摘要:

0. BEECK, W. A. COLE AND A. WHEELER 32 1 THE MEASUREMENT OF THE THERMAL PROPERTIES OF GASES AND VAPOURS ADSORBED ON SOLID SURFACES BY J. A. MORRISON AND J. M. Los Received I 6th January, I 950 A description is given of a low-temperature adiabatic calorimeter suitable for the measurement of thermal properties of gases and vapours adsorbed on solid surfaces. The calorimeter was designed particularly for the determination of the heat capacity and the heat of adsorption of adsorbed films with rather high precision. An interest in the measurement of the thermal properties of adsorbed films arises from a consideration of the utility of a knowledge of the thermo- dynamic properties in evaluating theories of adsorption. In addition, direct information may be obtained concerning the state of the adsorbed film, and the magnitude of the interaction between the elementary particles of the adsorbate and the adsorbent may be construed. Recently, Hill has given an extensive treatment of the thermodynamics of ad- sorption.Whilc some of the quantities used in this treatment are not ordinarily measured calorimetrically, for example, one usually measures C, rather than C,, by appropriate rearrangement of the relationships one can arrive at other thermodynamic quantities from the experimentally determined values. Of some interest is the entropy, from a knowledge of which may come information concerning the distribution of energy of adsorption sites.2 For the heat capacity measurements the assumption is made that the thermal properties of the adsorbent are unchanged by the adsorption process.While there is no experimental evidence concerning the point, it seems likely that the assumption is valid for physical adsorption pro- vided that the specific surface of the solid is not too large, but it may be seriously in error for chemisorption. To be used in testing theories of adsorption, the measurements should be performed with as high accuracy as is possible. In the heat capacity measurements, where the heat capacity of the film is obtained as a difference, the requisite precision is only obtained at quite low tem- peratures where the contribution of the heat capacity of the film to the total is a reasonable fraction. While this factor necessarily makes the experimentation somewhat more difficult, it leads to the use of simple Hill, J.Chem. Physics, 1949, 17, 520. Hill, ibid., 1949, 17, 762.322 THERMAL PROPERTIES O F GASES AND VAPOURS gases such as argon and nitrogen as adsorbates, which is desirable from the viewpoint of interpretation of results. It would be equally desirable t o use a very elementary solid as adsorbent, but here one is limited by requiring a total surface such that the effects being measured are sufficiently large. So far it has not been possible to obtain simple solids with a sufficiently high specific surface and recourse is had to titanium dioxide (rutile) of reasonably high chemical purity. In what follows is described the calorimeter assembly which is being used to investigate the adsorption of argon on titanium dioxide. Experimental A number of adiabatic low-temperature calorimeters have been described in the literature.While in principle all are similar, in detail there arc rather important differences which depend upon the specific purpose for which the calorimeter is to be used. The measurement of the heat capacity and the heat of adsorption of vapours adsorbed on solids places some special requirements upon the calorimetric apparatus, and in particular upon the calorimeter vessel itself. The adiabatic design is certainly the most suitable since the adsorption processes taking place in the calorimeter vessel may require a rather long time t o reach equilibrium due to unfavourable heat conduction between the par- ticles of the solid or to slow changes taking place on the surface of the solid.With the adiabatic type of low-temperature calorimeter, properly designed, it is possible to determine total heat capacities with a precision of 0-1 yo or better even if a half hour or longer is required for equilibrium to be reached. The requirement of outgassing the solid adsorbent in place in the calorimeter vessel at an elevated temperature makes it necessary that the vessel be easily accessible and that surrounding components be detachable. This is not ordinarily neces- sary in calorimetry where condensable substances are studied. In a previous investigation Morrison and Szasz described a calorimeter which was used to measure the heat capacity of adsorbed nitrogen. Several features of that design have been incorporated in the present one particularly in the design of the calorimeter vessel. However, the present design embodies sufficient departures to warrant an adequate description.Since the amount of liquid hydrogen available for the work was small, advantage was taken of a cryostat design used by Yost et al.4 Cryostat.-The cryostat assembly and details of the calorimeter vessel and associated parts are illustrated in the sectional drawings in Fig. I . The main assembly is hung within the Dewar flask 2 , in which liquid nitrogen is kept normally. The flask rests inside the outer case I, the lower part of w-hich is detachable from a flanged joint. The upper part of the case is held in a frame a sufficient distance from the floor to enable removal of the lower section without disconnection of electrical leads and vacuum lines.Not shown in the drawing are tubes set into the brass top plate which permit the introduction of the liquid nitrogen to the Dewar flask and the connection of the case to a high-capacity Kinney oil pump used for reducing the temperature below 77" K by pumping on the liquid nitrogen. Liquid hydrogen is introduced through the syphon 5 into the container 4 which is supported inside a vacuum space enclosed by the can 3. Gaseous hydrogen which is evolved is led through tube 7 to a vent system. Connection is also possible between this tube and the Kinney pump when it is desired to reduce the temperature to that of solid hydrogen. The heat leak into the hydrogen container is reduced by making all connections to it with thin- walled stainless steel tubing.Adiabatic Shield and Calorimeter Vessel.-The adiabatic shield 15 which is made of copper sufficiently thick to have a uniform temperature, is of the form commonly used in low-temperature calorimeters. The top and bottom are detachable from the cylindrical side a t flanged joints. The shield and the calorimeter vessel 17 are hung by Nylon cords within a vacuum space enclosed by the container 10, which is fastened to the hydrogen container with low- melting solder. Evacuation of the space is accomplished with a high capacity vacuum line connected to tube 6. A similar vacuum line is connected t o the space surrounding the hydrogen container through a tube not shown in the Morrison and Szasz, J . Ckem. Physics, 1948, 16, 280. 4 Yost, Garner, Osborne, Rubin and Russel, J .Amer. Chem. SOC., 1941, 63, 3488.J. A. MORRISON AND J. M. LOS 323 I / / / / I / / / / , / / / , /,///,,,////,,//,//// c 6" FIG. I .-Low-temperature adiabatic calorimeter assembly. I . Monel metal case. 2. Silvered Pyrex Dewar flask, 4% in. int. diam. x 24 in. inside depth. 3. Brass container (3/64 in. wall thickness). 4. Brass liquid hydrogen container, 680 ~ m . ~ volume. 5. Vacuum- jacketed metal syphon. 6. Vacuum line connection. 7. Hydrogen vent line. 9. Brass anchoring ring a t liquid nitrogen temperature. 10. Copper container (I /32 in. wall thick- ness). 12. Anchoring rings at tem- perature of hydrogen container-inner ring & in. thick brass, outer ring 1/32 in. thick copper. 13. Copper nickel filling tube (2 mm. ext. diam. and I mm. int.diam.). 14. Brass tapered joint. 15. Adiabatic shield, 1/16 in. thick copper. 16. Vapour pressure thermometer bulb, 7 cm.3 volume. 17. Calorimeter vessel, 1/16 in. thick aluminum, go ~ r n . ~ volume. 18. Platinum resistance thermometer. 8. Vacuum seal for lead wires. 11, Brass clamps for lead wires.324 THERMAL PROPERTIES OF GASES AND VAPOURS drawing. Cooling of the assembly is accomplished by admitting a small pressure of helium, I or z mm. of mercury, to these spaces. When liquid hydrogen is used the parts within the container 3 are first cooled to the temperature of solid nitrogen. The helium is then removed from the space between containers 3 and 10 and the liquid hydrogen admitted to the container 4. When the calori- meter reaches the desired temperature the helium is removed from the space inside container 10.With the vacuum lines used the pressures of helium can be reduced to less than I x I O - ~ mm. of mercury within 3 min. The cylindrical calorimeter vessel is made of pure aluminum, and is sealed by electric welding of the rim of the dished lid. The powdered solid adsorbent is packed not too tightly between thin perforated aluminum discs which make a sliding fit with the cylindrical wall. The dished top construction permits changing of the solid adsorbent in that the vessel may be opened by cutting off the welded seam in a lathe and may be resealed subsequently by welding. The resistance thermometer 18 enclosed in a platinum thimble ( 5 mm. diam., 70 mm. long and 0-2 mm. wall thickness) is cast into a brass tube with Woods metal.The calorimeter heater of glass-covered manganin wire is wound on the outer surface of the platinum thimble near the upper end and is thus held in the Woods metal. The outer surface of the brass tube is accurately machined with a IO taper which fits closely a similar taper on the inner surface of the re-entrant well of the calorimeter vessel. A threaded piece a t the lower end of the ther- mometer assembly engages a ring on the brass tube in such a way that i t will either tighten or loosen the tapered connection. This design has the distinct advantage of providing a metallic conducting path between the thermometer and heater and the calorimeter vessel. The leads from the thermometer and from the heater are wrapped around a cylindrical aluminum shield surrounding the glass cap of the thermometer to ensure their departure a t the temperature of the calorimeter.The calibration of the resistance thermometer may be checked periodically by direct comparison of the thermometer with a vapour pressure thermometer. The bulb of the vapour pressure thermometer 16 is attached to the adiabatic shield, and into it may be condensed pure oxygen or pure hydrogen. Electrical Wiring.