摘要:

J. Chem. SOC., Faraday Trans. I, 1985, 81, 2659-2666 Determination of Stability Constants using Linear-scan and Cyclic Voltammograms BY HAMADA M. KILLA Chemistry Department, Faculty of Science, Zagazig University, Zagazig, Egypt Received 21st September, 1984 A method for determining stability constants of metal complex ions by means of linear-scan and cyclic voltammograms is described. It has been tested by comparing the obtained stability constants of cadmium oxalate with these derived from polarographic measurements. The results agree well with each other and with those of others. Furthermore, different systems have been studied using this method and the results agree well with previously reported values. Cyclic voltammetry (c.v.) has been shown to be useful for investigating the mechanisms of the oxidation and reduction of organic compo~nds.l-~ In contrast to polarography, the products of cathodic reduction by C.V.remain near the electrode surface and can be reoxidized by reversing the direction of polarization. There is a well known relationship between the peak potentials (E,) obtained from linear-scan voltammograms (1.s.v.) and half-wave potentials (El) obtained from d.c. p~larography.~-~ This suggests that the former could be used, like the latter,s-ll for the determination of the stability constants of metal complex ions in solution. This is demonstrated in this paper for the cadmium(1r)-oxalate system. There are numerous previous polarographic s t ~ d i e s ~ ~ - l ~ of this compound with which to compare the present results. Moreover, the stability constants of other systems, e.g.Cd-l,3- diaminopropane, Cd-imidazole and Cu-oxalate, have been measured using the 1.s.v. method and the results agree well with previously reported polarographic values. EXPERIMENTAL Current against potential curves were obtained with an IBM EC/225 voltammetric analyser and an IBM X-Y recorder (7424 M). A conventional three-electrode cell with a saturated sodium chloride calomel electrode (SSCE) was employed. The working electrode in linear-scan and cyclic measurements was a PAR 9323 (stationary) hanging-mercury-drop electrode (HMDE) and the electrode area normally employed was 0.0408 cm2. The scan rate was 100 mV s-l. For tast polarography, a drop time of 1.0 s, a scan rate of 5 mV s-l and a sample time of 16.7 ms were used and a potential pulse 50 mV was used for differential pulse polarography. All measurements were made at 25 "C and an ionic strength of 1 .OO mol dm-3 was maintained using sodium nitrate.The concentrations of Cd2+ and Cu2+ ions were 2 x lo-* mol dm-3. Reagent-grade chemicals, deionized water and triply distilled mercury were employed in all cases. A maximum suppressor was not used. THEORY The theory of fast electrode processes has been given by RandlesZ0 and SevEik.21 The Randles-SeviEk equation for the peak current (i,) is i,, = 2.72 x lo5 n3 DiAfiC 2659 ( 1 ) 87-22660 DETERMINATION OF STABILITY CONSTANTS where C is the concentration of depolarizer in solution, D is the diffusion coefficient of the substance being reduced, A is the surface area of the electrode, n is the number of electrons involved in the electrode process and V is the scan rate.In d.c. polarography the limiting current is proportional to the concentration of the depolarizer in the bulk of the solution, but in the case of 1.s.v. it is the peak current (i,) which is proportional to concentration. The linear dependence of the peak current on the concentration of the reacting substance makes this method useful in quantitative analysis. At constant V and A , eqn (1) may be written as ip = k'IC (2) which is similar to the Ilkovic equation in d.c. polarography. Nicholson and Shain5 have shown that the value of E; can be obtained for a reversible stationary electrode polargram using the fact that it occurs at a point 85.17 % of the way along the wave. On the other hand, for a reversible process the peak potential (E,) obtained from a stationary electrode polarogram is related to the half-wave potential of d.c.polarography (E;) by Ep = E;-28.5/n. (3) It would thus appear from eqn (3) that AEp = AE:, which means that the shift in the peak potential (E,) caused by complexation will be the same as the shift in E;. Thus, eqn (3) may be written as (4) (E, )s - (Ep ) c = V ; ) s - V q C where c and s refer to the complexed and free ions, respectively. Using eqn (2) and (4), we can convert the DeFord-Hume22 equation for d.c. polarography in the presence and absence of complexing agent to the following expressions : where F,(X) represents the experimentally measured right-hand side of the equation, Bj is the overall stability constant of thejth complex, Cx is the analytical concentration of the complex-forming substance and Ep is peak potential of the linear-scan voltammetry wave. The term log (ip)s/(ip)c is normally small because it makes little contribution to the I;;.(X) function.Thus, eqn ( 5 ) can be simplified to give an expression of the form: where AE, = (Ep)s-(Ep)c is the shift in the 1.s.v. peak potential caused by complexation. For the determination of the stability constants of the Cd"-oxalate system, the more precise eqn (5) was used for the final calculations with the same steps as previously reported. 22 &(X) = antilog [0.434 (nF/RT) AE,] (6) RESULTS AND DISCUSSION The reversibility that occurs with C.V. is tested by the value of 59 n G--G=-mV where E"p is the anodic peak potential and PP is the cathodic peak potential.In the present work the value obtained was 32 & 1 mV, which is close to the calculated valueH. M. KILLA 266 1 Table 1. Linear-scan voltammetric data for the Cd-oxalate systema [ oxalate] 1% Wp)s /in01 dmP3 - Ep/V AEp/mV /(i,),] lo2 F,(X) lo4 Fl(X) lo4 &(X) lo5 F,(X) 0.0 0.01 0.02 0.03 0.04 0.06 0.08 0.10 0.15 0.20 0.25 0.30 0.35 0.40 0.45 0.598 0.615 0.630 0.637 0.643 0.65 1 0.659 0.667 0.68 1 0.691 0.698 0.707 0.710 0.715 0.719 - 17 32 39 45 53 61 69 83 93 101 109 112 117 121 - 0.0132 0.0 177 0.0222 0.0362 0.03 15 0.0457 0.0606 0.0555 0.0704 0.0655 0.0604 0.0704 0.0860 0.0860 - 0.0388 0.126 0.220 0.363 0.671 I .29 2.49 7.36 16.60 30.6 56.6 73.2 112.7 153.1 - 0.058 0.074 0.088 0.1 10 0.161 0.249 0.490 0.832 1.22 1.89 2.09 2.80 3.40 - 1.39 2.00 2.93 3.9 1 4.68 6.13 5.83 6.88 7.44 - 1.41 1.43 1.67 1.35 1.45 1.41 a I = 1.00 mol dmP3, [Cd2+] = 2 x mol dmP3 and T = 25 "C; log B, = 2.69, log B, = 4.04 and log B, = 5.16.of 29.5 mV for a two-electron reduction.,,7 24 Further confirmation of the reversibility of the Cd-oxalate system was obtained using the tast polarography technique under the conditions used for the C.V. technique. The plots of log [i/(id - i)] against & e were linear with a slope of the order of 3 1 & 1 mV, and by differential pulse p~larography~~ the half-width of the peak was 62 1 mV, indicating the reversibility of the reduction. Note that for a reversible system using 1.s.v. curves Ep is independent of scan rate ( V ) and ip is proportional to V;,' which indicates diffusion control and is analogous to the variation of id with hi in d.c.polarography. In our work the plots of ip against V ; were linear, indicating that reduction of the Cd-oxalate system in nitrate medium was diffusion controlled. The stability constants were calculated by the graphical extrapolation method using eqn (5). The 1.s.v. results for the present system are given in table 1. The plot of E;(X) against the total oxalate ion concentration is shown in fig. 1 and indicates the formation of three complex species, Cd(C,O,), Cd(C,O,)z- and Cd(C,O,):-, with overall formation constants of log B, = 2.69, log B, = 4.04 and log B, = 5.16, which agree well with the results of several other w o r k e r ~ .~ ~ - ~ ~ For further confirmation the C:d-1,3-diamin0propane,~~ Cd-imida~ole~~ and Cu-oxalate12 systems were checked using the 1.s.v. method under the conditions used for the Cd-oxalate system and the results are given in tables 2-4. The reduction of these systems was reversible and diffusion controlled. The results for the Cd-l,3-diaminopropane system are shown in fig. 2. The stability constants of these systems were calculated by graphical-extrapolation methods using eqn (6). The values agree well with previously reported polarographic values12* 2 6 g 27 when the different experimental conditions are taken into account. For an extra check tast polarography was used for the determination of the stability constants of the Cd-oxalate system under the conditions used for 1.s.v.measurements. The current was measured for a short interval near the end of the drop2662 n X W L i 7 1600 1200 800 100 DETERMINATION OF STABILITY CONSTANTS 0.2 0.4 0.1 0.2 [oxalate]/mol dm-3 Fig. 1. Plot of q(X) against [oxalate]. 0, F,(X); e, F,(X); x , F,(X); 0, F,(X). Table 2. Linear-scan voltammetric data for the Cd- 173-diaminopropane systema [ 1,3-diamino- propane] /mol dm-3 -E,/V AEJmV lo4 &(X) lo5 F,(X) lo7 &(X) lo8 F,(X) 0.0 0.0 1 0.02 0.03 0.04 0.06 0.08 0.10 0.15 0.20 0.30 0.40 0.598 0.708 0.723 0.733 0.743 0.755 0.763 0.769 0.783 0.790 0.803 0.813 110 125 135 145 157 165 171 183 192 205 215 0.0533 1.72 3.74 8.17 20.83 38.88 62.08 158.30 319.44 880.61 1921.11 5.33 8.60 12.47 20.42 34.72 48.60 62.08 105.53 159.72 293.35 480.27 - - - 4.36 5.29 2.3 1 5.70 2.25 5.91 2.01 6.84 1.96 7.84 1.97 9.68 1.93 11.93 2.0 1 a Z = 1.00 mol dm-3, [Cd2+] = 2 x lop4 mol dm-3 and T = 25 "C; log B, = 5.477, log B, = 7.59 and log B3 = 8.31.life. In our work the current was sampled for 16.7ms just before the drop was dislodged by the mechanically controlled drop-knocker. This sampling technique reduces the current oscillations observed in classical d.c. polarography. The DeFord-Hume22 expression was used for the final calculations. The results are given in table 5, which shows the presence of three complexes with stability constantsH. M. KILLA 2663 Table 3. Linear-scan voltammetric data for the Cd-imidazole systema [imidazole] /mole dm-3 -Ep/V AE,/mV 10, F,(X) lo3 F,(X) lo4 F,(X) lo5 &(X) lo6 &(X) 0.0 0.02 0.03 0.04 0.05 0.06 0.08 0.10 0.15 0.20 0.25 0.30 0.40 0.50 0.598 0.626 0.637 0.648 0.655 0.662 0.676 0.685 0.706 0.720 0.730 0.740 0.754 0.766 - 37 48 59 66 73 87 96 117 131 141 151 I65 177 - 0.179 0.423 0.997 1.72 2.97 8.86 17.87 91.96 274.08 597.94 1304.44 3887.80 991 3.54 - 0.846 1.38 2.47 3.42 4.94 11.06 17.86 61.30 137.04 239.17 434.8 1 971.95 1982.71 - - 2.93 4.93 5.84 7.40 13.20 17.36 40.53 68.27 95.47 144.77 242.86 396.44 - - - - 7.68 9.00 13.99 15.36 25.69 33.14 37.39 47.59 60.22 78.89 - - - - - - 13.75 12.36 15.12 15.07 13.76 14.86 14.31 15.18 ' I I = 1.00 mol dmP3, [Cd2+] = 2 x lov4 mol dmP3 and T = 25 "C; log B, = 2.69, log B, = 4.301, log B3 = 5.47 and log B, = 7.16.Table 4. Linear-scan voltammetric data for the Cu-oxalate systema [oxalate] - E,/V /mol dm-3 us.SSCE AEPImv lo5 F,(X) lo6 &(XI iOg e(X) 0.0 - 0.0 1 0.015 0.02 0.03 0.04 0.05 0.06 0.08 0.10 0.15 0.20 0.25 0.30 .0.004 0.143 0.157 0.166 0.177 0.182 0.188 0.192 0.201 0.208 0.218 0.225 0.230 0.236 147 0.96 161 2.85 170 5.74 181 13.54 186 20.00 192 3 1.94 196 43.64 205 88.06 212 152.03 222 331.66 229 572.57 234 845.70 240 1350.45 9.59 19.19 28.69 45.13 50.01 63.88 72.73 110.07 152.03 221.10 286.28 338.28 450.15 - 1.39 1.47 1.23 1.26 1.20 1.36 1.51 1.47 1.43 1.35 1 S O a I = 1.00 mol dmP3, [Cu2+] = 2 x mol dm-3 and T = 25 "C; log B, = 6.00 and log B, = 9.13. of log B, = 2.65, log B, = 4.10 and log B, = 5.07. These results are in excellent agreement with the values obtained above using I.s.v. polarograms. The agreement between the present 1.s.v.and polarographic stability constants validates the proposed 1.s.v. method. For further confirmation the present results are in a good agreement with previously reported values obtained using the d.c. polarograp hy technique. 12-12664 n 5 k0 5 300 2 00 100 DETERMINATION OF STABILITY CONSTANTS n 5 G- 52 Le . 500 * 300 - 100 0.4 1 0.05 0 . 1 0.2 [ 1,3 diaminopropane]/mol dm-3 Fig. 2. Plot of 4(X) against [1,3-diaminopropane], 0, F,(X); 0, F,(X); x , F,(X); 0, &(X). Table 5. D.c. polarographic data (tast mode) for the Cd-oxalate systema [oxalate] - E,/V log slope /mol dmA3 us. S k E AE;/mV (Is/Zc) /mV lo2 F,(X) lo4 4(X) lo4 F,(X) lo5 &(X) 0.00 0.01 0.02 0.03 0.04 0.06 0.08 0.10 0.15 0.20 0.25 0.30 0.35 0.40 0.45 0.587 0.603 0.618 0.626 0.630 0.638 0.649 0.655 0.668 0.679 0.688 0.695 0.698 0.703 0.709 - 016 03 1 039 043 05 1 062 068 08 1 092 101 107 111 116 122 - 0.01 37 0.0173 0.021 0.02 1 0.02 1 0.024 0.02 1 0.028 0.028 0.024 0.035 0.028 0.035 0.03 16 - 32.00 3 1 .OO 3 1 S O 3 1 .OO 3 1 .OO 3 1 .OO 31.5 3 1 .OO 30.50 30.50 3 1 .OO 30.50 30.00 30.00 - 0.0360 0.1 17 0.220 0.300 0.561 1.33 2.1 1 5.92 14.00 27.90 45.70 61.400 92.00 146.00 - 0.053 0.070 0.073 0.092 0.165 0.210 0.394 0.697 1.12 1.52 1.75 2.3 1 3.25 1.50 - 1.65 - 2.32 - 3.24 0.981 4.28 1.20 4.29 1.22 4.88 1.03 5.65 1.10 7.1 1 1.29 a I = 1.00 mol drn-,, [Cdzf], T = 2 x mol dm-3 and T = 25.00 "C; log B, = 2.65, log B, = 4.10 and log B, = 5.07.H.M. KILLA 2665 Table 6. Values of stability constants for different systems system stability average of relative method constants deviation error Cd--oxalate d.c.(tast) polarography B, = 1.18 x lo5 1 .OO x lo4 8.47% Cd--oxalate I.S.V. B, = 1.45 x lo5 7.00 x lo3 4.83% Cd--l,3-diaminopropane 1.s.v. B, = 2.05 x lo* 1.245 x lo7 6.07% Cd--imidazole I.S.V. B4 = 1.45 x lo7 7.85 x lo5 5.23% Cu--oxalate I.S.V. B, = 1.35 x 109 9.01 x 107 6.67% It can be seen from the present results that AE, and AE; under the same conditions were not in exact agreement, but in all cases were within experimental error (the maximum difference is 2 mV). Note that we have assumed linear diffusion, which is not strictly correct for HMDE with the scan rate and electrode radius used in this study, and errors of the order of 22; are expected. Moreover, the fast electrode conditions, 100 mV s-l, cause AE, and AE; to differ slightly, as shown in tables 1 and 5.Thus for a reversible, diffusion- controlled process, good agreement was found between the peak potential (E,) of the linear-scan voltammogram and the half-wave potential (E;) obtained from d.c. polarography. This means that AE, = AE+ and hence the DeFord-Hume expression could be used for the I.s.v. results. On the other hand, the DeFord-Hume expression22 was modified for the determination of the stability constants using results from a.c. polarography.89 28 The limitations of the original DeFord-Hume treatment22 have been pointed out by a number of 28f 29 particularly in regard to the accuracy required in E; measurements. Thus, any technique which enables E; values to be measured more accurately than by conventional d.c.polarography will lead to improved calculations of the stability constants. 'The convenient peak form of an 1.s.v. wave allows En to be determined simply and directly from the recorded polarogram, whereas in d.c. polarography E; must be determined from plots of E against log [(id-i)/q,28 which may lead to lower reproducibility. Information on the reversibility or irreversibility of the system and on the chemical reactions accompanying the electrode processes can be obtained directly from C.V. polarogram.6 Moreover, the use of stationary-electrode volotammetry offers the possibility of studying systems where maxima are a problem without the use of maximum suppressors, which are known to have many di~advantages.~Ol 31 The average of the deviation and the relative error of the higher complex (B3) for the Cd-oxalate system were calculated using I.s.v.measurements. Note that the calculations of the value of B, depends on B, and B,. The results obtained using I.s.v. compared with the deviation and relative error in the value of B, for the same system using d.c. (tast) polarography are given in table 6. Furthermore, the same calculations were carried out for the higher complexes formed in the Cd-l,3-diaminopropane, Cd-imidazole and Cu-oxalate systems and the results are given in table 6. Prof. R. H. Philp of the University of South Carolina is thanked for his assistance.2666 DETERMINATION OF STABILITY CONSTANTS Z. Galus, H. Y. Lee and R. N. Adams, J. Electroanal. Chem., 1962, 5, 17.R. P. Buck and L. P. Griffith, J. Electrochem. Soc., 1962, 109, 1005. W. Kemula and Z. Kublik, Bull. Acad. Pol. Sci., CI. 3, 1958, 6, 653. W. Kemula, Z. R. Grabowski and M. K. Kalinowski, Naturwissenschaflen, 1960, 22, 1 . R. S. Nicholson and I. Shain, Anal. Chem., 1964, 36, 706. Z. Galus, Fundamentals of Electrochemical Analysis (Wiley, New York, 1976), p. 231. ’ A. J. Bard and L. R. Faulkner, Electrochemical Methods (Wiley, New York, 1980), p. 218. A. M. Bond and G. Hefter, J. Electroanal. Chem., 1972, 34, 227. D. R. Crow, Polarography of Metal Complexes (Academic Press, London, 1969), p. 56. lo J. J. Lingane, Chem. Rev., 1941, 29, 1. l1 H. Irving, Advances in Polurography (Pergamon Press, New York, 1960), p. 42. l 2 W. B. Schaap and D. L. McMasters, J. Am. Chem. SOC., 1961,83,4699. l3 S. C. Khurana, J. K. Gupta and C. M. Gupta, Electrochim. Acta, 1973, 18, 59. l4 S. C. Khurana and C. M. Gupta, J. Znorg. Nucl. Chem., 1972, 34, 2557. l5 S. C. Khurana and C. M. Gupta, Talanta, 1972, 19, 1235. l6 S. C. Khurana and C. M. Gupta, J. Znorg. Nucl. Chem., 1973, 35, 209. l7 R. G. Bidkar, D. G. Dhuley and R. A. Bhobe, Indian J. Chem., Sect. A, 1977, 15, 63. P. D. Jadhav, R. G. Bidkar, D. G. Dhuley and R. A. Bhobe, J. Indian Chem. Soc., 1976,1,451. l9 A. R. Aggarwal, H. K. Arora, K. B. Pandeya and R. P. Singh, J. Znorg. Nucl. Chem., 1981,43, 601. 2o J. E. B. Randles, Trans. Faraday Soc., 1948,44, 327. 21 A. SevEik, Collect. Czech. Chem. Commun., 1948, 13, 349. 22 D. D. DeFord and D. N. Hume, J. Am. Chem. Soc., 1951,73, 5321. 23 P. Delahay, New Instrumental Methods in Electrochemistry (Interscience, New York, 1954), p. 137. 24 H. M. Killa, E. E. Mercer and R. H. Philp, Anal. Chem., 1984, 56, 2401. 25 J. M. Dillard and K. W. Hanck, Anal. Chem., 1976,48, 216. 26 K. D. Gupta, S. C. Baghel and J. N. Gaur, Monatsh. Chem., 1979, 110, 657. 27 M. Shivhare, K. N. Jain and M. Singh, J. Inorg. Nucl. Chem., 1981, 43, 2885. 28 A. M. Bond, J. Electroanal. Chem., 1969, 20, 223. 29 L. N. Klatt and R. L. Rouseff, Anal. Chem., 1970,42, 1234. 30 L. Meites, Polarographic Technique (Wiley, New York, 2nd edn, 1965). 31 J. Heyrovsky and J. Kbta, Principles of Polarography (Academic Press, New York, 1966). (PAPER 4/ 1635)

ISSN:0300-9599    

DOI:10.1039/F19858102659

出版商:RSC    

年代:1985

数据来源: RSC

摘要:

J . Chem. SOC., Faraday Trans. 1, 1985,81, 2667-2679 Activity of Lipase in Water-in-oil Microemulsions BY PAUL D. I. FLETCHER AND BRIAN H. ROBINSON Chemical Laboratory, University of Kent, Canterbury, Kent CT2 7NH AND ROBERT B. FREEDMAN AND CHRISTOPHER OLDFIELD Biological Laboratory, University of Kent, Canterbury, Kent CT2 7NH Received 24th September, 1984 The lipase-catalysed hydrolysis rates of several nitrophenyl alkanoate esters of varying alkyl chain length (C4-C16) have been measured both in aqueous solution and in water-in-oil (w/o) microemulsions (which are known to contain discrete droplets). Lipase retains its activity in w,/o microemulsions of water, heptane and sodium bis-2-ethylhexyl sulphosuccinate (AOT); the observed rates are consistent with the intrinsic activity of the enzyme (i.e.kcat/&,) being the same as in water. However, the observed conversion rates for C, and C, substrates are slower in the microemulsion system because of substrate partitioning to the oil-continuous phase, which results in a reduced concentration in the aqueous pseudophase. This conclusion is reached by comparing lipase and non-enzymic-(i.e. buffer) catalysed rates in both solution media. Again for the C, and c6 substrate, partition coefficients for the substrates in the limit of high molar ratio of H,O: AOT, as determined from the kinetic results, show good agreement with measured values in heptane + water mixtures. This suggests that lipase functions effectively in the water pseudophase of the microemulsion. Lipase in the microemulsion can also catalyse the hydrolysis of longer chain alkanoates (up to C,,).It can be inferred from the kinetics that such substrates partition to the interface where the lipase must also be active. In the case of AOT microemulsions, the pH profile of enzyme activity is not significantly altered compared with bulk water. The lipase retains > 60% activity in the microemulsion after incubation at 35 "C for 6 days. In w/o microemulsions of water, heptane, chloroform and cetyltrimethylammonium bromide (CTAB), the observed hydrolysis rates are significantly reduced and the intrinsic activity is reduced by a factor of twenty as compared with the AOT system. This is thought to be caused by inhibitory binding of CTAB to the protein. There is current interest in the use of water-in-oil (w/o) microemulsions as a medium for the study of enzyme properties. When solubilised in the small water droplets of the microemulsion, enzymes are afforded some protection from the denaturing effect of the oil solvent.There are novel synthetic possibilities for such systems. Enzymes studied to date include a-chymotrypsin, liver alcohol dehydrogenase and 1 y sozyme . l-' This paper, which describes the activity of a lipase in w/o microemulsion systems, has novel aspects. Lipases in general catalyse the hydrolysis of fatty acyl esters and are known to act at oil/water interfaces.8 Previous work on enzyme activity in microemulsions has been largely concerned with substrates and enzymes located in the water phase of the microemulsion. The substrates used in this work partition strongly to the oil domain of the microemulsion system.The microemulsions used in this study are (a) water + AOT + heptane and (b) water dispersed in CTAB solutions in a 50-vol % heptane+chloroform mixture. Microemulsions are simply prepared ; on mixing the reagents clear thermodynami- cally stable microemulsions are formed on gentle shaking. The AOT system has been extensively characterised using a variety of techniques including PCS (photon- 26672668 LIPASE IN WATER-IN-OIL MICROEMULSIONS correlation spectros~opy),~~ lo ultracentrifugation,ll SANS (small-angle neutron scattering)12 and a fluorescence technique.13 For compositions used in this study, the optically-clear microemulsions consist of small (similar sized) water droplets surrounded by a surfactant shell and dispersed in a continuous oil solvent.The droplet size increases linearly with the mole (or molar) ratio [water]/fsurfactant] (= R) and can be varied over the radius range 1-20 nm.s-13 Increasing the water volume fraction at constant R proportionally increases the concentration of droplets without signifi- cantly changing their size. We have recently studied the CTAB system using PCS, SANS and viscosity and observed qualitatively similar beha~i0ur.l~ In this paper kinetic data are presented for the lipase-catalysed hydrolysis of p-nitrophenylesters of fatty acids (C4-ClG) in the two microemulsion systems. The effect on the kinetics of altering droplet size (by varying R), droplet concentration, the oil solvent, temperature and pH of the dispersed water is discussed.The time course of enzyme inactivation in the microemulsion is also reported. In addition to lipase-catalysed reactions, non-enzymic-catalysed reactions of the same substrates in water and microemulsions have been studied. The subsequent data analysis allows rate effects due to (a) changes in substrate partitioning and (b) changes in intrinsic activity of the lipase in the microemulsion media to be identified. EXPERIMENTAL The lipase used was an extract from Chromobacterium viscosum. It was purchased from Genzyme Corporation and used without further purification. In a series of papers Isobe and coworkers 15-19 have described the purification and properties of two lipases from Chromo- bacterium viscosum. They identify lipase A (mol.wt. = 1.2 x 1 05, 2 subunits, isoelectric point 4.7) and lipase B (mol. wt. = 2.7 x lo4, single unit, isoelectric point 6.9). Both proteins have intrinsic viscosity values of ca. 3 cm3 g-l in water, which indicates they are close to spherical in shape. The lipase preparation used in this study has a molecular weight of 3.3 x lo4 by sodium dodecylsulphate polyacrylamide gel electrophoresis. Iso-electric focussing showed a major band at pH 7k0.2 and three minor bands in the pH range 4-6. A further band was observed at pH 6.5. Lowry analysis showed that 60 wt % of the commercial material is protein. It is concluded that the predominant lipase species present is the Lipase B of Isobe and coworkers. Substrates Ip-nitrophenyl alkanoates (C4-C16)] were purchased from Sigma.Substrate purity was determined by complete hydrolysis and measurement of the liberated p-nitrophenol spectrum. AOT may contain an acidic impurity (resulting from surfactant hydrolysis) which can seriously distort kinetic results.20 For these measurements, AOT was purchased from Sigma and was shown to contain negligible quantities of the acidic impurity.20 CTAB was a Sigma product. n-Heptane (Fisons) was distilled over sodium metal, stored over type 4A molecular sieve and filtered prior to use. Chloroform (May and Baker, analytical reagent) contains 1.5 vol % ethanol as stabiliser and was not purified further. Various buffers (indicated in fig. 3) were freshly prepared and used within 1-2 days. Rate measurements were performed using a Unicam SP-8 200 spectrophotometer.Tempera- ture control was to k0.2 "C. Rates were monitored by means of the absorbance of the p-nitrophenol liberated. Since p-nitrophenol has pK, = 7, the spectrum is a sensitive function of pH. The ratio of nitrophenol/nitrophenolate species is also sensitive to the microemulsion composition as microemulsification causes a pK, change. To obviate these difficulties, the monitoring wavelength was selected as the isosbestic point (Aiso) of the nitrophenol/nitro- phenolate couple. (For measurements in aqueous solution, it was necessary to add 4% ethanol to dissolve the substrates.) The relevant spectroscopic parameters were found to be: 4 vol % ethanol + water Riso = 346 nm and E = 4.8 x lo3 dm3 mol-l cm-l; AOT system, Aiso = 333 nm and e = 4.42 x lo3 dm3 mo1-l cml; CTAB system, Aiso = 347 nm and e = 4.80 x lo3 dm3 mol-l cm-l.A typical experiment was performed as follows. Aqueous solutions of lipase and buffer were added to a concentrated solution of AOT in heptane in a 10 cm3 volumetric flask. A solutionP. D. I. FLETCHER, B. H. ROBINSON, R. B. FREEDMAN AND c . OLDFIELD 2669 of the substrate in n-heptane was added and the solution made up to the mark with heptane. Gentle shaking produced clarification within 1 or 2 min. (Clarification time depends on the buffer used, the R values and the temperature.) The initial linear rate of increase of absorbance with time was then recorded at jliso. Measured rates were independent of the order of addition of reactants. Where necessary, small corrections were made for the non-enzymic-catalysed rates of hydrolysis. Partition coefficients for the C,-C, substrates between n-heptane and water were measured using a U.V.spectrophotometric method, with allowance for spontaneous hydrolysis in the water phase.21 RESULTS AND DISCUSSION THEORETICAL BASIS FOR THE INTERPRETATION OF THE KINETICS There was no significant solubility of active lipase in n-heptane. Therefore, in the microemulsion, the lipase is assumed to be exclusively associated with the droplets. The ester substrates, however, preferentially partition to the oil-continuous phase. The reaction scheme may be modelled, to a first approximation, as follows: fast enzyme-substrate slow (Km)w (Michaelis complex) kcat enzyme (E), + substrate (S), complex (ES), ---+ products (P) p , 1 I fast substrate (S), The subscripts w and o indicate whether the species is located within the water droplet pseudophase or oil-continuous phase, respectively, (Km)w is the Michaelis constant for the formation of the enzyme-substrate complex (ES) in the water droplet pseudophase and kcat is the first-order rate constant for reaction within the Michaelis complex. The substrate partition coefficient P, is defined as: PI, [SI, P =-.Concentrations are defined per unit volume of the water and oil phases. velocity of reaction ( u ) : Using the analysis of Berezin,22 the following expression is derived for the initial where the subscript T indicates that the concentration is expressed as number of moles per unit of total volume of the microemulsion medium (as opposed to per unit volume of one of the pseudophases).The apparent reaction parameter (Km)app is related to the ‘true’ parameter (Km)w in the water droplet pseudophase (where reaction occurs) as follows: where 8 is the volume fraction of the water droplet pseudophase. For a water-soluble substrate, i.e. P, % 1, then (Km)app = (Km),8. For an oil-soluble substrate (P, + 1) I S = (Km)w/Ps when 8 6 1.2670 LIPASE IN WATER-IN-OIL MICROEMULSIONS 0 2 4 6 [lipaseIT/pg cm-3 Fig. 1. Dependence of initial rate of lipase-catalysed hydrolysis of p-nitrophenyloctanoate in 0.1 mol dm-3 AOT ( R = 1 1.1) at 25.0 "C. 0 I I 1 2 3 [substrate];."103 dm3 mol-' Fig. 2. Lineweaver-Burk plot for the lipase-catalysed hydrolysis of p-nitrophenyloctanoate in 0.1 mol dm-3 AOT ( R = 1 1.1) in heptane at 25.0 "C.The pH is 8.15 (diglycine).P. D. I. FLETCHER, B. H. ROBINSON, R. B. FREEDMAN AND c . OLDFIELD 2671 For these experiments, 8 was typically in the range 0.01-0.15. Fig. 1 and 2 shows the rate behaviour for the lipase-catalysed hydrolysis of p-nitrophenyloctanoate. The rate of reaction is linearly dependent on enzyme concentration (fig. 1). The Lineweaver-Burk plot (fig. 2) (from which K , and kcat are generally obtained from the slope and intercept) shows a zero intercept, indicating that ( Z Q a p p is too large to be separated from kcat in this system. Hence (Km)app, which is a dissociation constant, is much greater than [SIT. Eqn (2) can then be expressed in the form It is simpler, in practice, to derive a second-order rate constant, k,, defined as follows : ' (7) = k2[E1 where k , = constant kcat/(Km)app = constant kcat (8) The constant term is necessary because the enzyme concentration [El is here expressed in g ~ m - ~ of total solution, since the molecular weight of the enzyme is unknown.Hence the units of k, are cm3 g-l s-l [from eqn (7)]. FACTORS INFLUENCING THE RATE OF LIPASE-CATALYSED REACTIONS IN AOT MICROEMULSIONS The influence of various experimental parameters on k, was examined. For a constant R value (R = [H,O]/[AOT] = 11.1) the rate constant k , was independent of [AOT] over the range 0.05-0.4 mol dm-3. In addition, k , is independent of the buffer (diglycine) concentration up to 2 x mol dmP3 diglycine (expressed as an overall concentration). This independence of the rate on buffer concentration is an important experimental check that sufficient buffering capacity is present in solution.The effect of pH on the rate constant is shown in fig. 3. Different buffers were used near their pK, values to ensure sufficient buffering capacity. A maximum is found in the range 6.0-7.5, as observed for the same reaction in bulk water.,' Since the pH cannot be measured absolutely in microemulsions, the pH refers to that of the water before solubilization. However, we estimate from the buffer range used that the pH is precise to < 1 pH units in the microemulsion. The effect of changing droplet size is shown in fig. 4. Droplet size was varied by changing R. Fig. 4 shows results obtained with three different buffers. The data show that k, is indeed constant in large droplets (high R), but increases as the droplet size is reduced, going through a maximum at R = 1 1.The radius of the water core of an R = 11 water droplet is ca. 2 nm. The temperature dependence of k , is shown in fig. 5. For all R values, the rate constant goes through a maximum at ca. 30 "C. The decrease in rate at high temperature is not the result of an irreversible loss of enzyme activity, since when the solubilised enzyme was incubated at 55 "C for 15 min and subsequently assayed at 25 "C, the observed rate was only 10% lower than that in a control that had not been incubated. This compares with 50% lower activity noted in fig. 5 when the assay is carried out at 55 "C. From eqn (8) it is clear that the temperature dependence of k, is a composite of the dependences of kcat, (Km)w, P, and of any effect of temperature on the microemulsion structure (e.g.size of water droplets). No simple interpretation of the temperature dependence of k, is therefore to be expected. However, note that the low-temperature data of fig. 5 for R = 1 1.1, when plotted as an Arrhenius plot, yield an apparent activation energy of 44 kJ mol-' compared with the value of 43 kJ mol-1 observed in bulk water.,'2672 LIPASE IN WATER-IN-OIL MICROEMULSIONS 600 4 I v) I 4 400 5 \ y" 200 0 Fig. 3. Variation of k, with aqueous pH in 0.1 mol dm-3 AOT (R = 11.1) in heptane. The substrate is p-nitrophenyloctanoate, the temperature is 25 "C and [buffer], = mol dm-3. In ascending order of pH the buffers used were citrate, maleate, citrate, cacodylate, phosphate, imidazole, collidine, tris, diglycine, glycine and phosphate.- I v1 - I m 400 - 5 ---. y" 200 - 0 20 LO R Fig 4. Variation of k, with mole ratio of H,O/AOT (= R). Substrate is p-nitrophenyloctanoate in 0.1 mol dmP3 AOT in heptane at 25.2 "C. The buffers used were diglycine (A), cacodylate (0) and phosphate (0). The stability of the enzyme in microemulsions was investigated over extended periods at 35 "C. Fig. 6 shows that the enzyme is very stable in microemulsions of all R values, with 6045% activity remaining after 6 days. The most rapid drop in activity is observed for the largest R values (largest water droplets); in microemulsions of R = 2.8 an initial increase in activity is observed, followed by a relatively rapid decrease.There is no obvious explanation for this latter effect.P. D. I. FLETCHER, B. H. ROBINSON, R. B. FREEDMAN AND c. OLDFIELD 2673 3 00 .. I m 4 I M 3 --- 2oo lu" 100 0 \'* \ '\ 1 I I 1 I I0 20 30 10 50 T/"C Fig. 5. Variation of k, with temperature for the reaction of C , in 0.1 mol dm-3 AOT at R values of 5.6 (O), 8.3 (A), 11.1 (O), 16.7 (A) and 27.8 (@). 100 n E 0 + E 50 0 1 I I 2 4 6 incubation tirne/days Fig. 6. Activity of lipase in 0.1 mol dm-3 AOT microemulsions as a function of time. The solutions were incubated at 35.0 "C and assayed a t 25.0 "C using the C , substrate. R values are 2.8 (A), 5.6 (O), 11.1 (O), 27.8 (A) and 55.6 (0). A comparison of the rates observed with p-nitrophenylalkanoates of different chain lengths provides some interesting data.Fig. 7 shows the variation of k, with R for the different substrates and the effect of substituting dodecane for heptane as the oil-phase solvent. The value of k, for the C,, substrate is only four-fold lower than that for the C , homologue. From eqn (7), k, is linearly dependent on the partition2674 LIPASE IN WATER-IN-OIL MICROEMULSIONS 500 200 - I v1 - I ," 100 5 1 y" 50 20 d' I 1 1 0 20 40 60 R Fig. 7. Variation of k, with R for different p-nitrophenylalkanoates (c4-Cl6). The solutions contain 0.1 mol dm-3 AOT at 25.3 "C with diglycine buffer. The substrates are C, (a), C, (+), C, (O), C,, (A), C,, (m), C,, (0) and C, in dodecane (as opposed to heptane) oil solvent 1 2 [ diglycine] -,-/ 1 0-3 mol dm-3 Fig. 8. Variation of kapp for C, hydrolysis in 0.1 mol dm-3 AOT ( R = 11.1) at 25.2 "C.P. D.I. FLETCHER, B. H. ROBINSON, R. B. FREEDMAN AND c . OLDFIELD 2675 coefficient I?,, and for homologous compounds P, would be expected to decrease by a factor of four for each additional -CH2- group.23 Hence for these substrates would be expected to vary by over a 107-fold range. The fact that the values of k, for all the substrates fall within a relatively narrow range indicates that some other factor is also varying. A decrease in (&JW with increasing chain length, in parallel with compensatory variations in partition behaviour, may offer an explanation, but it is also likely that for the longer chain substrates a build up of substrate in the interface region can occur. STUDIES OF NON-ENZYMIC HYDROLYSIS RATES IN MICROEMULSIONS In order to resolve effects on partition behaviour from effects arising from intrinsic properties of the enzyme, studies were carried out on related non-enzymic hydrolysis reactions in microemulsions since the results provide an alternative route to Ps.This approach is in many ways preferable to a direct determination of the partition coefficients since it is difficult to study the partition behaviour of the substrates in the microemulsions, and partition coefficients obtained from studies using bulk water and heptane phases may not necessarily be appropriate. Esterolysis reactions are subject to nucleophilic catalysis. The non-enzymic rates of hydrolysis of the p-nitrophenyl esters were determined in water and in AOT microemulsions as a function of R and of buffer, substrate and AOT concentration.Reactions in phosphate and cacodylate buffers showed a spontaneous rate unaffected by buffer concentration. However, at high concentrations of diglycine buffer, the rate showed a first-order dependence on diglycine concentration both in water an in microemulsions, and hence the reaction is catalysed by diglycine (fig. 8). The reaction also showed first-order dependence on substrate concentration. Since diglycine is charged and hence confined to the water-droplet pseudophase, the reaction scheme shown for the enzyme-catalysed reaction is applicable. The observed second-order rate constant k,(obs.) for buffer catalysis is then given by k,(obs.) = ( I C ~ ) ~ P , . This provides a route for the indirect determination of P, in the microemulsion system itself, by comparison of rates of diglycine-catalysed hydrolysis in microemulsions and in bulk water.The assumption is made that (k2)w, the rate constant in the microemulsion water droplets, is the same as the rate constant measured in bulk water. This assumption is most likely to hold at high Data for a comparison of reaction rates in water and in high-R microemulsions are collected in table 1. The data are for C , and C, substrates and include second-order rate constants for both lipase- and buffer-catalysed reactions and first-order rate constants for the spontaneous hydrolysis in phosphate buffer. For non-enzymic reactions the ratio of rates in the microemulsion to rates in water gives an estimate of P,.For the C, substrate this estimate is in the range (5.0-6.6) x whereas the partition coefficient directly determined for partition between water and heptane is 4.7 x For the C, substrate the rate ratios are an order of magnitude lower, as is the directly determined partition coefficient. The self-consistency of this set of data suggests that the partitioning of these short-chain substrates between the water and heptane pseudophases of AOT microemulsions roughly corresponds to that between bulk water and heptane. Furthermore, the trend in partition coefficients with substrates of varying chain length has been found21 to agree with that predicted from the theory of T a n f ~ r d . ~ ~ Most importantly, this set of data shows that incorporation of the lipase into AOT microemulsions has very little effect on its catalytic properties.The difference in 252676 LIPASE IN WATER-IN-OIL MICROEMULSIONS Table 1. Comparison of rates in aqueous solution and AOT microemulsion for buffer- and lipase-catalysed reactions ~ ~~ ~ ~~ ~ rate in AOT rate (micro- system rate in 4% EtOH + water (high R) /rate (water) microemulsion emulsion) C,/diglycine (3.0f0.3) x dm3 mo1-1 s-l (1.5f0.3) x dm3 mol-l s-la 5.0 x 10-3 C,/diglycine (2.2f0.5) x lop2 dm3 mol-l s-l (1.7f0.7) x lop5 dm3 mol-l s-la 7.7 x 10-4 C4/lipase (6.5f0.5) x lo4 cm3 g-l s-l 2.2 x 10-3 C6/lipase (5.0 & 1 .O) x lo5 cm g-l s-l 60 6 cm3 g-' s-' 1.2 x 10-4 C,/phosphate (3.5 k0.7) x s-l (2.3f0.5) x lo-' s-l 6.6 x 145 & 15 cm3 g-' s-' a These values are extrapolated values corresponding to R = co, otherwise R = 55.Values of P, observed experimentally between bulk heptane and water were (2.5 f 0.2) x for the C, ester and (1.4 & 0.2) x 10-4 for the C, ester.21 A comparison should be made of P, with data in the last column. 3 0 30 60 R Fig. 9. Variation of k, with R for the diglycine-catalysed hydrolysis of various alkanoate substrates in 0.1 mol dm-3 AOT at 25.3 "C. Substrates are C, (O), c6 (A), C, (0) and c16 (a). observed rates of lipase-catalysed reactions in water and in the microemulsion can be entirely explained by effects of substrate partitioning, since the rate ratios for the lipase catalysed reaction are close to those for the non-enzymic reactions and so there can be little change in the enzyme's catalytic parameters (i.e.kcat/Km) on incorporation into the microemulsions. This conclusion, together with the data of fig. 3 and 5 on pH and temperature dependence, implies that the enzyme in AOT microemulsions is comparable in catalytic behaviour to the enzyme in bulk water. This behaviour is in contrast to that observed for a-chymotrypsin in AOT microemulsions, where K , is increased one-hundred fold' as compared with bulk water. However, reactions in microemulsions show complexities which indicate that theP. D. I. FLETCHER, B. H. ROBINSON, R. B. FREEDMAN AND c. OLDFIELD 2677 1 A 1 1 1 0 10 20 30 R Fig. 10. Variation of first-order rate constant kapp with R for C , hydrolysis in 0.1 mol dm-3 AOT in heptane at 25.3 "C. initial model is a considerable oversimplification.Fig. 9 shows the variation of k,(obs.) for the buffer-catalysed ester hydrolysis as a function of R and of substrate chain length. The rate constants increase with decreasing R. This may be explained, in part, by the fact that the solubilised water in droplets of low R is known to be of lower polarity than that of bulk water.25 Hence the substrates may partition into water droplets of low R more favourably than they do into droplets of high R, leading to increased rates of hydrolysis. The complexity of microemulsions as a reaction medium is also demonstrated by the data of fig. 10, which shows the variation with R of the spontaneous hydrolysis rate in phosphate buffer. The rate goes through a maximum at R = 8 and decreases sharply at low R, in contrast to the situation for the diglycine-catalysed rate (fig. 9).With diglycine acting as a nucleophilic catalyst, water is not directly involved in the rate-determining step. For the spontaneous reaction in phosphate buffer, the rate- determining step presumably involves attack by H,O or OH-. Thus the two reactions will be influenced differently by effects on the activity of water, which is thought to drop rapidly26 with R as R decreases below 10. Hence the decrease in rate at low R in fig. 10 is mainly caused by the decrease in water activity, whereas the increase in rate at low R observed in fig. 9 is mainly determined by the decrease in water polarity and consequent increased partitioning. The combination of these effects may explain the variation in lipase-catalysed rates with R (fig.4). The range of rates from C, to c16 substrates is much less than that expected if substrate partitioning alone is significant. In other words the rate of buffer-catalysed hydrolysis of the C,, substrate is much higher than would be expected on the basis of the simple partition model used so far. This is due to the fact that the long-chain substrates are surface-active and will partition into the surfactant interface between the water and heptane pseudophases. For the long-chain alkanoates, the concentration at the interface will exceed that in the water droplet. It would appear that the substrate at the interface is able to react and so the observed rate will be higher than that predicted on the basis of the heptane/water partition coefficient.The model initially proposed neglects this interface entirely. Thus the apparently anomalously high reaction rates for the long-chain substrates are readily explained.2678 LIPASE IN WATER-IN-OIL MICROEMULSIONS 0 0 20 40 60 Fig. 11. Variation of k, with R for the lipase-catalysed hydrolysis of C, (0) and C, (0) in 0.1 mol dm-3 CTAB at 25.3 "C. The oil solvent is 50 vol % heptane+chloroform. R Table 2. Comparison of rates in aqueous and CTAB microemulsion for buffer- and lipase-catalysed reactions rate in CTAB rate (micro- (high R) emulsion) system rate in 4% EtOH + water microemulsion /rate (water) ~~ ~~ ~~ ~~ CJdiglycine (3.0 k 0.3) x dm3 mol-l s-l (1.2 0.4) x dm3 mol-l s-l 4.0 x C,/lipase (6.5k0.5) x lo4 cm3 g-l s-l 12+ 1.5 cm3 g-l s-l 1.8 x 10-4 STUDIES IN CTAB MICROEMULSIONS For comparison, some studies of lipase-catalysed hydrolyses were also carried out in CTAB + heptane + chloroform + water microemulsions. As with the AOT system, the observed initial rates were first order with respect both to enzyme and to substrate.Fig. 11 shows the second-order rate constant k, as a function of R in these CTAB microemulsions. Very significantly, the rate constants are approximately one-two orders of magnitude lower than in AOT microemulsions. The variation with R is also qualitatively different in that the rate is only slightly dependent on R. A further difference between CTAB microemulsions and AOT microemulsions is noted when me p n is vanea; in L ~ A D rnicroemuisions, m e rare is viriuaiiy inoepenoeni vi me pH of the solubilised water over the range 5-1 1.Results obtained with the enzyme a-chymotrypsin in CTAB microemulsions showed normal pH behaviour,' so this effect is neculiar to the linase and is not a nronertv of these nH buffers in CTAB microemulsions. The apparent independence of the second-order rate constant on R and onP. D. I. FLETCHER, B. H. ROBINSON, R. B. FREEDMAN AND c . OLDFIELD 2679 the water droplets. Some specific interaction between lipase and CTAB may occur, for example the formation of an oil-soluble complex of lipase and a few molecules of CTAB. The lipase in such an aggregate would have little or no interaction with the buffered water pseudophase and would be unaffected by its pH or other properties. This proposal is supported by our observation that lipases in aqueous solution interact with low concentrations of CTAB, suggesting very strong interactions between the enzyme and this particular surfactant. Table 2 shows the rates of diglycine-catalysed ester hydrolysis in CTAB micro- emulsions and a comparison with corresponding rates in bulk water.In contrast to the situation with AOT microemulsions (table l), where the ratios of rate in microemulsion to rate in water were the same for lipase- and buffer-catalysed reactions, the rate ratio for lipase in CTAB microemulsions is 20-fold lower than that for the buffer-catalysed reaction. This again suggests that incorporation of the lipase into CTAB micro- emulsions produces some specific effect on the lipase reducing its catalytic activity. We thank the S.E.R.C.(Biotechnology Directorate) and Tate and Lyle Research for financial support. P. D. I. Fletcher, R. B. Freedman, J. Mead, C. Oldfield and B. H. Robinson, Colloid Surf., 1984,10, 193. R. Hilhorst, C. Laane and C. Veeger, FEBS Lett., 1983, 159, 225. C. Grandi, R. E. Smith and P. L. Luisi, J. Biol. Chem., 1981, 256, 837. F. M. Menger and K. Yamada, J. Am. Chem. Soc., 1979, 101,6731. P. Meier and P. L. Luisi, J. Solid-phase Biochem., 1980, 5, 269. K. Martinek, Y. L. Khmelnitskii, A. V. Levashov, N. L. Klyachko, A. N. Semenov and I. V. Berezin, Dokl. Akad. Nauk SSR, 1982, 256, 1423. R. Verger and G. H. de Haas, Annu. Rev. Biophys. Bioeng., 1976, 5, 77. J. D. Nicholson and J. H. R. Clarke, in Surfactants in Solution, K. L. Mittal and B. Lindman (Plenum Press, New York, 1984), vol. 3, p. 1663. ' S. Barbaric and P. L. Luisi, J. Am. Chem. SOC., 1981, 103,4239. lo M. Zulauf and H. F. Eicke, J. Phys. Chem., 1979, 83, 480. l 1 B. H. Robinson, D. C. Steytler and R. D. Tack, J. Chem. Soc., Faraday Trans. 1, 1979, 75,481. l 2 B. H. Robinson, C. Toprakcioglu, J. C. Dore and P. Chieux, J. Chem. Soc., Faraday Trans. 1 1984, l3 N. J. Bridge and P. D. I. Fletcher, J. Chem. SOC., Faraday Trans. 1, 1983, 79, 2161. l4 P. D. I. Fletcher, B. H. Robinson and J. Mead, to be published. l5 T. Yamagluchi, N. Muroya, M. Isobe and M. Sugura, Agr. Biol. Chem., 1973 37, 999. l6 M. Sugiura, M. Isobe, N. Muroya and T. Yamaguchi, Agr. Biol. Chem., 1974,38, 947. *' M. Sugiura and M. Isobe, Biochim. Biophys. Acta, 1974, 341, 195. l8 M. Sugiura and M. Isobe, Chem. Pharm. Bull., 1975, 23, 1226. l9 M. Isobe and M. Sugiura, Chem. Pharm. Bull., 1977, 25, 1980. 2o P. D. I. Fletcher, N. M. Perrins, B. H. Robinson and C. Toprakcioglu, in Reverse Micelles: Biological and Technological Relevance of Amphiphilic Structures in Apolar Media, ed. P. L. Luisi (Plenum Press, New York, 1984), p. 68. 80,413. 21 C. Oldfield, P. D. I. Fletcher, R. B. Freedman and B. H. Robinson, to be published. 22 K. Martinek, A. V. Levashov, N. I. Klyachko, V. I. Pantin and I. V. Berezin, Biochim. Biophys. Acta, 23 C. Tanford The Hydrophobic Eflect (Wiley, New York, 1980). 24 M. Wong, J. K. Thomas and I. Nowak, J. Am. Chem. Soc., 1977,99,4730. 25 M. Wong, J. K. Thomas and M. Gratzel, J. Am. Chem. Soc., 1976,98, 2391. 26 R. Kubik and H. F. Eicke, Helv. Chim. Acta, 1982, 65, 170. 1981, 657,277. (PAPER 4/ 1650)

ISSN:0300-9599    

DOI:10.1039/F19858102667

出版商:RSC    

年代:1985

数据来源: RSC

摘要:

