive bovine serum albumin and bifunctional agent glutaraldehyde have been widely used.8-' Since the modified electrodes are prepared from homoge- neous solution, a uniform distribution of active enzymatic sites can be obtained in transparent films, with good adher- ence to the electrode and high enzyme stability.* In a previous paper16 we compared the amperometric response for the oxidation of glucose mediated by ferrocene monosulfonate with soluble glucose oxidase, to the response of an electrode with the enzyme immobilized onto the surface.The position and shape of the catalytic waves with the enzyme in solution and in a thin surface layer on the electrode were the same, indicating that a similar mechanism was operative in both cases. The purpose of this work is to study the effect of several factors that define the performance of amperometric enzyme electrodes with the aim of gaining a better understanding of their physico-chemical mechanisms : (a)the activity of immo-bilized enzyme and concentration of enzymatically active FAD; (b) the thickness of the hydrogel layer, in relation to characteristic reaction lengths for substrate and cosubstrate and the respective diffusion layer^;^ (c) comparison of the enzyme kinetics with native enzyme and the aerobic ping- pong mechanism of the oxidases.The present paper reports results on three different enzyme-modified electrodes : (i) silanized Pt electrodes treated with bifunctional reagent in order to bind the enzyme cova- lently to the electrode; (ii) thin enzyme layers obtained by crosslinking the enzyme with BSA and glutaraldehyde, and (iii) thick BSA-enzyme hydrogel films. Experimental Chemicals Glucose oxidase, GOx (EC 1.1.3.4 type VII-S from Asper-gillus niger) with activity 277 IU mg (Sigma Chem.Co.)was used. The buffer systems employed were 0.1 mol I-' acetic acid-sodium acetate (pH 5.5) and 0.1 mol 1-' potassium hydrogen phosphate (pH 7.0). Glucose solutions were pre- pared from a stock equilibrated in the anomers and kinetics herein are referred to the total glucose concentration. Doubly distilled water was further purified from a Milli-Q Reagent Water System (Millipore Corp.). Ammonium ferrocene monosulfonate, NH,[FeCp,SO,], was synthesized as described elsewhereI6 and pentaamine (pyrazine)ruthenium hexafluorophosphate, [Ru(NH,),(pyz)] PF, (pyz = pyrazine), was obtained by the method described by Taube.17 Bovine serum albumin (BSA) powder 96-99% (Sigma; fraction V) was used as received.t Present address : Department of Chemistry and Biochemistry,Concordia University, 1455 de Maisonneuve Blvd. West, Montreal, Electrodes Quebec, Canada. $ Permanent Research Staff of the Argentine Science Research Spectroscopic grade 99.98% platinum (Johnson and Matthey) Council (CONICET),Argentina. and glassy carbon (Tokai, Japan) 0.2 cm2 disc electrodes 988 embodied in epoxy (Araldite-Ciba Geigy) holders were employed. Platinum electrodes were subjected to cyclic volt- ammetry treatment between the limits of solvent decomposi- tion in 0.1 mol 1-' H,SO, solution until the typical cyclic voltammogram (CV) of clean Pt was obtained. '* Platinum was further oxidized at 2.1 V for 30 min in the same solution, washed with water and dried at 50°C under vacuum for 1 h." After treatment with a 10% toluene solution of 3-aminopropyltriethoxysilane (Aldrich Chemical Co.) for 1 h the electrode was washed successively with toluene, ethanol and water.Silanised Pt electrodes were treated with bifunctional reagent in order to bind GOx covalently by the method of Castner and Wingard13 (electrode type I) or to form thin enzyme membrane layers by crosslinking with BSA (electrode type 11).In the first case, after reaction with 2.5% glutaralde- hyde solution in pH 7 buffer, the electrode was treated with 2.5 mg 1-' GOx solution in the same solution at 4 "C for 4 h. The different steps in the surface modification were tested by recording CVs of attached redox probes such as ferrocene carboxyaldehyde chemically bound to the Pt aminosilane and 1-aminoanthraquinone bound to glutaraldehyde-Pt aminosilane.In the second case, the amino groups at the silanised surface were further crosslinked with 1.5 pl 20% BSA (Sigma Chem. Co.) and 2 pl 4% GOx solutions in acetic acetate buffer (pH 5.5) and 1 pl 2.5% glutaraldehyde solution by rapid mixing. After 2 h the gel-modified surface was thor- oughly washed with buffer solution. Thick hydrogel films (electrode type 111) were prepared by immobilizing GOx in BSA-glutaraldehyde hydrogel in paral- lel on carbon electrodes and on quartz windows by the fol- lowing procedure:13 Solutions of 2.5% glutaraldehyde, 17% BSA and GOx of different concentrations (1-6%) in buffer electrolyte (pH 7.0) in the ratio 6 : 5 :4 were mixed together on the surfaces.The thickness and uniformity of the films on the final hydrogel-modified electrode were achieved using a similar procedure to that described by Wingard et aE.:I2The cross- linking solution was dropped onto a Teflon plate, on top of which the disc electrode was then positioned, adjusting the space between both surfaces to the desired thickness (0.05 or 0.5mm) by means of a micrometric device. The gap was thus filled by the crosslinking solution. After 3 h of gel formation, the excess material was separated from the edges and the Teflon plate was removed. The resulting BSA crosslinked gels were held strongly to the surface, although there was no chemical link in this case, the only anchorage being through surface roughness.In all cases the electrodes were left overnight in buffer solu- tions at 4°C in order to allow the soluble components to leach out of the gels. The concentration of active GOx in the thick gels was determined spectrophotometrically with the quartz windows treated in the same manner as for the GC electrodes. The absorbance at 450 nm was measured in oxygen-saturated buffer and in 0.1 moll-' glucose solution. Assuming a differ- ential molar absorption coeflicient of 1.31 x lo4 mol 1-'cm-' , 20 the same as for native GOx, for the enzymatically active FAD, the concentration of GOx in the gels was esti- mated from the Lambert-Beer law and the thickness of the layer.The water content of the thick BSA films was determined on a larger sample of hydrogel by drying at 105°C until a constant weight was obtained. A typical water content value is 62% in BSA-glutaraldehyde crosslinked gels. Experiments were carried out at 27 "C and the electrodes were stored in buffer at 4°C. J. CHEM. SOC. FARADAY TRANS., 1994, VOL. 90 Equipment Cyclic voltammetry experiments were carried out with a purpose-built operational amplifier potentiostat and triangular-wave generator. Experimental data were acquired and processed with a PC-XT computer interfaced with a 12-bit AD-DA card. A 15 ml conventional three-electrode cell and a 3 ml miniature ce1121 were employed with a large Au foil auxiliary electrode, and a saturated calomel electrode (SCE) was used as a reference; all potentials are quoted with respect to the SCE.Digital Simulation An explicit point method was used to simulate digitally diffu- sion and enzyme kinetics in the gel layer at a planar electrode for a two-substrate enzyme (immobilized glucose oxidase) catalysing the anaerobic oxidation of glucose, with a soluble one-electron redox mediator, as described elsewhere.22 In the aqueous solution outside the gel layer a diffusion coefficient, Do = D, = D, = cm2 s-l, was used for sim- plicity, while D) is the value in the gel corrected for excluded volume DS = PJ'Dj (Pi is the partition coefficient of the speciesj determined experimentally, see below). Results Electrochemical Behaviour of Hydrogel-modified Electrodes The characteristics of hydrogel-modified electrodes for single electron transfer from redox species in solution such as fer- rocene sulfonate depend upon the preparation method of the gel electrode.Fig. 1 shows a typical CV of NH,[FeCp2S03] for GOx covalently attached to Pt via aminopropylsilane-glutaraldehyde (electrode type I). Instead of the CV peak characteristic of the oxidation of soluble NH,[FeCp,SO,] on clean platinum' a symmetrical sigmoidal oxidation wave centred at (0.37 V) and a small cathodic wave in the 14 I 0 0.2 0.4 0.6 EIV Fig. 1 Cyclic voltammetry of Pt electrode modified by silane-glutaraldehyde-GOx covalently attached: (-) 5 mmol I-' NH,[FeCp,SO,]; (-* * * .) same with 0.1 mol 1-glucose, buffer pH 7; 2 mV s-' J.CHEM. SOC. FARADAY TRANS., 1994, VOL. 90 reverse scan are apparent. Similar waves were reported, ”sZ3 however, no explanation was advanced by those authors. Bond and co-workers have recently proposed a micro-scopic model to explain the electrochemistry of several redox proteins at electrode^.'^ The model assumes that mass trans- port to the electrode occurs by radial diffusion when the density of surface active sites is low and linear diffusion pre- vails when the density of electrochemically active sites at the electrode is increased sufficiently to cause overlap of the diffu- sion zones. The results with BSA-GOx linked through silane-Pt bonds in the present work may be interpreted in terms of a partly blocked electrode where patches at the electrode surface accessible to the diffusing redox species are sufficient- ly separated by protein domains so that radial diffusion pre- vails.This is further confirmed since the sigmoidal-shaped curves (Fig. 1) with half-wave potential, El/,,equal to the formal redox potential of ferrocene monosulfonate, Eo’,l6 give linear plots of E us. log(i, -i)/i with a slope close to 2.303 RT/F, which is consistent with a reversible redox reac- tion at an array of small and well spaced micr~electrodes.’~ (Here subscript L denotes ‘limiting’ and the other symbols have their usual meanings.) Cyclic voltammetry of NH4[FeCp,S0,] in pH 7.0 buffer at BSA-GOx-glutaraldehyde (electrode type 111) shows a quasi-reversible and linear diffusion behaviour (AEp = 70-90 mV and i,, dip, = 1, where subscript p denotes peak, a and c denote anode and cathode, respectively), comparable to the CV of ferrocene monosulfonate in aqueous electrolyte. l6 Similar results were obtained with [Ru(NH,),(pyz)]’+ : AE, = 60-70 mV and ip,Jip, = 1.From the linear plots of peak current vs. the square root of sweep rate, the appar- ent diffusion coefficients in the gel, OK, were obtained from NH,[FeCp,SO,] (2.4 x cm2 s-l) and [Ru(NH,),(pyz)]’+ (3.5 x lop6 cm’ s-l) for an electrode reaction with fast kinetics using eqn. (1):26 i, = (2.69 x 105)~3/2~~ycg (1)u1/2 where Cg is the analytical concentration of reduced redox mediator. From these results it can be concluded that linear diffusion of the redox mediator is operative for BSA-glutaraldehyde hydrogel modified electrodes not covalently attached to the surface via silane and that the theory of voltammetry for aqueous solutions2’ can be applied in this case.Note that thick (0.5 mm) hydrogels with high water content (ca. 62%) containing soluble redox species show electrochemical and diffusion behaviour comparable to that of aqueous solutions. The enzyme, however, is present only in the hydrogel layer adjacent to the electrode surface. Dk = Pi D, ,where D, is the diffusion coefficient in aqueous solution and Pi, the square of the partition coefficient for R between the hydrogel and the external electrolyte. Redox Enzyme Catalysis The catalytic waves for anaerobic enzymatic oxidation of D-glucose by GOx immobilized at electrodes, with different redox mediators in aqueous solutions, are shown in Fig.2. In electrodes of types I1 and 111, the position and shape of the symmetric redox enzyme catalytic waves are the same as described for native GOx and redox mediator in solution, which has been shown elsewhere.16 These waves reflect the changes in surface concentration of the oxidized form of the redox couple. For electrodes of type I, with no BSA matrix, a very small catalytic current was observed as shown in Fig. 1 (dotted 100 80 60 %.a 40 20 0 0.2 0.4 0.6 0.8 Elv Fig. 2 Catalytic waves due to the oxidation of glucose (0.12 mol I-’) on a GC electrode modified by hydrogel (type 111) for different redox mediators: 1, 0.25 mmol 1-’ hydroquinone; 2, 0.5 mmol 1-’ [RU(NH~)~(PYZ)]*’;3, 0.5 mmol I-’-NH,[FeCp,SO,] and 4, 0.5 mmol 1 -Fe(CN):-line), which can be attributed to the small enzyme loading in that case.For thin enzyme layers covalently attached via silane to a Pt rotating-disc electrode (type 11), the catalytic limiting current increases with rotation frequency. This indicates that external mass transport of the redox cosubstrate becomes important under conditions of substrate saturation, i.e. high concentration of substrate in the external solution. The dependence of catalytic current on rotation frequency was found to follow the Koutecky-Levich relationship,’* as can be seen in Fig.3. The dependence of the catalytic currents on substrate con- centration for both types of gels (I1 and 111) was studied by 2.0I I 0.0 I I I 8 I I w-1 /2/HZ-1/2 Fig. 3 Dependence of the catalytic current at 0.55 V with rotation frequency for a Pt-silane-BSA-GOx-glutaraldehyde gel electrode (type 11) 01 1 I 1 1 I I C,/mmol I-' Fig. 4 Dependence of the catalytic current on substrate concentra- tion for glucose oxidation on GC electrodes with high GOx loading (20 pmol 1-I) in a thick BSA layer (type 111) in pH 7.0 buffer solution with NH,[FeCp,SO,] cosubstrate: (*) 2.5; (+) 5.0 and (m) 10 mmol 1-' chronamperometry. The electrode potential was stepped from 0.1 to 0.6 V where the limiting catalytic current was obtained for both couples.The potentiostatic transients were similar to those observed for native enzyme in solution and a steady- state current was achieved in a few seconds; the transient duration depended on the concentrations. The effect of enzyme concentration on the catalytic current for native GOx in solution and soluble NH,[FeCp2S03 J was reported in a previous paper.16 However, at high enzyme concentration there are practical limitations for these studies in solution since excessive foam is formed at the electrolyte/ air interface which may drag protein away from the solution. With enzymes immobilized in hydrogels, on the other hand, much higher enzyme loadings can be reached. Fig. 4 depicts typical curves of the catalytic limiting current us.substrate concentration for high loadings of GOx (20 pmol 1-I) and three concentrations of cosubstrate : An extended linear behaviour is apparent in all cases at low glucose concentra- tion; for 10 mmol 1-' ferrocene monosulfonate the linear behaviour is observed up to a substrate concentration of 30 mmoll-'glucose. Hydrogels formed by co-reticulation of BSA with enzyme and glutaraldehyde are known to provide a matrix support where enzymes can be immobilized with high stability.' While enzyme activity decays in a few days for soluble enzyme and electrode type 11, it is apparent that hydrogels of type I11 are very stable for at least two weeks when stored in buffer solution at 4 "C. Discussion The redox enzyme catalytic cycle for the anaerobic oxidation of substrate, S, by soluble cosubstrate, 0, catalysed by the enzyme in the hydrogel layer can be represented by eqn.(1)-(111): electrodeR-O+e (1) 20 + Ered Eox+ 2R + 2H' (11) E,, + S ki r ES Ered+ P (111) k-t J. CHEM. SOC. FARADAY TRANS., 1994, VOL. 90 The concentrations of R, 0,S and all forms of enzyme in the layer are given by CR, Co, Cs, CET, CER, CEO and CEs. Elec-trochemical reaction (I) takes place at the electrode surface with fast kinetics. Reactions (11) and (111) occur within the enzyme layer, 0 < x < I, adjacent to the electrode surface. Reaction (11) represents the overall oxidation of the reduced form of the enzyme active site (FADH,), by the soluble oxidant, 0, generated at the electrode surface and diffusing away into the reduced enzyme zone in the gel adjacent to the electrode.Reaction (111) represents the enzyme catalysis char- acterized by non-linear Michaelis-Menten kinetic^.^' Since reaction (I) is a one-electron transfer step and reac- tion (11) a two-electron reaction, 2 mol of 0 are needed per mol of Eredfrom which the stoichiometric coefficient of two arises in eqn. (11). The overall rate coefficient for the oxida- tion of FADH, can be split into two one-electron elemental steps: 0 + GOx(FADH2) GOx(FADH*)+ R + H+ (IIa) 0 + GOx(FADH') GOx(FAD) + R + H+ (IIb) where, in eqn. (11), k = k,kI,/(k,+ k,J, GOx(FADH,) = Ered and GOx(FAD) = Eox. Assuming a homogeneous distribution of enzyme in the layer for gels prepared from homogeneous solutions by cross- linking with 100% chemical yield,' under steady-state condi- tions the following system of simultaneous second-order differential equations can be obtained by combination of the kinetics of reactions (11) and (111) with Fick's law of diffusion in the gel layer: a2co 2kcat CET Do 3= 1 + KJC, + Ko/Co (3) where KO = k,,Jk and K, = k,,J(k, + k-') are the character- istic Michaelis-Menten constants for the ping-pong mecha- nism of glucose oxidase under aerobic condition^.^' Substitution by dimensionless variables3 ' y = x/l (4) 0s = CJK,; bo = c, D,/(2Ds K,) (5) and the Thiele modulus, 4, which compares the enzyme reac- tion rate to the rate of diffusion in the enzyme layer :31 and depends on the properties of the gel layer (CET, kc,,, and D,), but not on the concentrations of substrate and cos~bstrate,~~in eqn.(2) and (3) gives: (7) and with So= Do K0/2DsKs. Note that eqn. (7) and (8) are symmetrical and can be reduced to two particular cases of one-substrate problem, as will be shown below.32 This problem has been tackled by several authors'~~~-~~ and an analytical solution of the system in close form cannot be obtained because eqn. (7) and (8) are non-linear in oo and J. CHEM. SOC. FARADAY TRANS., 1994, VOL. 90 a,. However, approximate solutions for limiting cases can be obtained analytically and these solutions give an insight into the detailed processes that take place in the enzyme layer at the electrode The boundary conditions for amperometric enzyme elec- trodes with soluble reduced mediator in solution are3’ at y=o: aajay = o (9) and 60 = (PRCg Do/2Ds Ks) = and at y = 1: 8ao/8y = -Bi, oo(1) and aoday = ~i~[~;-as(i)~ (12) where Bi,= lkm,JD,,the dimensionless mass-transfer Biot number,” and km,i is the external mass transport constant for the solution outside the enzyme layer.The catalytic current is given by : i = KJP, r)[a~,/ayly=0 (13) We shall consider three limiting cases for which approximate analytical solutions can be obtained: Case I : When inequality (14) holds the oxidation of GOx(FADH,) by 0 [reaction (11)], is the rate-determining step in the hydrogel layer adjacent to the electrode (substrate zero-order approximation).Under these conditions, eqn. (7) becomes: with 42/Xo= (l/po) and po = (DO/2kCET)’/2 (the kinetic reaction layer thickness for the reoxidation of FADH, by analogy with the homogeneous cases).16 For electrode type I1 (thin layer) and from eqn. (A6) in the Appendix for high rates of external mass transport, tanh(l/pL,)< km,,/Do,we obtain by substituting in eqn. (13): Eqn. (16) is a Koutecky-Levich type equation36 for the rotating-disc electrode (RDE) where km,,= 1.554D213v-1/6 W1l2(the rate of mass transport), v is the kinematic viscosity and W the rotation frequency in s-’ which has been experi- mentally verified as shown in Fig. 3. The experimental diffu- sional gradient in Fig.3 is 1.23 x lo4 A-’ s-l12 and compares well with the value 1.5 x lo4 A-’ s-’l2 calculated using eqn. (16). The first term in the bracket is the inverse of the Levich convective diffusion equation and represents the external mass transport of the redox mediator. The second term is the inverse of the catalytic limiting current generated at the enzyme layer for infinite rate of external mass transport. For electrodes of type I with a thin enzyme layer (1 4 po), where tanh(l/p,) -+ l/po: 991 The thickness of the gel-enzyme layer was estimated to be of the order of 15 pm from the extrapolated data in Fig. 3. For extremely thin enzyme layers, tanh(l/po) -,0, and thus eqn. (19) predicts currents described by the Levich equation: The effect of enzyme catalysis is negligible compared with the con- vective diffusional flux of redox cosubstrate.For very thick gel layers [I % po, tanh(l/po) + 13, the same equation for the current as that obtained for soluble enzyme is found :’ i = FA(2Dk k’CET) ‘I2 C,* (18) where i is independent of the rotation frequency. Eqn. (18) is equal to the limiting catalytic current in an EC’ mechanism for a stationary electrode; however, note that Dk and k’ cor-respond to the values in the gel. In this case all the redox-enzyme catalysis and internal dif- fusion occur within the thick enzyme layer and because of the high water content in the hydrogel and the homogeneous dis- tribution of the enzyme, the kinetic pattern is similar to that of a homogeneous enzyme solution adjacent to the elec- trode.l6 Eqn. (18) has been derived on the assumption that all R species that reach the electrode surface are transformed into 0 under diffusion conditions [boundary condition in eqn. (lo)]. At lower electrode potentials, however, the surface con- centration of 0 is potential dependent and therefore the cata- lytic current is a function of potential as can be seen from the characteristic wave shape. For soluble enzyme with a large concentration of substrate, it has been shown’6*22 that a pseudo-first-order analysis of the redox catalysis leads to the equations derived by Saveant and Vianello to describe the EC’ voltammetric catalytic waves at low scan rate:37 . nFA(Dk kf)1/2Cg I= 1 + exp(-O*) (19) where 8* = ornF/RT(E -Eo’)-ln[(kfD’)1/2/k,], and k, = 2k&, o! is the charge-transfer coefficient, k, the rate coefi- cient for the heterogeneous electrode reaction and n the number of electrons exchanged in reaction (I) for an irrevers- ible system such as benzoquinone-hydroquinone. The half-wave potential for slow heterogeneous kinetics, is in this case a function of both the heterogeneous elec- trode reaction as well as the catalytic homogeneous rate coef- ficien ts : RT RTEl,, = Eo’ + -In Dkk, --In k, (20)2unF anF while for fast redox couples El,, x E0‘.37The position of the catalytic waves and the maximum catalytic current are dependent on the redox potential of the electron mediator and on the rate of reduced enzyme reoxidation [reaction (11)] as described by eqn.(19) and (20). For immobilized enzymes it is rather difficult to know accurately the actual concentration of the active enzyme at the electrode surface; however, the ratio of the catalytic limit- ing currents for different electron mediators at the same enzyme-modified electrode can be computed : and 0;can be measured independently from cyclic voltam- metry in the absence of substrate or from the Levich slope with the RDE. Thus, the reoxidation rates of GOx can be compared for different mediators with the same modified electrode. For instance, taking into account the values of 0; from cyclic voltammetry and of k from experiments with soluble enzyme and applying eqn. (18), the calculated ratio for [R~(NH~)~(pyz)l~ + as species 1 and NH,[FeCp,SO,] as species 2 is 1.55, in good agreement with the experimental value of 1.30.The difference (15%) would indicate that the kinetics of reaction (11) using GOx(FADH,) immobilized in hydrogels are not severely altered with respect to native enzyme in solution. This was further confirmed by studies on the substrate concentration dependence (see below). The simulated concentration profiles of oxidized mediator, 0,for thin (I = 50 pm) and thick (I = 300 pm) enzyme layers are compared in Fig. 5 for the conditions of case I. The profile for the thick layer is identical to the profile simulated for soluble enzyme with the same parameters as for the gel layer.22 It can be seen that the reaction layer for the reoxida- tion of the enzyme (FADH,) lies within the hydrogel layer, which validates the use of eqn.(18) and (19) for solutions and gel containing enzyme. Case II : With high enzyme loading an extended linear amperometric response with substrate concentration was found (Fig. 4), as desired for analytical purposes. Digital simulated concentra- tion profiles of gels with high enzyme loading in Fig. 6 show that close to the electrode surface, CEOe CET. Under these conditions Cs cannot be considered to be con- stant (as for case I), and for low Cs and when (K&) 9 1 + (Ko/Co), eqn. (8) becomes: Note from the solution of eqn. (All) in the Appendix that, in analogy with for the redox cosubstrate in case I, a sub- strate characteristic kinetic distance, ps, can be defined as: Ps = (4Kslkcat cET)1'2 (23) which describes the substrate depletion layer close to the elec- trode.This is the result of a high 4 = I/ps in the layer which may result either from a large CET,a thick enzyme layer, or a low diffusivity of substrate, D,. Inspection of the simulated profile in Fig. 6 shows that the enzyme layer behaves in this case as a membrane for substrate diffusing towards the elec- trode. By substituting eqn. (A14) into eqn. (13) and the value of the flux of 0 in the surface of the electrode obtained in eqn. Oa8 n 0 0.01 0.02 0.03 0.04 xjcm Fig. 5 Simulated concentration profiles of cosubstrate, C,/C,* , at steady state near the electrode surface for CET= 3.5 pmol-', K$C, = 0.1 and K&,* = 10 for different hydrogel layer thicknesses: (-) I, = 50 pm and (---) I, = 300 pm J.CHEM. SOC. FARADAY TRANS., 1994, VOL. 90 1.o 0.8 0.6 * 9 c, 0.4 0.2 0 / .. I I I 0 0.02 / 0.04 0.06 0.08 xlcm Fig. 6 Simulated relative concentration profiles at steady state for C,, = 20 pmoll-'; K$C,* = 10 and K,/C: = 0.1:(-) C,, (---) C, and (. . . * *) C,, (A15), the catalytic current is given by: which predicts the linear response observed in Fig. 4. Using the data obtained from independent experiments under differ- ent conditions for this system, a calculated slope of 8 A 1 mol-was obtained, close to the experimental value of 3.8 A 1 mol-l. Case III: For a low enzymatic rate, a low consumption of substrate is expected and os is relatively constant.Inspection of Fig. 7 suggests that there is some distance, 0 < E < 1, that corre- 1.o i,, ,, :0.8 .. 