-The electrical leads enter the cryostat through a vacuum seal 8. Forty-two leads of 30 AWG silk-covered copper wires rest in shallow grooves cut longitudinally in a brass male plug which fits a female piece of similar taper, and the small space remaining around the wires is filled with Apiezon wax. Heavier leads from the measuring system which are soldered to the leads coming out of the cryostat are brought in through ten brass clamping pieces 11.The assembly was first used for the intercomparison of resistance thermometers where twelve extra leads were required, so for calorimetry only thirty leads are used. The wires are wrapped twice around the brass ring g and held in place with Glyptal lacquer, which step serves to bring the wires to the temperature of liquid nitrogen and hence to minimize the loss of liquid hydrogen due to heat leaking along the leads. The loss could be further re- duced by using finer wires but in the present instance where considerable manipu- lation of the lower part of the wire assembly w-as necessary the use of sturdy leads seemed desirable. Contact of the wires with the hydrogen container is made by wrapping the wires individually on two rings 12 attached to the bottom of the hydrogen container.The assembly was first tried with just the inner ring but this did not provide a sufficient path to reduce the temperature of the wires and the second ring was added. From the rings 12, the wires are carried in a series of loops to the adiabatic shield. The thermocouple, thermometer and calorimeter heater leads going to the inside of the shield are wrapped in spiral grooves around the cylindrical side of the shield starting near the top and enter through hard rubber insulators near the bottom. The outer surface of the three components of the shield, the top, bottom and side, are covered with electrical heaters close wound and held in place with Glyptal lacquer.Elec- trical heaters are also wound on the two tubes leading to the calorimeter vessel and to the vapour pressure thermometer bulb, and on the bundle of lead wires just before it reaches the top part of the side of the shield. The positions of the difference thermocouple junctions are indicated by arrows on the drawing. Single junction difference thermocouples of Chromel P and constantan extend between the side of the adiabatic shield and five other parts, the two capillary tubes, the bundle of leads, the top and the bottom of the shield. Between the side of the shield and the calorimeter vessel is placed a three junction Chromel P constantan couple. A similar three junction couple is placed between the junction of the tube with the shield 14, and the point whereJ. A.MORRISON AND J. M. LOS 325 the tube is attached to the calorimeter vessel. These latter two are the most important of the difference thermocouples and hence are made of greater sensitivity . Filling Tube Assembly.-Gases or vapours are admitted to the calori- meter through the capillary tube 13. Control of the temperature of the tube is attained by balancing the heater wound on the tube against a thermal shunt (5 cm. of 18 AWG copper wire) fastened between the tube and the inner of the two rings 12. Connection of the shunt to the tube is made above the upper loop shown in the drawing and in such a way as to avoid having a cold spot on the tube. The shunt is placed under the winding of the tube heater and insulated from the tube with paper for a distance of 2 cm.before being soldered to the tube. A similar shunt is attached to the tube leading to the vapour pressure thermometer bulb 16. However, for calorimetric measurements this latter tube is left disconnected. Thermal contact of the calorimeter tube with the adiabatic shield is made with a tapered connection 14, which is tightened with a screw device. This joint may be undone to permit raising of the top part of the shield when a furnace is placed around the calorimeter vessel for degassing the adsorbent. Since much better control of the temperature of the assembly may be exercised by heating rather than cooling it is desirable to measure heats of adsorption rather than heats of desorption. In the measurement of the heat of adsorption it is necessary that the gas be at the same temperature as the calorimeter vessel before adsorption.An estimation of the heat exchange between the gas and the tube indicated that sufficient exchange was obtained in 15 cm. of the tube. Such a lengrh is provided in a loop just above the adiabatic shield. Operation.- Adiabatic conditions within the assembly are maintained by manual adjustment of the energy admitted to the electrical heaters on the three components of the adiabatic shield, on the two tubes and on the bundle of leads. The control of the energy put through the various components is made in such a way that the indications of the array of difference thermocouples are kept a t zero. The potentials of the thermocouples are read from two galvanometers of such sensitivity that for the triple junction couples temperature differences of o.002~ at 15" K or 0~0005" at 90" K can be detected.With maintenance of the indications of the thermocouples as near t o zero as possible, the temperature of the calorimeter vessel may be held constant within 0-001" for lengthy periods a t all temperatures 0.5' or more above the temperature of the hydrogen container. The resistance of the thermometer and the energy admitted to the calori- meter heater are determined with a White double potentiometer. Standard resistors used in the circuits are checked periodically against standards calibrated at the National Bureau of Standards. The establishment of a temperature scale for a series of resistance thermometers, one of which is used for the calori- metry, will be described elsewhere. The time of energy input to the calorimeter heater is determined with a synchronous clock run on a controlled frequency power supply and operated with a switch coupled directly to the heater switch.Calibration of the clock is made by direct comparison with time signals from the Dominion Observatory. Liquid hydrogen is made from compressed hydrogen in cylinders with a liquefier of the type used by Andrews A single filling of the liquid hydrogen container in Ehe cryostat is sufficient for 5 hr. of experi- mentation. For most measurements two operators are required, one of whom adjusts the adiabatic controls continuously. When pressure measurements are also required, a third operator is necessary. at a rate of about I l./hr.The authors are indebted to Mr. G. Clement of the Central Workshops who made most of the component parts of the assembly. Thanks are due to Mr. R. Reid and Mr. G. Ensell of the Laboratories for technical assistance. Division of Chemistry, NatioHal Research Laboratories, Ottawa, Canada. 5 DeSorbo, Milton and Andrews, Chem. Rev., 1946, 39, 403.326 VIBRATING CELLS FOR CONDENSER METHOD ADDENDUM.-The calorimeter has been designed for the purpose of examining systems involving physical adsorption only, but in its present form it could be used €or some particular measurements of heats of chemi- sorption. The modification of the assembly €or other experiments in- volving chemisorption would require, principally, a1 teration of the design of the calorimeter vessel.The operation of the calorimeter has been tested by a rather extended series of measurements of the heat capacity of the calorimeter vessel plus the solid adsorbent only', the heat capacity ranging from 0.6 joules/deg. to 60 joules/deg, The average precision as determined by the deviation of some 40 experimental heat capacity values from a smoothed C, against T relationship for the temperature range 12" to 92" K was found to be 0.04 yo. Below 30' K the precision falls to about 0.1 yo due to lowered sensitivity of. the resistance thermometer. It seems reasonable to Fxpect that for a limited temperature range above 30' K a precision of 0-01 yo could be achieved, which compares favourably with any other form of precise calorimetry. In most of the heat capacity measurements a small correction, 0.02 yo on the average, was made for heat exchange between the calorimeter vessel and the adiabatic shield. About 98 yo of this heat exchange takes place across the section of the filling tube between the calorimeter vessel and the shield, and the correction is cal- culated from the indications of the sensitive difference thermocouple placed across this section. Some measurements have been made of the heat capacity of adsorbed argon at a concentration corresponding to a coverage of the surface of about 0.4. The absolute increment in the heat capacity due to the ad- sorbed argon varies from 0-25 joules/deg. at 14" to 0.7 joules/deg. at 75" K, and the precision of the calorimeter is such as to permit the measurement of the increment to approximately I yo of its value. Only a limited number of measurements of heats of adsorption have been made at low coverages of the surface but the values so far obtained appear to be internally consistent to better than 0.5 yo. The results of these thermal measurements and their significance will be discussed elsewhere.