J. <?hem. SOC., Faraday Trans. I, 1985,81, 2681-2692 268 1 Experimental and Theoretical Investigation of the Kinetics of the Sorption of Water Vapour by Silica Gel BY JAN Y. ANDERSON,* HENRIK BJURSTROM, MICHEL AZOULAY AND Bo CARLSON Department of Physical Chemistry, The Royal Institute of Technology, S-100 44 Stockholm, Sweden Received 19th November, 1984 The kinetics and mechanism of the sorption reaction between water vapour and silica gel have been investigated in the pressure range 1-50 Torrt using thermogravimetry under controlled temperature and water-vapour pressure. Measurements at equilibrium were carried out in order to determine the pertinent equilibrium parameters and the extent of hysteresis. The influence of particle radius and temperature on the kinetics in the whole of the pressure range investigated are well described by theoretical curves fitted to experimental data by taking into account simultaneous mass- and heat-transfer.Calculations yield a diffusivity of 8 x m2 s-l. The sorption kinetics in the pressure range 15-40 Torr are shown to be controlled by heat transfer. A number of hygroscopic solids such as salt hydrates, molecular sieves and silica gel have been proposed as sorbents for water vapour in the storage of low-temperature heat.l However, the structural transitions that can take place when salt hydrates sorb or desorb water add complexity to the analysis of the overall sorption On the other hand, in the case of adsorbents the analysis of the mass- and heat-transfer in the gas-solid system is facilitated because adsorption on a surface is a fast process.Silica gel is a porous adsorbent: water absorbed by the gel is found either adsorbed on the walls of the pore or condensed in the capillaries formed by the pores. The smaller the pore size, the more predominant is the film adsorption and the less marked is the hysteresis between sorption and desorption at high water content^.^ In the present series of investigations of the factors influencing the thermal output charac- teristics of a chemical heat pump, i.e. this study and previous reports,6” we have chosen a gel with narrow pores (2 nm). The reaction with water vapour in this case is known to be reversible and it exhibits a small hysteresis.8,9 In the present investigation, which is aimed at determining the relative importance of mass- and heat-transfer in the reaction of water vapour with silica gel, a thermobalance has been utilized in order to carry out kinetic experiments on a few particles only.The model of Lee and RuthvenlO is used in the analysis of the results obtained. EXPERIMENTAL MATERIALS ’The silica gels used were Davison gels of types 125 and 127 from W. R. Grace and Co., delivered as irregular granules. (The specific surface area of the dry gel is 750 m2 g-l, its porosity is 0.45 for a particle density of 1200 kg m-3 and its mean pore diameter is 2.1 nm.ll) A pore-size analysis of adsorption equilibrium data (see below) using a model-less method” yielded pore sizes < 2 nm. The gel was used without any prior treatment. t I Torr = 101 325/760 Pa.268 12682 SORPTION OF WATER VAPOUR BY SILICA GEL PROCEDURE AND INSTRUMENTATION EQUILIBRIUM STUDIES The equilibrium water-uptake properties in an atmosphere of pure water vapour were determined gravimetrically using a Cahn RG electrobalance. The sensitivity of the balance in the pertinent mass range is 0.05 mg, which corresponds to an accuracy of < 1 % for the measured water content. A sample consisting of a few particles was placed in a glass pan suspended from the balance lever and surrounded by a jacketed glass tube connected to a thermostatting bath. Prior to the experiments the system was thoroughly degassed. The water-vapour pressure was kept constant by means of a thermostatted evaporator containing distilled water. Equilibrium sorption and desorption isobars were obtained by stepwise varying the temperature with pauses for equilibrium to be established. KINETIC STUDIES The present kinetic studies of the reaction of water vapour with silica gel have been conducted by means of a modified McBain thermobalance as described elsewhere4* l3 (with an accuracy of ca.5pg for a given run). The balance permits measurements on a few particles in a controlled atmosphere of water vapour. The samples consisted of ca. 10-15 particles spread out on the glass pan, all particles being well separated from each other, except when a pulverized sample was used. The dry weight of a sample was between 40 and 50 mg, as determined by exposing the sample to vacuum for 1 h at 343 K. The sample weight change was continuously recorded using a microcomputerized data-acquisition system which allowed for measurements every 30 s.The dead-time of the thermogravimetry (t.g.) apparatus, which corresponds to the time required for opening valves and pressure-levelling, was estimated to be 1-3 min, depending on the pressure range in which the balance was used [cf. the pressure curves in fig 2(a) and (b)]. The temperature change of the sample during sorption and desorption was checked by means of a silica-gel particle glued to a thermocouple junction located in the oven. RESULTS EQUILIBRIUM MEASUREMENTS All the adsorption isobars for the five different pressures (7.3, 8.9, 11.0, 12.9 and 15.7 Torr) and temperatures between a few degrees above the dew point of water vapour and 343 K can be reasonably well represented by a single curve in a Dubinin-Polanyi plot,14 i.e.water content w against Ap, where Ap = RTln ( p / p o ) [see fig. 1 (a) and (b)]. Thus, by fitting an expression of a double-Langmuir type:9 [where X = (p/po)T/al] to each isobar we find values of the parameters a,(i = 1-6) (cf. The difference caused by hysteresis between a sorption and a desorption curve is relatively more pronounced [see fig. 1 (b)]. The hysteresis can be observed for water contents above 0.20 g water per g dry silica gel. The equilibration times for both sorption and desorption processes are very large in the vicinity of this water content, as compared with the 2-3 h that are sufficient for lower and higher water contents. The last experimental value that exhibits hysteresis in a desorption series was recorded after 50-80 h, although equilibrium had not yet been established.If the equilibrium value is assumed to correspond to that obtained during the sorption series, t1,2 of the relaxation process is found to be of the order of one week. fig- l(b)l.0.30 - e, ; 0.20 a bo .-I I 02 3 bo 1. 0.10 J. Y. ANDERSON, H. BJURSTROM, M. AZOULAY AND B. CARLSSON 2683 0.30 - & m 0.20 .-I I 00 0- -.-- M 3 0 -10 1 2 3 4 5 6 7 8 9 10 55 * I 50 Z E 3 --- 2 45 Fig. 1. (a) Water content w plotted against the chemical-potential difference between sorbed and pure liquid water, Ap = - RT In @/po), for sorption isobars obtained at five different pressures: x , 7.3; A, 8.9; 0, 11.0; 0, 12.9 and +, 15.7 Torr. (b) Comparison between a sorption (0) and a desorption (@) isobars obtained at a pressure of 15.7 Torr.(-) Curve fitted according to eqn (1); (---) plot according to eqn (1) as recommended by Jury and Edwardss for the same type of gel; ( - - * .) differential enthalpy of sorption, AH.2684 SORPTION OF WATER VAPOUR BY SILICA GEL Differential heats of sorption, AH, at varying water contents have been derived from eqn (1) using Clapeyron's equation (cf. fig. 1): AH-AHo = - RTln (PIPo). Both the numerical values of AH and their decreasing trend with increasing water content are in good agreement with published data for similar silica gels.15 KINETIC MEASUREMENTS Typical relaxation curves recorded by t.g. at constant temperature and pressure are illustrated in fig. 2(a) and (b). Analysis of these curves shows that all of them follow a simple exponential expression when considering only times t > tlj2.The part of the relaxation curve used in the calculations can be described with a single rate constant, k. The initial part of the curves includes, as well as the dead time of the instrument, a non-exponential decay period. The main factors influencing the reaction rate are pressure, particle size and temperature. The rate constant is found to be insensitive to variations in the initial pressure when the pressure jump is a few Torr only. For pressure jumps > 5 Torr and for those < 1 Torr the rate constant will deviate. In all the present experiments the difference between initial and final pressure was therefore chosen to be in the interval 1 < Ipc-pi I/Torr < 5, the lower limit being set by the dead time of the t.g. instrument.Fig. 3 shows the dependence of the rate constant on pressure for sorption and desorption experiments conducted at 313 K with the fraction: rp = 0.5-1 mm. The minimum in k appearing at ca. 30 Torr is noteworthy. The influence of particle size on the rate constant has been investigated for desorption at 3 13 K and varying water-vapour pressures (cf. fig. 4) using three particle sizes, uiz. rp = 60-100 pm, 0.5-1 mm and 2-2.5 mm. The rate constant, k , is found to increase strongly with decreasing particle size. However, irrespective of the particle size and in the pressure region that has been investigated, the rate constant decreases monotonically with increasing pressure. The temperature dependence of the rate constant for three different pressures (5.3, 8.7 and 13.7 Torr) is displayed in fig.5: the rate constant increases with temperature. ANALYSIS SIMULTANEOUS MASS- AND HEAT-TRANSFER Lee and Ruthven have presented a model predicting the kinetics of the pertinent type of adsorption system when it is simultaneously controlled by internal mass transfer (diffusion) and external heat transfer.lO The solid sorbent is assumed to be a spherical particle for which external heat transfer may influence the diffusion of sorbate. Temperature gradients in the particle are neglected, and its surface is supposed always to be in chemical (but usually not thermal) equilibrium with the surrounding gas. The transfer of heat between the particle and the surrounding gas can be described with the heat-transfer coefficient h, a constant during the reaction.Starting with the coupled diffusion and heat-transfer equations and linearizing, Lee and Ruthven obtained an analytical solution which for small pressure jumps and constant diffusivity can be written as:J. Y. ANDERSSON, H. BJURSTROM, M. AZOULAY AND B. CARLSSON 2685 2 1 . 5 M E Q -... E ' 0.5 0 A I I 10 20 30 10 6 4 G c --- 4 2 0 28 26 24 22 A4 AAAAAA A A 20 0 10 20 30 40 50 tlmin 4 G . Fig. 2. Relaxation curves obtained by t.g. The solid curves are obtained by fitting an exponential expression to experimental data obtained at times t =- (a) Sorption at 313 K and 7.7 Torr using a pressure jump of 2.2 Torr; (b) desorption at 313 K and 21 Torr using a pressure jump of -2 Torr. Am (0, left-hand scale) denotes weight changes of the sample andp,(A, right-hand scale) denotes vapour pressure.2686 SORPTION OF WATER VAPOUR BY SILICA GEL 0.4 0.3 .- .'; 0.2 1 Y 0.1 0 0 0 4 I I I I 10 20 30 40 50 60 p,lTorr Fig.3. Experimental rate constants for sorption ( x ) and desorption (0) at 3 13 K as a function of vapour pressure, using a particle-size fraction with rp = 0.5-1 mm. The dotted curve refers to a calculated curve with rp = 0.70 mm and D = 8 x m2 s-l. The other curves are calculated for four different values of the diffusivity [(a) 00, (b) 15 x 10-lo, (c) 8 x 10-lo and (d) 5 x m2 s-l] using a particle radius of 1.126 mm. where D is the diffusion constant, rp is the particle radius and cc and j? are two dimensionless constants that depend on various material properties, i.e.a = 3hrp/pc, D ( 3 4 /3 = $(&) a where = 55.49 mol kgGto. P The qn are the roots of 3/3(qn cot qn - 1) = q i -a. After a brief initial period all terms in the infinite series in eqn (2) have decayed except the first one, and the process exhibits purely exponential behaviour characterized by a rate constant For isothermal diffusion, i.e. when mass transfer is predominant : k = qf D/r& (4) k = n2 Dlr2, ( 5 ) and for pure heat-transfer control 3h a DJ. Y. ANDERSSON, H. BJURSTROM, M. AZOULAY AND B. CARLSSON 2687 0 5 10 15 pC /Torr Fig. 4. Experimental rate constants for desorption at 313 K as a function of vapour pressure using three different particle-size fractions, rp: A, 2.0-2.5 mm; 0, 0.5-1.0 mm and +, 90-100pm.The illustrated curves are calculated using radii as indicated on the graphs. The derivative (aw/aT), is evaluated from a sorption curve (series 3 in table I). The extent to which a relaxation process can be said to follow one of these extremes depends largely on the quantity AH(aw/aT),: when it is large thermal effects will dominate, and when it is small diffusion will dominate. According to Lee and Ruthven, a process should be considered as purely mass-transfer controlled if EVALUATION OF DIFFUSIVITY AND PARTICLE RADIUS In order to compute diffusivities and average particle radii, non-linear curve-fitting was applied to all the experimental rate constant against pressure data in fig. 3 using eqn (4); values of cp and p values were taken from ref.(6). The differential heat of sorption is assumed to be 40 kJ mo1-1 for each calculation (except in fig. 7). Only the heat conduction in the gas surrounding the particles contributes to the heat transfer and so the coefficient h may be set equal to Av/rp, where Av is the thermal conductivity of water vapour [taken from ref. (16)]. m2 s-' and an average particle radius of 1.12 mm were obtained. The calculated value of the effective radius, which is the most sensitive parameter in the curve fitting (see fig. 3), is in reasonable agreement with the estimated particle size (rp = 0.5-1 mm). The rate constant is less sensitive to variations in the diffusion constant, which will create a large uncertainty in the estimated diffusivity A diffusivity of 8 x2688 SORPTION OF WATER VAPOUR BY SILICA GEL 0.5 0 .4 0 * 3 I E: 2 --- Y 0 . 2 0 - 1 0 / / / / / / 8 / 8 / / / 5 . 3 8.7 13.7 30 LO so 60 Pc /Torn Fig. 5. Experimental rate constants for sorption (filled symbols) and for desorption (open symbols) at different pressures and temperatures for the fraction with particle size rp = 0.5-1 mm. The experimental points refer to the pressures of 5.3 (0, a), 8.7 (A, A) and 13.7 (0, .) Torr. The illustrated solid curves are calculated using a particle radius of 1.06 mm, a diffusivity of 8 x loplo m2 s-' and the derivative (aw/aT), obtained from curve-fits of sorption data in fig. 1 (b) to eqn (1). The dashed curve pertains to D = 00 and a pressure of 5.3 Torr (i.e. pure heat-transfer control). value (ca. 50-100%). Variations in (aw/aT), as a result of hysteresis will probably introduce errors in the high-pressure region.From the calculated value of the diffusivity, 8 x lo-'* m2 s-l, the radius of the particles in the different samples in fig. 4 has been estimated by curve-fitting to each of the series of experimental data. For the larger particle sizes the particle radii obtained in this way (1.06 and 1.60 mm) correspond well to those estimated experimentally. For the pulverized sample, which was not spread out as a single layer of particles in the glass pan, the calculated particle radius (717 pm) diverges greatly from the true radius (90-100 pm). Assuming the same diffusivity as above and a radius of 1.06 mm, rate constants have been calculated for the different temperatures and pressures in fig.5. A D = co curve is also shown for the lowest pressure. Good agreement is obtained between theoretical and experimental values at the higher pressures. The deviation found at the lower pressures may be attributed to the temperature dependence of the diffusivity. DISCUSSION RELAXATION CURVES A relaxation curve obtained in a pressure-jump experiment may be divided into two regions, viz. a dynamic and a quasi-steady-state region. The dynamic region occurs immediately after the pressure jump and it is characterized by sharp concentrationJ. Y. ANDERSON, H. BJURSTROM, M. AZOULAY AND B. CARLSSON 2689 0 10 20 30 LO 50 PClTOrr 1 .o 0.8 n 9 0.2 0 10 20 30 LO 50 tlmin Fig. 6. Experimental and calculated relaxation curves during sorption at 3 13 K (a) at a pressure of 21 Torr and (b) at a pressure of 7.7 Torr for the fraction with rp = 0.5-1 mm.The relative weight change (w - wo)/(w, - wo) (left-hand scale) and temperature change in the particle AT (right-hand scale) as a function of time have been calculated for: (-) a particle radius of 1.126 mm and a diffusivity of 8 x 10-lo m2 s-' using values of (aw/aT), obtained from curve-fits of sorption data in fig. 1 (b) to eqn (1): (---) denoted by H, a pure heat-transfer-controlled process, see eqn (6); (---) denoted by D, a pure mass-transfer-controlled process, see eqn ( 5 ) . 88 F A R 12690 SORPTION OF WATER VAPOUR BY SILICA GEL 0 -0,005 cQ (D --. 3 fp, -0.010 -0.015 0 10 20 30 40 50 60 PlTOrr Fig. 7. The derivative (3w/3T), as a function of pressure at different temperatures ("C), evaluated from curve-fits of the sorption (-) and desorption (----) data in fig.1 (b) to eqn (1). gradients in the particle and a fast warming up or cooling down of the particle. In the quasi-steady-state region concentration gradients have been smoothed out in the particle [a concentration dependence of c(r) = A sin ( r / u ) / r is obtained1' in a spherical particle], and the temperature decreases slowly to the equilibrium value. The characteristic parameters of the process decrease exponentially with time, and this later part of the experimental curve can easily be analysed in terms of the Lee and Ruthven model. VALIDITY OF THE LEE AND RUTHVEN MODEL In the model of Lee and Ruthven the heat conduction in the solid particle is assumed to be much more efficient than heat transfer from the surface to the vapour phase (Biot number @ l), and all temperature gradients are accordingly supposed to arise at the exterior of the particle.At the low pressures prevailing in the present case free convection in the gas phase is minimal, and the transfer of heat to the gas can further be assumed to take place by way of conduction in a stagnant gas. Accordingly, the heat-transfer coefficient h can be set equal to 3LV/rp, i.e. Nusselt's number Nu = 2. As the heat conductivity of the silica-gel particle7 is 0.4-0.6 W K-l m-l while the heat conductivity of the water vapour16 is 0.02 W K-l m-l, the Biot number for heat transfer, BiH, becomes ca. 0.05, and the particles can be assumed isothermal. The diffusivity of water in silica gel may be explained by several microlevel processes : Knudsen diffusion in micropores, surface diffusion by molecules jumping from one site to a neighbouring site and capillary action (resulting from the balance between adhesion and cohesion forces).The first two processes probably dominate at low and medium water coverage. Note that a diffusivity value of 2 x lo-' m2 s-' applies for a pure Knudsen diffusion. The diffusion coefficient for water has been foundJ. Y. ANDERSON, H. BJURSTROM, M. AZOULAY AND B. CARLSSON 269 1 by Allanderl* to be 4 x 1O-lo m2 s-l at 293 K, for a silica gel characterized by (p0 = 1370 kg mp3 and w = 0.55) at saturation, using therrnogravimetry under con- tinuous flow of air containing water vapour at various partial pressures. Tempelhoff and Feldmann,19 in their inelastic neutron scattering studies, obtained a diffusion coefficient 25 x m2 s-l at 298 K and with a degree of coverage, 8, equal to four for a silica gel with almost the same pore characteristics as in the present case.Comparing the lifetimes given by Tempelhoff and Feldmann for the water-silica complexes at different degrees of coverage, and assuming a constant jump length, one can calculate a diffusion constant of 13 x m2 s-l at 8 = 1 . This is of the same order of magnitude as the value of 8 x m2 s-l obtained in the present study by fitting the rate-constant curve. In these calculations the diffusivity has been assumed to be independent of the water content of the gel, which is a questionable hypothesis for a wide pressure range.However, only rate constants measured at low pressures (pc < 15 Torr) contribute significantly to the calculated diffusivity [cf. fig. 3(b)], and the value obtained is thus an average for the low pressure or low water-content range. Furthermore, diffusivities are temperature-dependent and vary in the course of a reaction as a result of self-cooling or -heating, so that an averaged value is obtained. RELATIVE IMPORTANCE OF HEAT AND MASS TRANSFER Fig. 6(a) and (b) show two typical relaxation curves, together with simulated curves obtained from previously determined diffusivity and particle size and the Lee and Ruthven equations. Curves for the cases controlled by mass transfer and heat transfer, respectively, have been included for the sake of comparison.The curve in fig. 6(a) represents almost pure heat-transfer control, while that in fig. 6(b) corresponds to the mixed region. It is evident that both curves display very brief dynamic intervals; initially the temperature increases very quickly until the quasi-steady state is attained. Fig 6 ( a ) and (b) illustrate the relative importance of heat- and mass-transfer. In the pressure range 15 <p,/Torr < 40 it is found that alp < 3, which implies a pure heat-transfer control. At pc < 15 Torr both mass- and heat-transfer contribute to the rate constant. According to eqn ( 7 ) alp is sensitive to variations in (aw/dT),, which is largely responsible for the pressure and temperature dependence of the rate constant. Regarding the temperature dependence (see fig.5), note that an activation energy calculated from the rate constant against temperature data has no relevance, since it is governed mainly by the temperature dependence of (aw/aT),. As hysteresis effects may interfere, this derivative may not be constant, despite the temperature and pressure being constant. When the desorption branch is followed, (awl8 T)P is larger than for the corresponding sorption branch. This may explain the lower rate constants found for desorption in the medium-pressure region, where heat-transfer limitation occurs (cf. fig. 3 and 7). NOTATION constants depending on boundary values (i = 1-6) constants used in eqn (1) the Biot number for heat transfer heat capacity of silica gel diffusivity of water in silica gel activation energy of diffusion heat-transfer coefficient between particle and vapour differential enthalpy of sorption J kg-l K-l m2 s-l J mol-1 W K-l m-2 J mol-l 88-22692 SORPTION OF WATER VAPOUR BY SILICA GEL AHo enthalpy of vaporization for water J mo1-I rate constant equilibrium pressure water saturation pressure constraint pressure constraint pressure prior to a pressure jump distance to the centre of the spherical particle particle radius gas constant (n = 1,2, ...) nth root of the equation 3P(qn cot qn- 1) = 4: time half-value time temperature mass of water per mass of dry silica gel w a t t = O w a t t = c o dimensionless constants [eqn 3(a, 6)] degree of coverage heat conductivity of the solid particle heat conductivity of water vapour chemical potential difference density of silica gel constant = 55.49 min-l Torr Torr Torr Torr m m J molt1 K-l -a S S K g 8-l g 8-l g g-' This work has been financed by The National Swedish Board for Technical Development (S.T.U.), The Swedish Council for Building Research (B.F.R.) and The Swedish Natural Science Research Council (N.F.R.).We thank the N.F.R. for financial support to Joan de Pablo during his stay at the Department of Physical Chemistry of the Royal Institute of Technology, Stockholm, during which he assisted in some of the experiments. See e.g. Proc. Znt. Seminar Thermochemical Energy Storage, Stockholm 1980 [Swedish Council for Building Research (B.F.R.), Stockholm, 19801, document no. BFR D25; 1980. Reactions in the Solid State, vol. 22 of Comprehensitle Chemical Kinetics, ed. C. H. Bamford and C. F. H. Tipper (Elsevier, Amsterdam, 1980). G. Bertrand, M. Comperat, M. Lallemant and G. Wattelle, J . Solid State Chem., 1980, 52, 57. J. Anderson, J. de Pablo and M. Azoulay, Thermochim. Acta, 1983, 70, 291. R. K. Iler, The Chemistry of Silica (Wiley, New York, 1979). H. Bjurstrom and B. Carlsson, Proc. 3rd Multiphase Flow and Heat Transfer Symposium/ Workshop, Miami Beach, Florida, 1983 (Elsevier, Amsterdam, 1984), vol. B. p. 71 1. H. Bjurstrom, E. Karawacki and B. Carlsson, Znt. J . Heal Mas.\ Tramfer, in press. S. H. Jury and H. R. Edwards, Can. J . Chem. Eng., 1971, 49, 663. * S . H. Hubbard, Znd. Eng. Chem., 1954, 46, 356. lo L. K. Lee and D. M. Ruthven, J . Chem. Soc., Faraday Trans. 1, 1979, 75, 2406. l 1 W. R. Grace Ltd, manufacturer's information. l 2 S. Brunauer, R. Sh. Mikhail and E. E. Bodor. J . Colloid Interface Sci., 1976, 24, 451. l3 F. Lavanant, Thesis (Universite de Dijon, 1963). l4 M. M. Dubinin, J . Colloid Interface Sci., 1967, 23, 487. l6 Handbook of Chemistry and Physics, ed. R. C. Weast (Chemical Rubber Co., West Palm Beach, l 7 J. Crank, The Mathematics qf Dzflusion, (Oxford University Press, Oxford, 2nd edn, 1975). l9 K. Tempelhoff and K. Feldmann, 2. Phys. Chem. (Leipzig), 1975, 256, 369. 0. M. Dzhigit, A. V. Kiselev and G. G. Muttik, Kolloidn. Zh.. 1962, 24, 15. Florida, 58th edn, 1977). C . G. Allander, Trans. R. Inst. Technol., Stockholm, 1953, no. 80. (PAPER 411971)

ISSN:0300-9599    

DOI:10.1039/F19858102681

出版商:RSC    

年代:1985

数据来源: RSC

摘要:

J . Chem. Soc., Faraday Trans. I , 1985, 81, 2693-2702 Apparent Molar Volumes for Aqueous Solutions of the Homologous Series of a-Dodecyl-co- hydroxypoly( ox ye thylene) Surfac tan ts, C,,EO,. (j = 4,5,6,8 and 12), and of C,,EO, BY THELMA M. HERRINGTON* AND S. S. SAHI Department of Chemistry, University of Reading, Reading RG6 2AD AND CHRISTINE A. LENG Unilever Research Port Sunlight Laboratories, Quarry Road East, Bebington, Wirral, Merseyside L63 3JW Received 21st November, 1984 Apparent molar volume data for aqueous solutions of the series of a-dodecyl-o- hydroxypoly(oxyethy1ene) surfactants, C,,EO,(j = 4,5,6,8 and 12), and of C,,EO, are reported in the temperature range &70 "C. Theories of dilute solutions are applied to solute-solvent interactions assuming a rigid-particle model for the repulsive potential.Trends in the temperature dependence of the attraction are expressed in terms of the square-well potential for the micelle-water interaction and related to the dehydration of the micelles. Previous studiesll of the partial molar volumes of aqueous solutions of non-ionic surface-active agents have been concerned with the change in the partial molar volume of the surfactant molecule on micellization and analysing the data in terms of a mass-action model. However, for n-decyldimethylamine oxide the c.m.c. is ca. 0.02 mol kg-l at 25 "C and so it is difficult to obtain reliable thermodynamic data in the premicellar region and to derive meaningful aggregation numbers from the mass-action model. The non-ionic surfactants considered in this paper have even lower c.m.c.mol kg-l or less); also the c.m.c. is very sharp and they aggregate to form larger micelles with aggregation numbers of 30 or more. It was thus decided to use density data determined in the micellar region to investigate micelle-micelle and micelle-solvent interaction. In the model used to analyse the data the micellar aggregation number is assumed to be independent of concentration and deviations from ideality are analysed in terms of virial coefficients. These non-ionic surfactants form isotropic micellar solutions in a given temperature range with liquid-crystalline phases frequently occurring at higher concentrations. However, all are unusual in having a lower consolute temperature ; when the temperature is increased the solution becomes turbid in a narrow temperature range, called the cloud point.Above the cloud point the system separates into two isotropic solutions. The temperature at which phase separation occurs is a function of the amphiphile concentration. The minimum of the consolution curve is a critical point; the temperature and concentration at the minimum are called the critical temperature, T,, and the critical concentration, C,. As the cloud point is approached it is anticipated that micelle-micelle attraction should increase together with dehydration of the surfactant molecules. 26932694 MOLAR VOLUMES OF POLY(OXYETHYLENE) SURFACTANTS EXPERIMENTAL The surfactants used were a-dodecyl-co-hydroxytetrakis(oxyethy1ene) (C,,H,,[OCH,CH,], OH, abbreviated to C12E04), a-dodecyl-o-hydroxypentakis (oxyethylene) (C,,EO,), a-dodecyl- o-hydroxyhexakis(oxyethy1ene) (C,,EO,), a-dodecyl-whydroxyoctakis(oxyethy1ene) (C,,EO,), a-dodecyl-whydroxydodecakis(oxyethy1ene) (C,,EO,,) and a-decyl-cu-hydroxy- hexakis(oxyethy1enej (C,,EO,).The surfactants C,,EO,, C,,EO,, C,,EO,, and C,,EO, were prepared by synthesis. Their purity was estimated by gas and liquid chromatography as > 97% for C,,EO, and 99% for the rest. C,,EO, and C,,EO, were purchased from Nikkol Chemical (Japan). They were stated to be of 98 % purity; it was verified by gas and liquid chromatography that the samples contained no detectable methoxypolyethyleneglycol or lauryl alcohol. The C and H analyses agreed to better than 1 % with the theoretical values and the cloud points were in agreement with those of previous worker^.^^ Doubly-distilled deionized water was used in the preparation of all solutions.The distilled water was outgassed on a water pump and then further degassed by keeping at 10 "C above thermostat temperature for the volume measurements. Solutions were made up by weight.to give a precision of 40.1 mg. Densities were determined at 0, 25, 40, 50 and 70 "C using an Anton Parr density meter (DMA 601). The instrument was calibrated using water, air and KC1 solutions of known density. The density of water was taken from the compilation of Kel15 and the densities of KCl solutions from ref. (6) and (7). THEORY Solutions of non-ionic surface-active agents behave in a non-ideal manner, first because of the formation of micelles and then because of the effect of micelle-micelle and micelle-solvent interactions.The surfactants considered in this paper have a sharp c.m.c. and the micelles are assumed to be monodisperse with a single aggregation number independent of concentration.* By considering a single-phase micellar solution to be a two-component system where the solvent consists of water plus monomer at the c.m.c. and the micelles are the solute, the theories of McMillan and Mayerg and of Hilllo enable the micelle-micelle and micelle-solvent interactions to be calculated. According to the McMillan-Mayer theory for a solution of a solute in a solvent, the osmotic pressure n is given as a power series in the number density (the notation is defined at the end of the paper): Thus in this model the non-ideal behaviour of the system is characterized by the virial coefficients B6, etc.In dilute solution B*,, w B;; = -biz, the solute-solute cluster integral. The micellar molar mass is assumed to be independent of concentration. The postulate of a non-concentration-dependent aggregation number is consistent with the sharp c.m.c. found for the C,,EO, series. In the classic paper of Corkill and Walkerll the thermodynamic behaviour of C,,EO, with a sharp c.m.c. is contrasted with that of octylmethyl sulphoxide with a broad and ill-defined c.m.c., and it is concluded that the former compound forms a micelle of fixed aggregation number whereas the latter has a micelle of concentration-dependent aggregation number and the properties of the system are analysed in terms of a mass-action model. The apparent molar volume may be calculated from the solution density usingT. M.HERRINGTON, S. S. SAHI AND C. A. LENG 2695 The partial molar volumes of micelles, b, and solvent, 4, can be calculated from the apparent molar volume using y~ = 4 v- m, 84 v/am, V, = W- kf, mh aW/am,. (3) (4) In the above equations m, is the molality of the micelles, q is the micellar aggregation number and p i is the density of the surfactant monomer solution at the c.m.c. For these surfactants the c.m.c. are of the order of lop4 mol kg-l or less, except for C,,EO,, and as the lowest molalities used are ca. lo3 greater than this the corrections for monomer molality are of the same order as the experimental error. Now by,, the solute-solvent cluster integral, is related to the partial molecular volume of solute at infinite dilution by byl = -@+kTK.( 5 ) The solute-solvent interaction, NB:: (where (5) and by, is given by = - !&), can be calculated from eqn byl = -4nJom [l-exp(-col1/kT)]r2dr where d1 is the potential of mean force between one molecule of surfactant in a micelle and one of solvent in the pure solvent (including averaging of the force over all rotational coordinates) and r is the distance apart of the centre of the molecules. l2 Here the pure solvent is water plus surfactant monomer at the c.m.c., in a molecular ratio of ca. lo5: 1. RESULTS The apparent molecular volume with a micellar surfactant molecule was calculated from the solution density using 1 1 w=- --- +M,/p, md, 3 (7) where m is the molality of the surfactant monomer in the form of micelles and pi is the density of the surfactant solution at the c.m.c.(c.m.c. and p: values are given in table 1). The apparent molar volumes obtained from these densities were fitted to the equation where V p is the partial molar volume of the micellar surfactant molecule at infinite dilution. The values of V p and A are given in table 2. 4V= Vp+Am (8) DISCUSSION The value of V p in table 2 is the average partial molar volume at infinite dilution of each surfactant molecule in a micelle. It can be seen that each additional -CH2CH20- group contributes on average 38 cm3 mol-l to V p and this is not markedly temperature dependent. For CloEO, at 0 "C, V p is only 2 cm3 mol-l greater than that for C,, EO, and 36 cm3 mol-1 less than that for C,, EO, suggesting that each additional -CH2CH2- group contributes ca.36 cm3 mol-1 at 0 "C; the contribution2696 MOLAR VOLUMES OF POLY(OXYETHYLENE) SURFACTANTS Table 1. Monomer concentration and ‘solvent’ density at the c.m.c. mxa monomer concentration surfact ant e/oc /mol kg-’ / 1 0-5 mol dm-3 P:/g cm-3 C12E04 0 2.5 L, ‘1ZE05 0 1.0 H, 25 3.0 L, C12E06 0 1.3 H, 25 1.3 H, 40 3.3 L, C12E08 0 0.8 I, 25 1.1 H, 40 1.1 H, 50 1.0 H, 70 - 0 > 0.5 I, 0 1.7 H, 25 1.9 H, 40 50 - - 6.3 9.5 6.5 9.7 6.8 4.8 10.2 7.1 5.2 4.0b 2.6b 16.7 136 90 60 41 0.999 840 0.999 840 0.997 043 0.999 841 0.997 045 0.992 220 0.999 843 0.997 046 0.992 216 0.988 036 0.977 770 0.999 849 0.999 864 0.997 059 0.992 2 18 0.988 036 a The molality at the onset of the most dilute liquid crystalline phase.Extrapolated values. Table 2. Coefficients of the equation OV = V p +Am standard deviation A in 4V molality range /mol kg-l V? surfactan t e/”C /cm3 mo1-I /kg cm3 mo1-2 /cm3 mol-1 C12E04 C12E06 0 0 25 0 25 40 25 40 50 70 ‘lZE08 0 C12EOl2 0 C10E06 0 25 40 50 362.08 402.11 410.88 440.60 450.98 456. I4 51 1.45 524.15 534.78 540.25 554.06 655.62 404.38 41 6.42 422.72 424.75 -0.167 - 1.69 1.68 -4.18 - 1.31 - 0.05 - 8.85 - 4.0 1 - 1.79 - 1.20 1.95 - 16.25 0.932 0.383 - 5.33 -6.91 0.01 0.37 0.43 0.05 0.22 0.04 0.44 0.0 I 0.05 0.15 0.30 0.0 1 0.0 1 0.06 0.30 0.46 0.2 -+ 1.2 0.2 -+ 0.7 0.2 + 0.9 0.1 4 0.7 0.1 + 0.7 0.1 + 0.7 0.05 + 0.35 0.05 + 0.4 0.05 + 0.4 0.05 --* 0.4 0.05 + 0.4 0.2 + 0.5 0.2 --* 0.6 0.2 + 0.6 0.2 + 0.6 0.2 --+ 0.6T.M. HERRINGTON, S. S. SAHI AND C. A. LENG 2697 Table 3. Attractive contribution to the solute-solvent interaction coefficient sucrose + water 5 25 50 70 C,,EO, + water C,,EO, + water C,,EO, + water C,,EO, + water 0 0 25 0 25 40 0 25 40 50 70 165&62 111+35 112+ 10 99+4 103k 19 110+11 38f2 39+ 1 39k 1 39+ 1 42+3 206.48 210.38 213.75 215.68 360.94 400.97 409.77 439.46 449.87 454.99 510.31 523.04 533.63 539.08 552.79 477 477 477 477 423 482 48 1 534 533 53 1 670 669 669 669 666 270 266 262 260 61 80 70 94 82 75 159 145 135 129 112 a From ref. (8). is 34.6 and 33.4 cm3 mol-1 at 25 and 40 "C, respectively, from the values of V p for C,, EO, and C,, EO,. An average increment of 34.6 cm3 mol-l per -CH2CH2- group at 25 "C was found by Corkill et aZ., for the surfactants RNMe3Br (R = C,-c,,), RS0,Na (R = Clo-C14) and RSO(CH,),OH (R = C, or C,, n = 2,3 or 4).The value for each -CH2CH2- group may be compared with that from the series of n- alkyldimethylamine oxides in water [C,(CH,),NO, where n = 4,6,7 and lo];, this series exhibits a gradation in properties from C,(CH,),NO, which behaves like a medium-sized alcohol in water, to C,,(CH,),NO, which has a c.m.c. at 0.02 mol kg-l; the data were analysed by a mass action model and in the premicellar region the monomers had a value of V p that was ca. 5 cm3 mol-1 less than that of a micellar monomer. Each -CH2CH2- group contributed ca. 32 cm3 mol-l in the premicellar region and 33 cm3 mol-1 when in the form of micelles at 25 "C. However, the aggregation numbers of the micelles formed are much smaller than for the C,, EO, series (see table 3), e.g.q = 22 for C,,(CH,),NO and 13 for C,(CH,),NO. Certainly in the C,,EO, series the alkyl chain in the micelle interior appears to be more expanded than in the liquid form since the n-alkanes have increments of 33.4 cm3 mo1-l per -CH2CH2- for chain lengths C,-c,4 at 25 "C.13 The values for the solute-solvent interaction NBr: calculated from eqn (5) are given in table 3. Compressibility data for water were taken from the compilation of Bradley and Pitzer.14 The cluster integral b:, consists of two parts, an attractive and a repulsive contribution. The following treatment, previously used for aqueous solutions of non-electrolytes, l5 can be usefully applied here. Let R be the distance of closest approach of the surfactant micelle and solvent2698 MOLAR VOLUMES OF POLY(OXYETHYLENE) SURFACTANTS molecule, then repulsion exists for values of r < R and attraction for r > R, and the integral can be split into repulsive and attractive parts as follows: B,*," = 4n J [I -exp(-wl1/kT)]r2dr+4n J [I -exp(-cu11/kT)]r2dr (9) 0 R = S,,+@f, where S,, is the repulsive and @fl the attractive contribution and d1 has the appropriate value for the range of integration. If the form of the potential function 0.9, is known then the integration can be performed to yield B;",".The simplest potential function regards the micelle and water molecule as rigid spheres. For two hard spheres of diameters R, and R, (1 1) The water molecule can be considered to be a sphere of diameter 3.04 A (Corey-Pauling models).The micellar volume (and hence diameter) was calculated from the micellar mass using the value of the aggregation number in table 3 and assuming the density to be 1 g ~ m - ~ . For example, for C,, EO,, R, = 57.5 and the repulsive contribution to B:;" is NS,, = 7.0 x lo4 cm3 mol-l and NS,,/q = 423 cm3 mol-l for each monomer in a micelle. So that by difference from eqn (10) at 0 "C we have N@&/q = -61 cm3 mol-l. Values for the other surfactants are given in table 3, where data for sucrose + water are also given.I6 NS,,, the hard-sphere repulsive contribution to the micelle-water interaction, is relatively insensitive to the choice of aggregation number q, within a significant range, e.g. & 30% .NS,, is equal to the excluded volume of the micelle and water molecule, i.e. the volume excluded by the micelle and water sheath, and as R, %. R, this is fairly close to the micellar volume, although this approximation would be too drastic for the calculation of S,,. The effect of the aggregation number on the ratio NS,,/q is given by (12) For a micellar density of 1 g ~ m - ~ , R;/q is a constant, so that for fixed R , the q dependence of N S l l / q , the hard-sphere contribution for each monomer in a micelle, varies as the reciprocal of the micellar radius or as q1'3, resulting in a small effect. For example, for C,, EO, at 0 "C, q = 165 62 corresponding to values of NS,,/q of 423 cm3 mol-1 (q = 165), 408 cm3 mol-1 (q = 227) and 434 cm3 moP1 (q = 103); doubling q to 76 for C,, EO, at 0 "C gives NS,,/q as 640 cm3 mol-l compared with 670 cm3 mol-1 for q = 38.The aggregation numbers presented by Herrington et al., for the other C,, EO, surfactants show far less variation than for C,, EO,. However, the key point is that the results presented for NS,,/q (and consequently -N@P,/q) are not highly dependent on these particular aggregation numbers. Furthermore, including water bound to ethylene oxide groups is not appropriate in the calculation of water-ethylene oxide interaction presented here as V p itself includes solute-solvent interactions. The second question in our data analysis is the assumption of spherical micelles in the calculation of the repulsive contribution to the virial coefficient. For a value of q > 56 the hydrocarbon core is beyond the fully extended length of a C,, chain so that the micelle is not ~pherica1.l~ The greatest departure from sphericity would be expected for C,,EO,; however, the 'hard core' or excluded volume virial coefficient is also relatively insensitive to changes in shape for given q, and within the n S,, = 6 (R, + R,)3.NS,,/q = (R, + R,)3/q = Ri/q + 3Rg R,/q f . . . .T. M. HERRINGTON, S. S. SAHI AND C. A. LENG 2699 experimental error of our aggregation number. If other shapes are considered, such as a discoid lamellar phase cut-out or an oblate ellipsoid, the virial coefficients are 420 and 424 compared with 423 cm3 mol-l for a sphere. This is easily understood by the fact that the values of the Kihara shape factorfare 1.21 and 1.02, respectively, for these shapes (f= 1 for a sphere). [The molecular dimensions assumed were 16.6 A for the fully extended length of the hydrocarbon chain,' and 7.1 A for the length of the (EO), polyoxyethylene chain.,,] The value found for the virial coefficient NS,, / q was remarkably insensitive to the choice of molecular dimensions.For a discoid sha e respectively, and for an oblate spheroid from 430 to 423 cm3 mol-1 for minor axis values of 30 and 60 A, respectively. The definition S = 4uf for the second virial coefficient of rigid non-spherical molecule^^^ implies that the volume u, but not the detailed shape of the micelle, is important; f z 1 as long as the molecule is not highly asymmetric. The micelle-water attractive contribution to the virial coefficient in table 3 is expressed per monomer in the micelle, i.e. as --N@f,/q.A linear least-squares fit to a plot of - N@f,/q against j for j = 4,5,6 and 8 at 0 "C gives it varied from 417 to 423 cm3 mol-1 for extreme disc heights of 30 and 60 x , - N(Dfl/q = (25 & 3)j- (43 & 18) (13) and for j = 5,6 and 8 at 25 "C - N@fl/q = (26 & 5 ) j - (65 & 3 I). (14) The negative intercepts a t j = 0 imply that for dodecan-1-01 the attraction has reversed sign, i.e. the C,, H,,OH + H,O virial coefficient is entirely repulsive. The attraction is zero for j 2 2, which is in agreement with the solution behaviour of C,,EOj surfactants.20 C , , EO, is only slightly water soluble between 0 and 25 "C and f o r j < 3 these surfactants are insoluble in water. The attractive contribution to the virial coefficient may be related to the micelle-water interaction in terms of the square-well potential.This is a simple potential function, specified by the hard-core diameter, CT, the extent of the well, 10, and attractive well-depth, - E , and is suitable for discussing trends in behaviour. For the square-well potential - @f, = S,,(A3 - 1) [exp ( e / k T ) - I]. (15) The parameter 3, for the well-width makes this potential more suitable for systems with large solute entities in a solvent with small molecules than, say, the Lennard-Jones interaction. For simple liquids II = 1.5 is used, but for a small water molecule interacting with a micelle of radius 20-30 A values of 1 . 1 or 1.01 are more appropriate. Because the attractive well is described by two parameters it is necessary to fix one and calculate changes in the other.For micellar systems it seems appropriate to hold the width constant and describe hydration by a reduction in the well-depth. The micelle may be considered to have a monolayer hydration shell in which the number of water molecules decreases with increasing temperature. However, the trends calculated do not depend on this molecular interpretation. In table 4 values of e/kT are given for C,, EO, and C,, EO, as these data span the widest temperature intervals. 1 values of 1.01 and I . 1 are compared, corresponding to a water molecule a distance 0.29 and 2.9 A from C,, EO, and 0.32 and 3.2 A from C,, EO, and C,, EO,, respectively. It can be seen that E/kT is very sensitive to the choice of 2..If we assume that the water molecule is very close to the micelle, then for C,, EO, an increase in temperature from 0 to 40 "C yields a 20% decrease in - N@f,/q corresponding to a 10% reduction in the reduced well-depth e* = e/kT, and2700 MOLAR VOLUMES OF POLY(OXYETHYLENE) SURFACTANTS Table 4. Analysis of the attractive contribution to the virial coefficient in terms of the square-well potential C,,EO, + water 0 40 0 40 C,,EO, + water 0 70 0 70 1.01 1.01 1.1 1 . 1 1.01 I .01 1.1 1.1 94 75 94 75 159 113 I59 113 1.9 1.7 0.43 0.36 2.2 1.9 0.54 0.41 for C,, EO, a 30% decrease in the attractive potential is equivalent to a 14:< drop in E * . This is not just a temperature effect; in each case ~ / k has increased but at a slower rate than kT.In order to relate changes in c* to changes in hydration, we consider the cluster number n, defined by (16) where rmax is the separation of the particles at the first maximum in r2g(r); n gives a measure of the number of nearest neighbours surrounding a central particle.21 To first-order perturbation theory22 n = 8np J;max g(r) r2 dr Little work has been done on square-well potentials for values other than 1.522>23 so that g l ( r ) is not known. If as a first approximation we take the reduction in E/kT to be proportional to the dehydration, we see that for both C,, EO, and C,, EO, with a R value of 1.01 the decrease in E* for the micelle-solvent interaction calculated from the density measurements agrees very well with the 10% increase in aggregation numbers over the same temperature range obtained by diff'erent methods.The aggregation number may increase with dehydration because of the decrease in the head-group area. The small change in aggregation number with increasing temperature supports the choice of 1 as closer to 1.01 than 1.1. A feature common to the phase diagrams of these four C,, EOj surfactants is the cloud point, a lower consolute temperature. At the cloud point two phases are formed; the micelle-micelle attraction is so great that a surfactant-rich phase separates from a dilute surfactant phase. Where the head group is highly hydrated, as in binary aqueous mixtures of sucrose laurate or sucrose oleate,2*i a cloud point is not observed. For the C,, EOj series, the magnitude of the surfactant-water attractive contribution increases asjincreases at any given temperature.C,, EO, has an attractive contribution considerably less than that of sucrose, although the ether linkage is usually considered to have only approximately one-half the hydrogen-bond-forming capability of the hydroxy group (sucrose probably forms ca. 5 hydrogen bonds with water). Thus the overall picture presented is that dehydration of the polyoxyethyleneglycol molecule occurs with increasing temperature.T. M. HERRINGTON, S. S. SAHI AND C. A. LENG 270 1 We thank the S.E.R.C. for the award of a studentship (to S.S.S.) and Tate and Lyle Industries Ltd for generous support. We also thank John C. Schofield for helping with the density measurements on C,,EO,. GLOSSARY OF SYMBOLS cluster integral for one molecule of solute and one of solvent in pure solvent integral defining the interaction between solute and solvent critical concentration at T, radial distribution function Gibbs energy ethylene oxide content in C,, EO, Boltzmann's constant mole ratio of solute to solvent (N2/N1) molality of solute molality of micelles molar mass of solvent molar mass of solute cluster number number of moles of solute Avogadro's constant number of solvent molecules gas constant hard-sphere diameter of solvent molecule hard-sphere diameter of solute molecule repulsive contribution to the cluster integral absolute temperature critical or lower consolute temperature partial molecular volume of solvent molecular volume of pure solvent partial molecular volume of solute at infinite dilution total volume of solution partial molar volume of solute at infinite dilution apparent molar volume square-well depth micellar aggregation n um ber isothermal compressibility coefficient of solvent parameter defining width of square-well potential molecular chemical potential of pure solvent molecular chemical potential of solvent in standard state number density of solute (n2/V> density of solution at c.m.c.density of solution hard-core diameter attractive contribution to the configuration integral potential of mean force J. E. Desnoyers, G . Caron, R. DeLisi, D. Roberts, A. Roux and G. Perron, J . Phys. Chem., 1983,87, 1397. J. M. Corkill, J. F. Goodman and T. Walker, Trans. Furuday SOC., 1967, 63, 768. F. Harusawa, S. Nakamura and T. Mitsui, Colloid Polym. Sci., 1974, 252, 613. J. S. Clunie, J. F. Goodman and P. C. Symons, Trans. Furuduy Soc.. 1969, 65, 287.2702 MOLAR VOLUMES OF POLY(OXYETHYLENE) SURFACTANTS G. S. Kell, J. Chem. Eng. Data, 1975, 20, 92. L. A. Dunn, Trans. Faraday SOC., 1968, 64, 2951. T. M. Herrington and R. J. Jackson, int. Sugar J., 1983,85, 364. * T. M. Herrington, C. A. Leng and S. S. Sahi, Colloids and Surfaces, in press. W. G. McMillan and J. E. Mayer, J . Chem. Phys., 1945, 13, 276. lo T. L. Hill, J. Am. Chem. SOC., 1957, 79, 4885. l 1 J. M. Corkill and T. Walker, J . Colloid interface Sci., 1972, 39, 62. l 2 J. E. Garrod and T. M. Herrington, J . Phys. Chem., 1969, 73, 1877. l 3 M. L. Huggins, J . Am. Chem. Soc., 1941, 63, 116. l4 D. J. Bradley and K. S. Pitzer, J . Phys. Chem., 1979, 83, 1599. l 5 T. M. Herrington, A. D. Pethybridge, B. A. Parkin and M. G . Roffey, J . Chem. SOC., Faraday Trans. l6 T. M. Herrington and E. L. Mole, J . Chem. SOC., Faraday Trans. 1, 1982,78, 213. l7 C. Tanford, The Hydrophobic Efect (Wiley-Interscience, New York 1980). P-G. Nilsson, H. Wennerstrom and B. Lindman, J . Phys. Chem., 1983, 87, 1377. l9 A. Isihara, J . Chem. Phys., 1950, 18, 1446. 2o D. J. Mitchell, G. J. T. Tiddy, L. Waring, T. Bostock and M. P. McDonald, J. Chem. SOC., Faraday 21 C. N. Pings, Physicsof Simple Liquids, ed. H. N. V. Temperley, J. S. Rowlinson and G. S. Rushbrooke 22 J. A. Barker and D. Henderson, Ret;. Mod. Phys., 1976,48, 587. 23 D. Henderson, 0. H. Scalise and W. R. Smith, J . Chem. Phys., 1980,72, 2431. 24 T. M. Herrington and S. S. Sahi, in preparation. I , 1983, 79, 845. Trans. I , 1983, 70, 975. (North Holland, Amsterdam, 1968). (PAPER 4/ 1984)