0.6 * 9 c, 0.4 0.2 0 0 0.01 I 0.02 0.03 0.04 xjcm Fig. 7 Simulated relative concentration profiles at steady state for CET = 3.5 pmol l-', K&,* = 10 and Ko/C,* = 0.1: (-) C,, (---) C, and (* . . . .) C,, J. CHEM.SOC. FARADAY TRANS., 1994, VOL. 90 sponds to the limit of the enzyme reaction with a small change of substrate concentration, CT, within 0 < y < E < 1. Then eqn. (7) becomes: For (K&) 9 1 + (Ko/Co)from eqn. (A21) we find for this case: which describes the dependence of the catalytic current with substrate and cosubstrate.Note that eqn. (26) does not predict a linear behaviour with substrate concentration. Inversion of eqn. (26), instead, gives the same linear relation- ship verified experimentally for soluble enzyme :' which is predicted by the comprehensive theory of ampero- metric electrodes of Albery et al.' Fig. 8 depicts typical linear plots of C,*iP2 us. CC1 for both redox mediators investigated. These plots are similar to the plots obtained for soluble enzyme under similar conditions and reflect that for thick enzyme layers, where all the changes in Co and Cs are contained within the enzyme layer and no external diffusion would occur ; thus a similar kinetic pattern to that in solution is expected. Note, however, that the kinetic and diffusion parameters should have different values from those for glucose oxidase in aqueous solution.51 0 20 40 60 80 100 120 140 11 I Cil/l rno1-l Table 1 summarizes the intrinsic kinetic parameters for the anaerobic oxidation of glucose catalysed by native and immobilized glucose oxidase in BSA hydrogels with two redox cofactors : NH,[FeCp,SO,] and [Ru(NH,),(pyz)] * + For comparison, the kinetic data for native glucose oxidase under aerobic conditions3' are given in the last line. Since BSA and glucose oxidase in homogeneous solution are the precursors of the hydrogels, a near 100% chemical yield is expected with a homogeneous distribution of active sites within the enzyme layer.8 As a result, transparent films, with good mechanical resistance and high enzymatic stability, are obtained.Owing to the high water content (> 60%) the transport of substrate and redox cosubstrate by diffusion is not hindered in the hydrogels at the electrode, as shown by cyclic voltammetry for electrodes of type 111. The small decrease in the diffusion coeficient with respect to aqueous solutions is in the expected order for the solution excluded volume in the polymer according to the models for diffusion in heterogeneous media.4' The rate of reoxidation of enzyme FADH,, k' [reaction (11)], depends on the nature of the redox mediator, with the fastest rate for the natural cofactor dioxygen. Also, the enzyme turnover, Kat(= kcaJPJ, exhibits a lower value under anaerobic conditions for enzyme both in solution and immobilized onto the electrode.Note, however, that within the experimental error and the concentration range where eqn. (18) and (27) can be applied, k and k,,, for a given redox mediator have the same values for enzyme in solution and immobilized in BSA hydrogels. This is further evidence that the BSA hydrogels do not alter significantly the flavoprotein microenvironment, and there- fore the enzyme has great stability. Weibel and Bright2' have compared the kinetics of native glucose oxidase with the enzyme immobilized onto glass par- ticles, and reported a constant ratio for k/kc,, of 2 x lo3 mol 1-' for soluble and insoluble enzyme under aerobic condi- tions. Under anaerobic operation constant ratios (k/k,,J for soluble and BSA gel-immobilized glucose oxidase are also apparent in the present work: 3.9 x 10, and 1.6 x lo3 1 mol-for the ferrocene and the ruthenium complexes, respectively. As for aerobic conditions a ping-pong mechanism is valid for glucose oxidase and a one-electron redox mediator, however, the intrinsic kinetic parameters differ from those for aerobic kinetics and depend strongly upon the nature and charge of the redox cofactor.The different values observed for the apparent Michaelis constant Kk = KdP, [kl/kcat in reaction (111)] for both redox mediators and oxygen are more difficult to rationalize since they would suggest that the value of k, depends on the nature of the redox cosubstrate.K, has been obtained from the ratio between the slope and the intercept of a plot of C;l i-, us. C; (Fig. 8). Inspection of the simulated concentra- Fig. 8 Test plot of eqn. (27) for two soluble redox mediators: (0) tion profiles in Fig. 7 for hydrogels and in Fig. 2(b) 3 of ref. 225 and (x) 10 mmol I-' NH,[FeCp,SO,]; (m) 5.5 mmol I-' for soluble enzyme under conditions where eqn. (27) holds, CRU(NH3)5(PYZ)I2+ indicates that there is a significant contribution of mass Table 1 Kinetic parameters for the anaerobic oxidation of glucose catalysed by GOx in solution and immobilized in BSA hydrogels (type 111) system Da/cm2 s-' k"/l mol-' s-' k*tc/S -' KMd/molI-' NH,IFeCP2s031{ solution gel 4.0 x 1.2 x 105 310 0.088 2.4 x 1.1 x 105 460 150 0.140 solution 4.3 x 1.9 x 105 110 0.061 CRU~NH,)5(PYZ)l2+ (gel 3.5 x 1.8 x 105 110 0.078 O,, native GOx 1.9 x 10-5' 2.1 x lo6" 874* 0.033# a From eqn.(1). * From eqn. (18). 'From eqn. (27) and data in Fig. 8. From eqn. (27). 'Ref. 40. Ref. 38. Ref. 39. transport of substrate which has not been taken into account in deriving eqn. (28). Internal diffusion of substrate could therefore be the source of the observed differences in the values of Ks derived. It is generally accepted that internal diffusion of substrate affects the value of the apparent Michaelis constant.43 The apparent Ks for the hydrogels should also be greater than the value for soluble enzyme owing to the gel/aqueous solution partition of substrate that results from the excluded volume of water in the gel as is apparent from Table 1.Conclusions In BSA hydrogels very high stability of glucose oxidase (electrode type 111)can be obtained. The ping-pong mechanism for glucose oxidase is operative in electrodes modified with enzyme in BSA hydrogels. The rate coefficients k and k,,, for soluble native enzyme and immobilized enzyme have the same values within experi- mental error. However, k,,, and k depend on the redox medi- ator used in the anaerobic oxidation. The authors thank the Argentine Science Research Council (CONICET) for financial support and Lic. Luis Baraldo for the synthesis of the ruthenium complex. Appendix Case I : From eqn. (19, azo0 aYz Jf-0 Integrating eqn.(Al), with 4’/Xo= (2/p0)’ yields: ao(Y) = A 1 exdYllPo) + A, exd -Ybo) From the boundary conditions in eqn. (12) and (13): aty=O ao(0)= A, + A, = aty=1 OO(1) = A, exP(VPo) + A, exP(-VPo) = 2A sinh(l/p,) + a: exp(-I/po) and from eqn. (14): 2A,l a: 1(2)= -cosh(I/po) --ex& -l/po) = -Bio ao(1) y=l Po PO A, can be eliminated, replacing ao(l)for eqn. (A4), then: Case I I : From eqn. (22), It can be seen from Fig. 6 that if the enzyme gel layers are thick enough, then there is a region inside which reaction (I) controls the kinetics and reaction (11) is the rate-determining step outside that layer. The boundary between both zones is J. CHEM. SOC. FARADAY TRANS., 1994, VOL. 90 at y = E,~where: as(&)= P,C,*/K, = a,*; (2)= 0 Y=& 649) Then, integrating eqn.(A7): as = 4 exd4y) + A2 exp(-4Y) (AW at y = 0, aaday = 0, thus A, = A, and replacing in eqn. (A8): Integrating eqn. (A12): Since at y = E (Fig. 6), also = 0; A, an0 (A14)= (F)= -40; tanh(4s) y= 0 For thick films $E -+ a,tanh($e) + 1 and Case III: In this case, asx a: can be considered relatively constant within 0 < y < E (see Fig. 7), and: Integrating eqn. (A16): 00 = -” 1 + l/as*y2 + A,y+A, Since, at y = 0: and from Fig. 7 at y = E: (2)y=E= go(&)= 0 then, following the same procedure as for case 11, we obtain: References A. E. G. Cass, D. G. Francis, H. A. 0. Hill, W. J. Aston, I. J. Higgins, E. V. Plotkin, L. D. L. Scott and A. P. F. Turner, Anal.Chem., 1984,56,667. M. J. Green and H. A. 0. Hill, J. Chem. SOC., Faraday Trans. 1, 1986,82,1237. Y.Degani and A. Heller, J. Phys. Chem., 1987,91,1285. A. P. F. Turner, 1. Karube and G. S. Wilson, Biosensors, Funda- mentals and Applications, Oxford University Press, New York, 1987. A. Heller, Acc. Chem. Res., 1990,23, 128. B. A. Gregg and A. Heller, J. Phys. Chem., 1991,95, 5976. J. CHEM. SOC. FARADAY TRANS., 1994, VOL. 90 995 7 8 9 W. J. Albery, P. N. Bartlett, B. J. Driscoll and R. B. Lenox, J. Electroanal. Chem., 1992,323, 77. D. Thomas and G. Brown, Biochimie, 1972,54,229. L. D. Meloy and J. T. Maloy, Anal. Chem., 1975, 47, 299; 1976, 48,1597. 27 28 29 R. S. Nicholson and I. Shain, Anal. Chem., 1964,36,706. (a)J. Koutecky and V.G. Levich, Zh. Fiz. Khim., 1958,32, 1565; (b) J. M. Cooper, M. Alvarez-Icaza, C. J. Mc.Neil and P. N. Bartlett, J. Electroanul. Chem., 1989,272, 57. A. Fersht, Enzyme Structure and Mechanism, Freeman, New 10 11 12 13 14 15 C. Bourdillon, J. P. Burgeois and D. Thomas, J. Am. Chem. SOC., 1980,102,423 1. T. Yao, Anal. Chim. Acta, 1983,148,27. L. B. Wingard Jr., L. A. Cantin and J. F. Castner, Biochim. Biophys. Acta, 1983,748,21. J. F. Castner and L. B. Wingard Jr., Biochemistry, 1984,23,2203. J. K. Leypoldt and D. A. Gough, Anal. Chem., 1984,56,2896. K. Yokoyama, E. Tamiya and I. Karube, J. Electroanal. Chem., 1989,273,107. 30 31 32 33 34 York, 2nd edn., 1984. H. J. Bright and D. J. T. Porter, in The Enzymes, ed. P. Boyer, Academic Press, New York, 1975, vol.12, p. 421. D. A. Gough and J. K. Leypoldt, AppI. Biochem. Bioeng., 1981,3, 175. B. Atkinson and D. E. Lester, Biotechnol. Bioeng., 1974,26, 1299. T. Schulmeister and F. Scheller, Anal. Chim. Acta, 1985, 170, 279; 1985,171, 11 1. C. Bourdillon, J. M. Lava1 and D. Thomas, J. Electrochem. SOC., 16 17 E. Liaudet, F. Battaglini and E. J. Calvo, J. Electroanal. Chem., 1990,293, 55. P. Ford, D. F. P. Rudd, R. Gaunder and H. Taube, J. Am. Chem. 35 36 1986,133,706. J. K. Leypoldt and D. A. Gough, Anal. Chem., 1984,562896. P. N. Bartlett and R. G. Whitaker, J. Electroanal. Chem., 1987, Soc., 1968,90, 1187. 224,27. 18 E. Gileadi, Ekirowa-Eisner and J. Pencier, Interfacial Electro- 37 J. M. Saveant and E. Vianello, Electrochim. Acta, 1965, 10,905. 19 chemistry, Addison Wesley, New York, 1975. J. R. Lenhard and R. W. Murray, J. Electroanal. Chem., 1977, 78, 195. 38 39 M. Weibel and H. J. Bright, J. Biol. Chem., 1971,246,2734. G. Wilson and A. P. F. Turner, Biosens. Bioelectron., 1992, 7, 165. 20 21 22 23 24 25 M. Weibel and H. J. Bright, Biochem. J., 1971,124,801. P. N. Bartlett and R. G. Whitaker, Anal. Chem., 1989,61, 2803. F. Battaglini and E. J. Calvo, Anal. Chim. Acta, 1992,258, 151. P. Bianco, J. Haladnjian and C. Bourdillon, J. Electroanal. Chem., 1990,293,15 1. F. A. Armstrong, A. M. Bond, H. A. 0. Hill, B. N. Oliver and I. S. M. Psalti, J.Am. Chem. Soc., 1989, 111,9185. F. A. Armstrong, A. M. Bond, H.A. 0.Hill, I. S. M. Psalti and 40 41 42 K. E. Gubbins and R. D. Walker, J. Electrochem. SOC., 1965,112, 469. J. Crank, The Mathematics of Diflkion, Oxford University Press, London, 2nd edn., 1975. (a) L. Goldstein, in Methods in Enzymology, ed. K. Mosback, Academic Press, New York, 1976, vol. 44, p. 397; (b) K. J. Laidler and P. S. Bunting, in Methods in Enzymology, ed. K. Mosback, Academic Press, New York, 1980, vol. 64, p. 227. 26 C. G. Zoski, J. Phys. Chem., 1989,93,6485. A. J. Bard and L. R. Faulkner, Electrochemical Methods. Funda- mentals and Applications, Wiley, New York, 1980. Paper 3/06435B; Received 27th October, 1993
ISSN:0956-5000
DOI:10.1039/FT9949000987
出版商:RSC
年代:1994
数据来源: RSC
|
14. |
Solid-state hydrolysis of aspirin |
|
Journal of the Chemical Society, Faraday Transactions,
Volume 90,
Issue 7,
1994,
Page 997-1001
Matthew C. Ball,
Preview
|
PDF (560KB)
|
|
摘要:
J. CHEM. SOC. FARADAY TRANS., 1994, 90(7), 997-1001 Solid-state Hydrolysis of Aspirint Matthew C. Ball Department of Chemistry, Loughborough University of Technology, Loughborough , Leicestershire, UK LEI1 3TU The hydrolysis of solid aspirin has been studied, using singlecrystal material, over the temperature range 368-387 K at humidities varying between 7 and 24 kN m-2. The reaction follows the Avrami-Erofeyev equation, indicating that nucleation-and-growth kinetics are rate controlling, rather than the through-solution processes suggested previously. The Arrhenius parameters increase with increasing water vapour pressure and a com-pensation effect seems to operate. Water sorption and microscopy have been used to aid the kinetic analysis. Relatively few mechanistic studies of reactions of solid organic pharmaceutical compounds have been published.'-3 The main general reason for this is probably the relatively high volatility of organic reactants or products which leads to the use of low reaction temperatures; long reaction times are therefore necessary. Low melting points leading to possible eutectic formation between reactant and product are also a problem.An important factor which also affects the study of pharmaceutical materials is the general formulation require- ment of good shelf-life, which leads to the rejection as drugs of compounds that are too reactive. In addition, the range of mathematical models considered in the testing of pharmaceuticals is frequently restricted to order-of-reaction or shrinking-core equation^.^ This is pre- sumably on the grounds that rates and temperature coeff- cients, rather than mechanisms, are important in shelf-life studies.Such restrictions cause difficulties, however, because not all possible solid-state mechanisms can be described by these equations. This often results in an apparently poor fit of experimental data to model, and most importantly, the basic understanding of the mechanisms of many reactions is reduced. However, these strictures do not apply only to studies on pharmaceuticals, but to many other solid-state kinetic ~tudies.~ The hydrolysis of aspirin is one of the more closely studied reactions of a pharmaceutical compound. Work has been done on the solution hydrolysis as a function of pH,596 and on the hydrolysis of aqueous suspensions.' The effects of various alkaline soaps and lubricants on the suspension reaction' have also been studied.The solid-state reaction was first studied by Leeson and Mattocks,' who examined the effect of temperature and humidity on aspirin powder in the range 308-383 K. Later studies of the solid-state hydrolysis of aspirin'Ov'l and some have also been made. The effects of various excipients on the solid-state reac- tivity have also been ~tudied.'~,'~ In the Leeson and Mat- tocks study,g reasonable rates were obtained between 323 and 383 K with water vapour pressures between 6.1 and 31 kN m-2; S-shaped decomposition-time curves were obtained. The activation energy derived from this work (65 kJ mol-') is close to that derived from solution hydr~lysis.~ Later studies confirm the S-shaped curves but query the details of the mechanism suggested by Leeson and Mattocks.Hasegawa et aI." suggested that the acceleratory period fol- lowed a power law, and derived an activation energy of about 125 kJ mol-' for aspirin and two substituted compounds. A further study of 5-nitroacetylsalicylic acid suggested that the Avrami equation best described the solid-state hydrolysis.' The difficulties with the Leeson and Mattocks analysis are t Acetylsalicylic acid. outlined in the paper by Wu-Huang and Brooke." Unfor-tunately, this paper re-emphasises some of the problems out- lined above, for example by suggesting that the Avrami-Erofeyev equation is 'empirical '.The aim of the present work is clarify the kinetics and to consider a wider range of possible solid-state mechanisms for the hydrolysis. Experimental Sample The starting material was supplied by Monsanto (Asagran 7016) and consisted of single crystals some 2-5 mm in length and ca. 0.1 mm in diameter. Kinetic Study Runs were carried out in small sample tubes containing aspirin, which were heated in an aluminium block, under nitrogen, at constant temperature. The tubes containing the resulting mixtures of aspirin and salicylic acid were then treated on a vacuum frame at room temperature to remove salicylic acid, and the residual aspirin determined by reaction with sodium hydroxide and back-titration of the excess alkali with hydrochloric acid' (using 0.05 mol dm- solutions).Preliminary experiments were carried out to check the accu- racy and reproducibility of the extent of decomposition and the lack of volatility of aspirin under these conditions. The temperature range used was 368-387 K. The humidity was controlled by passing nitrogen (99.9%) through ther- mostatted water bubblers containing glass beads, modified from the design in ref. 16, before passing over the sample. Heating tapes were used between the bubbler and the block to prevent condensation of water vapour. The thermostat temperatures were always lower than the sample tem-peratures to produce relative humidities < 100%. The humid- ity range used was 7.0-24 kN m-' water vapour pressure." Sample weights were 20 mg (k10%) throughout.Water Sorption Two methods were used to determine the uptake of water on the aspirin sample. These were both gravimetric and involved, first, the use of 1 g samples placed in desiccators containing saturated salt solutions,' until constant weight was reached. The range of relative humidities covered was 0.1-0.86. The second method used a thermobalance; samples weighing 20 mg were exposed at 303 K to nitrogen gas humidified by passing through the same saturated salt solu- 998 tions. The weight increases were monitored continuously. The equilibration times using this method were short, and the results were very similar indeed to those obtained by the use of desiccators.Results Reaction Mechanisms The hydrolysis reaction proceeded in a single stage, as mea- sured by direct aspirin determination, producing S-shaped curves. The reaction mechanism was decided on by a com- parison of reduced-time plots (based on the time for half- reaction, to.5)for experimental results and model data for nucleation, phase-boundary and diffusion control of the reac- tion.” This method is very useful for direct comparison of experimental data. Typical reduced-time plots are given in Fig. 1, which includes data from Leeson and Mattocks’ paper’ and other model equations. It can be seen that the best fit between experiment and model data is for the Avrami-Erofeyev equation : -ln(l -a) = kt” where a is the proportion decomposed and n lies between 2 and 3.Because the value of n varies in the above equation, it cannot be used to derive consistent rate constants. The reciprocal of the half-time, l/to.5, for each experimental run was used throughout. This allowed comparable values to be used irrespective of the actual value of n. Effect of Temperature on Rate The effect of temperature is shown in Fig. 2, as an Arrhenius plot, from which values of activation energies and pre- exponential factors were derived; these are combined in Table 1. The error in the degree of conversion was deter- mined from the preliminary analysis of three lots of triplicate samples at low, intermediate and high degrees of conversion. This error was ca.9%, and this has been incorporated into Fig. 2 and Table 1. 1.00 -3 0.80 -E“8 0.60 -alU C .-t: 0.40 -P z d 0.20 -0.00 I I I 1 0 0.5 1.0 1.5 2.0 2.5 3.0 reduced time, f/ro.5 Fig. 1 Reduced-time plots for the hydrolysis of aspirin: (a) -ln(l -a) = kt”, n = 3, Avrami-Erofeyev equation; (b) -ln(l -a)= kt”, n = 2, Avrami-Erofeyev equation; (c) 1 -(1 -a)”’ = kt, contracting disc equation. (0)368 K, 24.0 kN m-’;@) 387 K, 7.9 kN m-2; (0)378 K, 13.6 kN m-2; (A) Leeson and Mattocks, 333 K, 16 kN m-’. J. CHEM. SOC. FARADAY TRANS., 1994, VOL. 90 -13.0 -r -12.5 -c Iv) 3 ac2 -12.0 -0 al w z W ,C -11.5 -~ -ll.O! ,I I t I I I I I I I I I I I I, 0.0025 0.0026 0.0027 0.0028 reciprocal temperature/K- ’ Fig.2 Arrhenius plot for the hydrolysis of aspirin. Water vapour pressure/kN m-2: (a)24.0, (b)13.6, (c) 7.9. Table 1 Arrhenius parameters for the hydrolysis of aspirin vapour pressure E, correlation /kN m-’ /kJ mol-’ log(A/s -1) coefficient 7.9 66.5 f6.0 4.11 k0.08 -0.99 13.6 73.0 f6.6 4.98 & 0.08 -0.99 24.0 84.7 7.6 6.52 +_ 0.08 -0.99 Effect of Water Vapour Pressure on Rate The effect of water vapour pressure on the hydrolysis rate constants is shown in Fig. 3. The rate decreases as the water vapour pressure rises, although the effect is not great. Water Sorption The results of the desiccator experiment are shown in Fig. 4. The isotherm shape is best described as type 4 in Brunauer’s classification.’’ A BET calculation gave a surface area of 0.17 m2 g-’ with cBET = 20. The external geometrical area was calculated to be about 0.1 m2g-’.“1 T (DI 121 0 T Vl I I I I 5.0 10.0 15.0 20.0 25.0 vapour pressure/kN m-* Fig. 3 Effect of water vapour pressure on hydrolysis rate. Reaction temperatures/K: (a) 368; (b) 378; (c) 387. The figure against each point is the relative humidity. J. CHEM. SOC. FARADAY TRANS., 1994, VOL. 90 0.8 l'O 1 0 20 40 60 80 100 relative humidity (76) Fig. 4 Water sorption isotherm for aspirin Microscopy A scanning electron micrograph (SEM) of material reacted at 330 K and 9.5 kN m-2 (relative humidity = 0.55) is shown in Fig. 5. This shows a series of oriented triangular etch pits which vary slightly in size.Optical microscopy indicated the presence of needles of salicylic acid which disappeared under the vacuum conditions of coating with metal and exami- nation in the scanning electron microscope. Discussion Water Sorption The sorption isotherm (Fig. 4) shows typical type 4 behaviour with a knee indicating monolayer coverage and a plateau indicating coverage of the external surface. This behaviour suggests that intergranular sorption is occurring at high rela- tive humidity. The agreement between the geometric surface area (ca. 0.1 m2 g-') and that calculated from the isotherm (0.17 m2 g- ') is reasonable, and suggests that little porosity is present. The knee occurs at about 0.4% weight gain and a relative pressure of 0.3, and the second plateau at about 1.0% gain at a relative pressure of 0.8, suggesting that approx- imately two equivalent layers of water are sorbed at the highest partial pressures.Approximate monolayer behaviour continues up to a rela- tive pressure of 0.6. The constant, c, indicates relatively strong interaction between the first layer of water and the crystal surface. Water commonly exhibits type 3 behaviour with low c.~' Fig. 5 Scanning electron micrograph of etch pits in aspirin. Bar = 25 pm. Reaction Mechanism The reduced-time plots (Fig. 1) follow the Avrami-Erofeyev equation, which suggests that nucleation-and-growth pheno- mena are rate controlling. The obedience to a given equation does not, of course, confirm a mechanism.The value of n in the Avrami-Erofeyev equation is compos- ite,22n = y + 6, where y is the time dependence of nucleation, and 6 the dimensions of growth of the nuclei. Various com- binations of y and 6 can arise and microscopy is an impor- tant tool for differentiating between the possibilities. Values of n are commonly low for reactions of organic compounds. It is clear from Fig. 5 that such nucleation, followed by growth, has taken place. The nuclei are three-dimensional, but they are shallow, indicating that the growth rate is very slow normal to the surface, and could be described as two- dimensional. Examination of Fig. 5 indicates that the etch pits vary slightly in size. This suggests that not all of them are produced at the same time, i.e. at the start of the reaction.This further suggests that y should have a value close to unity, i.e. dN/dt =f(t), where N is the number of nuclei. The best simple interpretation of n = 2-3, therefore, is that the number of nuclei varies slightly with time (y = 1) and that each is growing two-dimensionally (6 = 2). The observation of etch pits not only leads to an explana- tion of the S-shaped curves, but also provides more evidence against the Leeson and Mattocks' mechanism, in which the S-shaped curves were accounted forg on the basis of a through-solution mechanism, with the dissolution of the solid not being rate controlling. A nucleation-and-growth mecha- nism would suggest a non-uniform attack on the solid which may, in reasonable circumstances, leave visible traces, as observed.Kinetics The overall rate of hydrolysis decreases slightly with increase in water vapour pressure. This is a relatively unusual obser- vation and suggests that a through-solution mechanism is not operating, since increasing vapour pressure leads to increased sorption, as shown in Fig. 4, and therefore an increasing layer of water at the surface, which should produce faster hydrolysis. More insight can be gained from the Arrhenius parameters given in Table 1, rather than the rate constants. These show that both the activation energy and the pre-exponential factor increase with increasing vapour pressure, but that the latter increases more slowly than the former. These terms oppose each other in the Arrhenius equation, and the result is that the rate decreases with rising vapour pressure.The increase in activation energy suggests that different processes are occurring at higher partial pressures of water vapour. There are difficulties in assigning possible reaction steps to these energies because of the appearance of a compensation effect (see below). The variation of the activation energy with water vapour pressure is given in Fig. 6, which shows a linear relationship. The value of the activation energy extrapolated to zero pres- sure is about 57 kJ mol-', which is slightly lower than the activation energy of the homogeneous reactions and that obtained by Leeson and mattock^,^ who corrected their rate data for the effect of vapour pressure. Compensation Effect The relationship between activation energy (EJ, in kJ mol-', and pre-exponential factor (A) is given in Fig.7, showing what has been described as compensation behaviour, which is I I I 1 5 10 15 20 25 vapour pressure/kN m-2 Fig. 6 Effect of water vapour pressure on activation energy commonly expressed in linear form : log A = B + eE, where B and e are constants such that B = -4.68 and e = 0.133 in this case. Such behaviour has been discussed in both homogeneous and heterogeneous systems, and a set of possible explanations have been proposed23 as follows: (a) there is a characteristic temperature of onset of reaction; (b) the surface is energeti- cally heterogeneous; (c) there is more than a single active surface; (d) an interrelationship exists between reaction entropy and enthalpy. The last of these is the basis of the usual evidence for compensation behaviour, but we feel that the second is important in this case, in that it links the kinetic behaviour with the water sorption results.The compensation behaviour is occurring as an increasing amount of water is being sorbed by the sample up to a notional monolayer at PIP, = 0.30 (the highest relative pressure used in the kinetic study is 0.32). If the sample is taken to contain reactive sites (for example, emergent dislocations) with a range of energies [cJ: (b) above] then it can be assumed that these will sorb water (and therefore react) in proportion to that excess energy, with the highest energy/most reactive site sorbing water at the lowest vapour pressures, and requiring a low activation energy for reaction.Less reactive sites will sorb water at higher vapour pressures and will require more energy input for reaction to occur. These effects will manifest 1 60.0 I I I 4.O 5.0 6.0 7.O In(pre-exponential factor/s- ') Fig. 7 Compensation plot for aspirin hydrolysis J. CHEM. SOC. FARADAY TRANS., 1994, VOL. 90 themselves as a lower rate with increasing pressure at con- st ant temperature. Comparison with Previous Work The appearance of S-shaped curves for the decomposition, as shown by other workers,'-12 is confirmed. In the earliest of these studies, Leeson and Mattocks suggested a through-solution mechanism, governed by the limited solubility of aspirin in water, but modified by pH changes caused by dis- solution of the product acid.The solution hydrolysis was assumed to be rate-controlling and catalysed by dissolved salicylic acid, i.e. that any process involving the solid (dissolution, for example) was fast. This solution mechanism was effectively disproved by Hasagawa et a1.l' who showed that both normally hydrolysing material, and samples which had been subjected to vacuum treatment (to remove free acid), followed the same decomposition curve, indicating that the rate was independent of the amount of free acid. The present work indicates an alternative explanation, involving the solid in the rate-controlling step.The observation of etch- pits is important in this explanation. The major differences lie in the materials used: Leeson and Mattocks' and Hasegawa et a/." used aspirin powder, while the present work used single-crystal material. Single crystals are more likely to show nucleation-and-growth kinetics than powder^,'^ and it is interesting that the study of single crystals of 5-nitroacetylsalicylic acid also showed Avrami-Erofeyev behav- iour.12 It is interesting to speculate that the production of the powder has limited the range of active sites on the surface, and that those remaining are the most active, as shown by a comparison of activation energy. The suggestion that poten- tial reaction sites (germ nuclei) are present at the surface can also account for the differences in the solid-state behaviour of aspirin after treatment with soaps and other material^.'*'^*^^ These nuclei are probably the sites of emergent dislocations (the crystallography of these is being studied further), and they are certainly reactive, and so would be obvious sites for modification by surface active agents, thus giving rise to a changed hydrolysis mechanism (to contracting disc kinetics) after treatment, as has been 0bser~ed.I~ References 1 S.R. Byrn, J. Pharm. Sci., 1976,65, 1. 2 J. T. Carstensen, J. Pharm. Sci., 1974,63, 1. 3 K. A. Connors, G. L. Amidon and V. J. Stella, Chemical Stability of Pharmaceuticals, Wiley, New York, 1986, p. 115. 4 N. J. Carr and A.K. Galwey, Thermochim. Acta, 1984,79,323. 5 L. J. Edwards, Trans. Faraday SOC., 1950,46,723. 6 C. A. Kelly, J. Pharm. Sci., 1970,59, 1053. 7 K. C. James, J. Pharm. Pharmacol., 1958,10,363. 8 S. S. Kornblum and M. A. Zoglio,J. Pharm. Sci., 1967,!56,1569. 9 L. J. Leeson and A. M. Mattocks, J. Am. Pharm. Ass., 1958, 47, 329. 10 J. Hasegawa, M. Hanano and S. Awazu, Chem.Pharm. Bull., 1975,23, 86. 11 Y.Wu-Huang and D. Brooke, Int. J. Pharm., 1982,11,271. 12 M. Okamura, M. Hanano and S. Awazu, Chem. Pharm. Bull., 1980,28,578. 13 P. V. Mroso, A. Li Wan Po and W. J. Irwin, J. Pharm. Sci., 1982,71, 1096. 14 E. Nelson, D. Eppich and J. T. Carstensen, J. Pharm. Sci., 1974, 63, 755. 15 British Pharmacopoeia, The Pharmaceutical Press, London, 1963, p. 18. 16 R. E. Dodd and P. L. Robinson, Experimental Inorganic Chem- istry, Elsevier, London, 1954, p. 140. 17 R. C. Weast, Handbook of Chemistry and Physics, CRC Press, Boca Raton, FL, 61st edn., 1980, p. D 196. J. CHEM. SOC. FARADAY TRANS., 1994, VOL. 90 1001 18 J. F. Young, J. Appl. Chem., 1967,17, 241. Chemical Kinetics, ed. C. H. Brennan and C. F. H. Tipper, Else-19 J. H. Sharp, G. W. Brindley and B. N. N. Achar, J. Am. Ceram. vier, Amsterdam, 1980, vol. 22, p. 41. SOC.,1966,49, 379. 23 A. K. Galwey, Adv. Catal., 1977,26, 247. 20 S. Brunauer, The Adsorption of Gases and Vapours, Oxford Uni- 24 Ref. 22, p. 58. versity Press, London, 1943. 21 J. Hagymassy, S. Brunauer and R. Sh. Milhail, J. Colloid Inter- face Sci., 1969,29,485. Paper 3/06805F; Received 15th November, 1993 22 A. K. Galwey, in Reactions in the Solid State, Comprehensive