ISSN:0366-9033    

DOI:10.1039/DF9500800321

出版商:RSC    

年代:1950

数据来源: RSC

摘要:

326 VIBRATING CELLS FOR CONDENSER METHOD STUDIES IN CONTACT POTENTIALS 11. VIBRATING CELLS FOR THE VIBRATING CONDENSER METHOD BY J. C. P. MIGNOLET Received 2nd February, 1950 Cells are described for measuring surface potentials of films adsorbed on evaporated metal deposits using the vibrating condenser method of Zisman.1 Essentially they are electrically-driven, hollow tuning forks. Sources of error are pointed out and precautions to be taken, given. The accuracy is i I or z mV on individual contact potential measurements and 10 or 20 mV on surface potential determinations. A number of cells for the vibrating condenser method have been described.', 2p Two of them are suitable for work in vacuum, but Zisman, Rev. Sci. Instr., 1932, 3, 367. Potter, Physic. Rev., 1940, 58, 623.Hoskins, Rev. Sci. Instr., 1945, 16, 343. 3 Frost and Hurka, J . Amer. Chem. Soc., 1940, 62, 3335. Rosenfeld andJ. C. P. MIGNOLET 327 neither is adapted for work with metal deposits prepared in situ by thermal evaporation. The Zisman cell can certainly be developed in t h a t direc- tion. However, the writer had preferred t o s t a r t from a new type of basic cell. Experimental The Hollow Vibrating Fork Cell.-This cell is essentially a hollow tuning fork. One of the condenser plates is fixed to the inner wall of one limb and vibrates with it. The immobile plate is fixed to the top of the cell (Fig. I). The second limb, bearing a piece of iron sheet, serves for the electrical excitation of the vibration. The fork is made asymmetrical t o allow easy introduction and removal of the non-vibrating electrode.A T-shaped cell would do as well. A very rigid fixation of the fork tail is necessary. This is achieved by a series of 5 side tubes sealed with Picein in an iron block of about 40 kg. For a cell in Pyrex having the dimensions given in Fig. I, the fundamental frequency parallel to the plane of the h is about 450 c./sec. The length of the side limb is not critical for good resonance. Hence, the fundamental frequency can be slightly modified by altering the length of the side limb and leaving the condenser limb untouched. The glass cells stand an amplitude of 0-7 mm. or more. Since the condenser plates are 0.1 t o 0.3 mm. apart there is no risk of the cell breaking in normal operating conditions. We have adapted the hollow fork cell for studying adsorption on metal deposits prepared by thermal evapor- ation.In this case, the metal must be deposited on one of the condenser plates and not on the other. Accordingly one of the plates has to be removed during deposition and brought back for measuring. Two types of cell have been used in which the non-vibrating plate is displaced magnetically. Type I Cell.-Here, the non-vibrating plate moves in its plane by flexion of its support (Fig. 2). The plate bears a long iron rod fixed t o the tungsten lead by a pair of parallel thin tungsten wires disposed in a plane perpendicular t o that of the plate. The displacement is actuated by a magnet which attracts FIG. 2.-Interior of type I cell (filament and filament leads not W, tungsten lead WW, pair of tungsten wires, diam.0.2 mm. shown). XL, excitation-limb - 1 ~ : N , - V ; b m . & S, support of the stops 20mm e/ecfrode Fe, iron rod ; M, magnet FIG. I.-Basic cell. N-V, non-vibratingplate. sp, stops the iron rod and thus bends the two parallel tungsten wires. Two non-over- lapping positions are determined by two stops. Fig. 3 shows a cross-section of the condenser limb. The non-vibrating plate is a little less than half as wide as the vibrating plate and defines two distinct ‘ I areas ” on the latter. A metal film can be deposited on either area by passing a current through a hair-pinj 28 VIBRATING CELLS FOR CONDENSER METHOD filament, the leads of which (not shown) are sealed in the foot. The vibrating electrode consists of a nickel foil 0.3 mm.thick welded to two tungsten strips, the ends of which are sealed to the wall. The lead is a 0.3 mm. tungsten wire. It is welded on the upper strip. It crosses the wall a little higher and is stuck on t o thc external wall (through a glass sheath) on a length of about 10 cm. The condenser tube is covered from the junction with the side limb to the bottom with a metallic (nickel) deposit. This shield is prepared by thermal evaporation in a preliminary operation, the vibrating plate being protected by a screen. P F W FIG. 3.-Type I cell. P F FIG. 4.-Type I1 cell. P, Pyrex tube, diam. 20 mm. ; W, tungsten strip ; F, filament ; V, vibrating plate ; N-V, non-vibrating plate. Broken line represents second position of N-V (Fig. 4). Type I1 Gel€.-In this, the metal is deposited on the glass wall (Fig.4). The non-vibrating electrode is moved by rotation around the axis of the tube. The axle is a long, straight tungsten rod (diam. 1-5 mm.) sealed in the foot (Fig. 5 ) . A R -r . - FIG. 5.-Type I1 cell. Details of lower buckle and stop (crosses indicate welds). Cross-section of condenser tube a t level of junction with excitation limb. Dotted lines are vertical projections of lines not in the plane. A, axle ; R, rotating tungsten rod ; U-B and L-B, upper and lower buckles ; Fe I and Fe 2, main and secondary iron rods ; J, junction between Fe I, Fe 2 and R : Sp, stop ; F, filament (second lead not shown) ; F-L, filament lead ; N-V, non-vibrating electrode ; M, magnet; X-L, excitation limb.J. C. P. MIGNOLET 329 It also serves as a lead for the filament. (a) the non-vibrating electrode (through a nickel strip) ; (b) two molybdenum wire buckles fitting snugly around the axle for guiding the rotation.The lower buckle rests on a small horizontal piece of tungsten rod welded to the axle and serving a double purpose : (i) it prevents the rotating system from falling ; (ii) it acts as a stop, limiting the rotation of the tungsten rod to half a turn and thus defines two positions of the non-vibrating electrode; (c) two iron rods which, together with an external magnet, produce the rotation. The main iron rod is the one a t 180' of the non-vibrating electrode. Since the axle is slightly flexible, pushing the magnet nearer to the main iron rod draws the electrode away from the wall.This has two advantages : (i) it provides a means of adjusting the distance between the electrodes ; (ii) during the rotation from one position t o the other, the electrode is held apart from the metal deposit and does not scratch it. The second iron rod helps to reduce the condenser spacing in case of an inaccurate adjustment of the axle. In one of the two positions of the magnet, the excitation limb is in the way. This difficulty is obviated by choosing a magnet with a gap wide enough for the excitation limb t o pass through. The condenser tube is covered internally with a nickel deposit (shield) from the junction with the excitation limb to about 2.5 cm. of the bottom. The lower part is metallized during the experiment proper. Electrical contact with the vibrating electrode is effected through the shield and a small tungsten helix sealed about I cm.below the junction. The rotating system includes a tungsten rod (diam. 1-5 mm.) t o which are spot-welded : Sources of Error and Perturbating Effects Induction from the Electromagnet on the Cell and Amplifiers .- Shielding and grounding eliminates electrostatic but not magnetic induction Fortunately, there is a simple way of obviating this based on frequency doubling, i.e, the property of a non-polarized electromagnet to excite the vibration of a mechanical system at double the frequency of the current. Since the induction has the frequency of the excitation current, and the signal due to contact poten- tial has the frequency of the mechanical vibration, separation can be effected by selective amplification.Actually, we use a band-pass filter which eliminates induction from the mains as well. Of course, the second harmonic of the ex- citation current must be small. Triboe1ectricity.-Friction between the leads and glass or rubber gives rise t o irregular, and sometimes enormous, effects. Since even apparently immobile parts of the cell may vibrate a little, it is advisable to avoid loose contact between the leads and non-conductors. They must be sealed or kept apart. For instance, the leads in the foot tube cannot be isolated with either glass or rubber sheaths. They are kept a t a distance from one another and from the glass wall only by the seal and by a Pyrex bead a t the top. For the same reason, in type I cell the lead of the vibrating plate has t o be stuck on to the glass wall.Electrostatic Effects.-The inner wall of the condenser tube must be metallized, otherwise the measured contact potential varies with the hygro- metric or electric state of the outer wall. As an extreme case, rubbing the tube with a cloth produces an effect equivalent to that of a contact p.d. of a few hundred volts. The shield also eliminates effects due t o the inner glass wall which would not be detected so simply. So far, it has not been necessary to metallize the inner wall of the foot, but it would probably be a good precaution in very accurate work. Stray Vibrating Capacities.-Clearly, the capacity between the immobile electrode and the opposite area is only part of the capacity of the system con- nected t o the grid of the amplifier.No error is caused provided that the sur- faces forming the stray capacities do not vibrate appreciably. The question is primarily one of crowding. From that point of view type I1 cell is better than type I cell. Edge effects are also present and as a test the distance between the electrodes can be changed when the measured contact potential should not vary. Miscellaneous Details .-The equipment used comprises the following items: (i) for exciting the vibration, an oscillator and a push-pull amplifier when 18 W secure ample reserve of energy; (ii) for amplifying the signal, a L +330 VIBRATING CELLS FOR CONDENSER METHOD pre-amplifier (one 4060 Philips triode, battery operated), a three-stage amplifier (three 1T4 pentodes, battery operated), a band-pass filter for 450 c./sec.and finally an oscillograph. In good operating conditions, successive measurements agree to f I or z mV, but the sensitivity may fall t o f 10 mV or worse in some conditions (scaling-off the deposit, excessive frosting around the condenser tube in experi- ments at low temperatures, etc.). The accuracy obtained for surface potentials is f 10 or 20 mV as judged from the results on H and Xe films (see Part I). The same cell can be used for several series of experiments. For cleaning and replacing of the filament, the cell is cut just under the Dewar seal. Discussion The problem of designing a cell for the Zisman method is firstly one of vibration. We prefer the reverse solution, i.e. make a tube vibrate with an immobile plate in it, because then there is no need to introduce the vibration into the cell.The hollow vibrating cell is perhaps simpler as regards con- struction and production of the vibration. It seems more suitable for work in high vacuum since it contains no thin metallic diaphragm liable to cause trouble during the outgassings at 500' C. It has one weakness, it does not work with the condenser tube immersed in a viscous bath. This drawback, though real, is not important for fluid baths, e.g. liquid- air at low temperature, vapour baths at high temperatures, are available. In adsorption studies it is normal to study the variation of single surfaces, but since contact potential techniques require two surfaces, an auxiliary surface * has to be introduced.Then it is essential that its work function be known and above all remain constant. This may be a problem in itself or indeed the major problem. The value of contact potential as a tool4 in surface reactions on solids depends much on the practical possibilities of eliminating the auxiliary surface or of preventing it from changing. Perhaps a discussion of this question in relation with the cells described here is not superfluous. A surface undergoes a certain change and we propose to determine the difference in work function between the initial and final states. An interesting possibility exists if two surfaces, one in the initial and the other in the final state can be prepared and kept side by side in the same atmosphere (vacuum or gas). Then either they may be used directly as the condenser surfaces or two contact p.d.measure- ments may be made, using the same auxiliary surface. The stability of the latter can be controlled by measuring alternately with the two sur- faces. For instance, for non-volatile films, a metal layer may be deposited on one area of a cell of type I or I1 and covered with gas ; another metal layer deposited on the area and left bare. The difference between the two yields the surface potential of the film, When the change in the state of a surface is to be observed directly, care must be taken to prevent a simultaneous variation of the auxiliary surface. Once again, the case of non-volatile films is fairly simple. Measurements can be made in a vacuum and a bare metal (mostly tungsten which can be cleaned at will) chosen for the auxiliary surface.This is the solution adopted in the two-filament methods of Langmuir and Kingdon,6 Reiman It would be worth while developing vibrating cells offering that possibility too. t Zisman makes a plate vibrate in an immobile tube. and Bosworth and Rideal.' * The non-vibrating electrode in cells of type I and 11. 5 Langmuir and Kingdon, Physic. Rev., 1929, 34, 129. 6 Reiman, Phil. Mag., 1935, 20, 594. 7 Bosworth and Rideal, R o c . Roy. SOC. A , 1937, 162, I. t Incidentally, a hair-pin ribbon hanging near the wall could serve both as The ribbon could be drawn Bosworth : Proc. Roy. SOC. N . S. W., 1946, 79, 53. a source of metal and as an auxiliary surface. away from the wall for deposition and brought back for measuring.J.C. P. MIGNOLET 33 = The case where the surfaces have to be in a gas is more difficult, a t least theoretically. In the author's opinion the situation is as follows. There is no general, satisfactory way of preventing the auxiliary surface frcm changing when the temperature, pressure or composition vanes. In actual practice, however, since inert, poisoned surfaces are more readily obtained than active surfaces, it will be frequently possible to find a surface that, in a limited range of conditions, varies much less than the surface under investigation, e.g. massive nickel outgassed at 500' C and a bare nickel deposit relative to additions of H,, C,H, or C,H,, etc. As a first approximation, the effects observed may be attributed t o the bare surface. A limited control is possible with the two-area cells, where one area is available for it. The control area should be poisoned too but in a different way so as to get a control measurement different from zero. Then good constancy of the control measurement during an experiment is strong evidence that the auxiliary surface has not changed appreciably. Eventually, the experiment will have to be repeated with a chemically different auxiliary surface. Sometimes, it will be possible to get a control at some stage by freezing the film and applying methods available for non-volatile films.* I wish to express my deep gratitude to Prof. L. D'Or for help, encour- agement and advice ; and to Prof. M. Dubuisson for facilities in the early stage of this research. I also wish to thank hlessrs A. Debot, J. Haesen, and J. Sarlet for skilful technical help. Laboratoire de Chiinie gbnkrale, Universite', Libge, Belgique. *This would be an additional reason for developing cells with a hair-pin ribbon (of tungsten) as the auxiliary surface.

ISSN:0366-9033    

DOI:10.1039/DF9500800326

出版商:RSC    

年代:1950

数据来源: RSC

摘要:

J. C. P. MIGNOLET 33 = SEMI-CONDUCTIVITY AND CATALYSIS BY T. J. GRAY Received 19th Januavy, I950 The adsorption and interaction of gases on semi-conductors leads t o a vari- ation in semi-conductivity which can be successfully applied t o the study of adsorption catalysis. Fundamental information is simultaneously furnished regarding the semi-conductivity process. An experimental technique is de- scribed for a study of the adsorption of gases by variation in semi-conductivity. Although von Auwers and Kerschbaum observed the effect of oxygen adsorption on copper oxide systems, the significa-nt work on this system was due to Jusk and Kurtschatow,2 Wagner and co-workers,ap 4~ and Dubar.6 The work of Anderson? extended the observations of the effect of adsorbed gases to various other systems.1 von Auwers and Kerschbaum, Ann. Physik, 1930, 7, 129. 2 Jusd and Kurtschatow, Physik. 2. Sowj., 1932, 2, 453. Wagner, 2. physik. Chem., 1933. 22, 181. 4 Dunwald and Wagner, ibid., 1933. 22, 212. 5 Wagner and Hammen, ibid., 1938, 40, 197. IS Dubar, Ann. Physique, 1938, 9, 5. 7 Anderson, Proc. Roy. Soc. A , 1945, 184, 83.332 SEMI-CONDUCTIVITY AND CATALYSIS The application of the effect of adsorbed gases on semi-conductivity to a study of catalysis has already been described.sp s* lo A technique has been developed whereby semi-conductors can be examined during the adsorption, desorption or interaction of gases under conditions for which a high degree of reproducibility is possible. The current theory of semi-conductivity l1* l2 relates the conductivity at any temperature to a function of the number of defects.The varia- tion of conductivity with temperature is related to the activation of elec- trons either into surface levels with the production of a positive hole carrier, or into the conduction band from impurity levels in the case of intrinsic semi-conductors. The variation of conductivity during the adsorption and desorption of gases can, therefore, be taken as directly related to the amount of gas present on specific sites on the surface con- tributing to the conduction process. A study of the adsorption of gases requires a consideration of the initial adsorption of gas molecules, dissociation, surface diffusion and final interaction with the surface. Some or all of these stages may be in- volved and the overall rate of adsorption may be dependent on one or several stages.Although measurements of semi-conductivity may be considered as related to the final stage, a study of the kinetics of adsorption of oxygen on the copper oxide system has demonstrated the suitability of the method in helping to elucidate the various steps in the adsorption mechanism. Simultaneously the temperature coefficients of conductivity of the systems under varying conditions of adsorbed gas yields inormation relating to the surface energy levels associated with the conduction pro- cess and any variation which may occur in these levels from interaction or other cause. This leads to general information relating to the surface which is of particular significance in the general problem of adsorption. Orthodox methods of investigation of adsorption processes have nor- mally introduced the measurement of the amount of gas adsorbed and fre- quently an estimate of the surface area by the B.E.T.or other method. The areas involved in semi-conductivity measurements on films are of the order 50 sq. cm. superficial area, so that special methods are essential for the measurement of amounts of gas adsorbed. Expansion from calibrated bulbs at low pressure into accurately calibrated apparatus has been unsuccessful. I t may be possible to measure the amount of adsorption by controlled diffusion of hydrogen or oxygen into the system through palladium or silver thimbles respectively. This particular method is applicable to those systems where the incremental change of semi-conductivity on the adsorption of a gas is independent of pressure within certain limits for any particular temperature.The technique to be described covers the development of a greaseless reaction system containing the catalyst in the form of a glass-supported oxide film between conduction electrodes formed from an evaporated film of metal. The gas handling system has been developed with special attention to the measurement of very small quantities of gas adsorbed and desorbed. Resistance measurements are described for D.C. and A.C. by equipment capable of high accuracy with maximum speed of operation. Experimental Several forms of reaction tube have been designed and described. The general form comprises a glass cylinder on t o which a uniform metal film can be evapor- ated from a metal plated tungsten filament at the axis of the cylinder.The Garner, Gray and Stone, Proc. Roy. SOC. A , 1949, IW, 294. Gray, ibid., 1939, 197, 314. l o Garner, Gray and Stone, Trans. Faraday SOC., 1949. 00, 000. l1 Mott and Gurney, Electronic Processes in Ionic Crystals (Oxford, 1948). l2 Seitz, Modern Theory of Solids (McGraw-Hill, New York, 1940).T. J. GRAY 333 inner surface of the hard glass cylinder carries a series of conduction electrodes in the form of rings of thermal pure platinum foil fused into the surface a t suit- able intervals. The leads for the conduction electrodes are of 26 S.W.G. platinum To foreroc 350w Heufers 600w FIG. I. and lead out through intermediate seals (C 9 to C 19) t o the resistance measuring apparatus.Thermocouples are connected t o several of the conduction electrodes. The pumping system for this investigation requires special attention since very high pumping speeds are required to enable the pressure in the reaction FIG. 2.334 SEM I-CONDUCTIVITY AND CATALYSIS system t o be reduced from 0 - 5 mm. t o 10-7 mm. in 10-15 sec. A two-stage mercury vapour pump, Fig. I, has been developed with separate boilers for the primary umbrella jet and the secondary high pressure annular ring jet. The use of inner and outer cooling surfaces for the annular ring jet materially im- proves the pumping speed. Twin liquid oxygen traps are employed as protec- tion for the reaction system. The unit is backed by a two-stage rotary oil pump. FIG. 3.The main reaction system is illustrated in Fig. 2. The reaction tube a is protected from mercury vapour by dual liquid oxygen traps b. Large bore mercury cut-offs connect the tube with the main pumping lead or with the measuring and gas handling sections. A Pirani gauge c is included in the reaction tube unit. The measuring unit includes two McLeod gauges d, e, covering the FIG. 4. range 10-6 mm. to 15 mm., and two expansion bulbs of calibrated volume f, g. This unit, together with the associated mercury cut-offs, is housed in an air thermostat and can be operated externally. At one end it connects with the reaction tube system and at the other with the gas handling and desorption unit which is externally mounted (Fig. 3). This unit provides facilities for admittingT.J. GRAY 335 hydrogen or oxygen through the heated palladium or silver thimbles respec- tively h, i, these gases being stored in the reservoirs j , K. A gas inlet system I is provided and a calibrated gas burette wz. The desorption mercury vapour diffusion pump n is of the triple jet variety and returns the desorbed gas to the measuring system inside the air thermostat. The furnace unit (Fig. 4) incorporates two separate temperature regulated furnaces arranged to permit horizontal and vertical motion. Each furnace is constructed of two heavy walled stainless steel tubes with an exponentially wound nickel-chrome heater between the tubes insulated by asbestos and alundum. These are heavily lagged with asbestos and mounted in angle iron frames. Horizontal movement is facilitated by the use of large ball-bearings in V tracks with precision location of the furnace bores.Vertical movement is taken on four steel columns with a counterweight system. The furnace temperature is controlled by A.C. resistance thermometer controller operating from a nickel resistance thermometer adjacent to the reaction tube. The control is better than 0.05' C at all operating temperatures, although the control unit operation limit is better than 0-001' C and maintains heavier furnace units t o f 0.01' C cycle as measured by a Smith's Difference Bridge. Measurements of Resistance.-A review of the methods previously em- ployed in the measurement of the resistance of semi-conductors revealed no one method completely satisfactory for the projected measurements.The range of resistance t o be covered is I O - ~ - I O ~ ~ ohms, with an accuracy of 0.1 yo to 10' ohms and I yo to 1olZ ohms. In addition, the readings must be extremely rapid, variations of 5 x 105 to j x 104 ohms in less than one minute being encountered with active films. I t is also desirable to have A.C. and D.C. measurements avail- able to ensure the absence of any rectifying action. Two units have been de- veloped, the first an A.C. Wheatstone bridge and the second an electronic potentio- meter for D.C. measurements. A.C. Resistance Bridge .-Many A.C. Wheatstone bridges have been de- scribed, mainly in relation to conductance measurements, important contribu- tions being due to Jones and Joseph,13 Shedlovsky l4 and Luder.15 In general, satisfactory precision was obtained by controlling phase relationships, electro- static and electromagnetic coupling by suitable screening circuits and the use of the Wagner earth.However, the bridges so far described are relatively slow in operation and have not the range coverage for the study of semi-conductors. The bridge which has been developed employs normal ratio arms but the standard arm is replaced by two precision helical potentiometers (0.05 yo accuracy), each having 15 revolutions rotation covering 5400~. The value of these is 1000 ohms and IOO,OOO ohms. The bridge is fed a t mains frequency with an adjust- able metered voltage up to 10 V. The output from the bridge is taken by cathode follower, a high-gain amplifier and phase splitter, to the Y plates of a cathode ray tube with the X sweep operated a t mains frequency.After adjustment of the amplifier, a horizontal line is obtained for zero signal input, signal input being visible as inclination of the trace. The system has the additional advantage that phase shift in the bridge circuit immediately appears as an elliptical trace instead of a line and immediate correction may be made. Very satisfactory bridge balancing is obtained up to 108 ohms. Although a higher frequency of operation for the bridge would have been preferred, the relatively high and varying inductance of the IOO,OOO ohm helical potentiometer prevents its use above about zoo c./sec. In practice the bridge gives an accuracy of 0.1 % up t o 107 ohms with a deterioration t o 0 - j yo between 107 and I G ~ ohms ; this is combined with extreme rapidity of measurement.D.C. Resistance Measurements .-Although for preliminary work a pre- cision Kohler potentiometer was used with standard resistances for D.C. resist- ance measurements, i t was found that the time required for measurements was prohibitive. An electronic potentiometer has been developed with direct in- dication of I to 10 mV full scale deflection. This, in conjunction with a potential source and standard resistors, gives direct readings of resistance from 0.01 ohm to 1o1* ohms. I t consists of two bridge circuits, the first being tubes V, and V, and the resistor train R, while the second bridge is the current amplifying bridge V 3 ~ , B and V 4 ~ , B connected across the output of the first.The first bridge acts as a Schmitt voltage amplifier, while the second bridge acts as a current amplifier to permit connection to a current l3 Jones and Joseph, J . Amer. Chem. SOC., 1928, 50, 1049. l4 Shedlovsky, ibid., 1930, 52, 1793. l5 Luder, ibid., 1940, 62, 89 ; Rev. Sci. Instr., 1943, 14, I. The instrument is illustrated in Fig. 5.336 SEM I-CONDUCTIVITY AND CATALYSIS measuring instrument. Output is either by a pointer instrument or by galvano- meter and scale, the accuracy of the final resistance measurement being limited by the method of display. By suitable design the input tube is operated a t a grid current of less than 10-l~ A and normal operation gives an output of IOO pA for 10 mV input, although this can be increased 10-fold when required, with some sacrifice of stability.The potential drop across standard resistors is meas- ured using the standard as a grid resistor with the unknown in series and with a suitable applied potential. With the maximum standard grid resistor normally FIG. 5.-Electronic Millivoltmeter. employed, 5 x 106 ohms, the current in the resistor is greater than IOO times the grid current for a resistance of 1 ~ 1 2 ohms, which can thus be measured to I yo. The unit is spring mounted and supplied from a two-stage degenerative power supply with a stability of I in 25,000 ; all heaters are supplied by ac- cumulator. After the initial warm-up period of one hour an accuracy of better than I yo can be indefinitely maintained in the higher resistance region with considerably improved accuracy up to I G ~ ohms.General Method of Study.-The initial requirement in the present method of investigation is the successful plating of the parent metal of the oxide under investigation on to the 0.25 mm. tungsten evaporation filament. I t is found that no difficulties are encountered in plating tungsten providing the wire is cleaned by A.C. electrolysis in 10 yo caustic soda, washed carefully and immediately plated. A wide variety of cells have been used with the tungsten wire at the axis and the anode in the form of a cylinder or several rods distributed on the circumference of a circle with the tungsten wire a t the centre ; diaphragm cells have been used where applicable. Normal plating technique is employed except that the current density is increased from 2-5 times and agitation is ob- tained by nitrogen fed through the glass tube supporting the filament.The plated wire is carefully washed and spot welded into the reaction tube with the minimum delay. The reaction tube is thereafter rapidly reconnected t o the reaction system and evacuated. After thorough degassing, the filament is raised to just below evaporation temperature in the presence of atomic hydrogen obtained by high frequency discharge. This treatment cannot be unduly pro- longed as the atomic hydrogen rapidly penetrates the glass. Evaporation is then performed in hard vacuum by controlled heating a t very slow- evaporation rates. Uniformity is judged by the simultaneous appear- ance of the first transmission colour over the entire surface and by comparison of resistances between electrodes immediately measurement becomes possible.Thickness control is by resistance although actual thickness estimates are ob- tained subsequently by chemical analysis of the film after removal from the reaction system. The evaporated metal film is next oxidized t o the appropriate condition and the oxidation process followed by resistance measurements and pressure change. Frequently atomic oxygen obtained by high-frequency discharge is employed. The oxide film is processed to a semi-conducting condition and is then subjectedrr. J. GRAY 337 to prolonged evacuation at 10-7 mm. at a suitably selected temperature until a stationary value of resistance has been established. This stationary value is afterwards used as a reference value and the film is normally returned to this reference condition between successive adsorptions.The general procedure is to determine the temperature coefficient of the evacuated film and investigate any variation with oxidation state. Thereafter the characteristics of the adsorption of oxygen are determined over a range of temperatures and pressures, with particular attention to any modification of the surface resulting in non-reproducible results. During this investigation a maximum temperature is normally established, above which the state of the surface is liable to change. A suitable temperature can then be selected to obtain the greatest range of investigation. Temperature coefficients of the conductivity for varying states of adsorbed gases are determined, together with the kinetics of the processes concerned. Although desorption is studied a t various temperatures, the final desorption of the film is always carried out at the reference temperature. A detailed account of the results obtained for the adsorption of oxygen on the copper oxide system has already been given. The effect of variations of pressure is studied in detail and measurements of the amounts of gas adsorbed and desorbed made where possible. When the characteristics of the adsorption and desorption are determined, interaction of gases can be followed as described for carbon monoxide on copper oxide and the reaction between carbon monoxide and oxygen. Difficulty is, however, experienced in that the effect of such interaction may affect the surface, as does, in fact, occur in the above case a t zooo C. The author wishes to express his thanks for grants from Imperial Chemical Industries Ltd. and from Distillers’ Company Limited. Dept. of Inorganic and Physical Chemistry, The University, Woodland Road, Bristol 8.

