摘要:

EQUATION OF STATE AND THERMAL CONDUCTIVITY OF GASES AT HIGH PRESSURES AND ELEVATED TEMPERATURES BY J. SAUREL, R. BERGEON, P. JOHANNIN, J. DAPOIGNY, J. KIEFFER, B. VODAR Laboratoire des Hautes Pressions (C. R. N. S. and D. E. F. A.), 1, Place Aristide Briand, Bellevue (Seine et Oise), France Received 6th July, 1956 Experimental investigations of some thermodynamical properties of compressed gases at elevated temperatures have been undertaken. The method employed is the " internal heating " method, where the electrical furnace is within the high-pressure bomb, the walls of this bomb being water-cooled. The p, v, t measurements are made by a modified Amagat method ; a special electronic device has been developed for the detection of the mercury level in the capillary tubing of the piezometer.Measurements have been made on argon and nitrogen up to 1O00" C and loo0 atm. The thermal conductivity is measured in an apparatus composed of 2 coaxial cylinders, separated by a very narrow space. An electronic device composed of a breaker amplifier and a magnetic amplifier is used for the temperature control. So far only measurements on nitrogen have been made. Some p, u, t data on gases at very high pressures are obtained by the study, using ultra high-speed X-ray photography, of shockwaves generated in liquefied gases; as yet the data obtained on supercritical argon extend to 72,000 atm and about 1OOO" C. THERMAL CONDUCTIVITY The measurements so far published on the thermal conductivity of compressed gases are those of Stoliarov, Ipatiev and Teodorovitch 1 up to 500 atm and 300" C, of Keyes 2 up to 150 atm and 400" C, and of Michels and Botzen up to 2500 atm, and 75" C.Stoliarov and Keyes used the hot wire method; Michels employed the circular- plates method with a guard ring. In both cases the external heating method was used, the whole of the high pressure vessel being submitted to the temperature of the experiment. To attain higher temperatures one of us (P. J.) developed an apparatus based on the "internal heating'' method. The heat conductivity cell is in an electric furnace within the high-pressure vessel. The difiiculties arising from the limited strength of steel at elevated temperatures are thus completely avoided. But other problems arose, because of the high convective transport of heat in the gas between regions at elevated and at moderate .temperatures.This transport is favoured by the properties of the compressed gases, viz., large density, large thermal expansion, small Viscosity and large specific heat. The method using circular plates with guard ring is theoretically very satisfactory, but has been discarded because it needs too much space and because of the necessary cylindrical shape of the furnace. The hot-wire method was discarded for the reason that it does not give readily a thin uniform gas layer of known thickness, and therefore is not well suited for absolute measurements. Our apparatus comprised 2 coaxial cylinders, made of pure silver (fig. 1). The dimen- sions of the outside cylinder are 150 mm length, 20 mm ext. diam., 1 3 5 int.diam. TWO silver elements centre the internal cylinder, which is 88 mm long by 13.1 mm diam. The thickness of the cylindrical gas layer is consequently of about 02mm. Platinmi 64SAUREL, BERGEON, JOHANNIN, DAPOIGNY, KIEFFER, VODAR 65 platinum-rhodium thermocouples, four set in the internal cylinder and eight in the external cylinder, measure the temperature difference between the 2 cylinders and record the uni- formity of temperature along the external one. A few watts are supplied at the centre of the internal cylinder by a heater of platinum wire. The measuring cell com- posed of the 2 cylinders is entirely sur- rounded by a cylindrical electric furnace 320 mm long and having an internal diameter of 21 mm. The temperature control of this fur- nace is effected as follows.The sensing element is a platinum/platinum-rhodium thermocouple mounted against the inter- nal wall of the furnace. Its output is opposed to a known voltage taken from a high precision potentiometer. The unbalanced voltage, after a first ampli- fication by a breaker-type amplifier, controls a two-stage magnetic transducer which constitutes the voltage supply for the furnace. The controlled power may amount up to 4 kVA, the sensitivity being of 0.1 pV or less. An auxiliary device based on the phase of the unbalanced voltage, allowed the temperature fluctua- tion to be reduced to less than 0.01" C at about 400" C. A set of 4 compensating heating elements has been used for re- ducing the temperature gradient along the internal cylinder ; the temperature obtained is uniform to better than 0.01" C; 2 heater windings are located at the ends of the furnaces and 2 other ones in the centre elements.All these precautions are justified by the small value (1 to 4" C) which could be used for the temperature difference between the cylinders. Due to this small temperature difference and to the small thickness of the gas layer, the convective heat transport is reduced to a negligible value; this was checked by repeated measurements made with various tern- FIG. l.--a, auxiliary heating coils; b, centring element (silver) ; c, mobile centring element (silver) ; d, nickel-chromium ribbon (main heating winding); e, crosses indicat- ing the location of thermocouples; silica insulator ; g, high pressure vessel. perature differences across the gas layer, which gave the same value for the heat conductivity.The small thickness, moreover, minimizes the radiative heat exchange compared with the conductive heat transport. The heat conduction by the stand and the thermocouple wires which is assumed to vary only slightly with pressure, is elimin- ated by comparing the results for the same temperature under high pressure and at atmospheric pressure. Therefore we obtain the variation of the heat conductivity of the compressed gas relative to the heat conductivity at normal pressure which is known from other measurements. An accurate determination of the geometrical constant of the apparatus (essentially the thickness of the gap between the cylinders) has been made. The method used is based on the similarity between electric fields and heat fields between two bodies at uniform voltages or temperatures.The cylinders are spaced with the help of very small alumina cylinders, and thus are electrically insulated. The measurement of the electrical capacity, which is very accurate, gave the " mean geometrical constant " of the heat conductivity cell, including the regions around the ends of the cylinders ; this method avoids the use of a guard ring, the main purpose of which is to suppress the heat transport in regions where it is difficult to calculate the geometric constant. The apparatus has been operated now for many months up to 500" C and lo00 atm. The gas studied was nitrogen. Preliminary data, in the normal temperature region C66 THERMAL CONDUCTIVITY (up to 75" C), are in good agreement with those published by Michels and Botzen.3 At present we can confirm qualitatively the results of ref.(1) : at higher temperature a smaller effect of pressure on the heat conductivity of nitrogen is observed. EQUATION OF STATE: EXPERIMENTAL DATA The experimental study of p, w, T relations in gases undertaken by one of us (J. S.) is based on a constant density method. The " internal heating " method is used. This method has been, to our knowledge, used previously only once, and at moderate pressure up to 80 atm.4 Our apparatus has been already described.5 We here give only its essential features. The gas to be studied is contained in a bulb terminated by a capillary tubing, which is immersed in a mercury cup (fig.2). The pressure transmitting fluid is a gas (argon or nitrogen). An electromagnetic coil displaces the mercury cup, so that the piezometer can be filled with a known pressure and temperature. In another, slightly different arrangement, it is possible to introduce into the piezometer a gas different from the pressure-transmitting one. In both arrangements the measurements comprise main- taining the mercury in a capillary at a given level as temperature and pressure are varied. measuring setting filling j setting I FIG. 2.-Experimental device for p, v, T measurements ; " Monogas " method, The detection of the mercury level in the capillary is effected by an electronic device, without any contact electrode. This device has been described;6 it gives an accuracy of at least 0.2 mm in locating the level, even with small internal bores (0.06 mm).Pyrex glass, silica, and pure alumina piezometers have been used. From the constant density readings it is possible to obtain the isothenns. Fig. 3 gives the isotherms of nitrogen (R-type from " L'air Liquide ") up to 1OOO" C and loo0 atm. In the normal temperature regions the agreement is fairly good with previous data (Wouters 7 up to 150" C, and Bartlett up to 400" C ; the agreement is better with Wouters' data). It is hoped to obtain, from the high temperature p, w, T relations, some information on the repulsive term of the intermolecular potential energy. EQUATION OF STATE ; THEORETICAL COMPUTATIONS A repulsive potential of the exponential type seems the most valuable for de- scribing the intermolecular forces in phenomena involving high temperatures or high pressures, because quantum-mechanical calculations of short-range forces 9 indicate that actual repulsive forces are approximately of this form. Hitherto.SAUREL, BERGEON, JOHANNIN, DAPOIGNY, KIEFFER, VODAR 67 calculations of thermal and transport properties in particular have been made assuming such an interaction law ; second virial coefficients, viscosities and crystal properties have been calculated by several authors using the so-called " modified " Buckingham potential : For the (6 : n) potential of the Lennard-Jones type, the third virial coefficient has been obtained in the form of a power series in temperature, each coefficient of which involved double integrals and had to be determined by numerical integra- tion.10 Such series expansion could not be obtained for an (6 : exp) potential and one of us (R.B.) has effected a numerical integration of the triple integral by an electronic computer (" gamma " computer of the Compagnie des Machines Bull), which required a total (programming and computing) time that did not exceed very greatly that:of the above-mentioned series expansions. Isotherms of nitroqea FIG. 3.-Plot of nitrogen isotherms. The calculations have been made with a ranging from 12 to 15 because a values for simple molecules have been found to be in this range, according to second virial coefficient and viscosity data. With argon, for example, we compared experimental and calculated third virial coefficient for different sets of values of a, e*/k and r, (or bo) which have been proposed by different authors (table 1).TABLE 1 remarks bo U c*]k (" K) (cm3,mole) 119 52.7 13.5 tried by Bergeon 121.5 58.37 15 ref. 11 123.2 51.23 14 ref. 12 Observed values have been taken from Michels, Wijker and Wijker and cor- rected by trial and error, plotting (PV/RT - 1) V - (B + C/V) against the square of the density (V-2) and modifying B and C till we obtained a straight line (with a slight curvature at high densities) going through the origin. C(T) values so cor- rected are on a smooth curve (fig. 4). We note that the value a = 14 proposed by Mason and Rice seems satisfactory. There remains a slight discrepancy in the slopes of observed and calculated curves.68 THERMAL CONDUCTIVITY SHOCK WAVES AND EQUATION OF STATE AT VERY HIGH PRESSURES Measurements at very high pressures have been made in liquids using strong shock waves generated by an explosive charge.13 An X-ray flash synchronized with the detonation allows a measurement of the absorption of X-rays and of the velocity of the shock wave.The former gives the density of the compressed fluid; by determining both density and shock velocity the pressure attained can be calculated from the Hugoniot relations. By this method we obtained data on argon gas up to a pressure of 72,000 atm. Pressures above 100,000 atm could be measured in water and a few solids (Pyrex glass and aluminium). 1.K boo I I I 3 0 0 3 5 0 4 0 0 4 5 0 FIG. 4.Third virial coefficient. Expt. values : 0 from Michels, Wijker and Wijker (Physicu, 1949, 15, 627).* A values corrected.Calc. curves : full curve -- 01 = 14 (ref. 12). interrupted curve - - - - - - - u = 15 (ref. 11). dotted curve . . . . . . . . . . . . . . . u = 13-5 (tried by R. Bergeon). (6.12) Lennard-Jones potential. dashed curve - - - - - - - - - - - - Because of the particularly simple nature of argon, it seemed interesting to com- pare the experimental data with a theoretical equation of state.14 By fitting the Lennard-Jones and Devonshire equation of state 16 to Bridjgnan’s data at pressures up to 15,000 atm.17 we could obtain approximate isotherms by extra- polating both to higher pressures and higher temperatures. From these isotherms, the temperature of the shock wave could be determined (1450” K for 72,000 atm).This value seemed satisfactory, because similar temperature values are obtained by comparing experimental and theoretical internal energies calculated from the TABLE 2 expt. values calc. values 1) TCK) T(”K) pint from them. from calor. eq. of state eq. of state E (cmJlg) ( k d n z ) (kJ/mole) initial state 0.714 1 - 4.5 0.594 10,400 - 2.0 - 6,300 250 280 { 0475 72,000 + 29.3 - 35,000 1,450 1,380 &&-wave states 0518 30,000 f 7.0 - 15,400 480 550 * Whalley, Lupien and Schneider (Can. J. Chem., 1953,31,722 ; Can. J. Tech., 1955, 33, 111)~ whose C(T> values are not reported here, have determined for pressures up to 80 atm, PY values of argon, which are in good agreement with those of Michels and co-workers.SAUREL, BERGEON, JOHANNIN, DAPOIGNY, KIEFFER, VODAR 69 same Lennard-Jones and Devonshire model.Internal energies in a shock wave are given by the Hugoniot relation : Ef - Ei = HPi + PfXVi - vf), (i andfrefer to initial and final states). On the other hand, theoretical values are known.16 Table 2 gives the comparison and shows also values of internal pressures (3E/Jw), which are markedly negative, of the order of half the external pressure in magnitude. 1 Stoliarov, Ipatiev and Teodorovich, Zhur. Rz Khim., 1950,24, (2), 166. 2 Keyes, J. Amer. Chem. SOC., 1950, 72,433. 3 Michels and Botzen, Physica, 1953, 19, 585. 4 Yntema and Schneider, J. Chem. Physics, 1950, 18, 641. 5 Sam1 and Vodar, J. Rech. C.N.R.S., 1955,33,386. 6 Galperin, Saurel, Lecocq and Vodar, J. Phys. Radium, 1955,16,492. 7 Wouters, Diss. (Amsterdam, 1941). 8 Bartlett, Cupples and Tremearne, J. Amer. Chem. SOC., 1928,50,1275. 9 see, for instance, Slater, Physic. Rev., 1928,32, 349. losee, for instance, Hurschfelder, Curtiss and Bird, MoZecuZar Theory of Gases md 11 Schneider, J. Chem. Physics, 1955,23, 1644. 12 Mason and Rice, J. Chem. Physics, 1954, 22, 843. 13 Dapoigny, Kieffer and Vodar, J. Phys. Radium, 1955,16,733. 14 Bergeon, Kieffer and Vodar, J. Phys. Radium, 1955,16,813. 15 Lennard-Jones and Devonshire, Pruc. Roy. Soc. A, 1937,163,53. 16 Wentorf, Buehler, Hirschfelder and Curtiss, J. Chem. Physics, 1950, 18, 1484. 17 Bridgman, Pruc. Amer. Acud. Arts Sci., 1934, 70, 1. Liquids (1954), p. 228.

ISSN:0366-9033    

DOI:10.1039/DF9562200064

出版商:RSC    

年代:1956

数据来源: RSC

摘要:

IONIZATION OF PIPERIDINE IN METHANOL TO 12,000 ATM BY s. D. HAMANN AND W. STRAUSS C.S.T.R.O. Division of Industrial Chemistry, High Pressure Laboratory, Sydney University, Australia Received 4th June, 1956 This paper reports the first measurements of the effect of pressure on the ionization of a weak electrolyte in a non-aqueous solvent. The electrical conductances of methanolic solutions of piperidhe, piperidinium bromide, sodium bromide and sodium methoxide have been measured to 3000 atm at 25" C, and to 12,000 atm at 45" C. The results show that the basic ionization constant of piperidine in methanol at 45" C increases from 28 x 10-6 mole kg-1 at 1 atm, to 3100 x 10-6 mole kg-1 at 12,000 atm. This is a greater pressure effect than has been found in aqueous solutions of weak bases ; it can be ascribed to the proportionally greater increase in the dielectric constant of methanol at high pressures.In earlier papers 192 we showed that pressure causes a large increase in the ionization of weak acids and bases in water, and that the increase arises from the enhanced solvation of the free ions at high pressures. We have now extended our measurements to solutions of a weak base in methanol, to see how the pressure effect depends upon the nature of the ionizing solvent. The base was piperidine. It would have been preferable to use one of the methylamines whose ionization constants had previously been measured in water to high pressures,l. 2 but unfortunately they are too little ionized in methanol to give significant ionization constants by the conductance method.EXPERIMENTAL MmoD.-The experimental procedure was the same as in the earlier work.192 The conductance measurements to 3000 atm at 25" C were made in a glass cell described previously ; 1 those to 12,000 atm at 45" C were made in a Teflon cell.2 Great care was taken to keep moisture from the reagents and the conductance cells. MATERxALs.-hdytiCa1 grade methanol was refluxed with fresh quicklime, then distilled from magnesium activated by iodine. It was finally distilled several times from anhydrous copper sulphate. The product had a specific conductance of 1.1 x 10-6 ohm-1 cm at 45" C. The sodium bromide, sodium methoxide, piperidine and piperi- dinium bromide were specimens which had been used in earlier work.3 RESULTS ACCURACY The sources and magnitudes of inaccuracies were the same as in the earlier measure- ments? CONDUCTANCES The conductance of each electrolyte was measured for a range of concentrations and pressures : some typical results are given in tables 1 to 3.Tables 1 and 2 illustrate how the conductances change with pressure for particular concentrations. Table 3 shows how they change with concentration for particular pressures. The quantities listed are molal conductances A', calculated from the relation A' = 1000 K/C, 70S. D. HAMANN AND W. STRAUSS 71 where K is the specific conductance in ohm-1 cm, corrected for the contribution of the solvent, and c is the concentraton of electrolyte in mole kg-1. TABLE ~.-MOLAL CONDUCTANCES IN ~ A N O L AT 25" C electrolyte : sodium bromide sodium methoxide piperidinium bromide piperidine 0.0348 conc/mole kg-1 : 0.00137 0.000756 0.00106 press./atm 1 71.5 74.5 82.3 1.12 1000 59.0 61.5 66.7 1-66 2000 51-5 53.8 56.3 2-32 3000 44-23 47.8 48.9 3.05 TABLE 2.-MOLAL CONDUCTANCES IN METHANOL AT 45°C electrolyte : conc/mole kg-1 : press./atm 1 1100 2500 4000 5400 6800 8200 9600 1 1000 12000 sodium bromide OW930 803 67-9 56.4 46.7 42.6 38.0 31.7 26-5 23.1 19.9 sodium methoxide piperidinium bromide 0.0140 0.0110 80.8 70-2 58.3 51.4 46.3 4343 37.9 327 28.0 24.0 86.0 75-3 62.7 525 44.3 39.0 305 25-8 22-5 20.0 piperidine 0.3967 0.344 0-520 0.84 1.25 1-73 2.34 2.68 3.18 3-56 3.79 TABLE 3 .-CHANGE OF MOLAL CONDUCTANCE WITH CONCENTRATION IN METHANOL AT 45" c electrolyte pressurelatm A' sodium bromide conc/mole kg-1: OW075 0.001 37 0.00269 1 74.3 71-5 69.6 3000 45.4 44.8 44.2 sodium methoxide conc/mole kg-1: O.Oo609 0.0140 0.0312 1 705 64.1 58-7 3000 459 44.4 43.7 piperidinium bromide conclmole kg-1: 0.000532 0.00216 0.00415 1 84-9 78.4 752 3000 48.6 47.5 46.1 piperidine concfmole kg-1: 0.0348 0.1283 0.4548 1 1.118 0.620 0.328 3000 3-05 1.78 0.858 IONIZATION CONSTANTS The ionization of piperidine in methanol is represented by the formula CsHllN + CH30H + CSHIINH+ + CH3O-, and the basic ionization constant K is defined as K = (QCsHIlNH+) (QCH~O-)/WSHIIN Y the a's being molal activities.Tables 4 and 5 list values of K calculated from our experi- mental results by the method described previously.172 IONIZATION OF PIPERIDINE TABLE 4.-IONIZATION CONSTANT OF PIPERIDINE IN METHANOL AT 25" C pressurelatm 1Oa Klmole kg-1 pressurelatm 106 Kpnole kg-1 1 6.1 lo00 21.9 100 7.2 2000 56 250 8.6 3000 126 500 14.6 TABLE 5.-IONUATION CONSTANT OF PIPERIDINE IN METHANOL AT 45" c pressurelatm 106 Klmole kgl pressurelatm 106 Klmole kg-* 1 2 8 6800 480 1 100 8.6 8200 860 2500 38 9600 1400 4000 103 1 1000 2300 5400 240 12OOo 3100 DISCUSSION CONDUCTANCES There are two marked differences between the high pressure behaviour of A' for strong salts in methanol and in water.In methanol, A' for a particular concentration is reduced much more by pressure than it is in water. Also, the concentration dependence of A' which is almost unaffected by pressure in water, is greatly reduced at high pressures in methanol. These changes can be judged from the effect of pressure upon the quantities Ao' and B' in the Kohlrausch relation, where c is the molal concentration of the salt and &' is its molal conductance at infinite dilution.Table 6 lists some values of Ao' and B' for the two solvents. A' = Ao' - B'd, TABLE &-THE QUANTITIES &' AND B' AT 25" c electrolyte pressurelatm A,' B'expt. (i) water as solvent : KC1* 1 149 90 3000 158 75 KOCOCH3 t 1 113 80 3000 117 87 (ii) methanol as solvent : NaBr-$ 1 79 190 3000 46 44 CsHllNHBr $ 1 89 225 3000 50 60 * ref. (1). j- measurements made as part of some earlier work (ref. (2)). $ this work. B 'calc. 94 85 85 75 171 83 179 85 In water at 3000 atm, &' is slightly greater than it is at 1 atm ; at higher pres- sures it decreases. In methanol, however, it shows a steady and much larger decrease over the whole range to 12,000 atm.This difference in behaviour is probably due to the greater relative increase in the viscosity 4 of methanol at high pressures. The experimental values of B' in table 6 are subject to fairly large uncertainties, possibly as much as f 20 units at 3000 atm. Nevertheless they show clearlyS. D . HAMANN A N D W. STRAUSS 73 that pressure causes a much larger decrease in B' for methanol solutions than for aqueous solutions. This can be understood on the basis of the Debye-Onsager theory of electrolytic conduction. Onsager calculated B in the relation where the A's are molar conductances and x is the concentration in mole 1.-1. For a 1 : 1-electrolyte his theory gives 5 where D is the dielectric constant of the solvent, 77 is its viscosity in poises, and T is the temperature ("K).Changing to molal units, where p is the specific gravity of the solution. The quantities p, q, D and A( are all pressure dependent. Bridgman 49 6.7 has measured p and 77 for methanol and water at high pressures, and Kyropoulos8 has measured D. Some values of Ao' are given in table 6. From these data it is possible to calculate B': the results are shown in the last column of table 6. Clearly the Onsager values of B' change with pressure in much the same way as the experimental values. For methanol the numerical agreement is not good, but it is known that, even at atmospheric pressure, the Onsager formula applies only approximately to methanol solutions.9 The decrease of B' in methanol at high pressures is caused principally by the large increase in viscosity of the solvent, which reduces the electrophoretic effect. IONIZATION CONSTANTS At 45" C the ionization constant K of piperidine in methanol increases from 28 x 10-6 mole kg-1 .at 1 atm to 3-1 x lO-3mole kg-1 at 12,000 atm.The value of K for piperidine in water at atmospheric pressure and 45" C is 1.2 x 10-3 mole kg-1.10 From this it might be said that methanol at 12,000 atm is a better " ionizing " solvent than water at 1 atm. But it should be emphasized that the change of ionization with pressure arises only partly from the changed properties of the solvent ; at least half of the pressure effect is due to the compression of the ions (the factor (a) below). This was not appreciated by Kritschewsky,ll who attributed the whole of the pressure effect to the change in dielectric constant of the medium, and was thereby forced to adopt unrealistic values for ionic radii.In fig. 1 we compare the effects of high pressures on the ionization of ammonia in water and of piperidine in methanol, The quantity ACi - A??; is the differ- ence between the standard free energy of ionization at the pressure p atm and the corresponding value at 1 atm. It is defined by a -0 A G ~ - AG, = - Some density measurements in this laboratory3 have shown that aAc"/ap for the ionization of piperidine in water at 1 atm is less negative than it is for the ionization of ammonia in the same solvent. We can safely conclude from this that the plot of A G - A S against pressure for the piperidine + water system would lie above the curve for the ammonia + water system.The difference between the two experimental curves in fig. 1 must therefore be ascribed to the change of solvent. In previous papers 1 s 2 we have suggested that the increase in ionization of a weak electrolyte at high pressures is caused by the lowering of the free energy of solvation of its ions. This can be estimated by calculating the Born solvation74 IONIZATION OF PIPERIDINE energy 12 of a pair of ions of about the same size as those of the weak electrolyte, and allowing for (a) the change in the mean radius of the ions with pressure, and (6) the change in the dielectric constant of the solvent with pressure. We have given2 the results of this calculation for the ions Cs+ + F- in water.We have now made similar calculations for the same pair of ions in methanol. The factor - 5 1 I I I 1 '. 0 2 500 5000 7 5 0 0 10000 12500 Pressure/atm FIG. 1.-Full curves : the ionization free energies of weak bases ; dotted curves : the theoretical solvation energies of the ions Cs+ + F-. All the data are for 45" C. (a) is, of course, unaltered by the change of solvent but the factor (b) is more im- portant for methanol than for water because of the larger percentage increase in the dielectric constant of methanol at high pressures.* Fig. 1 shows that the change from water to methanol shifts the predicted solvation energies in the same direction as it does the experimental free energies of ionization. The work described in this paper was carried out as part of the programme of the Division of Industrial Chemistry of the Commonwealth Scientific and Industrial Research Organization, Australia, 1 Buchanan and Hamann, Trans. Faruhy SOC., 1953,49,1425. 2 Hamann and Strauss, Trans. Faraday Soc., 1955,51, 1684. 3 Hamann and Limy Austral. J. Chem., 1954,7,329. 4 Bridgman, Proc. Amer. Acad. Arts Sci., 1925, 61, 57. 5 MacInnes, Princbles of Electrochemistry (Reinhold Pub. Corp., New York, 1939), 6 Bridgman, Proc. Amer. Acad. Arts Sci., 1913,49, 1. 7 Bridgman, Proc. Amer. Acad. Arts Sci., 1912, 48, 307. 8 Kyropoulos, 2. Physik, 1926,40,507. 9 Unmack, Murray-Rust and Hartley, Proc. Roy. SOC. A, 1930,127,228. 10 Hantzsch and Sebaldt, 2. physik. Chem., 1899,30,258. 11 Kritschewsky, Acta physicochim., 1938, 8, 181. 12 Born, 2. Physik, 1920, 1,45. p. 327.