ISSN:0300-9599    

DOI:10.1039/F19858102693

出版商:RSC    

年代:1985

数据来源: RSC

摘要:

J . Chem. SOC., Faraday Trans. I , 1985, 81, 2703-2710 Relationship between the Entropy of Transfer of a Solute and the Thermodynamic Functions of Mixed Solvents BY EAMONN DE VALERA, DAVID FEAKINS* AND W. EARLE WAGHORNE* Department of Chemistry, University College, Belfield, Dublin 4, Ireland Received 3rd December, 1984 Tne entry of a solute into a solvent will affect the interactions between solvent molecules by making, breaking, strengthening or weakening solvent-solvent bonds. In a binary mixed solvent the relative partial molar entropies, si, and excess entropies of mixing, ASE, are related to these interactions. Thus a relationship between the entropy of transfer, A S P , of a solute and these thermodynamic parameters might be expected. A general relationship is developed and successfully applied to the A S P for alkali-metal halides in methanol-water mixtures.The results of this analysis show that, for this system at least, A S P is dominated by the effects of changes in solvent-solvent interactions, in contrast to the enthalpy of transfer, AH?. Combination of the results of this analysis with those previously reported for AH? allows a complete, quantitative explanation of the thermodynamics of solvation in the alkali-metal halides in methanol-water mixtures. Recently' we showed that the enthalpies of transfer, AH?, of a solute in a mixed solvent could be related, quantitatively, to changes in the enthalpies of solvent-solvent bonding in the mixed solvent. This model was found to account satisfactorily for AH? of simple electrolytes in methanol-water mixtures, where they pass through broad maxima as functions of solvent composition, and in acetonitrile-water mixtures, where the (AH?, x,) profiles show sharp minima.In this paper we derive an analogous relationship for the entropy of transfer, ASP, and apply this to data for the alkali-metal halides in methanol-water systems. The variations in the ASP data with solvent composition largely result from the changes in solvent-solvent bonding. The results of this analysis are combined with those from the analysis of the corresponding AH? data and show that the treatment represents satisfactorily the free energies of transfer, AGP. THEORY The derivation of the expression for the entropy of transfer, ASP, is based on the same model as that used to derive the corresponding expression for the enthalpy of transfer,l AH?, and follows the same line of development.Our approach can be summarized as follows. A solute dissolved in a solvent is supposed to occupy a cavity in which n solvent molecules become its nearest neighbours. In forming this cavity each of these n molecules must break some of its solvent-solvent bonds, giving rise to an increase in entropy equal to - m AS*, where a is a fraction of the molar entropy of solvent-solvent bonding, AS*. Secondly the solute may cause a weakening or strengthening of solvent-solvent bonds over a number of molecular diameters. On average N solvent molecules will be affected (note N 2 n). If we postulate that the associated entropy change is 27032704 SOLUTE-SOLVENT ENTROPY EFFECTS proportional to AS* we can set it equal to -/?NAS* where /? is the proportionality constant and is negative if bonds are strengthened.Not all of the solvent-solvent bonds will be affected equally, nor necessarily in the same sense; i.e. some may be strengthened and others weakened. Thuspis the weighted average of the proportionality constants for the various modified interactions. Finally the solute is supposed to interact with the modified solvent, giving rise to an entropy change AS%. Thus the entropy of solvation of a solute in some solvent A, (ASp)A, is given by (ASP), = (ASg)A- <a: n i +FA WA) A ~ A * -k Asss (1) where the superscripts O indicate the values in the pure solvent. ASss accounts for the change in entropy arising from the change in the standard states of the solute.A solute dissolved in a mixture of A with some second solvent B will have nA and nB molecules of A and B, respectively, as nearest neighbours and will, on average, influence the interactions of N A and NB molecules of A and B, respectively. Clearly each of the processes detailed for solvation in A will operate in the mixed solvent and will contribute to the entropy of solvation. There is, however, an additional configurational-entropy change which must be considered in the mixed solvent. This configurational entropy, ASc, arises from the possibility of a difference between the compositions of the solvent near the ion and in the bulk, i.e. from preferential solvation. This has been considered previously2* and is given by ASc = - NA R In (NAINxA) - N , R In ( N B / N x H ) .(ASp)AB = (ASg)AB - (aA nA + P A NA) AS: - (a, nB +pB NB) AS;+ ASc + ASss (2) Thus for the entropy of solvation in the mixed solvent we obtain (3) where the symbols have the obvious meanings. We shall assume that the terms ASss in eqn (1) and (3) are equal if the mole-fractional standard state is adopted for the solutions. The entropy of transfer from A to the mixture of A and B is given by A S P = (ASp)AB-(ASp)A - ( a B ?l~ +& N , ) A s ; -I- (a: +& wA ) ASA* f A s c . (4) Before proceeding we must consider the entropies of solvent-solvent bonding, A s * . This is done in the following way. Consider a reservoir containing a large number of moles of solvent, N ; thus N % 1. The solvent is at equilibrium with its vapour, supposed to be a dilute gas between the molecules of which there are no intermolecular forces.If one mole of the solvent is vaporised reversibly, the process may be supposed to take place in two stages. First the intermolecular bonds are broken to form a hypothetical fluid at the same vapour pressure; the fluid is then expanded to form the dilute gas. The entropy of solvent-solvent bonding in the present problem is then ASo*, defined by A$"* = $ p - S p * . ( 5 ) gy is the molar entropy of the real liquid and Sy* that of the hypothetical fluid. (The hypothetical fluid was incorrectly defined in our earlier paper.) AS"* can be obtainedE. DE VALERA, D. FEAKINS AND W . E. WAGHORNE 2705 entropy of vaporisation to an ideal gas at the vapour pressure of the (6) AS"* = -ASv + R. from the molar liquid, ASv, as We now note that the entropy S o f a mixture of xA moles of A and xB moles of B is given by and the entropy of mixing ASM by S= x A S A + X B S B (7) where SA and SB are the partial molar, and sA and sB the relative partial molar entropies of A and B, respectively.The excess entropy of mixing is given by We can write, by analogy with eqn (7) and ( 5 ) : S* = x A S ~ + X B S ~ (10) and AS: = gi-S: (1 1) thus To make eqn (4) amenable to further development we can introduce the simplifying approximations : n i = n,+n, = n (13) a; = a, = aB = a (1 5 ) We shall make no attempt to justify these approximations in detail, but note that they led to equations which accurately represent the data considered here and, from a preliminary investigation, those for a number of other systems.We introduce the possibility of preferential solvation via and nA xA n B P x B - A value of unity for p (or P) indicates random (i.e. non-preferential) solvation, while values of 0 < p (or P) < 1 and p (or P) > I indicate preferential solvation by A and B, respectively. For the simplest case p = P and is the mean equilibrium constant for the successive2706 SOLUTE-SOLVENT ENTROPY EFFECTS replacement of A by B around the solute and is independent of the solvent composition. Eqn (17) leads to for n A and nB and eqn (18) to analogous equations for N A and NB. Introduction of these into eqn (4) leads to ASP = (ASg)AB-(ASf2)A Finally we must consider the value of (ASg),,.In treating the corresponding enthalpy term1 we made the approximation that (AHg)AB varied linearly with the composition of the solvent in the vicinity of the ion. The same approximation for (ASg)AB leads to Introducing this into eqn (21) leads to (AS% )AB = NA(AS$i )A + NB(ASf2 ) B e (22)E. DE VALERA, D. FEAKINS AND W. E. WAGHORNE 2707 i - 2 Fig. 1. Plots of -298 A S p / x , H , O H against -298 A S ~ / X C H , O H (T in K) in methanol-water solvent systems for: 0, LiC1; 0, NaCl; A, KCl; 0, RbCl and ., CsC1. APPLICATION TO THE METHANOL-WATER SYSTEM Eqn (23) can be applied to A S P data for any solute in any solvent system, provided that the degrees of preferential solvation, p and P, are known. In the methanol-water system there is evidence1 that preferential solvation is weak, i.e.p and P are both 1. If we set p and P to unity eqn (23) reduces to the particularly simple relationship A S P = XB [(AS~),-(AS~),]-(~U?+PN) A S E + x B ( ~ + P N ) ( A p A * -APB*) (24) which we recast in the form ~- As’ = - (m +PN) A S E / ~ B + [(ASg)), - + (m +PN) (As:* - As”,*). (25) xn Fig. 1 shows plots of -298 ASP/xMeoH against -298 ASE/xMeoH for the alkali-metal chlorides in these systems (T is in K). The plots for NaBr and NaI were similar and are omitted for clarity. The A S P data were taken from ref. (4) and the A S E values from combination of the data in ref. (5) and (6). The linearity of the plots in fig. 1, particularly those for the LiCl and NaCl data, is striking and provides strong support for the theory outlined above.There is some curvature in the other plots, but this is accentuated by the form of eqn (25), i.e. by division by xMeOH. In all cases the variations in - 298 A S E / ~ M e O H account for over 98 7; of those in - 298 ASp/xMe0H. Thus the variations in -298 A S P , including the minima (see fig. 1) result simply from the variation in A S E and not, for example, from any additional effects due to changes in the solvent structure in this region. Such changes in solvent structure are,2708 SOLUTE-SOLVENT ENTROPY EFFECTS Table 1. Model parameters for the solvation of alkali-metal halides in methanol-water solvent systems -298 ASPa -298 ASP'b electrolyte /cal mol-l /cal mol-l (an +/?N). (om +pN)Hd LiCl 6903 7565 4.7 5.6 NaCl 7842 709 7 4.7 6.1 KCl 7769 7404 4.3 5.7 RbCl 7082 7309 4.2" 5.6 CSCl - 7541 4.7e 5.9 NaBr 8101 8353 6.3' 8.0 NaI 88 14 960 1 8.0" 10.1 a Experimental values taken from ref.(2). ASP' = (ASE),,,3,,,, - (ASg)H20 +(m fPN) From the slopes of the plots These (AFGo - As&oH), from the intercepts of the plots in fig. 1. in fig. 1. plots show significant non-linearity (cf. fig. 1). Values from the analysis of the corresponding AH? data, from ref. (1). of course, not precluded and may account for the variation in AS", but they do not need to be involved specifically to account for the variations of ASP. Table 1 lists the values of (cvz+BN) obtained from the analysis of these ASP data. In each case this is positive, showing that the net effect of the solute on the solvent is to break solvent-solvent bonds. This result is in agreement with the results of the earlier analysis1 of AH?, but differs from those of studies of other structure-related properties, such as viscosities, which reflect the effects of all of the bonds formed in the system, including those between the solute and solvent.The effect of these solvent-solute bonds is, of course, reflected in the [(ASf$)B-(ASE)A] term in eqn (25)- The agreement between the values of (cvz + P N ) obtained from analyses of the ASP and AH? data for these systems is reasonable. However, the values from the ASP analysis are in all cases lower. Combination of the results of this analysis with those of the corresponding analysis of the AH? data' permits a consideration of all of the transfer parameters of the electrolytes in this solvent system.Fig. 2 shows plots of the experimental values of - 298 ASP, AH? and AGP, the free energy of transfer. for LiCl and NaI ; also shown are the values obtained ziia these analyses. The data for the other electrolytes are similar. The success of the theory in representing the variations in AGP, AH? and A S P with changing solvent composition is remarkable. In the case of LiCl data the agreement between the experimental and calculated values is to within experimental error up to 60 wt:< methanol, although there are relatively small systematic deviations beyond this. The ability of this simple theory to account for all of the transfer parameters in these systems. and for the AH? values of solutes in a variety of aqueous and non-aqueous solvent systems,1' 7 q * indicates that it provides a good basis for understanding the changes in solvation which accompany changes in the solvent.Let us consider briefly the transfer of LiCl from water to methanol in more detail, since these data are in the best agreement with the predictions of the theory. For transfer between pure solvents eqn (25) reduces to ASP = (ASg),-(ASE),\+(ctn+pN)(ASF -AS::) (26)E. DE VALERA, D . FEAKINS AND W. E. WAGHORNE 10 9 8 7 6 5 4 - i d 3 2 2 y 1 E - Y . Q 0 - 1 - 2 - 3 - 4 -5 2709 Fig. 2. Plots of -298 ASP (open), AH? (filled) and AGP (half-filled) for 0, LiCl and A, NaI against wt % methanol for methanol-water solvent systems. Lines represent values calculated via the model (see text). and for AH? we have the corresponding equation:] AH? = ( A H ~ ) - - ( A H ~ ) , + ( ~ ~ + / ~ N ) ( A H " , * - A ~ * ) .(27) Using the appropriate datas we find values of - 298(an + p N ) (AFG,, -AS",*,,,) and (m + P N ) (AT;, - A8",*,,,) of 8058 and - 8940 cal mol-l, respectively,.f. and of 4980 cal rnol-l, respectively. Thus the enthalpic effects of changes in solvent-solvent bonding are largely compensated by the corresponding entropic effects. It is interesting that the contribution of [(AS&,,, - (ASg)H,O] to A S P is relatively small. Thus, in agreement with an earlier s~ggestion,~ A S P is largely determined by changes in the entropy of solvent-solvent bonding, A T * , while AH? reflects changes in both solvent-solvent, AHo*, and solute-solvent, AH$, interactions; AG? results almost entirely from - - 2 9 8 [ ( A ~ g ) , , , H - ( A s ~ ) , p O l and [(AHg)MeOH-(AH~)H,O] of - 138 and ' E. de Valera, D. Feakins and W. E. Waghorne. J . Chem. Soc., Furaday Trans. 1, 1983, 79, 1061. B. G. Cox, W. E. Waghorne and C. K. Pigott, J . Chem. SOC., Faraday Trans. 1, 1979, 75, 227. B. G. Cox and W. E Waghorne, Chem. SOC. Retl., 1980, 9, 381. E. de Valera, D. Feakins and W. E. Waghorne, J. Chem. SOC., Faraday Trans. I, 1980, 76, 560. t 1 cal = 4.184 J.2710 SOLUTE-SOLVENT ENTROPY EFFECTS R. F. Lama and B. C.-Y. Lu, J. Chem. Eng. Data, 1967, 12, 318. F. Goelles, Monatsch Chem., 1961, 92, 981. G. Carthy, D. Feakins and W. E. Waghorne, J . Chem. SOC., Chem. Commun., 1984, 588. Natl Bur. Stand. (U.S.), Circ. 500 (US. Government Printing Office, Washington D.C., 1961), pp. 9 and 103. 'I B. G. Cox and W. E. Waghorne, J. Chem. Soc., Faraday Trans. I , 1984,80, 1267. (PAPER 4/2053)

ISSN:0300-9599    

DOI:10.1039/F19858102703

出版商:RSC    

年代:1985

数据来源: RSC

摘要:

J. Cltern. SOC., Furuday Trans. 1, 1985, 81, 271 1-2722 Metal-organic Chemical Vapour Deposition (MOCVD) of Compound Semiconductors Part 1.-Optimisation of Reactor Design for the Preparation of ZnSe BY J. IWAN DAVIES, GUANGHAN FAN? AND JOHN 0. WILLIAMS* Department of Chemistry, U.M.I.S.T., P.O. Box 58, Manchester M60 1QD Received 5th December, 1984 Optimisation of the reactor design for the preparation of ZnSe by metal-organic chemical vapour deposition (MOCVD) at atmospheric pressure has been achieved using TiO, ‘smoke ’ experiments. Formal expressions derived from gas flow dynamics adequately describe the experimental situation in the laminar flow regime and the ‘stagnant layer’ observed above the hot graphite substrate holder (susceptor) can be treated in terms of boundary layer theory.The required dependences of the stagnant layer thickness (6) on distance along the substrate holder (x), gas velocity (U,) and temperature ( T ) are experimentally observed. Growth of ZnSe from dimethylzinc and H,Se on silica and GaAs( 100) substrates has been achieved in this horizontal reactor at atmospheric pressure in a hydrogen flow. The growth rate (G) has been studied as a function of distance along (x) and across (z) the susceptor and as a function of temperature: G is independent of temperature between 555 and 645 K and is controlled by the diffusion of reactants across the stagnant layer. Over the past ten years there has been increasing interest in the use of metal-organic chemical vapour deposition (MOCVD) for the preparation of thin film structures of metals, semiconductors and insulators, following the pioneering work of Manasevit.’ Since 1980, two international conferences have been held on this topic2? and a variant of the method - metal-organic vapour phase epitaxy (MOVPE) - is capable of preparing ‘ quantum well ’ structures and thin layer (ca.100 %.) ‘ superlattices ’ of Group 111-V compound semiconductors that are equal to or superior in quality to those prepared by molecular beam epitaxy (MBE).4~5 However, studies of the fundamental processes occurring during crystal growth by MOCVD are at a relatively primitive stage and information is currently lacking in relation to both gas-phase and surface reactions. This is particularly the case so far for 11-VI materials which have been less studied than their 111-V counterparts.Epitaxial layers of selected materials such as ZnS and ZnSe have, however, been grown on a variety of single crystalline substrates by MOVPE.6-8 Some of their luminescence, electrical and structural properties have been reportedg* lo but there is little information available on the details of crystal growth and reactor design. In any epitaxial process a knowledge of gas flow dynamics and the parameters that affect the growth is essential if good quality and reproducible layers are to be obtained. An important role of the theory of epitaxy is to predict the conditions within the growth reactor and how these affect characteristics such as thickness, morphology and -f Present address: Changchun Institute of Physics, Chinese Academy of Sciences, People’s Republic of China.271 12712 CHEMICAL VAPOUR DEPOSITION OF SEMICONDUCTORS doping uniformity of the grown layer and the chemical efficiency of the overall process. The required degree of control of parameters that affect the growth, viz. flow rates, concentration of reactants and temperature of substrate, may then be achieved. In this paper we report on the gas flow patterns in a horizontal, cold-walled, atmospheric pressure reactor of the Bass1’ design used for the growth of the 11-VI system ZnSe,-, S, (0 < y < The experimental technique employed TiO, ‘smoke’ and the results are analysed in terms of accepted simplified theories of gas flow dynamics involving temperature, momentum and concentration boundary layers. A model is presented which satisfactorily approximates to the experimental conditions.THEORY GAS FLOW DYNAMICS AND TEMPERATURE DISTRIBUTION In MOVPE it is generally assumed (and indeed there is good evidence available to support this assumption) that the rate of crystal growth is diffusion controlled. The change in the chemical potential ( p ) of the reactant species above the substrate is thus the driving force for the reaction. Knowledge of the flow dynamics and the gas-phase reactions that compete with the surface processes is, therefore, crucial for optimum growth of epitaxial layers. One of the main problems, however, is to find a suitable approach to describe the gas flow patterns within reactors which may vary in dimensions and where concentration and temperature gradients inevitably exist unless carefully controlled.This has been tackled in the case of silicon epitaxy and satisfactory, descriptive models have been obtained based on relatively simple expressions derived from an engineering approach to boundary layer theory. 12-23 It has, however, been pointed out recently for particular systemsz4 that experimental fits to these theoretical equations are not in themselves sufficient proof that the rate- controlling step in the crystal growth process is diffusion across the boundary layer above the substrate. In the case of GaAs growth, for example, the reagents Ga(CH,), and ASH, do not react in the gaseous phase at room temperature and can, therefore, be thoroughly mixed before reaching the deposition zone. The boundary layer thickness can then be greater than the cross-sectional dimension of the reactor and diffusion across a narrow boundary layer is inappropriate to describe the crystal growth process.We shall see later that for the preparation of ZnSe employing the reagents Zn (CH,), and H,Se - two reagents which do react at room temperature and which must, therefore, be separated for as long as possible before introduction into the deposition zone - a boundary layer approach describing the situation, particularly in the entrance zone of the reactor, is appropriate. FLOW CHARACTERISTICS Solutions of the Navier-Stokes equations are generally useful in predicting the nature of local flow and temperature.,, In order to solve these equations we need to consider a number of dimensionless groups.The dynamic behaviour of a gas flowing through a tube at constant temperature may be characterised by the Reynolds number, Re. This is given by where U, is the free stream velocity (m s-l), p is the carrier gas density (kg m-3), h is the height of the channel (m) and p is the dynamic viscosity of the carrier gas (kg m-l s-l). The parameter Re is effectively a ratio of the inertial force to the viscous force and so at very low values (Re 4 1) the viscous forces dominate and the effectsJ. I. DAVIES, G. FAN AND J. 0. WILLIAMS 2713 of internal friction must be considered throughout the medium. At higher numbers (Re 3 1) the zone of flow may be sub-divided into a region of inviscid flow and a viscous boundary layer region. However, the flow is still laminar in nature and the velocity profile is parabolic whereas at much higher values of Re a transition to turbulent flow takes place.Strong mixing now occurs and forces perpendicular to the main flow direction, such as vortices, arise beyond a critical value of Re given by Schlichting,, as Recrtrrr = 2300. If forced flow is absent, then we must consider whether free or forced convection is taking place. Free convection could exist if we have a temperature gradient, for example, between a hot susceptor and a water-cooled wall, in the same direction as the gravity vector. The Rayleigh number, Ra, is given by Ra = gD C P p 2 h 3 ( A T ) / p where g is the gravitational acceleration (m s-,), is the coefficient of thermal expansion (K-l), Cp is the specific heat capacity (J kg-l K-l), AT = Tsusceptor - 300 (K) and IC is the thermal conductivity (J m-l s-' K-l).Free convection will occur when Ra > RacRIT = 1707.25 Another dimensionless quantity that is used to describe the flow is the Grashof number, Gr, which is given as Gr = gDh3(AT)/v2 where v is the kinematic viscosity of the carrier gas (m2 s-l). The ratio Gr/Re2, given as Gr/ Re2 = g/3h3(A T ) / U , effectively represents the ratio of buoyancy forces to inertial forces and differentiates between free convection caused by gravity and forced convection.14~ 1 7 - 1 9 7 26 The criteria after Sparrow et aL2' are as follows: 0 < Gr/Re2 < 0.3 forced convection (viscous forces predominate) 0.3 < Gr/Re2 < 16 combined free/forced convection 16 < Gr/Re2 free convection (buoyancy forces predominate).The range of Gr/Re2 values normally found in epitaxial reactors is 0.0041 .0.26 Ban17 performed flow visualisation experiments with TiO, over the range 0.16 < Gr/ Re2 < 12.5 and found that above a critical value, Gr/Re2 = 0.5, fully developed spirals existed and below this value an upwards deflection of the streamline was observed. ENTRANCE EFFECTS AND BOUNDARY LAYERS 'Temperature measurements within epitaxial reactors have been performed by Sedgwick et al.15 and l7 Ban used a specially designed thermocouple and found that there were very strong temperature gradients in the first 15 mm above the susceptor whilst the remainder of the reactor was at a fairly constant temperature. This was explained by the Ban, Berkman and Goldsmith model,ls in terms of a split region whereby the gaseous reactants are thoroughly mixed in the turbulent layer before diffusing through the laminar layer to the substrate.Sedgwickf5 used laser Raman scattering to show the same temperature distribution but only in the vicinity of the leading edge of the susceptor. The various stagnant layer models put forward by Eversteyn et a1.l27 l3 also show a clear region below the white haze of TiO, smoke2714 CHEMICAL VAPOUR DEPOSITION OF SEMICONDUCTORS in the upper layer. All these observations suggest that the temperature and perhaps the velocity profiles are not yet fully developed and that a finite length is required for them to become so. Gilinglg9 2o describes this finite length as the ‘entrance length’ ( x ) , and below this value, the velocity and temperature profiles are still developing.The entrance length (in cm) of the velocity profile is given by xu = 0.04 h2Ua p / p = 0.04 hRe from Schlichting22 and that of the temperature profile, XT = 0.28 h2U, = 0.28hRe from Hwang and Cheng.28 Sedgwick15 found that near the leading edge of the susceptor, a boundary layer type distribution was observed, but much further along, a more linear profile was observed. Leys and Veen~liet~~ have developed a critical velocity for a certain position along the susceptor, below which the velocity profile is fully developed: UCRIT = 4vx/h2. According to this expression and using a typical velocity for growth such as 20 cm s-l, the velocity profile would not be fully developed for ca.3.5 cm along the susceptor. A boundary layer forms because the flow of a fluid ceases immediately adjacent to the surface of a solid body. The flow of the fluid is therefore retarded and the velocity undergoes a continuous change from its bulk value to zero. This region of velocity change is very narrow and is termed the boundary layer. It can be described as a region of rapidly increasing momentum, temperature and concentration gradients perpendicular to the susceptor and mass transport occurs by diffusion through this ‘static’ layer. The boundary layers for the above parameters are not identical but are almost coincidental in the gaseous systems we are considering.21* 22f 2 9 7 30 The typical expression for the boundary layer thickness, S,(x), is given by 6,(x) = A ( v x / U , ) k A has been found to have differing values: A = 5 from ref.(21), A = 5.83 from ref. (22), A = 1 from ref. (23). If we are depositing in such a developing region, the boundary layer thickness is important especially when growing in diffusion-limited regimes such as MOVPE. Eversteyn et a1.l2? l3 used the parameter 6 in their equations describing the growth rate of Si from SiH,. It has also been suggested that the flow of carrier gas is more stable in the thermal boundary layer region than in the fully developed profile.31 B r a d ~ h a w ~ ~ also developed a boundary layer theory to describe growth rates and showed the velocity dependence on the parameter 6, and from plots of the growth rate with position along the susceptor, Eversteyn et a1.l29 l3 were able to find 6 and also show its velocity dependence.We have, therefore, seen above how hydrodynamic theory is able to explain the existence of flow characteristics such as turbulence, convection, entrance effects and boundary layers. EXPERIMENTAL EXPERIMENTAL ARRANGEMENT OF FLOW PATTERN APPARATUS The apparatus used to generate smoke patterns in our epitaxial reactor is shown in fig. 1 and is similar to that used by Eversteyn et al.l2 Palladium-diffused hydrogen is passed throughJ. I. DAVIES, G. FAN AND J. 0. WILLIAMS 2715 mass flow controllers , eihaust I I I I I I I I I I I ' susceptor I I I I I I I I I I I Fig. 1. Schematic outline of the experimental arrangement for the growth of ZnSe/TiO, smoke experiments including inset showing reactor dimensions.Bubblers: A, Zn(CH,), or TiC1, ; B, dopant or H,O. separate vessels containing TiCl, and H,O and then combined to form a TiO, smoke, entering the reactor through both the nozzle and side-arm. The flows are controlled by electronic mass flow controllers. Owing to the well known gas-phase reaction of dimethylzinc with hydrogen selenide,6-8 the flows of these two reactants are kept separate until immediately prior to the growth region. Therefore, two smoke inlets were used to simulate the gas mixing around the susceptor. The susceptor is heated and its temperature controlled by an r.f. coil. The reactor itself is water cooled, has an internal diameter of 40 mm and is 230 mm long. A silica nozzle transports the dimethylzinc flux to the susceptor leading edge.The susceptor itself is a graphite block and the sloping face, which is at a tilt angle of 17.4", is 55 mm in length. At the rear, the susceptor is 25 x 25 mm but as the block is sloping, the free height at the leading edge is 25 mm compared with 10 mm at the rear. A diagram of the reactor dimensions is shown in the inset to fig. 1. The temperature of the susceptor was varied from room temperature to 623 K, which is the chosen growth temperature for ZnSe. By adjusting the mass flow controllers of the H,O line, the total flow in the reactor may also be changed. Photographs of the smoke patterns were taken with a camera positioned to obtain an image of the length of the reactor. GROWTH APPARATUS The growth apparatus is also shown in fig. 1. The same silica reactor is used as in the earlier smoke experiments but the gas handling system is different.Hydrogen is supplied via a palladium diffuser (Johnson Matthey model H480/ 17) and is used to transport the starting materials to the reactor. Dimethylzinc, supplied in standard stainless-steel bubblers by Alfa products, was cooled to - 10 "C in a temperature bath (Haake model D3L) to obtain the necessary vapour pressure. Hydrogen selenide, supplied as a 5000 VPM mixture in high purity hydrogen (BOC Special Gases), was also diluted in hydrogen to obtain the necessary mole2716 CHEMICAL VAPOUR DEPOSITION OF SEMICONDUCTORS Table 1. Ratio of Gr/Re2 for H, in the epitaxial reactor (12 = 0.0175 m) T/K U , = 10 cm s-' U , = 25 cm s-' U , = 40 cm s-' 500 6.9 775 10.5 1.10 0.43 1.68 0.66 fraction.Typical mole fractions were 2 x for dimethylzinc and 6 x lod4 for hydrogen selenide. The total flow rate of hydrogen was 2.25 dm3 min-I. As the dimethylzinc and hydrogen selenide react at room temperature according to a well known homogeneous gas-phase the dimethylzinc-hydrogen mixture was passed down the central nozzle, described earlier, and was mixed with the hydrogen selenide, supplied via the main reactor inlet, in the region of the susceptor. Growth was performed on (i) a silica plate and (ii) (100)-oriented GaAs surfaces, etched in a 5 : l : 1 solution of H,SO4-H,0,-H,O, and cleaned as described previou~ly.~~ Growths were performed in one particular flow regime and thicknesses of the layers examined as a function of distance along the susceptor and also of the temperature of the susceptor. The thicknesses of the ZnSe layers on GaAs were measured with an optical microscope and those of the ZnSe on silica by using an Art ion laser system (Spectra Physics model 165).Light of wavelength 457.9 mm was impinged on the ZnSe layer and the incident and transmitted intensities were monitored as a function of distance both along (x direction) and also across (z direction) the susceptor. Relative thicknesses of the layer were calculated from Beer's law. RESULTS AND DISCUSSION APPLICATION OF HYDRODYNAMIC THEORY TO THE ZnSe REACTOR Typical values of Re under our experimen tal conditions are in the range 1 < Re < 1 00, using values of the density and the dynamic viscosity of the carrier gas, H,, reported by Giling.l9 As our susceptor (the substrate holder) is a sloping block, we calculated values relevant to the maximum, minimum and average free height (h) of the tube.From this range of values, we see that the conditions fall within the laminar region of an inviscid flow with a viscous boundary layer close to the reactor wall and suscep t or. Values of Ra calculated for our reactor were 657 and 187 at 500 and 775 K, respectively. These fall well below RacRrT = I707 25 and so we conclude that no free convection occurs in our system. Table 1 lists the Gr/Re2 ratios for the experimental conditions in our epitaxial reactor for an average free height of 17.5 mm. We should, according to the criteria of Sparrow et al.,27 be in the combined free/forced convection region at low flows but in a laminar flow regime at higher flow rates.For a typical flow regime ( U , = 25 cm s-l) and growth temperature (575 K) in our reactor, calculated values for x, and xT were 10 and 70 mm, respectively. This thermal entrance length extends beyond that (55 mm) of our graphite susceptor and therefore we would expect to be depositing in a developing temperature profile region. The velocity profile is, however, fully developed over the majority of the growth region. Nevertheless, it seems that some of our growths may occur in boundary layer regions. OBSERVATIONS OF THE FLOW PATTERNS Some photographs of typical flow patterns are shown in plate 1. The photographs show the TiO, smoke stream clearly and the studies can be summarised into two sections.J .Chem. SOC., Faraday Trans. I , Vol. 81, part 1 I Plate I Plate 1. Photographs of TiO, smoke flow patterns: ( a ) U , = 10cm s-l, T = 290 K ; (h) U , = 10 cm s-l, T = 350 K. J. I . DAVIES, G. FAN AND J. 0. WILLIAMS (Facing p . 27 17)J. I. DAVIES, G . FAN AND J. 0. WILLIAMS 2717 First, the room temperature measurements show that the smoke stream from the nozzle falls directly onto the susceptor at very low flows (ca. 10 cm s--l). The nozzle flow is mixed with part of the side-arm flow at the leading edge and the remainder at the susceptor top. If the flow velocity though the nozzle is increased the smoke does not fall directly onto the susceptor at the leading edge but settles half-way up the slope. In general, at room temperature, without a heating effect from the susceptor, no buoyancy is observed and there is no upward deflection of the smoke screen.All the flows appeared to be laminar, but for one flow condition, at high velocity (ca. 40 cm s-I) with the nozzle pulled away from the susceptor, large vortices were observed. This was probably due to the sharp leading edge of the graphite block. Secondly, we studied the effect of heating the susceptor from room temperature to the growth temperature of 623 K. As the temperature is increased the smoke pattern rises as reported by Eversteyn et a1.l29 l3 The thickness of the layer appears to increase with an increase in temperature and the smoke pattern is more diffuse. At low flow rates (ca. 10 cm s-l) severe upturning of the nozzle flow occurred and part of the smoke screen spiralled backwards towards the main reactor inlet.As the flow was increased to ca. 20-25 cm s-l the thickness of the clear layer decreased and the smoke pattern became more diffuse, compared with the thin ‘wisps’ of smoke observed at lower flows. No upturning of the smoke pattern occurred at this flow velocity and the screen appeared to be completely laminar, only diverging into a ‘wavy’ formation downstream of the susceptor. Increasing the flow to 40 cm s-l makes the smoke pattern even more diffuse. However, what is obvious when we increase the velocity is the narrowing of the clear ‘stagnant’ layer. It also appears that the thickness of the clear layer varies along the length of the susceptor but this is hardly surprising as the susceptor slope gives different values for the free height, h, as the length, x, is increased.Some interesting observations arise when the central nozzle is retracted from the susceptor leading edge by ca. 3 cm. At low velocities (ca. 10 cm s-l) vortices are observed, probably by the sharp leading edge of the susceptor. At higher velocities (ca. 20-25 cm s-l) this effect diminishes but the smoke appears to separate into two parts whereby one portiop falls to the bottom of the reactor before it reaches the susceptor leading edge, and the other portion flows over the susceptor in a normal laminar fashion. We also saw part of the smoke screen traversing the sides of the susceptor between it and the reactor wall. The relationship of the observed clear layer above the substrate to various parameters will now be discussed.INTERPRETATION OF FLOW PATTERNS USING BOUNDARY LAYER THEORY Our TiO, smoke experiments have allowed us to observe, at least qualitatively, the flow patterns and any disturbances that arise in the particular type of epitaxial reactor we employ to grow ZnSe. From the photographs taken (see e.g. plate l), we can clearly see a region of very low density of TiO, particles beneath the dense smoke screen characterising the laminar flow regime of the system. The effect of susceptor temperature is obvious by the enhanced raising of the smoke pattern at higher temperatures, with the velocity of the flow also having an effect on the thickness of the clear layer produced. The question arises as to whether this clear layer is due to an increasing amount of stagnation in the gas flow towards the susceptor as proposed in the Eversteyn l3 or whether it is due to the thermal diffusion of the heavy TiO, smoke particles. Giling3* suggests that the clear boundary layer observed in these experiments is entirely due to the thermal diffusion effect.This enhances the diffusion of the lighter particles but causes the heavier particles to reflect from the hot susceptor .2718 5- L- E 5 3- Y 2- x 1- CHEMICAL VAPOUR DEPOSITION OF SEMICONDUCTORS 3.5- 3.0- E E 2.5- 2= z 2.0- 1.5- I 1 I 1 I I I I 1 2 3 1 , 5 6 7 ,d/mmt 2.41 I I I , I 1 1.8 2.0 2.2 2.1, 2.6 2.8 3.0 3.2 3.4 U(x,)-t/sl m-4 I I 1 I I I I 1 18 19 20 21 22 23 24 25 26 Ti/ K i Fig. 2. Dependences of (a) d(x) on xi for four different susceptor temperatures, T = 373 K (a), 473 K (x), 573 K (0) and 623 K (A); (b) 6(x) on U(x,,)+; (c) 6(x) on Ti.J. I.DAVIES, G. FAN AND J. 0. WILLIAMS 2719 An alternative explanation to the ‘stagnant’ layer model of Eversteyn et a1.12- l3 has been put forward by JGza and Cermak26 based on the Gr/Re2 ratio. If we work in the region of Gr/Re2 < 0.3 so that we are in a laminar flow situation, with some turbulence at high Re, then the ‘stagnant’ layer can than be viewed as a diffusion boundary layer. This boundary layer results from the thermal diffusion of TiO, particles perpendicular to the surface. Berkman et al.18 proved that by theory, a diffusion boundary layer can be expressed by 6, = 1.549(vx/Um); which is identical in form to the classical equation as derived by Schlichting.22 Schlichting also showed that the parameter S* was the displacement boundary layer where and S, z 6* for laminar flow.The bulk gas flow is, therefore, displaced by a minimum distance, S,, from the susceptor. We can envisage in our TiO, experiments, the bulk gas flow, represented by the smoke, being displaced by this diffusion boundary layer. In the light of the expression for 6, given by Berkman above, we have measured the thickness of the clear layers from our photographs for varying flow and temperatures and observed how our ‘stagnant layer thickness’, S(x), varies with temperature, velocity and distance along the susceptor. These values have been plotted to observe whether any agreement with the equation for SE exists. Fig. 2(a) shows the variation of the boundary layer parameter S(x) with distance along the susceptor, where the thickness of the layer increases with distance and temperature.Theoretically, the boundary layer thickness should decrease with x4 for a tilting susceptor.12 However, the values do appear to decrease after x = ca. 32 mm and perhaps this might suggest the existence of an entrance effect. Thereafter, the velocity and temperature profiles are fully developed. The variation of layer thickness with velocity is shown in fig. 2(b). The position chosen on the susceptor was x~, = 9 mm in order to avoid the effect of the nozzle and U(x,) is the velocity at this point. A straight line agreement is shown from a plot of S(x,) against l / U ( x , ) ; . This is in agreement with the boundary layer equation.The factor (vx/Ucc,) in the equation can be transformed; knowing that Y = p/p and considering the ideal gas law, we have S(X) = A(pRTx/MpU,)i 6* = +S,(X) where R is the gas constant (J K-l mol-l), M is the molar mass (kg mol-l) and p the pressure (N m-2). All other terms are as previously stated. Therefore, S(x) should be proportional to Ti and this is confirmed by reference to fig. 2(c), where the flow velocity U = 25 cm s-l and x, = 9 mm. GROWTH OF ZnSe We have shown that the TiO, smoke patterns can give us a good, qualitative idea of the gas flow patterns. We have also attempted to fit simple existing formulae to our experimental results. The best flows seemed to arise at a velocity of 20 cm s-l. We used this flow regime to grow epitaxial ZnSe on GaAs substrates under the conditions described in the Experimental section.The reason for the choice of GaAs is that its structure (sphalerite) is similar to that of ZnSe and both have almost identical unit-cell dimensions. The only variables that were changed were the susceptor temperature and the nature of the substrate. We grew ZnSe on a silica plate covering the whole area of the susceptor. To measure the thickness variation, the argon ion laser with the wavelength set at 457.9 nm was2720 N I E: 1.1- 2 W c3 0.9- CHEMICAL VAPOUR DEPOSITION OF SEMICONDUCTORS 0 0 0 5 10 15 20 25 30 35 x/mm 0 0 40 1 45 1 50 1 0 0 0 I , r 1 1 2 1 6 8 10 12 1 I 16 z/mm 5i (4 -3.8 1 -1.9 10 20 30 1 0 x/mm 0 0 0 0 0 0 - 1 . 6 - 1 1 1.3 1.1 1.5 1.6 1.7 1.8 1.9 2.0 10:' K / T Fig.3. Dependences of ( a ) G(x) on x for silica, (b) G(z) on z for silica, (c) G(x) on x for GaAs(100) surfaces, ( d ) In G on 1/T for GaAs(100) surfaces.J. I. DAVIES, G. FAN AND J. 0. WILLIAMS 272 1 36 used. With theabsorptioncoefficient a(ZnSe) = 5.62 x lo4 cm-l at this and using III, = exp ( -ad) where I is the intensity collected in the power meter (mw), I, is the original intensity corrected for silica (mw), the layer thickness d (cm) was calculated. The silica was then cut normal to its surface and the layer thickness measured with an optical microscope. The value of the thickness given by the laser measurement was 1.30 pm where the value measured under the microscope was 1.45 pm. For the remaining thickness measurements with the laser, the growth rate was normalised by the factor L.12.The growth rate, G(x), is plotted against distance (x) in fig. 3(a). We find that the growth rate drops away towards the end of the susceptor, suggesting a depletion of reactants. This should have been corrected by the tilt of the susceptor but we have already seen how the boundary layer thickness increases with x , suggesting that a lower growth rate should indeed be obtained. Fig. 3(b) shows the variation of rate across the width of the susceptor at x = 41 mm. We observe a channel of approximately equal growth rate down the centre of the slice, but ‘fall off’ towards the edges. This effect may be caused by the strong temperature gradients in the lateral direction caused by the water cooled wall in this type of reactor. Such an effect was observed by Gilinglg using interference holography.Growths were then performed on three GaAs(100) substrates at 623 K, covering the length of the susceptor. Fig. 3(c) again shows the depletion of the growth rate as a function of distance along the susceptor. It is more pronounced in this case than for silica but it must be remembered, however, that three non-identical substrates arranged linearly, and separated by intervals of several mm, were used in this case and this might explain the peculiarity in the curve. The trend is significant, however. We then grew ZnSe on GaAs( 100) substrates at various susceptor temperatures from 523 to 803 K at 40 K intervals. The growth rates were measured for x = 30 mm and varied from 0.001 to 0.023 pm min-l.The temperature dependence of the growth rate (G) is shown in fig. 3 ( d ) as a plot of In G against 1 / T. A plateau region is observed between 555 and 645 K where the growth rate is highest and essentially constant. We believe this to be consistent with a growth process that is controlled by diffusion of reactants across a boundary layer, where the temperature gradient is virtually constant, to a substrate surface that catalyses the formation of ZnSe. The growth temperature selected for ZnSe (623 K) falls within this plateau region. CONCLUSIONS Well established formulae based on hydrodynamic considerations have been used in an attempt to describe the observations of TiO, smoke patterns in our MOVPE reactor. We appreciate that the stagnant layer present may not be a true boundary layer but it appears to be representative of it.Measurements from the photographs show that the boundary layer thickness is related to the parameters x, U , and T. The variation in boundary layer thickness may then be used to explain the growth rate of, in our case, zinc selenide. We then observed how the growth rate of zinc selenide diminished towards the downstream end of the susceptor. This should have been corrected by the relatively large susceptor tilt angle used in our experiments but the probable reason why this does not happen is the unknown extent of the gas-phase reaction between dimethylzinc and hydrogen selenide. Although the susceptor end is reasonably close to the source of reactants, there must be a significant depletion as a result of this reaction by the time the substrate is reached.The temperature variation 89 FAR 12722 CHEMICAL VAPOUR DEPOSITION OF SEMICONDUCTORS of the growth rate showed that a maximum growth rate was achieved between 555 and 645 K and in this region the deposition process was shown to be diffusion- controlled. Using this reactor and the conditions described, good, single-crystalline, epitaxial layers of ZnSe and ZnSe,-,S, (0 < y < 1) on GaAs(l00) substrates have been grown and characterised. Certain aspects of this work will appear in later parts of this series. We thank the S.E.R.C. for supporting this work. J.I.D. acknowledges the award of a C.A.S.E. studentship with the Allen Clark Research Centre (Caswell) and G.F. is supported by a Royal Society/Chinese Academy of Sciences Exchange Program.H. M. Manasevit, Appl. Phys. Lett., 1968, 12, 136; H. M. Manasevit and W. I. Simpson, J . Electro- chem. SOC., 1969, 116, 1725. J. Cr.vst. Growth, 1981, 55, (Proc. 1st Int. Con$ MOCUD, ed. J. B. Mullin). J . Cryst. Growth, 1984, 68, (Proc. 2nd Int. ConJ MOCUD, ed. J. B. Mullin). A. C. Gossard, Treatise Mater. Sci. Technol., 1982, 24, 13. S. J. Jeng, C. M. Wyman, G. Costrini and J. J. Coleman, Mater. Lett., 1984, 2, 359. W. Stutius, Appl. Phys. Lett., 1978, 33, 656. P. J. Wright and B. Cockayne, J. Cryst. Growth, 1982, 59, 148. J. 0. Williams, E. S. Crawford, J. L1. Jenkins, T. L. Ng, A. M. Patterson, M. D. Scott, B. Cockayne and P. J. Wright, J . Mater. Sci. Lett., 1984, 3, 189. lo J. 0. Williams, T. L.Ng, A. C. Wright, B. Cockayne and P. J. Wright, J . Cryst. Growth, 1984,68,237. l 1 S. J. Bass, J. Cryst. Growth, 1975, 31, 172. l2 F. C. Eversteyn, P. J. W. Severin, C. H. J. van der Brekel and H. L. Peek, J. Electrochem. SOC., 1970, l3 F. C. Eversteyn and H. L. Peek, Philips Res. Rep., 1.970, 25, 472. l4 R. Takahashi, Y. Koga and K. Sugawara, J . Electrochem. SOC., 1972, 119, 1406. l5 T. 0. Sedgwick, J. E. Smith Jr, R. Ghez and M. E. Cowher, J. Cryst. Growth, 1975, 31, 264. l6 V. S. Ban and S. L. Gilbert, J. Cryst. Growth, 1975, 31, 284. l7 V. S. Ban, J . Electrochem. SOC., 1978, 125, 317. ' P. Blanconnier, M. Cerclet, P. Henoc and A. M. Jean-Louis, Thin Solid Films, 1978, 55, 375. 117, 925. S. Berkman, V. S. Ban and N. Goldsmith, in Heteroepitaxial Semiconductors for Electronic Devices, ed. G. W. Cullen and C. C. Wang (Springer, Berlin, 1978). l9 L. J. Giling, J . Electrochem. SOC., 1982, 129, 634. 2o L. J. Giling, J . Phys. (Paris), 1982, 43, C5-235. 21 F. Rosenberger, Fundamentals of Crystal Growth I (Springer, New York, 1979), chap. 5. 22 H. Schlichting, Boundary Layer Theory (McGraw-Hill, New York, 6th edn, 1968). 23 D. J. Tritton, Physical Fluid Dynamics (Van Nostrand-Reinhold, New York, 4th edn, 1980). 24 F. Rosenberger, Abstr. 6th Int. ConJ Vapour Growth and Epitaxy (Atlantic City, July, 1984). 25 B. J. Curtis and J. P. Dismukes, J . Cryst. Growth, 1972, 17, 128. 26 J. J6za and J. Cermak, J . Electrochem. SOC., 1982, 129, 1627. 27 E. M. Sparrow, R. Eichhorn and J. L. Gregg, Phys. Fluids, 1959, 2, 319. 28 G. J. Hwang and K. C. Cheng, J. Heat Transfer, 1973, 95, 72. 2g M. R. Leys and H. Veenvliet, J . Cryst. Growth, 1981, 55, 145. 30 M. L. Hitchman, J . Cryst. Growth, 1980, 48, 394. 31 Y. Kamotani and S . Ostrach, J. Heat Transfer, 1976, 98, 62. 32 S. E. Bradshaw, Int. J . Electronics, 1967, 23, 381. 33 A. M. Patterson and J, 0. Williams, J . Electronic Mater., 1984, 13, 621. 34 L. J. Giling, R.S.C. 1st School MOCVD (Queen Mary College, London, 2630th March 1984), 35 A. M. Patterson, PhD Thesis (U.C.W., Aberystwyth, 1984). 36 P. Lemasson, A. Etcheberry and J. Gautron, Electrochim. Acta, 1982, 27, 607. personal communication. (PAPER 4/2063)