ISSN:0956-5000
DOI:10.1039/FT9949000997
出版商:RSC
年代:1994
数据来源: RSC
|
15. |
Morphology and polymorphism in molecular crystals: terephthalic acid |
|
Journal of the Chemical Society, Faraday Transactions,
Volume 90,
Issue 7,
1994,
Page 1003-1009
Roger J. Davey,
Preview
|
PDF (998KB)
|
|
摘要:
J. CHEM. SOC. FARADAY TRANS., 1994, 90(7), 1003-1009 Morphology and Polymorphism in Molecular Crystals :Terephthalic Acid Roger J. Davey and Steven J. Maginn ZENECA , R& T Dept, The Heath, Runcorn, Cheshire, UK Steven J. Andrews, Simon N. Black, Anne M. Buckley, Denis Cottier, Phillip Dempsey, Rodney Plowman, Joanne E. Rout, David R. Stanley and Anne Taylor lCl Chemicals and Polymers, R&T Dept, The Heath, Runcorn, Cheshire, UK The phase characterisation and morphologies of forms I and II of terephthalic acid are presented. The structural nature of the II + I polymorphic phase transition has been explored and form II shown to be more stable at room temperature and pressure. A new mechanism is proposed for the apparent stability of multiply twinned samples of form I.Terephthalic acid, used in the production of polyester, is manufactured by the oxidation of p-xylene. In order to purify it from other reaction by-products and to isolate it in a solid form appropriate for subsequent processing, the crude product is recrystallised from water in a multi-stage process in which the solution temperature and pressure are gradually reduced from values in excess of 250°C and 50 bar to atmo- spheric values. These extreme conditions are dictated by the solubility of terephthalic acid (TA) in water and ultimately yield material of purity greater than 99.9%. In terms of its solid-state chemistry TA is of some interest. Its crystal structure was first determined by Bailey and Brown' who found two polymorphic structures (referred to here as I and I1 but sometimes designated ct and /3) both with triclinic unit cells, space group Pi.The unit-cell data are given in Table 1. As discussed by Bailey and Brown' and later by Berkovitch-Yellin and Leiserowitz,* in both struc- tures molecules are linked through hydrogen-bonded carbox- yl dimers into infinite chains. These chains are held in two-dimensional sheets by Van der Waals and C-H..*O interactions in which the carbon to oxygen separations of 3.79 and 3.48 A in forms I and 11, respectively, lie in the expected range.3 The relative positions of adjacent H-bonded chains in each layer differ somewhat between the forms, as does the offset between molecules in adjacent layers. In form I the chains are directed along [llO] with the sheets lying in the (001) plane such that the dimer motifs in one layer lie approximately above those of the adjacent ones.In form I1 the chains lie along [lo01 with the sheets in the (011) plane such that the carboxyl dimers of one chain lie over the phenyl rings of the adjacent layers. These differences are shown in Fig. 1, which shows views normal to the molecular plane for each structure. Neutron diffraction: solid-state NMR' and molecular mechanics calculations6 have explored the nature of proton disorder in form I of TA and shown that at room temperature the acid protons are almost fully disordered with respect to the carboxyl dimers. Significant work has been carried out on the purification of TA*'v8 however, compara- Table 1 Crystallographic data for TA forms I and I1 (after ref.1) form I form I1 alA 7.73 9.54 blA 6.443 5.34 clA 3.749 5.02 aldegrees Bldegreesyldegrees 92.75 109.15 95.95 86.95 134.6 104.90 Fig. 1 Crystal structure of TA. (a)form I, (b)form I1 tively few studies of the transformation between polymorphs I and I1 have been reported. Bailey and Brown' assumed that form I was the more stable because it had the higher density and because form I1 'tended to become rarer on storage'. This assumption was followed by Saska and Myersong who calculated the lattice energies of the two forms and performed powder diffraction on a range of samples. They failed to observe form I1 but concluded that it was an accessible struc- ture since its calculated lattice energy was only 1.0 kcal mol-' smaller than the -41.0 kcal mol-' of form I.Using Raman spectroscopy Gerasimov et al." were able to observe the phase transition between forms I and I1 and to identify that the transition temperature, 349 K, was sensitive to sample pressure. They concluded that the transition was reversible and that form I was more stable at atmospheric pressure. The current studies concentrate on two aspects of TA crystal chemistry, first its crystal morphology, and in particu- lar the morphogenesis of industrially prepared products, and secondly the structural nature of the polymorphic transform- ation between forms I and 11. Experimental Commercial samples of purified TA were supplied by ICI Chemicals and Polymers Ltd.Recrystallisations were per-formed by dissolution of this material in water and aqueous acid in an autoclave at 250°C and 650 psi.? Crystallisation was effected by cooling and crystals were recovered at room temperature and pressure. Both commercial and rec-rystallised materials were characterised by a number of tech- niques. Morphologies were studied using optical and scanning electron microscopies (SEM), and optical geometry. For industrial samples this included embedding in resin and sectioning with a microtome knife in order to examine the internal microstructure. Phase characterisation was carried out by a combination of X-ray powder diffraction, Weissen- berg and synchrotron radiation (SR) Laue single-crystal X-ray diffraction, and low-frequency Raman spectroscopy.It is worth commenting here that phase characterisation of samples by powder diffraction rather than single-crystal tech- niques is not reliable because of the similarities in the diffrac- tograms of the two forms. These are shown in Fig. 2, (calculated using the CERUIS molecular modelling software' ') where it is clear that phase assignment can only be achieved using minor reflections such as those in the 28 region around 30". It was our experience that limited resolution and preferred orientation effects increased the uncertainty and, although it was clear that no method of sample preparation gave pure phases, quantification of phase composition was not possible.This problem may be the cause of Saska and Myerson's inability to observe form 11.' Ultimately it was found that a combination of single-crystal Of--0 10 20 30 40 50 60 100 j ao -60-40 -jl-0 t 1 psi = 6894.76 Pa. J. CHEM. SOC. FARADAY TRANS., 1994, VOL. 90 XRD and Raman spectroscopy of both powders and single crystals provided the most reliable combination of character- isation techniques. Phase transitions were studied by hot- stage optical microscopy, differential scanning calorimetry (DSC), Raman spectroscopy and single-crystal Laue diffrac- tion. The latter employed station 9.7 of the synchrotron radi- ation source at Daresbury, UK, together with a special temperature-controlled furnace.Crystals (ca. 0.3 x 0.1 x 0.1 mm3) were held between kapton foils; expo- sure times were 3 s, a wavelength range of ca. 0.25-1.5 %i was used, the beam was attenuated using 0.18 mm A1 foil and the crystal to film distance was 64mm. Diffraction patterns were recorded on film every 5 "C in the range 80-150 "C and inter- preted using the programs SPOTIN, NEWLAUE and GENLAUE. l3 Lattice energy and morphological calcu-lations were performed using the program HABIT together with potential constants from Lifson et all4 This program utilises the Hartmann-Perdok approach to morphology pre- diction, calculating attachment and slice energies for selected surfaces. Results and Discussion Morphological Predictions The predicted morphologies of forms I and I1 are shown in Fig.3. For form I1 the corrected H-atom fractional co- ordinates were taken from ref. 9. In both cases elongated prisms are predicted, for form I the elongation is in the [llO] direction, while in form I1 it is along [lOo], i.e. in both cases in the direction of the H-bonded chains. Lattice energies of -33.04 kcal mol-' and -31.19 kcal mol-' were obtained for forms I and 11, respectively. These values compare well with the measured heat of sublimation, 33.3 kcal mol-l lS and suggest that form I will be more stable. This is in contra- diction to calculations of Berkovitch-Yellin and Leiserowitz2 who calculated the reverse order of stability with lattice ener- gies of -37.2 and -36.8 kcal mol-' for forms I1 and I, respectively.Experimental Morphologies Experimentally the morphologies of the two polymorphs have not been well characterised. Bailey and Brown' refer to two habits, needles and plates, for form I and equi-dimensional parallelepipeds for form 11. They defined the needle axis of form I as [Ool], with many crystals twinned. (a1 001) . (20i) (100) Fig. 3 Predicted morphologies of TA. (a)form I, (b)form I1 Plate 1 Weissenberg photograph of a twinned form I crystal Plate 2 Single-crystal Laue diffractograms of form I1 at (a) 30 and (b) 150"C Davey et al. (Facingp. 1005) J. CHEM. SOC. FARADAY TRANS., 1994, VOL. 90 Saska and Myerson reported non-faceted commercial materials and demonstrated a change to a faceted morphol- ogy on ageing.16 Ozaki and Shigeya~u".'~ reported more detailed studies of the commercial materials.They concluded that these ovoid particles were aggregates of form I crystals, ordered such that their c axes were approximately coincident (f16) with the long axis of the ovoid and their short axis with [1001. Form I: Appearance and Morphogenesis Form I crystals were only observed in this study when p-toluic acid was included as an additive and, in agreement with Bailey and Brown, these crystals were always found to be twinned parallel to their needle, c, axes. In preparations using pure TA, form I1 was always found to be the major phase. Commercial material, on the other hand was always found to be form I and is shown in Fig. 4.Fig. 4(a)is an SEM view of the exterior of a typical ovoid with Fig. 4(b) and (c) showing sections at increasing magnification. These data demonstrate the polycrystalline nature of these particles and show how the c-axis fibres of TA are highly oriented within domains. This is consistent with Ozaki and Shigeyasu's con- clusion that these particles comprise ordered arrays of form I crystals. However, these observations and many others made during the course of this work failed to find any consistent geometric relationship between the internal fibrular micro- structure and the overall particle morphology as suggested by Ozaki and Shigeyasu.'* Overall these observations concern- ing Form I lead to the conclusion that the observed micro- structure of commercial products is a direct result of the tendency of form I to twin during growth.Thus it may be imagined that during the industrial purification process form I crystals nucleate and grow at high temperatures and super- saturations to yield bundles of highly twinned needles which, being susceptible to mechanical abrasion parallel to the twin axis, eventually lose their faceted morphology and become rounded in appearance. Weissenberg photographs such as that shown in Plate 1 demonstrated that within a twinned crystal the common directions are [110] and [Ool), (i.e. one of these is the same in both twins and the other exactly reversed). Because these directions are separated by an angle of 77.83", the two lattices are incommensurate across the twin plane.The observation that twinning is promoted by p-toluic acid suggests that the surface termination of hydrogen-bonded chains may be an important feature of the twinning process, perhaps because it promotes alternative non-cyclic H bonds across the twin plane, in response to the need to over- come the steric effect of the incorporated p-toluic acid mol- ecules. This is consistent with the observed microstructure of industrial products and results in the preservation of the chain direction across the twin boundary together with a twist of 37.06 in the molecular planes as indicated in Fig. 5. It is worth noting that in the crystallisation of adipic acid, addi- tion of caproic acid is also known to promote twinning.lg Form 11: Morphology Fig.6 shows a typical selection of crystals recovered from a hydrothermal recrystallisation experiment carried out in this study. Weissenberg photographs showed all such crystals to be form I1 although morphologically they exist both as hex- agonal, c-axis needles and rhombs. Optical gonimetry of a needle showed it to be bounded by (loo), (010) and (110) forms. In rhombic crystals the (100) face dominates the mar-phology with the longest dimension being along [Ool]. These two habits possibly represent populations of crystals nucle- ated at different times during cooling. It is interesting to note that for form I1 as for form I pre- ferred growth does not take place along the H-bonded chains ' 30pm . Fig.4 Commercially produced TA. (a) Exterior view, (b) internal section, (c) internal section showing details of Crystalline array. as predicted, but rather in the direction normal to these. This may be the result of the solvent, water, binding strongly to these surfaces and suppressing growth. Similar effects have been found in both succinic2' and adipic acid.lg Fig. 5 Diagrammatic representation showing the relative orienta- tion of form I twinned crystals Polymorphic Transformation Having investigated the conditions under which the two forms of TA may be prepared, further studies were made in order to examine the interconversion between the forms. Hot-stage Microscopy Initial investigations were made by heating crystals of both forms and observing any gross changes in their opacity or morphology.Crystals and ovoids of form I showed no changes on heating to 200°C. Form 11 crystals on the other hand behaved in a dramatic way. First, it was found that over a range of temperatures between 75 and 100°C these crystals underwent a spontaneous shape change from rhombic to rectangular. This is shown in Fig. 7-at various stages in during heating. Fig. 7(a) shows a form I1 crystal at room temperature. Fig. 7(b)was taken at 90°C and Fig. 7(c) at 95 "C.The morphological change is seen to proceed in two steps, with half the crystal transformed in (b) and the trans- formation complete in (c). Secondly, for many crystals this change was associated with the release of sufficient mechani- cal energy to cause the crystals to 'jump' off the microscope stage.The morphological change was found to be reversible on cooling to about 30°C although it was observed that many crystals were able to retain the high-temperature mor- phology, only reverting to rhombs upon the application of slight stress using a metal point. Fig. 6 Hydrothermally grown form I1 crystals J. CHEM. SOC. FARADAY TRANS., 1994, VOL. 90 Fig. 7 Morphological changes occurring during phase transform- ation: form I1 crystal at (a)room temperature, (b)90 "C,(c)95 "C Thermal Analysis DSC of bulk samples of form I1 showed a weak endotherm over the temperature range 80-110°C with a related exo-therm on cooling. Fig. 8 shows an example of the endotherm and also the effect of thermal cycling which progressively shifts the peak maximum to higher temperatures.Measured values of the endotherm lay in the range 0.07-0.2 kcal mol- '. This variability is most probably due to the presence of vari- able amounts of form I in the samples used. As discussed J. CHEM. SOC. FARADAY TRANS., 1994, VOL. 90 2.0-1.6-80 90 100 110 120 TrC Fig. 8 DSC traces for a form I1 sample of TA above it was not possible to provide quantitative measure- ment of phase composition. Samples of form I showed no evidence of phase changes on heating as observed by thermal analysis. These data lead to the tentative conclusion that at atmo- spheric pressure form I1 is the stable form converting to a second form on heating.In order to characterise the structur- al nature of the change further studies were carried out using single-crystal XRD and Raman spectroscopy. Raman Spectroscopy Fig. 9 shows the Raman spectra measured on single crystals of forms I and 11. Powder samples of these forms give identi- cal spectra and they are consistent with those observed pre- viously by Gerasimov et al." who assigned the three lowest bands to librational vibrations and the 132-133 cm-' band to intramolecular vibrations of the carboxyl group relative to the phenyl ring. It is important to note, however, that our phase assignment is the reverse of theirs but based on crystals that have been independently characterised by single-crystal XRD. It seems likely that they assigned the high-temperature spectra as form I1 on the basis of Bailey and Brown's unsub; stantiated conclusion that form I was stable at room tem- perature.Using a thermostatted stage it was possible to heat single crystals of form I1 and simultaneously record their morphologies and Raman spectra. Fig. 10 shows a typical example of a crystal heated to 95°C. The morphologcal change is accompanied by a change in the Raman spectrum ?, I' I' I I I\ 70 90 110 130 wavenumber/cm-Fig. 9 Raman spectra of TA. (a) form I and (b) form I1 ' 50pm ' 7-70 90 110 130 wavenumber/cm-' Fig. 10 Change in Raman spectra and morphology observed on heating a form I1 crystal to 95 "C: (a)at room temperature and (b) at 95 "C from that of form I1 to that of form I.On cooling the sample the form I1 spectrum reappeared. It is thus clear that form I1 is the stable phase at atmospheric pressure and room tem- perature, converting to form I on heating. On the basis of the changes in the spectra with temperature tentative conclusions can be drawn concerning the structural changes involved. At the transition temperature it is evident that the intramolecu- lar band retains its position whilst the lattice bands drop to lower frequencies. According to Gerasimov et al.," the tem- perature dependences of these bands show no discontinuities as the transition temperature is traversed. The 82 cm-band, however, has a temperature coefficient about seven times greater than the other bands. Following the work of de Ville- pin et aL21 on the transition in malonic acid it seems likely that this lattice band relates to a twisting of the dimer ring and libration in the direction of the H-bonded chains.The monotonic temperature dependence suggests conservation of the chain direction. Single-crystal Laue Diflraction In this technique, diffraction patterns were recorded over a range of temperatures until a change in the pattern revealed that a structural change had taken place. All the crystals examined exhibited highly streaked patterns due to large mosaic spreads. This gave difficulty in data processing but ultimately two crystals were chosen which gave data of suffi- cient quality. Plate 2 shows the low-(3OoC) and high-(150"C) temperature patterns from one of these crystals.The method-ology of data interpretation is given elsewhere13 but Fig. 11 shows the predicted patterns that best match the Fig. 12 pho-tographs. These confirm the Raman data that the room-temperature form I1 converts to form I on heating, and indicate that form I crystals generated in this way are untwinned. Table 2 gives details of the data fit and the rela- tive orientation of the crystal. This shows that both before and after heating the direction of the c axis is virtually unchanged (-179.206 cf 179.838), implying that during the transformation the c-axis direction is preserved. Hence, from the known morphology of form I1 crystals (see above), Fig. 12 has been drawn illustrating the relationship between the observed morphological change and the crystal structures of the polymorphs.The change is such that, in agreement with the Raman spectra, the H-bonded chains are preserved but the molecular sheets are not. Taken together the data suggest that this transformation is Martensitic in nature; no phase boundaries are present in transforming crystals and its extent is a function only of temperature. The stresses built up when the molecular chains slide to their new positions and rotate to reform the sheets are rel- eased as mechanical energy. (a1 J. CHEM. SOC. FARADAY TRANS., 1994, VOL. 90 Table 2 SR Laue phase characterisation temperature of crystal]"C form showing best fit orientation angles/degrees 30 11, rms = 0.107 nm -117.14, 28.62, 150 for 118 spots I, rms = 0.158 nm -179.21 14.28, 7.25, for 88 spots 179.84 Thermodynamic Considerations The observations reported here are consistent with a first- order phase transition.The measured endotherm of 0.2 kcal mol-' is in reasonable agreement with the calculated lattice energy difference of Berkovitch-Yellin and Lieserowitz' of 0.4 kcal mol-' and suggests that the transition is driven by the increased disorder associated with the form I structure. This is consistent with the kn~wn~-~ proton disorder in the car- boxyI dimers. Assuming that the form 11 structure is ordered this amounts to an entropy increase of R In 2 which at 100°C contributes -0.5 kcal mol-I to the Gibbs energy change and drives the transition.A similar phenomenon exists in 4-formylbenzoic acidz2 also related to proton dis- order. Using the Clapeyron equation the effect of pressure on the transformation temperature can be predicted. Taking the cal- culated densities, the volume change associated with the tran- sition I1 -+I is -3.81 cm3 mol-' (shrinkage). This implies that as the pressure increases so the transition temperature decreases. Taking the transition temperature as 85 "Cand the enthalpy change as 0.2 kcal mol-' then at a pressure of 34 bar the transition temperature decreases to 79 "C. Conclusions This study has considerably advanced the understanding of TA crystal chemistry. Form 11 is now understood to be the more stable polymorph at ambient temperatures and pres- sures.Its transformation to form I occurs over the tem- perature range 75-100 "Cand is driven by the contribution of proton disorder to the Gibbs energy change. The tendency of form I crystals to grow as twins parallel to their c axes explains the observed microstructure of commercial products. In contrast to single crystal samples, these ordered arrays of microcrystals are able, unexpectedly, to remain in the form I structure indefinitely under conditions where transformation to form I1 would be expected. This is presumably a form of collective stabilisation which is possible since the morpho- logical change and release of stress observed as an inevitable form 11 form 1 Fit. 12 Relationship between morphological and structural changes Fig.I1 Predicted Laue diffractograms using the data of Table 2 during the 11 +1 transformation in TA J. CHEM. SOC. FARADAY TRANS., 1994, VOL. 90 1009 consequence of phase transformation in single crystals is impossible in these twinned arrays since it would require a concerted inter-crystal process. The precise nature of this concerted change can be envisaged from examination of Fig. 5 and 12 where it is clear that a twinned form I crystal can only revert to the form I1 structure if both twins simulta- neously change from a rectangular to a rhombic morphology. 9 10 11 12 13 M. Saska and A. S. Myerson, Cryst. Res. Technol., 1985,20,201. V. F. Gerasimov, N. K. Zharikov, V. S. Korobkov, I. V. Ovchin-nikov and L.V. Rud’, Zh. Prikl. Spektrosk., 1981, 34, 308. CERIUS molecular modelling software, Molecular Simulations Ltd, St John’s Innovation Centre, Cambridge, UK. H. L. Bhat, S. M. Clarke, A. El Korashy and K. J. Roberts, J. Appl. Crystallogr., 1990, 23, 545. J. R. Helliwell, J. Habash, D. W. J. Cruickshank, M. M. The probability of this taking place in an ordered, multiply twinned sample such as shown in Fig. 4 is extremely low and hence form I is stabilised. 14 Harding, T. J. Greenhough, J. W. Campbell, I. J. Clifton, M. Elder, P. A. Machin, M. 2. Papiz and S. Zurek, J. Appl. Crystal- logr., 1989,22,483. R. Docherty and K. J. Roberts, J. Cryst. Growth, 1988,88, 159. 15 C. A. Lucchesi and W. T. Lewis, J. Chem. Eng. Data, 1968, 13, 389. References 16 M.Saska and A. S. Myerson, AIChE J., 1987,33,848. 1 M. Bailey and C. J. Brown, Acta Crystallogr., 1967, 22, 387 and 1984, C40,1762. 2 Z. Berkovitch-Yellin and L. Lieserowitz, J. Am. Chem. SOC., 1982,104,4052. 3 G. R. Desiraju, Material Science Monographs, 54, Crystal Engin- eering, Elsevier, Amsterdam, 1989. 4 P. Fischer, P. Zolliker, B. H. Meier, R. R. Ernst, A. W. Hewat, J. D. Jorgensen and F. J. Rotella, J. Solid State Chem., 1986, 61, 109. 5 B. H. Meier and R. R. Ernst, J. Solid State Chem., 1986,61, 126. 6 S. N. Black and D. Pullen, J. Solid State Chem., 1989,79, 293. 7 S. Gaines and A. S. Myerson, AIChE Symp. Ser., 1982, 78,42. 17 18 19 20 21 22 T. Ozaki and M. Shigeyasu, Maruzen Sekiyu Giho, 1974,19,88. T. Ozaki and M. Shigeyasu, Maruzen Sekiyu Giho, 1975,20,81. R. J. Davey, S. N. Black, D. Logan, S. J. Maginn, J. E. Fair- brother and D. J. W. Grant, J. Chem. Soc., Faraday Trans., 1992,88,346 1. R. J. Davey, J. W. Mullin and M. J. L. Whiting, J. Cryst. Growth, 1982,58,384. J. de Villipin, M. H. Limage, A. Novak, N. Toupry, M. Le Pos- tollec, H. Poulet, S. Ganguly and C. N. R. Rao, J. Raman Spec- trosc., 1984, 15,41. M. Haisa and S. Kashino, Acta Crystallogr., Sect. B, 1976, 32, 857. 8 P. M. Brown, M. Marquering and A. S. Myerson, Ind. Eng. Chem. Rex, 1990,29, 2089. Paper 3/06151E; Received 14th October, 1993
ISSN:0956-5000
DOI:10.1039/FT9949001003
出版商:RSC
年代:1994
数据来源: RSC
|
16. |
Enhancement in the optical CO sensitivity of NiO film by the deposition of ultrafine gold particles |
|
Journal of the Chemical Society, Faraday Transactions,
Volume 90,
Issue 7,
1994,
Page 1011-1013
Masanori Ando,
Preview
|
PDF (384KB)
|
|
摘要:
J. CHEM. SOC. FARADAY TRANS., 1994, 90(7), 1011-1013 1011 Enhancement in the Optical CO Sensitivity of NiO Film by the Deposition of Ultrafine Gold Particles Masanori Ando*t Light and Material Group, PRESTO, Research Development Corporation of Japan (JRDC), Midorigaoka I, lkeda, Osaka 563, Japan Tetsuhiko Kobayashi and Masatake Haruta Osaka National Research Institute, AIST, Midorigaoka 1,lkeda, Osaka 563, Japan In the presence of reducing gases thin NiO films show reversible changes for visible-near-IR light absorption, a phenomenon which can be applied to the optical detection of CO in air. The deposition of highly dispersed Au on NiO films by a deposition-precipitation method significantly enhances the optical sensitivity and the response rate, allowing CO detection under a substantially lowered operating temperature.Optical gas sensors have recently attracted growing interest because of several advantages over conventional electrical gas sensors.' These are (1) resistivity to electromagnetic noise, (2) fire resistance and (3) the capability of remote control and information transfer through an optical fibre network. We have already reported an optical method to detect reducing gases in air, which directly utilizes the optical absorbance change in the thin films of transition metal oxides such as Mn,O,, Co,O, and NiO by CO and H, in air.2 The reversible change in the visible-near-IR absorbance of the films can be assigned to the change of electron or positive hole density during a catalytic reaction of anionically adsorbed oxygen on the surface.Among the three oxides listed above, NiO film shows the greatest optical change for CO in air. However, the drawbacks are that appreciable change in absorbance can be observed only at temperatures above 250 "C and that the sensitivity to low concentrations of CO is not sufficiently high. Therefore, for the application of NiO films to CO gas sensors, it is necessary to improve both the sensitivity and response rate at low temperatures. The deposition of gold on NiO has already been proven to be effective in enhancing the absorbance ~hange,~ due to an appreciable increase in the catalytic activity., Since an increase in the absolute amount of absorbance change (AA) by depositing gold is not necessarily accompanied by an increase in the absorbance change ratio (AA/Aair), which cor- responds to the sensitivity, the present study has been under- taken to find the optimum conditions for gold deposition which lead to an improved sensing performance of the NiO film.Experimental Gas-sensitive films consisting of small gold particles depos- ited on NiO were prepared in the following manner. A mixed solution of toluene and butan-1-01 containing nickel 2-ethylhexanoate with an Ni content of 3 wt.% was spin-coated at 2000 rpm on a glass-plate substrate (18 mm x 18 mm x 0.1 mm), dried at room temperature in air and pyrolysed at 380°C in air. The thickness of the NiO films prepared was estimated by ellipsometry (DHA-XA2, Mizojiri Optical Co., Ltd.) to be about 85 nm.Gold hydroxide was then deposited on the NiO film from a neutralized aqueous solution of HAuCl, (25 ml for an 18 mm x 18 mm NiO film) at 70°C. The sample was then washed, dried and calcined at 300 "C in air (deposition-precipitation method5). In order to t Also at: Osaka National Research Institute, AIST, Midorigaoka 1, Ikeda, Osaka 563, Japan. find the optimum deposition conditions for Au to obtain the largest absorbance change ratio (AAIA,,,) around 400-1000 nm, the concentration of HAuCl, and pH of the aqueous solution were varied from 0.2 to 4.0 mmol 1-' and from 5 to 7, respectively. A gold-free NiO film was also prepared directly on the glass substrate for comparison. The dispersion of gold deposited on NiO was observed for the Au/NiO films separated from the glass substrate by trans- mission electron microscopy (TEM), using an Hitachi H-9000 instrument operating at 300 kV, which enabled us to see Au particles with diameters >1 nm.Gold content in the film was also determined by X-ray fluorescence analysis. The absorption spectra of all films used in this study were obtained in a controlled atmosphere with transmitted visible- near-IR light (12 = 400-1700 nm) at 150-250 "C. The spec- trometer (UV-3100PC, Shimadzu Co., Ltd.) was operated with a resolution of <0.001 in absorbance, so that a 3% absorbance change ratio for a film with an absorbance of 0.033 can be reproducibly detected. Test gases were fresh air (atmospheric air without the addition of CO) and air contain- ing 1 vol.% CO.All gases were dried by passing through silica gel and molecular sieve columns at room temperature and a molecular sieve column at 0°C prior to use. Film samples were pretreated at the operating temperature in fresh air for at least 1 h before measurements were made. Results and Discussion Table 1 summarizes the optical CO sensing properties for the Au/NiO films obtained under various conditions of Au depo- sition and for the Au-free NiO film. The absorbance change (AA) due to reaction with CO gas was enhanced by an increase in the amount of Au deposited up to a certain surface coverage obtained at a concentration of 1.0 mmol 1-' HAuCl, and at pH 6. However, within the upper limit, an increment in AA did not necessarily contribute to an increase of AAIA,,, .This is because the increase in AA is accompanied by a larger increment of absorbance in the Au/NiO film in fresh air due to light absorbance by the Au particles. On the other hand, it was difficult to clarify the relationship between AA and the particle size of Au, because the particle diameter of Au in every sample ranged too widely (from several nm to 100-200 nm) to obtain reliable mean particle diameters. For the present purpose, to maximize AA/Aair, an Au concentra-tion of 0.5 mmol 1-' and a pH of 7 were selected as the optimum conditions among those tested. Fig. l(a) and (b)show the absorption spectra obtained at 200°C for an NiO film and an Au/NiO film prepared under optimum conditions (hereafter, abbreviated as opt-Au/NiO J.CHEM. SOC. FARADAY TRANS., 1994, VOL. 90 Table 1 Optical CO sensing properties of Au/NiO films and an NiO film Au deposition conditions ~~ ~ concentratjon/HAuCl, mmol I - PH operating temperaturePC Aair 103 AA~ AA/A,i, b- b- 200 0.0286 2.0 0.069 b- b- 250 0.0356 5.7 0.160 0.2 6.0 200 0.0387 6.3 0.162 0.2 7.0 200 0.0280 4.1 0.145 0.5 5.0 200 0.0928 8.1 0.087 0.5 6.1 200 0.0641 8.1 0.127 0.5 7.1 200 0.0288 5.1 0.175 0.5 7.1 250 0.03 14 8.6 0.273 1.o 5.0 250 0.1111 2.6 0.023 1.o 6.0 200 0.0840 9.4 0.112 1.o 7.0 200 0.06 14 6.9 0.113 2.0 6.0 250 0.0534 1.8 0.034 4.0 6.0 250 0.0753 0.9 0.013 4.0 7.1 250 0.0253 0.3 0.01 1 Absorbances at 900 nm.Change in absorbance due to introduction to 1 vol.% CO in air. Without Au. film) in fresh air and in air containing 1 vol.% CO.The differ- ence spectra shown in Fig. 2(a) and (b)were obtained by sub- tracting the absorption spectra in air containing CO from those in air. In the absorption spectrum of the opt-Au/NiO film [Fig. l(b)], dipolar plasmon excitation of small Au particles6 was observed with a maximum at a wavelength of around 600 nm. The atomic ratio of Au to Ni in the film was 0.11 :1. A broad absorption band due to NiO was also observed in the whole visible-near-IR region. As shown in Fig. l(a) and 2(a),CO reduced the absorbance of the Au-free NiO film over the whole visible-near-IR region.The decrease of the absorbance can be ascribed to a decrease of positive hole density in NiO due to a decrease in oxygen anion density on the surface of NiO during the cata- lytic oxidation of C0.2 The difference spectrum [Fig. 2(a)] indicates that the absorbance change (AA) monotonically 400 600 800 1000 1200 1400 1600 wavelength/nrn Fig. 1 Visible-near-IR light absorption spectra for NiO films without and with Au deposition. (a)NiO film in air (-) and in air containing 1 vol.% CO (---); (b)opt-Au/NiO film in air (-) and in air containing 1 vol.% CO (---); 200°C. varies with wavelength. The absorbance change ratio (AA/Aair) slightly depends on the wavelength [Fig.3(a)],exhibiting a broad maximum of about 6.9% for 1 vol.% CO at 900 nm. On the other hand, Fig. l(b) and 2(b)show that the absorp- tion spectrum change of the opt-Au/NiO film is much more sensitive to CO in air than that of the Au-free NiO film at the 8r-----lF-? 400 600 800 1000 1200 1400 1600 wavelength/nrn Fig. 2 Difference absorption spectra of an NiO film and an Au/NiO film in air with and without 1 vol.% CO. (a) NiO film; (b)opt-Au/ NiO film; 200°C. 0.20,, I ' I ' I ' I ' 1 ' I[ 0.16 .k 0.12 3 a 0.08 0.04 0.00400 600 800 1000 1200 1400 1600 wavelengthlnm Fig. 3 Absorbance change ratio of an NiO film and an Au/NiO film by 1 vol.% CO in air as a function of wavelength. (a) NiO film; (b) opt-Au/NiO film; 200 "C.J. CHEM. SOC. FARADAY TRANS., 1994, VOL. 90 same temperature. The absolute value of the decrease in absorbance (AA) is enhanced by about a factor of 2 in magni-tude. The change ratio of absorbance (AA/Aair)of the opt-Au/ NiO film for 1 vol.% CO at 900 nm reached 17.5% (Fig. 3), the value of which is more than 2.5 times larger than that of the Au-free NiO film. This enhancing effect is considered to be reflection of an increase in catalytic activity for CO oxida-tion by depositing Au.~Such enhancement in the absorbance change by depositing Au would be useful for the more sensi-tive detection of CO in air. The opt-Au/NiO film can also detect CO at concentrations much lower than 1 vol.%. For example, AA and AA/A,,, measured for 100 ppm CO in air were 0.0015 and 4.7%, respectively, at 200 "C, and 0.0027 and 10.5%, respectively, at 250 "C.In Fig. 2(b) and 3(b)an appreciable wavelength dependence of AA and AAIA,, can also be seen. AA and AA/A,,, in the plasmon band at around 600 nm were somewhat suppressed compared with those in other wavelength regions. For thin metal films deposited on the transparent metal oxide, the eva-nescent wave from the metal oxide was effectively absorbed by the surface plasmon of thin metal films at the interface.' This may be the most likely reason for the suppression of the AA and AAIA,, at near 600 nm.8 Another possibility is the absorbance change of Au ultrafine particles as a result of interactions between atmospheric gas and Au ultrafine par-ticles and/or NiO, as seen in other systems comprising ultra-fine metal particles and metal oxide^.^.^ Fig.4 shows the response curves of the absorbance change due to CO at 900 nm and at 200°C for the NiO film and the opt-Au/NiO film. In the whole temperature region investi-gated from 150 to 250°C, the opt-Au/NiO film showed a larger response rate than the Au-free NiO film in both the absorbance-decreasing step (by exposure to CO-containing air) and the absorbance-increasing step (by exposure to air free of CO). The difference in the response rate between the opt-Au/NiO film and the Au-free NiO film became greater with decreasing temperature. Fig. 5 shows the absorbance change ratio (AA/Aair) of the films as a function of operating temperature.In both the Au-free NiO film and the opt-Au/NiO film, the change ratio increased upon increasing the temperature from 150 to 250°C. In the case of the NiO film, no absorbance change was detected below 150°C. In contrast, the opt-Au/NiO film showed reversible changes in absorbance even at tem-air 0 h2 -1 I 0 -2 s-9 m -30 F time/min Fig. 4 Optical response of an NiO film and an Au/NiO film with atmosphere periodically changing from air to air containing 1 vol.% CO. (a)NiO film; (b) opt-Au/NiO film; 900 nm, 200 "C. 1013 I *I 0.20.2 --m 3 Q 140 160 180 200 220 240 260 TrC Fig. 5 Absorbance change ratio of an NiO film and an Au/NiO film by 1 vol.% CO in air as a function of operating temperature. (a)NiO film; (b)opt-Au/NiO film; 900 nm.peratures below 150°C. The temperature for a given value of the absorbance change ratio shown in the opt-Au/NiO film is about 50°C lower than that for the Au-free NiO film. This implies that the temperature for operation as an optical CO sensor can be lowered by 50°C when Au is deposited with high dispersion. Conclusions Deposition-precipitation of small Au particles on NiO film from an aqueous solution of HAuC1, at pH 7 was found to be effective in enhancing the sensitivity and response rate of the film for the optical detection of CO in air. The Au/NiO thin film can be used to detect CO at temperatures about 50°C lower than can the Au-free NiO thin film. Although an absorbance change ratio of 3% can be measured using a spectrometer, the Au/NiO film can detect CO in air over a wide concentration range from below 100 ppm to above 1 vol.% at temperatures as low as 200°C.In addition, as the Au/NiO film has an absorbance as small as 0.03 at 900 nm, it is possible to put a number of films into the light beam to control the dynamic range of transmitted light intensity. The optical CO sensitivity is assumed to reflect the change in oxygen anion density on the film surface induced by the adsorption and catalytic oxidation of CO. This suggests a potential applicability of the visible-near-IR light absorption measurement as a new technique to investigate the elemental processes of surface reactions over heterogeneous catalysts. References 1 K. Eguchi, in Gus Sensors, ed. G.Sberveglieri,Kluwer, Dordrecht, 1992, p. 307. 2 T. Kobayashi, M. Haruta, H. Sano and B. Delmon, Proc. 3rd Znt. Meeting on Chemical Sensors, Cleveland, 1990, p. 318. 3 T. Kobayashi, M. Haruta and M. Ando, Sensors Actuators B, 1993, 13/14, 545. 4 M. Haruta, N. Yamada, T. Kobayashi and S. Iijima, J. Catal., 1989,115,301. 5 S.Tsubota, M. Haruta, T. Kobayashi, A. Ueda and Y. Nakahara, in Preparation of Catalysts V, ed. G. Poncelet, P. A. Jacobs, P. Grange and B. Delmon, Elsevier, Amsterdam, 1991, p. 695. 6 J. Turkevich, Gold Bull., 1985, 18, 125. 7 K. Matsubara, S. Kawata and S. Minami, Appl. Spectrosc., 1988, 42, 1375. 8 M. Ando, T. Kobayashi and M. Haruta, Proc. Symp. Chem. Sens. I1 (183rd Meet. Electrochem. Soc., Proc. 93-7), Honolulu, 1993, p. 690. 9 A. Yanase, H. Komiyama and K. Tanaka, Jpn. J. Appl. Phys., 1988,27,L164. Paper 3/06392E; Received 26th October, 1993
ISSN:0956-5000
DOI:10.1039/FT9949001011
出版商:RSC
年代:1994
数据来源: RSC
|
17. |
Adsorption and decomposition of methanol on TiO2, SrTiO3and SrO |
|
Journal of the Chemical Society, Faraday Transactions,
Volume 90,
Issue 7,
1994,
Page 1015-1022
Nina Aas,
Preview
|
PDF (1094KB)
|
|
摘要:
J. CHEM. SOC. FARADAY TRANS., 1994, 90(7), 1015-1022 Adsorption and Decomposition of Methanol on TiO, SrTiO, and SrO Nina Aas, Thomas J. Pringle and Michael Bowkert Leverhulme Centre forInnovative Catalysis and Surface Science Research Centre, University of Liverpool, P.O. Box 147, Liverpool UKL69 3BX The adsorptive and reactive behaviour of TiO,, SrTiO, and SrO towards methanol have been investigated using temperature-programmed desorption and XPS. TiO, was found to be reduced during the adsorption/desorption process and showed predominantly CH, formation at low coverages. At higher coverages, alternative paths led also to the desorption of CH,OH and HCHO though CH, was the main product desorbed at ca. 500 K. The product pattern indicates that a methoxy is the major surface intermediate.CO and H, were evolved at ca. 600 K, possibly from the decomposition of a formate species, while H,O was desorbed between 300 and 700 K. On SrO the coincident evolution of CO and H, at 580 K, with H,O and CO, being minor products, indicated a formate intermediate as the dominant one on this surface. The product pattern on SrTiO, resembled that on TiO,, although the main product in this case was CH,OH, with CH, being produced in much smaller quantities and only at much higher temperature than on TiO, (620 K). The source of these differences is discussed in terms of the defect chemistry of the surfaces involved. Because of their thermal stability and their easy preparation into high-surface-area materials, oxides have a long history of being used as supports for catalysts.' However, they can also operate as catalysts themselves.Kung2 has summarised some of the reactions catalysed by transition-metal oxides. These range from dehydrogenations, selective oxidations and reductions, to the water-gas shift reaction. In spite of their inherent advantages mentioned above, oxides are complex materials, a fact which is also pointed out by Kung., They are generally polycrystalline in nature, which means that the surface is far from homogeneous. Furthermore, their surfaces contain defects, can be enriched or depleted in one element and can vary in chemical state (e.g. hydroxylation). These and further factors listed by Kung underline the importance of well controlled preparation conditions and the subsequent careful characterisation which is essential for understanding the surface chemistry and hence the reactivity of oxides.The interaction of methanol with oxide surfaces represents a good test of surface chemistry. It is as part of a broader programme aimed at extending the knowledge of the reacti- vities of oxide surfaces that the interaction of methanol with TiO, (rutile), SrTiO, and SrO has here been investigated. TiO, in its various crystal modifications is widely used as a catalyst support and is probably most associated with the SMSI effect.,,* It has also been employed in photoassisted reactions,' as have SrO and SrTiO,. The latter has been investigated in the photocatalytic water splitting reaction under zero applied potential in its pure form6 and in com- bination with NiO.' SrO has found little use in catalysis,*q9 but it has been investigated in terms of its acid-base properties" and its formation of different surface states upon heating.ll More recently, there has been some interest in doping La203 used in the oxidative coupling of methane with Sr,' 2*1 and the alkaline-earth-metal oxides and carbon- ates have been re-examined with XPS because of their poten- tial usefulness in superconducting materials.t SrTiO, is a mixed oxide. It can be prepared by mixing SrCO, and TiO, in a 1 : 1 ratio and heating at 1273 K for a prolonged period.' In terms of crystallographic structure it belongs to the class of perovskites, although its basic building t Present address : Department of Chemistry, University of Reading, Whiteknights Park, Reading, UK RG6 2AD.blocks, which are TiO, octahedra, are the same as those of TiO, (rutile). In the 1970s, the photodecomposition of H,O under conditions of band-gap illumination on both SrTiO, as well as on TiO, inspired a lot of work relating these obser- vations to the structural and electronic properties of the oxides under both oxidising and reducing conditions. These studies were generally carried out on single crystals and established the link between the oxide reactivity and the exis- tence of surface defects.'6-22 For instance, in the case of SrTiO,( 11l),the stoichiometric surface showed photoelectron peaks indicative of Sr2+, Ti4+ and 0,-including some residual hydroxy groups.The surface was inert to 0, and H2 exposure even under illumination. If this oxide was bom- barded with argon ions, oxygen ions were preferentially removed, which led to an increase in the observed Sr: 0 ratio and to the appearance of Ti3+ in the spectrum. The surface thus produced was very reactive to 0, resulting in reoxidation of Ti4+.I6 TiO, single-crystal studies have shown a range of simi-larities with SrTi03.'9-22 Lo et al. found that oxygen defi- ciency could be introduced by prolonged heat treatment at 1400 K19,,0 as well as by argon-ion bombardment. On 4 Ti02(100), this led to a new surface structure as was seen in the change of the LEED pattern and in the appearance of Ti3+ in the XPS spectrum. Both Lo et a/.,' and Heinrich et d2'observed an enhanced reactivity of the reduced surfaces towards adsorbates containing oxygen.The different nature of the defects introduced by either heat treatment or argon- ion bombardment has been discussed by Gopel et All these studies have in common that they are mainly con- cerned with the adsorption of either H,, 0, and H,O, that is adsorbates with direct relevance to water splitting. The overall conclusion from these studies is that the reduced sur- faces tend to adsorb 0, and H,O dissociatively and become more oxidised while the oxidised surfaces tend to adsorb these gases molecularly. More recent work on TiO, single crystals has concentrated on determining point defects2, and on the adsorption of other small molecules like meth-formic acid2' and ammonia.26 These studies utilise the control over the surface stoichiometry that is achieved on single crystals and attempt to correlate results from adsorp- tion experiments on these with similar results from oxide powders.We have here adopted a different approach. By using J. CHEM. SOC. FARADAY TRANS., 1994, VOL. 90 methanol adsorption as a test reaction it is aimed to probe the surface of three related oxides and especially to establish Ti 2P3/25000 fl*-"ihow the mixed oxide surface is composed. Experiment a1 The methanol decomposition experiments were carried out in a small vacuum chamber which was equipped with a VG Monitorr quadrupole mass spectrometer operated by an Opus PC5 computer.The working background pressure was 5 x lop7Torr. The samples were mounted in a small silica tube which could be resistively heated via a Pt/Rh wire wound around it. A chromel-alumel thermocouple was inserted into the oxide powder to obtain accurate temperature measurements. The heating was carried out in a constant-current mode which meant that the heating rate varied over the range 300-800 K. In the range of interest it was 3 K s-l, and it varied little over the region of a peak. TiO,, SrTiO, and SrCO, were supplied by Johnson Matthey and were of 99% purity. SrO was prepared from SrCO, by calcining it at 1200 K in air for 1.5 h and crushing it before it was inserted into the vacuum chamber where it was further calcined at 973 K for 1 h.Methanol of 99.5% purity was supplied by BDH. It was further purified by cycles of freeze-pump thawing. Prior to use each oxide was heated in 1 x lop4 Torr oxygen for 1 h at 773 K to remove any carbon contami- nation, and it was then briefly annealed at 773 K in ~acuo before the adsorption experiments took place. XP spectra of the three oxides were generated using a VG ESCA 3 mk I instrument. The oxides were pressed into discs and mounted on a Cu sample holder. They underwent the same heat treatment at 773 K in vacuo as described above before the spectra were recorded. The working background pressure was 5 x lo-' Torr, and the spectra were obtained with Al-Ko! radiation of 160 W using a pass energy of 50 eV and an angle of electron escape of 45".Results XPS Investigation of the Oxides Fig. 1 shows the Ti 2p, 0 1s and Sr 3p/C 1s regions of the spectra obtained from the three oxides. The relevant peak positions from these are summarised in Table 1, and they are quoted relative to C 1s = 285.0 eV. The Ti 2p region from both TiO, and SrTiO, consists of two single peaks in the approximate ratio of Ti 2p3,, :Ti 2p,,, = 2 : 1. For TiO,, the separation of the two peaks is 5.7 eV and the FWHM of Ti 2p,,, is 1.9 eV. The values for Ti in SrTiO, are identical, underlining the structural and chemical similarity of Ti in the two oxides. A comparison of the Ti 2p3,, peak position with literature values indicates that it is in the oxidation state 4 +.Carley et aL2' found the most intense Ti4+ peak in oxidised Table 1 XPS peak positions (binding energy in eV) of the most intense peaks in TiO, ,SrTiO, ,SrO and SrCO, TiO, 458.7 530.0 SrTiO, SrO 458.4 529.7 531.5 268.6 269.2 289.7 SrCO, 531.4 269.1 289.7 SrCO,b 531.5 269.0 289.5 SrOb 528.2 268.1 a C 1s peak other than the normal carbon reference at 285.0 eV. Values quoted from ref. 14. t 1 450 460 470 E,IeV 12000,. . . , , . . . I..3 I., Fig. 1 A, XP spectra showing the Ti 2p regions of (a)TiO, and (b) SrTiO, . ByXP spectra showing the 0 1s regions of (a)SrO, (b)TiO, and (c)SrTiO,, C, XP spectra showing the Sr 3p/C 1s regions of (a)SrO and (b)SrTiO, . J. CHEM. SOC.FARADAY TRANS., 1994, VOL. 90 Ti foil at 459.0 eV, and the two Ti 2p peaks were separated by 6.0 eV. The argon-ion bombardment of their oxide film led to the appearance of suboxides with Ti3+ at 457.5 eV and Ti2+ at 455.3 eV, that is, they were well separated from the Ti4+peak. Sayers and Armstrong28 who investigated a whole range of different titanium oxide powders and electrodes with XPS observed similar differences between Ti in different oxi- dation states as quoted above. They concluded that Ti4+ in TiO, is very stable unless the compound is subjected to severe ion bombardment. Their work also included SrTiO, which in terms of Ti 2p peak positions, peak separation and FWHM was found to be identical to TiO, . The 0 1s region shown in Fig.1B confirms the similarity between the two oxides. The peak positions are identical within the limits of detectability, and in both cases the signal originates from more than one peak. Assuming the presence of two peaks of Gaussian shape, a peak fit for TiO, yields the main peak at 529.9 eV (FWHM 2.1 eV) and a second smaller peak at 531.8 eV (FWHM 2.4 eV) with an area ratio of these two peaks of 1 : 0.3. For SrTiO,, the equivalent peak posi- tions are 529.5 eV (FWHM 2.1 eV) and 531.6 eV (FWHM 2.4 eV), and the area ratio is 1 : 0.6. With reference to the liter- ature quoted above these peak positions are indicative to oxide and hydroxide, respectively. It is concluded that in spite of the heat treatment in uucuo the surfaces of both oxides are extensively hydroxylated.Fig. 1B further shows that the 0 1s peak of SrO differs significantly from those of the other two oxides. Its position is at higher binding energies and its peak shape is asym- metrical with a tail to lower binding energies. Fig. 1C shows the combined Sr 3p/C 1s regions of SrTiO, and SrO. Besides the Sr peaks which are located at binding energies typical of Sr2+ 14,28 and which are identical in both oxides the main feature here is the additional peak at 289.7 eV in SrO (indicated by the arrow). By comparison with the literature', and with the spectrum of the starting material SrCO, from which SrO had been made this peak was identi- fied as carbon in carbonate. The spectrum of SrCO, has not been included but the relevant peak positions are listed in Table 1, as are the literature values for SrCO, and SrO.On this basis, the 0 1s region of SrO must be interpreted as being dominated by oxygen from carbonate. A peak fit of this region yielded the best results when assuming that it consist- ed of three Gaussian peaks. The first peak is located at 528.7 eV (FWHM 2.0 eV) and is attributed to oxide, the second a,t 530.7 eV (FWHM 2.0 eV) and identified as hydroxide, and the last one at 531.8 eV (FWHM 2.1 eV) which is typical of carbonate. The fitted peak area ratio is 1 : 5.5 : 7.6. The presence of a mixture of oxygen-containing species on the surface of SrO is further confirmed by estimating the surface composition of the materials.Table 2 lists the atomic ratios of the main components in the three oxides. These were calculated according to Penn2' using the XPS peak areas, the photoionisation cross-sections tabulated by Sco- field,30 the detection efficiency taken as E-' (E being the kinetic energy of the photoelectron involved) and the mean free path of the photoelectrons based on Penn's formula. The Table 2 Atomic ratios of the main components in the three oxides and in the carbonate which were derived according to ref. 29 Ti/O Sr/O Ti/Sr Sr/C" C/O TiO, 0.58 SrTiO, 0.23 0.30 0.78 SrO 0.40 2.50 0.16 srco, 0.38 1.22 0.3 1 'C 1s peak area of carbon in carbonate, located at 289.7 eV. ratios confirm that SrO is not pure oxide, that TiO, is essen- tially stoichiometric and that SrTiO, may be slightly deficient in Ti or enriched in Sr and 0 in the selvedge region.Thermal Desorption of Methanol adsorbed on TiO, The product distribution following the highest exposure of 3.0 Torr s is shown in Fig. 2A. The main products are CO, CH,, CH,OH and H, ,with H,O, HCHO and (CH,),O pro- duced in smaller quantities. CO, was seen to be desorbed only at the higher temperature of 580 K and it was too small to be seen on this scale of the figure. The evolution of these species as a function of exposure is illustrated in Fig. 2B-G. The exposures used here were 0.12, 0.24 and 0.5 Torr s. Fol-lowing the first exposure CH, is the major product being desorbed over a wide range of temperature, but with main peaks at ca.460 K and ca. 650 K; small amounts of CH,OH and HCHO are also seen. The increase in exposure leads to an increase in the amount of CH, desorbed and to a decrease in the desorption temperature of the high-temperature state. The CH, evolution is now accompanied by CO, CH,OH, HCHO, H,O and H,. The intensity of all these desorption peaks increases with further exposure, the two stages of desorption being observed for CH,, CH,OH and HCHO while CO and H, are only evolved in a single-temperature state. After dosing 3.0 Torr s of CH,OH the two desorption states are merged together to yield one broad peak. The H,O evolution occurs over the temperature range 350-700 K, and CH,OH is desorbed at around 415 K, slightly lower than CH,, while HCHO is mainly evolved between 400 and 600 K.