ISSN:0366-9033    

DOI:10.1039/DF9562200070

出版商:RSC    

年代:1956

数据来源: RSC

摘要:

GENERAL DISCUSSION Dr. R. E. Duff (Los Alarnos) said: I would like to comment briefly on the work being done at Los Alamos mentioned by Dr. Cottrell. Dr. Fickett and Dr. Wood, who are doing work on the theoretical equation of state, are certainly cognizant of the implications of detonation experiments for the determination of intermolecular potentials. Extensive calculations of detonation parameters have been made with the free volume equation of state and many intermolecular potentials. The results obtained at fixed composition are now being prepared for publication. I believe it fair to say that none of the calculations gives satis- factory agreement with measurements of detonation pressure and detonation velocity as a function of loading density when the values of the constants in the intermolecular potentials are in the range of those determined from measurements of virial coefficients, viscosity, etc.The calculations have shown, however, the great value of a reliable measurement of detonation temperature. At near crystal density there is a difference of more than 1000” K in calculations of detonation temperature made with different intermolecular potentials. Mr. Boyd is making an attempt to measure the detonation temperature of a homogeneous explosive by a microwave method which should circumvent some of the difficulties encountered in past experiments based on optical spectroscopy. Two papers by Dr. Walsh and co-workers presenting additional equation of state measurements on 15 liquids and 27 solids will be published shortly in J.Chem. Physics (liquids) and Physic. Rev. (solids). Dr. R. A. Buckingham (University College, London) said: Whilst we may accept Dr. Cottrell’s conclusion that an inverse power is seriously inadequate for representing repulsive potentials, it is also important to recognize the limita- tions of a simple exponential form, which is almost equally empirical. Thus, when we use an expression such as to represent a non-valent interaction (fig. 1) the parameter a may serve at least three distinct purposes: (a) it characterizes the steep rise of the potential in region I, (6) it describes the curvature near the minimum in region 11, (c) it de- scribes the rate of diminution of the first-order Coulomb and exchange forces in 111. No single value of a is likely to be wholly sufficient for these thee purposes.In particular one must resist the conclusion that values of a obtained empirically from experimental data relating to solids and gases at low or medium temperatures and normal pressure, provide reliable information about the potential in I ; in many instances these values are more relevant to region 11. For region I, data at genuinely high temperature or pressure are essential. At the other limit, the “effective” value of a in region I11 may be much less than is often assumed, because the correct form of the potential contains terms such as r2 exp (- alp), r3 exp (- alr) which diminish relatively slowly. Dr. Cottrell also mentioned the evidence on the repulsive potential I provided by collision experiments in which the collision energy corresponds to 300eV or more.Here we should anticipate a more complete failure of expression (1) due to the fact that, as R -+ 0, for two similar atoms with nuclear charge 2 (R and Y now being in atomic units). This suggests that a form such as V(R) = (Z2/R)(1 + plR + p2R2 + p3R3 + . . .) exp (- aR) 7576 GENERAL DISCUSSION for the first-order energy, the coefficients p1, p2, p3 being chosen to give the correct initial behaviour, may be a better way of interpreting the results of such experi- ments and may even have a validity extending to larger values of R. .Dr. J. S. Rowlinson (Manchester University) (communicated) : In discussing polyatomic molecules Dr. Cottrell states that chloroform and similar substances deviate little from the principle of corresponding states and therefore observations at low pressures show little influence of the (presumably) steeper repulsion forces between such molecules.An analysis of this effect has been reported by both Hamann and Lambert, in work more recent 1 than that which Dr. Cottrell quotes and by us at Manchester.2 It is clear that chloroform and symmetrical halides of the type XY4 and xY6 do not obey the same reduced equation of state as the inert gases or nitrogen, molecules for which the 12 : 6 inverse-power potential is usually accepted as a satisfactory approximation at low and moderate densities. This is shown most readily by a comparison2 of the reduced vapour pressure curves and of the entropies of evaporation. It is true that such molecules obey a reduced equation similar to that of benzene and other hydrocarbons of like molecular weight, but the behaviour of these is complicated by the orientational forces between their molecules.The discussion of the effects of a steepening of the repulsive part of the radial potential is complicated by the simultaneous presence in all poly-atomic molecules of orientational forces, and it has been shown 3 that it is probable that both types of departure from a simple 12 : 6 potential lead to the same macroscopic effects. Nevertheless the effects are clearly there even at low pressures and it is only the details of their interpretation that are not entirely clear. Prof. A. R. Ubbelohde (Imperial College) said: If the force fields between dissimilar molecules A and B are considered, it is usual to take the van der Waals attraction parameter as the geometric mean of that for the pairs AA and BB, cm = (E~E&.Does any analogous rule apply for repulsive forces? Dr. E. wballey (N.R.C., Canada) said: Dr. Cottrell rightly points out that a wide range of potential-energy curves will fit the second virial coefficients of simple gases to within the experimental errors of the data. In order to obtain 1 Hamann and Lambert, Austral. J. Chem., 1954, 7, 1, 18, 219 ; 1955, 8,21. 2 Rowlinson, Trans. Furuhy Soc., 1954, 50, 617 ; 1955, 51, 1317. Clegg, Rowlinson and Sutton, Trans. Furahy Soc., 1955,51, 1327. 3 Rowlinson, Austral. J. Chem., 1954,7, 397.GENERAL DISCUSSION 77 more information it is necessary to measure the virial coefficients over a very much wider temperature range.On the other hand the viscosities of simple molecules are not well represented by any potential which has been so far tried.1 The experi- mental and theoretical results differ quite significantly. For example, if one tries to determine 12 : 6 parameters from the viscosity, one finds that the parameters are strongly dependent on temperature, and Elk apparently drops to zero at tem- peratures of a few hundred degrees.2 Here, then, is a promising line of attack on the problem of intermolecular forces. The viscosities of simple gases obey the principle of corresponding states to within approximately the experimental error.1 Consequently calculations of the viscosities for more complicated potentials of the form ought to yield valuable information.Dr. S. D. Hamann (C.S.Z.R.O., Sydney) (communicated): Dr. Cottrell has re- ferred to an attempt by McManamey, Pearse and myself 3 to estimate the form of the repulsive potential between polyatomic molecules. I should like to mention that in 1954 Lambert and I 4 refined the original model by taking account of shell-shell and centre-centre interactions as well as the shell-centre interactions. For tetra- hedral molecules with large peripheral atoms we derived a total intermolecular energy which could be represented closely by a 28 : 7 bireciprocal formula. We computed 4 s 5 a number of the bulk properties of materials having this steep- walled potential and showed that it is a decided improvement on the 12 : 6 model for substances like CF4, SnC4, C(CH3)4.There is little doubt that these molec- ules have repulsive forces which, at least initially, rise more steeply than those of the inert gases. Dr. K. M. Guggenheimer (Glmguw University) said: In a forthcoming paper it will be shown that if the logarithm of pressurep is plotted against the volume V as measured by Bridgman,6 the points fall near straight lines for metals, non- metals and compounds, for pressures between 30,000 and 100,OOO atm. The same regularity is found for measurements by Walsh and Christian7 in the pressure range of 1OO,OOO-4oO,O00 atm. Fig. 1 gives some examples. + = .f(rlro) cs Rb K Na LI CsBr C Th U 4 . d 9 I I 1 1 I t I I I I 1 1 - -3 * 4 * S -6 a 7 -8 -9 I v FIG. 1 .-Pressure-volume relation for pressures between 35,000 and lo0,OOO atm.Results of this kind are to be expected if the law of repulsion and thence the free energy are dominated by an exponential factor exp (- kr) (and such a factor is provided by Laguerre functions) and if the compression is one-dimensional, which may be the case in a cylindrical arrangement at high pressures. 1 Whalley, Can. J. Chem., 1954,32,485. 2 Whalley, Trans. Amer. SOC. Mech. Eng., 1954,76, 983. 3 Hamann, McManamey and Pearse, Trans. Faruday Soc., 1953,49, 351. 4 Hamann and Lambert, Austral. J. Chem., 1954, 7, 1, 18, 219 ; 1955, 8, 21. 5 Hamann, Lambert and Thomas, Austral. J. Chem., 1955, 8, 149. 6 Bridgman, Proc. Amer. Acad., 1944-1948.76, 55. 7 Walsh and Christian, Physic. Rec., 1955, 97, 1544.78 GENERAL DISCUSSION Also plots of log (electrical resistance) against the volume lead to straight lines.Even in liquids the same kind of regularity as in solids can be found, namely, p = A exp (-- aV), valid between 5,000 and 50,000 atm. for substances measured by Bridgman. Any other expression for describing the isotherm would need more parameters . Dr. J. S. Rowhson (Manchester University) said: I would like to make a plea for a change of nomenclature in the paper of Buckingham and Pople. They discuss certain properties of gases which can be written as power series in the molar density. They call BQ the second virial coefficient of the property Q. However, if Q is taken to be the pressure, then it is not BQ but the ratio (BQIAQ) which is universally called the second virial coefficient. Such a ratio always has the units of molar volume whatever the dimensions of the property Q.Moreover, the use of such a ratio would make much easier comparisons between the effects of intermolecular forces on different properties. For example, if Q is the dielectric polarization then BQ is - 600 cm6 mole-2 for CH3F at 50" C, but the ratio (BQ/AQ) is -8 cm3 mole-1, and it is this latter figure which can be more properly compared with the second virial coefficient of the pressure series which is - 170 cm3 mole-1. Such a comparison is revealing, for the former coefficient cannot be measured with any greater absolute accuracy than the latter (generally several cm3 mole-1). Furthermore, the former coefficient must include the whole of the experimental error of the latter, as well as any error incurred in its own measurement, under the usual experimental conditions where Q is measured as a function of pressure and not directly as a function of density. Thus the use of the ratio (BQIAQ), rather than BQ itself, gives a more reliable measure of the relative contributions of inter- and intramolecular effects and shows more clearly the accuracy which can be expected in the measurement of the intermolecular effects.Dr. A. D. Buckingham (Oxford University) said: The suggestion has been made by Dr. J. S. Rowlinson, and by Ashton and Guggenheiml that it would have been preferable to have expanded the observed equilibrium property Q in inverse powers of the molar volume Vin the form rather than as in eqn. (1) of the paper by Pople and myself. The advantage of this amended form is said to be the fact that Bb and the other " electromagnetic virial coefficients " would then have the same dimensions for all properties, and would be comparable to the virial coefficients in the equation of state.I should like to point out that in addition to these advantages, there is the disadvantage that, unlike ours, the proposed expansion is inapplicable to those cases when AQ is zero (such a praperty would be the depolarization of the light scattered elastically by particles which are spherical when isolated). Dr. Rowlinson has also made the point that the published data for the Clausius-Mossotti function of methyl chloride at 50" C indicate that Bb = BD/AD = - 8 cm3 mole-1, and this is small compared with the observed second virial coefficient (- 170 cm3 mole-1).He claims that this implies that the measurement of BD by observing E and P might lead to large experimental errors. In Oxford we are trying to overcome this difficulty by building an apparatus which will measure the dielectric constant and the density, so that we shall not need very accurate P, V, T data. We intend to examine substances (such as paraldehyde and AsF3) for which BD is expected to be large, and are hopeful that some knowledge of hydrogen bonds will be ob- tained through an examination of substances like the alcohols. 1 Ashton and Guggenheim, Proc. Physic. SOC. B, 1956, 69, 693.GENERAL DISCUSSION 79 While there is an abundance of information about the concentration dependence of electromagnetic properties of solutions, nevertheless these data are less valuable theoretically than vapour results, for they can lead to quantitative intermolecular- force knowledge only after the use of additional, and possibly bold, assumptions.However, in the case of the Clausius-Mossotti function qualitative information can sometimes be obtained. Because there are great experimental advantages in working with solutions rather than with gases, it is hoped soon to examine the effects of solute interactions on the properties of mixtures. Dr. B. Vodar (Bellewe) said: In reply to the interesting question of Dr. Longuet-Higgins concerning the effect of temperature, there are not sufficient experimental data available on the behaviour of a spectral band as a function of temperature at constant density.Nevertheless, even if these are obtained by somewhat indirect methods they are approximate results1 which lead us to believe that the effect of temperature is not large, which is, of course, in agree- ment with the Margenau statistical theory of broadening and shifts. In this theory no thermal motion is taken into account, and temperature may be intro- duced only in a Boltzmann factor describing the number of molecules with a given interaction energy. Recently we studied the effect of temperature over a large interval in a molecular spectrum. The spectrum studied was the " pressure induced " electronic ab- sorption of 0 2 at 7600A compressed up to 6000 atm. (These results by J. Robin will appear soon in Compt. rend.). The intensity of this band, plotted against the density of the gas, fits a single curve, independent of the temperature of the experiments, which varied from - 78" C up to 300" C .This means that no tem- perature effect is observable in that particular case, which is of course different from the electronic atomic bands discussed above. Concerning the interesting mechanism proposed by Prof. Margenau as possibly responsible for the satellite bands, I would mention that the satellite bands are not always a specific high pressure (or high density) phenomenon. Some of them are observed at quite low pressures also (see, for example, the review pub- lished by s. Robin and Mme. s. Robin I), particularly in emission spectra. Under this condition, only binary collisions are important, and the satellite band produced must be due to binary collisional processes.The " cluster theory " suggested by Prof. Margenau seems to be valid only for explaining the bands observed at elevated pressures and not for describing the totality of the experimental data, but it may constitute the explanation of the specific high pressure phenomena, such as the appearance of a new satellite at very high pressures. The only general statement concerning the origin of the satellite bands, is that they are due to close collisions. Thus they are related to the structure of the " collisional complex ", and may prove to be useful for describing the pro- perties of this complex. The detailed mechanisms responsible for the satellite bands, in addition to the one proposed by Prof. Margenau, may be sought in one at least of the following effects : the splitting of the energy levels of the absorber in the field of the per- turber, and/or some kind of kinetic effect, Mr.Galatry will say something about a very preliminary attempt he made along the second of the above-mentioned lines. Mr. R. P. Bell (Oxford University) said: The calculated effect of foreign gas upon the atomic lines appears to be rather sensitive to the form assumed for the intermolecular forces, as shown in fig. 9 and 10 of the paper by Robin, Bergeon, Galatry and Vodar. Does this mean that the experimental study of such effects is likely to prove a useful way of getting information about intermolecular forces, especially about repulsive forces at short distances ? 1 Robin and Robin (Mme.S.), J. Phys. Radium, 1956, 17, 143.80 GENERAL DISCUSSION Dr. L. Galatry (Bellewe) said: One possible interpretation of the satellite bands may be sought in a kinetic-type model. In this model it seems reasonable to use the Lorentz-Weisskopf collision theory. But it is known that the more advanced form of this theory, which considers only long distance approaches, gives for the line shape a dispersion-type curve without secondary maxima which could be identified with the satellite bands. It is consequently interesting to investigate the effect of central collisions for which the W.K.B. approximation used by Weisskopf is no longer valid because the perturber is going into the strongly repulsive potential regions. It is clear that such strong collisions increase in number with the density.We made a very preliminary attempt to investigate the effect of such collisions. During the electronic excitation process the localization space of the electron is increased. We assumed a very crude picture according to which the atom is compared to a rigid sphere, the radius of this sphere having two different values optically active, ro and rl, (rl > ro) when the atom is respectively in the ground state or the excited state. For mathematical simplicity our calculations have been so far limited to a " cage model " which is obviously adequate for only very high-pressure phenom- ena, and is not convenient for explaining the general binary processes responsible for the satellite bands observed at moderate pressures. The potential in the cage has been chosen as a " square well potential " with infinitely high walls according to the above assumptions.As a result we find an oscillatory band shape with a great number of closely- spaced maxima, the separation of which corresponds to the small vibrational frequency in the potential wall of the cage. The envelope of this oscillatory curve is a band which should correspond to the experimental shape ; this envelope shows a maximum definitely distinct from the unperturbed line. We hope to extend such calculations to binary collisions and to more realistic types of potential-energy curves. Mr. H. de Kluiver (Amsterhm) said: In the van der Waals laboratory the influence of argon on the mercury resonance line, 2537A, has been studied in absorption at densities up to 250 Amagat at 52°C.A medium quartz Hilger FIG. 1. spectrograph with a dispersion of 8*32&mm at 2537A was used. With this dispersion two well-resolved satellites were observed at distances of 80 cm-1 and 171 cm-1 respectively from the principal mercury line (fig. 1).GENERAL DISCUSSION 81 These two satellites have also been observed by Kuhn and Oldenberg, and Preston in absorption and emission below 1 atm.l.293 The second satellite, at 171 cm-1 from the main peak, corresponds to the S1 satellite reported by Robin. The absorption, as measured at the position of the two satellite peaks, is given as a function of the argon density in fig. 2. Hq atom 0 The absorption in the peak of the171cm-' sotellite The obsorption in the peak of the 80cm' sotellite -f? FIG.2. At low densities the absorption at both frequencies increases approximately linearly with the argon density. At higher densities the absorption of the 171 cm-1 peak continues to rise, in contrast with the behaviour of the other satellite. The frequency separation of the satellites from the main peak is constant up to ap- proximately 180 Amagat. Excessive overlapping of the components precludes accurate measurements above this density. Dr. Berni J. Alder (California University) said: The so-called molecular chaos assumption used in the theory of transport processes appears to be a good ap- proximation not only at low density but also at quite high density in a system of rigid spherical molecules. The assumption that the pair distribution function in velocity space can be written as a product of singlet distribution functions must break down, however, at very high density, since there the probability of a given pair of molecules' colliding again after a small number of collisions with other molecules becomes very high.This implies that there is a correlation between the velocities of a given pair of molecules in two succeeding encounters and hence the pair distribution function cannot be factorized. For molecules with attractive potentials where a pair of molecules can be bound together at quite low density and undergo quite a few collisions before separation, the chaos approximation breaks down at lower densities. We have shown the limitation of the molecular chaos approximation by solving exactly on large electronic computing machines the simultaneous equations of motion of a few hundred particles having either a rigid core or a square well potential of attraction.From the history of the path of each molecule it is possible to evaluate the velocity autocorrelation function. An immediate consequence of the molecular chaos approximation is that this autocorrelation function should decay exponentially with time, since this is a Markoff process. For rigid spheres up to a density which corresponds to a volume twice as large as that of close packing, the exponential decay is observed. .At densities higher than that this autocorrelation function seems to develop wiggles, which are characteristic of the behaviour of bound states such as an oscillator. 1 Kuhn and Oldenberg, Physic.Rev., 1932,41,72. 2 Preston, Physic. Rev., 1937, 51,298. 3 Kuhn, H., Proc. Roy. SOC., 1937, 158,212.82 GENERAL DISCUSSION The calculation, therefore, indicates that the Enskog theory of transport properties for hard spheres is valid up to quite high densities. We hope that these machine calculations, when carried out far enough, will enable us to evaluate all the transport properties of dense systems with the single limitation of the small sample involved. Ah. R. P. Bell (Oxford University) said: Eqn. (3.5) and (3.6) of Dr. Whalley's paper illustrate a general point which sometimes causes confusion in interpreting gas imperfections. It is often stated (e.g. eqn. (1) of the paper by Buckingham and Popk) as a general result of statistical treatment that if a property Q of an imperfect gas is written in the form Q = AQ + BQ/V + CQ/P + .. ., then BQ arises from intersections between pairs, CQ from intersections between triplets, and so on. This is not the case for eqn. (3.6) unless there is some special relation between K2 and K3; in particular, if K3 = 0, both BQ and CQ arise from pair interactions. This apparent contradiction arises from the assumption usually made in the statistical treatment that the energy of a triplet can be calculated in terms of the interactions of the pairs which it contains. This will not be the case when chemical interactions are involved, and there is thus a real difference between the two methods of treatment if we are considering the third or higher virial co- efficients.Dr. E. U. Franck (Giittingen University) said: Dr. Whalley showed the Enskog effect to be the main contribution to the thermal conductivity of dense gases near the critical region. In exceptional gases such as hydrogen fluoride, however, the contributions due to the transport of internal energy of clusters or to internal circulation might prevail even at high densities. Because of cluster formation by hydrogen bonds, gaseous hydrogen fluoride at atmospheric pressure possesses an outstanding high thermal conductivity.1 Specific heats attain corresponding high values? Recent measurements showed that the specific heat at constant volume of HF around the critical region also exceeds the values expected for a gas composed only of dipole-containing monomers by a factor of three to five 3 (C, (expt.) = 1.06 or 1-57 cal/g deg.at p = 0-30 or 0.20 g/cm3 at 200" C). The influence of polymerization, therefore, might predominate over collisional transfer of energy in hydrogen fluoride gas even at high compression. Dr. B. Vodar (Bellevue) said: In addition to the data presented in our paper, I wish to give some preliminary experimental results, on the thermal conductivity of nitrogen at elevated temperatures. These results, recently obtained by Mr. P. Johannin, are summarized in fig. 1 and extend at present up to 300" C and about 1000 atm. The evaluation of experimental results for higher temperatures is in progress. Dr. J. S . Rowlinson (Manchester University) said: I would like to ask Dr. Vodar what is the temperature of the mercury surface in the apparatus described in fig.2 of their paper. If it is at all high then the amount of mercury in the compressed gas will be large and hard to estimate. In recent experiments at Manchester we have shown that the volatility of mercury is appreciably enhanced by the presence of a compressed gas. For example, as low a pressure as 30 atm of n-butane is sufficient to cause an increase of 30 % in the mercury concentration in the gas phase at 200" C. I am sure that Dr. Vodar and his colleagues who were amongst the first to study and interpret the solvent power of compressed gases, will not have overlooked this point, but it would be interesting to know how serious the correction is in this type of apparatus. Dr. B. Vodar (Bellevue) said: We were aware of the corrections due to the solubility of mercury in the compressed gas, and we were very interested to find 1 Franck and Spalthoff, Nuturwiss., 1953, 40, 580.2 Franck and Meyer, 2. Elektrochem., to be published. 3 Franck and Spalthoff, 2. physik. Chem., N.F., 1956, 8, 255 ; 2. Elekrwchem., to be published.GENERAL DISCUSSION 83 in the paper published by Dr. Rowlinson an evaluation of the magnitude of these corrections which are quite important even at 150" C. h X I D t n r i r y in Arnoqoti FIG. 1. filling setting mcatur/ng setting 1 FIG. 2. But in our experimental method using " internal heating " we are in a very favourable situation, as even when the bulb of the piezometer (see fig. 2 of the paper) is at the highest temperature (lOOOo C), the mercury level in the capillary84 GENERAL DISCUSSION which is outside and below the furnace, is always kept at a temperature close to room temperature (or even lower, i.e.close to the temperature of the cooling water, as was shown by various experiments where the temperature of the capillary was explored with the help of an auxiliary thermocouple). In our case the conditions are different from those of the usual method, where the whole pressure vessel is heated from outside when the temperature of the mercury is the same as that of the piezometer bulb. The calculation of the error in P, V, T measurements due to the presence of dissolved mercury represents a problem slightly different from the one treated by Dr. Rowlinson. In our case, where the main volume of the gas is hot and the mercury is cold, the concentration of mercury is mainly conditioned by the vapour pressure of Hg at room temperature.This must then be corrected to take into account both the Poynting effect and the effect of intermolecular forces. As- suming equilibrium, it is well known that the Hg concentration in the main part of the gas volume (i.e. in the hot gas) will be smaller than in the neighbourhood of the liquid Hg level because Hg, the heaviest component, tends to concentrate in the colder part. In order to calculate the value of the correction to apply in this case, Mr. Bergeon (paper to appear in J. Rech. C.N.R.S.) estimated the thermodiffusion ratio by treating the gas phase as a mixture of two gases interacting according to the usual 6 : 12 law, and by using available results of the kinetic theory of trans- port phenomena.As a final result, it was shown that for the measurements reported in our paper (i.e. for moderate densities for which it is possible to make a reasonable estimate of the thermodiffusion ratio) the relative magnitude of the correction for the measured pressure is only of the order of 10-4. Dr. E. U. Franck (Gottingen University) said: According to the experiments of Dr. Hamann and Dr. Strauss the degree of ionization of weak electrolytes in water and methanol is increased by application of high pressure. Corresponding results have been obtained with supercritical steam as a solvent. Steam at high density is a fairly good solvent at least for quartz and certain salts of alkali metals.Some of them behave as electrolytes in the dense vapour phase. This has been proved already for NaCl between 378 and 393" C by Fogo, Benson, and Copeland.1 Recently 2 we measured the electrolytic conductance of KCl, KOH and HCl in liquid water and steam up to 750" C and 2800 kg/cmz. The mole fraction x of the solute varied from 1.8 x 10-5 to 1.8 x 10-4 (corresponding to 0.001 and 0.01 m solutions at standard conditions). The electrolytic cell was an autoclave made of Ni + Co+ Cr-containing alloy with 35 cm3 internal volume, platinum lined and with a Bridgman closure. The central electrode was introduced by means of a long thick-walled stainless steel tube and was insulated by a long narrow tube made of sintered corundum. The density of the fluid solutions was calculated from measured P, T-values by means of the P, T, V-data for pure water published by Kennedy? Smoothed values of the equivalent conductivity obtained with KCl solutions are presented in fig.1 in a three-dimensional form. At densities below p = 0.2 g/cm3 KC1 is a weak electrolyte. It appears to be almost completely dissociated, however, beyond p = 0-5 gIcm3. Because of decreasing ionic mobility the equi- valent conductivity is descending again at higher densities. The maximum conductivity attained exceeds the value for liquid aqueous solutions at standard conditions by a factor of ten. With increasing mole fraction of KCl the maximum of conductance is lowered and shifted to higher density. Using reasonable assumptions concerning viscosity and dielectric constant of supercritical steam the ionic mobilities at infinite dilution of the solute and the 1 Fogo, Benson and Copeland, J.Chem. Physics, 1954,22,212. 2 Franck, 2. physik. Chem., N. F., 1956,8, 92, 107, 192. 3 Kennedy, Amer. J. Sci., 1950,248, 540.GENERAL DISCUSSION 85 dissociation constants K for KCl, KOH and HCl can be calculated. The K-values are slightly decreasing with rising temperature and strongly increasing with rising density (e.g. K(KC1) = 3.6 x 10-5 and 3.7 x 10-6 mole/l. at p = 0.30 g/cm3 and t = 450" C and 750" C resp. K(KC1) = 1.2 x 10-2 mole/l. at p = 0.70 g/cm3 and t = 500" C). From the energy and entropy of dissociation the average hydration number of 6 to 7 for K+ or C1- ions at supercritical temperatures is obtained. The average hydration energy connected with the attachment of one water molecule to the central ion has a value of 12-13 kcal/mole. FIG.1.-Electrolytic equivalent conductivity A of KCI in water at subcritical and super- critical conditions. Constant mole fraction x(KC1) = 1.8 x 10-5 (corresponding to 0001 m at standard conditions). While the results for KOH are quite similar to those for KCl the maximum conductance of HCl in supercritical steam does not occur until a density of 0.80 g/cm3. Thus HC1 at medium densities of 0.30 to 0.60 g/cm3 proves to be con- siderably weaker as an electrolyte than KC1 and KOH. Conclusions as to the degree of hydrolysis of alkali chlorides and the ionization constant of pure water at supercritical conditions can be derived from this result.Dr. E. Whalley (N.R.C., Ottawa) said: The third virial coefficient of argon has been measured in two other laboratories besides those discussed,l and in view of the high uncertainty in the values of third virial coefficients, it seems to be worth while to compare these data with the theoretical calculations. Mr. J. Jeener and Mr. M. Lambed (University of BrusseZs) said: The theories of solutions 2 have been able, in recent years, to predict many of the thermodynamic properties of mixtures of simple molecules : excess free energy, volume, enthalpy, compressibility. . . . A further test of these theories was attempted by measuring the variation of volume on mixing under pressure for the system CCb + C(CH3)4. This allows one also to calculate the excess free energy at high pressure from its low 1 Holborn and Schultze, Ann.Physik., 1915,47, 1089. Holborn and Otto, 2. Physik., 1924, 23, 7 7 ; 30, 320. Masson and Dolley, Proc. Roy. Soc., A, 1923, 103, 524. Tanner and Masson, Pruc. Roy. Suc. A, 1930, 126,268. 2 Prigogine, Bellemans and Mathot, MoZecufar theory of solutions (North Holland Publishing Co., Amsterdam), to be published soon. Prigogine, Bellemans and Englert-Chwoles, J. Chem. Physics, 1956, 24, 518.86 GENERAL DISCUSSION pressure value (which is known from vapour pressure measurements) and its derivative with respect to pressure, the excess volume. The glass apparatus shown in fig. 1 is filled with weighed quantities (about 0-5 c m 3 each) of the pure components A and B by vacuum distillation and sealed at a and b.The level of the mercury at c is known from the resistance of a thin platinum wire which is stretched in the 0.5 mm diameter glass capillary. This level is measured as a function of pressure before and after mixing, and the differ- ence gives the excess volume. The liquids can be mixed by taking the apparatus in its metal envelope out of the thermostat, turning it upside down and shaking. I-I FIG. 1. 0 6 0 0 atm. 200 4 0 0 0 FIG. 2.-Excess volume of the system neopentane 4 carbon tetrachloride as a function of pressure (mole fraction = 05). expt. value at 23.6" C. - - - - - - c alc. value 1 at 5" C. The results of the first measurement on the system CCb + C(CH3)4 (at 23.6" C) are shown in fig. 2 together with the prediction of the theory of Prigogine, Bellemans and Englert-Chwoles (at 5" C).It is seen that the predicted and experimental excess compressibilities are quite large (about 20 % of the compressibilities of the pure substances). It appears likely that, at moderately high pressures (say lo00 atm), the excess volume might change its sign. Further measurements are in progress and it is hoped that a detailed account will be published soon. Mr. R. P. Bell (Oxford University) said: Hamann and Strauss have attributed the large effect of pressure on basic dissociations in methyl alcohol partly to a change in ionic radii, and partly to an increase in the dielectric constant of the solvent. These equilibria are not simple dissociations, but involve an acid-base reaction with the solvent, and it therefore seems likely that the effect of pressure is to increase the acidic properties of methyl alcohol, probably by modifying its hydrogen-bonded structure.This structural change would also affect the di- electric constant, but it does not seem reasonable to attribute the change in the apparent basic strength of amines purely to a dielectric effect. prof, A. R. Ubbelohde (ImperiaZ College), said: The reported influence of pressure on the dielectric constant of methanol, and the studies on aqueous solu- tions above the critical temperature of water reported by Dr. Franck, raise the general problem about the " structure " of water and other hydrogen-bonded solvents at high densities and temperatures, around and above their critical temperatures.GENERAL DISCUSSION 87 Near the melting point at ordinary pressures, water, for example, is generally thought to contain a considerable proportion of clusters with the tetrahedral ice structure. Some of the anomalous properties of water as a solvent are associ- ated with the presence of such clusters.Do any of the properties of aqueous solutions show abnormal changes with temperature and density in regions where tetrahedral clusters largely disappear? Dr. E. U. Franck (Giittingen University) said: Accurate determinations of the electrolytic conductivity of supercritical steam solutions have not yet been carried out within a temperature region of about f 3" C of the critical point of water. Beyond this boundary no exceptional behaviour of electrolytic con- ductance has been observed hitherto which could positively be ascribed to the influence of typical critical phenomena, i.e. abnormal cluster formation, etc. At temperatures beyond 400" C the concentration of well-defined hydrogen- bonded water polymers as occurring in liquid water at room temperature seems to be unimportant even at densities as high as 0.80 g/cm3. Evidence for this con- clusion is derived from the measured conductivity of HC1 at supercritical tempera- tures, which coincides within about 10 % with that of KC1 at corresponding degrees of dissociation. Obviously there is little or no anomalous mobility of hydrogen ions. As the mechanism of this extra mobility depends upon polymer water molecules with hydrogen bonds the concentration of the latter appear to be very small at supercritical temperatures. Dr. L. E. Sutton (Oxford University) said: It may be suggested that the solvent properties of water are not a consequence of its structure per se but of certain molecular characteristics which also are the cause of this structure. It has long been recognized that water can solvate either cations, through the donor property of its oxygen atom, or anions, through the acceptor property (for hydrogen-bond formation) of its hydrogen atoms; and that this solvating power is responsible for the special solvent power. The same molecular characteristics cause its macro- molecular solid or liquid structure. High temperatures will tend to reduce such solvation ; but high pressures will favour it ; and this may explain the solubility of electrolytes under the supercritical conditions. The importance of water in such studies may be due also to the fact that it is one of the relatively few good solvents which can be heated to high temperatures without decomposing. Other substances of marked donor or acceptor character (or both) might also be effective, provided that they will stand the necessary temperature.