ISSN:0366-9033    

DOI:10.1039/DF9500800331

出版商:RSC    

年代:1950

数据来源: RSC

摘要:

rr. J. GRAY 337 PROTON RELAXATION AND CATALYST ACCESSIBILITY BY P. vT7. SELWOOD AND F. K. SCHKOYER Received 24th January, 1950 The accessibility of a solid catalyst may be defined as the ratio of activity shown by a fixed weight of solid catalyst to that of the same weight of catalyst in true solution in the reacting medium. This quantity may readily be evalu- ated by the nuclear induction experiment. The proton relaxation time in water is determined, for example, for supported iron oxide on alumina. Cata- lyst accessibility is then found by relating the measured relaxation time to that observed in solutions of ferric sulphate in water. The accessibility is further corrected for 180O shielding on the under side of the surface. The accessibility then becomes a = 2M,/M, where M and M , are the apparent molar concen- trations of promoter in supported form and in true solution respectively.This paper is a further progress report on application of the nuclear induction experiment to problems in heterogeneous catalysis. The results given below supplement to some degree the information from mag- netic susceptibility measurements on similar systems. In particular, the nuclear induction experiment may be used t o compare the extent t o which, in different catalyst samples, supported transition group oxides are accessible t o the molecules of a reactant. The supported systems Spooner and Selwood, J . Amer. Chem. Soc., 1949, 71, 2184.338 PROTON RELAXATIOK AND CATALYST ACCESSIBILITY studied were ferric oxide, chromic oxide, and cuyric oxide, all on high- area y-alumina.Susceptibility d a t a and some catalytic activity d a t a have already been reported for these systems. 2-4 Experimental The nuclear induction experiment is that of Bloch, Hansen, and P a ~ k a r d . ~ Their procedure was modified as follows. A raised rim was added to the flat pole pieces of the magnet. This compensates for the radial inhomogeneity of the simple gap.* In place of the paddle used by Bloch to regulate the flux linkage between the primary and secondary coils this experiment employs an auxiliary coil, wound coaxially both with the primary and secondary coils which is in parallel with a variable resistance and a variable capacitance. A simple phase shifting network has been added to the external synchronizing signal circuit, and a square-wave generator and an audio oscillator are used t o calibrate the oscilloscope sweep.The modulation of the direct current field is accomplished by either a 60-cycle signal from the mains or by an audio oscillator and power amplifier, if other frequencies of modulation are desired. There is no radio frequency amplification in the part of the circuit containing the secondary coil. Preparation and analysis of samples have been described .*-4 Preparation consists, in brief, of impregnation of the support by a nitrate solution, followed by ignition. Catalyst samples of different active oxide concentrations were obtained by varying the concentration of the impregnation solution. For reasons described elsewhere,' the catalytic activities of supported oxides are most easily compared a t constant space velocity.This is achieved by using the technique of mechanical dilution. To compare the activities of a series of chromia-aluminas of varying chromia content, each sample except the most dilute is mixed mechanically with pure support until all samples contain the same percentage of chromia. A series of samples ranging from 2 t o 30 yo chromium content would thus all be reduced to 2 yo, but they would differ in the proportion of alumina used as a support t o that used merely as a mechanical diluent. For the first results t o be given below a sample containing 17.1 Yo iron as iron oxide supported on alumina was progressively diluted mechanically with alumina. The proton relaxation times were determined as a function of weight of iron in the several samples.The remaining results on supported oxides of iron, chromium, and copper were obtained on samples all diluted to a fixed percentage of active oxide. The actual percentage of supported element in each case is indicated in the Tables. Proton relaxation times are reported as a function of original concentration before mechanical dilution. In all relaxation studies, I g. of catalyst sample was moistened with I ~ m . ~ of water. Water was chosen as the proton source in all these experiments, which must be regarded as preliminary t o studies involving the use of gaseous proton sources. Results In Table I there are given relaxation times as a function of iron concentration for a series of iron oxides supported on alumina, and prepared by progressive TABLE I PROTON RELAXATION TIMES ON IRON OXIDE- ALUMINA DILUTED FROM 17.1 yo IRON Actual yo Fe 1 T (sec.x I$) 17.1 12.8 8.55 4-27 2-14 3-90 5-00 7-00 10.8 15'5 2 Selwood, Ellis and Wethington, J . Amry. Chem. SOC., 1949, 71, 2181. 3 Eischens and Selwood, ibid., 1947, 69, 1590. 4 Selwood and Dallas, ibid., 1948, 70, 2145. 5 Bloch, Hansen and Packard, Physic. Rev., 1946, 70, 474. 6 Rose, ibid., 1938, 53, 715. 7 Eischens and Selwood, J . Amer. Chem. Soc., 1948, 70, 2271.P. W. SELWOOD AND F. K. SCHROYER Original % Fe T (sec. x 103) 0'37 0.70 2'5 3'0 9'5 17.1 1'0 5'8 6.9 8.2 7'3 10.7 14'5 19.5 30.1 12'0 TABLE I11 PROTON RELAXATION TIMES ON COPPER OXIDE- ALUMINA DILUTED TO 1.2 yo Cu Original yo Cu 1'2 2'1 3'3 4'9 10'0 T (sec. x 103) 3'55 5'39 8.80 5-87 12.9 339340 PROTON RELAXATION AND CATALYST ACCESSIBILITY mechanical dilution of a sample containing 17.1 yo iron.The members of this series differ only in percentage of iron actually present during the measure- ments. In Tables 11, 111, and IV there are given relaxation times for supported oxides of iron, chromium and copper, respectively. These results were obtained by the method described above of diluting the various members of a series down to a fixed concentration. The members of all these series differ in the ratio of alumina used as support to that used as a mechanical diluent. For reasons to be indicated below i t was necessary to remeasure proton relaxation times as a function of concentration for aqueous solutions contain- ing the ions Fe+++, Cr+++, and Cu++.These data are all given in Fig. I. Discussion We shall first compare the relaxation times found over the hetero- geneous catalysts with those found for solutions. A convenient method for doing this is to compare molar concentrations of paramagnetic ions necessary to give equal relaxation times for the two cases of heterogeneous and homogeneous catalysis. Thus, we shall say that M is the concentra- tion of supported paramagnetic ions in moles of ions per litre of the water used as a proton source. The concentration is found to give a relaxation time of T . Then M o is the molar concentration of an aqueous solution which will also give a relaxation time T . It is assumed that the para- magnetic species is the same in each case ; but if the moment of a sup- ported ion is different from that of the same ion in true solution,2 then a correction which may be made is It will be noted that the relaxation time varies inversely as the square of the moment.The only important case in which p o and p are notably different is for iron oxide where p is about 70 yo of p o . A final term will correct for the fact that a dissolved ion is accessible t o protons from all sides, but a supported ion may be reached from one side only. We have then a = 2M0/M, where a may provisionally be referred to as the catalyst accessibility. In Fig. z there are given accessibilities for supported iron oxide on Tcorrected = Trneasured (psupported) ' / ( P o in solution)2- 60 I I I FIG. 2.-Accessibility of iron oxide supported on alumina.alumina, calculated from the data of Table 11. It will be noted that the accessibility shows an expected increase as the original iron concentration is diminished. Fig. z bears a superficial resemblance t o a plot of sus- ceptibility against original iron concentration, but it must be remembered that the measured susceptibility is the average susceptibility of all ironP. W. SELWOOD AND F. I<. SCHROYER 34 I ions in the system, while the relaxation time depends on the moment of those iron ions exposed on the surface, and thus accessible to incoming molecules of reactant. For supported chromia there are already available catalytic studies on the dehydrocyclization of n-heptane.' In Fig. 3 there is plotted FIG. 3 .-Catalytic activity as a function of accessibility for chromia-alumina.catalyst activity against catalyst accessibility. I t is clear from the linear relation observed that for this system and reaction it is possible to predict activity from a measurement of nuclear relaxation. In Fig. 4 there is plotted the accessibility of supported copper oxide FIG. 4.-Accessibility of copper oxide supported on alumina. as a function of original concentration. Here we have the anomaly of accessibilities considerably greater than I 00 yo. This at once illustrates the crudeness of our approach to catalyst accessibility, and suggests how the approach may be refined. Turning to the paper of Bloembergen,342 PROTON RELAXATION AND CATALYST ACCESSIBILITY Purcell, and Pound,8 we have reference to their eqn.(54), which is as follows : I 12,rr'y'vNp' - (1) - = T 5kt ' where y is the gyromagnetic ratio, the viscosity, N the number of para- magnetic ions per ~ m . ~ , p the effective moment, k the Boltzmann constant, and t the absolute temperature. It is clear from this equation and from the results reported here that the viscosity environment of a water mole- cule near the surface of a catalyst must be considerably larger than that in normal water. This conclusion is, of course, entirely consistent with current speculations concerning the nature of adsorbed films of water on the active surface. A preliminary measurement of v / v o gives the value of 3, (where 17 is the viscosity factor for the supported ions, and q o for the true solution). FIG. s.-Relaxation time as a function of (yo Fe)2/s and of (yo Fe)% for iron oxide-alumina.There will now be presented a provisional theory by which the relaxa- tion times may be predicted for different members of a catalyst series. It will be noted from eqn. ( I ) above that I/T - N. The experimental results of Bloembergen, Purcell, and Pound as well as those reported here show that this relation is obeyed over a wide concentration range for iron, chromium, and copper salts in solution. But it is also to be noted that at very low concentrations there is a deviation which is better repre- sented by I / T N N%. The reason for this deviation is not known, but Bloembergen, Purcell and Pound, Physic. Rev., 1948, 73, 679.P. W. SELWOOD AND F. K. SCHROYER 343 it is believed to be significant in the interpretation of the results on sup- ported oxides given below.Examination of the data of Table I where any interaction between adjacent iron ions is constant, shows that I / T - (% Fe)21s over a wide concentration range, but that I / T - (yo Fe)'l2 is obeyed over a wider concentration range. These observations are shown in Fig. 5 . It will be noted that [( % Fe)e/s]s/r = (yo Fe)lln, and also that the total surface of iron ions exposed must be proportional to (Yo Fe)'h. Hence it may be said that the exposed surface of the iron particles behaves in the hetero- geneous nuclear relaxation experiment in exactly the same way that ionic concentration behaves in the homogeneous experiment. Let it be assumed then that the function of paramagnetic ions per cm.3 in solution is taken by ions per cm., in the surface of the solid catalyst.We shall further assume that the concentration of such ions in the surface is proportional to the total surface. Take two samples of differing paramagnetic ion content and dilute them mechanically to the same concentration by the addition of pure support. Susceptibility measurements have shown that such supported oxides are aggregated into particles, even though the concentration of active oxide be very small. Let n, and n, be the number of such particles of radius rl and Y , respec- tively, and of density p in the two samples. p n,42!3 = p nlrir4ia or n, In, = vi/vi. Then The surfaces of supported oxide S,, and S , will be related by s2/S1 = (4n %&) / ( P '%v:), then, substituting for n,/n,, we have S,/S, = (rp;)/(v;v;) = vljv2.Going back to our original assumption that T,/T, = S,/S,, we have T1/T2 = Y,/Y, = C,'Ia/C~ls, where C1 and C, are the original percentages of paramagnetic ion present in the undiluted samples. The conclusion is, therefore, that the re- laxation time should be proportional to the cube root of the original concentration in a series of catalysts all members of which are diluted mechanically to a fixed concentration. Fig. 6 shows relaxation time plotted as a function of the cube root FIG. 6.-Relaxation time as a function of the cube root of original concentration for supported oxides of iron, chromium and copper on alumina. of the original concentration for supported oxides of iron, chromium and copper. In general The data are derived from Tables 11, I11 and IV.344 SCATTERING OF X-RAYS AT SMALL ANGLES it is confirmed that relaxation time in these systems is roughly linear with the cube root of original concentration. At extremes of concentration some deviations may be expected parallel to the deviations from I/T - N noted above. At high concentration deviations may be noted owing to increasing exchange interaction between adjacent ions. The agreement is best for supported copper oxide which, according to susceptibility measurements] is the most magnetically dilute. In conclusion it may be observed that the nuclear induction experi- ment gives the same kind of information as may be obtained from sus- ceptibility measurements on the same systems, but that the nuclear in- duction effects are related to the surface ions rather than to the whole mass of paramagnetic ions. The information obtainable by nuclear induction is similar to that from ortho-para hydrogen conversion studies. But the nuclear experiment has a much wider range of applicability. This work was done under contract with the Office of Naval Research. Departments of Chemistry and Physics, Northwesterm University, Evanston, Illinois, U.S.A .