(CH3),0 is only observed following the highest exposure and it is desorbed at 440 K. At this stage, repeated adsorptions and desorptions did not change the surface, that is, the desorption spectra were identi- cal for identical doses. However, at the end of these experi- ments the oxide was heated in oxygen (0.05 Torr for 15 min at 773 K) and then exposed to methanol. Now evolutions of CO, and H,O were seen below 373 K, the desorptions of CH,OH and HCHO were very much increased and the desorption of CH, decreased by around 50%. Repeated exposures of methanol followed by thermal desorption resto- red the original intensities. Thermal Desorption of Methanol adsorbed on SrO The product distribution following an exposure of 0.15 Torr s of methanol is shown in Fig.3. At this low dose the spectrum is dominated by the coincident evolution of CO and H, at 580 K. At higher exposures only the CO and H, desorptions continue to increase. The evolution of these as a function of exposure have therefore not been included. At the same tem- perature of the CO and H, desorptions there are also smaller amounts of CO,, H,O, HCHO and (CH,),O. HCHO and (CH,),O were too small to be included in the figure. CH,OH is evolved at 405 and 470 K and is of low intensity compared with CO and H,. Thermal Desorption of Methanol adsorbed on SrTiO, The product distribution is shown in Fig. 4A, and the dose dependence of the fragments is illustrated in Fig.4B-E. The spectrum is dominated by the evolution of CH,OH around 440 K with H,, H,O and CH, being lesser products. Coin- cident with CH, there are also traces of HCHO, CO, CO, and H, which are too small to be seen on this scale. (CH,),O evolution is not observed at all. From Fig. 4E it is seen that CH, is the main product following the lowest exposure. Although this resembles the behaviour of CH,OH on TiO, J. CHEM. SOC. FARADAY TRANS., 1994, VOL. 90 20000rI 300 400 500 600 700 800 900 1000 300 400 500 600 700 800 "300 400 500 600 700 800 3000 , . , ' I 3000 1 ' 1 .. 11 '' 11 '" D ElE h cv) C .- h v)c, C .- = 20004 = 4 2000 v- v- ([IC 0,.- m C 01.- v) v) d d 5:* * 1000 I (5:1000 E 2v) 9 10 400 500 600 700 800 '300 400 500 600 700 800 300 400 500 600 700 800 TIK TI TIK 5000 G 300 400 500 600 700 800 TIK Fig.2 A, Thermal desorption spectra showing the product distribution from 3.0 Torr s of methanol adsorbed on TiO, at room temperature: (a) CH,, (b) CH,OH, (c) CO, (d)H,, (e)H,O, (f)HCHO, (9)DME. B, Thermal desorption spectra showing the evolution of CH, after dosing methanol onto TiO, at room temperature. The doses are 0.12, 0.24 and 0.5 Torr s (scale x 5). C, Thermal desorption spectra showing the evolution of CH,OH after dosing methanol onto TiO, at room temperature. The doses are 0.12, 0.24 and 0.5 Torr s (scale x 4). D, Thermal desorption spectra showing the evolution of CO after dosing methanol onto TiO, at room temperature.The doses are 0.12, 0.24 and 0.5 Torr s (scale x 1). E, Thermal desorption spectra showing the evolution of H, after dosing methanol onto TiO, at room temperature. The doses are 0.12, 0.24 and 0.5 Torr s (scale x 1).F, Thermal desorption spectra showing the evolution of H,O after dosing methanol onto TiO, at room temperature. The doses are 0.12,0.24 and 0.5 Tom s (scale x 6.7). G, Thermal desorption spectra showing the evolution of HCHO after dosing methanol onto TiO, at room temperature. The doses are 0.12,0.24 and 0.5 Torr s (scale x 20). J. CHEM. SOC. FARADAY TRANS., 1994, VOL. 90 As was pointed out in the Results section, the binding energy values of Ti 2p and 0 1s in TiO, are identical to those 5x1051.1.1.1.1.in SrTiO,.As far as the 0 1s position is concerned, noI4x1 O5 w.-% I v 3~10~1 (4 (4 0 300 400 500 600 700 800 900 1000 TfK Fig. 3 Thermal desorption spectra showing the product distribu- tion from 0.15 Torr s of methanol adsorbed on SrO at room tern- perature: (a)H,, (b) CO, (c) H,O, (d) CO, ,(e)CH,OH the desorption is shifted much higher in temperature to 620 K. Following the second exposure, all the other species have also appeared and with the exception of CH,, these are pre- dominantly evolved in the lower-temperature state. At satura- tion, the CH,OH peak has broadened and is shifted by 10 K to lower temperatures. When this oxide was heated in oxygen in the same way as TiO,, the result was again to decrease the evolution of CH, and increase that of H,O.Discussion Surface Cbaracterisation by XPS The XPS results have shown that it is difficult to prepare SrO in a pure form. Mohri et aL8 indicated that heat treatment above 873 K was necessary to observe any catalytic activity of SrO for the isomerisation of but-1-ene. They prepared SrO from Sr(OH), and attributed the lack of activity at lower pretreatment temperatures to the presence of CO, on the surface which blocked the reaction. When the oxide was heated by itself, they observed the evolution of H20 at 623 K and CO, at 973 K. Heating to higher temperatures did not increase the amount of CO, evolved which was estimated to be 40 times as much as would be needed to make a mono- layer on the oxide surface.Coluccia and Tenck" found that heating the sample at 953 K led to the complete bulk decom- position of SrCO, to SrO, although they could still detect some CO, on the surface with IR. In the present case, SrO had to be transported through air for the XPS analysis which will have led to CO, and H,O uptake. The carbonate and hydroxide thus formed which dominate the spectrum could not be decomposed properly as the sample heating on this instrument was limited to around 770 K. Consequently, a spectrum of pure SrO could not be obtained. As far as the surface stoichiometry of SrO is con- cerned no conclusions can therefore be drawn. (SrO has a rocksalt structure. It is cubic face-centred and each Sr atom is surrounded by six 0 atoms and vice versa.) However, it is assumed that the oxide surface was predominantly composed of oxide during the thermal desorption experiments as it had been heated to 973 K before these were carried out.simple correlation between the binding energy and the metal oxide bond character in general has been found so far.,' For Ti it is concluded that it is in the same oxidation state in both oxides. In ref. 27 and 28 it was stated that it is very difficult to reduce either oxide by heating in uucuo or even heating in H, at the temperatures employed here. Only after argon-ion bombardment was it possible to observe Ti3+. The results presented here are consistent with this statement. Only Ti4+ is detected and the atomic ratio Ti :0 indicates a stoichio- metric TiO, surface.The TiO, surface is often considered to be best represented by the (110) plane, as its preparation by cleavage in vucuo requires the breakage of the fewest number of bonds. It is visualised as consisting of bridging oxygen atoms in the [Ool] direction on top of every second row of in-plane Ti atoms. This ensures that half the Ti atoms have six-coordination and half have five-c~ordination.~~.~ For SrTiO,, two model surfaces have been proposed by Heinri~h.~~The (111) surface consists of a layer of SrO on which Ti atoms are located in three-fold oxygen sites. The (100) surface is composed of alternating layers of SrO and TiO, and the crystal termination is either the one or the other or a mixture of both. A recent STM study of this surface suggested that it always has a TiO, face which con- sists of a (2 x 2) ordered oxygen vacancy structure with Sr scattered on top.34 In the work presented here, SrTiO, was found to be depleted in Ti and enriched in Sr and 0 at the surface.This is contrary to the observations of Lo and Somorjai who were interested in compositional changes of an SrTiO,( 11 1) surface following different treatments. l9 They established that heating the stoichiometric surface at 873 K in uucuo resulted in a decrease in both the Sr :0 and Ti : 0 ratios. This change was found to be a reversible process as the original values were restored by cooling the sample to room temperature. Sputtering led to the preferential removal of 0 and Ti, and subsequent heating caused diffusion of Sr into the bulk.In contrast to this, high-temperature sputtering produced an Sr-rich surface. As far as the SrTiO, powder is concerned, it must be assumed that these models are useful as a first approach to understanding the surface structure of the two oxides, but that they are too simplified to account for all the observations. The exposure of a range of different planes as the surface as well as the presence of defects will influence the results, and these factors have not been considered in the models. Lastly, the XPS results revealed that the surfaces of the oxides are hydroxylated. In the Introduction, it was already quoted that Ferrer and Somorjai'6 found the surface of SrTiO,(lll) to contain residual hydroxy groups after it had been cleaned and heated in vacua TiO, is well known to dehydroxylate on heating in UUCUO.~~The greatest loss of OH takes place up to 600 K after which a slow further decrease in the number of hydroxy groups with time and temperature is obtained, although it never reaches zero.This is consistent with the results presented here. It is interesting to note that although SrTiO, underwent the same heat treatment as TiO, its surface remains more hydroxylated. Thermal Desorption Experiments The mechanism of methanol adsorption and decomposition is generally explained in the following way.36*37 The initial adsorption of methanol can be either molecular [step (l)] or dissociative [step (2)] : 1020 J. CHEM.SOC. FARADAY TRANS., 1994, VOL. 90 80004x105ri. h a. 6000 .: 7 :*:i ?... : ! .. 'i :i:..4 ..Y . -.I . i.*.Q .. . ...Cn 4000 2: v aPv) rn i! 2000 I.I.I.1.0 0 300400 5006007008009001000 31 1 400 500 600 700 800 0-300 400 500 600 700 800 TIK TIK 4000 5000r----TIK h .:3000 32 v-Q .P 2000 v 2 v) ro E 1000 I.I.l.1.0 300 400 500 600 700 800 10 TIK Fig. 4 A, Thermal desorption spectra showing the product distribution from 3.0 Torr s of methanol adsorbed on SrTiO, at room temperature; (a)CH,OH, (b) H, ,(c)H,O, (d) CH,. B, Thermal desorption spectra showing the evolution of CH,OH after dosing methanol onto SrTiO, at room temperature.The doses are 0.12, 0.24 and 0.5 Torr s (scale x 5). C, Thermal desorption spectra showing the evolution of H, after dosing methanol onto SrTiO, at room temperature. The doses are 0.12, 0.24 and 0.5 Torr s (scale x 1). D, Thermal desorption spectra showing the evolution of H,O after dosing methanol onto SrTiO, at room temperature. The doses are 0.12, 0.24 and 0.5 Torr s.(scale x 6.7). E, Thermal desorption spectra showing the evolution of CH, after dosing methanol onto SrTiO, at room temperature. The doses are 0.12, 0.24 and 0.5 Torr s (scale x 5). In fact, the coincident desorption of CH,OH, HCHO and H,CH,OH, *CH,OH, (1) has been taken as proof of the existence of methoxy groups CH,OH, -+ CH,O, + Ha (2) on a ZnO surface.36 If the surface is in a reduced state and hence binds the oxygen in methoxy strongly a further reac-The dissociative adsorption is thought to occur at a coordi-tion is possible.In order to replace this oxygen vacancy onnatively unsaturated Ti cation (Ti,) and an oxygen anion the surface the C-0 bond may be broken and the oxygen[step (311: atom incorporated as surface lattice oxygen, 0, [step (7)] : CH,OH, + 0, + Ti, -,CH,O, + OH, (3) CH,O, +CH,, + 0, (7)The methoxy species thus formed can react in several ways, one of which is by recombining [the reverse of step (2)] to This frees the methyl group which can react with adsorbed H form methanol. This would lead to molecular desorption. or OH from step (3)to be released as CH4 [step (8)] or it can Alternatively, the methoxy group can dehydrogenate further decompose further to adsorbed C and H [step (9)].to yield HCHO and H, in the vacuum [steps (4)-(6)] : CH3, + Ha+CH4, (8) CH,O, +CH,O, + Ha (4) CH,, -+C, + 3H, (9) CH,O, +CH20, (5) The removal of hydroxy groups or oxygen deposited in steps CH,O, +CO, + H2, (6) (3) and (7) can occur in two ways, either through dispro- J. CHEM. SOC. FARADAY TRANS., 1994, VOL. 90 portionation of neighbouring hydroxy groups [step (lo)] or through the reaction of adsorbed H with adsorbed 0 and OH [step (1 l)]. OH, + OH, -+ H,O + 0, Kim pointed out that oxygen could also be removed through the oxidation of carbon deposited in step (9) [step (1211: c, + 0,-+ cog The evolution of (CH,),O has generally been explained as a bimolecular reaction which requires two neighbouring methoxy species [step (13)]: (13) .~~On ZnO Bowker et ~1 found that adsorbed methoxy groups could undergo a further reaction. Here they coexisted with formate on the surface, the former decomposing into CH,OH, HCHO, H, and CO at 380 K while the latter decomposed into mainly CO and H, together with small amounts of CO, and H,O at 580 K.Methoxy groups are initially oxidised by basic centres on the surface, leading to the binding of the methoxy carbon to another oxygen atom with rapid deprotonation to leave formate [step (14)]. The formate decomposes according to steps (15) and (16). CH,O, + 30, -+ HCOO, + 20H, (14) HCOO, + CO, + Ha + 0, Kim et observed on the (011) facetted surface of~11.~~9~~ single-crystal Ti0,(001) and TiO, powder (both anatase) that molecularly adsorbed CH30H was desorbed predominantly at 275 K and was completely removed at 365 K.Onishi et using UPS, also found methanol to be adsorbed molec- ularly on Ti0,(110) and (441).In the work presented here, it is not possible to decide whether methanol is adsorbed molecularly [step (l)] or dissociatively [step (2)]. In any case, the dissociation according to steps (2) and (3) occurs once the thermal desorption experiment is started, as a range of decomposition products are observed which are attributed to the adsorbed intermediate methoxy groups. Different decom- position paths are in competition with each other and their relative probability will depend on the temperature, the coverage, the state of the adsorption site and the availability of other species like H and 0.At low coverage, CH, forma-tion is favoured.This must imply that the surface is in a par- tially reduced state containing coordinatively unsaturated centres and vacancies. On the one hand, the XPS results indi- cated that the original surface is stoichiometric and consists of Ti4+ only. On the other hand, the oxidation experiment described in the Results section showed that the TiO, surface is reduced during the thermal desorption of methanol. The introduction of vacancies must therefore be attributed to pre- liminary desorption experiments which were carried out.This would leave the surface in a partially reduced state made up of both Ti3+ and 0 lattice vacancies. These then have a reduced coordination and will react to restore saturation. The adsorption of methanol hence takes place according to step (3) and the filling of the 0 vacancy is complete once the C-0 bond of the methoxy group is broken [step (7)]. This becomes an irreversible process as far as methanol is con- cerned and results in CH, formation [step (8)]. In this way, through every decomposed molecule of CH,OH the surface gains one atom of oxygen. If this was a continuous process it would be expected that the adsorption behaviour would change from that of a reduced surface to that of an oxidised surface with increasing exposure.In that case, the evolution of CH, would be expected to decrease with the number of lattice oxygen vacancies. As this is not the case here, it must be inferred that the oxygen that is deposited in step (7) does not remain on the surface but is removed during the TPD experiment. The evolution of H20 takes place over a wide temperature range and is most likely to account for the loss of 0 from the surface [steps (10)and (ll)]. In fact, it might be speculated that the first loss of H,O initiates further decom- position of adsorbed methanol through the creation of further vacancies. After the thermal desorption, the surface is essentially unchanged, that is, it is in the same state as before the adsorption of methanol. It is worth noting that Kim et al.also found the oxide to be reduced during use and that they detected H,O desorption over the temperature range 350-600 K during thermal desorption of methanol from TiO, anatase.,,*,’ At higher methanol exposures, the reversible adsorption becomes more probable [reversal of step (2)]. The higher con- centration of methanol and methoxy groups on the surface means that the existing vacancies can be filled and that at the same time other reaction paths are available. This is seen in the increased evolutions of CH,OH, HCHO, H, and CO. CH, formation still occurs and it takes place at two different temperatures. This is indicative of two different sites on the surface which participate in this reaction. These two sites also give rise to methanol and HCHO, the former being formed through the reversal of steps (2) and (3), while the latter is formed according to steps (4) and (5).The H liberated through step (4) is not desorbed [step (6)] but instead is con- sumed in the reformation of methanol [reversal of steps (2) and (3)] or alternatively remains on the surface as OH. From Fig. 2A it is seen that the formation of CH, is more favour- able at 440 K than at 580 K while CO and H, are evolved only at the higher temperature. The coincident evolution of CO and H, is attributed to the further oxidation of HCHO formed from methoxy [steps (4)-(6)] which is likely to proceed via a formate intermediate. The fact that H, is of lower intensity than CO is due to its consumption in the pro- duction of H,O [step (1 l)] and CH, [step (8)].Kim et uE.,~ observed the evolution of CO which they attributed to the oxidation of carbon deposits at tem-peratures in excess of 600 K. In the present work CO evolu-tion took place at slightly lower temperatures and it was accompanied by H, evolution. This indicates that it is linked to the decomposition of methoxy groups rather than being a result of carbon oxidation. Repeated adsorption experiments also did not reveal any major decrease in activity, so a block- age of the surface by carbon had not occurred. It is con- cluded that steps (9) and (12)do not take place here. The evolution of (CH,),O according to step (13) is only feasible if two methoxy species are adsorbed next to each other.It is therefore not surprising that this product is only evolved following the highest exposure and that it is desorbed before other reactions become dominant and remove methoxy. Similarly, Fig. 2A shows that following the highest exposure the CH,OH desorption peaks at low temperatures and is lost before other reactions like the formation of CH, became dominant. These results are in qualitative agreement with those of Suda et and Ramis et al.,’ who using IR found that methanol and other alcohols were adsorbed dissociatively to form alkoxides on TiO, (rutile) which they attributed to the dehydroxylated state of the surface. Cunningham et ~l.,~’ who also studied alcohol adsorption on TiO, (rutile) follow- ing extensive heating in uucuo, also observed the evolution of CH, together with some higher alkenes.The work presented here also indicates that the adsorbed surface species on TiO, is the methoxy group. This contrasts strongly with the behaviour of methanol on SrO where the dominant desorptions are those of coincident CO and H, at 580 K. This pattern has been attributed to the existence of formate on the surface of ZnO as outlined above [steps (14)- (16)]. Both the temperature of desorption and the intensity of the products are very similar to the results obtained on Zn0.36 The reversible adsorption of CH,OH also takes place but it is more limited than on TiO,. On this oxide only one type of site is present which leads to one adsorbed species and one set of decomposition peaks.Formate was not detected on TiO, (anatase) or Ti0,(00) by Kim et a1.24137 On SrTiO,, there are again two types of sites available as indicated by the two desorption states. However, the thermal desorption of adsorbed methanol differs from that on both the other oxides. As was shown with XPS this surface is more highly hydroxylated and possibly enriched in 0 and Sr/ depleted in Ti. Both these factors would decrease the likeli- hood of forming CH,. This is indeed confirmed by the fact that CH, is only evolved in the higher-temperature state. This CH, evolution is qualitatively similar to that from TiO, in that it is affected by the oxidation of the surface, but it is shifted to significantly higher temperatures and it is of lower intensity.It may correspond with the high-temperature peak seen in the low-coverage data from TiO,. The increased abundance of oxygen on this surface has successfully blocked the low-temperature route to CH, in favour of CH,OH desorption. Methoxy groups are the likely intermediate to form CH, ,since coincident, minor products of CO and form- aldehyde are seen. Methane formation presumably occurs by CO bond dissociation to fill what was previously an anion vacancy. A significant difference between this material and TiO, itself is the lack of CO evolution (compare Fig. 2 and 4), indicating no formate formation. This perhaps relates to the greater stability of the methoxy on the mixed oxide which in turn also alters the selectivity pattern.These results are in agreement with the trend observed by others that a more reduced surface would tend to bind the oxygen of methanol strongly hence leading to dissociative adsorption followed by other reactions. In order for disso- ciation to take place a pair like Ti3+ next to an oxygen vacancy is required which forms the reactive centre. On a more oxidised surface such a centre is less likely to exist. The adsorption therefore occurs on a single site and is more likely to be molecular. In this respect, although Kim et al. showed methanol to dissociate on a well oxidised Ti02(001) surface24 (they detected methoxy groups by XPS after room-temperature adsorption), the most likely reaction during thermal desorption was the recombination to form CH,OH in the gas phase since reactive centres (vacancies) for further decomposition were not available.With respect to the results shown here, this is similar to the behaviour of methanol on SrTiO,. The fact that two reaction sites are present on SrTiO, as well as on TiO, must be attributed to similar surface structures. The basic building block of both oxides is the TiO, octahedron. The inclusion of Sr cubes around these octahedra modifies the behaviour of TiO,. The presence of Sr has influenced TiO, structurally and electronically, making it more difficult to remove surface hydroxy groups and shifting the high-temperature decomposition of methoxy groups to even higher temperatures. Although XPS results show an enrichment of Sr and 0 on SrTiO, this does not indi- cate the presence of an extended SrO structure, evidenced by the lack of formate formation and decomposition on SrTiO, .This is attributed to the very much changed environment of Sr in SrTiO, as compared with Sr in SrO. From this point of view, an STM study of SrTi0,(100) found the surface struc- ture to be dominated by an ordered oxygen/Ti structure, and J. CHEM. SOC. FARADAY TRANS., 1994, VOL. 90 this is reflected well in the surface behaviour of SrTiO, powder.34 The authors thank the SERC for the funding of the postdoc- toral position to N.A. uia the IRC in Surface Science. References 1 G. C. Bond, Heterogeneous Catalysis: Principles and Applica- tions, Clarendon Press, Oxford, 1974, ch.4, pp. 57-65. 2 H. H. Kung, Transition Metal Oxides: Surface Chemistry and Catalysis, (Stud. Surt Sci. Catal. 45), Elsevier, Amsterdam, 1989, ch. 1, pp. 1-5. 3 S. J. Tauster, S. C. Fung and R. L. Garten, J. Am. Chem. SOC., 1978,100,170. 4 G. L. Haller and D. E. Resasco, Adu. Catal., 1989, -36, 173. 5 R. I. Bickley, Specialist Periodical Report : Catalysis, The Royal Society of Chemistry, London, 1982, pp. 309-332. 6 F. T. Wagner and G. A. Somorjai, J. Am. Chem. SOC., 1980, 102, 5494. 7 K. Domen, A. Kido, T. Onishi, N. Kosugi and H. Kuroda, J. Phys. Chem., 1986,90,292. 8 M. Mohri, K. Tanabe and H. Hattori, J. Catal., 1974,32, 144. 9 G. Zhang, H. Hattori and K. Tanabe, Appl. Catal., 1988,36, 189. 10 K. Tanabe, Catalysis Science and Technology, ed.J. R. Anderson and M. Boudart, Springer-Verlag, Berlin, 1981, vol. 2, pp. 231-273. 11 S. Coluccia and A. J. Tenck, J. Chem. SOC., Faraday Trans. I, 1983,79,1881. 12 M. T. Xu and J. H. Lunsford, Catal. Lett., 1991, 11, 295. 13 S. J. Conway, J. A. Greig and G. M. Thomas, Appl. Catal. A, 1992,86,199. 14 M. 1. Sosulnikov and Yu A. Teterin, J. Electron Spectrosc. Relat. Phenom., 1992,59, 11 1. 15 Gmelins Handbuch der Anorganischen Chemie, System-Nr. 41, 8. Auflage, Verlag Chemic, Weinheim, 1951, p. 433. 16 S. Ferrer and G. A. Somorjai, Surf: Sci., 1980,94,41. 17 S. Ferrer and G. A. Somorjai, Surf: Sci., 1980,97, L304. 18 R. G. Egdell and P. D. Naylor, Chem. Phys. Lett., 1982,91,200. 19 W. J. Lo and G. A.Somorjai, Phys. Rev. B, 1978,17,4942. 20 W. J. Lo, Y. W. Chung and G. A. Somorjai, Surf: Sci., 1978, 71, 199. 21 V. E. Heinrich, G. Dresselhaus and H. J. Zeiger, Solid State Commun., 1977,24,623. 22 W. Gopel, G. Rocker and R. Feierabend, Phys. Rev. B, 1983,28, 3427. 23 C. A. Muryn, P. J. Hardman, J. J. Crouch, G. N. Raiker, G. Thornton and D. S-L. Law, Surf. Sci., 1991,251/252,747. 24 K. S. Kim and M. A. Barteau, Surf: Sci., 1989,223,13. 25 H. Onishi, T. Aruga, C. Egawa and Y. Iwasawa, SurJ Sci., 1988, 193, 33. 26 E. Roman and J. L. de Segovia, Surf: Sci., 1991,251/252,742. 27 A. F. Carley, P. R. Chalker, J. R. Riviere and M. W. Roberts, J. Chem. SOC., Faraday Trans. I, 1987,83,351. 28 C. N. Sayers and N. R. Armstrong, Surf. Sci., 1978,77,301. 29 D. R. Penn, J. Electron Spectrosc. Relat. Phenom., 1976,9, 29. 30 J. H. Scofield, J. Electron Spectrosc. Relat. Phenom., 1976,8, 129. 31 N. S. Mclntyre, in Practical Surface Analysis by Auger and X-Ray Photoelectron Spectroscopy, ed. D. Briggs and M. P. Seah, John Wiley, Chichester, 1983, ch. 10, p. 410. 32 G. S. Rohrer, V. E. Heinrich and D. A. Bonnell, SurJ Sci., 1992, 278, 146. 33 V. E. Heinrich, Prog. Surf: Sci., 1983,14, 175. 34 T. Matsumoto, H. Tanaka, T. Kawai and S. Kawai, Surj Sci., 1992,278, L153. 35 A. H. Boonstra and C. A. H. A. Mutsaers, J. Phys. Chem., 1975, 79, 1694. 36 M. Bowker, H. Houghton and K. C. Waugh, J. Chem. SOC., Faraday Trans. I, 1981,77,3023. 37 K. S. Kim, M. A. Barteau and W. F. Farneth, Langmuir, 1988,4, 533. 38 Y. Suda, T. Morimoto and M. Nagao, Langmuir, 1987,3,99. 39 G. Ramis, G. Busca and V. Lorenzelli, J. Chem. SOC., Faraday Trans. I, 1987,83,1591. 40 J. Cunningham, D. J. Morrissey and E. L. Goold, J. Catal., 1978, 53, 68. Paper 3/06196E; Received 18th October, 1993
ISSN:0956-5000
DOI:10.1039/FT9949001015
出版商:RSC
年代:1994
数据来源: RSC
|
18. |
FTIR study of carbon monoxide adsorption on ceria: CO2–2carbonite dianion adsorbed species |
|
Journal of the Chemical Society, Faraday Transactions,
Volume 90,
Issue 7,
1994,
Page 1023-1028
Claude Binet,
Preview
|
PDF (850KB)
|
|
摘要:
J. CHEM. SOC. FARADAY TRANS., 1994, 90(7), 1023-1028 FTIR Study of Carbon Monoxide Adsorption on Ceria: COZ- Carbonite Dianion Adsorbed Species Claude Binet,* Ahmed Badri, Magali Boutonnet-Kizlingt and Jean-Claude Lavalley Laboratoire Catalyse et Spectrochimie, URA CNRS 04 14, ISMRA-Universite, 6, Boulevard du Marechal Juin--14050 Caen Cedex, France CO has been adsorbed in doses at room temperature by ceria which had been reduced with CO and then evacuated at 873 K, or only evacuated at this temperature, as well as by a Pd/CeO, catalyst, H,-reduced and evacuated at 823 K. From comparison of the observed IR spectra with literature results concerning matrix- isolated Cs,CO, species, the formation of two carbonite COZ-species adsorbed over reduced ceria may be inferred.One of these species, with Czvgeometry and with IR bands at around 1300 and 1160 cm-’, was formed before the other one, with a C,structure and characterized by IR bands at 1270, 1071 and 771 cm-’.A band at 1465 cm-’ was only tentatively assigned to a third carbonite species. COZ-species ultimately lead to C0:-species when they are heated in vacuum. Ceria is a component of the three-way post-combustion cata- lysts used in treatment of automobile exhaust gases. The nature of the species formed on ceria through CO adsorption at room temperature depends on the degree of reduction of ~eria.’-~Experimentally, by increasing the temperature, the reduction of ceria may proceed in two stages more or less well separated and dependent on the specific area of the ~eria.~Beginning at about 473 K, a reduction of the surface alone was first observed, which became almost complete at 573-633 K and then a reduction of the bulk occurred.For low temperatures of reduction, carbonates were the main species observed when CO was adsorbed at room tem-perature. For reduction temperatures as high as 823 K, formate species were almost exclusively ~bserved;~.~ the origin of the hydrogen involved in the formate species may be, when used, the H,-reducing agent, but it may also be the result of reduction of the hydroxylated surface. Hydrogen could be eliminated by evacuation of the sample at higher temperatures ; indeed no formate species were observed through the CO adsorption onto ceria H,-reduced at 673 K and evacuated at 10oO K,’ but the proposed formation of carbonate or carboxylate species through CO dispro-portionation into C and CO, may not be beyond contro- versy.Whereas, for example, the mechanism for CO chemisorption on the TiO, (1072) face at low CO exposures was tentatively identified as a dissociative adsorption at defect sites,6 a large extent of CO disproportionation on metallic oxides at room temperature is rather unusual. So, in this work, we re-investigated the IR spectra of the species resulting from CO adsorption by ceria that had been reduced at high temperature (873 K). Usually, a reducing treatment at a temperature, T,, is fol- lowed by an outgassing at the same or a higher temperature than the previous one. The effect of the last stage of pretreat- ment has to be considered, bearing in mind that defect sites on basic oxide surfaces may be created simply by evacuating at high temperature.More or less spectroscopically well char- acterized ‘carbonite’-like species were evidenced through CO adsorption on these defect sites.7 In this work, CO was adsorbed by ceria either (i) reduced by CO at 873 K and evacuated at this temperature or (ii) only evacuated at 873 K. The reducing agent, when used, was not H, but CO in order to avoid hydrogen storage by the ceria. Furthermore, CO is an IR spectroscopic molecular probe t Present address: Institute for Surface Chemistry, Box 5607, 11486 Stockholm, Sweden. commonly used in transition-metal supported-catalyst studies.In the case of Pd/CeO, catalysts, where CeO, is a reducible support, work is in progress in our laboratory using CO for both the metal and support characterizations, and also for the metal/support interface study during the reduction process. In order to examine the influence of the metal on the reduction of ceria, CO was adsorbed by a Pd/CeO, (2.5 wt.% Pd) catalyst reduced with H, at 823 K and evacuated at the same temperature. Under these pretreatment conditions, Pd was completely recovered with ceria and was inaccessible to adsorption of CO.* Whatever the sample used (CeO, reduced with CO, CeO, only evacuated or Pd/CeO, reduced with H2), the resulting spectra from adsorption of CO were qualitatively the same.Because the intrinsic quality of these spectra is variable, the results presented here are sample selected. Experimental Ceria (from Rhane-Poulenc) has a specific BET surface area of 160 m2 g-’. It was pressed into a disc (15 mg ern-,). The treatment of the disc was carried out in situ in the IR cell and was accomplished in two steps: (a) an oxidation with oxygen (P = 20 kPa, T = 723 K, t = 30 min) followed by evacuation at the same temperature; (b) a reduction treatment by CO (P = 2 kPa, T = 873 K, t = 20 min) followed by evacuation at 873 K; this reduction treatment was carried out three times. Then, the specific BET surface area decreased to 40 mz g-l. Some experiments were performed with a sample for which no reducing agent was used in step (b); in that case a longer evacuation time was considered (P = lop3 Pa, t = 1 h).The Pd/CeO, catalyst was prepared from ceria and Pd(NO,), + 2H,O by a microemulsion method.’ The same pretreatment as described above was applied by replacing CO with H, and reducing at 823 K.Note that, at ambient atmosphere, carbonation of ceria is unavoidable. During the oxidation step of the pretreatment a major part of the adsorbed carbonate species was eliminated, but more firmly held carbonates, called ‘core-carbonates ’,, were hardly affected even during the high-temperature reduction step. In this last step of the pretreatment no gaseous oxygen equilibrated with the sample, consequently the thermal decomposition of the core-carbonate species would possibly lead to defect sites on the ceria surface.The IR spectra were recorded at room temperature with a Nicolet 60 SX FTIR spectrometer (resolution 4 cm-’). The spectra presented in this paper were obtained from the differ- ence between the absorbances of the sample after and before adsorption of CO. Results Overall Characterization of the Adsorbed Species CO was adsorbed in doses at room temperature by ceria that had been reduced at 873 K. The resulting spectra are given Fig. l(a) and (b). Then CO was admitted under 0.5 kPa pres- sure [Fig. l(c)]. For the first dose two rather wide bands are observed at 1160 and 1300 cm- ' [Fig. l(a)]. The correspond- ing adsorbed species are said to be of type A. From different experiments, the 1160 cm-' band appeared to have at least two unresolved components at 1150 and 1170 cm-'; the same observation could be made for the 1300 cm-' band with components at 1297 and 1317 cm-'.However, owing to the lack of reproducibility of such components, this analysis remains doubtful and only the band envelope will be con- sidered further. When further quantities of CO were intro- duced [Fig. l(b) and (c)] bands from another adsorbed species appeared at 771, 1071 and 1270 cm-' (species B). In particular [Fig. l(c)], very narrow bands characteristic of two formate species adsorbed over reduced ceria3 were observed with the following vibrational modes: v,(OCO) at 1600 and 1569 cm-', v,(OCO)at 1365 and 1352 cm-' and S(OC0) at 790 cm-'.Bands of variable definition were observed at around 1625,1495,1455,1380 and 927 cm-'. The thermal stability of the above-mentioned species was investigated (Fig. 2) by heating the sample in vacuum suc- cessively at 573 K [Fig. 2(b)] and 673 K [Fig. 2(c)]; spectra were scanned after cooling the sample to room temperature. The effect of evacuation on the spectra may be evidenced by comparing Fig. 2(a) with Fig. l(c); these two curves concern the sample under vacuum and under 0.5 kPa CO. The only noticeable difference is the relative intensity of the 1160 cm- ' band (species A) which decreases upon sample evacuation. The partial vacuum instability of species A may be used for the characterization of these species.Upon heating the sample at 573 K [Fig. 2(b)] the narrow formate bands located near 1590, 1360 an 790 cm-' disappear. While bands due to species B are not affected, a decrease in the intensity of the bands for species A is observed [Fig. 2(b)] as soon as a band emerges at 1205 cm-'. Although suggested by other experi- ments, a correlation between the thermal conversion of species A and the band at 1205 cm-' remains questionable. However, a contribution of the thermal decomposition of formate species to the band at 1205 cm- 'is implausible, such a band not being observed when formate species alone are thermally desorbed.' Bands located around 1625 cm- ' and F 1700 1500 1300 1100 900 700 wavenumber/cm-' Fig. 1 IR bands from CO adsorption at room temperature by ceria reduced with CO at 873 K, then evacuated at 873 K.After admission of successive doses: 6 pmol g-' (a); 12 pmol g-' (b); in the presence of 0.5 kPa CO (c). A, B, bands due to species A and B; F, formate species. J. CHEM. SOC. FARADAY TRANS., 1994, VOL. 90 wavenum ber/cm --Fig. 2 IR bands from CO adsorption at room temperature by ceria reduced with CO at 873 K, then evacuated at 873 K (cf: Fig. 1). Vacuum thermal desorption of adsorbed species : at room tem-perature (a), at 573 K (b), at 673 K (c)(samples were cooled to room temperature before scanning). at 925 cm- ' also disappear when the sample is heated at 573 K and may tentatively be assigned to oxalate species, other bands for these species being confused within the unresolved spectra.Such species were observed through CO, decomposi-tion on bulk polycrystalline alkali metals with absorptions at 1650, 1350 and 920 ern-'.'' After heating the sample at 673 K [Fig. 2(c)] bands at 1495, 1455, 1380, 1070 and 860 cm-' were nearly the same as those observed through adsorption of CO by unreduced ceria3 and assigned to polydentate car- bonate species. Bands at 1489 and 1390 cm-' were also observed from CO adsorption on reduced ceria5 but these were assigned to carboxylate species. Though the n(CO,) mode at 860 cm-' and the v,(CO,) mode at 1070 cm-' favour carbonate species, contributions of carboxylate species to the bands observed at 1495, 1455 and 1380 cm-' may not be excluded. In order to investigate the effect of using a reducing agent like CO during sample pretreatment, CO was adsorbed by ceria which had been evacuated at 873 K.The resulting spec- trum [Fig. 3(a)] is qualitatively similar to that obtained 8 0.12 C m $ $P m 0.08 0.04 1700 1500 1300 1100 900 700 wavenumber/cm -' Fig. 3 CO adsorption by ceria outgassed at 873 K for 30 min. Spectra in the presence of 0.7 kPa CO: just after admission of CO (a) and after 5 min (b). J. CHEM. SOC. FARADAY TRANS., 1994, VOL. 90 through CO adsorption by ceria which had been reduced with CO at the same temperature (873 K) [Fig. l(c)]. So it may be inferred that outgassing at high temperature is a suffi- cient condition for obtaining surface-reduced ceria.A slight variation of the spectra us. time appears [Fig. 3(b)] mainly involving an increase in intensity of the bands located at 1587, 1293, 1012 and 862 cm-'; these wavenumbers match those observed upon adsorption of CO, by reduced ceria fairly well, and are assigned to a bidendate carbonate species. Bearing in mind that carbonate species, as impurities in ceria, are not completely eliminated during the oxidation step of the sample pretreatment at 673 K, but are decomposed upon evacuation at higher temperature (873 K), this first evacuation was followed by: (i) a re-equilibration of the sample with dioxygen (14 kPa) at 873 K, and then (ii) a second evacuation step at 873 K without further carbonate decomposition. When CO was adsorbed by such a pretreated sample, spectra were in complete disagreement (Fig.4) with those from a sample not re-equilibrated with dioxygen at high temperature (Fig. 3). The observed spectra are very similar to those obtained upon CO, adsorption by unre-duced ceria3 and due to the following adsorbed species: hydrogencarbonate (1612, 1390, 1218 cm-'), bidentate car-bonate (1563, 1296, 1015, 857 cm-'), polydentate carbonate (1456, 1348, 1040, 857 cm-'). The band at 1508 cm-' may be either due to some polydentate carbonate3 or associated with a weak and unresolved band at around 1310 cm-' and then assigned to carboxylate species.' Note that the reduced state of the ceria surface may be distinguished from an unreduced one through the highest wavenumber v(OC0)mode3 of the bidentate carbonate: 1587 cm-' for the reduced state (Fig.3) and 1563 cm-I for the unreduced one (Fig. 4). These results suggest that the decomposition of adsorbed carbonate species, in vacuum and at high temperatures, is a reductive process for the ceria surface. For the sake of generalization of the above results, CO was admitted on a Pd/CeO, catalyst (see the Experimental) that had been reduced with H, at 823 K and evacuated at this same temperature. When a small quantity of CO was adsorbed, the resulting spectrum [Fig. 5(a)] was nearly the same as the parent one [Fig. l(a)]concerning CO adsorption by CeO, pre-reduced by CO. The change of the spectrum upon evacuation of the sample clearly shows a decreased intensity of the two bands A [Fig.5(b)]. This bears out the above-mentioned partial instability of species A under vacuum and the post-formation of species B as well as the other observed species. Note that evacuation at 823 K is suf- ficient to rid the catalyst of the hydrogen involved in the for- mation of formate species, although the catalyst was pre-reduced by H, . 0 1700 1500 1300 I100 900 700 wavenumber/cm-' Fig. 4 CO adsorption by aria outgassed at 873 K for 30 min (cf. Fig. 3) and further oxidized (14 kPa O,), then re-evacuated for 30 min at 873 K. Adsorption of 60 pmol g-' CO: just after admission (a), after 40 min and then evacuation (b). For scaling convenience, intensity of spectrum (b)is reduced to half-size. 0.151 0.121 B iioo 1500 1300 iibo 960 760 waven urn ber/cm-' Fig.5 CO adsorption by Pd/CeO, catalyst reduced with H, at 823 K and evacuated at 823 K. CO was admitted in doses, 6 pmol g-' (a),then evacated at room temperature (b). The reactivity of the species adsorbed over evacuated ceria was investigated through dioxygen adsorption at room tem- perature (Fig. 6).By comparing the spectra before [Fig. qa)] and after [Fig. 6(b)] adsorption of dioxygen we note a quasi- complete disappearance of species A and B upon oxidation. The species leading to the large band at 1465 cm-' has a similar behaviour, the underlying bands at around 1380 and 1500 cm-' from carbonate species being unaffected. The two v,(OCO) and v,(OCO) modes of the formate species are shifted from 1570 and 1362 cm-' to 1545 and 1357 cm-', respectively; moreover, its S(CH) mode at 1372 cm-' becomes active.The large downward shift of the formate v,(OCO) mode and the appearance of the 6(CH)band may be considered as proof of the reduced state of the evacuated ceria surface and of its reoxidation by adsorbing 0, at room temperat~re.~The band at 1600 cm-', which may be tentati- vely assigned to oxalate species, is unchanged. Upon oxida- tion [Fig. 6(b)] new wide bands appear around 850, 1073, 1293 and, much less clearly, 1570 cm-'. They may be assign- ed to the n(C03) and to the three v(C03)modes of carbonate species, which might be thought of as in bidentate geometry ; however, the width of the more clearly isolated bands at 1293 and 850 cm-is larger than usually ~bserved.~ The weak, but very narrow band at 1126 cm-' may be assigned, with a high level of confidence, to superoxide 0, species.'' F Lo FF 0.1 2 8 0.08 C ; f 0.04 0 I 1 1700 15b0 1300 11'00 900 700 wavenumber/cm-' Fig.6 CO adsorption by CeO, outgassed at 873 K for 3 h: admis- sion of 6 pmol g-' CO (a),then 1 pmol g-' 0, (b) Carbonite Nature of the Adsorbed Species A and B The nature of species A and B may be deduced by comparing their spectra with those obtained from the interaction of CO, with alkali metals, products of which were trapped in argon or dinitrogen matrices.I2 When the alkali metal was caesium, two Cs,CO, isomeric species were evidenced, one with C,, and the other with C, symmetry: For each of these species two bands described as v,(OCO) and v,(OCO) modes, with about the same intensity, were observed at 1328.7 and 1186.7 cm-' for the C,, structure, and at 1320.0 and 1050.0 cm-' for the C, geometry.12 From the relative intensity of the twin bands and wavenumber value considerations, these spectra correlate nicely with those due to species A and B, respectively, as described above.When the Cs,CO, species in dinitrogen matrices were con- sidered, a thermal process from the IR source was involved in a geometric rearrangement of the C,, structure into the C, one.', Although species A formed before species B here, such a thermal interconversion of species A into species B induced by an IR source does not seem plausible at room temperature where the thermal energy is already high.By analogy with the Cs,CO, compounds we propose that species A are CO, species adsorbed with a C,, geometry, while the parent species B are of C, structure. In order to bear out these assignments, isotopically substituted species were studied through adsorption of I3CO and Cl80 by either ceria that had been reduced with CO at 873 K or by the Pd/CeO, catalyst that had been reduced with H, at 823 K. The results were qualitatively independent of the sample used, but the isotope wavenumber shift measurements entailed a precision which may be, to a large extent, dis- turbed by a lack of experimental reproducibility due to (i) the envelope nature of the measured bands and (ii) the formation of secondary species.So, the l60 l8O substitution is shown from adsorption of CO by the Pd/CeO, catalyst (Fig. 7), while the 12C -+ I3C substitution shift is evidenced from 0.~~1 0.28 1 u: I I I I 1700 1500 1300 1100 900 760 wavenurnber/crn-Fig. 7 CO adsorption by Pd/CeO, catalyst reduced with H, at 823 K and evacuated at 823 K. Cleo adsorption under a pressure of 0.1 kPa followed by evacuation (a). For comparison, the spectrum from Ci60 adsorption (cf:Fig. 5) is reproduced (b). J. CHEM. SOC. FARADAY TRANS., 1994, VOL. 90 0.08O.l0I 0.06 (0I) 0.02 o! I I I 1 1 1700 1500 1300 1100 900 700 wavenum ber/cm -Fig. 8 CO adsorption at room temperature by CeO, reduced with CO and evacuated at 873 K: 13C0(a),"CO (b) adsorption of CO onto ceria alone (Fig. 8).Wavenumbers for the valence v, and v,(OCO) and bending S(OC0) modes are reported Table 1 for the C,, geometry (species A) and Table 2 for the C, geometry (species B). For comparison, literature values for the related Cs,CO, compounds are also given in the first part of these tables. When Ci80 was adsorbed on ceria the species A then obtained were exclusively C'60'80 and not the isotopomers C1602 or CI8O2, so the following scheme may be suggested for the formation of species A K2") : c =-02-CenL 0:.,'p ce In this scheme, CO interacts with ceria, (i) through its oxygen atom, with a surface oxygen vacancy (a),and (ii) through its carbon atom with an 0,-surface anion.There is, however, a slight discrepancy between the above scheme for the species A and the reported structure for Cs2C02 (C2,). In this last structure (see before), two Cs atoms interact with both the carbon and the oxygen part of CO,, while in the aria surface case Ce interacts only with the oxygen part of CO, . The former structure was found to be slightly more stable based on theoretical studies', and is conceivable for species A as near-surface half-embedded species. However, from the prior appearance of species A through CO adsorp- tion and from the observed partial desorption of these species upon evacuation, it may be inferred that the formation of species A does not have a high activation energy, as the embedded structure of adsorption would suggest.So we discard the embedded structure for species A. Species B, which appears after species A was formed when Table 1 Wavenumbers (cm-') of Cs,CO, species (literature values") and of CO3-species adsorbed onto ceria with C,, geometry species vibration I2CO2 13C0, C'80160 C'80, Cs,CO," v,(OCO) 1329 1295 1318 1305 v,(OCO) 1187 1166 1168 1152 &OCO) 745 738 723 706 CO;-/ceria* v,(OCO) 1317 1290 1310 --v,(OCO) 1150 1134 1140 S(OC0) not observed " C,". Species A. J. CHEM. SOC. FARADAY TRANS., 1994, VOL. 90 CO was adsorbed in doses onto ceria, would be merely a local structural modification of species A. In this hypothesis C'*O adsorption onto ceria (l60)would lead to the C'60180 species, as in the case of species A.However, only the CI6O2 species B were observed. Consequently, species B formation proceeds through several exchanges with the O2-anions of ceria; such a suggested surface mobility of the species, or of their precursor, was also previously observed in the case of the formation of formate species3 Apart from the observed isotopic shift for species A and B, note the I2C-+ I3C isotopic shift for the v,(OCO) and v,(OCO) bands of the formate species. The observed shifts of -44 and -19 cm-' with respect to these vibrational modes are nearly the same as those observed for the formate anion in C,, ge~rnetry,'~consequently such a geometry is inferred for the formate species adsorbed over ceria. Discussion As in the ionic compound (CS+),CO;-,'~ species A and B could be called adsorbed C0;- dianion species.Such ionic species, called 'carbonite~',~~ were evidenced upon adsorp- tion of CO by MgO and CaO previously heated under vacuum at high temperat~res.~9'~*' Carbonite species also formed through CO adsorption at 140 K onto La203 pretreated at 1200 K;16 only the C, geometry was then observed with IR bands at 1393, 983 and 760 cm-'. The wavenumber splitting between the v,(OCO) and v,(OCO) modes is greater for C0;-species (C,) adsorbed over La,03 (1393-983 cm-') than in the case of their adsorption over ceria (1266-1071 cm-', Table 2) or for the parent Cs,CO, species (1320-1050 cm-I), but the mean wavenumer is always around 1180 cm-'.The wavenumber splitting between the v,(OCO) and v,(OCO) vibrations is related to the coupling between the v(C0) vibrations and its variation may be due mainly to a modification of the angle between the two CO bonds.16 Carbonite species were also considered as a possi- bility for assigning bands in the 1050-1200 cm-' spectral range from CO adsorption at room temperature onto Tho, evacuated at 973 K.I7 Two pairs of bands may be seen at 1315-1157 cm-' and 1255-1078 cm-' in spectra5 of CO adsorbed by ceria evacuated at lo00 K, but they were assign- ed as parts of the spectra of carbonate species, in contradic- tion with the present work. Carbonite species are oxygen ~ensitive.~ When dioxygen is admitted to the sample at room temperature (Fig.6), the bands due to species A and B disappear, as well as the band located at 1465 cm-'. This last band is superimposed on car- bonate species bands and its ill-defined profile makes its study by isotopic-exchange difficult. However, the band at 1465 cm-' matches an isolated band at 1475 cm-' of car- bonite species over MgO (T' species in ref. 15) very well or bands at around 1480 cm-'of COT species on CaO.' For the corresponding species a structure was proposed with two uncoupled v(C0) vibrations7 and showing a double-bond Table 2 Wavenumbers (cm-') of Cs,CO, species (literature values") and of COi-species adsorbed onto ceria with C,geometry species vibration 12C02 13C0, Ci802 CS,CO2" Va(OC0) 1320 1280 1290 V,(OCO) 1050 1026 1029 WCO) not observed CO:-/ceriab va(OcO) 1266 1216 -V,(OCO) 1071 1052 -WCO) 773 765 -a C,.Species B. character in the v(C=O) mode at 1480 cm-1,7*15 according to an adsorption geometry such as:15 ,I MM where M is a metal cation. However, the v(C-0) mode, observed at around 850 cm-' for such a species over Ca0,7 is not clearly detected in the present work. Nevertheless, if the band at 1475 cm-' is assigned to such carbonite species, denoted species C, three types of carbonite species are observed over the ceria surface, namely: where O* stands for an oxygen atom interacting with a surface. Carbonite species are formed through a non-redox inter- action process of CO with some 0,-surface sites. In this work, using formate species as redox molecular probes, it may be asserted that the ceria surface is reduced.On the unreduced ceria surface, carbonite species were not observed upon CO ad~orption.~ The sensitivity of the carbonite species to oxidation may explain this, but also special sites would be needed for their formation, and these would presumably be produced only upon evacuation at high temperature. The apparent similarity between a non-reducible basic oxide (CaO, MgO) evacuated at high temperature and ceria, as far as the carbonite species formation is concerned, favours pri- marily the role of the defect sites rather than the reduction state criterion; such defect sites on basic oxides were involved with the isotopic oxygen exchange between CO and the surface, for which only surface atoms with very low coordi-nation were considered, excluding two-dimensional point defects on flat surfaces.l8 When ceria was evacuated at 873 K, and then reoxidized with 0, at 673 K and re-evacuated at 873 K, no carbonite but only carbonate species were observed through CO adsorption (Fig.4), whereas carbonite species were produced after the first evacuation pretreatment (Fig. 3). This suggests that when ceria was only evacuated at high temperature, defect and reduced sites were created by thermal desorption under vacuum of ceria impurities, such as core-carbonate species. These sites were destroyed through re-equilibration of the surface with gaseous dioxygen. Evac- uating the ceria surface cleaned (carbonate-free) in this way at 873 K did not produce this type of reduced site. Basic oxides are easily carbonated under ambient atmosphere, so the role of the decomposition of carbonate species would also be con-sidered in the formation of defect sites on these oxides.When ceria was reduced with H, and then evacuated at temperatures lower than those used here (873 K), not carbon- ite but formate species were observed, although the ceria surface was already reduced. Formate and carbonite species may be involved with the same sites, formate species then being produced, as active hydrogen would be stored in ceria. In order to investigate the possibility of conversion of car- bonite into formate species, dihydrogen was adsorbed by ceria precovered with carbonite species : at room temperature no conversion of carbonite species was observed and upon heating to 373 K their evolution was the same as that observed without dihydrogen.So it may be concluded that dihydrogen is not reactive with Cot- at low temperature (below about 423 K) and that the thermal instability of car- bonite species prevents the observation of a reaction, if any, with thermally activated forms of adsorbed hydrogen. Carbonates are the adsorbed species ultimately observed either upon thermal decomposition (Fig. 2) or upon 0, adsorption at room temperature (Fig. 6). In the former J. CHEM. SOC. FARADAY TRANS., 1994, VOL. 90 process carbonate species look like polydentate core-carbonates while, in the latter, they have a more open adsorption structure such as bidentate geometry.A simple interpretation for this difference would be that the oxygen involved in the COi--+ Cog-oxidation comes from the gas phase upon 0, oxidation, while it has a surface or a sub- surface origin during the thermal process, therefore leading to more tightly bound carbonate species. According to this interpretation of the thermal change for COZ- species, their oxidation by a surface or subsurface oxygen (initially 0,-) would proceed with the reduction Ce4+ -+ Ce3+, in an appar- ent contradiction with the prereduced Ce3+ state of the surface. However, the reduction of ceria is a two-stage process, i.e. a reduction of the surface followed by that of the bulk.Surface-bulk reduction-state equilibration may arise, upon heating, through migration from the bulk to the ceria surface (i.e. bulk reduction) or through reorganization of the surface, bringing about the above contradiction. However, the above interpretation of the formation of car- bonate species, involving the reducibility of the oxide, suffers a lack of generality, since such carbonate species were observed even in the case of irreducible oxides. So, in order to discard any reduction of the oxide, carbonate species may also be produced from disproportionation of adsorbed species:7, co ___+2co;-(c0):-+ c0;-Unfortunately anionic CO polymers, (CO); -, were not clearly observed in the present work; the transient band observed at 1205 cm-'(Fig.2) may be attributed, but not specifically, to such an unstable species. Disproportionation of C0;- into oxalate and carbonate species was also con- sidered.15 When CO was adsorbed by MgO, long exposure at room temperature and a moderate increase of temperature favoured the disproportionation of adsorbed species, leading to Cog- and presumably C (Boudouard reaction)." Such a disproportionation was found to operate on ceria at room temperat~re,~but the present work shows that the major species formed through CO adsorption onto ceria at room temperature is C0;- and not Cog-. Conclusion Adsorbed COZ- was the main species produced when CO was admitted onto ceria which had been reduced and evac- uated at high temperature (873 K).These species were also found to be formed when ceria was only evacuated at high temperature; but in that case the thermal decomposition, under vacuum, of impurity carbonates is thought to be involved with the formation of defect and reduced sites needed for Cog-adsorption. Carbonite species were also observed upon CO adsorption by a reduced and evacuated Pd/CeO, catalyst. Two types of carbonite species, labelled A and B, are clearly evidenced, one with C,, geometry, the other with a C, adsorption structure. Another carbonite species with double- bond character (a C=O bond) is tentatively postulated. Upon heating, the adsorbed species ultimately observed are carbonates. The CO5- structure may be considered to be adsorbed CO, (surface 02-+ CO,) and COi-may be adsorbed CO (surface 0,-+ CO).So the result of the thermal process is an oxidation of CO to CO,, but it could not be assessed whether this overall oxidation proceeds with CO disproportionation or with bulk or subsurface ceria reduction. References 1 C. Li, Y. Sakata, T. Arai, K. Domen, K. Maruya and T. Onishi, J. Chem. Soc., Faraday Trans. I, 1989,85,929. 2 C. Li, Y. Sakata, T. Arai, K. Domen, K. Maruya and T. Onishi, J. Chem. SOC.,Faraday Trans. I, 1989,851451. 3 C. Binet, A. Jadi and J. C. Lavalley, J. Chim. Phys., 1992, 89, 1779. 4 A. Laachir, V. Perrichon, A. Badri, J. Lamotte, E. Catherine, J. C. Lavalley, J. El Fallah, L. Hilaire, F. Le Normand, E. QuCmC- re, G. N. Sauvion and 0.Touret, J. Chem. Soc., Faraday Trans., 1991,87,1601. 5 C. Li, Y. Sakata, T. Arai, K. Domen, K. Maruya and T. Onishi, J. Chem. Soc., Chem. Commun., 1991,410. 6 K. E. Smith and V. E. Henrich, Phys. Rev. B, 1985,32,5384. 7 M. A. Babaeva, D. S. Bystrov, A. Yu Kovalgin and A. A. Tsyga- nenko, J. Catal., 1990, 123, 396. 8 C. Binet, M. Boutonnet-Kizling, A. Jadi and J. C. Lavalley, J. Chem. SOC.,Faraday Trans., 1992,88,2079. 9 A. Jadi, These, Caen, 1990. 10 J. Paul, F. Hoffmann and J. Robbins, J. Phys. Chem., 1988, 92, 6967. 11 C. Li, K. Domen, K. I. Maruya and T. Onishi, J. Am. Chem. Soc., 1989, 111, 7683. 12 Z. Kafafi, R. Hauge, W. Billups and J. Margrave, Znorg. Chem., 1984,23, 177. 13 J. Contreras, J. Mol. Struct., 1986, 147, 301. 14 M. Babaeva and A. A. Tsyganenko, React. Kinet. Catal. Lett., 1987,34,9. 15 A. Zecchina, S. Coluccia, G. Spoto, D. Scarano and L. Mar- chese, J. Chem. SOC.,Faraday Trans., 1990,86,703. 16 A. Tsyganenko, J. Lamotte, J. P. Gallas and J. C. Lavalley, J. Phys. Chem., 1989,93,4179. 17 J. Lamotte, J. C. Lavalley, V. Lorenzelli and E. Freund, J. Chem. Soc., Faraday Trans. I, 1988,81,215. 18 R. Huzimura, Y. Yanagisawa, K. Matsumura and S. Yamabe, Phys. Rev. B, 1990,41,3786. Paper 3/073281; Received 13th December, 1993