ISSN:0366-9033    

DOI:10.1039/DF9562200075

出版商:RSC    

年代:1956

数据来源: RSC

摘要:

B. CHEMJCAL REACTIONS AND TRANSFORMA- TIONS AT HIGH PRESSURES THE ROLE OF THE SOLVENT IN CHEMICAL REACTIONS, AS REVEALED BY HIGH PRESSURE STUDIES BY K. J. LAIDLER Dept. of Chemistry, The University of Ottawa, Ottawa, Canada Recehed 18th June, 1956 Pressure effects on equilibrium constants and on reaction rates depend respectively upon overall volume changes and upon volumes of activation. The factors influencing these volume changes are discussed with particular reference to aqueous solutions and to ionic reactions. It is shown that for such systems the volume depends principally upon the first power of the charge on the ion (in contrast to the entropy, which depends on the square of the charge). Observed overall volume changes and volumes of activa- tion are found to be capable of interpretation mainly from the standpoint of electro- striction effects.The significance of the temperature coefficients of the volume changes is considered from the same point of view. Studies of the effects of hydrostatic pressure on reaction rates and equilibria in solution, besides being of considerable interest in themselves, are important in throwing light on the detailed mechanisms of reactions and on the effects of solvents on reaction rates and equilibria. In this paper an attempt is made to analyse and discuss what has been learnt from such studies, particularly from those on aqueous solutions and with ionic reactions. The basic relationship between the equilibrium constant K of a reaction and the pressure p is the van’t Hoff equation 1 where A V is the total molar volume change when the reactants are converted into products. For the variation of the rate constant k of a reaction with pressure van’t Hoff derived the analogous equation A v* fF)T =- RT 9 where AV* is the molar volume change when the reactants are converted into the activated state.The derivation of eqn. (2) from eqn. (1) involves the assumption that the rate of a reaction depends only on the conditions prior to the formation of the activated complex. Pressure effects are frequently discussed in terms of absolute rate theory, but it is important to realize that eqn. (1) is no more dependent on the assumptions inherent in that theory than is the Arrhenius law ; in particular, the question of the correctness of the factor kT/h and the significance of the entropy of activation are in no way involved.88K . J . LAIDLER 89 An alternative way of writing eqn. (2) is (SAF*/Sp), = A V*, (3) where AF* is the free energy of activation. This equation applies whether the free energy of activation is defined, according to absolute rate theory, as - RTln (kh/kT), or, according to Hinshelwood and co-workers,2 as - RT In k. Values of AV* can be calculated from plots of log k (or AF*) against p, and lists of such values are given in various places.39 4 Plots of this type are generally fairly linear, and even when they are not AV* is still related to the slope at any point. The values quoted are usually the volume changes at atmospheric pressure, calculated from the initial slopes, and it is these values that will be considered in this paper.SIGNIFICANCE OF VOLUME CHANGES The problem of understanding pressure effects on equilibria and rates thus reduces to one of Understanding the volume changes that occur in the overall re- action or in the activation process. In a discussion of volumes of activation Evans and Polanyi 5 pointed out that two effects must be considered : the changes in the volumes of the molecules themselves, and changes of volume in the neigh- bouring solvent molecules. Hamann and his collaborators6*7 have shown that for the rates of certain reactions involving changes in polarity the solvent effects predominate. This view has been supported by Bums and Laidlerp who showed for a group of reactions of widely different ionic types that the volumes of activa- tion are primarily determined by the influence of changes of polarity in the reaction system on the surrounding solvent molecules.In brief, when ions of the same sign come together, or when two neutral molecules unite to form a polar substance, there will be a contraction due to the binding of solvent molecules; conversely, if the electrical field is weakened during the reaction (as when two ions of opposite sign come together) there will be a release of bound water molecules and a consequent expansion. These considerations apply both to total volume changes and to volumes of activation, the general rule being that the activated complex has properties that lie somewhere between those of the reactants and the products. For both rates and equilibria there is a great need for a theoretical or empirical treatment of partial molal volumes that will make it possible for these quantities to be estimated in a reliable manner on the basis of the known characteristics of the molecules or ions involved; such a treatment would allow predictions to be made about the effects of pressure on equilibria and rates, and would significantly increase our understanding of the role of the solvent in reactions.Recently Couture and I8 have made some progress by an analysis of the partial molal volumes of ions of various kinds. For monatomic ions in aqueous solution it has been found possible to correlate these values satisfactorily with the valencies z of the ions t and the crystallographic radii r. If the partial molal volumes of the ions are expressed in ml per mole and are referred arbitrarily to a value of zero for the proton, and r is expressed in A, we find that the following equations are obeyed : for cations (4) and for anions (5) The mean deviations of the experimental values from these equations are 2.4 and 2-7ml respectively.A simple interpretation of these equations is that the - V+ = 16 + 4.9r3 - 202+, v- = 4 + 4.9r3 - 20 12- I. - -f Throughout this paper z represents the number of unit charges on the ion, and is positive for cations and negative for anions. The symbol 1 z I is used to remesent the absolute value of the number of unit charges.90 ROLE OF THE SOLVENT volume occupied by an ion in aqueous solution may be considered to be made up of two terms, as follows : (i) a term related to the intrinsic volume of the ion itself, and therefore pro- (ii) a term related to the electrostriction of water molecules in the neighbour- We are at present attempting to extend the treatment to uncharged molecules and to polyatomic ions, and are trying to include the effects of charge distributions in the molecules or ions.A complete treatment of pressure effects on equilibria and rates awaits a more detailed understanding of these partial molal volumes. In the meantime we may consider some reactions between ions, and see how empirical equations such as (4) and (5) above may aid in the understanding of the behaviour of such systems. Before doing this we may note that eqn. (4) and (5) show a linear dependence of the volume on the charge on the ion.This means that if we consider an ionic reaction the volume change will depend in part on the quantity, portional to r3, and hood of the ion, and therefore related to 1 z 1. A'A + BIB -2 c'c + D'D, which may be written as Zz. If this quantity is zero, i.e. if there is no charge neutralization (e.g. if all charges are of the same sign), the volume change may be expected to be small, since it arises only from variations in the radii. A negative Zz means that the reaction has been accompanied by charge neutral- ization, and here AV is expected to be positive; conversely, a positive Ez will lead to a volume decrease. These remarks apply equally to volumes and to volumes of activation. The entropy changes accompanying reaction or the process of activation, however, depend upon different factors.Powell and Lather9 and others10 have suggested that the entropies of ions also depend upon I z I, but it has recently been shown 11 that various lines of evidence point to the fact that the true depend- ence is upon 22. This being so the entropy change in the above reaction will depend upon which will be written as 2 2 . There is a number of reactions for which 2 2 and Zz differ markedly-frequently even in sign-so that we should not in general expect to find close correlations between volumes and entropies. Burris and Laidler4 recently suggested, on the basis of the results for a series of reactions of various ionic types, that there is a correlation between entropies and volumes of activation, but probably such correlations are not as general as was supposed (cp.later). OVERALL VOLUME CHANGES.-we may now consider briefly the application of these ideas to the ionic equilibrium zC2 + zD2 - zA2 - zB2, Tl+ + 2Fe3+ J Tl3+ + 2Fe2+. Table 1 shows the crystal radii for these four ions, and the partial molal volumes that are calculated using eqn. (4). For three of the ions there are observed values,* and these are in reasonably satisfactory agreement with the calcu1ated.t The calculated AV for the reaction is - 8.9 ml, and from this one concludes that there will be an increase in equilibrium constant with increasing pressure; a -/- The uncertainty in the experimental values is probably as great as 5 ml in some cases particularly for the ferric ion.K . J . LAIDLER 91 pressure of lo00 atm should in fact increase K by a factor of 1.44.The relatively small volume change found is a reflection of the fact that Ez is zero. The value of 2 2 2 on the other hand is - 8, and one would predict an entropy increase; the observed value is 50.4 cal/mole deg. TABLE PA PARTIAL MOLAL VOLUMES OF IONS - - V+ V+ (CalC) (obs) T1+ 1.49 12.2 14-7 ~ 1 3 + 1 -05 - 39.3 - Fe2+ 0 8 3 - 21.2 - 25.3 Fe3 + 067 - 42.5 - 44.6 A V (calc) for Tl+ + 2Fe3+ + Tl3+ + 2FeZ+ = - 8.9 ml. ion crystal radius Much larger volume changes are expected when there is a charge neutralization. In the process HCO3- H+ + CO32- there is a neutralization when the reaction occurs from right to left. The value of Z,Z for the process from left to right is 2, and a volume decrease is therefore expected ; the actual value is - 27.8 ml, and a pressure of lo00 atm would increase the dissociation constant by a factor of 3-12, The value of 1 2 2 is here 4, and A s is - 35 cal/mole deg.It follows from eqn. (4) and (5) that as a rough rule the volume change is although there may be important deviations from this arising from the contribution of the r3 term, The empirical equation for entropies 11 similarly leads to the con- clusion that the entropy change in an ionic reaction will be approximately given by and this has been found to be obeyed quite satisfactorily.11 VOLUMES OF ACTNATION.-h considering the effects of pressure on the rates of ionic reactions one is essentially considering volume changes in processes such as AV- - 20 ZZ (6) AS'== - 10 222 (7) AZA + p 7- XZA+ZB f , where X, is the activated complex; the charge on this is the algebraic sum of those on the reactants.From the preceding discussion one may distinguish between two types of reactions, according to whether the reacting ions are of the same, or opposite signs. (i) In the former case Zz is zero, and only a small volume change is expected. Two reactions of this type in aqueous solution have been studied experimentally, namely, and CH2BrCOO- 3- S ~ 0 3 ~ - --j CH2(S203)C002- + Br-, and the volumes of activation at 25.0" C are - 6.1 12s 3 and - 4.8 4 ml respectively. For reactions of this type the value Of k2 is equal to 2zAzBy and a negative entropy of activation is therefore expected. As a rough rule the entropies of activation of ionic reactions are equal to - 1ozAzB ; the actual values for these two reactions are - 11.6 12 and - 17.0 13% 4 cal/mole deg., in satisfactory agreement.A simple view of the first of these reactions is that it occurs through an activated complex of structure CH2CIC00- + OH- -+ CH20HC00- + CI-, H H \ / I HOB--. ....... C .... ....el 4- coo- 692 ROLE OF THE SOLVENT The total charge t on this complex is 2, which is equal to the total charge on the reactants. The fact that AV* is negative suggests that the total charge on the complex is somewhat greater than 2, and this would be the case if its structure is something like H H coo- the total charge on which is 3 (the net negative charge is, of course, 2). Such an activated complex may be expected to bind water to a greater extent than the reactants, and therefore to give rise to a negative AV*.Similar remarks-apply to the second reaction. (ii) If the reaction is between ions of opposite signs, Zz and xz2 are both negative, the latter again being equal to 2zAzB. Positive volumes and entropies of activation are therefore expected, and are generally found.4 An example is the reaction Co(NH3)5Br2f + OH- -4 Co(NH3)sOH2+ + Br- for which AV* is 8.5 ml,4 and AS* is 21.7 ; 1 4 ~ 4 the latter value agrees closely with the prediction of eqn. (7). During the formation of the activated complex in this reaction 2 z is - 2, and application of eqn. (6) leads to 40 ml for the volume of activation. That the true value is much less than this is not surprising, since undoubtedly in the activated state the charge neutralization has not proceeded to completion.As our understanding of ionic volumes increases it may become possible to draw conclusions about the degree of polarity of activated complexes on the basis of the values obtained for volumes of activation. The conclusions with regard to pressure effects on rates are that the effects will be very small for reactions between ions of the same sign, and that the volumes will in this case not show any correlation with the entropies. For reactions be- tween ions of the opposite sign, on the other hand, there may be some correlation with entropies of activation, and the volumes of activation will in general be smaller than would be predicted by eqn. (6) owing to the fact that neutralization will seldom be complete. TEMPERATURE DEPENDENCE OF VOLUME CHANGES One other aspect of pressure studies that throws some light on the behaviour of the solvent during reaction is the temperature dependence of the volume changes. A convenient way of measuring such a temperature dependence is to make use of the thermodynamical relationship SAS& =- 6AV]8T.(8) SAS*/Sp = - 8A V* JST. (9) It follows that if one measures overall entropy changes, or entropies of activation, over a range of hydrostatic pressures, one can calculate the temperature coefficient of the overall volume change or of the volume of activation. This procedure is mathematically equivalent to calculating A V or AV* at each temperature and so determining the temperature coefficient directly. There appear to be in the literature no data on equilibria to which this pro- cedure can be applied, although Fajans and Johnson 15 have presented and dis- cussed some temperature coefficients for the partial molal volumes of individual The analogous equation applicable to rates of reaction is f- By this is meant the sum of the absolute values of the charges on all groups.K .J. LAIDLER 93 ions in aqueous solution. One interesting result that they point out is that there is something of a linear relationship between the temperature coefficient and the volume itself. This is understandable in the light of empirical relationships such as eqn. (4) and (3, which imply that electrostriction effects play a very im- portant part in connection with ionic volumes. An ion with a large volume therefore tends to be one in which the water molecules are not strongly held, and the temperature coefficient of weakly-held water will be similar to that of unbound water.If the water molecules are strongly bound, however, they will not be as much disturbed by a rise in temperature; in these circumstances there will therefore be low volumes and low temperature co- efficients. The correlation between temperature coefficients and volumes cannot be ex- pected to be very close, for various reasons. One of these, which is parti- cularly important for complex ions, is illustrated by the fact that a low volume may arise from the strong binding of a small number of water molecules, or from the weaker binding of a large number. In the former case there will be a correla- tion between temperature coefficient and volume, but in the latter case the tempera- ture coefficient may be quite high.It will be seen later that there are two reactions in which it appears that the activated complexes bear a " smeared " charge which tends to bind a number of solvent molecules in a relatively loose manner. As far as rates are concerned there are six reactions for which suitable data are available (table 2). In fig. 1 the entropies of activation for these six re- actions are plotted against the pressure, and it is to be seen that there is a fairly linear dependence of AS* on the pressure. From the slopes of the lines the values of 6AY*/6T have been calculated (table 2). Another interesting relationship may be pointed out. If for each of the six reactions the TAS* values obtained at the various pressures are plotted against the corresponding energies (or heats) of activation the dependence is found to be linear, and the slope unity.This re- sult means that the influence of pressure 00 m U G; \ o w e s 5 In 0 U r? I x 0, 8 U o g 00 I Y f v) on the free energy of activation hF* is much less than that on its componen& AH* and TAS* ; in other words, there is almost complete compensation between changes in AH* and TAS*. Such a compensation has been noticed many times94 ROLE OF THE SOLVENT previously,l9 although not when pressure is the variable. It has been observed in both equilibrium and rate studies, and for both solvent and substituent effects. A striking case of the same type of linear relationship has recently been observed by Pap& and Canady in connection with the effects of substituents on the ionization constants of phenols,2* anilines and other compounds.FIG. 1.-Plots c TAS* against pressure for six reactions. The values of 8,V*/6T are proportional to the slopes of the lines, and are listed in table 2. Since such effects are found in equilibrium studies as much as in rate studies it is clear that they are not to be explained in terms of the nature of collisions during the course of reaction. The most plausible explanation seem to be in terms of the effects of changes in polarity of the reaction system on the behaviour of the neighbouring solvent molecules. Factors that tend to increase the binding of solvent molecules during the course of reaction (or during the activation process) bring about a decrease in TAS or TAS*, and at the same time a decrease in AH or AH*, and as a rule there appears to be an almost exact compensation between the two factors.As a result, solvent effects of this kind frequently escape recog- nition when equilibrium constants or rates are only studied at one temperature. Turning now to a consideration of the values of 8hV*/8T shown in the last column of table 2, for three reactions (1, 2 and 3) the values are negative, while for three (4, 5 and 6) they are positive. Reactions 1, 2 and 3 are all of the same type, and involve a significant increase in polarity as the activated complex is formed (this is revealed by the negative AV* and AS* values). The solvent molecules are therefore more tightly bound in the activated state than in the initial state, and in view of this it is not difficult to understand why the coefficient of expansion of the system in the activated state should be less than that in the initial state. In reaction 6 there is a decrease of polarity, and a consequentK.J. LAIDLER 95 reduction in tightness of binding, resulting in a higher temperature coefficient for the activated state as compared with the initial state. The results for reactions 4 and 5 cannot be explained on the simple view that there are correlations between volumes and their temperature coefficients. As discussed earlier, a low volume results from the weak binding of a number of water molecules ; this may be the case if the charge is “ smeared ” over a large area instead of being concentrated in a smaller one.This type of explanation may well apply to reaction 5, the activated complex for which may resemble structure I1 shown above. This has a larger total charge than the reactants (3 instead of 2) and will therefore occupy a smaller volume. Owing to the “ smearing ” of the charge, however, none of the water molecules will be bound as strongly as in the initial state, and the temperature coefficient of the volume of the activated state will therefore be greater than that of the initial state. A similar situation may exist in reaction 4, although the reason is now some- what different. The methyl group in the ortho-position will screen the solvent molecdes from the ions that have been formed incipiently in the activated state.There is more binding of solvent in the activated state as compared with the initial state (as indicated by the AV* of - 26.6 ml), but it appears that, owing to the screening, these molecules are not as strongly bound as they are in the initial state; in the initial state there may be an appreciable amount of fairly strong binding by the nitrogen atom in the amine, whereas in the activated state the nitrogen atom will be shielded. This interpretation receives some support from the fact that its frequency factor is 250 times that for the reaction of the para-substituted compound; 1 8 ~ 1 7 the higher AS* for the ortho-compound is very probably due to the exclusion of bound solvent molecules. Canady and Papee 20 have recently discovered similar entropy effects in their calorimetric studies of the ionizations of urthu-, meta- and para-substituted phenols and other compounds.In connection with effects of the kind discussed in this section it is noted that the 6V/6T values will be affected markedly by the solvent. Values for the molal coefficients of expansion of various solvents are listed in table 3, in which the figure for water is much less than for other solvents. The 8A V*/8T value for reaction 5 is close to that for the expansion of water, and is therefore consistent with the hypothesis that during the formation of each activated complex a single water molecule becomes released from a state of complete immobility to one which corresponds to the state in ordinary liquid water. This hypothesis is undoubtedly too simple, however.TABLE 3.-MOLAR COEFFICIENTS OF EXPANSION OF SOLVENTS T=25”C solvent HZ0 CzHsOH CH3OH CH3COCH3 CHCI4 -00879 6 V/6T (mI/mole deg.) *08535 *05153 a1085 -1093 CONCLUDING REMARKS.-Although some of the COnClUSiOnS drawn in this paper are speculative, it does seem that most of the AV* and 8AV*/6T values that have been obtained are capable of explanation in terms of relatively simple concepts regarding the effective volumes occupied by molecules and ions. The results provide considerable support for the view that electrostriction effects play a predominant role in connection with these volume changes. Since the rate of a chemical reaction is proportional to the concentration of activated complexes, the problem of understanding the nature of activated com- plexes is of paramount importance in chemical kinetics.It is becoming increasingly clear that interactions between activated complexes and solvent molecules play a very important role,4921 and high pressure studies, when interpreted in the light of these effects, contribute much to our knowledge of these matters.96 ROLE OF THE SOLVENT 1 Van’t Hoff, Vorlesungen iiber theoretische und physikaiische Chemie (Braunschweig, 2 Stubbs and Hinshelwood, J. Chem. SOC., 1949, 1180. 3 Moelwyn-Hughes, The Kinetics of Reactions in Solution (Clarendon Press, Oxford, 4 Burris and Laidler, Trans. Faraday SOC., 1955,51, 1497. 5 Evans and Polanyi, Trans. Faraday SOC., 1935,31,875. 6 Buchanan and Hamann, Trans. Faraday SOC., 1953,49, 1425. 7 David and Hamann, Trans. Faraday Soc., 1954,50, 1188. 8 Couture and Laidler, Can. J. Chem., 1956,34. 9 Powell and Lather, J. Chem. Physics, 1951, 19, 1139. IoCobble, J. Chem. Physics, 1953, 21, 1443, 1446, 1451. Connick and Powell, J. Chem. Physics, 1953, 21, 2206. 11 Laidler, Can. J. Chem., 1956,34. 12 Perrin, Trans. Faraday SOC., 1938, 34, 144. Williams, Perrin and Gibson, Proc. 13Kappanna, J. Indian Chem. SOC., 1929, 6, 45. Kappanna and Patwardhan, J 14Brernsted and Livingston, J. Amer. Chem. SOC., 1927,49,435. 15 Fajans and Johnson, J. Amer. Chem. SOC., 1942,64, 668. 16 Gibson, Fawcett and Perrin, Proc. Roy. SOC. A, 1935,150,223. 17 Weale, J. Chem. SOC., 1954, 2959. 18 Evans, Watson and Williams, J. Chem. SOC., 1939, 1345. 19 Evans and Polanyi, Trans. Faraday SOC., 1936,32, 1333. Bell, Trans. Faraday Soc., 1937, 33, 496. Barclay and Butler, Trans. Faraday SOC., 1938, 34, 1938. Laidler and Eyring, Ann. N. Y. Acad. Sci., 1940,39, 303. 1901), vol. 1, p. 236. 1943, chap. 11. Roy. SOC. A, 1936,154,684. Indian Chem. SOC., 1932,9, 379. 20 Pap&, Canady and Laidler, Can. J. Chem., 1956, 34. 21 Laidler and Landskroener, Trans. Faraday Soc., 1956,52, 200.