ISSN:0366-9033    

DOI:10.1039/DF9500800337

出版商:RSC    

年代:1950

数据来源: RSC

摘要:

344 SCATTERING OF X-RAYS AT SMALL ANGLES STUDY OF CATALYSTS BY SCATTERING OF X-RAYS AT SMALL ANGLES BY A. GUINIEK Received 30th January, 19 jo The possibilities of using small-angle scattering of X-rays for the determin- ation of the degree of dispersion of catalysts are discussed. It is concluded that these measurements cannot alone provide precise information about the different size and shape of an irregular array of particles of a powder catalyst, but in simple cases or when they are supplemented by information obtained by other methods, such as by electron microscopy. the quantitative use of X-ray data is possible. One of the essential factors which determine the properties of a catalyst is its mean grain-size, or the total area of the grains. The object of this paper is to show t h e possibilities of using small-angle scattering of X-rays for the experimental determination of the degree of dispersion of matter in catalysts. Experiment a1 I t may briefly be recalled that the occurrence of scattering of X-rays a t small angles (Fig.I ) is characteristic of the heterogeneous state of the specimen ; i t does not occur unless the specimen shows local variations of electron density. The scattering is limited t o an angle of approximately A/d, h being the wave- length of the primary beam and d the mean diameter of the grains. With current experimental techniques it is difficult t o study scattering a t angles less than 10' from the primary beam; consequently, with readily accessible wavelengths, particles of diameter greater than 600 A cannot be studied.If, on the other hand, the particles are too small, scattering occurs over a larger angle, but becomes too weak t o be detected. The conditions for convenient study of a specimen of a fine powder by the low-angle scattering method are as follows : (a) the particles should be immersed in a medium of considerably different electron density ;A. GUINIER 345 ( b ) the dimensions of the particles should be of the order 20 to 500 A. The first condition is usually realized in powder catalysts ; the second limits the use of the method to the most finely divided materials. C FIG. I. Results and Discussion The results of the experimental work is a curve relating the intensity I of scattered X-rays to the angle of scattering t. Consideration must now be given to the deduction from this curve of the size of the particles.It is known that when the particles are identical and sufficiently far removed from each other for interparticular interferences to be neglectEd, the " radius of gyration ' I , a quantity representing the size of the particles] can be determined from the experimental data. (The radius of gyration is the root-mean square ofthe distances from all the electrons in the particle to its centre of gravity.) The radius of gyration is deduced by plotting log I against c2 ; the plot tends towards a straight line as E tends to zero, and the limiting gradient of the curve gives the radius of gyration. This result has been satisfactorily verified by comparison of an electron- microscope photograph and an X-ray scattering photograph for a colloidal gold preparation containing very uniform grains. When the preparation is somewhat less homogeneous, calculation of radius of gyration from the scattering curve still gives a good approximation to the mean value for the particles.It should be noted that a knowledge of the radius of gyration is not sufficient for determination of the dimensions of a particle, which requires some assumption as to its shape. For example, a sphere and a series of elongated ellipsoids, which may have the same radius of gyration, cannot be distinguished by the scattering method. The case of a specimen with identical and widely dispersed particles, allowing accurate application of the low-angle scattering technique, is unfortunately rare in the study of catalysts. Often there is reason to believe that the grains are of widely varying size and shape ; moreover, in powder catalysts the grains are in contact, and distortion of the scatter- ing curve by interparticular interferences may be expected. Heterodisperse Specimens.-Consideration may first be given to the case in which the grains are dispersed but of varying size.Evidently the observed intensity oj scattering will be the sum of the scattering intensities from different groups of particles of given size. For each of these groups the plot of log I against c2 will tend to become linear a t small angles, but the log I - c2 plot will show greater curvature and be concave upwards. Several authors 3, 49 have attempted t o analyze these complex curves in order to determine the distribution of particle sizes from low- angle scattering data.The most correct method is that of Shull and Guinier, J . Chim. physique, 1943, 40, 133. Turkevjch and Hubbell, Physic. Rev. A , 1948, 73 (ii), 1250. Hosemann, Z. Elektrochem., 1940, 49, 535. Jellinek and Fankuchen, Ind. Eng. Chem., 1945, 37, 158. Shull and Roess, J . APPZ. Physics, 1947~ 18, 295, 308.346 SCATTERING OF X-RAYS AT SMALL ANGLES Roess 3 ; a particular particle shape 1s assumed, all particles in the specimen being similar in this respect, and a distribution lsw (e.g. a Maxwellian distribution) of sizes about a specified mean value is intro- duced, this mean value being an arbitrary parameter. An a priori calculation of the diffusion curves as a function of the parameter is then possible, and the parameter is determined by selecting the curves ccr- responding most closely with the experimental data.This procedure certainly provides a solution to the problem but it depend? on two assumptions-the shape of the particles, and the size-distribution law. I f it is desired to determine a mean or integral value (giving, for example, the total surface area), these assumpticns may affect the result, and the scattering technique alone does not give a unique solution. It may be used if the assumpticms made are justified by other available information about the catalyst, in such a way that the possible results lie in a rather narrow range. Thus Fankuchen et ~ 1 . ~ and Shull and Roess have obtained results for total surface areas of alumina catalysts which agree with values determined by other methods.Compacted Powders.-The influence of interparticular interferenes is particularly important with particles of very uniform shape and size, the significant factor being the regularity of disposition of the particlcs. We have verified that solutions of macromolecules of known identity give scattering curves which change considerably with concentration.e FIG. aa.-Curves of scattering intensity ; log I plotted against P. Con- tinuous curve : colloidal solution of silver. Dotted curve : flocculated colloidal silver. (The radii of gyration deduced from the linear portions of these curves are, respectively, 48 and 50 A. The mean grain diameter measured on the electron micrograph is 120 A, which, in the case of a sphere, corresponds to a radius of gyration of 47 A.*4 ring appears, instead of a central maximum, in diagrams for con- centrated solutions. However, if the particles are of differat sizes, even if they come into contact, there is no frequently repeated value of the distance between neighbouriag particles ; under these conditions a ring shGuld no longer occur in the diagram. We have found, in fact, that the same scattering curve is obtained from a dilute cclloidal soluti~n of silver and the same colloid after flocculation, although in the latter case the density of the silver is considerably increased (Fig. 2 ) . We con- sider that this result shows theories applying to dilute solutions to remain roughly valid, in general, for powder catalysts.Fournet and Guinier, Compt. rend., 1947, 224, 1848.FIG. 2b.-Electron microgr-al)li of the colloidal silver (magnification x 82,000). [To face page 346.A. GUINIER 347 In conclusion, a low-angle scattering measurement is carried out in the following way. The intensity of scattering is measured as a function of the scattering angle, and log I is plotted against 9. If the curve has a linear portion of appreciabk length, the particles are sufficiently uniform in dimensions, and the radius of gyration may be deduced. If thz contrary is observed, the particle sizes are widely dispersed, and an analysis of the size distribution can be made with certainty only if assump- tions are made, the validity of which determines the value of the results. Comparison with the Debye-Scherrer Method.-Low-angle scattering depends only on the dimensions of the unit domain of matter, and not on the internal arrangement of atoms in that domain.Thus the measure- ment ~f width of Debye-Scherrer lines for crystalline materials may in- dicate a smaller grain-size than thc scattering method, if each grain comprises several crystallites of different orientation or if the crystal lattice is distorted. We have examined two carbon-black samples, one of which gave broad Debye-Scherrer lines and the other lines so diffuse that they resembled rings from an amorphous material ; low-angle scattering from both samples was similar in extent. This shows that the grains were of similsr size in b t h cases, but that the state of crystal- lization was different ; one sample was almost amorphous carbon, whereas the other approached graphite in structure.This example shows how study of the central part of the diffraction diagram augments information given by classical techniques, and prevents erroneous interpretations based only on the study of D2bye-Scherrer line-w idth. Comparison with the Electron Microscope .-From a comparison of X-ray data with results given by the electron microscope, it becomes evident that the latter instrument gives much more precise information about the specimen. The electron microscope does not suffer from restrictions inherent in the X-ray methcd, and the distribution of par- ticle sizes in a heterogeneous sample can lx analyzed. The X-ray method, however, offers two advantages; firstly, the specimen can be examined while in suspension in a liquid.For electron-microscopic examination the specimen must be dried ; the drying may alter the particles or pro- duce coagulation, so that the true free particle in the system is nct actually observed in the microscope. The second advantage is in the examination of porous materials containing minute cavities ; these produce regions of heterogeneous electron density which, if of appropriate size, cause strong low-angle scattering. Active charcoal, for example, produces strong xattering, although prior to activation, scattering from the same sample is only weak. This is attributed to the formation of porosities some tens of Angstroms in Fize. In general, the electron microscope shows only the apparent contour of the grain and does not didase internal cavities. It is just such porosity that may be important in the action of catalysts. Conclusion.-For systems as complex as an irregular array of par- ticles of different size and Fhape, such as a powder catalyst, the X-ray scattering method above cannot give a precise and complete answer to the prcblem of granulometry. Nevertheless, in simple czses, or when complementary information secured by other physical methods is avail- able, quantitative use of X-ray data is possible. I t should not be forgotten that, even in the general case, low-angle scattering may be used qualitatively ; two catalysts of the same material may be compared, the form of the log I - curves showing which is the more dispersed or more homogeneous. We have shown that the quali- tative selection of two varieties of carbon black used as rubber fillers was the same as that made by study of the filled rubber. In such a case X-rays provide a convenient and practical test of the quality of the product. Conservatoire National des Arts et Me'tiers, 292 Rue Saint-Martila, Paris 3""'.