ISSN:0300-9599    

DOI:10.1039/F19858102711

出版商:RSC    

年代:1985

数据来源: RSC

摘要:

J. Chem. SOC., Faraday Trans, I , 1985,81, 2723-2732 Binding of 2-Naphtholate Ions to a Water-in-oil Cetyltrimethylammonium Bromide Microemulsion The Enthalpy and Entropy of Interaction BY 0. A. AMIRE* Department of Chemistry, University of Ife, Ile-Ife, Nigeria AND HUGH D. BURROWS Departmento de Quimica, Universidade de Coimbra, 3000 Coimbra, Portugal Received 10th December, 1984 The binding of 2-naphtholate ions by a water-in-oil cetyltrimethylammonium microemulsion has been studied by u.v.-vis. spectroscopy. By assuming non-cooperative multiple equilibria in which the complexes obey Poisson's distribution law, the binding constants for the reaction were determined as a function of temperature. The reaction was found to be essentially enthalpy driven. Furthermore, it was observed that the entropy of reaction was small and positive below 25 "C. These thermodynamic parameters suggest that the binding involves both hydrophobic and electrostatic interactions.Substrate binding to inverted micelles and water-in-oil (w/o) microemulsions is currently a subject of great intere~t,l-~ generated by the fact that inverted micelles and w/o microemulsions have been used by many research groups as catalysts of chemical reactions in micro-environments which are thought to mimic the active sites in enzyme^.^ In fact, the catalysis of water-soluble enzymes entrapped in w/o micro- emulsions has been rep~rted.~ In this paper we report on the binding of 2-naphtholate ions by a cetyltrimethyl- ammonium bromide w/o microemulsion. The effects of temperature, protonation of the naphtholate anion and partitioning of the protonated anion into water and oil pseudophases on the values of the binding constants have not been reported.We have derived a binding isotherm which takes these effects into account. We also report on the enthalpy and entropy of the interaction between 2-naphtholate ions and a cetyltrimethylammonium bromide w/o microemulsion. EXPERIMENTAL MATERIALS Cetyltrimethylammonium bromide (CTAB), B.D.H. reagent grade, was recrystallised from acetone containing a small amount of methanol6 and dried in an oven at 60 "C to remove all traces of water. n-Heptane and chloroform, of reagent and analytical grade, respectively, were purified by distillation and then stored over sodium wire or calcium chloride.2-Naphthol, May and Baker analytical-reagent grade, was recrystallised as previously described. Other reagents used were of the highest commercial purity available. 2723 89-22724 BINDING OF NAPHTHOLATE IONS TO A MICROEMULSION PREPARATION OF SOLUTIONS BINDING EQUILIBRIUM Samples of dried recrystallised CTAB to give 0.045-0.055 mol dm-3 solutions were weighed on a Mettler balance into 5.0 cm3 bottles. These were later kept in an oven at 60 "C to remove any water absorbed during weighing. The samples were allowed to cool in a dessicator and later made up of 5.0 cm3 by addition of 1 : 1 n-heptane+chloroform mixture. Similarly, different amounts of 2-naphthol were carefully weighed on a Mettler microbalance and dissolved in 0.05 mol dm-3 NaOH solution to give 2-naphtholate solutions of various concentrations, usually between 1.02 x and 1.24 x lop2 rnol dmp3. The solutions were stored in brown bottles in a refrigerator before use.Microvolumes of 2-naphtholate solutions which gave 1 .O x 10 were added with a Hamilton microsyringe to the previously prepared inverted CTAB micelles and shaken vigorously until turbidity disappeared. We found that ultrasonnication was unnecessar~.~ The samples were then thermostatted in a water bath for 1 h. Spectra were recorded using a Pye Unicam SP8-400 or SP6-500 u.v.-vis. spectrophotometer with thermostatted cell compartments. rnol dmP3 overall concentration and [H,O]/[CTAB] PARTITION EQUILIBRIUM Appropriate quantities of 2-naphthol to make a 0.010 mol dm-3 solution in 10.0 cm3 total volume were weighed on a Mettler microbalance into 10.0 cm3 standard bottles.These were then dissolved by addition of 1 .O cm3 of 0.05 mol dmp3 solution of sodium hydroxide and made up to 10.0 cm3 with 0.05 mol dm-3 sodium chloride solution. The solution was titrated with 0.200 mol dm-3 hydrochloric acid sdution using a microsyringe to give a pH value of ca. 6 or any other pH required. The volume was noted and the final ionic strength calculated (ca. 0.05 mol dm-3). 0.054 mol dm-3 solutions of CTAB were prepared as described above in 5.0 cm3 bottles and 50.0 mm3 of the 0.010 mol dm-3 2-naphthol solution at pH 6.0 and incremental amounts of the 1 : 1 n-heptane + chloroform mixture were added. The resulting solutions were thermostatted for 3 h. Fluorescence spectra were then recorded using an Aminco-Bowman spectrophotofluorimeter with X- Y recorder attachment, as described previo~sly.~ METHODS DETERMINATION OF EQUILIBRIUM CONSTANT At a water-to-surfactant ratio of ca.10, it has been suggested that the water in the core of reversed micellar aggregates is highly structured by hydrogen bonds stabilized by the dipole moments of the headgroups and that micellar size is independent of the temperature.* In such a system substrate binding, occurring parallel to the other reactions, may be important, as suggested by the spectra in fig. 1. Subsequent partitioning of the uncharged conjugate acid is also possible. Neglecting activity coefficients4 (total concentration of 2-naphtholate ion, [BN,], is ca. mol dmP3) we may define the following equilibrium constants: where [BN-] is the molar concentration of unbounded naphtholate ion, [BNH] is the molar concentration of protonated naphtholate and [H+] is the molar concentration of hydrogen ions at the centre of the water pool.The subscripts p and ap in the concentration terms refer to polar and apolar media. K,, obs and K, are, respectively, the observed and thermodynamic dissociation constants for 2-naphthol solubilized in the centre region of the water pool and Kp is the partition coefficient of 2-naphthol between the water pool and the n-heptane + chloroform phase. El Seoud has postulated, for a similar sy~tern,~ that in the centre region of the water pool the pK, of the more hydrophilic acids are essentially the same as, and the pK, of the less0.A. AMIRE AND H. D . BURROWS 2725 hydrophilic acids were closer to, those found in ordinary water. It can easily be shown from eqn (1)-(3) that Ka = &, obs(l + l/kp)* (4) The interaction between CTAB-stabilized microemulsion and 2-naphtholate ion may be represented by stepwise incorporation of naphtholate ion from the core water into the surfactant/oil interfacial region : k M + BN M.(BN) k' k M*(BN)+BN@M*(BN), 2k' k M*(BN),-,+BN M-(BN), where M represents the CTAB-stabilized microemulsion aggregates. The charge on the 2-naphtholate ion, BN-, has been dropped to avoid symbolic misrepresentation of the net charge on the complex, M .(BN),. The rate constant for the association of naphtholate ions into the interfacial region is k, while nk' is the dissociation rate constant for the nth association step of the multiple equilibria.The equilibrium constant for the nth step, K,, = [M -(BN),]/[M - (BN)n-l] [BN-I, where K, = k/nk' = K/n. K is the overall equilibrium constant. It can be shownlo, l1 that nk' and [MT] and [MI are the total and equilibrium molar concentrations of CTAB microemulsion aggregates, respectively. Ignoring premicellar aggregates, the total molar concentration of CTAB-stabilized microemulsion is related to the critical micelle concentration, c.m.c., by [MT] = ([CTABI-[c.m.c.])/n', where n' is the aggregation number. In this simplified system the interaction is assumed to be non-cooperative and the complexes are assumed to obey Poisson's distribution law. As can be easily seen in eqn (6), no account was taken of any other parallel reaction involving the substrate.This may be corrected for by introducing a protonation term [BN-] [H+]/Ka. If we add this term we can then easily show that Furthermore, if the protonated substrate becomes neutral and the solubility in the oil + surfactant medium relative to water is significant, a partitioning term may be introduced and in eqn (7) K, is replaced by K,, obs and we have As shown in a previous publication' where A,, A and A , are absorbances at 353 nm of BN- alone, in the presence of CTAB microemulsions and in the presence of saturating amounts of CTAB microemulsions, respec- tively. A , was obtained by extrapolation of a plot of A against CTAB microemulsion concentration to zero value. Where the binding constant is small, as in the present case, the2726 BINDING OF NAPHTHOLATE IONS TO A MICROEMULSION value of A , was obtained by the method of Magid et aL4 Kp was determined as described below and [H+] was approximated from the pH of 2-naphtholate solution before addition to the reverse micellar solution.y At a known pH, a plot of A - A , (Kp+l)[H+] ~- & - A KaKp against [MT] allows one to obtain K.DETERMINATION OF PARTITION COEFFICIENT The ultraviolet emission from the excited 2-naphthol (BNH*) at 350 nm12 is the dominant emission at high concentrations of CTAB in aqueous medium. The intensity of BNH; emission is some two orders or magnitude less than that of BNHZp,l2 and the measured emission intensity is proportional to the fraction of BNH* in the apolar medium.The method of Tong and Glesmann13 as modified by Harris and Selinger12 is therefore applicable to the distribution of 2-naphthol between the water pool in the microenvironment and surfactant + oil medium. It is possible to show, adopting the same methods of derivation as Harris and Selinger,12 that the relative intensity, 4, is related to the partition coefficient by (9) where x = vp/vap, vp is the volume of water introduced to the reverse micellar system, vap is the volume of the reverse micelle and solvent, and Kp and k are the partition coefficient, as defined in eqn (3), and proportionality constant between [BNH*] and the fluorescence emission intensity, respectively. From a plot of 1/4 against x, both Kp and k may be obtained using the least-squares method. RESULTS AND DISCUSSION The u.v.-vis.absorption spectra of 2-naphthol solubilized into the water cores of CTAB-stabilized microemulsions with water-to-surfactant molar ratios of 10 and 30 are shown in fig. 1. The naphtholate peak14 has undergone a bathochromic shift of ca. 6 nm, in agreement with our previous observation in aqueous mediurn.l5 A shoulder has developed at 332 nm and further developed to a full peak at a water-to-surfactant ratio of 10. This is a new feature, which was not noticeable in an aqueous system, and it has been ascribed to 2-naphthol formation resulting from protonation by hydrogen bonding or direct hydrolysis. Since the initial pH is very high (ca. 12.0), it seems that hydrogen bonding must be very important in the water + CTAB + n-heptane +chloroform system, in agreement with the highly structured entrapped water in similar systems.8 Increasing the water-to-surfactant ratio did not change the position of Amax (353 nm), an indication that the solubilization site must be identical with that previously suggested for it in aqueous solution of CTAB;15 i.e.the anion is solubilized with its naphthalene ring sitting inside the hydrocarbon region of the surfactant/oil interface. The absorption spectra as a function of CTAB concentration is shown in fig. 2. It is remarkable that at low CTAB concentration, the 332 nm shoulder has emerged as a prominent peak. This again suggests that protonation was competing with microemulsion binding equilibrium. 1 /4 = [k,/k] x + 1 / k EMISSION SPECTRA The behaviour of the 2-naphtholate ion in the water core of CTAB-stabilized microemulsion has been investigated at various pH.The fluorescence emission spectra at pH 6, 9 and 12 are shown in fig. 3(a). They all show the hypsochromically shifted emission band of BNH* at 308 nm. Harris and SelingeP ascribed the 350 nm peak of 2-naphthol in an aqueous solution of CTAB micelles to the excited-state dissociation of BNH* to BN-*. The peak has undergone a hypsochromic shift of ca. 8 nm in the H,O + CTAB + oil system.0. A. AMIRE AND H. D. BURROWS 2727 1- I p"""$ 0.3 h m u ." 4 0.2 m e, e 3 0.1 D 0.0 310 320 330 340 350 360 370 380 390 LOO X/nm Fig. 1.Absorptionspectra ofthe2-naphtholateionin theH,O + CTAB + n-heptane+chloroform system at pH 1 1.70, I = 0.02 mol dm-3, T = 28.0 f 0.1 "C.[CTAB] = 0.0498 mol dm-3 and [HN,] = 0.8455 x ml dm-3; A and 0 represent water-to-surfactant molar ratios (MI) of ca. 10 and 30, respectively. The result for w = 10 is multiplied by 1.5. 0.32 - 0.16 - 0.14 1 \ 0.10 * 320 330 340 350 360 370 380 h/nm Fig. 2. Absorption spectra as function of cetyltrimethylammonium bromide concentration: 0, 0.05498 mol dm-3; x , 0.05246 mol dm-3; A, 0.04961 mol dm-3; a, 0.04774 mol dm-3. I = 0.05 mol dm-3, pH 12.70, [BN,] = 1.0 x mol dm-3, w = 10.0 and T = 15.9 "C.2728 BINDING OF NAPHTHOLATE IONS TO A MICROEMULSION 80 70 '5 60 % 50 v, LO 5 '- 30 20 10 0 h v) c, v x +I ._ c Y 300 320 340 360 380 LOO 420 440 460 480 500 520 540 2.5 X/nm 119 2 .o 1.5 8.5 9.0 9.5 10.0 103 x Fig. 3. (a) Fluorescence emission spectra of 2-naphthol in the H,O+CTAB+n- heptane +chloroform microemulsion as a function of pH: x , 6.0; .,9.0; 0, 12.0. Excitation was at 302 nm; an isobestic point in the absorption spectra of BNH and BN- in aqueous medium.15 I = 0.05 mol dmP3, [BN,] = 1.0 x mol dm-3 and w = 10.0.(b) Reciprocal relative intensity of 2-naphthol at 24k 1 "C as a function of x( z vP/vap) (see text for the definitions). I = 0.05 mol dm-3, [CTAB] = 0.0544 mol dm-3 and [BN,] = 9.901 x lop5 mol dm-3. The 358 nm peak has been chosen for the determination of the partition coefficient. A plot of 1/# against x( = vp/vap), according to eqn ( S ) , is shown in fig. 3 (b). From the slope and intercept of the plot, the partition coefficient, Kp was obtained. At 24 "C, Z = 0.05 mol dm-3 and a total concentration of the 2-naphtholate ion, [BN,], of ca.1 .O x 20. This value, which is of the same order of magnitude as those obtained in a similar ~ystem,~ suggests a marked preference of the 2-naphtholate ion for the microenvironment water than for the n-heptane+chloroform medium. A value of Kp < 10 would suggest a significant partitioning effect.17 Although Kp has been incorporated into our corrected equation for the sake of completeness, it is doubtful if it makes any contribution to the final value of K. Magid et aZ.4 have suggested molecular partitioning as an important factor in the binding equilibrium of substrate to Aerosol OT microemulsion, although this work does not support this view. mol dm-3, the value obtained from fig. 3 (b) is 2000. A. AMIRE AND H.D. BURROWS 2729 0.2 0 . 1 2.0 2.5 3 .O 3.5 [M,]/10-3 mol dm-3 Fig. 4. Plot of [(A -A,,)/(Am -A)]-(&+ 1) [H+]/K,K, against [MT]. See the text for further explanation. pH = 12.70, I = 0.05 mol dm-3, [BN,] = 1.0 x lop4 mol dm-3, w = 10.0 and T = 24.0kO.l "C. Table 1. Binding constants for the reaction between the 2-naphtholate ion and the H,O + CTAB + n-heptane + chloroform microemulsiona T/"Cb 15.9 20.0 24.0 28.0 K/dm3 m o P 143.5 135.6 127.9 117.4 a I = 0.05moldm-3, pH 12.7. kO.1 "C. +2.5dm3moIp1. THERMODYNAMIC DATA Fig. 4 shows a plot of A - A , (Kp+l)[H+] ~- A , - A Ka Kp against [MT]. A good linear plot has been obtained and the values of K were obtained by a least-squares fit of the data. The values of the equilibrium constants as a function of temperature are shown in table 1.The change in standard enthalpy, - A H e , was obtained from A H e (K)p =-Re Fig. 5(a) shows that the plot of In K against 1/T is linear. The change in standard enthalpy was obtained from the slope of the graph by a least-squares procedure. The value of - A H e obtained from fig. 5(a) is an average over the temperature range covered. The standard free-energychange, AGe, wascalculated from A G e = - RT In K at 20 "C. The standard entropy change, AS*, which is also an average over the same temperature range, was obtained from the thermodynamic relationship AG* = A H 0 - T A P at 20 "C. The values of the thermodynamic parameters thus obtained are shown in table 2.2730 BINDING OF NAPHTHOLATE IONS TO A MICROEMULSION 4.501 I I 1 I , I t 3.30 3.35 3.40 3 .L5 F-, .@' % 11.9 11.8 2 85 29 0 29 5 300 T/ K Fig. 5. (a) Plot of In K against 1/T. The experimental conditions are as for fig, 4. (b) Plot of RT In Kagainst T. The tangent to the smooth curve over experimental points gives the standard entropy change at a given temperature. The experimental conditions are as for fig. 4. Table 2. Thermodynamic parameters for binding of the 2-naphtholate ion to the H,O + CTAB + n-heptane +chloroform microemulsion' T/"C AHe/kJ mo1-l ASe/J mol-' K-I AG*/kJ mo1-l 20.0 k 0.1 - 1 1.86 & 0.84 0.36 & 0.04 - 11.96k0.24 a I = 0.05 mol dm-3, pH 12.7. It has been suggested* that the Stoke's radius of the Aerosol OT microemulsion system is strongly dependent on temperature at large values (w 3 11) of water- to-surfactant molar ratios and that at lower w the radius is independent of temperature and the water in the core is highly structured.The CTAB microemulsion may not be drastically different from the Aerosol OT microemulsion in this regard. This assumption is borne out by the plots in fig. 5, which are the expected results when compared with those obtained for similar systems in an aqueous mediu111.l~ From table 2 we see that the standard enthalpy change is small and negative. The enthalpy change required in transferring aromatic hydrocarbons from an aqueous environment into an organic phase has been observed to be slightly endothermic or atherma1.18-20 Thus, we expect the solubilization of 2-naphtholate ions into the hydrocarbon part of CTAB microemulsion to be endothermic, contrary to what has been observed in this work.Emerson and H01tzex-l~ proposed that when a charged hydrocarbon is being incorporated into a charged micelle from an aqueous medium, there are two opposing effects: the hydrophobic and dielectric effects. It could be easily shown in this case that the dielectric or electrostatic effect is exothermic. Thus a small negative enthalpy change, as observed for this reaction, suggests the predominance of the dielectric effect over the hydrophobic effect. From the thermodynamic expression0. A. AMIRE AND H. D. BURROWS 273 1 the variation of A S 0 with temperature may be 0bser~ed.l~ A plot of RT In K against T is shown in fig. 5 (b). The standard entropy change, ASe, is the slope of the tangent to the curve at a given temperature.Fig. 5(b) shows that the process of adding a charged aromatic hydrocarbon, in this case a 2-naphtholate ion, from the core water to the surfactant/n-heptane +chloroform interfacial region of the CTAB micro- emulsion is not entirely driven by the negative standard enthalpy change, except above 25 "C when the standard entropy change is small and negative. Below this temperature the standard entropy change for the process is remarkably small but positive. The unitary entropy change observed in transferring aromatic hydrocarbons from a polar to a non-polar medium is ca. 20 e.u.19 Allowing for some uncertainty in our results, the standard entropy change observed in this reaction is approximately an order of magnitude smaller than the literature values. A possible explanation of the positive sign is that the microemulsion core water, which is known to be highly structured even in the absence of a solute, tries to force the naphthalene ring structure of the 2-naphtholate ion into the oil + surfactant region (hydrocarbon region) in order to increase the entropy and attract the charged headgroup.The small numerical value may originate from the fact that the naphthalene ring structure becomes immobilized after ejection from the microemulsion water core, possibly an indication that the ring does not penetrate deep into the n- heptane +chloroform medium or that the electrostatic interaction between the negatively charged oxygen atom and the positively charged CTAB headgroup is relatively very strong. The negative sign of the standard entropy change at higher temperatures (> 25 "C) observable in fig.5 (b) is probably important. Various anomalous physical properties2' of the aqueous solutions of the quaternary ammonium ions have been explained on the basis of an increase in hydrogen bonding in the solvent surrounding the hydrophobic alkyl groups.18 N.m.r. studies22 showed that at 0 "C a solution of tetrabutylammonium ions is highly structured. At room temperature, and as the temperature is increased to 37 "C, the tetrabutylammonium ion becomes a water structure breaker. The trimethylammonium headgroup ofthecetyltrimethylammonium ion, the CTA+ ion, is expected to show similar behaviour but to a much lesser extent. If the trimethylammonium headgroup becomes a water structure breaker at high temperatures, the entropy of the water molecules in the microemulsion water core is expected to increase, and consequently the AS* that accompanies the incorporation of the naphthalene ring structure into the surfactant/oil interfacial region would decrease as the temperature is increased and may eventually become negative, as has been observed in the results presented here.Taken together with the enthalpy result, the entropy data appear to be consistent with previous suggestion from fluorescence and lH n.m.r. spectroscopy that the naphthalene part of the 2-naphtholate ion is solubilized into the hydrocarbon region of CTAB micelles in aqueous medium with the negatively charged oxygen close to the positive charge of the trimethylammonium ion.(The extent of both the hydrophobic and electrostatic effects would differ.) Thus, the interaction between the 2-naphtholate ion and an H 2 0 + CTAB + n-heptane +chloroform microemulsion has both enthalpic and entropic origin. The enthalpic contribution, the most important contribution in this study, arises from chargexharge interaction and the entropic one from a hydrophobic interaction between the naphthalene ring and the hydrocarbon structure at the surfactant/n-heptane +chloroform interfacial region. We thank Prof. L. B. Aladekomo, Physics Department, University of Ife, for use of his spectrofluorimeter and technical assistance. We also thank Mr Abiodun Ojo, Chemistry Department, University of Ife for assistance.27 32 BINDING OF NAPHTHOLATE IONS TO A MICROEMULSION J.H. Fendler and L-J. Liu, J. Am. Chem. Soc., 1975, 97, 998. F. Nome, S. A. Chang and J. H. Fendler, J . Chem. SOC., Faraday Trans. I , 1976, 72, 296. J. H. Fendler, Acc. Chem. Res., 1976, 9, 153 and references therein. L. J. Magid, K. Kon-no and C. A. Martin, J . Phys. Chem., 1981, 85, 1434. K. Martinek, A. V. Levashov, N. L. Klyachko, V. I. Pantin and I. V. Berezin, Biochem. Biophys. Acta, 1981, 657, 277. J. Ulminus, B. Lindman, G. Lindblom and T. Drakenberg, J . Colloid Interface Sci., 1978, 65, 88. S. A. Amire and H. D. Burrows, J. Chem. SOC., Faraday Truns. I , 1982,78, 2033. M. Zulaf and H. F. Eicke, J . Phys. Chem., 1979, 83, 480. 0. A. El Seoud, A. M. Chinelatto and M. R. Shimizu, J . Colloid Interface Sci., 1982, 88, 420. lo C. L. Kwan, S. Atik and L. A. Singer, J . Am. Chem. SOC., 1978, 100,4783. l 1 A. Yekta, M. Aikawa and N. J. Turro, Chem. Phys. Lett., 1979, 63, 543. l 2 C. M. Harris and B. K. Selinger, Z. Phys. Chem., N.F., 1983, 134, 65. l 3 L. K. J. Tong and M. C. Glesmann, J. Am. Chem. Soc., 1957, 79,4305. l4 D. M. Hercules and L. B. Rogers, Spectrochim. Acta, 1959, 15, 393. l5 0. A. Amire and H. D. Burrows, in Surfactants in Solution, ed. K . L. Mittal, in press. l 6 C. M. Harris and B. K. Selinger, J. Phys. Chem., 1980, 84, 891. M. F. Emerson and A. Holtzer, J. Phys. Chem., 1967, 71, 3320. H. S. Frank and M. W. Evans, J. Chem. Phys., 1945, 13, 507. l9 W. Kauzmann, Adv. Protein Chem., 1959, 14, 1. 2o C. Tanford, The Hydrophobic Eflect, Formation of Micelles and Biological Membranes (Wiley, New I1 W-Y. Wen, in Water and Aqueous Solutions, ed. R. A. Horne (Wiley, New York, 1972), p. 613. z 2 M-M. Marciacq-Rousselot, A. de Trobriand and M. Lucas, J. Phys. Chem., 1972, 76, 1455. York, 2nd edn, 1982), chap. 4. (PAPER 4/2089)

ISSN:0300-9599    

DOI:10.1039/F19858102723

出版商:RSC    

年代:1985

数据来源: RSC

摘要:

J . Chem. SOC., Faruduy Trans I , 1985. 81, 2133-2144 Studies of the Effect of Calcination on the Dispersion and Reduction of Nickel Supported on Alumina by X-Ray Photoelectron Spectroscopy, X-Ray Diffraction, Chemisorption and Catalytic Activity BY SANKARASUBBIER NARAYANAN* AND KADIYALA UMA Regional Research Laboratory, Hyderabad 500 007, India Received 10th December, 1984 The influence of calcination temperature on alumina and on alumina-supported nickel has been systematically studied. Identification of the nickel species has been made using the techniques of X-ray photoelectron spectroscopy (X.P.S.) and X-ray diffraction (X.r.d.). Hydrogen and oxygen adsorption measurements were used to calculate the dispersion and crystallite size of the metal. Calcination of y-alumina between 873 and 1073 K, before loading the netal, reduces the interaction of the metal and support, thereby increasing the area of metal avaiiable for hydrogen adsorption and for the hydrogenation of benzene.X.P.S. and X.r.d. indicate the presence of surface and bulk NiA1,0, on catalysts calcined at temperatures > 873 K. Benzene hydrogenation is a facile reaction and depends directly on the available metal area, irrespective of the nickel content of the catalyst. Supported-nickel catalysts find wide applications in a number of industrial processes such as hydrogenation, hydrocracking, methanation and steam reforming. The deactivation of nickel catalysts due to sintering and compound formation during high-temperature reactions is a major problem.' The activity of the supported catalyst depends to a large extent on the nature of the carrier used:2'3 the active species involved in the reaction may be the metal or the metal oxide.However, the support alters the reducibility of the metal ion, the dispersion, the crystallite size of the metal and the sintering The difficulty in reduction has been attributed to aluminate f ~ r m a t i o n , ~ ~ 7 9 but has not been unambiguously pr~ved.~Even though the concept of metal-support interactions is used in explaining the variation in catalyst characteristics and activity, the nature of the interaction which occurs in the preparation process and during the thermal treatment, and also the type of nickel species present, are still not clear.loT1l The phase of alumina and the calcination temperature of catalyst and support do play a major role in influencing the support-metal interaction.The presence of metal is reported to have a pronounced effect on the phase transformation and surface area of the support.12 The surface chemistry of nickel supported on alumina is complex, and much further work is needed before it can be fully understood and related to catalytic beha~i0ur.l~ There have been few systematic studies of the sintering of supported-nickel systems in an attempt to identify the influence of the support and the active component on each other. Our earlier studies14 on the reduction of nickel in mordenite revealed that incomplete reduction, and not just sintering and agglomeration of the metal alone, was responsible for the low dispersion of nickel. Nickel aluminate formation has also been suggested.l4 21332734 STUDIES OF NICKEL-ALUMINA CALCINATION The present investigation aims to identify the nature of the metal-support interaction in the nickel-alumina system. The influence of support on the active nickel species is studied in two ways: (i) by changing the structure and the phase of alumina before its use as support and (ii) by changing the structure of the alumina-supported nickel after loading the metal. The techniques of hydrogen chemisorption X-ray diffraction (X.r.d.) and X-ray photoelectron spectroscopy (X.P.S.) have been used to characterise the catalyst, together with benzene hydrogenation. EXPERIMENTAL Harshaw alumina A1 11 1-61E was crushed and particles of 1000-1400 pm were used as support (uncalcined alumina).The catalysts were prepared by pore-volume impregnation of the support with a solution of nickel nitrate of appropriate concentration and subsequently dried at 393 K overnight. Three sets of nickel-alumina catalysts were prepared. Set I (support- calcined): Alumina was calcined in air at various temperatures between 873 and 1673 K. The calcined supports were used for the preparation of a ca. 8% nickel catalyst. Set I1 (catalyst- calcined): Uncalcined alumina as received (with no pretreatment) was used to prepare a ca. 8% nickel catalyst. This was divided into five parts and each part was calcined at different temperatures in the region 873-1673 K. Set I11 (catalysts of different percentages by weight): Catalysts with varying nickel concentrations were prepared using uncalcined alumina.The impregnated catalysts, after drying, were pretreated in hydrogen for 5 h at 723 K and passivated with a mixture of nitrogen and air. The nickel content was estimated by atomic absorption. B.E.T. surface areas and hydrogen and oxygen adsorptions were measured using a constant-volume all-glass hgh-vacuum apparatus, capable of giving 1 x Torr* pressure. A mercury-penetration porosimeter (Micromeritics Autopore 9200) was used for measuring the pore-size distribution, pore volume and pore radius. The metal area, dispersion and crystallite size were calculated using hydrogen uptake at room temperature. Before adsorption the catalyst was reduced for 2 h at 723 K and degassed at the same temperature to reach The extent of reduction to metallic nickel was calculated15 from the oxygen adsorption at 700 K.The assumption was that the unreduced nickel was present in the form of nickel oxide and that at this temperature the metallic nickel would be converted into nickel oxide in the presence of oxygen. The theoretical oxygen uptake was calculated assuming that all the nickel was present in its metallic form. The ratio of observed oxygen uptake to the theoretical value gave the degree of reduction.16 The metal surface area was calculated17 by assuming the dissociative adsorption of hydrogen with one hydrogen atom on each surface nickel atom and taking the cross-sectional area of nickel to be 6.5 A2. The dispersion (D), defined as the ratio of available surface nickel atoms to the total number of atoms, is calculated from the equation D(%) = 1.17X/Wf where X is the hydrogen uptake in pmol g(catalyst)-l, W is the weight of catalyst andf is the fraction of nickel oxide reduced to nickel.This was based on the assumption17 that the unreduced nickel was present in a separate dispersed layer in intimate contact with the support. The average crystallite diameter was calculated from the equation d(nm) = 101 /D(% ). X.P.S. measurements were made using Vacuum Generators ESCA model 3MK I1 spectrometer with an A1 Ka source (hv = 1486.6 eV) as described previo~s1y.l~ All binding energies were charge-corrected to the C 1s line for adventitious carbon, assuming a binding energy of 285 eV for C 1s. Samples mounted onto the probe could be heated, reduced and evacuated in the preparation chamber and transported to the analyser chamber without exposure to the atmosphere. The accuracy of these measurements was & 0.3 eV.X.r.d. measurements were carried out with a Philips PW 1051 X-ray difractometer. Torr. * 1 Torr = 101 325/760 Pa.S. NARAYANAN AND K. UMA 2735 160 7 120 E 2 3 80 2 6 M N --- w 4 2 40 TI K Fig. 1. Effect of calcination on the physical characteristics of alumina. 0, B.E.T. surface area; A, average pore volume; 0, average pore diameter. Benzene hydrogenation, used as a model reaction to test the catalytic activity, was performed in a vertical, fixed-bed, continuous-flow reactor.'* The catalyst (ca. 0.5 g), diluted with ceramic beads, was packed in between layers of glass beads.Before reaction the catalyst was reduced in flowing hydrogen for 2 h at 723 K. Benzene was added from a calibrated motorized syringe. The products were analysed by Hewlett Packard model 5840 A gas chromotograph with a Silar column. Activity is defined as the number of moles of benzene converted to cyclohexane per gram of catalyst per hour. The turnover number is defined as the number of molecules of benzene converted per atom of exposed nickel per second. RESULTS AND DISCUSSION EFFECT OF CALCINATION ON SUPPORT Alumina was calcined at various temperatures in the region 873- 1673 K for 6 h. As the calcination temperature increases, the total surface area and average pore volume decrease whereas the average pore diameter increases (fig. 1). This change in the characteristics is prominent when the calcination temperature is increased above 1273 K.At a high calcination temperature there is a closure of the narrow pores, resulting in a loss of pore volume and surface area and in an increase in the average diameter of the pores. The X.r.d. pattern of the calcined alumina (fig. 2) indicatesclearly the transformation of the amorphous phase into the more crystalline a-alumina which is complete at 1473 K. ?-Alumina will lose hydroxyl groups at temperatures near 1073 K. SoledlS explained this by a mechanism in which two adjacent particles coalesce and a terminal oxide ion becomes incorporated as a bulk anion bridging the two particles. In this manner the particle grows and the effective surface area decreases. Complete dehydroxylation produces a-alumina.The marked decrease in the surface area, pore volume and the increase in pore diameter beyond 1473 K is attributed to the formation of a-alumina.2736 t x U .r( 2 U C .- STUDIES OF NICKEL-ALUMINA CALCINATION I I I I I 1 70 60 50 LO 30 20 261' Fig. 2. X.r.d. patterns for alumina calcined at various temperatures. (1) Uncalcined, (2) 873 K, (3) 1073 K, (18) 1173 K, (4) 1273 K, (20) 1373 K, (5) 1473 K, (6) 1673 K. Table 1. Characteristics of nickel catalysts prepared with calcined alumina as support calcination B.E.T. H, uptake temperature nickel surface area p mol g reduction catalyst /K (wt% 1 /m2 g-' (catalyst)-' (% 1 7R uncalcined 7.2 171 67 58 8R 873 7.9 137 I63 99 9R 1073 8.1 104 179 95 1 OR 1273 8.4 96 159 85 11R 1473 8.3 - 24 ca. 100 12R 1673 7.7 - < 24 ca.100 EFFECT OF CALCINATION OF SUPPORT ON HYDROGEN ADSORPTION (SET I) Hydrogen adsorption at room temperature was carried out on nickel catalysts (ca. 8 % nickel) prepared using alumina calcined at different temperatures (873-1 673 K). Some of the properties of these catalysts are given in table 1 . Hydrogen uptake is high for catalysts prepared using alumina calcined at temperatures in the range 873-1 273 K, catalyst 9R showing the maximum. For catalyst 1 1 R it is much lower than for catalyst 7R, which was prepared using an uncalcined support and with no heat treatment prior to metal loading. The percentage reduction of nickel oxide was the lowest for catalyst 7R. The metal surface area and dispersion follow the same trend as hydrogen uptake. Fig.3 shows the variation of metal area and dispersion with calcination temperature of the support: there is a maximum around 1073 K. The nickel crystallite size is almost constant in the temperature region 873-1273 K and also for catalyst 7 R with uncalcined alumina as support. It increases dramatically at 1473 K. An increase in the calcination temperature of the support prior to metal loadingS. NARAYANAN AND K. UMA - - - - - 2737 32 28 24 E n 2 0 3 5 .z r: a, .o, - 1 6 2 5 v) .Y * m m a * em 12 - 8 - 4 2 00 7 150 c z E W bD N 1 g 100 a - c 50 0 n I I I I I I I 473 6 73 673 1073 1273 1473 T/K Fig. 3. Effect of calcination of support on the dispersion of nickel (set I). A, Percentage dispersion; 0, crystallite size; 0, metal area. isexpected to increase the resistance ofthecatalyst to sintering.l However, Bartholomew et aZ.l observed that the high-temperature treatment of alumina (1 173 K) resulted in a support more prone to thermal degradation in a hydrogen atmosphere.High- temperature treatment of the support causes a loss of hydroxyl groups, which is responsible for the decrease in the interaction of the support and As has been reported,89 9 7 l4 metal-support interaction, whether due to electron transfer between metal and support or due to the formation of compounds such as nickel aluminate, will certainly make the reduction of nickel ions more difficult. In the present case, the decrease in the metal-support interaction favours the reduction of nickel oxide and hence the consequent increase in hydrogen uptake, percentage dispersion and metal area when compared with the uncalcined support catalyst 7R.For catalysts prepared on alumina calcined at temperatures > 1473 K, the decrease in hydrogen uptake could be due to the low-surface-area a-Al,O,. In the absence of free hydroxyl groups in a-alumina there is little or no metal-support interaction. Hence, impregnation of nickel on the low-surface-area alumina forms multiple layers of metal on the surface. During the reduction of these catalysts at 673 K, Ni2+ ions are easily reduced to metallic nickel and become very mobile. The observed growth of particles from 4 to 30 nm indicates the agglomeration of metallic nickel (fig. 3). The decrease in metal area may be ascribed to crystallite growth, and is consistent with the observation1 that metal migration and crystallite growth occurred more readily on an Ni-Al,O, catalyst where the support was calcined at 1 173 K.Calcination of the support at temperatures > 873 K before metal loading seems to result in improved reducibility of nickel oxide compared with the catalyst prepared using an uncalcined support (catalyst 7R). It is difficult to prepare a well dispersed nickel-alumina catalyst, especially with a low metal content (say 10% of below). The above procedure may help in achieving better dispersion by eliminating metal-support interactions to some extent. Gavalas et aL21 used X.P.S. and X.r.d. to identify the nickel species in NiO on a-alumina (ca. 2 wt% Ni) calcined at 1123 and 1323 K. The X.P.S. spectrum of the2738 - - - STUDIES OF NICKEL-ALUMINA CALCINATION 6 0 m 2 .0 P E 40 2 k z E + 20 m u W U v 673 1073 1473 TI K Fig. 4. Hydrogenation of benzene at 443 K: variation of turnover number (a) and activity (0) with calcination temperatures (set I). WHSV = 4.3 mol h-l g-l. sample, calcined at 1323 K and subsequently extracted by acid to remove NiO, indicated the presence of NiAl,O,. In contrast to this, the samples calcined at 1123 and 1323 K had X.P.S. spectra similar to that of NiO, also confirmed by X.r.d. However, X.r.d. did not show the presence of NiA1,0, in the acid-extracted or calcined samples. This observation confirms that, generally, NiO deposited on a-alumina forms multiple layers of metal oxide, as there is very little interaction;,O however, some NiA1,0, formation cannot be ruled out, especially for low metal concentrations and at high calcination temperatures.HYDROGENATION OF BENZENE OVER NICKEL ON CALCINED ALUMINA (SET I) Benzene hydrogenation was carried out at 443 K over these catalysts. The activity and turnover number (fig. 4) follow the same trend as the dispersion and metal surface area, viz. passing through a maximum at ca. 1073 K (fig. 3), indicating an increase in metal area with decreasing crystallite size of nickel. In fig. 5 the variation in crystal- lite size and turnover number with the metal area is shown. The metal area increases as the crystallite size decreases. The turnover number varies smoothly with the metal area per unit weight of nickel. Benzene hydrogenation depends directly on the available metal surface and is a facile reaction. These catalysts have the same weight percentage of nickel (ca.8 wt%). Hence, the difference in activity is mainly due to the calcination temperature of the support, which in turn affects the dispersion, crystallite size and metal area. EFFECT OF CALCINATION ON NICKEL-ALUMINA CATALYSTS (SET 11) Catalysts containing 8 wt% of the metal, prepared by impregnation of nickel nitrate hexahydrate into uncalcined alumina were calcined at different temperatures (873-1673 K) in air for 6 h. The colour of the samples turned from green to grey and finally to blue. The blue colour was retained by these samples (14 F-l7F)S. NARAYANAN AND K. UMA 2739 r 180 40 80 120 160 200 metal area/m2 g(Ni)-' Fig. 5. Hydrogenation of benzene at 443 K: variation of crystallite size (m) and turnover number (A) with metal surface area (set I).WHSV = 4.3 mol h-l g-l. even after two years. Physical properties of the samples, such as B.E.T. area, pore volume and pore diameter, were of the same order as that of the support calcined samples (set 1) at the same temperature. It was difficult to measure hydrogen adsorption on these samples. Even after reduction in hydrogen at 723 K for several hours, the samples did not change in colour. Only for sample 7F, which was not subjected to calcination in air, could 58% reduction be achieved. In the case of a catalyst 13F (calcined at 873 K, grey in colour) the hydrogen adsorption after reduction was only 5 pmol g (catalyst)-l and the reduction was only 11 % .X.P.S. DATA X.P.S. is surface-sensitive and is a valuable tool in identifying surface species. Binding-energy values of the samples, together with full widths at half-maximum (f.w.h.m.) measured from the X.P.S. spectra, are listed in table 2. All the samples have Ni 2p3,2 primary peaks around 857 eV as well as satellite peaks around 863 eV. Even after reduction there was no appreciable change in the binding-energy values of these samples or in their f.w.h.m., indicating that no new nickel species was formed after reduction. A comparison of the binding energies of these samples with the standard values for NiO, NiO and NiAl,O, suggests the presence of nickel aluminate, which is difficult to reduce, in all the calcined catalysts. X.R.11. DATA X.r.d. patterns for alumina and nickel-alumina (set 11) calcined at various temperatures are shown in fig.2 and 6, respectively. Values of 28 for the major identifiable peaks in the case a-Al,O,, NiO and NiAl,O, are listed in table 3. At 1473 K and above, alumina is present in the a-phase and is very crystalline. In the case of nickel-loaded alumina samples, unlike in the case of the support alone, new peaks at 28 values of 37.01 and 45.1" appear at temperatures as low as 873 K. Above 1473 K another peak at 65.7" is identified, along with the major a-Al,O, peaks. The new peaks2740 STUDIES OF NICKEL-ALUMINA CALCINATION Table 2 Binding energy values (in eV) of Ni-A1,0, calcined in air at various temperaturesa sample T I K Ni 2p,,,, satellite AEsat 0 1s A1 21, 13F 14F 15F 21F 16F 17F Ni014 ~ i 0 1 4 NiA1,O4l4 873 856.6 (3.8) 1073 857.4 (3.7) 1273 856.3 (3.5) 1373 856.0 (3.8) 1473 856.6 (3.2) 1673 856.2 (3.8) - 852.6 (2.4) - 854.2 (3.6) - 857.0 - 857.226 863.6 864.2 863.4 863.1 863.2 862.4 - 861 .O 863.1 - 7.0 6.8 7.1 7.1 6.6 6.2 - 6.8 6.1 - 53 1.4 532.2 53 1.2 530.9 53 1.9 53 1.3 (3.6) (3.6) (3.5) (3.4) (3.1) (3.1) - 533.6 530.6 (3.4) - 74.2 75.2 74.2 74.0 74.9 74.2 (3.0) (2.8) (2-3) (2.7) (2.6) (2.5) - - - - a Values in parentheses represent full widths at half-maximum.7 F J t I I I I 70 60 50 40 30 20 2elo Fig. 6. X.r.d. patterns for Ni/AI,O, calcined at various temperatures (set 11). (7F) Uncalcined, (13F) 873 K, (14F) 1073 K, (19F) 1173 K, (15F) 1273 K, (21F) 1373 K (16F) 1473 K (17 F) 1673 K. (at 28 = 37.01, 45.1 and 65.7") are characteristic of NiAl,O,.NiO peaks, especially those at 37.5 and 43.5", are difficult to identify at temperatures > 1473 K, since they appear together with a-Al,O, peaks. Moreover, at this temperature and at low nickel concentrations it is doubtful whether nickel could be present as NiO. If NiO is one of the bulk species, along with nickel aluminate, then the characteristic peaks for NiO should be identified at low temperatures when the a-alumina peaks are absent. TheS. NARAYANAN AND K. UMA 274 1 Table 3. X.r.d. data: 28 values identified for 8% Ni-A1,0, catalysts NiO a-Al,O," (ASTM) NiAl,O,b - - 25.5 35.4 38.0 37.5 37.0 1 43.6 43.5 - 45.1 52.8 57.6 63.05 - 66.7 68.4 __ - - - - - - - - - 65.7 _- - - - a Identified in the X.r.d. pattern of the calcined support (fig.2). Identified in the X.r.d. pattern of the calcined catalyst (fig. 6). absence of characteristic NiO peaks below 1373 K precludes the presence of NiO. Hence it is reasonable to assume from the X.r.d. patterns of these samples that nickel on an alumina support forms NiAl,O, at temperatures as low as 873 K, thereby influencing reduction and dispersion. As mentioned earlier, it may be worthwhile to prevent the formation of NiAl,O, by calcining the support between 873 and 1073 K before loading in order to obtain a better dispersion. X.r.d. and X.P.S. evidence indicates the formation of both bulk and surface nickel aluminate species. This is in agreement with our earlier observation14 on nickel mordenite, where the large crystallites and poor dispersion are explained on the basis of incomplete reduction due to NiA1,0, formation.Calcination of nickel-alumina in air at temperatures > 873 K is expected to form nickel a1~minate.l~ Lo Jacono et al., and Cimino et used magnetic, optical and X.r.d. evidence to support arguments for a surface spinel in the case of NiO on y- and q-alumina calcined at 873 K. It has also been reported that (especially at low nickel concentrations) nickel aluminate spinel is formed.23 Srivastava et al.,, suggested the presence of nickel aluminate spinel as a low-concentration ' skin ' rather than as a bulk phase in 5-20% NiO-Al,O, calcined at 723 K. Vedrine et al.9 failed to detect NiAl,O, either by e.s.r. spectroscopy or X.r.d. in the case of 15 % Ni/A1,0, calcined in air at high temperatures.However, they explained the difficulty in reduction as being due to a change in the electronic properties of nickel oxide because of its interaction with the support rather than to the formation of NiAl,O,. EFFECT OF NICKEL CONTENT ON DISPERSION OF METAL AND BENZENE HYDROGENATION ACTIVITY (SET 111) Catalysts with nickel contents between 5 and 50 wt% were used in this study. As the nickel content increases, both the crystallite size and the percentage of reduction increase, whereas the metal area passes through a maximum at ca. 10-20 wt% (fig. 7). At high nickel contents, after satisfying the surface forces of the support, which are responsible for the strong interaction with the metal oxide, multiple layers of metal oxides are formed. These become easily reducible and mobile, agglomerating to form large crystallites.2742 STUDIES OF NICKEL-ALUMINA CALCINATION nickel (wt.%) Fig. 7. Effect of nickel content on the dispersion of metal (set 111). @, Crystallite size; A, metal area; 0, percentage reduction. Fig. 8. Effect nickel (wt. %) of nickel content on benzene hydrogenation activity at 443 K (set 111). WHSV = 4.3 mol h-' ggl. 0, activity; A, turnover number.S. NARAYANAN AND K. UMA 2743 Hydrogenation of benzene was carried out over nickel catalysts of varying metal contents. Activity and turnover number vary with the degree of nickel loading (fig. 8), passing through a maximum at ca. 10-20 wt% at which the metal area reaches a maximum and the crystallite size a minimum. The turnover number for benzene hydrogenation is directly dependent on the available metal area and is independent of the metal loading.Bartholomew et al.25 found that the reduction to metallic nickel increases from 29 to 97 % on increasing the metal loading from 0.5 to 23 wt% nickel, suggesting a strong metal-support interaction at low concentrations of metal. Wu and Herculesx1 also suggest a strong interaction between NiO and y-Al,O,, especially at low metal concentrations, this interaction increasing with calcination temperature (cf. set 11). Even though these authors did not clearly identify the type of interaction, they explained the difficulty in reducing nickel ions at low metal loadings (2-10 wt% nickel) as being due to the preference of Ni2+ ions for tetrahedral sites in y-Al,O, rather than for the easily reducible octahedral sites. X.P.S.measurements by Dufresne et ~ 1 . ~ ~ also confirm the presence of Ni2+ species in the tetrahedral environment of Al,O, at low nickel concentrations. An increase in the nickel content forces the nickel ions to occupy the octahedral sites at a given calcination temperature, presumably because the tetrahedral sites are filled. It seems that there is an optimum percentage of nickel at which the metal area and the activity reach maxima (fig. 7 and 8). Below this, when the crystallites are very small, the interaction between the metal and the support is strong. This explains the difficulty in reduction at low metal loading, especially below 10%. However, this problem can be solved to some extent if the support is calcined at temperatures between 873 and 1073 K to eliminate the hydroxyl groups which are responsible for fine dispersion of metal particles and also for the strong metal-support interactions.CONCLUSIONS A metal-support interaction exists between nickel oxide and alumina, especially at low metal contents. X.P.S. and X.r.d. evidence strongly suggests the formation of NiA1,04 on the surface as well as in the bulk. This interaction can be eliminated to some extent if y-alumina is calcined between 873 and 1073 K before loading the metal. Benzene hydrogenation, which is a facile reaction, depends directly on the available metal surface area. We thank the Director of our Laboratory, Dr G. Thyagarajan for his encouragement and support. K.U. thanks the Council of Scientific and Industrial Research (CISR), New Delhi, India for the award of a Research Fellowship.C. H. Bartholomew, R. B. Pannell and R. W. Fowler, J. Cataf., 1983,79, 34. M. Houalla, F. Delannay and B. Delmon, J . Phys. Chem., 1981,85, 1704. W. F. Taylor, D. J. C. Yates, J . Phys. Chem., 1964,68, 2962. M. Lo Jacono, M. Schiavello and A. Cimino, J . Phys. Chem., 1972,75, 1044. C . H. Bartholomew and W. L. Sorensen, J . Cazal., 1983, 81, 131. J. R. H. Ross, M. C. F. Steel and A. Zeini-Isfahani, J . Caraf., 1978, 52, 280. R. B. Shalvoy, P. J. Reucroft and B. H. Davis, J . VQC. Sci. Technof., 1980, 17, 209. J. C. Vedrine, G. Hollinger, and T. M. Duc, J . Phys. Chem., 1978,82, 1515. Trans. I , 1983, 79, 2013. M. Wu and D. M. Hercules, J . Phys. Chem., 1979, 83, 2003. ' M. Houalla and B. Delmon, J. Phys. Chem., 1980,84, 2194. lo G. Wendt, D. Hentschel, J. Finster, R. Schollner, S. Hanafi and R. S. Mikhail, J . Chem. SOC., Faraday2744 STUDIES OF NICKEL-ALUMINA CALCINATION l 2 D. J. Young, P. Udaja and D. L. Trimm, Catalyst Deactiuation, ed. B. Delmon and G. F. Froment (Elsevier, Amsterdam, 1980), p. 331. l 3 J. B. Peri, J. Catal., 1984, 86, 84. l4 S. Narayanan, Zeolites, 1984, 4, 23 1. l5 C. H. Bartholomew and R. J. Farrauto, J. Catal., 1976, 45, 41. l6 C. H. Bartholomew, personal communication. l 7 D. G. Mustard and C. H. Bartholomew, J. Catal., 1981, 67, 186. l8 B. Coughlan, S. Narayanan, W. A. McCann and W. M. Carroll, J. Catal., 1977, 49, 97. l9 S. Soled, J . Catal., 1983, 81, 252. 2o A. Brenner and R. L. Burwell, Jr, J. Catal., 1978, 58, 353. 21 G. R. Gavalas, C. Phichitkul and G. E. Voecks, J. Cat& 1984, 88, 54. 22 A. Cimino, M. Lo Jacono and M. Schiavello, J . Phys. Chem., 1975,79, 243. 23 L. A. Gambaro, J. L. G. Fierro, L. G. Tejuca and A. L. Agudo, SurJ Interface Anal., 1982, 4, 234. 24 R. D. Srivastava, J. Onuferko, J. M. Schultz, G. A. Jones, K. N. Rai and R. Athappan, Znd. Eng. 25 C. H. Bartholomew, R. B. Pannell, J. L. Butler and D. G. Mustard, Ind. Eng. Chem. (Prod. Res. Dev.), 26 P. Dufresne, E. Payen, J. Grimbolt and J. P. Bronnelle, J . Phys. Chern., 1981, 85, 2344. Chem. (Fundam.), 1982, 21,451. 1981, 20, 296. (PAPER 4/2095)

ISSN:0300-9599    

DOI:10.1039/F19858102733

出版商:RSC    

年代:1985

数据来源: RSC

摘要:

J. Chem. Soc., Faraday Trans. I, 1985,81, 2745-2756 X-Ray Photoelectron-spectroscopic Studies of Carbon-fibre Surfaces Part 5.-The Effect of pH on Surface Oxidation BY CAROL KOZLOWSKI~ AND PETER M. A. SHERWOOD*~ Department of Inorganic Chemistry, University of Newcastle upon Tyne, Newcastle upon Tyne NE1 7RU Received 12th December, 1984 Carbon fibres, electrochemically treated in a variety of different electrolytes, have been analysed using X.P.S. and s.e.m. The amount and type of surface oxide varies considerably depending upon the pH of the electrolyte. Much more surface is produced in acidic environments. Carbon dioxide is evolved during treatment in acidic environments, and oxygen, together with clouding of the solution, in basic environments. Different oxidation mechanisms are suggested at different pH values.There is currently considerable interest in the surfaces of carbon fibres that are subsequently used for the production of composite materials. There have been a number of X-ray photoelectron-spectroscopic (X.P.S.) studies of carbon-fibre surfaces [ref. (1) and references therein and ref. (3)-( 14)]. We have reported a series of studies in this area,l? 15-17 and in this paper we examine the nature of the surface electrochemical oxidation in aqueous environments at different pH values. This work uses X.P.S. and scanning electron microscopy (s.e.m.) to examine the electrochemically oxidised surfaces of type 2HT (high-temperature) fibres. EXPERIMENTAL The electrochemistry was carried out using a standard three-electrode glass cell.A bundle of carbon fibres (ca. 3000 filaments) acted as a working electrode and a piece of platinum foil acted as the counter-electrode. A saturated calomel reference electrode (SCE) was used. This reference electrode was placed as close to the working electrode as possible by means of a Luggin capillary. All solutions used triply distilled water, AnalaR nitric acid and sodium nitrate and AristaR sodium hydroxide and were thoroughly deaerated with nitrogen gas prior to use. The pH of the solutions was varied using small amounts of sodium hydroxide and nitric acid. The resulting pH was recorded on a digital pH meter. Samples for X.P.S. analysis were prepared by polarising the fibres to various positive potentials for 20 min in the various solutions. The fibre samples were immersed and removed from solution whilst still in circuit, washed thoroughly in triply distilled water and dried in an oven at 100 "C.All ample transfers were made in air. X-ray photoelectron spectra were obtained using an AEI (Kratos) ES200B X-ray photo- electron spectrometer operated in the FRR mode using Mg Ka X-radiation (240 W). Data were collected using an Apple I1 microcomputer linked to an IBM 370/168 computer, the latter being t Present address : Department of Chemistry, Kansas State University, Manhattan, Kansas 66506, U.S.A. 27452746 X.P.S. STUDIES OF CARBON FIBRES used for most of the data analysis.18 The base pressure in the sample chamber was in the range 10-8-10-9 Torr.? The curve fitting was carried out by using a non-linear least-squares curve-fitting program with a Gaussian Lorentzian product function.l8, l9 The C 1s binding energy of the ‘graphitic’ peak was taken as 284.6 eV for calibration purposes.This peak overlapped directly with the signal one would expect from the hydrocarbon contamination often used in X.P.S. for calibration, although we would expect that there is a little, if any, hydrocarbon contamination on the fibres. S.e.m. was performed using a Jeol JSM T20 microscope. The vacuum inside the microscope was A. The resolution of this electron microscope was ca. 150 A. Torr, with a beam current of RESULTS AND DISCUSSION We have not shown an overall X-ray photoelectron spectrum for any of these fibres, since they are all essentially similar showing only a C 1s peak and an 0 1s peak in the case of oxidised fibres.No N 1s peak could be observed. Checks were made for any fibre-surface decomposition,17 but no significant decomposition occurred during the collection time of the spectra reported here, except in the case of nitric acid (2 mol dm-3) as reported before.17 ANODIC OXIDATION IN 0.5 MOL D M - ~ SODIUM HYDROXIDE SOLUTION We discuss below the results for 0.5 mol dm-3 sodium hydroxide solution, although we found similar results at substantially greater concentrations. FEATURES OF THE c IS SPECTRUM Fig. 1 shows the fitted C 1s spectrum of fibres polarised to several potentials. Much less surface oxidation is seen than in previous studies in nitric acid.17 The ‘graphitic’ C 1s peak exhibits an asymmetric tail towards higher binding energy, as seen in untreated fibres,l but unlike the case of nitric acid17 it is retained even when the fibres are polarised to high anodic potentials.This suggests that the graphitic nature of the fibre surface remains largely intact even after vigorous electrochemical oxidation. The X.P.S. peak width, expressed as f.w.h.m. (full width at half-maximum), of the ‘graphitic’ peak remains constant at 0.98 0.01 eV (k 2 x standard deviation) for fibres polarised to different positive potentials, except for fibres treated at 2.5 V, where the peak broadened to 1.04 eV. We believe that the increased peak width at 2.5 V arises from unresolved 8-carbon peaks20 that occur as a result of chemically shifted carbon atoms adjacent to the oxidised carbon atoms. The substantial rise in surface -CO,H groups at this potential would be expected to give a large (0.7 eV) P-shift.20 The ‘oxide’ region of the C 1s spectrum consist of two main oxide components with chemical shifts of 1.6f0.1 and 4.0kO.l eV from the ‘graphitic’ peak.Two other components are observed with shifts of 6.0 and 6.9 eV in all cases. The main components arise from alcohol (-OH) groups (1.6 eV) and carboxylic acid/ester (-C0,R) groups (4.0 eV). This alcohol functionality was not found in fibres treated in nitric acid and provides a useful clue to the surface oxidation mechanism. Table 1 shows how the relative peak areas with respect to the ‘graphitic’ peak vary with potential for each chemically shifted species. Carboxy-type functionality predominates at higher potentials, in contrast to the nitric acid results. The plasmon-loss feature at 6.9 eV is present at each potential although its intensity fluctuates slightly, suggesting that there may be some extrinsic character as suggested previously.7 1 Torr = 101 325/760 Pa.ow n c, .r( T -e W x c, .r( + z .r( C. KOZLOWSKI AND P. M. A. SHERWOOD p I 1 : 290 285 290 205 290 205 2747 290 28 5 I 290 285 binding energy/eV Fig. 1. Carbon 1s region after steady-state polarisation for 20 min in 0.5 mol dm-3 sodium hydroxide solution to various potentials (vs. SCE).2748 X.P.S. STUDIES OF CARBON FIBRES Table 1. Relative C 1s peak areas with respect to the ‘graphitic’ peak for anodic oxidation of fibres in 0.5 mol dmP3 NaOH solution area ratios? chemical shift from the graphitic peak/eV potential/V 6.9 6.0 4.0 1.6 0.5 0.043 0.048 0.068 0.087 1 .o 0.0 12 0.040 0.044 0.046 1.5 0.028 0.049 0.102 0.067 2.0 0.006 0.072 0.099 0.061 2.5 0.030 0.066 0.160 0.100 a Area ratios are accurate within k0.003.The assignment of the species at 6.0 eV is uncertain. It may arise from -CO,:type groups, for carbonate is formed on graphite polarised in potassium hydroxide.21 However, it may also partly or wholly arise from n -+ n* shakeup satellite contributions. Its intensity increases slightly with potential. FEATURES OF THE 0 IS SPECTRUM Fig. 2 shows the fitted 0 1s spectra of the fibre samples. At low potentials two oxygen species are present, but as the carboxy functionality increases (as shown by the C 1s spectra) the fibres become increasingly polar and may lead to the adsorption of water, which might explain the third peak (535.5 eV) in the 0 1s spectrum.The two main components of the spectrum are due to )C=O [at lower binding energy (531.6 eV)] and )C-0- [at higher binding energy (533.1 eV)] groups on the fibre surface. These binding energies vary by k0.6 eV depending upon the potential. This variation may well arise from the steady change in surface environment caused by varying amounts of the two groups as the potential is altered. At higher potentials the higher-binding-energy component increases dramatically until its area ratio is greater than that of low-binding-energy component. This is due to the way in which the amount of -OH species increases, since the -OH contribution arises from carboxy and hydroxy groups while =O species arise only from carboxy groups. WORKING-ELECTRODE SOLUTION AND GASES EVOLVED The main gas evolved at the anode at potentials above 0.5 V was oxygen.At potentials above 1.5 V the solution in the working-electrode chamber darkened. At 3.0 V the solution turned dark brown and fragments of fibre were dispersed in the solution. The U.V. spectra for these solutions collected after polarisation to 2.0, 2.5 and 3.0 V showed an absorbance maximum between 204 and 208 nm. This peak is representative of n -+ n* transitions in conjugated carbon<arbon bonds, suggesting that small carbon-fibre fragments are present in the solution. ANODIC OXIDATION IN 2 MOL DM-3 SODIUM NITRATE SOLUTION FEATURES OF THE c 1 S SPECTRUM Fig. 3 shows the fitted C 1s spectrum of fibres polarised to several potentials.The overall oxidation increases with potential. These results are very similar to those foundC. KOZLOWSKI AND P. M. A. SHERWOOD 2749 536 534 532 530 . -_ 536 534 532 530 536 534 532 530 536 534 532 530 536 534 532 530 binding energy/eV Fig. 2. Oxygen 1s region after steady-state polarisation for 20 min in 0.5 mol dm-3 sodium hydroxide solution to various potentials (us. SCE). in the nitric acid case.17 The ‘graphitic’ C 1s peak loses its asymmetric tail at higher potentials due to exfoliation of the carbon-fibre surface as seen in nitric acid.17 The ‘oxide’ region of the C 1s spectrum consists of three main components with chemical shifts from the ‘ graphitic’ peak of 2.1 f 0.2 and 4.1 0.1 eV and > 6.0 eV.These components arise from carbonyl and relatzd groups (2.1 eV) and carboxylic acid/ester (-C0,R) groups (4.1 eV). The peak above 6.0 eV arises from satellite and plasmon processes.2750 X.P.S. STUDIES OF CARBON FIBRES 1.W 290 285 290 285 2 90 285 29 0 20 5 binding energy/eV Fig. 3. Carbon 1s region after steady-state polarisation for 20 min in 2 mol dm-3 sodium nitrate solution to various potentials (us. SCE). FEATURES OF THE 0 1s SPECTRUM The fitted 0 1s spectra of the fibre samples showed two oxygen species at all potentials with binding energies 533.0 k 0.3 and 53 1.4 k 0.3 eV due to )C-0- and )C=O groups, respectively. In all spectra the peak at highest binding energy had the greater intensity, although this intensity did vary slightly (the ratio of the peak areas varied from 0.54 to 0.66).ANODIC OXIDATION IN SOLUTIONS OF DIFFERENT pH In this section the results for solutions of different pH are discussed. Fibres were polarised to 2.0 V for 20 min in 2 mol dm-3 solutions of nitric acid, sodium nitrate at pH values 0.9, 7.0 and 11.9, and sodium hydroxide. FEATURES OF THE c 1s SPECTRUM The ‘graphitic’ C 1s peak has a f.w.h.m. that varies from 1.01 to 2.10 eV with an exponential tail seen only in the sodium hydroxide case for the reasons discussed earlier.C. KOZLOWSKI AND P. M. A. SHERWOOD .- ii 8. r ; . * . . s. I : :\ , - . - . ; : - - ' . . .. i , . + 8 f ;iY'+ . a . 275 1 \ I I I I I 8 6 4 2 0 chemical shiftlev Fig. 4. Carbon 1s region after steady-state polarisation for 20 min in solutions of various pH to 2.0 V (us.SCE). The unfitted spectra (fig. 4) clearly show the effect of pH on the amount of surface oxidation. Acidic solutions give rise to substantial surface oxidation, in contrast to a1 kaline solutions. The results of the fitted spectra are shown in table 2. A number of general trends can be seen. First, the relative C 1s area associated with carboxylic acid type groups increases with pH (shown as peak no. 2/peak no. 1 in table 2). The position of group 3 varies from 2.12 to 1.60 eV. We believe that the values from 2.05 to 2.12 eV are due to )C=O and related groups and the values from 1.74 to 1.60 eV are due to )C-OH type groups. In reality the situation is probably more complex than this, with the possibility of structures of the sort in scheme 1, where the net effect is to give Scheme 1.2152 X.P.S.STUDIES OF CARBON FIBRES Table 2. Relative C 1s peak areas with respect to ‘graphitic’ peak for anodic oxidation of fibres in 2 mol dm-3 solutions of different-pH electrolytes polarised to 2 V for 20 min electrolyte chemical shift ‘ graphitic ’ from‘ graphitic ’ peak peak/eV relative area area ratioa f.w.h.m./eV nitric acid 2.12 (peak 1) 4.02 (peak 2) 5.70 (peak 3) 2.05 (peak 1) 4.17 (peak 2) 6.43 (peak 3) 2.05 (peak 1) 4.19 (peak 2) 6.41 (peak 3) 1.74 (peak 1) 4.20 (peak 2) 6.65 (peak 3) sodium hydroxide 1.60 (peak 1) 4.00 (peak 2) 6.00 (peak 3) 6.90 (peak 4) sodium nitrate (pH 0.9) sodium nitrate (pH 7.0) sodium nitrate (pH 11.9) errors ( & 0.01) 0.65 1 0.292 2.10 0.191 0.037 0.673 0.326 1.44 0.220 0.074 0.660 0.384 1.31 0.253 0.079 0.543 0.538 1.28 0.292 0.040 0.095 1.181 1.01 0.1 12 0.032 0.068 ( L 0.001) ( f 0.004) (& 0.005) a Area ratio = peak no.2/peak no. 1 . two equivalent oxygen atoms leading to a C Is binding energy intermediate between that of )C=O and )C-OH. We would expect a straightforward )C=O group to have a shift of ca. 3.0 eV19l57l6 and a straightforward )C-OH to be ca. 1.6 eV;l? 1 5 7 l6 thus in this case the intermediate behaviour is exhibited for all but the highest pH values. This resembles the situation discussed by Boehrn2, for graphitic oxide, which can be electrochemically produced from graphite in aqueous nitric FEATURES OF THE 0 1s SPECTRUM Fig. 5 shows the fitted 0 Is spectra of the fibre samples, and table 3 summarises the results.The 0 Is spectra reflect the changes observed in the C 1s region. Thus if we consider the two equivalent oxygen atoms discussed above and suppose that they have a binding energy intermediate (532.3 eV) between that of )C=O (531.5 eV) and )C-OH (533.5 eV) then the changes seen in table 3 can be appreciated. At low pH one peak is seen at 532.3 eV, but as the pH is increased more individual )C-OH and )C=O character is present, so that for NaOH (2.0 mol dm-,) separate peaks are seen for these two groups. In addition carboxylic acid/ester (-C0,R) groups are present and were seen (in the C 1s region) to increase with pH. These groups would be expected to give two peaks at 531.6 and 533.1 eV (the results for 0.5 mol dm-3 sodium hydroxide have been discussed above) and would overlap with the peaks dis- cussed above.The absence of any peak at ca. 533 eV in the case of 2 mol dm-, HNO, (when the relative amount of carboxylic acid/ester groups are at their lowest) may be due to differences in escape depth for the C 1s and 0 1s regions (the 0 1s region is more sensitive to surface groups).C. KOZLOWSKI AND P. M. A. SHERWOOD 2753 HNO, 536 534 532 530 536 534 532 530 '.. I 1 536 534 532 530 I I 536 534 532 530 . ~- 536 534 532 530 binding energy/eV Fig. 5. Oxygen 1s region after steady-state polarisation for 20 min in solutions of various pH to 2.0 V (us. SCE). 90 FAR 12754 X.P.S. STUDIES OF CARBON FIBRES Table 3.a Relative 0 1s peak areas [column (a), total area = 1.01 and peak positions [column (b), binding energy in eV] from curve-fitted spectra for anodic oxidation of fibres in 2 mol dm-3 solutions of different-pH electrolytes polarised to 2 V for 20 min nitric acid 1 .o 532.3 sodium nitrate (pH 0.9) 0.051 536.4 0.734 532.8 0.209 531.4 sodium nitrate (pH 7.0) 0.053 536.2 0.662 532.7 0.338 531.7 sodium nitrate (pH 11.9) 0.064 536.7 0.607 533.1 0.328 531.5 sodium hydroxide 0.060 535.7 0.378 533.5 0.562 531.5 a Binding energies are accurate to & 0.2 eV and relative areas to & 0.004.WORKING-ELECTRODE SOLUTION AND GASES EVOLVED Different gases were evolved depending upon the pH. Thus as discussed previously in sodium hydroxide solution oxygen is evolved. In nitric acid solution carbon dioxide is evolved. We believe that these observations are explained by the different initial steps in the oxidation processes. A similar situation to that on graphite24 probably occurs.Thus in acid carbon dioxide is evolved according to the following electrode process: C(s) + H,O -+ C(s)O(ads) + 2H+ + 2e followed by processes such as 2C(s)O(ads) -+ CO, + C(s). In a base, oxygen is evolved according to the electrode process C(s)+OH- -+ ‘C(s)OH(ads)’+e followed by processes such as 4‘C(s)OH(ads)’ -+ 4‘C’+2H20+0,. Many possible intermediate reactions for graphite have been discussed by K o k h a r o ~ . ~ ~ In the equations above ‘C’ implies a breakup of the graphite lattice while C(s) implies that the lattice stays intact. The working-electrode solutions for fibres polarised in sodium hydroxide (0.5 mol dm-3) and sodium nitrate (pH 11.9) turned dark brown, and fragments of fibres were dispersed in the solution as described above.OTHER FEATURES Sodium was present on the fibres treated in sodium nitrate and sodium hydroxide. A large number of these ions were removed after washing in triply distilled water, but some sodium ions always remained. Sodium is well known to intercallate with graphite, and the same is probably true of carbon fibres. S.E.M. RESULTS Plate 1 shows an s.e.m. photograph of fibres polarised in 2 mol dmP3 sodium hydroxide. Clearly the physical mechanism of oxidation is different to that of fibresJ . Chem. SOC., Faraday Trans. I , Vol. 8 1, part I 1 Plate 1 Plate 1. Scanning electron microscope picture of fibres which had been polarised to 2.0 V (us. SCE) for 20 min in 2 mol dm-3 sodium hydroxide solution.Magnification: x 10000. C. KOZLOWSKI AND P. M. A. SHERWOOD (Facing p . 2754)C . KOZLOWSKI AND P. M. A. SHERWOOD 2755 oxidised in acidic s01utions.l~ Circular holes are formed in the fibre surface. These appear to be areas of localised attack and may provide suitable keying points to which the resin can adhere. There is no evidence of an ‘oxide’ layer as in the case of acidically treated fibres. CONCLUSIONS This work shows the important effect that pH and the nature of the electrolyte has on the resulting surface composition of electrochemically treated carbon fibres. The presence of hydroxide ions (high pH) leads to less surface oxidation but a breakup of the surface leading to small carbon-fibre fragments in the electrolyte. In high-pH environments oxygen is evolved, but in neutral and in acidic environments carbon dioxide is produced.At the edge sites of untreated carbon fibres carbon-oxygen complexes are already present. These complexes can be of several types including )C-OH, )C=O and -CO,H/ester. Alcohol and aldehyde groups will be more readily oxidised than the graphite lattice itself. Thus alkaline electrolytes may only be able to oxidise edge-site functionalities already present on the surface. They appear to be unable to oxidise the main graphitic structure. This would explain the holes produced on the fibre surface after treatment in sodium hydroxide solution, the edge sites being those of localised attack. We are grateful to the S.E.R.C. for the provision of equipment and for a studentship to C.K., and to the Royal Society for the provision of equipment. This work has been carried out with the partial support of the Procurement Executive, Ministry of Defence. A. Proctor and P. M. A. Sherwood, J. Electron Spectrosc. Relat. Phenom., 1982, 27, 39. B. M. Parker, Report RAE-TM-MAT-347 BR 75399 Order No. 81-14007, 1980. L. T. Drzal and G. E. Hammer, Report Order No. AD-A100863 (NTIS). 1981. G. Gynn, R. N. King, S. F. Chappell and M. L. Deviney, Report Order No. AD-A108233 (NTIS). 1981 K. Waltersson, Fibre Sci. Technol., 1982, 17, 289. A. M. Trushnikov, M. A. Koxykina, S. P. Papkov, V. Ya. Varshavskii, I. L. Kumok and A.- A. Konkin, Vysokomol. Soedin., Ser. B, 1982, 24, 865. B. Beck, E. Widyani and J. P. Wightman, NASA (contract Rep.), NASA-CR-169798, NAS 1.26:169798, 1983.K. W. Wolf, NASA (Contract Rep.), NASA-CR-169651, NAS 1.26:169651, 1983. ’ T. Kudo, G. Kawamura and H. Okamoto, J. Electrochem. Soc., 1983, 130, 1491. l o T. Takahagi and A. Ishitani, Carbon, 1984, 22, 43. l 1 A. Ishitani, Polym. Prepr. (Am. Chem. Soc., Diu. Polym. Chem.), 1983, 24, 221. l 2 G. Dagli and N. H. Sung, Report AMMRC-TR-83-46, Order No. AD-A136558 (NTIS), 1983. l 3 D. L. Messick, D. J. Progar, J. P. Wightman, NASA Tech. Memo, NASA-TM-85700, NAS l 4 D. R. Lowde, J. 0. Williams, P. A. Attwood, R. J. Bird, B. D. McNicol and R. T. Short, J. Chem. l5 A. Proctor and P. M. A. Sherwood, Carbon, 1983, 21, 53. l6 A. Proctor and P. M. A. Sherwood, Surf. Interface Anal., 1982,4, 212. 1.15:85700, 1983. Soc., Faraday Trans. I , 1979, 75, 23 12. C. Kozlowski and P. M. A. Shenvood, J. Chem. Soc., Faraday Trans. I , 1984, 80, 2099. P. M. A. Shenvood, in Practical Surface Analysis by Auger and Photoelectron Spectroscopy, ed. D. Briggs and M. P. Seah (Wiley, London, 1983), appendix 3, pp. 445-475. lB R. 0. Ansell, T. Dickinson, A. F. Povey and P. M. A. Sherwood, J. Electroanal. Chem., 1979,98,79. 2o B. J. Lindberg, J. Electron Spectrosc. Relat. Phenom., 1974, 5, 149. 22 H. P. Boehm, U. Hofman and A. Clauss, Proc. 3rd Conf. Carbon, 1957, Buffalo, N.Y., (Pergamon, H. Binder, A. Kohling, K. Richter and G. Sandstede, Electrochim. Acta, 1964, 9, 255. Oxford, 1958). 90- 12756 X.P.S. STUDIES OF CARBON FIBRES 23 R. V. Subramanian, V. Sundaram and A. K. Patel, Proc. 33rd Annu. Con$ Reinforced PlasticslCom- 24 Ngo Dai Wet, D. V. Kokoulina and L. I. Krishtalik, Elecktrokhimiya, 1972, 8, 225. 25 G. N. Kokhanov, Elecktrokhimiya, 1971, 7, 1606. posites (Society of the Plastics Industry, 1978) section 20-F, p. 1-8. (PAPER 4/2109)

ISSN:0300-9599    

DOI:10.1039/F19858102745

出版商:RSC    

年代:1985

数据来源: RSC

摘要:

J,. Chem. SOC., Faraday Trans 1, 1985, 81, 2757-2761 A Shape-selective Platinum-loaded Mordenite Catalyst for the Hydrocracking of Paraffins by the Chemical Vapour Deposition of Silicon Alkoxide By MIKI NIWA,* YOSHIMI KAWASHIMA AND YUICHI MURAKAMI Department of Synthetic Chemistry, Faculty of Engineering, Nagoya University, Furo-cho, Chikusa-ku, Nagoya 464, Japan Received 3 1 st December, 1984 A platinum-loaded H-mordenite (PtHM) catalyst has been modified by the chemical vapour deposition of silicon alkoxide in order to improve its shape-selectivity for the hydrocracking of paraffins. The paraffins reacted on the silica-coated PtHM according to their molecular size, and reactant shape-selectivity was achieved by choosing the extent of modification. Silica was only deposited on the external surface of the zeolite to form mono-, di- or tri-layers of SO,, with the internal acid and metal sites unpoisoned, thus effectively narrowing the pore exit size only.Platinum metal on the external surface was not completely covered by the silicon oxide. We have proposed the use of chemical vapour deposition (c.v.d.) of silicon alkoxide to control the pore-opening size of zeolites, exemplified by the cracking of octane isomers,l and have studied the characterization of the modified mordenite.2 Because the molecular size of silicon alkoxide is larger than that of the mordenite pore, it can not enter the pore and interact only with the external surface. A silica layer embedded on the external surface effectively narrows the pore, and thus a shape-selective catalytic reaction is achieved thereon.Zeolites are well known as supports for well-dispersed metal^,^ and metal-loaded zeolites possess catalytic activities for cracking, alkylation, isomerization, oxidation etc. The finely controlled modification of pore size would thus have wide application to shape-selective catalytic reactions4 over metal-loaded zeolites. Since these zeolite systems are believed to have a bifunctional activity based on the metal and the zeolite, the influence of foreign atoms on the c.v.d. process should be considered. The use of this type of catalyst in a hydrogen atmosphere would be advantageous, since it would prevent coke deposition and increase the life of the ~ a t a l y s t . ~ The purpose of present study is to apply the c.v.d.method to platinum-loaded mordenite, and to examine the structure of the silica layer which is effective in pore-size engineering. EXPERIMENTAL CATALYST PREPARATION Na-Mordenite was supplied by the Catalysis Society of Japan as a reference catalyst (JKC-2-M10). It was converted into the ammonium form by exchanging the sodium cation with ammonium ion in NH,NO, solution. The NH,-form of the mordenite thus obtained was introduced into a (Pt(NH,),)CI, solution for exchange in order to obtain the platinum-loaded mordenite. Cation exchange was performed at 353 K for ca. 20 h. The Pt-loaded mordenite was calcined at 573 K in an oxygen-nitrogen mixture, followed by reduction at 573 K in flowing hydrogen. The platinum content (0.13 wt”/C, ) was determined by X-ray fluorescence spectroscopy. 27572758 CATALYSIS USING A Pt-LOADED MORDENITE C.v.d.was carried out as described elsewhere.2 The alkoxide was deposited at 593 K, and the modified mordenite was calcined in oxygen at 573 K to remove the coke residue. The extent of modification was selected by choosing the amounts of zeolite and silicon alkoxide, and is shown by the weight gain upon deposition, e.g. SiPtHM (3.2 wt%). The weight gain upon deposition was not influenced by the platinum loading. CATALYTIC REACTION Hydrocracking was performed by the normal pulse technique. Hydrogen was used as carrier gas and was purified by a liquid-nitrogen trap which was installed prior to the reactor. The catalyst was pretreated in hydrogen at 623 K for 2 h. 1 mm3 of the octane isomer or a mixture of isomers was injected, and the products were analysed by using a liquid paraffin column operating from room temperature to 373 K.Hydrogenation of cyclohexene was carried out using the same apparatus with a p,p’- oxydipropionitrile column at 323 K. CHARACTERIZATION The acidity of the zeolite was checked by the temperature-programmed desorption (t.p.d.) of ammonia. The experimental method has been described in our previous paper.s The adsorption of carbon monoxide and hydrogen was carried out in order to evaluate the exposed metal surface area. The adsorption took place at room temperature on a sample which had been reduced at 623 K. The chemisorption of hydrogen was measured using a static system, and the amount retained by the catalyst was evaluated by extrapolating the equilibrium pressure to zero.The chemisorption of carbon monoxide was performed using the pulse method with helium as a carrier gas. X-Ray photoelectron spectroscopy was used to determine the surface concentrations of silica and alumina in the zeolites using a Vacuum Generators’ XPS LAB-5 instrument. RESULTS AND DISCUSSION HYDROCRACKING OF OCTANE ISOMERS ON MODIFICATED MORDENITES Before making activity measurements, the stability of the catalyst activity was checked by repeatedly injecting paraffins. The activities of PtHM and SiPtHM were stable for over six pulses, while HM was deactivated rapidly with increasing number of pulses. As long as the reaction was carried out in the conditions given earlier, the platinum-loaded mordenite was not deactivated.Fig. 1 gives the conversions of octane isomers on PtHM and on SiPtHM with various amounts of deposited silica. Three kinds of isomers were converted in a similar degree on the unmodified PtHM, as shown in Fig. l(a); the activity of PtHM was significantly altered by the modification, and the degree of conversion depended ,on the type of paraffin as well as on the amount of silicon deposited. On the SiPtHM (3.2 wt%) catalyst the degree of conversion of 2,2,4-trimethylpentane (TMP) was extremely small, while other species were still very reactive. Furthermore, on the SiPtHM (3.4 wt%) catalyst not only 2,2,4-TMP but also 3-methylheptane (MH) were not converted, only octane being reactive. Finally, the SiPtHM (3.7 wt % ) catalyst lost activity for paraffin cracking.In other words, the conversion of paraffins was controlled by c.v.d. according to molecular size; only the largest paraffin (TMP) was excluded from the SiPtHM (3.2 wt% ) catalyst, and the second largest (MH) also failed to react with the SiPtHM (3.4 wt%) sample. Even the smallest paraffin (octane) did not react on the SiPtHM sample with > 3.7 wt% loading. The PtHM catalyst could thus be modified by the c.v.d. of the alkoxide to obtain shape-selectivity for paraffin hydrocracking.100 80 h b? 60 0 ." k a L O 0 20 0 ( W / F ) / g h mol-' Fig. 1. Hydrocracking of octane (O), 3-methylheptane (A) and 2,2,4-trimethylpentane (0) over (a) PtHM, (b) SiPtHM (3.2 wt%), (c) SiPtHM (3.4 wt%) and ( d ) SiPtHM (3.7 wt%). Amount of zeolite (W) divided by the flow rate (F = 0.091 mol h-l) is shown on the abscissa.Table 1. Characterization of the zeolites used chemisorption : amount t.p.d. :a NH, chemisorbed X.P.S. desorbed/mmol g-l /pmol g-' cyclohexene thickness conversion sample lower peak higher peak HZC CO Si/Ald of SiO,/nm (%)" 4.9 - 0.0 7.1 HM 0.34 (415) 1.17 (683) - - PtHM 0.42 (411) 1.03 (683) 6.10 5.11 - - SiPtHM (1 .O wt % ) - - - - - 3.72 - - SiPtHM (1.6 wt%) - - - 2.47 5.2 0.06 SiPtHM (3.7 wt%) 0.43 (413) 1.00 (683) 5.61 2.93 7.6 0.6 5.1 SiPtHM (4.7 wt%) 0.31 (413) 1.09 (673) 5.67 2.98 9.1 0.8 2.7 a Values in parentheses are Tmax/K. Twice the number of hydrogen molecules chemisorbed. Total amount of platinum loaded, 6.67 pmol g-l. Cross sectional surface areas of Si 2s arid A1 2p were assumed to be 0.855 and 0.5735, respectively.' Hydrogenation experimental conditions: reaction temperature, 373 K; catalyst weight, 20 mg.f %/A1 ratio by X.r.f. is 4.9. CHARACTERIZATION OF SiPtHM ACIDITY The acidity of the SiPtHM catalyst was measured by the t.p.d. of ammonia, as shown in table 1. T.p.d. spectra were characterized by double peaks which appeared at ca. 410 and 680 K. These were assigned to weakly and strongly acidic sites, respectively.6 Acidity measurements showed that neither platinum loading nor c.v.d. of the alkoxide modified the intensity and number of acid sites. METAL SURFACE AREA The state of the platinum metal was evaluated by measuring the surface area of exposed metal. The amount of H, chemisorbed on the PtHM changed little upon2760 CATALYSIS USING A Pf-LOADED MORDENITE c.v.d., while that of CO gradually decreased (table 1).On the basis of hydrogen chemisorption it was considered that the metal surface was not covered by the deposited silicon. Suppression of CO chemisorption is probably caused by a narrowing of the pore exit size, because the kinetic diameter of CO is 0.376 nm and the reaction of octane with 0.43-nm pores is suppressed significantly by a deposition of > 3 wt% alkoxide. Using a non-steady pulse method in CO chemisorption may be a plausible reason for this. THICKNESS OF SILICA LAYER AND SURFACE DENSITY OF SILICON OXIDE Silicon/aluminum ratios at the surface were determined for the PtHM and SiPtHM samples by X.P.S. The ratio for PtHM was almost in agreement with the bulk composition, which justified the accuracy of the X.P.S.measurement of surface composition. The Si/A1 ratio measured by X.P.S. increased with the degree of deposition, although the total Si/Al ratio changed little. Surface enrichment of silicon oxide was therefore confirmed, because X.P.S. supplies information concerning the sub-surface. The Si/Al ratio can be correlated with the thickness ( t ) of the SiO, layer and the electron escape depth (d) as follows: Si/A1 = t(n + l)/(d- t ) + n where n denotes the Si/A1 ratio of HM. The thickness of SiPtHM (3-4 wt % ) samples, which exhibited shape-selectivity in hydrocracking, was estimated to be 0.6-0.8 nm, the electron escape depth being assumed as 2 nm. In other words, the deposited SiO, formed mono- to tri-layers on the external surface in this region.Because the silicon oxide was enriched on the external surface of the mordenite, and the acid sites were not poisoned by the deposition, the internal surface of the zeolite was not modified by this method. In other words, the silicon deposited by this method exists only on the external surface. The surface density of the silicon oxide can therefore be calculated from the external surface area and the amount of silicon oxide. Because the weight gain included not only the silicon oxide but also the coke residue, the amount of silicon oxide was measured by subtracting the amount of coke. The surface densities thus measured on HM (4.8 wt % ) and PtHM (2.4 wt % ) were 15.5 and 15.0 molecule nm-l, respectively. These surface densities were compared with the cation site density of mordenite, which was estimated from its structure (8.6 nm-,).Thus, layers of SiO, approximately two molecules’ thick were estimated on these zeolites. In conclusion, mono- to tri-layers of silicon oxide were formed on the external surface of the modificated zeolite which possessed shape-selectivity. HYDROGENATION OF CYCLOHEXENE In order to determine the surface conditions of platinum in the zeolite, the hydrogenation of cyclohexene was selected as a test reaction. Because the molecular size of cyclohexene (kinetic diameter 0.60 nm)8 is sufficiently larger than the expected pore size of the SiPtHM (> 3.0 wt % ) sample, it could only interact with the external surface of the zeolite. At 373 K the cracking activity was neglected, and the platinum loaded on the external surface was active only for olefin hydrogenation. The sole product was cyclohexane, and no inherent activity of the zeolite HM in this reaction was observed.By utilizing this condition, a test reaction was performed to determine whether the platinum on the external surface was deactivated by the silicon oxide or not. As shown in table 1, the SiPtHM (3.7 and 4.7 wt%) catalysts lost the inherent activity of PtHM by only 30-70% , showing that the platinum metal on the external surface was not completely coated. On the other hand, the acid site on the externalM. NIWA, Y. KAWASHIMA AND Y. MURAKAMI 276 1 surface must be covered with silicon oxide for the SiPtHM (> 3.7 wt%) sample, because the modified zeolite lost its cracking activity completely.This may indicate the structure of the deposited silicon oxide. The silica may not easily be bonded to the metal surface, or the metal particle may be too large. This work was partially supported by a Grant-in-Aid for Energy Research from the Ministry of Education, Science and Culture, Japan (no. 58045064). M. Niwa, S. Morimoto, M. Kato, T. Hattori and Y. Murakami, Proc. 8th Int. Congr. Catal., (Verlag Chemie, Berlin, 1984), vol. IV, p. 701. M. Niwa, S, Kato, T. Hattori and Y. Murakami, J. Chem. SOC., Faraday Trans. 1, 1984, 80, 3135. P. Gallezot, Catal. Rev. Sci. Eng., 1979, 20, 121. (a) S. M. Csicsery, in Zeolite Chemistry and Catalysis ed. J. A. Rabo, (ACS Monogr. 171, American Chemical Society, Washington D.C., 1976), p. 680; (b) E. G. Deruoane, in Catalysis by Zeolites. ed. B. Imelik, C. Naccache, Y. Ben Taarit, J. C. Vedrine, G. Coudurier and H. Praliaud (Elsevier, Amsterdam, 1980), p. 5 ; (c) P. B. Weisz, in Proc. 7th Int. Congr. Catal., ed. T. Seiyama and K. Tanabe (Kodansha, Tokyo, 1981), p. 3. (a) P. A. Jacobs, J. A. Martens, J. Weitkamp and H. K. Beyer, Faraday Discuss. Chem. Soc., 1981, 75, 353; (b) F. Ribeiro, C. Marcilly and M. Guisnet, J. Catal., 1982, 78, 267; 275. C. V. Hidalgo, H. Itoh, T. Hattori, M. Niwa and Y. Murakami, J. Catal., 1984, 85, 362. J. H. Scofield, J. Electron. Spectrosc. Relat. Phenom., 1976, 8, 129. D. W. Breck, Zeolite Molecular Sieves (Wiley, New York, 1974), p. 636. (PAPER 4/2 190)

ISSN:0300-9599    

DOI:10.1039/F19858102757

出版商:RSC    

年代:1985

数据来源: RSC