ISSN:0366-9033    

DOI:10.1039/DF9562200088

出版商:RSC    

年代:1956

数据来源: RSC

摘要:

THE DECOMPOSITION OF BENZOYL PEROXIDE IN SQLUTION AT HIGH PRESSURES BY A. E. NICHOLSON AND R. G. W. NORRISH Dept. of Physical Chemistry, University of Cambridge Received 11 th July, 1956 The rates of decomposition of solutions of benzoyl peroxide in carbon tetrachloride have been measured at temperatures of 60" C and 70" C in the pressure range 0 to 3000 kg/cm2. The results show that the decomposition at high pressures has the same mechan- ism as at atmospheric pressure and consists of a unimolecular decomposition into free radicals accompanied by a radical-induced chain decomposition. The effect of pressure is to decrease the rate of the unimolecular decomposition but to increase the rate of the chain decomposition. It is concluded that the effect of pressure on the rate of initiation in the peroxide- catalyzed polymerization of styrene is small and not an important factor in the large increase observed in the rate of polymerization. In the course of a study of the peroxide-catalyzed polymerization of styrene at high pressures, Merrett and Norrish 1 have shown that the rate of polymer- ization is greatly increased by the application of high pressures. The profound effect of pressure upon the overall rate of polymerization might be due to an in- crease in either the rate of initiation or the rate of propagation or to a decrease in the rate of termination, or to a combination of these factors.The initiation process consists of the decomposition of a molecule of benzoyl peroxide to give free radicals with the subsequent attack on the monomer mole- cules by the free radicals to start growing polymer chains.The rate of initiation is thus dependent upon the rate of decomposition of the benzoyl peroxide and the investigation reported in this paper was undertaken to determine the effect of pressure upon the rate of initiation. Nozaki and Bartlett 2 studied the decomposition of benzoyl peroxide at zero pressure in a wide range of solvents and expressed the rate of decomposition by the following equation : (1) - dP/dt = klP + kcP3/2, where P = concentration of peroxide. According to the conclusions of Nozaki and Bartlett, the first term in the above equation represents a unimolecular de- composition and the second term a chain decomposition induced by the free radicals present in a solution of decomposing benzoyl peroxide.The value of the unimolecular constant kl does not change very much from one solvent to another but the value of k,, the constant governing the radical-induced chain decomposition, is markedly dependent upon the solvent due to the differences in the reactivities of the radicals participating in the chain decomposition. In peroxide-catalyzed polymerization the number of growing polymer chains started is dependent upon the rate of the unimolecular decomposition. In choosing a solvent in which to study the effect of pressure upon the peroxide decomposition, it was therefore important to select one in which the unimolecular decomposition was not overshadowed by the chain decomposition. Carbon tetrachloride was used because it satisfied this condition, kl/kc being approximately 0.5 at 70" C.D 9798 DECOMPOSITION OF BENZOYL PEROXIDE EXPERIMENTAL MATERIALS.-The benzoyl peroxide was purified by dissolving in chloroform, pre- cipitating with methyl alcohol, filtering off and then drying in wucuo at 30" C. The carbon tetrachloride was redistilled before use. METHOn-The decomposition was followed by allowing aliquot portions of the benzoyl peroxide solution to react at prescribed temperatures for given times, then quenching the reaction by cooling in ice and estimating the unchanged reactant. For the experiments at pressure, the apparatus used by Merrett and Norrish 1 was employed, the peroxide solution being confined in glass tubes over mercury in tubular steel containers which fitted into the thermostatted steel reaction vessel.Pressures up to 3000 kg/cm2 were applied to the reaction vessel by a primary pump coupled with an intensifier unit, using as hydraulic fluid throughout the apparatus a thin lubricating oil with low setting point, Shell Tellus 11. For experiments at " zero " pressure the re- actant was contained in sealed evacuated tubes. In the preparation of the tubes used in the pressure experiments, 2 ml of the solution of benzoyl peroxide were placed over mercury in glass tubes having a constriction at the top. It was found that a slight reaction took place between the solution and the mercury but that this could be reduced to negligible proportions by using narrow tubes and by having a small glass piston floating on the surface of the mercury, so that the area of the mercury surface in contact with the solution was reduced to a minimum.The tubes were attached to a vacuum line, the contents frozen in liquid air, and evacu- ated for 15 min. They were then allowed to warm to room temperature with evolution of any dissolved air, then frozen and evacuated again. This procedure was repeated and then the tubes were sealed off at the constriction. The lower ends of the tubes were now broken off under mercury and the tubes placed in stainless steel containers filled with mercury. Each glass tube now contained an oxygen-free sample of solution con- fined over mercury at atmospheric pressure. Three such tubes were prepared and placed in the thermostatted high-pressure reaction vessel which was then sealed and the pressure applied.One of the products of the decomposition of benzoyl peroxide is carbon dioxide. This gas remains in solution at the high pressures used during the experirnents but, on releasing the pressure, is liberated. The reaction tubes had therefore to be long enough to prevent the carbon dioxide from expelling the solution when the pressure was released. Fhe tubes used were 15 cm long and had an internal diameter of 7 mm. At the end of the experiment, the pressure was released and the tubes removed and cooled rapidly in ice. A 1 ml sample of the solution was then pipetted into a solution of sodium iodide in acetic anhydride and 10 min were allowed for complete liberation of iodine. Water was then added and the iodine titrated with a 001 N or 0.002 N solution of sodium thiosulphate for peroxide solutions having initial concentrations of 0.2 M and 0.05 M respectively.No indicator was used in titrating the solution having an initial concentration of 0 2 M but the 0.05 M solutions were titrated to a starch end-point, a positive correction of O5ml of 0002N thiosulphate being applied. The thiosulphate solutions were standardized at frequent intervals against a standard solution of potassium iodate. For the measurements at zero pressure in sealed tubes, the same evacuation and analytical procedure as above was followed. RESULTS The rates of decomposition of peroxide solutions having initial concentrations of 0 2 M and 0.05 M have thus been measured under oxygen-free conditions at 60" C at pressures of lo00 and 20oO kg/cm2 and at 70" C at pressures of 1O00, 2OOO and 3000 kg/cm2.The pressure range over which measurements could be made was limited by the freezing point of the solvent, which Bridgman 3 has shown is 60" C at 2500 kg/cm2 and 70" C at 2850 kg/cm2; but because of the ease with which liquids supercool at these high pressures, it was possible to measure the rates at 70" C and 3000 kg/cm2 although this pressure is greater than the freezing pressure at this temperature. If the solution did freeze, it was apparent because the decomposition was almost completely halted. The rate of decomposition has also been measured at " zero " pressure at 60" C, 70" C and 80" C.A . E. NICHOLSON AND R . G. W. NORRISH 99 The concentrations were corrected for the temperature and pressure, the correction for the compressibility of the solvent being made using the values for the compressibility of carbon tetrachloride given by Bridgman.4 Eqn.(1) valid for zero pressure has been found in this work to apply equally well to the decomposition in carbon tetrachloride at high pressures and it may be concluded that there has been no significant change h the mechanism as a result of the application of pressure although, as will be shown, the values of the rate constants are affected. Employing the treatment of Nozaki and Bartlett? we obtain on integration where a = kl/k,, PO is the initial concentration of peroxide and P the concentration after a time t. The value of the ratio a was determined experimentally by using data from two I / P"2 FIG.1.-Graphs for the determination of a. 0, 80°C; X = 1.5; 0 + (> 0, 70°C; X = 1.0; 0 + @, 60°C; X=O. The numbers on the graphs indicate the pressure in 103 kg/cm2. runs with different initial concentrations of peroxide and samples removed after the same time intervals in the two runs. If PI and P2 are the concentrations after equal times in the two runs, and C is a constant, we have which may be rearranged to give Hence, by plotting I/ y'P1 against l / l / P 2 , a can be obtained from the slope and intercept of100 DECOMPOSITION OF BENZOYL PEROXIDE the graph. kl is obtained by plotting In (" - ;r) against t and kc from the relation- ship kc = ki/a. The results obtained are shown in fig. 1-4. In fig. 1, l/z/P~ is plotted against (l/+Pz) + X, where X is an arbitrary figure chosen so that the graphs may all be shown on the same diagram and from the slopes and intercepts of the graphs values of a are obtained. Using these values of a, the values of log Time i n hours FIG.2.-Decomposition of benzoyl peroxide at 60" C. 0 kg/cm2 0, X' = 0.420 ; 8, X' = 0.260 ; 1000 kg/cm2 0, X'= 0196; 8, X'= 0110; 2000 kg/cm2 0, X'= 0.112; 8, X'= 0.060. The numbers on the graphs indicate the pressure in 103 kg/cm2. and log r+) - X' is plotted against t in fig. 2, 3 and 4, where X' is an arbitrary figure chosen so that several graphs may be shown on the Same diagram. The values of the constants kl and kc obtained are summarized in table 1. TABLE 1 temp. "C pressure, kglcm* a kit-1 kc(mole/l)-4 h-' 60 1 lo00 2000 70 1 1000 2000 3000 80 1 *368 -134 -070 494 -210 -108 -083 648 -0052 1 000339 -0025 1 -0210 -01 54 -01 12 -0104 -0810 ~0141 -0253 -0358 -0425 -0733 -104 -125 *125A .E . NICHOLSON AND R . G . W. NORRISH 101 Time in hour, FIG. 3.-Demmposition of benzoyl peroxide at 70" C. , X' = 0.300 ; lo00 kg/cm2 0, X'= 0.140; 0, X'= 0.265; 2000 kg/cm2 0, A"= 0150; 0, X'= 0.080; 3000 kg/m2 0, X'= 0125 ; 0, X'= 0070. The numbers on the graphs indicate the pressure in 103 kg/cmZ. 0 kgIm2 0, X' = 0480 ; FIG. 4.-Dewmposition of benzoyf peroxide at 80" C in vacuo. 0, x'= 0 5 8 ; 0, X'=0.39.102 DECOMPOSITION OF BENZOYL PEROXIDE DISCUSSION The activation energies calculated from these results are given in table 2. The variation in the activation energies with pressure is within the experimental error.In cases where the effect of pressure on the activation energy has been accurately determined, it has been found to be small and not the most important factor 0 . 4 - U & TABLE 2.-ACTIVATION ENERGIES pressure El Ec kg/cm2 cal Cal 1 32,600 25,600 lo00 34,200 24,100 2000 33,800 24,100 in the variation of the rate constant with pressure.5 The values of El and E . are in good agreement with the values obtained with benzene as a solvent by Nozaki and Bartlettz (33,300 cal and 25,230 cal respectively). The effect of an increase in tern- perature will therefore be to increase the importance of kl and this is shown by the increase in the value of a with temperature at any particular pressure. As a result of this increasing importance of kl, benzoyl peroxide is more efficient in initiating polymerization at higher temperatures as reported by Redington.6 The effect of pressure on the two rate constants is best shown by plotting log (k,/kO) against pressure, where kp is the constant at a pressure p and ko is the constant at zero pressure.The results are plotted in this way in fig. 5 and it will be seen that an increase in pressure has opposite effects on the two constants kl and k,, the former being decreased by pressure and the latter increased. The scheme postulated by Nozaki and Bartlett 2 based on the formation of phenyl benzoate in the decomposition is as follows : Pressure in kq/cm2 FIG. 5.Variation of rate constants with pressure. ki (PhCOO)2 + 2PhCOO. k2 2PhCOO. -j C02 + PhCOOPh k3 PhCOO. + (PhCOO)2 + PhCOOPh + C02 + PhCOO..All the observed decomposition products of benzoyl peroxide can be fitted into the above type of scheme. The benzoate radicals may decompose further into phenyl radicals and carbon dioxide. The phenyl radicals so formed may combine with each other to form diphenyl and may also propagate the chain decomposition. Chain transfer with the solvent may also occur. The expression for the rate of decomposition will have the same form whether benzoate radicals or phenyl radicals or solvent radicals are the chain propagators. From the above scheme we derive the expression (5) for the rate of reaction, where P is the concentration of the peroxide. This is to be compared with eqn. (1) from which it will be seen that while kl is a true unimolecular constant, kc is composite.- dP/dt = klP + k3(k1/k#P3I2A. E. NICHOLSON AND R . G . W. NORRISH 103 The effect of pressure on the rate of a chemical reaction is given by the expression of Evans 7 (see also Glasstone, Laidler and Eyring 8) : d l n k AV* V * - y i dP RT RT ' -_--- _ - ~ - where Y* is the partial molal volume of the activated complex and 6 that of the reactants. The direction of the pressure effect on the rate constant thus depends on whether the activated complex has a greater or smaller volume than that of the reactants. In a unimolecular decomposition we should expect the activated complex to have a larger volume than the original molecule. Such a reaction would there- fore be expected to show a decrease in rate with increased pressure and this has indeed been found to be the case for the unimolecular component of the present reaction.A decrease in the rate constant with increase in pressure was also ob- served in the only other unimolecular decomposition upon which the effect of pressure has been studied. This is the decomposition of phenyl methyl benzyl ally1 ammonium bromide in chloroform which Gibson, Perrin and Williams9 showed has a rate of decomposition at 3000 kgIcm2 which is 0.65 of the rate at atmospheric pressure. In a bimolecular reaction the activated complex is concluded to have a smaller volume than the volume of the reactants, resulting in an increase in the bimolecular constant with pressure, as has been found for the many bimolecular reactions studied by Perrin.10 k2 and k3 may therefore be expected to increase with pressure, and since kl has been shown to decrease the effect of pressure on k, is made up of two opposed factors, k3 which increases and (kl/k2)4 which decreases.Since kc in fact increases with pressure, the effect of the former must be greater than that of the latter. The net effect of pressure is to increase the overall rate of decomposition, and according to eqn. (1) the increase will be the greater the higher the initial concentra- tion of peroxide P. The rate of initiation of polymerization will depend on kl and to a first approximation is uninfluenced by the chain reaction of decomposition since it does not remove radicals. In addition the chain decomposition in the presence of styrene will tend to be suppressed due to preferential reaction of the peroxide radicals with styrene rather than unchanged peroxide. Thus, judging from the effect of pressure on kl which is about halved by the application of 3000 kgIcm2, the rate of initiation of polymerization may be ex- pected to be slightly reduced by pressure, and the cause of the tenfold increase in the rate of polymerization of styrene observed by Merrett and Norrish at 3000 kg/cm2 must be sought elsewhere than in the initiation of reaction. We are indebted to the Distillers Company for a grant for apparatus and for a maintenance grant to one of us (A. E. N.) during the tenure of which this work was carried out. 1 Merrett and Norrish, Proc. Roy. SOC. A, 1951,206,309. 2 Nozaki and Bartlett, J. Amer. Chem. SOC., 1946,68, 1686. 3 Bridgman, Physic. Rev., 1914, 3, 126. 4Bridgman, Proc. Amer. Acad., 1931, 66, 185. 5 Shu-lin 'Peng, Sapiro, Linstead and Newitt, J. Chem. Soc., 1938,784. 6 Redington, J. PoZymer Sci., 1948,3, 503. 7 Evans, Trans. Faraduy SOC., 1938 34,49. 8 Glasstone, Laidler and Eyring, The Theory ofRate Processes (New York, 1941), p. 470. 9 Gibson, Perrin and Williams, Proc. Roy. SOC. A , 1936,154,684. 10 Perrin, Trans. FararJqv SOC., 1938,34, 144.