ISSN:0366-9033    

DOI:10.1039/DF9500800344

出版商:RSC    

年代:1950

数据来源: RSC

摘要:

ELECTRON MICROSCOPY AND SMALL ANGLE X-RAY SCATTERING BY JOHN TURKEVICH,~ HARRY H. HUB BELL^ AND JAMES HILLIERC Received 27th February, 1950 The application of the electron microscope to the study of catalytic sub- stances is discussed. The substances commonly used as heterogeneous catalysts reveal under electron microscope examination a diversity of form : amorphous materials, small spherical bodies, fibres and plates. Applications are cited in the use of the electron microscope for the study of catalyst preparation and sintering. The difficulties of this method of the examination of the fine structure are presented. An apparatus for small angle scattering of X-rays for particle size determination is described. Correlations are presented on the particle size determinations by the electron microscope with those obtained from the low angle scattering of X-rays.Examination of the fine structure of catalysts is a matter of some importance in any comprehensive study of heterogeneous catalysis. One aspect of such a study has been the thorough investigations carried out on the adsorptive characteristics of surfaces for gases. Van der Waals' adsorption, activated adsorption and heat of adsorption measure- ments have all been utilized for characterizing the extent, the chemical character and the heterogeneity of catalytic surfaces. Another aspect of such an examination of fine structure is the study of solids by X-ray diffraction.1 Application of this tool using the Bragg equation for powders is, however, limited since most catalytic bodies are so finely divided that they either give no lines or at best produce diffuse diffraction bands.Crystallite size may often be determined from the width of such bands and in recent years the diffuseness of the central spot of the usual X-ray powder diagram, the low-angle X-ray scattering, has been used to char- acterize in a semi-quantitative way the particle size of specimens. The applicability of the small angle X-ray scattering has not been fully tested by the direct correlation with the results of electron microscopy. The invention of the electron microscope offered an opportunity to determine the morphology of catalyst preparations with a resolution which approaches 15 %i under most favourable cases. A survey of various catalytic solids 2-s disclosed amorphous materials such as activated alumina (Fig.I ) , silica gel, cracking catalyst ; spherical metal particles such as colloidal platinum ; fibrous materials as vanadium pentoxide a Chemistry Department, Princeton University, Princeton, New Jersey. C R.C.A. Laboratories, Princeton, New Jersey. Physics Department, Middlebury College, Middlebury, Vermont. Jellinek and Fankuchen in Advances in Catalysis (Academic Press, New Ruska, Kolloid-Z., 1940, 92, 276. Rampino, Kavanaugh and Nord, Proc. Nut. Acad. Sci., 1943, 29, 247. Shekhter, Roginski and Isalo, Acta Physicochim., 1945, 20, 217. Turkevich, J . Chem. Physics, 1945, 13, 135. Roginski, Shekhter and Sakharova, Compt. .rend. U.B.S.S., 1946, 52, 687. Roginski, Tret-yakov and Shekhter, J . Physic. Chem. SOC. Russ., 1949, 23, York, 1948), p.256. ' Turkevich and Hillier, Ind. Eng. Chem. (Anal.), 1949, 21, 475. 50, 1152. 3 48J. TURKEVICH, H. H. HUBBELL AND J. HILLIER 349 (Fig. 2), asbestos (Fig. 3 ) ; plate-like substances such as aluminium oxide hydrate, tungsten oxide, activated charcoal, kaloin, bentonite clay. Supported catalysts such as platinum on asbestos or platinum on charcoal show well-recognizable dense particles scattered along a fibre or distributed on the surface of the plates. Sintering process is readily visualized as an agglomeration and fusion of fine particles while steam deactivation of activated alumina can be morphologically recognized a s the formation of lath-like growth with a concomitant decrease in the surface area. Electron microscopy also furnishes the catalytic chemist with a powerful tool to investigate the ageing of many hydrous oxides. In the alumina sol this phenomenon expresses itself in the aggregation of fine granules into flexible fibrils which ultimately combine to form crystal- lites while in the vanadium pentoxide sol the short fibrils lengthen into long rods which coalesce to form successively elliptical tactoids and rect- angular cry~tallites.~ The general conclusion that can be drawn from an electron microscope examination of catalysts and related colloid systems is the presence of diversity of form and a rather wide distribution of par- tic19 size.A clear cut correlation between particle size and catalytic activity is therefore difficult. A programme was therefore initiated by Dr.P. C. Stevenson lo and one of us (J. T.) on the electron microscope examination of the various modes of preparation of the classical colloidal system, finely divided gold. It was found possible to separate the nucleation process from the growth process in such a way as to produce particles whose average radius can be varied from 100-1000 b while the deviation from the mean radius was at most 12 yo. While such colloidal gold is unsuited for preparation of catalytic material its study has not only furnished samples for calibration of other methods for the determination of particle size but has also given an insight into the mechanism of nucleation and growth of finely divided particles. There are several difficulties involved in the use of the electron micro- scope for catalyst examination.The material studied must be so dis- persed that each morphologically significant body stands out as a separate unit. This is often difficult to attain especially when the intermolecular forces are great as in the case of the forces between graphite plates or when the elementary particles are fused together by previous treatment at high temperatures, Another difficulty is associated with the fact that significant structure is revealed only at a resolution close to the limiting one attained by the present-day microscopes. Thus an examination of a material with a resolution of only a IOO b often yields information of limited value. Further difficulty is the demand made on the specimen, that it withstand high vacuum and not be excessively thick (less than 4000 A).Finally, i t must be emphasized that the electron microscope presents us with a silhouette of the inorganic material on the collodion mounting screen and if there is an internal sub-structure it will not be revealed unless fortuitously obtained by the process of dispersion. It was therefore felt desirable to invoke the aid of small angle X-ray scattering to reveal whether there is any sub-structure in the particles which appear in the electron microscope as compact units and whether there are particles in the specimen which are invisible because they fall in size below the resolution limit of the microscope. The first step in this programme was a study of the experimental aspects in the determination of the small angle scattering of X-rays and a calibration of the method using citrate colloidal gold as the specimen.ll This sample of gold is quite uniform in size at IOO A radius and 12 % deviation from the mean.Watson, Heller and Wojtowicz, J. Chem. Physics, 1948, 16, 998. lo Stevenson, Some Experiments on Colloidal Gold (Princeton University l1 Turkevich and Hubbell, submitted to J . Amer. Chem. SOC. Thesis, I 949).350 ELECTRON MICROSCOPY AND X-RAY SCATTERING value of 107 was obtained from the slope of the linear portion of the logarithm of the intensity against the angle squared of the scattering curve. This is in good agreement with the prediction of the exponential curve suggested by Warren,l2 Fankuchen,' Guinier l3 and others. The experimental scattering curve also revealed a flat portion which could be interpreted as evidence of the maximum predicted by Guinier l3 but modified to give a flat portion because of the use of slits instead of pin holes and because of the finite distribution in the size of the colloidal particles.The small but significant distribution in the gold particle size did not express itself in a curvature of the above-mentioned linear portion of the scattering curve plotted as the logarithm of intensity against the angle squared. It is proposed in the present communication to examine the correlation between particle size as determined by the electron microscope and small angle X-ray scattering. Experimental The apparatus is illustrated in Fig. 4 and consists of a North American Phillips X-ray unit, a long collimating system and a small angle camera adapted both for photographic survey work and careful intensity measurements by - - - -& - - - - , I i I X . U ..".C 4U.'..* FIG. +-Small angle X-ray scattering apparatus (3 actual size). scanning with a Geiger-Muller counter tube. The experimental results were plotted as the logarithm of intensity against the scattering angle squared. Filtered chromium, copper and iron Ka radiation was used. The particle size distribution was obtained using the method of resolution of such curves proposed by Fankuchen, Jellinek and Solomon.l* l2 Biscoe and Warren, J . Appl. Physics, 1942, 13, 364. l3 Guinier, Ann. Physique, 1939. 12, 161. l4 Jellinek, Solomon and Fankuchen, Ind. Eng. Chem. ( A n d ) , 1946, 18, 172.FIG. I .-Electron micrograph of Alorco activated alumina ( x I 6,joo).FIG. 2.-Electron micrograph of vanadiunl pcntoxicle sol ( x 204,000). [To iace $age 350.FIG. 3 .-Electron micrograph of chrysotile asbestos ( x 31,000).FIG. 5.--Faraday gold sol ( x IOO,OOO).FIG. 6.--Tannin gold sol ( x 124,000)..J. TURKEVICH, H. H. HUBBELL AND J. HILLIER 351 The sodium citrate gold sol was prepared by adding 45.0 ml. of water to 50.0 ml. of chlorauric acid solution containing 0.1 mg. of gold per ml. and bringing the solution to the boiling point. 5.0 ml. of I yo sodium citrate solution was added t o the boiling solution and the boiling continued with good mechanical stirring for about 7 min. Faraday sol was prepared by treating IOO ml. of chlorauric solution contain- ing 0.5 mg.of gold and 0.5 milliequivalents of potassium carbonate with 2.0 ml. of a saturated solution of yellow phosphorus in diethyl ether and stirring con- tinuously at room temperature. When the reaction was judged to be complete from the depth of colour developed, the deep ruby-red colloid was heated t o the boiling point and filtered air was drawn through the solution to oxidize any remaining phosphorus, Tannin gold sol was made by neutralizing to litmus IIO ml. of auric chloride solution containing 0.01 mg. of gold per ml. with sodium carbonate solution and then heating it to the boiling point. A fresh I ?& tannin solution was added dropwise until no further development of colour resulted. Un- activated charcoal was a coconut char. It was steam activated at IOOO~C t o give the activated charcoal.I t was then impregnated with chlorplatinic acid and heated to give 10 yo platinum by weight. The final colour of the colloid was deep wine-red. Activated alumina was the commercially available Alorco alumina. Discussion For sodium citrate sol the agree- ment between the radius of the particles as determined from the small angle scattering with the result of 100 A as determined with the electroii microscope, is excellent. In the case of the Faraday sol and the tannin TABLE I.-PARTICLE SIZE DISTRIBUTION AS DETERMINED BY X-RAY SCATTERING The results are presented in Table I. Material - Sodium citrate gold sol . Faraday gold sol . Tannin gold sol . Activated alumina . Unactivated charcoal . Activated charcoal . 10 yo Platinum on charcoal . Radius (A) 107 61 107 75 170 56 68 208 91 38 I00 I20 246 34 I05 169 29 33 36 % - 91 7 95 5 61 28 9 69 28 3 88 2 2 I 0 2 61 29 I 0 sol the electron microscope reveals a wide spread in particle size, the size ranging from 15 to 40 with a most probable value of 25 A for the Faraday sol (Fig.51, and 15 to 75 A with a most probable value of 55 A for the tannin sol (Fig. 6). The results of the X-ray scattering do not reveal the same type of heterogeneity in the samples but do indicate a preponderance of particles of radius close to the maximum radius as revealed by the electron microscope. Activated alumina reveals under3 52 STABLE AND UNSTABLE ISOTOPES the electron microscope a porous structure which is difficult to characterize in terms of pore and particle size. The latter may be estimated as close to 20 A in radius. The small angle scattering indicates larger particle size and an apparent distribution in size which may be associated with the porous structure. The X-ray data on the three charcoals is con- sistent within the group in that steam activation results in the production of particles of smaller " radius " and impregnation with metallic platinum not only increases the total scattering curve by a factor of ten but reveals particles of a radius of 29-36 A. The electron micrographs disclose that these samples are aggregates of plates that are difficult to disperse and therefore difficult to characterize quantitatively. In conclusion we wish to state that mere examination of the X-ray scattering does not reveal the morphology of the specimen nor the dis- tribution in size, but must be supplemented by electron microscope examination. Chemistry Department, Princeton University, R. C . A . Laboratories, Princeton, N . J .