ISSN:0956-5000
DOI:10.1039/FT9949001023
出版商:RSC
年代:1994
数据来源: RSC
|
19. |
FTIR study of adsorption and transformation of methanethiol and dimethyl sulfide on zirconia |
|
Journal of the Chemical Society, Faraday Transactions,
Volume 90,
Issue 7,
1994,
Page 1029-1032
Maria Ziolek,
Preview
|
PDF (472KB)
|
|
摘要:
J. CHEM. SOC. FARADAY TRANS., 1994, 90(7), 1029-1032 FTIR Study of Adsorption and Transformation of Methanethiol and Dimethyl Sulfide on Zirconia Maria Ziolek,t Odette Saur, Jean Lamotte and Jean-Claude Lavalley" Laboratoire Catalyse et Spectrochimie, URA CNRS 4 14-ISMRAUniversite, 6,Boulevard du Marechal Juin-14050 Caen Cedex, France An FTIR study shows that methanethiol is adsorbed dissociatively on ZrO, leading to thiolate species and associatively, giving rise to H-bonded and coordinated species, less stable than the CH3S- species. Only coordinated and H-bonded species are formed by dimethyl sulfide adsorption. Heating in the presence of a gas phase shows that, besides the dismutation of CH,SH into (CH,),S and H,S, side reactions occur on the surface leading to methoxy and formate species.The adsorption and transformation of alcohols on metal oxides have been widely studied.'.' Less attention has been devoted to thiols and sulfide compounds, although their adsorption, transformation and decomposition are important for protection of the atmosphere. Recently, the reaction between methanol and hydrogen sulfide, leading to the formation of methanethiol and dimethyl sulfide, has been studied on various metal oxide^.^ Alumina and zirconia were found to be highly active. However, their selectivity was different: the main product on alumina was dimethyl sulfide, whereas on zirconia, a less acidic and more basic catalyst, methanethiol was predomi- nant. Whereas no study has been published on the adsorption of sulfur compounds on zirconia, a few investigations have been reported on alumina.Sugioka et ~l.,~9' study the cracking of ethanethiol and higher alkenic thiols and sulfides on various metal oxides, compared the adsorption of ethanethiol and ethanol on alumina. A comparison between adsorption of methanethiol, dimethyl sulfide and hydrogen sulfide on A1203 has been presented by Saur et d6Mashkina et al.'-' studied the decomposition of methanethiol to dimethyl sulfide on that oxide. The aim of the present study is to apply FTIR spectros- copy to studies of the adsorption and transformation by heating of methanethiol and dimethyl sulfide on zirconia. Experimental Zirconia was prepared from hydrolysis of zirconium pro- pylate. After washing and drying, the precipitate was dried at 400 K and then calcinated at 720 K.The nitrogen BET surface area was 80 m2 g-l. The powder was pressed into discs containing about 10 mg cmF2 for IR studies, which were evacuated at 623 K in the cell for 2h prior to use. FTIR spectra were recorded on a Nicolet MX-1 spectrometer. In the figures presented here the spectrum of the activated cata- lyst has been subtracted. Methanethiol and dimethyl sulfide were pure commercial products. For the adsorption experiments, known amounts were introduced at room temperature. Results The spectrum of the activated ZrO, sample presents two v(0H) bands at 3770 and 3670 cm-' due to residual hydroxy groups, types I and 11, respectively," and v(C0) bands near t Present address: Faculty of Chemistry, A.Mickiewicz Uni- versity, 60-780 Poznan, Poland. 1550, 1480 and 1380 cm-' due to carbonates. This latter impurity explains the negative absorbance in the 1550-1350 cm-' range in the subtracted spectra (Fig. 1). CH,SH Adsorption IR spectra obtained upon introduction of increasing amounts of CH,SH are presented in Fig. 1. Admission of the first portion of CH,SH [Fig. l(a)]does not lead to any band in the v(SH) region (2600-2400 cm-I). The adsorbed species are characterized by v(CH,) bands at 2980 cm-', very weak [va, v;(CH,)], 2924 cm-' [v,(CH,)], 2845 cm-' [2Sa(CH3)], 1437 cm-' [S,(CH,)] and 1314 cm-' [S,(CH,)]. All these bands are close to those observed in the gas phase.'' Another one is very apparent at 959 cm-', at a wavenumber intermediate between those due to the coupled vibrations 6(SH) and q(CH3), observed at 1072 and 802 cm-' in the gas phase." Increasing the amount of CH,SH introduced leads to the appearance of a v(SH) band near 2500 cm-' [Fig.l(b)] and to another one, at 1065 cm-', readily apparent under an equilibrium pressure of CH,SH. A broad band near 3300 cm- is also noted. Simultaneously the inten- sity of the v(0H) bands at 3770 and 3670 cm-', due to residual hydroxy groups, tends to disappear. Below 900 cm-',the zirconium oxide is very poorly trans- parent to the IR beam. However, a significant band near 850 cm -' appears, which could be assigned to associated OH group vibrations [S(OH)].The species which are formed upon CH,SH adsorption on ZrO, tend to disappear upon evacuation, even at room tem- perature [Fig. l(c)]. Increasing the evacuation temperature (up to 423 K) leads to a decrease of the intensity of all infra- red bands, whereas the v(0H) bands recover their intensity. 3800 3300 2800 2300 1800 1300 800 wavenumber/cm-' Fig. 1 IR spectra of adsorbed CH,SH: (a) 50 pmol g-' added; (b) in the presence of a gas phase (ca. 130 Pa) in the cell; (c)and then evacuation at room temperature These results suggest the formation of two types of species. One, more strongly adsorbed, is characterized by a strong band at 959 cm-' and the lack of any band in the v(SH) region. These two features prove that dissociative adsorption occurs.In particular, breaking the S-H bond suppresses the coupling between the 6(SH) and rll(CH3) modes. Only the rl,(CH,) vibration is then expected. The 959 cm- ' wavenum-ber is, in agreement, close to that found for CH3SD in the gas phase (1007 cm-'). It is difficult to characterize the new OH groups formed from the dissociative adsorption : II+ CH3SH-SH -Zr-0--Zr-0-They could lead to a broad band near 3600 cm-' [Fig. l(c)] which, as in the case of alcohol dissociation on alumina, would mean that a weak interaction persists between the new OH groups and the methylthiol species formed: -Zr-0-The lack of transparency of zirconia below 800 cm- ' pre-vents the detection of the v(C-S) band expected near 700 cm-I.By analogy with results obtained with CH,OH," two CH,-S species would be expected, mono- (type I) and bi- dentate (type 11) CH3 y3I S S /\I -Zr--Zr Zr-I I1 Unfortunately, the v(CH,) and 6(CH3) vibrations are not sensitive enough to the type of species formed and only observation of the v(CS) region would give information about their occurrence. The more weakly bonded species, reversibly adsorbed at room temperature, are characterized by broad bands at about 2500 cm-' [v(SH)] and 1065 cm-' [G(SH), rll(CH3)], showing that they are associatively adsorbed. The concomi- tant perturbation of the hydroxy groups, especially the most basic one," suggests the following adsorption mode CH3SH. H-_ /0' I -Zr-Other species such as coordinated (A) and hydrogen-bonded (B) 5343 GHYH S IS HI species cannot be excluded, as proposed for alumina.6 Since ZrO, is less acidic and more basic than Al2O3,l2 the occurrence of the latter species is highly probable.The forma- tion of several types of associatively adsorbed species could well explain the broadness of the v(SH) band. (CH,),S Adsorption Spectra resulting from introduction of (CH3),S, at room tem- perature, on activated zirconia are shown in Fig. 2. Bands J. CHEM. SOC. FARADAY TRANS., 1994, VOL. 90 O.*l I 13800 3400 3000 2600 1500 1300 1100 900 wavenumber/cm-' Fig. 2 IR spectra of adsorbed (CH,),S: (a) 100 pmol g-' added; (b) in the presence of a gas phase (ca. 130 Pa) in the cell; (c) and then evacuation at room temperature appear at ca.2995 [shoulder, v:(CH,)], 2983 [v,(CH,)], 2921 [v,(CH,)], 2850 [weak, 26(CH3)], 2834 [26(CH3)], 1432 CSACH3)I, 1308 C6JCH,)I, 1032 Cr,l(CHJI, 981 Crl(CH3)Iand 914 cm-' [weak, rl,(CH3)]. Their wavenumbers are very close to those observed in solution.' Their appearance affects the residual hydroxy groups: the 3770 and 3670 cm- ' band intensity tends to decrease whereas a broad band is noted at about 3400 cm-'. Evacuation at room temperature partly restores the surface OH bands and partly removes all of the other above-mentioned bands [Fig. 2(c)]. They com- pletely disappear after evacuation at 423 K. Since (CH,),S can act only as a base, two types of species are expected, according to whether the adsorption sites are Lewis or Brnrnsted acidic in nature.After evacuation at room temperature, most of the persisting species are coordinatively chemisorbed : +-Zr-since the surface OH groups are partly restored. Compared to results obtained on alumina,6 these species are more weakly held, as indicated by the lower v,(CH,) wavenumber (2921 cm-' instead of 2935 cm-'), in agreement with the weaker Lewis acidity of ZrO, relative to Al,03 .I2 The formation of other species at higher coverage involves the perturbation of surface residual OH groups, implying that they are H-bonded to these hydroxy groups. Type I1 OH groups are expected to be more acidic than type I, we there- fore assign the more perturbed band at 3400 cm-' to species (C): C\H3,CH3 S C D The occurrence of a broad band near 3600 cm-' at the expense of the intensity of the 3770 cm-' band suggests that interaction with type I OH groups also occurs (D).(CH3),S Transformation Introduction of 1.3 x 10' Pa of (CH3),S onto activated ZrO, gives rise to a spectrum to that already presented in Fig. 2(b). Heating to 523 K does not change the spectrum [Fig. 3(a)]. J. CHEM. SOC. FARADAY TRANS., 1994, VOL. 90 o.8 1 N 0.6 I I 1 3100 2900 27001500 1300 1100 900 wavenumber/cm-Fig. 3 IR spectra of (CH,),S species after heating in the presence of (CH,),S gas: (a) at 523 K; (b) then at 623 K After heating at 623 K, a new band is observed at 1150 cm- ', whereas shoulders tend to appear near 2820 and 1050 ern-'.Furthermore, the intensity of a broad band at about 950 cm -'tends to increase slightly. The 1150 cm-' band can be assigned to methoxy species since CH30H adsorption on ZrO, led to the appearance of strong IR bands at 2925, 2818, 1162 and 1060 cm-'. Such an assignment is in agreement with the occurrence [Fig. 3(b)] of shoulders near 2820 and 1050 cm-'. As for the 950 cm-' band, it could be due to a small amount of SCH, species. However, it is difficult to confirm this assignment since the other bands due to the formation of these species, in the v(CH,) and 6(CH,) frequency range, are quite similar to those observed when using (CH3),S. In other words, the wavenum- bers of v(CH3) and 6(CH,) are not sensitive to the sulfur atom substitution. CH,SH Transformation Fig.4 shows the IR spectra of species formed after admission of 1.3 x 10, Pa of CH,SH onto ZrO, at room temperature, and heating to 423 [Fig. 4(a)],523 [Fig. 4(b)] and 623 K [Fig. 4(c)]. Heating to 423 K causes a decrease in the inten- sity of the band at 957 cm-' characteristic of methanethiol chemisorption in the form of thiolate species, whereas new bands appear notably at 1032, 980 and 914 cm-'. They are characteristic of dimethyl sulfide formation. Four other bands at 2884, 1574, 1385 and 1370 cm-' are also observed. They can be assigned to formate species on ZrO,, since they are similar to those observed by He and Ekerdt14 from CO, adsorption on zirconia. Increasing the temperature to 523 K [Fig.4(b)] increases the intensity of bands due to adsorbed (CH,),S and formate I I 3100 2700 2300 1700 1500 1300 I 1 1100 1 900 wavenumber/cm- Fig. 4 IR spectra of CH,SH adsorption and then heating in pres-ence of CH,SH gas: (a) 423 K; (b) then at 523 K; (c) then at 623 K species. It also clearly gives rise to a new band at 1150 cm- ', very close to that mentioned above [Fig. 3(b)] and assigned to methoxy species. Its intensity increases with those due to formate species. Furthermore, a broad band in the v(SH) region (ca. 2500 cm-') is registered. Heating to 623 K [Fig. qc)] gives a spectrum quite similar to that observed after (CH,),S adsorption (Fig. 2). The bands due to formate species are then very weak; the broad band at 2500 cm-' persists, but it is easily removed by evacuation at room tem- perature.Discussion Surface species formed from CH,SH and (CH,),S adsorption on zirconia, at room temperature, are similar to those pre- viously observed on Al,03 .6 The transparency of zirconia between 10oO and 800 cm-' allows us to discriminate well between CH,SH dissociative and associative adsorption. In the present work, attention has been paid to the surface species formed upon heating the adsorbates in the presence of a gas phase at increasing temperature. By heating (CH,),S adsorbed on ZrO,, new bands appeared in the IR spectra which were assigned to methoxy species. On the basis of these assignments, one can propose dissociation into CH30 and CH,S species by heating (CH,),S on activated ZrO, .Such a dissociation has already been observed on highly activated silica.' 5916 0 mH3 SCH3 /\ I I Si Si +CH,SCH, -Si Si Note that the 1150 cm-' band is close to that characteristic of type I methoxy species:" 7H3 0 I -Zr-Their formation supposes that the methyl group resulting from the (CH3),S dissociation reacts with monodentate O2-surface groups on the activated zirconia. To our knowledge, there are no data in the literature on the surface oxygen arrangement. Note that this site can result from surface dehy- droxylation with the departure of a proton from type I OH groups. The formation of formate species arises from CH3SH decomposition (Fig. 4), but not from (CH3),S (Fig. 3).Neither does it arise from methanol or methoxy species decomposi- tion, as shown when heating CH,OH over ZrO, : no formate species were detected, even after heating at 623 K., We there- fore conclude that formate formation arises from decomposi- tion of CH,SH reversibly adsorbed on ZrO, , irreversibly adsorbed species do not give rise to HCO, (Fig. 1). One possibility is the intramolecular H,S elimination from CH,SH: the CH, radical formed would be chemisorbed on basic 0,-surface sites, possibly in the form of a dioxymethylene species. The disproportionation of the latter would give rise to both formate and methoxy species, which indeed appear at the same time. Formate and methoxy species are not stable after heating to 623 K [Fig. 4(c)],whereas it was found that the methoxy species arising from methanol decomposition were quite stable under such condition^.^ It is possible that, at this tem- perature, the methoxy species react with CH,SH leading to (CH,),S, thus explaining the disappearance of the methoxy species.However, the main pathway to form dimethyl sulfide is the following reaction: 2CH,SH +(CH,),S + H,S The formation of H2S explains the presence of a broad band near 2500 cm-' (Fig. 4). The formation of (CH,),S requires breaking of the C-S bond of one CH,SH molecule and breaking of the S-H bond of the other. It has been shown, on silica-support catalysts, that C-S and H-S bond dissociation occurs easily at low temperature. By analogy with dimethyl ether formation from methan01"~'~ we propose the following mechanism: CH3SHI HIHSI /\-0-Zr Zr-+CH,SCH, CH,SCH, +H2S +-0-Zr 0 Zr-Conclusions IR spectroscopy can easily distinguish between species formed from CH,SH or (CH,),S adsorption on zirconia.Methanethiol is mainly dissociatively chemisorbed and leads to CH,-S species, well characterized by a band at 959 cm-'. Dimethyl sulfide gives mainly coordinated and H-bonded species, leading to bands at 1032 and 982 cm- 'in this char- acteristic frequency range. Heating in the presence of a gas J. CHEM. SOC. FARADAY TRANS., 1994,VOL. 90 phase shows that CH,SH dismutes into H,S and (CH,),S even at 423 K. Surface side reactions are evidenced, leading to the formation of methoxy species, whatever the starting sulfur organic compound, and formate from CH3SH decom- position.The authors acknowledge D. Dudko and D. Dziurka for help with the experimental work. References 1 C. D. Chang, Catal. Rev., 1983,25,1. 2 V. Moravek and M. Kraus, J. Catal., 1984,87,452. 3 M.Ziolek, J. Kujawa, 0. Saur and J. C. Lavalley, J. Phys. Chem., 1993,97,9761. 4 M.Sugioka and K. Aomura, Int. Chem. Eng., 1974,13,755. 5 M. Sugioka, T. Kamanaka and K. Aomura, J. Catal., 1978,52, 531. 6 0.Saur, T. Chevreau, J. Lamotte, J. Travert and J. C. Lavalley, J. Chem. Soc., Faraday Trans. I, 1981,77,427. 7 A. V. Mashkina, V. R. Grunvald, V. I. Nasteka, B. P. Borodin, V. N. Yakovleva and L. N. Khairulina, React. Kinet. Catal. Lett., 1990,41, 357. 8 A. V. Mashkina, S. N. Koshelev, V. N. Yakovleva, V. R. Grun- vald, B. P. Borodin and V. I. Nasteka, React. Kinet. Catal. Lett., 1991,43,381. 9 A. V. Mashkina, V. R. Grunvald, B. P. Borodin, V. I. Nasteka, Y. N. Yakovleva and L. N. Khairulina, React. Kinet. Catal. Lett., 1991, 43, 361. 10 M. Bensitel, V. Moravek, J. Lamotte, 0.Saur and J. C. Lavalley, Spectrochim. Acta, Part A, 1987,43, 1487. 11 0.Saur, J. Travert, J. C. Lavalley and M. Chabanel, C.R. Acad. Sci. Paris, 1980,29lC, 227. 12 J. C. Lavalley, Trends Phys. Chem., 1991,2,305. 13 J. P.Perchard, M. T. Fore1 and M. L. Josien, 3. Chim. Phys., 1964,61,645. 14 M. Y. He and J. G. Ekerdt, J. Catal., 1984,87,381. 15 K. Klostermann and H. Hobert, J. Catal., 1980,63,355. 16 C. H.Rochester and R. J. Terrel, J. Chem. SOC., Faraday Trans. I, 1977,73, 596. 17 X. Montagne, R. Boulet, E. Freund and J. C. Lavalley, Stud. Sutf: Sci. Catal., 1989,48,695. 18 H.Knozinger, A. Scheglila and A. M. Watson, J. Phys. Chem., 1968,72,2770. Paper 3/06930C; Received 22nd November, 1993
ISSN:0956-5000
DOI:10.1039/FT9949001029
出版商:RSC
年代:1994
数据来源: RSC
|
20. |
CoAPO molecular sieve acidity investigated by adsorption calorimetry and IR spectroscopy |
|
Journal of the Chemical Society, Faraday Transactions,
Volume 90,
Issue 7,
1994,
Page 1033-1039
Jochen Jänchen,
Preview
|
PDF (966KB)
|
|
摘要:
J. CHEM. SOC. FARADAY TRANS., 1994, 90(7), 1033-1039 1033 CoAPO Molecular Sieve Acidity investigated by Adsorption Calorimetry and IR Spectroscopy Jochen J8nchen,* Mart P. J. Peeters, Jos H. M. C. van Wolput, Jillus P. Wolthuizen and Jan H. C. van Hooff Schuit Institute of Catalysis, Eindhoven University of Technology,P.O. Box 513,5600MB Eindhoven, The Netherlands Ursula Lohse Centre of Heterogeneous Catalysis, Rudower Chaussee 5,042489Berlin-Adlershof, Germany The acidic properties of cobalt- and silicon-substituted AIPO-5, -11 and -44 have been characterized by adsorp- tion calorimetry and IR measurements. Adsorption calorimetric measurements indicate that the adsorption potential of the samples for acetonitrile is enhanced upon cobalt incorporation. The calorimetrically measured heats of adsorption indicate the formation of strong acid sites, due to this cobalt incorporation, as well as the presence of weaker acid sites, probably terminal P-OH groups.IR measurements of adsorbed ammonia and acetonitrile indicate that some Brsnsted-bonded ammonia and acetonitrile are present, in combination with Lewis-bonded species. The Brsnsted-bonded species presumably originate from interaction of the base with the weakly acidic P-OH groups. The Lewis-bonded species are due to the coordinative interaction of the lone pair of the base with framework cobalt ions, Crystalline aluminophosphates, silicoaluminophosphates and widely used in the characterization of acidity, this base has isomorphously substituted aluminophosphates represent a been included in the present study.new class of molecular sieves which are considered as poten- tial catalysts and adsorbents. '-' The partial replacement of Ps+ by Si4+ results in a negative charged framework. This Experimental charge deficiency may be compensated by protons, giving All samples used were prepared hydrothermally in PTFE- bridging hydroxy groups as in alumosilicates. Several papers lined static autoclaves at 463 or 473 K for 4-48 h using an have appeared dealing with the acidic properties of these so-organic templating agent. The synthesis procedure generally called SAP0 molecular followed those given in the patent The molar Aluminophosphates containing bivalent cations such as compositions of the reaction gels of the Co-modified + +Cot on framework positions replacing A13 (MeAPOs) also aluminophosphates and the Co concentration of the products give rise to the formation of acidic sites.Some related contri- are given in Table 1. Pseudoboemite (Condea) for the AFI butions dealing with the characterization of such Co-and AEL systems or aluminium isopropoxide in the case of substituted AlPOs by spectroscopic methods, TPD and structure 44, Co acetate or CoSO,, phosphoric acid (85%) catalytic tests have appeared.' '-I4 However, the nature of the and the appropriate organic component were added suc-corresponding centres is still not clear. The expected bridging cessively with high shear mixing. SiO, sol was used in the hydroxy groups have not yet been directly observed by 1R.l' reaction mixture for the synthesis of Si-containing samples.In the present work acidic sites in A1P04-5, AlPO,-11, X-Ray diffraction and adsorption measurements on all CoAPO-5, CoAPO-11 and CoAPO-44 in comparison with samples used showed them to be highly crystalline and pure. the corresponding SAPOs and CoAPSO-44 are investigated The adsorption calorimetric investigations were performed with IR spectroscopy and adsorption calorimetry.' at 303 K on a SETARAM microcalorimeter of Calvet type Ammonia and acetonitrile have been used as probe molecules connected to a standard volumetric adsorption apparatus. for these experiments. Acetonitrile is an attractive probe mol- The equilibrium pressure was determined by a high-precision ecule since it allows the investigation of both Lewis and membrane manometer (MKS Baratron). Nitrogen isotherm Brrzrnsted acidity.The stretching mode of the protonic sites of measurements were carried out on a Sorptomatic 1900 the molecular sieves shifts and the CZN stretching vibra- facility from Carlo Erba. Prior to the adsorption, the samples tions is influenced by interaction with the base. Characteristic were carefully activated in high vacuum at 670 K for 14 h. shifts of the v(CN) modes can be used to discriminate IR spectra were measured on a Bruker FTIR spectrometer between Brnrnsted and Lewis sites as well as between Lewis (IFS 113v) equipped with a vacuum cell. Self-supporting discs centres of different chemical nature.' Because ammonia is with a thickness of 6 mg cmP2were used.The spectra were Table 1 Molar compositions of the reaction mixtures for the CoAPOs, the CoAPSO-44 and the SAPO-44 and the Co concentration of the products in wt.% CoAPO-5/1 0.98 1 0.04 0 1.5 TEA" 40 CoAPO-5/2 0.96 1 0.08 0 1.5 TEA 40 CoAPO-1 1 0.98 1 0.04 0 1 DPA' 40 0 0 0 1.2' 2.3' 1.1* CoAPO-44 0.80 1 0.40 0 1 CHAd 50 1 8.7' CO APSO-44 1 1 0.40 0.40 1 CHA 50 1 6.1' SAPO-44 1 0.85 0 0.40 1 CHA 50 1 0 Triethylamine. 'From AAS. Diisopropylamine. Cyclohexylamine. From wet chemical analyses. recorded by co-adding 500 scans at room temperature with a resolution of 1 cm-'. Activation of the samples was per-formed at 720 K in high vacuum for 1 h. After the sample had been cooled to room temperature the spectrum of the unloaded sample was taken, followed by contacting the sample with deuteriated acetonitrile or ammonia for 30 min at a pressure of ca.0.9 or 0.4 Torr, respectively, then another spectrum was recorded. After this the loading of the samples was reduced stepwise by lowering the equilibrium pressure to 0.04 Torr (acetonitrile) and by desorption (15 min) at room temperature. The last two spectra were recorded after desorp- tion for 1 h at 353 and 573 K, respectively. The desorption of ammonia was carried out at room temperature (15 min) and at temperatures of 423,473, 523 and 573 K for 1 h. Results and Discussion Adsorption Measurements To obtain more insight into the adsorption equilibrium of the acetonitrile on AFI and AEL molecular sieves and their Co derivatives, the adsorption isotherms are discussed first.They are depicted in Fig. 1 and 2. The incorporation of Co into the lattice enhances the adsorption potential of the samples at low adsorption values. At constant equilibrium pressure, fol- lowing the dashed lines for instance, the loading of aceto-nitrile increases with the Co content of the samples. The dashed lines also indicate the loadings used for the IR experi-ments. Fig. 1 and 2 also show that at the same pressure AEL -2.00 1.50 E 1.oo 0.50 0.00 0.01 0.1 1 10 100 pressureflorr Fig. 1 Adsorption isotherms of acetonitrile on AFI-type samples measured at 303 K; A, A1P04-5; H, CoAPO-5/1; 0,CoAPO-5/2; dashed lines indicate the pressure and the loading of the IR experi-ment (first and second spectra in Fig.8) 3.00 2.50 -2.00 -0 1.50 E \ rp 1.oo 0.50 0.00' ' """" '"c-sc" ' """" ' """ 0.0 1 0.1 1 10 100 pressu re/Torr Fig. 2 Adsorption isotherms of acetonitrile on AEI molecular sieves at 303 K; 0,AlPO,-ll; D, CoAPO-11; dashed lines also due to equilibrium conditions of the first and second spectra in Fig. 7 J. CHEM. SOC. FARADAY TRANS., 1994, VOL. 90 Table 2 Adsorption capacities of acetonitrile, methanol (at relative pressure 0.5) and nitrogen (at relative pressure 0.4) on some AlPOs and CoAPOs adsorption capacity/cm3 g- sample ace toni trile methanol nitrogen AlPO4-5 0.12 0.14 0.11 CoAPO-5/1 0.1 1 -0.08 CoAPO-5/2 0.12 -0.1 1 AlP0,- 1 1 0.09 -0.09 CoAPO- 1 1 0.09 -0.08 CO APO-44 --0.16 COAPSO-44 --0.16 SAPO-44 --0.14 adsorbs more acetonitrile than AFI.This is due to the smaller pore diameter of the AEL structure.20 This has to be taken into account if the adsorption heats of both types of molecular sieves are compared. The adsorption capacities also can be evaluated from the isotherms. A reliable value of the adsorption capacity is obtained using the amount adsorbed at a relative pressure of 0.5, to prevent contributions of capillary condensation. Table 2 lists results obtained with acetonitrile, methanol and nitro- gen. The adsorption capacity of methanol on A1P04-5 is close to the ideal pore volume (0.147 cm3 g-I). The other adsorbates give somewhat lower capacities, as reported in the literature. The adsorption capacities found for AEL are close to the theoretical pore volume of 0.09 cm3 g-', confirming the high crystallinity of the samples. The lower adsorption capacity of CoAPO-5/1 could be due to pore blockage by some non-framework material in the one-dimensional pore system since good crystallinity is found in the XRD measure-ments.Fig. 3 and 4 show the calorimetrically measured differential heats of adsorption of acetonitrile on AFI, AEL and their Co derivatives. The heat curves of all samples decrease contin- uously with increasing amount adsorbed except for a small increase for AFI at higher loadings. The heat curves exhibit at least two steps (CoAPO), indicating the existence of acid sites of different strengths.Adsorption calorimetric measure- ments of acetonitrile on silica gelz1 which has only silanol groups and no Lewis sites showed that heats >60 kJ mol- do not occur. On the other hand, adsorption experiments on zeolite HY gave heats of chemisorption of 80 kJ mol-' for acetonitrile, characteristic of the H-complex of acetonitrile on Bransted-acid sites. After careful shallow-bed dehydroxyla- 120 t 100 7-80 E 7 60 c1 40 20 0.00 0.50 1.oo 1.50 2.00 2.50 a/mmol g-' Fig. 3 Differential molar heats of adsorption of acetonitrile on AFI molecular sieves as a function of the amount adsorbed; arrows indi- cate the position at the heat curve where IR spectra 1 and 2 in Fig.8 were recorded; symbols as in Fig. 1 J. CHEM. SOC. FARADAY TRANS., 1994, VOL. 90 120 I I 100 a r 180E 7$ 60 40 ' 20 I I I 1 I 0.00 0.50 1.oo 1.50 2.00 2.50 almmol g-' Fig. 4 Differential molar heats of adsorption of acetonitrile on AEL molecular sieves as a function of the amount adsorbed; arrows indi- cate the position at the heat curve where IR spectra 1 and 2 in Fig. 7 were recorded; symbols as in Fig. 2 tion of this HY at 870 K the adsorption heat for the strongest sites amounted to ca. 90 kJ mol-'. This was most probably due to Lewis sites created by dehydroxylation.22 Based on the observed heats of adsorption it can be concluded that the acid sites found on CoAPO-5 (100 kJ mol- ') and CoAPO-11 (110 kJ mol-') are stronger than the bridging hydroxy groups or the Lewis sites in HY.They are present at the levels of 0.1 mmol g-' for CoAPO-5/1, 0.2 mmol g-' for CoAPO-5/2 and 0.15 mmol g-' for CoAPO-11. The second step in the heat curves of the CoAPOs, beginning at 70 kJ mol-' (CoAPO-5) and 80 kJ mol- ' (CoAPO-1 l), respec- tively, is due to medium-strong sites, probably P-OH, which are known to be weaker than bridging OH but stronger than silanols. No OH other than the P-OH groups could be detected in the CoAPOs by IR measurements (see below). The concentrations of these medium-strong sites can be determined from the heat curves and amount to 0.15 mmol g-' for CoAPO-5/1, 0.2 mmol ggl for CoAPO-5/2 and 0.25 mmol g- for CoAPO-11. P-OH sites also occur in pure AlPOs, which could explain the smaller amount (0.2 mmol g-') of stronger bonded acetonitrile in these samples.It is interesting that the complete heat curves of the AEL molecular sieves (Fig. 4) seem to be shifted towards higher adsorption heats by ca. 10 kJ mo1-' compared to those of the AFI (Fig. 3). This is due to different contributions of the dispersion interaction caused by differences in the pore size. The same effect is found for the adsorption of hydrocarbons on AEL and AFI, e.g. for propane, a molecule of similar size but n~n-polar.