ISSN:0366-9033    

DOI:10.1039/DF9562200097

出版商:RSC    

年代:1956

数据来源: RSC

摘要:

POLYMERIZATION OF STYRENE AT HIGH PRESSURES USING THE SECTOR TECHNIQUE BY A. E. NICHOLSON AND R. G. W. NORRISH Dept. of Physical Chemistry, University of Cambridge Received 11th July, 1956 The polymerization of styrene, photosensitized by benzoyl peroxide, has been studied at 30" C at pressures in the range 1 to 3000 kg/cmZ. The average lifetimes of the growing polymer chains have been determined by the rotating-sector method. Rates of polymer- ization and molecular weights have also been measured and the results used to calculate the propagation and termination constants. The results show that pressure has very little effect on the rate of initiation but that the propagation constant increases exponentially with pressure. The termination constant decreases rapidly up to lo00 kglcm2 and then decreases slowly as the pressure increases further.The increase in the propagation constant with pressure is attributed to the fact that propagation is a " slow " bimolecular reaction. The decrease in the termination constant is related to the accompanying increase in the viscosity of the monomer with pressure. The effect of pressure on the polymerization of styrene is considerable. Merrett and Norrish 1 have shown that the velocity of reaction at 60" C is increased some ten-fold exponentially over a range of 1 to 5000 kg/cmZ, while the average molecular weight increases with pressure up to 3000 kglcm2 and thereafter remains approximately constant. In the previous paper we have presented evidence that the rate of initiation is only slightly decreased by pressure and it becomes clear that the origin of the effects must be sought in the propagation and termination reactions.With this end in view we have determined the values of the constants of pro- pagation and termination at various pressures up to 3000 kg/cm2 for the photo- sensitized polymerization using benzoyl peroxide as photocatalyst. This was done by measuring the lifetimes 7 of the growing polymer chains by the rotating- sector method, and combining the results with corresponding determinations of the degrees of polymerization derived from measurements of viscosity. EXPERIMENTAL APPARATUS.-The high pressure apparatus was identical with that used by Merrett and Norrish 1 except that a quartz window, mounted on a modified form of the outlet plug, made it possible to irradiate the styrene in the high-pressure reaction vessel.An adaptation of the packingless method of mounting developed by Poulterz was used. The window was a cylindrical block of fused quartz, 3, 2, or 1 in. thick, with optically polished ends and was mounted directly against an optically flat steel surface as shown in fig. 1. Under these conditions there can be no leak because the intensity of stress on the bearing surface is greater than the hydrostatic pressure by the principle of the un- supported area. It was necessary for the steel surface to be precision ground to a good optical finish and this was difficult with the V30 steel of which the outlet plug was made. The difficulty was solved by surface hardening this face before precision grinding.was found convenient, in practice, to hold the window in place by means of a brass collar which prevented the window dropping off before the pressure was applied. Provision was also made for a safety window consisting of three quartz discs +in. thick to give protection in the event of fracture of the window. The quartz windows normally stood 104A . E. NICHOLSON AND R. G . W. NORRISH 105 between ten and forty applications of pressure before conchoidal fracture occurred, Pressures were measured by means of a strain gauge. PURIFICATION OF REACTANTS.-styrene was separated from the hydroquinone inhibitor by shaking three times with 10 % potassium hydroxide and then three or four times with distilled water, subsequently separating and drying over anhydrous sodium sulphate for an hour.The monomer was then degassed, placed in the distillation apparatus and distilled with a still surface at room temperature under an applied vacuum of 0.01 mm, the middle fraction being collected in the reaction tubes. The benzoyl peroxide was recrystallized at frequent intervals by dissolving in chloroform and precipitating with methyl alcohol, afterwards filtering off and drying in vacuo at 30" C. FIG. l.-Optical system and high pressure reaction vessel. A, mercury lamp; E, safety windows; J, piston; B, 1 % copper sulphate solution ; C, rotating sector ; G, reaction tube ; L, lens ring; D, glass filter; H, n-hexane; M steel container. F, quartz window ; K, oil ; METHOD OF mma-The Pyrex reaction tubes had flat end plates as shown in fig.1. They were washed with chromic + sulphuric acid, sulphur dioxide solution and distilled water and dried in wacuo before use. The catalyst was introduced in the required quantity as a standard solution in A.R. chloroform into the reaction tubes, which had previously been weighed, and the chloroform was removed under low vacuum. The styrene was then distilled into the reaction tubes, which were cooled in liquid nitrogen, under an applied vacuum of 0.01 nun. The quantity of styrene distilled into the reaction106 POLYMERIZATION OF STYRENE tube was carefully adjusted so as always to give the same path length for the light (4 cm). The reaction tube was then sealed off in wacuo and the monomer + catalyst mixture frozen and stored over liquid nitrogen until required.PRocmvRE.-Immediately prior to carrying out an experiment the reaction tube containing the styrene and the catalyst was weighed, so that the catalyst concentration was accurately known. The end of the tube was then broken under mercury and the reaction tube transferred to its stainless steel container. To ensure that the reaction tube was correctly aligned with the light beam, the steel container had a set of screws which ensured that the reaction tube was accurately centred in the steel container and able to move along a vertical line only. The steel container had external studs which determined its position in the high-pressure chamber. The experimental arrangement is shown in fig. 1. n-Hexane was used as the hydraulic fluid H in the reaction vessel so that there should be no.absorption of light between the window F and the reaction tube G, fresh hexane being used for each experiment. The hexane was separated from the light oil K, used as the hydraulic fluid throughout the rest of the apparatus, by a piston J which was of the usual type utilizing a Bridgman packing. The positioning studs of the stainless steel container rested on a spring (not shown in fig. 1) which ensured that the reaction tube G was always in contact with the window F as shown in fig. 1. The rotating sector C, having sectors of 90" and thus giving equal periods of light and darkness, was driven by a synchronous motor using a gear box which enabled a wide range of speeds to be obtained. The sector was painted black and was so placed that the diameter of the light beam at the point where it was cut by the sector was small, Under these conditions, the time taken by the sector edge to cross the light beam was small compared with the length of the light flash and so the light flashes approximated to the square form required by sector theory.The light source A was a Siemens 1 k W high-pressure point-source mercury lamp operated from 200 V 50 clsec. The lamp circuit contained a wattmeter and a rheostat. Any slight variation in the voltage was compensated by adjusting the rheostat so that the wattmeter reading and hence the output of the lamp remained constant throughout an experiment. The fastest sector speed gave a light flash of 013 sec duration. This was sufficiently long to prevent any apparent effects due to the fact that, since an a.c.lamp was used, the light output fluctuated at 50 c/sec. Heat radiation in the light beam was removed by a 1 % solution of copper sulphate in the tank B. The solution was cooled by means of a spiral of copper tube through which a constant flow of cold water was main- tained. After passing through the tank B, the light beam was focused to a narrow parallel beam by three convex quartz lenses as shown. The glass filter D, through which the beam passed before entering the reaction vessel ensured that only light of wavelengths greater than 3 3 0 A was used to irradiate the styrene. The transmission curves of the glass filter, 4 cm of pure styrene and 4 cm of a 0.074 solution of benzoyl peroxide in styrene were measured with a spectrophotometer.A comparison of these transmission curves showed that the light used to irradiate the styrene was sufficiently weakly absorbed for a uniform concentration of active centres to exist throughout the reaction system. At the end of an experiment the mixture of monomer and polymer was transferred to a beaker of known weight and weighed. It was then dissolved in methyl ethyl ketone to give an approximately 1 "/, solution and the solution was poured into a large volume of methyl alcohol to precipitate the polymer. The precipitated polymer, after standing overnight, was filtered into a sintered-glass Gooch crucible, washed with methyl alcohol and then dried at 30" C under a pressure of 0.1 mm. The polymer was found to reach constant weight in about 24 h.All reaction rates were measured gravhetrically in this way. RESULTS MEASUREMENTS OF THE LlFETIME OF CHAIN PROPAGATION The rate of polymerization of a specimen of styrene having a particular catalyst con- centration, was determined under steady illumination. Then, with the lamp running at the same wattage, the rates of polymerization of samples of styrene, having the same catalyst concentration, were determined under intermittent illumination for a wide range! of sector speeds. Finally, as a check that the light output from the lamp had not altered appreciably during the series of experiments, the rate under steady illumination was determined again.A . E. NICHOLSON AND R. G . W. NORRISH 107 It was necessary to correct for the dark reaction. The theory of the sector method given by Burnett and Melville 3 has been extended by Matheson et af.4 to include the dark rate and leads to the following expressions. n2 tanhm + - - n, - _ n2 - - (1 -9) tanh rn tanh (xmr) ns - _ - 2[x tanh m + tanh (xmr)] + [(l - x2)2 tanh2 m tanh2 (xmr) + 4(x tanh m + tanh (xmr))(x tanh rn + x2 tanh (xmr))l* (3) where r = ratio of the dark to light period, x = ratio of the dark rate to the rate under steady illumination, no = average concentration of active centres under intermittent illumination, n, = average concentration of active centres under steady illumination, nl = concentration of active centres at beginning of the dark period, n2 = concentration of active centres at end of the dark period, m = number of times that the duration t of the light flash is greater than T, the The rate of reaction is proportional to the concentration of active centres and so the ratio of the rates under intermittent and steady illumination varies in the same way as the ratio of the concentrations of active centres n0/ns.To determine T, the ratio of the rate for a particular sector speed to the rate under steady illumination was plotted against 2[x tanh m + tanh (xmr)] average lifetime of the active centres. I.0- 0.9 - E 2 c + + 0 . 8 - I I - 1.0 00 1.0 lo9 t 0 0 FIG. 2.-Typical curve showing .variation of rate of polymerization with sector speed. log t, where t was the duration of the light flash. The values of n0/ns for different values of m were calculated for the particular dark rate from eqn. (l), (2) and (3) above.Theoretical curves of n0/ns against log t were then calculated for various assumed values of T using the fact that log t = log m + log T. The theoretical curve which best fitted the experimentally determined points gave the value of 7, the average lifetime of the108 POLYMERIZATION OF STYRENE active centres. As a typical example, the experimental points for a catalyst concentra- tion of 0.0393 M at lo00 kg/cm2 are shown in fig. 2, together with the theoretical curve for = 0.28 sec and a dark rate 7-6 7; of the rate under steady illumination. MOLECULAR WEIGHT MEASUREMENTS The degree of polymerization was determined from measurements of the viscosities of solutions of polymer in benzene at 25" C. The B.S.S. viscometer used had a flow time of 130 sec for the pure solvent. The molecular weight was calculated from the intrinsic viscosity by means of the relationship given by Mayo, Gregg and Matheson 5 for unfrac- tionated samples of polystyrene in benzene solution at 25" C.Mn = 178,000 [q]1'37. The intrinsic viscosity in this expression is based on concentrations expressed in g/lOO ml. - COMPRESSIBILITY OF STYRENE To calculate the rate constants at any particular pressure it was necessary to know the monomer concentration at that pressure and so the compressibility of styrene had to be determined. The piezometer used was similar to that of Tait,6 consisting of a bulb provided with a capillary silvered on the inside by which any volume changes in the fluid were appropriately measured. The piezometer, whose total volume had been accurately determined, was filled with styrene, placed in a stainless steel tube containing mercury and then immersed in the oil in the pressure vessel.pressure relative density The pressure was applied to the monomer by means of the mercury and the volume of the 1 1.OOO 0.897 styrene determined by the position of the lo00 0942 0952 mercury in the capillary. The extreme height 2000 0.906 0.990 reached by the mercury in the silvered capillary 3000 0.884 1.015 was determined by the height to which the silvering had been dissolved away by the mercury. A correction was applied for the compressibility of the glass of which the piezometer was made. This was done by the method given by Bridgman 7 assuming the cubic compressibility of the glass to be 2.5 x 10-6.The measurements were made at 30" C -C 0.1" C and the results are set out in table 1, together with the densities calculated from these values, taking the density of styrene at atmospheric pressure and 30" C to be 0897 as given by Patnode and Scheiber.8 TABLE 1 g'ml kglcmz volumes A separate determination was necessary for each pressure. RATE OF POLYMERlZATION The rate of polymerization, average lifetime of the active centres and molecular weights of the polymers formed have been measured at pressures of 1,1000, 2000. and 3000 kg/cm*. All measurements were made at 30" C. The experiments at atmospheric pressure were carried out in the pressure apparatus. This method, although not as rapid or convenient as the usual dilatometric technique, was preferred because it enabled the results obtained at high pressure to be compared with results obtained at atmospheric pressure, using exactly the same experimental arrangement.All measurements were confmed to the first 10 % of polymerization and the majority to the first 7 %. At atmospheric pressure the rate of polymerization was observed to decrease slightly as the reaction proceeded, as reported by Melville and Valentine? This effect was most noticeable with the higher peroxide concentrations. Lifetime measure- ments were therefore confined to peroxide concentrations between 0074 M and 0.0135 M. The 0.0135 M solution gave a steady rate over that part of the reaction studied and the 0-074 M solution gave rates which decreased only very slightly as the reaction proceeded.The effect was not observed at high pressures. The rate of polymerization was found to be proportional to the catalyst concentration raised to the power 0.50 f 0.02 at atmospheric pressure. As the pressure was increased, this index was found to decrease from 0 5 0 at atmospheric pressure to 0.45 at 3000 kg/cmz. A similar trend was found by Merrett and Norrish in their study of the peroxide-catalyzed reaction at 60" C. In that case, the index was found to decrease from a value of 050 at zero pressure to about 040 at 3000 kg/cm2 and thereafter remained constant or rose slightly.A. E . NICHOLSON AND R. G . W. NORRISH 109 The following scheme is assumed for the polymerization, where C represents a catalyst molecule, R a free radical, M a monomer molecule, P, a growing polymer chain of n units and M, a " dead " polymer molecule of n units.initiation propagation P, + M --t Pn+l k2 termination Pn + P, --t M,,, k3 transfer P, + M -+ M, 3- Pi k4 The stationary concentration n of growing centres will then be given by dn/dt = klf(Z)CM - k3n2 = 0, where M is the monomer concentration. I Kq /sq cm / IOOOKq/sq cm 0 0 1: 0 2000Kq/sq fCGQKq/sq 3 0 c m c m 10 2 0 Rote in mole/t. sec I 10s FIG. 3.-Variation of the reciprocal of the degree of polymerization with rate of polymerization at different pressures. Iff; is the average degree of polymerization, (4) The rate of polymerization R is given by R = kznM and thus, substituting for n in eqn. (6) Hence by plotting 1/75 against the rate R we are able to obtain the value of k3/2k22 from the slope of the graph.was calculated from the molecular weights and l/F is plotted against R in fig. 3 which shows how the slope k3/2k22M2 varies with pressure.110 POLYMERIZATION OF STYRENE Another relationship is needed to calculate k2 and k3 and this is provided from lifetime no. of active centres per unit volume no. of active centres disappearing per unit volume per second studies. If 7 is the average lifetime of the growing centres we have 7 = .. 7 = n/lc3n2. (8) r = k2Mik3R. (9) Substituting for n in eqn. (S), Hence by measuring lifetimes at different rates, k2/k3 may be calculated. The results of such measurements together with the corresponding values of l/F and the catalyst con- centrations are set out in table 2. If we neglect transfer, it is also possible to calculate k&) from the results given in table 2.Assuming no transfer the expression for 7; becomes (10) = (2 x rate of polymerization)/rate of initiation = 2R/klf(l)CM. Hence klf(I) == 2R/FCM. (1 1) TABLE 2 catalyst rate of pressure concentration polymerization z k2/k3 1 /F kgicm* mole/l. mole/l. sec sec 1 0.0185 4.32 X 10-5 0.19 0.95 X 10-6 3.70 x 10-3 0.370 6-56 X ,, 0.17 1-29 X ,, 5.45 x ,, 0.0740 7.42 x ,, 0.12 1.03 x $, 6-50 x ,, 1000 0.0393 1.12 x 10-4 0.28 3-42 X ,, 1.76 x ,, 0.0786 1-29 X ,, 023 3.24 x ,, 213 x ,, 2000 0.0205 1.60 X ,, 0.40 6.55 x ,, 5-90 x 10-4 0.0410 2.01 x ), 0.3 1 6.75 X ,, 7.93 x ,) 3000 0.0084 214 x ,, 0.72 1.58 X 10-5 1.78 x ,, The values of the individual rate constants k2 and k3 together with klf(I) calculated from the above results are given in table 3.Recent work 22 has shown that termination OCCUTS by combination of two growing polymer chains. The constants are therefore calculated assuming termination by self-neutralization of chains ; if disproportionation is the mechanism of termination, the values of kp and k3 given must be doubled and the value of klf(l) halved. TABLE 3 pressure k2lk3 k2 k3 k i f (1) k22'k3 (1. mole-1 w-1) (1. mole-1 sec-1) (1. mole-1 sec-1) kglcm2 1 1-09 x 10-6 079 x 10-4 725 6.65 x 107 19.2 x 10-7 lo00 3-33 ), 3.59 y, 108 3.25 ,, 9.9 7, 3.05 ,, 9.8 YY 2000 6.65 ,, 1.37 X 10-3 206 3000 1.58 X 10-5 6.30 ,, 400 2-54 ,, 10.5 ,, DISCUSSION RESULTS AT ATMOSPHERIC PRESSURE The results obtained at atmospheric pressure are set out in table 4, together with those of other workers corrected to 30" C for the purposes of comparison, assuming an activation energy of 4 5 kcal/mole for kzlk3.There is a scatter of about a factor of 8 for k2/k3 and of 3 for k22/k3, but the results presented here can be regarded as consistent with each other even if theA. E . NICHOLSON AND R . G. W. NORRISH 111 absolute values are modified later. The value of k2/k3 is in good agreement with those of Matheson and of Majury and Melville while the value of k221k3 lies between the results of Matheson and of Melville and Valentine. reference this work Melville and Valentine 9 Matheson et. al. 10 Bamford and Dewar 11 Burnett 12 Majury and Melville 13 Grassie and Melville 14 method sector sector sector viscosity dilatometer dielectric constant interferometer TABLE 4 kdk3 1-09 x 10-6 5.46 ), 1.09 Y Y 7.60 >, 7-42 ,, 1.14 3.36 ,, k22/k3 7-9 x 10-5 12.9 ,, 6.1 YY 17.5 y, - - - k2 72.5 26 55 23 - - - k3 6.65 X 107 5.25 X 106 5.05 x 107 3-03 X 106 - - - EFFECT OF PRESSURE ON RATE CONSTANTS The values of k ~ f ( l ) which depend on the light intensity (table 3) are seen to be sensibly constant for the experiments at pressure, when the mercury lamp was run at 850 W.The value of this constant at atmospheric pressure is not directly comparable for in these experiments the lamp was run at 1000 W. The FIG. 4.-Variation of rate constants with pressure. measurement of the other constants k2 amd k3 is of course unaffected by this change of intensity. Thus the initiation process is seen to be very insensitive to pressure change in agreement with the results of our study of the effect of pressure on the thermal decomposition of benzoyl peroxide described in the previous paper.The effect of pressure on the propagation and termination constants is best shown by plotting the logarithm of (k)p/(k)l against pressure where is the value of the constant at a pressure P and (k)l is the value at atmospheric pressure. The results are plotted in this way in fig. 4 and show that the propagation constant increases exponentially with pressure. Chain growth is a bimolecular reaction112 POLYMERIZATION OF STYRENE with which a large decrease in entropy is associated, as has been pointed out by Dainton and Ivin,ls Evans 16 and others. It would therefore be expected to show the characteristics of a ' slow ' bimolecular reaction which Gibson Fawcett and Perrin 17 and Perrin and Williams 18 have shown experimentally to be strongly accelerated by pressure. The conclusions of Merrett and Norrish 1 in this respect are fully confirmed. The termination constant shows a rapid decrease up to 1000 atm, followed by a slow but steady decrease above this pressure.We believe that this effect on k3 has its origin in the increase in viscosity produced by pressure and to be of the same nature as the decrease in k3 resulting from changes in the viscosity of the medium produced by other means. Norrish and Smith19 showed that the acceleration in the velocity observed in the later stages of the polymerization of methyl methacrylate is caused by a decrease in k3 with increasing viscosity as polymerization proceeds and ascribed it to the progressive immobilization of the propagating centres with respect to the termination reaction.This effect was confirmed by Burnett and Melville 20 and by Matheson ef aZ.10 for styrene. It is of interest to discover whether any empirical relationship exists between the termination constant and the viscosity of the medium. No data are available for the effect of pressure on the viscosity of styrene but measurements have been made on several similar liquids (Bridgman 7). These data are set out in table 5, the logarithm of the viscosity being given because the viscosity increases very rapidly with pressure. TABLE 5 pressure, kglcm2 1 loo0 benzene toluene o-xy lene p-cy mene 0OOo 1.765 O-OOO 1.796 OOOO 1,767 0000 I.800 log '1 { 301 log 1 { 301 170 75 c 90 75 c log 90 L{ 75 301 c log L{ 170 75 30: c 0347 0.08 1 0274 0-065 0311 0.057 0.333 0.087 The figures for the viscosity of toluene are given in more polated values at 2000 and 3000 kg/cm2 in table 6 together with values of k3.TABLE 6 pressure k3 rl 4 5 kglcm* 1 6.65 x 107 000525 00725 1OOO 3.25 ,, 000990 00995 2000 3-05 ,, 0.0 1 89 0.138 3000 2-54 ,, 0,0302 0174 4Ooo - 0498 (3000) 0897 0-597 - 0.689 1.194 0.749 detail with inter- the corresponding k;2/; 0482 0323 0.42 1 0442 It is thus seen that k3d; is approximately constant and the termination constant is thus apparently inversely proportional to the square root of the viscosity. It is of interest to study the parts played by the propagation and termination constants in the effect of pressure on the overall rate of polymerization at 30" C and to compare this effect with that found by Merrett and Norrish 1 at 60" C.The rate of polymerization R is given by R = (k~/k3*)(ki f(l)C)JM% . Since the rate of initiation is not greatly affected by pressure the rate of polymer- ization will increase rapidly up to 1000 kg/cm2 due to the combined effects ofA . E. NICHOLSON A N D R . G. W. NORRISH 113 pressure on k2 and k3. Above 10oO kg/cm2 the rate will increase exponentially but more slowly, because k2 is now the controlling factor, k3 decreasing only slightly. Merrett and Norrish found that the effect of pressure on the rate of polymerization at 60" C was similar, there being an initial rapid increase UP to 2000 kg/cm2 after which the rate continued to increase exponentially but not so rapidly.From this we may infer that at 60" C the initial rapid decrease in the termination constant continues up to 2000 kgIcm2. As derived from table 5, the viscosity at loo0 kg/cm2 and 30" C is approximately equal to the viscosity at 60" C and 2000 kg/cm2, and it would seem that the levelling off of the curve of log [(k3)p/(k3)1] against pressure is controlled by the viscosity of the medium. In fig. 3 it is clear that the lines pass through the origin within the limits of accuracy of the results. The value of k4/k2, which would be equal to the intercept, is thus too small to determine from our results, involving as it does a long extra- polation. In addition, any increase in k4 with pressure would be masked by the increase in k2.It is not possible to say, therefore, from these results whether pressure has any effect upon the transfer constant k4. The transfer reaction is, however, a bimolecular reaction and would be expected to increase with pressure. If this is the case then, since and while k3 is decreasing with pressure k4 is rapidly increasing, then in the limit = 2k2n/(k3n2 -I- 2k4nM), and if k2 and k4 increase approximately to the same degree as is to be expected, the molecular weight will tend to a constant value. This constant value, however, will be dependent in a secondary way on catalyst concentration as found and ex- plained by Merrett and Norrish. We conclude that the effect of pressure on the rate and degree of polymerization of styrene can be explained in terms of the observed increase in the propagation constant, together with the more complex effect of pressure on the termination constant. One of us (A. E. N.) wishes to thank the Distillers Company for a maintenance grant, during the tenure of which this work was carried out. 1 Merrett and Norrish, Proc. Roy. SOC. A, 1951,206, 309. 2 Poulter, Physic. Rew., 1930,35, 297. 3 Burnett and Melville, Proc. Roy. SOC. A, 1947, 189,456. 4 Matheson, Auer, Bevilacqua and Hart, J. Amer. Chem. SOC., 1949,71,497. 5 Mayo, Gregg and Matheson, J. Amer. Chem. SOC., 1951,73,1690. 6 Tait, Report of the voyage of H.M.S. " Challenger ", Phys. and Chem., 1881, 11, 7 Bridgman, Physics of High Pressure (1931), p. 103. 8 Patnode and Scheiber, J. Amer. Chem. SOC., 1939,61,3449. 9 Melville and Valentine, Trans. Faraday Soc., 1950, 46,210. 10 Matheson, Auer, Bevilacqua and Hart, J. Amer. Chem. Soc., 1951,73, 1700. 11 Bamford and Dewar, Proc. Roy. Soc. A, 1947,192,329. 12 Burnett, Trans, Faraday SOC., 1950,46, 772. 13 Majury and Melville, Proc. Roy. SOC. A, 1951, 205,496. 14 Grassie and Melville, Proc. Roy. SOC. A, 1951, 207,285. 15 Dainton and Ivin, Nature, 1948, 162,705. 16 Evans, Trans. Faraday SOC., 1947,43,2277. 17 Gibson, Fawcett and Perrin, Proc. Roy. SOC. A, 1935, 150,223. 18 Perrin and Williams, Proc. Roy. SOC. A, 1937, 159, 162. 19 Norrish and Smith, Nature, 1942, 150, 336. 20 Burnett and Melville, Nature, 1945, 156, 661. 21 Bridgman, Physics of High Pressure, 1931, table XXII, p. 341. 22 Bevington, Melville and Taylor, J . Polymer Sci., 1954, 12, 449. Appendix A.

ISSN:0366-9033    

DOI:10.1039/DF9562200104

出版商:RSC    

年代:1956

数据来源: RSC

摘要:

THE KINETICS OF SOME ORGANIC REACTIONS UNDER PRESSURE BY S. D. HAMANN AND D. R. TEPLITZKY C.S.I.R.O. Division of Industrial Chemistry, High Pressure Laboratory, Sydney University, Australia Received 4th June, 1956 Measurements have been made of the effect of pressure on the rates of addition of iodine to allyl alcohol in water and in nitromethane solutions, and of bromine to stilbene in methanol solution. These reactions proceed through a partially ionic transition state and simple electrostatic theory suggested that they should be accelerated by pressure. They were found to be accelerated by factors of 5-10 at 3000 atm. Some values are also given for the rates of three Menschutkin reactions at pressures up to 15,000atm. The influence of structural changes on the pressure effect is briefly discussed.It is now known that the effect of pressure on an ionic reaction in solution is closely related to the electrical nature of the reaction. Reactions which produce ions are accelerated by pressure, whereas reactions which remove ions are retarded. This rule has a theoretical basis in the fact that pressure increases the electrostatic interaction energy between a charged group and the surrounding solvent mole- cules ; 1 a rise in pressure therefore makes it easier to form ions and more difficult to remove them. The experimental justification for the rule is to be found in the work of Perrin and his colleagues, reviewed by Buchanan and Hamann,l and in the recent measurements of David and Hamann 2 and Burris and Laidler.3 There seems no reason to doubt its general validity.If a reaction involves no net change in the number of ionic charges it is still possible for the electrostatic factor to play a major part in determining its change of rate under pressure. This may happen if the formation of the transition state involves the temporary appearance and separation of ions which subsequently recombine and neutralize is the dark halogenation scheme : 4 slow R\ --c +x-x+ R P- <: their charges. &I example of this type of reaction of an olefin in a polar solvent, which follows the X + fast R\ /R R / \R . . \&<j +x- _3 x-c-c-x (1) k/ '**' -- The transition state is likely to be much more polar than the initial state and we should therefore expect the reaction to be accelerated by pressurn.The present paper describes some measurements which we have made to test this conclusion. They relate to the addition of iodine to allyl alcohol and of bromine to trans- stilbene : (2) (3) The rates of both reactions were measured at 0" C and at pressures up to 3000 atm. The solvents were water and nitromethane for reaction (2) and methanol for reaction (3). 114 CH2 : CH . CH20H + I2 + CH21. CHI. CH2OH ( 3 3 5 . CH : CH . GHs + Br2 + c&5. m r . C m r . c&5.S . D . HAMANN AND D . R . TEPLITZKY 115 We have also measured the acceleration, with pressure, of the rates of three Menschutkin reactions in methanol solution. We selected the following reactions : C6HsN(CH3)2 -I- CH3I -+ CeHsN(CH3)3+ + I- (4) C6HSN(CH3)2 4- C ~ H J B ~ -+ C~HSN(CH~)~(C~H~)+ + Br- (5) C6HSNfCH312 3.C2Hd -+ C~HJN(CH~)~(C~H~)+ 3- 1- (6) in order to see what influence structural changes have on the pressure effect. These measurements were taken to 15,OOO atm. EXPERIMENTAL APPARATus.-The measurements on the halogenations were made in a 3000 atm steel bomb similar to that described previously.1 The pressures were developed by an intensifier and measured on a Bourdon gauge. The reaction mixtures were contained in 1Oml glass tubes, blackened on the outside and sealed by internally ground glass plugs which could slide into the tubes to transmit the pressure. There was seldom any leakage of liquid past these plugs. The Menschutkin reactions were carried out in the 15,000atm apparatus described in part 2.2 MAmRIALs.-The bromine, iodine, sodium bromide and potassium iodide were of analytical quality and were not further purified.B.D.H. nitromethane was distilled from phosphorus pentoxide and the fraction boiling at 101.1" C collected. Methanol was purified by the method of Lund and Bjerrum.5 The trans-stilbene, supplied by Light and Co., melted sharply at the correct melting point and was used without further purifica- tion. Light's allyl alcohol was distilled twice, to give a product which boiled between 96.9 and 97.0" C. The dimethyl aniline, methyl iodide, ethyl bromide and ethyl iodide were B.D.H. specimens, purified by distillation. THE IODINATION OF ALLYL ALCOHOL IN WATER.-Precooled solutions of ally1 alcohol in water, and of iodine in water con- taining 0.05 mole kg-1 of KI, were mixed rapidly and then compressed, within a period of 5 min, to the desired pressure.After a measured time the pressure was released and two aliquots of the mixture were titrated with standard thiosulphate solution to determine the amount of unchanged iodine. This procedure was repeated for a number of different time intervals at each pressure. IN moMErHANE.-The method here was much the same as for the aqueous solutions except that it was unnecessary to have potassium iodide present in order to dissolve the iodine. The existence of two liquid phases during the titration with thiosulphate did not affect the end-point. THE BROMINATION OF t&.?tlS-STILBENE IN METHANOL Bartlett and Tarbell 6 have reported that solutions of bromine in methanol are unstable over long periods. We found this to be so, and for that reason we prepared a fresh bromine solution for each rate determination.The amount of unreacted bromine left at the end of each run was estimated by adding excess thiosulphate and back-titrating with standard iodate solution. MEN- REACI'IONS IN METHANOL These were carried out on a much smaller scale than the halogenations, the total amount of reaction mixture being only 0.2 ml. The extent of reaction was measured by titrating the halogen ions with silver nitrate, using an adsorption indicator. RESULTS AND DISCUSSION HALOGENATIONS The iodination of allyl alcohol in water in the presence of KI is a second-order - dDZl/dt = k2tI21" (7) reaction,7 and the rate of disappearance of iodine is given by116 KINETICS OF ORGANIC REACTIONS where [A] is the concentration of ally1 alcohol and [I21 is the concentration of iodine.Table 1 lists our experimental values of k2, based on the rnulal concentra- tion scale. TABLE THE RATE OF IODINATION OF ALLYL ALCOHOL IN WATER + 0-05 mole kg-1 KI AT 0" c Initial concentrations : [I21 = 0-01 mole kg-1, [A] = 0.10 mole kg-1 P/atm 1 lo00 2000 3000 104 x k2/sec-1 kg mole-1 1.57 2.6 4.1 7.2 It is apparent that the reaction is considerably accelerated by pressure. But this is not an unambiguous demonstration that the slow step in (1) has been accelerated. The reason lies in the fact that it was necessary to have KI present in the water in order to dissolve sufficient iodine. This means that most of the iodine is present in the form of 13- ions, which are much less active iodinating agents than I2 molecules.7 We therefore need to know how the equilibrium I- + I2 + 13- is changed by pressure. Some measurements in this laboratory8 have shown that pressure shifts the equilibrium towards the 13- complex, in other words it reduces the concentration of the active reagent, 12.It follows that the acceleration shown in table 1 would be even greater if the actual concentration of free iodine molecules were used in formula (7). The complicating effect of the tri-iodide equilibrium in water can be avoided by using nitromethane as the solvent, since this dissolves sufficient molecular iodine in the absence of iodide ions. But our experiments suggest that the re- action mechanism is slightly different in nitromethane. This is shown by the fact that the reaction now has a third-order kinetic form : - d[I2]/dt = k3[12]2[A].Similar changes of order from second to third have been observed in other cases where the dielectric constant of the solvent was reduced9 The change probably means that an extra iodine molecule is needed to " solvate " the I- ion iq (l), by forming the 13- complex. If this is so, the reaction remains electrically similar to (1) and should be accelerated by pressure. Table 2 shows that it is. TABLE 2 . T H E RATE OF IODINATION OF AUYL ALCOHOL IN NITROMETHANE AT 0" c Initial concentrations : [I21 = 0.0101 mole kg-1, [A] = 0.0505 mole kg-1. Platm 1 500 1000 2000 2500 10 x k3/sec-1 kg2 mole-2 1.27 2.5 3.9 7-6 11.3 The experiments of Bartlett and Tarbell6 have established that the addition of bromine to stilbene in methanol is not quite as simple as formula (3) suggests.The reaction produces, in addition to the dibromo-derivative, a considerable amount of 1 -bromo-Zmethoxy-1 : 2-diphenyl ethane : CH30H f Br2 + C6H5CH : CH . c6H5 --+ C6H5. CHBr . CHOCH3 . C6H5 + HC + Br- (9) Theoretically this is not a serious complication, because it is almost certain that the rate-determining step for reaction (9) is the same as for the bromination (3); the difference between the two processes lies in the subsequent fast reaction of the carbonium ion with a bromine anion or with a methanol molecule. Re- action (9) does, however, interfere with the primary step because it produces Br- ions, and these tie up some of the free bromine as Br3- ions, which are notS.D. HAMANN AND D. R . TEPLITZKY 117 active in the addition. In consequence the rate falls off very rapidly as the reaction proceeds. This diffculty can be avoided by having a large excess of Br- ions in the initial reaction mixture, but we then need to know the effect of pressure on the equilibrium Br2 + Br- + Br3-. This has not been measured, but it is safe to assume that the tribromide ion will be favoured by high pressures, just as the tri-iodide ion is.8 It follows that any observed acceleration of the reaction with pressure Will be less than the real increase in the rate constant for the attack of Br2 molecules on the double bond. I I I 1 i 000 2000 3 0 0 0 0 P re ssure/a t m FIG. 1.-The rates of some halogenations at high pressures. Curve I : iodine + ally1 alcohol in water at 0" C; curve I1 : iodine + ally1 alcohol in nitromethane at 0" C; curve I11 : bromine + stilbene in methanol at 0" C.The quantity k,/kl is the ratio of the rate constant at P atm to that at 1 atm. The reaction obeys the second order rate formula - dEBr2l/dt = ~21Br21CS1, (10) where [S] denotes the molal concentration of stilbene and @r2] denotes the total molar concentration of bromine. Table 3 lists our experimental values of k2. TABLE 3.-THE RATE OF BROMINATION OF trUlZS-STILBENE IN METHANOL -I- 0.63 mole kg-1 NaBr AT 0" C Initial concentrations : @3r2] = 0004 mole kg-1, [S] = 00038 mole kg-1 Platm 1 1000 2000 3000 102 x k2/sec-1 kg mole-1 4*3* 8.7 16 27 * cf. Bartlett and Tarbell's value 6 of 4.0. The results for the three halogenations are plotted in fig.1. The reactions are accelerated by pressure to about the same extent as Perrin's class of " slow " reactions 10 which form free ions from neutral molecules.1 It could be inferred from this, that the transition state for reaction (1) must be quite highly ionic. MENSCHUTKIN REACTIONS These were straightforward reactions obeying a second-order kinetic law : - d[RXl/dt = k2I-ItBl. (1 1) where [RX] is the molal concentration of alkyl halide and @] is the molal con- centration of amine. Table 4 lists our results for reactions of this type.118 KINETICS OP ORGANIC REACTIONS All these reactions involve the development of ionic charges and their accelera- tion by pressure is understandable on that basis. There are, however, some signi- ficant differences between the behaviour of individual reactions, and these must be due to structural differences in the reactants since the temperature! and solvent were the same throughout.The changes of structure are of two types. TABLE TH THE RATES OF SOME MENSCHUTKIN REACTIONS IN METHANOL AT 25" C Initial concentrations: WX] = [B] = 0.6 mole kg-1 (a) C~HSN(CH~)Z -I- cH3I + C6HsN(CH3)3+ 4- I- P/atm 1 1500 3000 6000 9OOO 12000 15000 105 x k2/sec-1 kg mole-1 4 5 * 20 44 113 303 680 1200 (b) C~HSN(CH~)~ 4- C2HsBr + C~HSN(CH~)Z(C~H~)+ + Br- Platm 1 1500 3000 6000 9000 12000 15000 107 x kzlsec-1 kg mob-1 9.4 60 129 370 1050 3040 4500 (C) CsHsN(CH3)2 -I- C2HsI -+ CSHSN(CH~)~(C~HS)+ + I- P/atm 1 1500 3000 6OOO 9OOO 12000 15000 106 x k2/sec-1 kg mole-1 2.6 11.3 31 86 206 530 1080 * cf. Evans, Watson and Williams' value 11 of 4.42.CHANGE OF HALOGEN AToM.-The replacement of the bromine atom in reaction (b) (table 4) by an iodine atom in reaction (c) reduces the amount of acceleration at high pressures. This effect is also found in Perrin's results 10 for the reactions between pyridine and n-butyl bromide and iodide, and in some measurements of the relative rates of solvolysis of ethyl chIoride,l% bromide,:! and iodide 2 in com- pressed methanol. The data are summarized in table 5. where k,lkl is the ratio of the rate constant at P atm to that at 1 atm. TABLE 5.-REACTIONS OF DIFFERENT ALKYL HALIDES reaction solvent temp. "C k3000/ki klS0oolkl solvolysis of C~HSCI methanol 65 7.9 136 9s C2HsBr 93 65 5.4 60 Y Y C2HSI 9, 65 4.7 19 pyridine + n-CdHgBr acetone 60 8.3 - - 9s + n-C4H9I 9s 60 6.5 dimethyl aniline + C2HsBr methanol 25 13.7 480 Y9 Y 9 + C2Jw ¶, 25 11.9 420 We believe that the trends shown in table 5 can be explained by the electrostatic theory of pressure effects.1 The smaller ions (CI- < Br- < I-) have the greater solvation energies and these in turn are more affected by compression of the medium.CHANGE OF ALKYL GRom.-The change from methyl iodide (a) to ethyl iodide (c) leads to an increase in the effect of pressure on the reaction rate. This is con- sistent with Perrin's observation 10 that the acceleration of Menschutkin reactions by pressure increases with the complexity of the reactants. It can hardly be an electrostatic effect and probably arises from steric factors which might be reflected in the entropy of activation.It certai~3y merits further study. The work described in this paper was carried out as part of the programme of the Division of Industrial Chemistry of the Commonwealth Scientific and Industrial Research Organization, Australia, We should like to thank Dr. A. H. Ewald and Mr. H. G. David for their assistance, and Prof. T. G. Hunter for providing us with accommodation and facilities. One of us (D. R. T.) s indebted to the C.S.I.R.O. and the University of Sydney for the grant o la studentship.S . D. HAMANN AND D. R. TEPLITZKY 119 ADDENDUM H. G. DAVID AND S. D. HAMANN. Since the paper by Hamann and Teplitzky 13 was written, we have succeeded in extending the pressure range of our kinetic measurements to 45,000 atm at room temperature.Our apparatus was similar to one developed by Bridgman 14 except for an important difference in the method used to seal the liquid reaction mixture against leakage. In his compressibility measurements Bridgman 15 used Stillman’s device 16 of enclosing the liquid in a lead capsule and relying on the plasticity of the lead to transmit the pressure applied by a piston. We soon found that the lead capsules have a number of disadvantages in chemical experiments. These are : (i) the lead is by no means inert, chemically : (ii) it takes about 18 h to seal the capsule by cold-soldering 17 and this delay can cause serious errors in kinetic work ; (iii) it is necessary to cut the capsule in order to remove the liquid, which means that a new capsule is needed for every experiment.Carboloy Hard steel Mild steel Tef!on(f?T.F.E.) Po lye t h yle’ne Liquid FIG. 1.-An arrangement for chemical experiments at 45,000 atm. We find that the arrangement shown in fig. 1 has none of these disadvantages. The reaction mixture is contained in a polyethylene tube fitting closely into the pressure vessel and closed by a tapered Teflon plug. A small force on the Carboloy piston pushes the plug into the tube until it forms a seal at the mouth. Thereafter an increase in the force tightens the seal and compresses the polyethylene longitudi- nally. In this way the pressure on the piston is transmitted, with some frictional loss, to the liquid. Extrusion of the container is prevented by mild steel chamfer rings. After an experiment, the tube and plug can be pushed out of the cylinder, the plug extracted and the liquid recovered without damage to the container.The same‘tube can be used repeatedly in spite of the fact that small cracks soon develop in its base. These may leak very:slightly at atmospheric pressure, but they apparently close up when the tube is compressed. It is necessary to know the relation between the actual pressure in the tube and the pressure applied to the Carboloy piston. We established this by observing some phase changes for which Bridgman had measured the transition pressures. Fig. 21 20 KINETICS OF ORGANIC REACTIONS shows that there is a linear relation between the two pressures and that an applied pressure of 56,000 atm is needed to produce an internal pressure of 45,000 atm.This loss of 20 % is notably higher than the corresponding loss of 8 % for a lead capsule at the same pressure. Evidently the elastic deformation of polyethylene requires a greater force than the plastic deformation of lead. Prtsrurt on p i s t o n / t ~ ~ otm FIG. 2.-A calibration-of thezapparatus by some known phase transitions. The letters LIand S denoteithelliquid and solid states, and the Roman numerals specify different solid phases. Using this apparatus, and an experimental procedure similar to that employed at lower pressures,18 we have measured the rate of the first-order solvolysis of 0-5 M allyl bromide in methanol : CH2=CH-CH2Br + CH3OH 3 CH2=CH-CH20CH3 + H -F + Br - (1) at 23°C. At that temperature methanol should freeze at about 29,000 atm but in our experiments it remained liquid, presumably because of the presence of the dissolved allyl bromide, and the tendency of methanol to " super-press." 3 Also, some tests showed that a pressure of 25,000 atm is insufficient to freeze pure allyl bromide at 23"C, so that there is very little chance that it froze out of solution in the kinetic experiments. In all the runs the pressure was developed slowly, to allow time for the heat of compression to dissipate. After decompression the contents of the tube were weighed and titrated with alkali to find the extent of reaction. The results are listed below : TABLE 1.- Pltttlll 1 3,000 5,000 10,500 15,Ooo 20,400 -FIRST-ORDER RATE 2-50 25 55 240 510 850 CONSTANTS FOR REACTION (1) AT 23°C Platm 107 klsec-1 25,000 1,200 30,000 2,300 35,000 3,200 40,000 4,300 45,000 4,900 The large acceleration of the reaction with increasing pressure arises from the fact that the transition state is more polar than the initial state and is therefore favoured by an increase in pressure.19S.D . HAMANN AND D. R . TEPLITZKY 121 The main object of these experiments was to see whether the slope of the plot of log k against P continues to decrease above 20,000 atm, or whether it shows signs of increasing. An upward trend could mean that pressure is beginning to have an effect on the internal structure of the molecules and ions, as distinct from its influence on their electrostatic interaction energies. The results show no sign of an upward trend. 1 Buchanan and Hamann, Trans. Faruday Suc., 1953,49,1425. 2 David and Hamann, Trum. Faraday Suc., 1954,50,1188. 3 Burris and Laidler, Trans. Faraday SOC., 1955,51, 1497. 4 Roberts and Kimball, J. Amer. Chem. Suc., 1937,59,947. 5 Lund and Bjerrum, Ber., 1931,64,210. 6 Bartlett and Tarbell, J. Amer. Chem. Soc., 1936, 58,466. 7 Mosset and Berthoud, J. Chim. Phys., 1936,33,272. 8 Ewald and Hamann, Austrul. J. Chem., 1956, 9, 54. 9 Ingold, Structure and Mechanism in Organic Chemistry (Bell & Sons, London, 10 Perrin, Trans. Furaduy Suc., 1938,34, 144. 11 Evans, Watson and Williams, J. Chem. SOC., 1939, 1345. 12 David, Hamann and Lake, Austral. J. Chem., 1955, 8, 285. 13 Hamann and Teplitzky, this Discussion. 14 Bridgman, Pruc. Amer. Acud., 1940,74,21. 15 Bridgman, Pruc. Amer. Acud., 1941,74, 399. 16 Stillman, Engineering, 1900, 69, 183. 17 Bridgman, Pruc. Amer. Acad., 1949, 77, 129. 18 David and Hamann, Trans. Faruduy SOC., 1954,50, 1188. 19 Buchanan and Hamann, Trans. Furaduy Suc., 1953, 49, 1425. 1953), p. 665.

ISSN:0366-9033    

DOI:10.1039/DF9562200114

出版商:RSC    

年代:1956

数据来源: RSC

摘要:

HIGH PRESSURES AND STERIC HINDRANCE IN LIQUID-PHASE REACTIONS BY K. E. WEALE Dept. of Chemical Engineering, Imperial College of Science and Technology, London, S.W.7 Received 15th June, 1956 The rate constants of the reaction between N : N-dimethyl o-toluidine and methyl iodide in acetone, which is retarded by the ortho-effect, have been determined at 1 atm and 3000 atm ; and equilibrium concentrations of the quaternary iodide at 1 atm have been measured. The effect of pressure on the Arrhenius parameters of the reaction is the same as for unhindered Menschutkin reactions in acetone. The reaction between neopentyl chloride and sodium ethoxide in dry ethanol does not result in a measurable yield at 3000atm. These findings, and other observations on the polymerizability of poly-substituted ethylenes at high pressures, indicate that reduction of steric hindrance by molecular deformation does not take place.The rates of many reactions in liquids are considerably accelerated or retarded by high pressures. It is known that if ionic charges appear or disappear during the rate-determining step of a reaction there is respectively acceleration or re- tardation by pressure.1929 394 This is because of the volume changes associated with the electrostriction or liberation of surrounding molecules during the forma- tion of the transition state. The ionic solvation mechanism is well-established but cannot apply to every type of reaction responsive to pressure, and other effects must sometimes be operative. The acceleration of radical-catalyzed polymer- izations may well be due to a more favourable mutual orientation of neighbouring molecules, as first suggested by Conant and Peterson ; 5 and the increase in rate (9.4 times at 5000 atm, 25.2" C) 4 of the acid-catalyzed cyclization of citral- dehyde to p-menth-1-ene-3 : 8-diol can be ascribed to an increase with pressure of the proportion of molecules having the conformation necessary for ring-closure.A fourth process by which high pressures could affect reaction rates is though molecular deformation. In particular, alteration of bond angles and bending of the extended partial bonds in the transition state might lead to an increase in the rate of a reaction which is sterically hindered. This possibility was mentioned by Perrin and Williams,6 and more recently the acceleration of the hindered Menschutkin reaction between N : N-dimethyl o-toluidine and methyl iodide in methanol, at up to 5000 atm, was thought due to such an effect.7 The Arrhenius parameters, A and E, of this reaction are both decreased at high pressures while in the seven unhindered Menschutkin reactions measured by Perrin and co- workers in acetone 8 there is an increase in both A and E.Subsequent theoretical calculations by Strauss and Hamann 9 on the solvation of large ion-pairs indicated that the reversed behaviour of the parameters of the hindered reaction might be due to the use of methanol as solvent. The rate constants have therefore been measured in acetone solution at 1 atm and 3000 atm. The possibility of effecting the sterically-hindered reaction between neopentyl chloride and sodium ethoxide in dry ethanol at 3000 atm has also been investigated.EXPERIMENTAL The high-pressure equipment has previously been described.7 B.D.H. methyl iodide and N : N-dimethyl o-toluidine were purified by the methods used in the earlier work, and 122K. E. WEALE 123 A.R. acetone was dried by the method of Conant and Kirner.10 Both reactants were initially 0-1 N (uncorrected), and analysis was by titration with 0.01 N silver nitrate (eosin). neo-Pentyl chloride was made from the alcohol and thionyl chloride by the method of Gerrard and Tolcher 11 (10 g alcohol yielded 2.9 g chloride, b.p. 83", n g 1,404. Attempts to prepare the bromide yielded mainly tert.-amyl bromide). Sodium was purified by melting under sulphur-free xylene, and dissolved in dry ethanol.Both reactants were initially 0 1 N (uncorrected), and 0.01 N silver nitrate (dichlorofluorescein) was used for titration. Reactions at ordinary pressure were carried out in sealed glass tubes, and at 3000atm in inverted glass tubes with a mercury seal. The temperatures were constant to within 0.1 ", and the high pressure was maintained to within f 50 atm. RESULTS THE MENSCHUTKIN REACTION at 3000atm was not allowed to proceed beyond 10 % to 12 % reaction, to avoid separation of crystals of the quaternary iodide. Satisfactory second-order constants were calculated from the usual equation. At 1 atm the reaction slowly attains equilibrium and it was necessary to determine the equilibrium concentration vT I lo3 FIG.1.-Menshutkin reaction in acetone at 1 atm (lower line, left-hand scale), and at 3000 atm (upper line, right-hand scale). of iodide, Xe, from runs of long duration (up to 300 h). Satisfactory second-order con- stants for the forward reaction were calculated, assuming a first-order reverse reaction, from the equation, where a is the initial concentration of each reactant, and x that of the iodide after time t. Corrected values of a and Xe at the four temperatures employed, and the values of the equilibrium constant Kare given below. The rate constants at both pressures are recorded in table 2. They are corrected for thermal expansion, and for the i s o t h e d compression of the solvent at 3000 atm (Bridgman 12). TABLE EQUILIBRIA IN THE MENSCHUTKIN REACTION AT 1 ATM to c 39.4 44.0 48.9 56.4 t (mole 1.--1) 0.0972 0.0965 0.0957 00947 Xe (mole 1.-1) 000914 0.00728 0.00589 000417 K 1.177 091 5 0.730 0509124 STERIC HINDRANCE TABLE 2.-RATE CONSTANTS OF THE MENSCHUTKIN REACTION IN ACETONE 39.4 44.0 48.9 56.4 atm ," 106 (l.mole-1 sec-1) 3.37 471 667 11.1 33.3 36.4 39-3 43.7 47.0 500 347 : 48-4 61.2 78.1 3000 atm {E 106 (1. mole-1sec-1) 21.5 f 26.9 The relations between log k and 1/T are shown in fig. 1. The Arrhenius parameters and relative rates are recorded in table 3, together with the corresponding quantities for the same reaction in methanol. TABLE 3.-ARRHENIUS PARAMETERS AND RELATIVE RATES OF THE MENSCHUTKIN REACTION IN ACETONE AND METHANOL acetone methanol P P E (kcal/mole) 1 4-5 1 14-3 1 9.13 20.9 3000 6-29 15.4 3000 8.4 18-6 THE HALIDE SUBSTITUTION was studied by keeping portions of the solution of sodium ethoxide and neopentyl chloride in ethanol at both 1 atm and 3000 atm, and at 91*5", for up to 100 h.No measurable reaction was found at either pressure. DISCUSSION At 3000atm the parameter A of the Menschutkin reaction in acetone is in- creased 60 times, and E is increased by 1.1 kcal. This behaviour is similar to that of unhindered Menschutkin reactions in the same solvent, which show increases in A of up to 11 1 times, and in E of up to 1.4 kcal at 3000 atm.8 In the absence of any abnormal changes, of the kind thought to exist from the measure- ments in methanol, the acceleration of the hindered reaction by pressure must be attributed to the ionic solvation effect.The bimolecular substitution of neupentyl halides is greatly retarded by non- bonded repulsions between the methyl groups and the ions entering and separating. The unreactivity of the chloride at 1 atm and 91.5" is expected from the measure- ments of Dostrovsky and Hughes.13 These give a second-order constant for the bromide of about 6 x 10-7 under the conditions of the present experiments (compared with about 6 x 10-2 for ethyl bromide), and the rate constant for the chloride is likely to be about 50 times smaller. However, a measurable reaction would have been produced in 100 h by an increase in rate of 10 to 100 times at 3000 atm. Any acceleration is certainly smaller than this, and there is evidently no large reduction of steric hindrance at 3000 atm. Other work relating to hindered reactions at high pressures is concerned with the polymerizability of various poly-substituted ethylenes.Among such sub- stances, isu-amylene, (CH&C: CHCH3, has been found not to polymerize at 5000 or loo00 atm,l4 while u : p-dimethyl styrene, (c(jHS)(c&) . C : CH . CH3, at 5000 atm,14 and methyl u-tert.-butyl acrylate CH2 : C . (t.-Bu) . CO . om3, at 5000 and loo00 atm,ls yield only dimer. The absence of higher polymers is con- sidered due to steric hindrance to chain propagation (but not to head-to-head dimerization) which persists at the highest pressures employed. The evidence discussed does not support the view that molecular deformation is a factor in determining reaction rates in liquids at these pressures. PerhapsK. E. WEALE 125 an effect would be revealed by more detailed and extensive measurements, or else considerably higher pressures may be necessary. There is not sufficient evidence about molecular deformation in pure liquids at high pressures, e.g. from dielectric constant measurements, to allow an estimation of the second possibility. 1 Buchanan and Hamann, Trans. Faraduy SOC., 1953,49, 1425. 2 David and Hamann, Trans. Furoday Soc., 1954, 50, 1188. 3 Burris and Laidler, Trans. Furuday Soc., 1955, 51, 1497. 4 Harris and Weale, J. Cheni. SOC., 1956, 953. 5 Conant and Peterson, J. Amer. Chem. Soc., 1932,54,628. 6 Perrin and Williams, Proc. Roy. SOC. A, 1937, 159, 162. 7 Weale, J. Chem. Soc., 1954, 2959. 8 Perrin, Trans. Faraduy SOC., 1938,34, 144. 9 Hamann, personal communication. 10 Conant and Kirner, J. Amer. Chem. SOC., 1924,46,245. 11 Gerrard and Tolcher, J. Chem. SOC., 1954, 3640. 12 Bridgman, Proc. Amer. Acud. Arts Sci., 1913, 49, 1. 13 Dostrovsky and Hughes, J. Chem. SOC., 1946, 157. 14 Sapiro, Linstead and Newitt, J. Chem. SOC., 1937, 1784. 15 Holmes-Walker and Weale, J. Chem. SOC., 1955, 2295.

ISSN:0366-9033    

DOI:10.1039/DF9562200122

出版商:RSC    

年代:1956

数据来源: RSC

摘要:

THE VELOCITY OF ETHYLENE POLYMERIZATION AT HIGH PRESSURES BY R. K. LAIRD, A. G. MORRELL AND L. SEED Imperial Chemical Industries Limited, Research Dept., Alkali Division, Northwich, Cheshire Received 19th June, 1956 The rate of polymerization of ethylene has been measured at pressures from 400 to 2,000 atm and temperatures between 20 and 260" C. Various free-radical producing initiators, y-rays, and purely thermal means have been used to induce the reaction. The measured rates depend on high powers of the ethylene pressure, but the alter- native expressions in terms of the ethylene fugacity are simpler, and are shown to be com- patible with the usually accepted free-radical chain mechanism of vinyl polymerizations. The differences in the kinetics of the various reactions studied are attributed to differences in the mechanism of their initiation. Below 180" C there is kinetic evidence that polymer separates as the reaction proceeds.The implications of the use of fugacities to represent concentration terms in a non- ideal gaseous system are considered in an appendix. Polymerization of ethylene has been studied extensively at low pressures ;Is 2 Pease 3 9 4 found the activation energy to be 35 kcal/mole and the reaction to be bimolecular, whereas Storch 5 noted the importance of traces of oxygen and ob- tained an activation energy of 42 kcal/mole. It is now accepted that the reaction proceeds by a free-radical chain mechanism and many different means of initiation have been used? The effect of pressure was studied by Cramer 6 up to 80 atm and more recently Kooijman and Gijsen 7 worked at 250 atm with a variety of initiators.We present kinetic measurements which were made at pressures up to 2,000 atm under conditions where the products had molecular weights of the order of 30,000. The high-pressure polymerization of ethylene has been investigated in this labor- atory since the discovery of polythene in 1933 *. 9 and, as the commercial manu- facture requires the highest pressures currently used in any large-scale industrial process, the kinetics are of considerable interest. Various free-radical producing initiators, including di-tert.-butyl peroxide (D.T.B.P), azodi-isobutyronitrile (ADIB), and acetoxime, have been used. The uncatalyzed thermal polymerization and that initiated by y-rays have also been investigated, The first polymerizations of ethylene to liquid products by nuclear radiations were those of Mund and Koch 10 followed by Lind, Bardwell and Perry.11 More recently, descriptions of the polymerization of ethylene by y-rays from cobalt 60 have been published by one of the present authors12 and by Lewis, Martin and Anderson.13 EXPERIMENTAL APPmATus.-Preliminary investigations, in which ADIB was used as the initiator, were carried out in simple high-pressure equipment using a cylindrical stainless steel reaction vessel of 80-ml capacity with a rounded bottom, and at the top a Bridgman closure14 with a gas lead.A reciprocating stirrer was used, actuated by external solenoids which enclosed the gas lead into which the stirrer shaft extended.The reaction temperature was controlled by immersion of the vessel in an oil thermostat and the pressure was 126R. K . LAIRD A. G . MORRELL AND L . SEED 127 measured by a manganin resistance gauge connected to the hydraulic oil of the pressure generating system. Polymerization rates were measured at constant pressure from a record of the rate of addition of oil to the pressure generator. Later, apparatus specifically designed for kinetic measurements was used. This is shown in fig. 1 ; it comprised a cylindrical reaction vessel of 80-ml capacity closed at the top by a valve, and at the bottom by mercury through which a flexible drive for a rotary stirrer passed. The stirrer motor was housed in a separate, oil-filled, pressure vessel connected to the reaction vessel by a mercury-filled U-tube.The reaction vessel was heated by an external furnace which was controlled from a platinum resistance thermometer. The reaction temperature was measured by an internal thermo- couple and, under normal reaction conditions, had a maximum varia- tion of f 0.1" C. The pressure was measured by a manganin resistance gauge situated in the oil system and sensitive to 4 atm pressure change at 1500 atm. Rate measurements were derived from the rate of injection of mercury into the vessel which was main- tained at constant pressure. The upper limiting reaction rate which could be measured was set by the condition at which the exotherm- icity of the reaction led to loss of temperature equilibrium within the vessel. The y-ray equipment was a simple 800ml autoclave with a pocket into which the irradiation source could be inserted.The in- ternal temperature, measured by a thermocouple, varied by & 3" C and was 2 to 5" C lower than that of the external furnace. The pres- sure was measured by a Bourdon gauge, and, since there was no I r i L. . g a r i n l e t s t irrer mercury stirrer drive -1 FIG. 1.-The reaction vessel. recompression equipment, the reactions were carried at constant volume. Cobalt 60 sources of 4.2 and 0.45 curies were used. In this relatively crude apparatus the absorption of the incident energy by the ethylene could not be measured; simple calculations indicate that about 10 % was absorbed by the pocket and a rather smaller amount by the compressed ethylene itself. MATERx~.-Erhy~ene: a fractionated gas with total impurities less than 0015 %; a detailed analysis is given in table 1.TABLE ANALYSIS OF E!"HYLENE IMPURITIES (p.p.m.) co2 H2 co saturates unsaturates 0 2 3.6 0.9 60 6.2 probably < 20 < 20 Other materials : commercial grades not specially purified. PROCEDURE.-The initiator, normally in dilute solution in benzene, was introduced into the cold reaction vessel, which was then purged several times with ethylene. In the oxygen-initiated reactions the vessel was evacuated and oxygen admitted, usually as air, followed by a little ethylene. Under a low pressure of ethylene the vessel was raised to128 ETHYLENE POLYMERIZATION the reaction temperature by a standardized procedure; the pressure was then raised with ethylene and the reaction system isolated.Under typical conditions the heating-up period occupied about 1 h and within 10 min of applying the pressure useful rate measure- ments could be obtained. Absolute polymerization rates were calculated from the dilato- metric measurements by use of compressibility data for ethylene 15 and polythene.16 Reactions were quenched by removing the furnace and cooling the vessel with a compressed-air blast. The products were dissolved in hot xylene and reprecipitated in methanol ; they were characterized by their solution Viscosity and infra-red absorption spectrum. In radiation experiments the autoclave was evacuated and purged with ethylene four times, after which excess ethylene was charged at room temperature. The temperature was then raised in a controlled manner and the excess pressure was bled off.After introduction of the source the reaction was followed for a drop in pressure of, at the most, 50 atm, which corresponded to a conversion of less than 5 %. The reaction was stopped by removing the source. Above 100" C, significant yields of polymer were obtained by thermal reaction for which corrections were made by parallel experiments without the introduction of the Pressure (arm) FIG. 2.Effect of pressure on polymerization rate. A, thermal reaction (230" C) ; B, initiator- acetoxime 17 x 10-3 molell. (200" C). radioactive source. Reaction rates were calculated from the weight of polymer produced in a measured reaction time, after subtraction of the induction period. Proportionate corrections to a constant rate of energy dissipation were made for varying ethylene densities and the decay of the sources.RESULTS The results are summarized in table 2 and fig. 2 to 8 which show the variation of the polymerization rates with pressure, temperature, and initiator concentration. The rate expressions are given in the table as functions of the ethylene pressure P and of the ethylene fugacity f. The fugacities were derived, with 1 atm as the standard state, from thermo- dynamic data assembled in this laboratory 17 from the compressibil- ity measurements of Michels and Geldermans.15 These measurements extended only to 150°C and extra- polations to higher temperatures were required. In experiments carried out at temperatures below 140" C, the ob- served rates accelerated continuously as the reaction proceeded up to at least 50 "/, conversion and the rates reported are those measured at a fixed conversion, 10 % for the DTBP-initiated reactions, and 1 % for the ADIB reactions in which pressure was varied.In the experiments where the ADIB concentration was varied, the rate was measured at a fixed time of 1 h after the start of the reaction. Above 180" C the reaction against time curves were linear and the acetoxime-initiated and uncatalyzed reaction rates were measured at conversions between 5 and 30 %. The measured rates of the oxygen-initiated polymerization were very irreproducible and only one series of results, showing the effect of pressure, is reported. Here a slow initial rate was followed by rapid polymerization which extended from about 20 to 70-80 % conversion, when the rate fell away. The results quote the maximum observed rate, at about 50 % conversion, in each experiment.M initiator c.acetoxime DTBP ADIB oxygen y-rays y-rays overall polymerization rate (moles/l. h) 1.81 x lO7exp - c- :;'"I 2-37 x 1011 exp - 1- Y I TABLE 2.-RESULTS 95 % confidence limits of exponents and activation energies. E experimental ranges p4'9 f 2'1 p2.6 CO'4 f 1'1 p2'2 c0'7 f 1'6 P (atm) 1200-2000 1200-2000 380-1 050 400-1200 1100-1800 473-1 185 675- 1 3 75 T (" C) 200-264 180-2 10 120-140 51 230 20-125 100-200 C (moles/l. x 103) - 0.85-3402 1.4-6.8 10.2-6 1 -2 0.45 and 4.2 curies 0.45 and 4.2 curies P orf 4-6-53 1 *9-2.3 2.1-3.0 0.8-163 1.8-2.6 1.4-1'9 1.0- 1.3 0.6-0.9 2-1-33 1.0-1.5 1.4-207 0.9-1 *8 2.6-3.6 1'4-2.1 C - 0.3-0.5 0.7-1.1 - 0.6-102 0*04-1*5 E (callmole) 26,700-34,500 28,800-35,OOO 16,400-4 1,900 - - 2,100-8,800 7,800- 1 8,400130 3 ETHYLENE POLYMERIZATION - I a * t I I 1 I I 1 I 200 4 0 0 600 800 1000 1 2 0 0 14C Pressure (otm) FIG.3.-Effect of pressure on polymerization rate. C, initiator-D.T.B.P. 1-37 x 10-3 mole/]. (130" C) ; D, initiator-A.D.I.B. 20 x 10-3 mole/l. (51" C). L I I Pressure (airl, 1000 1 5 0 0 2 0 0 0 FIG. 4.-Effect of pressure on polymerization rate. Initiator-oxygen 2 x 10 -3 mole/l. (250° C).R . K. LAIRD, A . G . MORRELL AND L . SEED 131 Two distinct expressions are required to describe the reaction rate of the y-initiated reaction, one between 20 and 100" C, and the second between 100 and 200" C.Induction periods were observed varying from 30 h in the slowest reactions, to 5 min in the fastest. I , I I 4 0 0 600 800 DO0 I200 1400 Pressure (arm) FIG. 5.-Effect of pressure on polymerization rate, y-ray initiation. 0.0 - - 1.0 - c u * e 0 0, 01 -2.0- -30- FIG. 6.-EEect of temperature on polymerization rate. A, thermal reaction 1500 atm ; B. initiator-acetoxime 17 X 10-3 mole/l., 1500 atin ; C, initiator-D.T.B.P. 1.37 x 10 -3 molell. 880 atm.132 ETHYLENE POLYMERIZATION I.Or I 0 I I I I I I I 2.0 2-2 2.4 2.6 2.8 3.0 3.2 3-4 3. + FIG. 7.-Effect of temperature on polymerization rate. y-ray initiation : A, 4.2 C source, 1175 atm ; By 0.45 C source, 1175 atm ; C, 4.2 C source, 675 atm; D, 0.45 C source, 675 atm. concentrotion I: lo3 (mole/I.) FIG. 8.-Effect of initiator concentration on polymerization rate.A, initiator-D.T.B.P. 880 atm, 130" C; B, initiator-acetoxime 1500 atm, 200" C; C, initiator-A.D.T.B., 800 atm, 51" C.R . K . LAIRD, A . G. MORRELL A N D L. SEED 133 DISCUSSION POLYMERIZATION RATES Interpretation of the overall rates of polymerization obtained in this work is difficult owing to the non-ideality of the system and consequent lack of know- ledge of the correct “ effective ethylene concentration ” to use. As table 2 shows, the empirical expressions obtained in terms of pressure contain much higher powers than those which are usually found in the more commonly studied free- radical polymerization systems.18 Even higher powers of concentration, as high as 20 in the uncatalyzed thermal reaction, are involved if the ethylene density is used.A reduced density expressed by the reciprocal of the van der Waals (V-b) term yields expressions similar to those involving pressure. However, the use of fugacity gives simpler empirical expressions which can be reconciled with acceptable reaction mechanisms. In the following discussion therefore, fugacity is used to represent the effective ethylene concentration ; the implications of this procedure are considered in the appendix. The mechanism for the uncatalyzed thermal polymerization may be represented by : ki kP kt M + M -+ R R + M + R R + R + P initiation, propagation, termination, where M represents the ethylene, R a free radical of any chain length, and P the polymer product. With the usual assumption of stationary states the overall polymerization rate is given by (1) where CM represents the effective concentration of ethylene.The second power of the fugacity in the empirical expression supports the assumption that, in the uncatalyzed reaction, the primary radical arises from a simple reaction between two ethylene molecules, and that termination is also a second-order process. We define a “ primary ” radical as one from which the propagation reaction pro- ceeds with the velocity constant kp. If, in the so-called “ catalyzed ” polymerization, the primary radical is pro- duced by the first-order decomposition of the initiator, the overall rate expression becomes which is fitted almost exactly by the empirical expression for the acetoxime- initiated reaction. Higher powers of the ethylene concentration would occur if ethylene activated the catalyst or if the catalyst decomposition fragment were not wholly efficient in starting polymerization chains.As the DTBP-initiated rate has a fugacity exponent of 1.6, one or both of these mechanisms may apply. However, the decomposition of DTBP has not been observed to deviate from a first-order law in many different environments.lQ* 20~21 Furthermore, the methyl radicals produced by its decomposition would be expected to be efficient chain initiators, although contrary evidence has been reported by Eltenton,22 and a similar 3/2 exponent of the ethylene concentration was observed by Rice and Sickman 23 in the low pressure polymerization initiated by azomethane. Ex- tremely high apparent efficiencies were found in all the thermal reactions but, because of a very large amount of chain transfer, precise efficiencies were not calculable.Neither was it possible to compare in this respect one initiator with uncatalyzed rate = kpki’C~’]kt’, or kCM2, catalyzed rate = kCMCc*, (2)134 ETHYLENE POLYMERIZATION another which operated at a different temperature. Nevertheless, we are surprised that the DTBP-initiated rate depends on a higher power of the ethylene fugacity than do the acetoxime and ADIB rates. The apparent dependence of the rate of the latter reaction on a power of the ethylene fugacity of less than unity must be attributed to the limitations of the measurements or to the use of fugacities as a measure of ethylene concentration. If the termination reaction were a first-order process rather than a second, the half-power of the initiator concentration in eqn.(2) would be raised to the first. Table 2 shows that termination is then apparently second-order in the acetoxime-initiated reaction, as in the uncatalyzed polymerization, but becomes increasingly first-order at the lower temperatures of the DTBP- and ADD-initiated polymerizations. This change of mechanism may be related to the occurrence of accelerating reactions at the lower temperatures. These, together with similar catalyst exponents, are encountered in vinyl polymerizations where, as the reaction proceeds, polymer separates from the monomer solution.24-28 At low tempera- tures the ethylene + polyethylene system is certainly of this type, but we have, as yet, no precise observations of the phase relations involved.The oxygen-initiated polymerization is a special example of a catalyzed reaction since the catalytic effect of oxygen far outlasts its lifetime, which under our re- action conditions was only a few minutes. In contrast to its catalytic effect at temperatures above 200" C, oxygen has a marked inhibitory effect on the DTBP- initiated reaction at 140" C; this dual role of initiator and inhibitor has been observed in other vinyl systems.18 We would expect the initial step in the high- temperature polymerization to involve at least one ethylene molecule and hence a fugacity exponent of not less than 3/2 in the overall rate expression. The lower power in the empirical expression may be significant and may indicate that at 230" C the reactivity of oxygen towards ethylene is so great, that the rate-determining step in the initiation process is the thermal decomposition of the oxidation products rather than their initial formation.ACTIVATION ENERGIES The overall activation energy of the polymerization is given by the simplified expression : E = Ep + x E ~ - xE*, (3) where the value of x lies between 3 and 1 and depends on the relative amounts of first-order and second-order termination. Ep is usually between 3 and 7 kcal/mole in vinyl polymerizations, and Et is 0 to 5 kcal/moleP The main contribution therefore arises from Ei, and depends on the mechanism of initiation. Consider now initiation by D.T.B.P. ; taking x = 0.7 as given by the exponent of the catalyst concentration in the overall rate expression, (Ep - 0.7Et) = 3 kcal/mole as determined from the yray experiments, and an overall activation energy E of 29 kcal/mole, Ei is found to be approximately 38 kcal/mole.This corresponds closely to the reported values for the thermal decompositions of DTBP,19-20 and while the correspondence should not be overstressed in view of the wide confidence limits for E, the apparent agreement suggests that, if the addition of catalyst fragment to ethylene is rate controlling, it has a similar activation energy. In the acetoxime-initiated and the uncatalyzed polymerizations similar calculations indicate the activation energy of the initiation step to be some 50-55 kcal/mole. This appears to be rather high for the decomposition of the acetoxime molecule, which presumably breaks at the N-0 bond,3*-31 but it is more reasonable for the uncatalyzed initiation, which, at the least, requires the opening of the carbon-carbon double bond and may even involve the production of the methylene radicals.32 The chain-transfer reactions are of the type, CH3.+ RH + R* + CH4,R. K . LAIRD A. G. MORRELL AND L. SEED 135 for which activation energies of about 10 kcal/mole have been reported3 They will, however, contribute little to the overall activation energy since the rate of transfer will be negligible compared with that of the normal propagation reaction. THE Y-RAY-INITIATED POLYMERIZATION In the y-ray-initiated polymerization, if the absorption of energy, and hence the rate of production of initiating radicals, is assumed to be independent of tem- perature, the measured temperature dependence of the polymerization rate should then arise only from the propagation and termination reactions.In the experi- ments below 100" C the activation energy of just over 3 kcallmole corresponds well with the accepted values as may be seen when Ei is put equal to zero in eqn. (3). Below 100" C, therefore, the reaction may be identified with a polymerization process in which active radicals derived from y-ray fission of ethylene add rapidly to ethylene to start polymerization chains whose rate of propagation is almost wholly determined by the effective ethylene concentration. Above 100" C a different mechanism is apparent from fig. 5 and 7, and table 2. Both the activation energy and the exponent of the pressure term are considerably larger above 100" C than below.This conclusion is firm, because, although the confidence limits of the determinations are wide, there is little overlap. Very much faster polymerization rates occurred in these high-temperature experiments and a convenient interpretation of the increased reaction rate is that an additional initiation mechanism had become important. This was presumably the reaction with ethylene of some less active radical derived from the irradiation process. Ei is now calculated to be about 14 kcal/mole and since there is an ethylene concentration term in the initiation rate, the exponent of fugacity in the overall rate expression should be raised by at least one-half, as is seen from table 2 to be the case.Mass-spectrographic evidence 34-36 indicates that hydrogen atoms are very likely products of electron impact on ethylene. There is also evidence that hydrogen atoms are not extraordinarily reactive towards ethylene,37 and, except at high temperatures, molecular hydrogen acts as a retarder in ethylene polymer- ization which suggests that the hydrogen atoms formed by some initial reaction are not wholly efficient in re-starting polymerization chains. As hydrogen atoms appear to be a major product of y-irradiation, it would be convenient to identify them as the entities requiring an energy of activation for reaction with ethylene. A requirement of 14 kcal/mole corresponds to the " high " values given for this reaction,38 although 5 kcal/mole or less is more commonly assumed.The identity of the fragments which are responsible for the initiation of the low-temperature polymerization is even more uncertain. The electron impact data can be recon- ciled with the production of methyl and methylene radicals, but Eltenton's work 22 suggests that these are not demonstrably more reactive than hydrogen atoms. On the other hand, he found vinyl radicals to be extremely reactive, and in y-ray irradiation, but not in electron impact experiments, final electron capture by ethylene might result in production of vinyl radicals. Cognate observations were made by Le Roy and Steacie 39 in the mercury-sensitized photo-decomposition of ethylene, where the products of the primary decomposition produced rapid polymerization at high tmeperatures, but very little at 25" C.CONCLUSION The essential points we wish to present in this paper are, first, that in studies of the high-pressure reactions of ethylene it is necessary to find some means of representation of the ethylene concentration in kinetic expressions; use of fugacities offers a partial solution. Secondly, the free radical polymerization of ethylene shows considerable variation with the reaction by which the primary addition of the first ethylene molecule is effected.136 ETHYLENE POLYMERIZATION APPENDIX THE EFFECT OF PRESSURE ON THE RATE OF POLYMERIZATION OF ETHYLENE In considering the influence of the ethylene pressure on this polymerization system it is necessary to distinguish between the concentration effect and the purely hydrostatic effect, and the approach which has been found to be most convenient is that of the transition state theory.We apply this theory, in the first place to the propagation reaction, R + M $ X --f R', and make the assumption that the reactants are in equilibrium with the activated complex X The f terms refer to fugacities which must be used in a non-ideal system if K is to be a true constant. If, at a constant temperature, the standard state used to define these fugacities is a pressure of 1 atm, K is a pressure equilibrium constant ; it will vary with temperature but will be independent of pressure. The reaction rate is not a function of the fugacity of the activated complex, but its concentration, and the h a 1 rate expression derived by this treatment must therefore contain a term GX, which relates the fugacity of the activated complex to its concentration. The actual expression may be written as propagation rate = O f d R , @x (4 in which K* is a modified constant obtained from K by removal of the vibrational factor associated with the decomposition of the activated complex.If similar treatments are applied to all the component reactions, of for example the uncatalyzed thermal reaction, with the assumption that the fugacity of the propagating radicals remains constant, we obtain Comparison of this expression with (1) of the main paper shows that the overall velocity constant k is given by In this relation the equilibrium constants K* are independent of pressure with the standard state which has been chosen, and therefore the variation of the conventional velocity constant k with pressure is given only by that of the @ terms. The fugacity of any component i of a non-ideal gaseous mixture is given by the generalized expression : 1 RT logfi = log xi + - 1; Gi dP, in which Xj and Gi are the molar fraction and partial molar volume of the ith component respectively. The molar fraction may be expressed in terms of the concentration and the total molar volume Y of the mixture by which we then obtain the expression : In a complex polymerization system it is impossible to determine the partial molar volume of the individual activated complexes or to predict how they will depend on the pressure ; the expression is therefore of limited use, but as the @ terms appear as ratios in the overall rate expression, they may to some extent compensate for one another.The substitution of fugacity for concentration then accounts for the concentration effect of the ethylene pressure in this non-ideal system, but does not take into account the hydrostatic effect which is defined by the variation of the @ terms of the activated complexes. In the consideration of the effect of pressure on reactions carried out in solutions it is usually convenient to use the concentrations or activities of the reacting species.R. K . LAIRD, A . G . MORRELL AND L. SEED 137 The equilibrium constants involved are therefore based on concentration units and vary with the pressure 40 as well as the temperature. The rate expressions in these units con- tain the familiar exponential term, the “ volume of activation,” 41 although here also the effect of pressure on the activity coefficients of the activated complexes should be con- sidered.In gaseous systems where fugacities are employed the volume term does not arise. We acknowledge gratefully the varied assistance given us by other members of this laboratory both in the experimental work itself and also in helpful discussions of the results. 1 Stanley, J. SOC. Chem. Ind., l930,49,349T. 2 Steacie, Atomic and Free-Radical Reactions (Reinhold Publishing Corporation, 3 Pease, J. Amer. Chem. SOC., 1930,52, 1158. 4 Pease, J. Amer. Chem. SOC., 1931, 53, 613. 5 Storch, J. Amer. Chem. SOC., 1934, 56, 374. 6 Cramer, J. Amer. Chem. SOC., 1934, 56, 1234. 7 Kooijman and Ghijsen, Rec.trav. chim., 1947, 66,247. 8 Perrin, Research, 1953, 6, 1 11. 9 Hunter, Chem. and Ind., 1955, 396. 10 Mund and Koch, Bull. SOC. Chim. Belg., 1925,34, 119. 11 Lind, Bardwell and Perry, J. Amer. Chem. SOC., 1926,48, 1563. 12 Symposium on Utilisation of Radiation from Fission Products (A.E.R.E. C/R 1231 13 Lewis, Martin and Anderson, Chem. Eng. Prog., 1954,50,249. 14 Bridgman, The Physics of High Pressures (Bell and Sons, London, 1949), p. 32. 1s Michels and Geldennans, Physica, 1942, 9, 967. 16 Parks and Richards, Trans. Faraduy SOC., 1949,45,203. 17 Dick and Hedley, to be published by the D.S.I.R.42 18 Bawn, The Chemistry of High PoZymers (Butterworths Scientific Publications Ltd., 19 Raley, Rust and Vaughan, J. Amer. Chem. SOC., 1948,70,88 and 1336. 20 Murawski, Roberts and Szwarc, J. Chem. Physics, 1951, 19,698. 21 Brinton and Volman, J. Chem. Physics, 1952,20,25. 22 Eltenton, J. Chem. Physics, 1947, 15,465. 23 Rice and Sickman, J. Amer. Chem. SOC., 1935, 57, 1384. 24 Burnett and Melville, Nature, 1946, 158, 553. 25 Bengough and Norrish, Proc. Roy. SOC. A, 1950,200,301. 26 Thomas and Pellon, J. Polymer Sci., 1954, 13, 329. 27 Bamford and Jenkins, Proc. Roy. SOC. A, 1953,216, 515. 28 Bamford and Jenkins, Proc. Roy. SOC. A, 1955,228,220. 29 Melville and Burnett, J. Polymer Sci., 1954, 13,417. 30 Steacie and Shaw, J. Chem. Physics, 1935, 3, 344. 31 Taylor and Bender, J. Chem. Physics, 1941,9, 761. 32 Egloff, Schad and Lowry, J. Physic. Chem., 1931,35, 1825. 33 ref. (2), p. 500. 34 Kusch, Hustrulid and Tate, Physic. Rev., 1937, 52, 843. 35 Roberts and Johnson, Anal. Chew., 1948,20,690. 36 Field, J. Chem. Physics, 1953, 21, 1506. 37 Melville and Robb, Proc. Roy. SOC. A, 1949, 196, 494. 38 ref. (2), p. 439. 39 LeRoy and Steacie, J. Chem. Physics, 1942, 10, 676, 40 Guggenheim, Trans. Faraday SOC., 1937,33,607. 41 Evans and Polanyi, Trans. Faraday SOC., 1935, 31, 875. 42 Din, Thermodynamic Functions of Gases, vol. 2 (Butterworths Scientific Publications New York, 2nd edn., 1954). M.O.S. Harwell, 1953). London, 1948). Ltd., London, 1956)