~',~~ Therefore the differences of the heats of chemisorption mentioned above between CoAPO-5 and CoAPO-11 do not only originate from differences in acid strength.Secondly, the heat of physisorption of acetonitrile on A1P04-5, is 40 kJ mol-' (see Fig. 3) which is only 6 kJ mol-' higher than the heat of condensation. This difference amounts to 20 kJ mol-' for methanol and 35 kJ mol-' for ammonia, indicating a much higher specific term of the adsorption heat arising from a coordinative bond to frame- work Al. This could be verified by 27Al DOR NMR measure- ments on samples loaded with ammonia, methanol and a~etonitrile.~~Ammonia adsorption leads to the formation of large amounts of five- and six-coordinated Al, whereas only tetrahedrally coordinated A1 is observed after adsorption of acetonitrile. Hence coordinatively bonded acetonitrile on lattice A1 in the AFI and AEL structures can be excluded for the interpretation of the IR spectra, in contrast to what is reported in ref.11. From the calorimetric measurements on the CoAPO and AlPO, samples it follows that incorporation of Co results in the formation of strong acid sites. However, no exact infor- mation can be obtained regarding the character of these centres from the calorimetric measurements. Hence, suitable IR investigations are necessary to identify these sites. There- fore IR spectra have been recorded beginning at the aceto- nitrile loadings marked by the arrows in Fig. 3 and 4, but first ammonia was used to discriminate between Lewis and Brnrnsted acidity. IR Measurements A commonly used technique for studying the acidic proper- ties of zeolites with IR is ammonia or pyridine adsorption, which may result in ammonium ions (1450 cm-') or in coor- dinatively bonded ammonia (Lewis sites, 1610 cm- I).Inte-gration of the peak areas gives information on the amount of both types of acid site if the molar absorption coefficients are known, but will not reveal the strength of the sites. Apart from calorimetric measurements, differences in strength can be revealed by measuring IR spectra after desorption of the samples at increasing temperatures. The results of these mea- surements are displayed for a series of AlPO, CoAPO and SAP0 samples in Fig. 5 and 6. Fig. 5 shows that the amount of Brernsted-bonded ammonia is highest for SAPO-5. The amount for SAPO-11 is almost half that of SAPO-5. This is in agreement with the ratio of the concentrations of Brernsted sites in SAPO-5 (0.6 mmol g-') to SAPO-11 (0.3 mmol g-') estimated by adsorption calorimetric measurements with .-m C2 25 E 20B v c 15 n6 10 w =52 m n Fig.5 Relative concentration of Brmsted-bonded ammonia on AFI and AEL molecular sieves estimated by integration of the IR band at 1450 cm-' at different loadings (adjusted by stepwise desorption after adsorption at 0.4 Ton) Fig. 6 Relative concentration of the Lewis-bonded ammonia on AFI and AEL molecular sieves estimated by integration of the 1610 cm-IR band of ammonia at different loadings (adjusted by stepwise desorption after loading at 0.4 Torr J. CHEM. SOC. FARADAY TRANS., 1994, VOL.90 Table 3 OH stretching vibrations and their shift after adsorption of acetonitrile (equilibrium pressure 0.04 Torr) sample Si-OH ________~ AlP0,-5 --CoAPO-5/1 CoAPO-5/2 -SAPO-5 3743 AlP0,-11 --CoAPO-1 1 SAPO-11 3743 COAPO-44 -COAPSO-44 3739 SAPO-44 3746 HY(18)' 3738 " Relative amount of P-OH. v(0H) before adsorption/cm -v(OH) shift after adsorption/cm -P-OH Bransted OH Si-OH P-OH Bronsted OH 3677 0.95" --610 3677 0.8 --610 3677 1.2 --590 3678 0.25 3626 3600 310 620 3677 0.87 --610 3677 1.8 --600 3678 0.85 3631 -620 3677 1.34 --555 -3675 0.3 3628 3593 -3676 0.4 3625 3600 ---3629 3550 320 Value for the centre of gravity of the two pseudo-bands. Si :A1 = 18. ammonia.25 The quantity of Bransted-bonded ammonia on the AlPOs and CoAPOs is significantly lower than in SAPO-5.Moreover, the bonded ammonia can be desorbed much more easily. This amount correlates for the AlPO and CoAPO samples with the relative P-OH concentration given in Table 3 (column 4). Therefore the comparable quan- tities in CoAPO-11 and SAPO-11, which has bridging hydroxy groups, can be explained in terms of the rather high concentration of P-OH in CoAPO-11. The amount of Lewis-bonded ammonia is highest for the CoAPOs. More- over, the strength of these sites is clearly higher (Fig. 6) than for the other samples, confirming the results of the calorimet- ric measurements. However, as mentioned before, acetonitrile is a better base for discriminating acid sites by IR measure-ments.The IR difference spectra of CD,CN-loaded AEL type molecular sieves are shown in Fig. 7. Whereas in Fig. 7(a) three spectra of A1P04-11 with decreasing amounts of the probe are depicted, Fig. 7(b) and 7(c) show the spectra of CoAPO-11 and SAPO-11 with four different loadings. The broad band observed at ca. 3100 cm-l [Fig. 7(a)] is due to the stretching vibration of the P-OH group (3677 cm-') after interaction with acetonitrile. This is a broad band because of resonance of the perturbed OH stretching mode with the vibrations of the base against the OH, v(0H) In~~v(OH-CD,CN).~~the region above 2400 cm-' changes in the zeolite OH stretching vibrations can be mea- sured, whereas below 2400 cm- l changes in the vibrations of the acetonitrile are observed.A number of bands can be detected in the range 2265-2320 cm-l due to the C=N stretching mode of the sorbed acetonitrile. The band at 2114 cm-' is due to the symmetric vibration of the CD, group.27 The position of this band is shifted by a few cm-' compared to the gas-phase value due to a small interaction of the mol- ecule with the zeolite framework. The weak band at 2250 cm-'is due to the asymmetric vibration of the CD, group of acetonitrile.'' The band at 2265 cm-' originates from the CEN stretching vibration of physisorbed acetonitrile. The frequency of this band is close to the frequency of the free acetonitrile. Acetonitrile is quite easily desorbed, since decreasing the pressure from 0.9 to 0.04 Torr at room tem- perature results in a drastic decrease of its intensity (see Fig.2 and 4). The band at 2281 cm-' is assigned to CEN stretch- ing after interaction with P-OH groups. As this band appears, the P-OH band at 3677 cm-l disappears and vice uersa, indicating a correlation between these bands. The shift of v(CrN) is very close to that observed in zeolite Y and ZSM-5 after sorption on Si-OH (2278 cm-1).29 The some- what higher wavenumber is in agreement with the higher strength of the P-OH group compared to the Si-OH 0.40 nv) 4-.-5 0.30 ti 2265 1+ W Q) 0.20 0 (Ps2 0.10 .o 0.00 1500 2000 2500 3000 3500 4000 wavenumber/cm-' 1 2309 22651I 3 0.30+ W a) 0.20 C 2 g 0.10 9 0.00 1500 2000 2500 3000 3500 4000 wavenumber/cm-0.40 nv) .-+5 0.30+ W Q) 0.20 0 m 2290 I ve2 0.10 I 9 0.00 1500 2000 2500 3000 3500 4000 wavenurnber/cm-' Fig.7 IR difference spectra of CD,CN-loaded AEL molecular sieves:(a)AlP0,-11, (b)CoAPO-11, (c)SAPO-11; 1,0.9 Torr; 2,O.M Torr; 3, desorption at 295 K; 4, desorption at 353 K J. CHEM. SOC. FARADAY TRANS., 1994, VOL. 90 group. The bands between 2312 and 2320 cm-' are assigned to acetonitrile interacting with some non-framework alu-minium (Lewis acids). The shift should be compared with non-framework aluminium in zeolite Y (2323-2332 cm-').30,31 Differences in the observed shifts may be due to differences in the coordination number of this non-framework aluminium. Fig.7(a)(and 2) also illustrates the weak inter- action of acetonitrile with the AlPO-11 framework. All ace- tonitrile can be desorbed by degassing the sample at 295 K for 15 min. Fig. 7(b)shows the same spectra for the CoAPO-11 sample. The bands of P-OH groups interacting with acetonitrile (3100 cm-') and the vibrations of the CD, group of the ace- tonitrile (21 14 and 2250 cm- ') are observed at the same posi- tions as in the AlPO-11 sample. In the C=N region the same peaks are observed as in the non-substituted sample, together with a strong band at 2309 cm-'. This band should be ascribed to Lewis-bonded acetonitrile on framework cobalt because no other band than the P-OH can be detected in the v(0H) region. There is no indication of other OH such as Al-OH or Co-OH which could also interact with aceto- nitrile. On the other hand the v(CGN) shift is too high to be due to Brsnsted-acid sites since a characteristic shift of ace- tonitrile bonded to a Brsnsted-acid site is between 2290 and 2300 cm111.26 The Lewis-bonded acetonitile on the CoAPO- 11 is strongly adsorbed, since it is still present after degassing the sample at 353 K for 1 h.After this treatment the broad band at 3100 cm-' (P-OH stretching interacting with acetonitrile) is still observed. This should be due to some stronger acidic P-OH groups in the neighbourhood of Lewis-acid sites (framework cobalt). Fig. 7(c) shows the spectra measured using a SAPO-11 sample. Here a somewhat different behaviour of the sample is observed in the OH stretching region.The bands disap- pearing after acetonitrile adsorption at 3734, 3678 and 363 1 cm-' are due to Si-OH, P-OH and bridging hydroxy groups, respectively. After acetonitrile adsorption these bands are observed at 3300 cm-' (Si-OH), 3100 cm-' (P-OH) and ca. 2600 cm-' (bridging hydroxy groups). The last band is split into two pseudo bands due to resonant interactions between the OH stretching and overtone bending modes of perturbed bridging OH groups.26 The different OH stretch-ing frequencies measured and their shifts after adsorption of acetonitrile on the AlPOs, CoAPOs and SAPOs are sum- marized in Table 3. According to the fact that the shift in v(0H) is a measure of the acid strength it follows that CoAPOs contain medium acid P-OH only.The strength of the Br~rnsted sites in the SAPOs is comparable to that in HY. In accordance with this, Fig. 7(c) shows that acetonitrile bonded to Si-OH and P-OH can be desorbed quite easily. The band in the CEN region at 2266 cm- 'is again ascribed to physisorbed acetonitrile. The band at 2284-2290 an-' is ascribed to Brsnsted-bonded acetonitrile. The apparent shift of this band is due to the contribution of acetonitrile bonded to a P-OH group at high loadings. The band at 2290 cm-', however, remains even after desorption of the sample at 353 K. The band at 2320 cm-' is ascribed to Lewis-bonded ace- tonitrile on non-framework aluminium.'6 Comparable behaviour is observed for the AFI samples.Since the OH stretching region gives no extra information, we will focus on the vibrations of the probe molecule. Fig. 8(a) shows the CSN region of the probe molecule adsorbed on AlPO-5. Bands due to physisorbed acetonitrile (2266 cm- '), acetonitrile interacting with P-OH (2280-2284 cm-') and Lewis-bonded acetonitrile on non-framework aluminium (2321 cm-') are observed. Again all acetonitrile can be desorbed by degassing the sample at room temperature for 15 min, confirming the weakness of the interactions (see Fig. 1 0.10 0.00 1 I 2200 2250 2300 2350 2400 wavenumber/cm-' 0.30 (b1 n .g 0.23 3 4 I I Q-0.15 t /ko90 m e $ 0.08 9 - 0.00 5 I I I 2200 2250 2300 2350 2400 ' wavenum ber/cm -0.30I I 2264I An Y.-5 0.20 Q) C 0.10 sP 0.00LL------------Ll 2200 2250 2300 2350 2400 wavenumber/cm -' 0.40 I I h v)c I I I.z 0.30 3 g v Q) 0.20 0 (De 4 O.'O 5 0.00 I 1 I 2200 2250 2300 2350 2400 wavenum ber/cm -' Fig.8 IR difference spectra of the CsN stretching vibration region of CD,CN adsorbed on AFI molecular sieves: (a) A1P04-5, (b) CoAPO-5/1, (c) CoAPO-5/2, (6)SAPO-5; numerals refer to experi-mental conditions as in Fig. 7; 5, desorption at 573 K and 2). v(CN) of acetonitrile in Lewis complexes has a charac- teristic value determined by the chemical nature of the metal ion. The frequency is a function of the ratio of the ionisation potential of the ion and the ionic radius.16 Therefore v(CN) for Al-based Lewis sites in AlPOs are higher than the stronger Co-based Lewis centres in the CoAPOs.Fig. 8(b) shows the spectra measured on CoAPO-5/1. Again a band at 2304-2309 cm-' is observed, due to aceto- nitrile bonded to framework cobalt. The characteristic fre- quency is too high to be due to Brsnsted-bonded acetonitrile. The band is broadened, probably due to a distribution of the acid strength of these sites. X-Ray powder diffraction and X-ray absorption measurements using synchrotron radiation indicated that two different Co-0 bond lengths existed in c0AP0-5.~~This might explain the distribution in acid strength in CoAPO-5 samples. As can be seen, these sites are rather strong, since they are observed even after desorption of the sample at 573 K.The band at 2290 cm-' is due to a small amount of more strongly acidic P-OH groups. The spectra of the sample with a higher cobalt content [Fig. 8(c)] are comparable to the spectra of the sample with the lower cobalt content, except for the higher intensity of all bands. Again a band at 2304-2310 cm-' is observed, even after high-temperature degassing of the sample, showing the exis- tence of strong coordinatively bonded acetonitrile on lattice co. Acetonitrile adsorption on SAPO-5 [Fig. 8(d)] results in spectra similar to those measured with SAPO-11. At high loadings (0.9 Torr) bands due to physisorbed acetonitrile (2264 cm- '), acetonitrile interacting with P-OH groups (2284 cm-') and Lewis-bonded acetonitrile on non-framework aluminium are observed (2318 cm- '), After some desorption of the sample, the bands of Brsnsted-bonded ace- tonitrile together with Lewis-bonded acetonitrile remain (spectra 3, 4).In contrast to the CoAPOs no band is detected at 2310 cm-', since no cobalt is present in the samples. Table 4 summarizes some of the characteristic V(CEN) frequencies of the acetonitrile sorbed in zeolites,26 AlPOs and CoAPOs on different sites. Until now, pure AlPOs have been compared with samples containing either cobalt or silicon. The presence of both species in one sample has been studied using CoAPSO-44. The spectra for these samples are displayed in Fig. 9. Fig. 9(a) shows the region of the OH stretching vibrations of SAPO- 44, CoAPO-44 and CoAPSO-44.The bands at 3746, 3676, 3625 and 3593 cm-' are ascribed to Si-OH, P-OH, bridg-ing OH (high-frequency band) and to bridging OH (low-fre- quency band), respectively. The high-frequency band is ascribed to undisturbed bridging OH groups, whereas the low-frequency band is due to bridging OH groups pointing into the double six ring.33 The intensities of these bands in the CoAPSO-44 samples are lower, although comparable amounts of silicon were used to prepare the samples. This is due to some clustering of silicon in the CoAPSO sample, which does not lead to the formation of acid sites.6 Acetonitrile adsorption results in a shift in the OH stretch- Table 4 Characteristic V(C=N) frequencies of CD3CN sorbed on different sites sorption state sorption site wavenumber/cm-' physisorbed H-bonded zeolite walls ZSi-OH 2264-2266 2278 EP-OH 2280-2284 coordinatively bonded bridging OH non-framework A13+ Co2+ (framework) 2290-2300" 2320-2332" 2309 " See ref.26. J. CHEM. SOC. FARADAY TRANS., 1994, VOL. 90 2.00 1 n t.-a C 1.80 4 W $ 1.60 C c$ 1.40 G 1.20 I I I I I I 3300 3400 3500 3600 3700 3800 wavenumber/cm- 1.oo 30.80*.- I 2286 2317 (6) C 3 $ 0.60 %PO-44 W Q, 5 0.40 f!2< 0.20 - cO@SO-44 CoAPQ-44 0.00 I I 1 2200 2250 2300 2350 2400 wavenumber/cm-' Fig. 9 IR spectra of the CHA-type molecular sieves in the 0-H stretching mode region (a) and the difference spectra of the adsorbed CD3CN (desorption at 295 K) (b) ing vibrations as described before (not displayed).The differ- ence spectra in the region of the CSN vibrations after room-temperature desorption of the sample are shown in Fig. 9(b). The band at 2286 cm-' is again due to acetonitrile adsorbed on Brernsted sites and some P-OH groups. The fact that a low-frequency shoulder is still observed after room-temperature degassing of the sample is ascribed to the smaller windows of the CHA related structure, and the large amount of Brsnsted-bonded acetonitrile, slowing down the diffusion; desorption for 15 min is not long enough to evac- uate all loosely bonded species, due tol the lower mobility of the probe molecule in the framework of SAPO-44.The CoAPSO-44 structure shows an asymmetric band at 2307 cm-' due to interaction with lattice cobalt. The low-frequency shoulder most probably represents interaction of acetonitrile on the (smaller amount of) Brsnsted sites in this sample compared to SAPO-44. The lower intensity of the band at 2308 cm-' (again due to a Lewis complex) in the CoAPO-44 sample compared to the CoAPSO-44 sample, although comparable amounts of cobalt are present in the samples, is due to the formation of non-framework cobalt in the CoAPO-44 sample during calcination. The non-framework Co most probably does not contribute to the Lewis acidity as has been proved by IR measurements on acetonitrile-loaded CoAPO- 1 1 previously calcined at 970 K in order to remove some lattice Co.The results indicate that no strong Brsnsted-acid sites could be found after incorporation of cobalt in the AlPO structures, but that rather strong Lewis-acid sites are formed. Incorporation of cobalt on hypothetical aluminium sites would result in the formation of weakly acidic P-OH groups if the interaction of the P-OH group and the lattice cobalt is weak. The longer bond length of the Co-0 (2.02 A) compared to A1-0 (1.72 A), calculated at the restricted Hartree-Fock level using a minimal STO-3G basis set, is in J. CHEM. SOC. FARADAY TRANS., 1994, VOL. 90 1039 agreement with this assumption. This and asymmetric tetra- hedral coordination would result in Lewis-acid sites, since the Co2+ is no longer (fully) tetrahedrally coordinated.Additional proof of the nature of the acid sites was obtained from catalytic experiments where C,D, olig~merization~~in CoAPO was studied.35 No H-D L. E. Iton, I. Choi and J. A. Desjardins and V. A. Maroni, Zeo-lites, 1989, 9, 535. S. Dzwigaj, M. Briend, A. Shikholeslami, M. J. Peltre and D. Barthomeuf, Zeolites, 1990, 10, 157. S. G. Hedge, P. Ratnasamy, L. M. Kustov and V. B. Kazansky, Zeolites, 1988, 8, 137. J. A. Martens, M. Mertens, P. J. Grobet and P. A. Jacobs, in exchange with the lattice was found, which confirms the absence of active Brernsted sites in CoAPO samples. Innovation in Zeolite Materials Science, ed. P. J. Grobet, W. J. Mortier, E. F. Vasant and G. Schulz-Ekloff, Elsevier, Amster- dam, 1988, p.97. Conclusions As follows from the adsorption calorimetric measurements, the incorporation of Co into the AlPO structure results in the S. A. Zubkov, L. M. Kustov, V. B. Kazansky, I. Girnus and R. Fricke, J. Chem. Soc., Faraday Trans., 1991,87, 897. P. P. Man, M. Briend, M. J. Peltre, A. Lamy, P. Beaunier and D. Barthomeuf, Zeolites, 1991, 11, 563. formation of strong acid sites. The heats of the most acidic centres amount to 100 kJ mol-' for CoAPO-5 and 110 kJ mol-' for CoAPO-11, which is significantly higher than on non-dealuminated HY zeolite. The heat curves of the CoAPOs show a stepwise course corresponding to two differ- ent sites. The heat curve of CoAPO-11 is shifted by CQ. 10 kJ mol-' towards higher heats compared with CoAPO-5 owing to the smaller pore diameter of CoAPO-11. Therefore the dif- ference in the heats of chemisorption should be due to differ- ent terms of the dispersion interaction and not to differences in acid strength of the centres.From the IR spectroscopic measurements it follows that in 11 12 13 14 C. Halik, J. A. Lercher and H. Mayer, J. Chem. SOC., Faraday Trans. I, 1988,84,4457. R. Khouzami, G. Coudurier, B. F. Mentzen and J. C. Vedrine, in Innovation in Zeolite Materials Science, ed. P. J. Grobet, W. J. Mortier, E. F. Vasant and G. Schulz-Ekloff, Elsevier, Amster- dam, 1988, p. 355. R. A. Schoonheydt, R. de Vos, J. Pelgrims and H. Leeman, in Zeolites, Facts, Figures, Future, ed. P. A. Jacobs and R. A. van Santen, Elsevier, Amsterdam, 1989, p. 559. J. Chen, G.Sankar, J. M. Thomas, R. Xu, G. N. Greaves and D. Waller, Chem. Muter., 1992, 4, 1373. K. Nakashiro and Y. Ono, Bull. Chem. SOC.Jpn., 1993,66,9. R. J. Gorte, G. T. Kokotailo, A. I. Biaglow, D. Parrillo and C. Pereira, in Zeolite Chemistry and Catalysis, ed. P. A. Jacobs, the OH stretching vibration range of the AlPOs and the CoAPO and SAP0 molecular sieves a P-OH band at 3677 cm-' is present, while for the SAPOs and CoAPSO-44 a band assigned to Brernsted sites around 3625 cm-' was found. These bands shift downwards by lo00 cm-' after 15 N. I. Jaeger, L. Kubelkova and B. Wichterlova, Elsevier, Amster- dam, 1991, p. 181. L. M. Kustov, S. A. Zubkov, V. B. Kazansky and L. A. Bondar, in Zeolite Chemistry and Catalysis, ed. P. A. Jacobs, N. I. Jaeger, L. Kubelkova and B.Wichterlova, Elsevier, Amsterdam, 1991, p. 303. adsorption of acetonitrile by the sample, as is typical for 16 H. G. Karge, in Catalysis and Adsorption by Zeolites, ed. G. strong Brernsted sites in zeolites. No bands are observed in this region for CoAPO-5, CoAPO-11 and CoAPO-44 other than the P-OH band which shifts by only 600 cm-', indi-cating a lower acid strength. After adsorption of deuteriated acetonitrile on the sample at room temperature four bands in the range 2265-2325 cm-', characteristic of the CN stretching vibration, were found. The bands at 2265 and 2280 cm-' disappeared after room-temperature desorption of the sample and are asigned to physisorbed and to weakly bonded acetonitrile on the P-OH, respectively. The bands at 2290 and 2320 for the SAPOs and 2310 cm-' for the CoAPOs remain after desorp- 17 18 19 20 21 22 23 Ohlmann, H.Pfeifer and R. Fricke, Elsevier, Amsterdam, 1991, p. 133. E. Paukshtis and E. N. Yurchenko, Russ. Chem. Rev., 1983, 52, 242. B. M. Lok,C. A. Messina, R. T. Gajek, T. R. Cannan and E. M. Flanigen, Eur. Pat. Appl. 0.103.117, 1984. S. T. Wilson and E. M. Flanigen, US Put. 4.567.029, 1986. J. Janchen, H. Stach, P. J. Grobet, J. A. Martens and P. A. Jacobs, Zeolites, 1992, 12,9. B. V. Kuznetsov, N. A. Tuan and T. A. Rachmanova, Ads. Sci. Technol., 1989,6, 27. V. Bosacek and J. Janchen, in preparation. E. G. Derouane, in Guidelines for Mastering the Properties of Molecular Sieves, ed. D. Barthomeuf, E. G. Derovane and W. tion at 353 K. From literature data and for quantum-chemical reasons we assign the 2290 cm-' band to the H-complex with the Brarnsted sites in the SAPOs and the stronger high-frequency bands at 23 10 cm- of the CoAPOs to coordinatively bonded acetonitrile on lattice Co.The band 24 25 Holderich, Plenum Press, New York, 1990, p. 255. M. P. J. Peeters, J. W. de Haan, L. J. M. van de Ven and J. H. C. van Hooff, J. Phys. Chem., 1993,97,8254. H. Stach, I. Girnus, J. Janchen, E. Loffler, U. Lohse, B. Parlitz and B. Zibrowius, Book of extended abstracts, IXth Interna- tional Zeolite Conference, Montreal, 1992, RP142. at 2320 cm-' is assigned to the interaction with Lewis sites on non-framework A1 species that are very often found in zeolites. The absence of strong Brarnsted sites is confirmed by results of the alkene oligomerisation.26 27 28 29 A. G. Pelmenschikov, R. A. van Santen, J. Janchen and E. Meijer, J. Phys. Chem., 1993,97, 11071. P. Venkatesvarlu, J. Chem. Phys., 1951,19,293. C. Angel and M. Howell, J. Phys. Chem., 1969,73,2551. P. 0.Sockert, F. D. Declerck, R. E. Sempels and P. G. Roxhet, The authors are indebted to Prof. R. A. van Santen J. Chem. SOC., Faraday Trans. I, 1977,73,359. (Eindhoven) and Dr. A. G. Pelmenschikov (Novosibirsk) for helpful discussions of the IR results. We thank Mr. E. The-unissen (Eindhoven) for the computation of the T-0 dis-tances, Mr. E. M. van Oers (Eindhoven) for the nitrogen isotherm measurements and Mr. D. Schielinski (Berlin) for 30 31 32 R. E. Sempels and P. G. Roxhet, J. Colloid Interface Sci., 1976, 55, 263. P. G. Roxhet and R. E. Sempels, J. Chem. SOC., Faraday Trans. I, 1974, 70,2021. J. W. Couves, G. Sankar, J. M. Thomas, J. Chen, C. R. A. Catlow, R. Xu and G. N. Greaves, in Proc. IXth Int. Zeolite technical assistance. The financial support of parts of this work by the Stichting Scheikundig Onderzoek in Nederland (SON) is acknowledged. 33 34 Con$, Montreal 1992, ed. R. von Ballmoos, J. B. Higgins and M. M. J. Tracy, Butterworth-Heineman, Boston, 1993, vol. 1, p. 627. B. Zibrowius, E. Loffler and M. Hunger, Zeolites, 1992,12, 167. L. Kubelkova, J. Novakovh, B. Wichterlova and P. Jiru, Collect. References 1 E. M. Flanigen, B. M. Lok, R. L. Patton and S. T. Wilson, in New Developments in Zeolite Science and Technology, ed. Y. Mrakame, A. Iijima and J. W.Ward, Kodansha, Tokyo, 1986, p. 103. 35 Czech. Chem. Commun., 1980,45,2290. M. P. J. Peeters, J. H. C. van Hooff, R. A. Sheldon, L. M. Kustov and V. B. Kazansky, in Proc. IXth Znt. Zeolite Con- ference, Montreal 1992, ed. R. von Ballmoos, J. B. Higgins and M. M. J. Tracy, Butterworth-Heineman, Boston, 1993, vol. 1, p. 651. B. Kraushaar-Czametzki, W. G. M. Hoogervorst, R. R. Andrea, 2 C. A. Emeis and W. H. J. Stork, J. Chem. SOC., Faraday Trans., 1991,87, 891. Paper 3/05865D ;Received 29th September, 1993
ISSN:0956-5000
DOI:10.1039/FT9949001033
出版商:RSC
年代:1994
数据来源: RSC
|
|