ISSN:0366-9033    

DOI:10.1039/DF9562200126

出版商:RSC    

年代:1956

数据来源: RSC

摘要:

LIQUID-PHASE FREE-RADICAL DISSOCIATIONS AT HIGH PRESSURE BY A. H. EWALD Sydney University, Australia Received 27th June, 1956 C.S.T.R.O. Division of Industrial Chemistry, High Pressure Laboratory, The rates of dissociation of 2 : 2'-azo-bis-isobutyronitrile (AZBN) and pentaphenyl- ethane (PPE) have been measured in toluene at pressures up to 10,OOO and 1500 atm respectively and were found to be decreased by hydrostatic pressure. The effect is as predicted by considering only the volume change due to a 10 % extension of the reacting bond in forming the activated state, although when using iodine as a scavenger with AZBN one finds a much larger pressure effect, presumably due to changes in the " cage effect ". The pressure effect on the dissociation equilibrium of nitrogen tetroxide in carbon tetra- chloride solution has been measured at pressures up to 1500 atm and was found to be ten times as great as predicted from the changes in molecular volumes.The effect of pressure on the rates of ionic reactions can largely be explained in terms of the changes in the energy of solvation brought about by pressure.l* %39 4 The rates of free-radical dissociations are known to be little influenced by the nature of the reaction medium and it is therefore unlikely that solvation energies are major factors in the energy changes of such reactions. One would thus expect the effect of pressure on these reactions to be much smaller than for ionic reactions. In the present investigation the effect of pressure on three free-radical dis- sociations has been observed.The rate of dissociation of 2 : 2'-azo-bis-isobutyro- nitrile (AZBN) in toluene has been measured in two ways; the disappearance of AZBN was followed directly by observing the light absorption at three character- istic wavelengths,s and the formation of free radicals was observed indirectly by using iodine as a scavenger and following its disappearance on a spectro- photometer.6 By the direct method, measurements were made at 1 and 1500 atm at 70" C, while the scavenger method was used to follow the reaction up to 10,000 atm at 62.5" C. The use of iodine in measuring the dissociation of AZBN has been investigated in some detail by Hammond et aZ.6 who found the efficiency of the scavenging reaction to vary in different solvents. It is known that the " cage effect " is very pronounced in this dissociation due to the possibility that the two primary free- radical fragments react to eliminate nitrogen and form the stable tetramethyl- succinodinitrile before they diffuse apart (Hamill et aZ.).7 This effect is specific to AZBN and should not apply to the dissociation of pentaphenylethane (PPE) which was investigated as the second substance.The use of iodine to measure the rate of this type of dissociation was first introduced by Ziegler, Ewald and Orth8 and was widely used by Bachmann.9 Since the formation of triphenylmethyl iodide in this reaction with iodine is reversible these workers found it necessary to introduce pyridine and ethanol to react further with the iodide. In the present investigation the need for this was avoided by using a large excess of PPE and following only the first few percent of the dissociation, in the manner of Bawn and Mellish.10 The dissociation of PPE was measured in toluene at 70" C and at pressures up to 1500 atm.The third system investigated was the equilibrium between nitrogen tetroxide and nitrogen dioxide in carbon tetrachloride solution. This equilibrium was measured at 22" and 51.5" C and at pressures up to 1500 atm. 138A . H . EWALD 139 EXPERIMENTAL All the reactions were followed by means of a Unicam S.P. 500 spectrophotometer which was modified to accommodate a high-pressure absorption cell.11 At pressures up to 1700 atm (up to 3000 atm in one case) the reactions were carried out directly in this cell and could be observed continuously.For reactions above 3000atm the reaction mixture was enclosed in either an all-glass or an all-Teflon reaction tube having at one end a well-fitting plug which could slide into the tube to transmit pressure to the reaction mixture. These cells were immersed in oil in a steel bomb of 1-in. int. diam., of the type described by David and Hamann.2 At the end of a run the reaction cell was removed from the bomb, chilled, and the contents were then analysed in an ordinary stoppered spectrophotometer cell. 2 : 2'-AZO-bk-k'UBUTYRONITRILE This was prepared by the method of Thiele and Heuser 12 and the final product melted at 104-105" C. A.R. toluene was carefully purified by the usual methods and finally fractionally distilled from sodium and stored over sodium.The solutions for the direct measurements were made up by weighing and were approximately 0.6 M. The rate constants were calculated from the observed optical densities I) at 350, 370, 280nip4 by the equation (1) The method of least squares was used to find the best value of k for any one plot of D against time t. The results are quoted in table 1 and are the means of the values found at the three wavelengths. The reactions using iodine as a scavenger were carried out in the manner of Bawn and Mellishlo using a 20- to 50-fold excess of AZBN over iodine. The solutions of AZBN in toluene were made up by weight and were freed of oxygen by bubbling purified nitrogen saturated with toluene through them. These solutions were then mixed with similarly deoxygenated solutions of iodine so that the reaction mixtures were approxim- ately 0.001 m in iodine and 0.057 m in AZBN, except in a few cases when half this con- centration of AZBN was used in order to check the first-order character of the reaction.The reaction was followed by measuring the iodine absorption at 500mp at short intervals. In order to eliminate any optical effects due to the pressure applied to the cell the optical density was also measured at 950 mp since the reaction mixture was known to be quite transparent at this wavelength. These measurements were used as a correction to the readings at 500 mp.* A plot of D against time gave straight lines and the reaction rates were calculated from the slope by the equation lc = - ( l / t ) In (DIDO). Here E' is the molal extinction coefficient of iodine at the temperature and pressure of the particular experiment, 1 is the optical path length of the absorption cell, and concentrations are expressed in moIe/kg. Values of E' were determined in separate experiments using analysed solutions of iodine in toluene.TABLE 1 .-DISSOCIATION OF 2 : 2'-AZO-b~S-~SOBUTYRONITRlLE presssure k x 10s (sec-1) k x lo5 (sec-1) k x lo5 (sec-1) 70" C 62.5" C 62.5" C direct direct scavenger atm 1 5-50 1-87 0-892 1500 4.47 1-52 0552 IN TOLUENE k (scavenger) k (direct) 0-447 0363 Some rate constants obtained by the two methods are shown in table 1. The third column of the table gives the rate constant at 62.5" C calculated from the direct measure- ments at 70" C by using the energyiof activation AE* = 31.1 kcal/mole found by Taliit- Erben and Bywater 5 and confirmed in the present investigation.Further results obtained by the scavenger method are shown in fig. 1. The upper of the two rate constant curves shows the results of a run in which particular care was taken to exclude all oxygen by * This method of correction was used with all the spectrophotometer measurements.140 FREE- R A D I C A L D IS SO C I AT10 N S handling the solutions only in an atmosphere of carbon dioxide. The higher rate con- stants obtained in this run show the inhibiting effect of oxygen on the scavenging reaction between iodine and the free radicals, although there was no actual induction period as was found in the PPE system. The top curve on fig. 1 shows values of loglo(k,lkl) calculated for individual runs.The close agreement in this ratio between differentkruns shows that it is not affected by the presence of a trace of oxygen. FIG. 1.-The dissociation of AZBN in toluene at 62.5" C. PENTAF'HENYLETHANE This substance was prepared by coupling diphenyhnethyl chloride with triphenyl- methyl sodium in the manner of Bachmann and Wiselogle.13 The final product melted with decomposition between 168 and 172" C. The reaction with iodine was carried out in the same manner as for AZBN using solutions 0.001 m in iodine and between 0.015 and 0.025 m in PPE. The reaction, however, seemed to be more sensitive to the presence of oxygen. Although nitrogen was bubbled M i n u t e r FIG. 2.-Optical density at 500 mp of PPE + iodine in toluene at 70" C.through the solutions and these were handled afterwards only in an atmosphere of carbon dioxide, the plots of optical density against time showed an induction period during which iodine disappeared at a slower rate than during the remainder of the reaction. This induction period usually lasted for only a small part of the time during which the reaction was followed. Fig. 2 shows the reaction curves obtained in one series of runs.A . H . EWALD 141 The rate of dissociation of PPE at 70" C was calculated from the slope of the linear portion of a plot of the optical density at 500mp against time, by using eqn. (2). The mean of the rate constants found at 1 atm was 2.5 f 025 x 10-5 sec-1 which is rather lower than the value found by Bachmann 9 in xylene (5.8 x 10-5 sec-1).Although the spread of the rate constants found in different runs was rather great the ratios kp/ki calculated from different runs were in much better agreement. These ratios are shown in fig. 3. c I m FIG. 3.-The dissociation of PPE in toluene at 70" C. NITROGEN TETROXIDE Nitrogen tetroxide was prepared by heating dried lead nitrate to 450" C in a current of dry oxygen. The gas was passed through two U-tubes packed with Raschig rings and phosphorus pentoxide and was condensed in a tube cooled in an ice + salt mixture. It was once more distilled through phosphorus pentoxide before use. The carbon tetrachloride used to make the solutions had to be very pure and absolutely dry in order to obtain reasonably stable solutions. A.R. carbon tetrachloride was there- fore fractionally distilled immediately before use and was exposed to air and light as little as possible.The dissociation constant of nitrogen tetroxide is given by where u is the degree of dissociation and mo the total concentration of nitrogen oxides expressed as moles NOz/kg. Nitrogen dioxide in CC4 has a strong absorption band with a maximum at 435 mp and a pronounced shoulder at 550-555 mp, while nitrogen tetroxide is quite transparent in this spectral region. Unfortunately it was not possible to find the molecuiar absorption coefficient of nitrogen dioxide because of the instability of the very dilute solutions required in order to extrapolate to zero concentration and complete dissociation. The dissociation constant of 12.35 X 10-4 (mold) found by Attwood and Roleffson 14 at 25" C, however, shows that a becomes very small in approximately 0 1 m solutions.If, then, u -g 1 we can write (3) as K = 2uZm0, (3 '1 D = "NO2 amo. (4) D cc mo4, (5) and the optical density of the solution is given by Combining these two equations we find and this relation was found to apply within the range of 0 3 to 1.0 m solution which was used in the equilibrium measurements.142 FREE-R A D I C A L DISSOCIATIONS From eqn. (3’) and (4) we also find that K1/Kp = D12/Dp2 (6) (7) and the change of the free energy of dissociation with pressure is then given by AGp - AG1 =RT In (Kl/Kp) = RT In (D12/Dp2). Since the optical densities are proportional to the volume concentration while the equilibrium constants are expressed in mold units it is necessary to correct the densities by dividing them by the appropriate relative volume of the carbon tetrachloride solutions.These values were obtained from Bridgman’s measurements.1s The solutions were made by weighing to between 0.1 and 0.9 m and were transferred through syphons operated by dry air. The equilibrium measurements were made at 435 mp except in the most concentrated solutions when the means of measurements at 550 and 555 mp were used. The ratios of the equilibrium constants and the changes in the free energy of dis- sociation are shown in table 2. TABLE 2 N204 + 2NO2 in CC14 KllKp AGp - AG1 (cal/mole) temp. ’ C 750 atm 1500 atm 750 atm 1500 atm 23-2 2.122 3-61 1 446 763 21.5 2.054 3.697 425 77 1 22.5 4025 825 23.1 3.775 788 mean 22 208 3.77 435 787 51 248 435 588 953 52 2.13 3.77 42 1 863 mean 51.5 2-30 4.06 540 908 DISCUSSION Although it has been shown that the Stearn-Eyring 16 theory does not correctly describe the effect of pressure on ionic reactions1 it could be expected to apply more nearly to the homolytic bond fissions occurring in free-radical dissociations. In the dissociation of AZBN the first bond broken is a C-N bond.5 If we assume a 10 % stretching of this bond and a constant cross-section of the molecule in passing from the initial to the activated state, we can calculate from a crude molecular model, using covalent radii for the bond lengths and van der Waals radii for the atoms, that the volume of activation should be AV* = 4.3 cm3/mole.From the equation A P d . RT -- _ - - d i n k and from the values of the rate constants shown in table 1 we find from the direct measurements AV* = 3.8 cm3/mole and from the scavenger measurements If we make a similar rough calculation of the volume of activation of the dis- sociation of PPE we fmd a value of AV* = 10.5 cm3/mole, while the measured values of the rate of dissociation at 1 and 1000 atm give a value of AV* = 13.1 cm3/mole.It is thus apparent that even in this crude form the assumption that the effect of pressure is purely a volume effect and that in the activated state there is a 10 % extension of the reacting bond gives the right sort of answer for these reactions. The much larger effect of pressure on the rate of dissociation of AZBN found by the iodine scavenger method must be attributed to changes in the “ cage effect ” Av* = 9-36 ~1113/mole.A .H . EWALD 143 due to pressure. The ratio of these rates to the rates found by the direct method (which are independent of the fate of the free radicals after dissociation) are shown in table 1 and are an indication of the efficiency of AZBN as a source of available free radicals. A much more stringent test of the assumption that the effect of pressure on free- radical dissociations is solely due to the volume changes involved in breaking a bond, should be provided by the measurements of the dissociation constant of nitrogen tetroxide. Not only should the result of the calculation here be inde- pendent of any assumptions regarding the activated state, but there are also much more accurate data available on which to base a theoretical calculation of the volume change.By using X-ray diffraction data on N2O4 17 and electron diffraction data on NO2 18 for the bond lengths and van der Waals radii for the atoms one finds that there is a volume increase of 2.1 cm3 when one mole of nitrogen tetroxide dissociates. From the data of table 2 one can on the other hand derive that the pressure effect on the dissociation constant is equivalent to a volume increase of about 23 cm3/mole. In this case therefore the assumptions made about the effect of pressure on the dissociation are quite inadequate. It is concluded that there is some interaction between the solvent and the nitrogen oxides which is changed by the application of pressure. The retardation of free-radical dissociations by pressure found in these experi- ments is of the kind one would expect from general considerations but which so far has never been demonstrated experimentally.Merrett and Norrish 19 refer to some measurements of the rate of dissociation of benzoyl peroxide under pressure, but these appear never to have been published. It follows from the present results that the large acceleration of polymerization reactions by pressure, which has been observed,Ig is not due to an increase in the rate of initiation but must be due to some other step in the polymerization. It is a pleasure to acknowledge the help and stimulus derived from many dis- cussions with Dr. S . D. Hamann and the help of I’vlr. H. G. David in running the high-pressure equipment. The work described in this paper was carried out as part of the research pro- gramme of the Division of Industrial Chemistry of the Commonwealth Scientific and Industrial Research Organization, Australia. 1 Buchanan and Hamann, Trans. Faraday SOC., 1953,49,1425. 2 David and Hamann, Trans. Faruahy SOC., 1954,50, 1188. 3 Hamann and Strauss, Trans. Faraday Soc., 1955,51,1684. 4 Burris and Laidler, Trans. Faraday SOC., 1955,51, 1497. 5 TalSt-Erben and Bywater, J. Ainer. Chem. Soc., 1955, 77, 3710, 3712. 6 Hammond, Sen and Boozer, J. Amer. Chem. SOC., 1955,77,3244. 7 Roy, Nash, Williams and Hamill, J. Amer. Chem. Soc., 1956, 78, 519. 8 Ziegler, Ewald and Orth, Annalen, 1930, 479, 277. 9Bachmann and Osborn, J. Org. Chem., 1940,5,29. 10 Bawn and Mellish, Tram. Faraday SOC., 1951,47, 1216. 11 Ewald and Hamann, Austral. J. Chem., 1956, 9, 54. 12 Thiele and Hewer, Annalen, 1896, 290, 1. 13 Bachmann and Wiselogle, J. Org. Chem., 1936, 1, 354. 14 Attwood and Roleffson, J. Chem. Physics, 1941, 9, 506. 15 Bridgman, Proc. Amer. Acad. Arts Sci., 1930, 66, 213. 16 Glasstone, Laidler and Eyring, Theory of Rate Processes (McGraw-Hill, New York, 17 BroadIey and Robertson, Nature, 1949,164,915. 18 Maxwell and Mosley, J. Chem. Physics, 1940. 8, 738. 19 Merrett and Norrish, Pvoc. Roy. Suc. A , 1951,206, 309. 1941), p. 470.

ISSN:0366-9033    

DOI:10.1039/DF9562200138

出版商:RSC    

年代:1956

数据来源: RSC