ISSN:0366-9033    

DOI:10.1039/DF9500800348

出版商:RSC    

年代:1950

数据来源: RSC

摘要:

3 52 STABLE AND UNSTABLE ISOTOPES STABLE AND UNSTABLE ISOTOPES IN CATALYTIC RESEARCH BY JOHN TURKEVICH,* FRANCIS BONNER,** DONALD SCHISSLER * AND PETER IRSA** Received 6th February, 1950 A discussion is given of the advantages and disadvantages of the use of stable and unstable isotopes in catalytic research. The application of radio- active carbon t o the study of the reduction of carbon dioxide by carbon is pre- sented. The results of the use of radioactive hydrogen for the study of the catalytic double bond migration are cited as evidence for the theory of a hydrogen switch mechanism. A discussion is given of the use of deuterium and mass spectrometry t o determine the number and position of the deuterium atoms in molecules containing varying amount of the deuterium atoms.This use of the stable isotope of hydrogen is illustrated by examples of the analysis of the reaction of ethylene and deuterium on a nickel wire catalyst. The number of radioactive isotopes available t o the catalytic chemist and of particular use to him is limited. There are the exceptionally weak ,&emitting tritium, the weak 8-emitting 14C, a sulphur isotope and one of phosphorus. On the other hand, there are no radioactive isotopes of nitrogen and oxygen of convenient life. Definite advantages are gained in the use of radioactive isotopes in that they may be employed in minute amounts and still be detected by inexpensive counting equipment. Furthermore, in many cases it is possible to monitor continuously the activity during the course of an experiment.It should be pointed out, however, that the use of a radioactive isotope gives only an overall picture of the exchange reactions that take place when there are two or more non-equivalent atom positions in a complex molecule. To determine the position of the radioactive tracer atom in such a molecule, one must chemically degrade the molecule in well-defined steps and determine in what fragment the radioactivity persists. Such chemical degradations * Chemistry Department, Princeton University. ** Chemistry Department, Brookhaven National Laboratory.TURKEVICH, BONNER, SCHISSLER AND IRSA 353 may be subject to rearrangements and exchanges and are, a t best, time- consuming. Not only is the specific radioactivity of the sample inde- pendent of the position of the tracer atom in the molecule, but it is also independent of the distribution of the tracer atoms among several mole- cules.Thus, the specific radioactivity will be the same if there is one molecule with two tracer atoms and another with no tracer atom, or if two molecules are present each with one tracer atom. The availability of stable isotopes, deuterium, 13C, 15N, l*O, together with the use of the mass spectrometer for analysis, offers a method for solving the problem of determining the distribution of the tracer atoms among and within molecules. In the mass spectrometer the molecule is broken down by electron impact to give a pattern characteristic of the molecule. The mass spectrum so obtained is governed in its main features by the rupture of valency bonds in the molecule. A number of model deuterium-substituted compounds, not only chemically pure but isotopic- ally pure, have been prepared t o determine the pattern relationship between the protium model compounds and the various deuterium model compounds. The spectra of the model compounds are used to convert the positive ion currents for the various masses obtained on the mass spectrometer, into composition of the various polydeuterated compounds.An Example of the Use of 14C.-As an illustration of the use of a radioactive tracer atom in a heterogeneous reaction, consider the re- action between carbon dioxide containing 14C and charcoal which was investigated at the Brookhaven National Laboratory by Bonner and Turkevich. The apparatus is illustrated in Fig.I. The radioactive I FIG. I.-Apparatus for the study of the reaction between charcoal and WO,. carbon dioxide was generated from the Oak Ridge barium carbonate by mixing it with lead chloride and heating it to 400° C.1 The charcoal granules were placed in one arm of a thermal syphon E, made in part out of quartz, so that it could be heated to IOOOO C. The other arm of the thermal syphon contained a trap for freezing out the carbon dioxide F, a manometer for measuring the pressure changes D and a counter G. The latter was a bell-jar counter with a compartment sealed next to the mica window so that the gases flowing by could be monitored continu- ously. The counter was calibrated for self-adsorption and a coincidence correction applied. The dead space of the system was measured with * Zweibel, Miller and Turkevich, J .Amer. Chem. Soc., 1948, 71, 376. Ilf3 54 STABLE AND UNSTABLE ISOTOPES helium for various reaction vessel temperatures. By making appropriate measurements one was able to determine the specific activity either of the carbon dioxide or of the carbon monoxide. The experimental study revealed two sets of phenomena, At relatively low temperature of 500' C there was no change of either pressure or radioactivity with time. This was interpreted to indicate that there is no exchange between the carbon dioxide in the gas phase and either the carbon on the surface or any adsorbed oxygen-carbon compound present on the surface. Condensation of the carbon dioxide showed several per cent non-condensable gas whose specific gravity was never less than that of the original carbon dioxide.This suggests the reaction CO, + charcoal = CO + oxygenated charcoal with the carbon dioxide molecules approaching the surface, depositing their oxygen on the surface and the product CO molecules containing the same carbon atom that was originally present in the reactant CO, mole- cules. This point of view has been suggested recently on the basis of kinetic and theoretical considerations by Gadsby and co-workers. At higher temperatures of 735-848" C, over 95 yo of the carbon dioxide was converted into carbon monoxide within at most four minutes from the start of the reaction, with no increase in pressure nor change in the specific radioactivity. A full analysis of the kinetics will appear in a forthcoming p~blication.~ Suffice i t to indicate that the changes in pressure cor- relate satisfactorily with the changes in the specific radioactivity.The mechanism that is proposed consists of a very rapid reduction of the carbon dioxide to a radioactive carbon monoxide with the formation of an oxygenated surface, followed by a slow decomposition of the latter to give the second molecule of carbon dioxide, which is now non-radioactive. An Example of the Use of 3H.-Another study of the use of radio- active tracers in heterogeneous reaction was carried out by Smith and Turkevich on the correlaion of the double bond migration in the butene-I+ butene-z system with the exchange of tritium from the catalyst to these hydrocarbon molecules. The detection of tritium is a bit troublesome because of the low energy of the 0.0145 MEV pray of this tracer nucleus.One is forced to introduce the tritium compound with argon into a counter and essentially make a count for each point of the rate curve, The results of Smith and Turkevich established that the catalysts for the migration of the double bond are also materials which readily exchange their hydrogen with the butene hydrogen, and that the rates of the two processes are comparable. Furthermore, on the basis of these experi- ments the catalyst was formulated as a configuration of atoms (in a mole- cule or on a surface of a solid) which contains a hydrogen donor and a hydrogen acceptor separated by about 3-5 A. Examples of substances which satisfy these criteria are H,S04, H ,PO,, aluminium silicate, copper pyrophosphate and many metal catalysts partially loaded with hydrogen.Such a " hydrogen switch " mechanism, whereby the catalyst gives one hydrogen to the substrate while it abstracts another, may be applicable to alkylation, polymerization, and cracking of hydrocarbons. Verifica- tion of such a mechanism must, however, await the development of a technique of identifying the distribution of deuterium atoms within and among complex hydrocarbon molecules. An Example of the Use of 2H.-At present Schissler, Irsa and Turkevich are studying the interaction of ethylene and deuterium on a nickel wire using an apparatus and technique similar to that of Twigg and RideaL5 Whereas the previous workers determined the extent of addition from the pressure drop and the extent of exchange from the * Gadsby, Long, Sleightholm and Sykes, Proc.Roy. SOC. A , 1948, 193, 35.7. Bonner and Turkevich, J . Amer. Cham. SOC. (forthcoming publication). 4 Turkevich and Smith, J . Chem. Physics, 1948, 16, 466. 5 Twigg and Rideal, PYOC. Roy. SOC. A , 1939, 171, 55.TURKEVICH, BONNER, SCHISSLER AND IRSA 355 thermal conductivity analysis of the hydrogen gas, we have analyzed the hydrocarbons in the General Electric Company mass spectrometer of the Brookhaven National Laboratory. To facilitate the interpretation of the mass spectra of the resulting deutero compounds, the analysis was made on the original ethylene and ethane portion and also on an ethane portion obtained by removing the ethylene from the former by FIG. 2.-Behaviour of ethylenes during the course of reaction between ethylene and deuterium: (I) C2H4-X (11) C2H3D-+ (111) C2H2D2-0 (IV) C,HD,-@ (V) C2D4-@.Values at end of the reaction for all isotopic ethylene molecules 0. treatment with bromine vapour followed by distillation. The mass spectra of the deutero ethylenes was obtained by difference. In order to translate these mass spectra from the positive ion currents of the various masses corresponding to the various deuterated ethanes (ethyl- enes) and their decomposition products on electron impact, one must know the mass spectra of the pure isotopic compounds C,D,, C2D5H, . . . C2H6 (CZD4, CZD,H, . , . C,H4). Since such mass spectra are known only for one deuterated ethane C2H5D, a set of patterns for various iso- topic ethanes and ethylenes was calculated from the patterns of the pro- tium compounds assuming that the ease of removal of a hydrogen and a deuterium are the same and that the variations in the pattern of the deuterium substituted compounds are governed solely by probability considerations.It should be pointed out that the calculated pattern agrees with the pattern observed for C2H,D in all masses but mass 29, where the calculated ion current is less than the experimentally observed one. This mass, however, is notoriously sensitive to impurities of higher hydrocarbons in the sample and the experimental value may be high for that reason, Further assurance of the validity of such model patterns is the fact that, by successively subtracting the model patterns of the C,D,, .. . C,H, (C2D4 . . . C2H4), a small residue is left for the ion current in the mass 29-24 region. Finally, the validity of the results obtained3 56 STABLE AND UNSTABLE ISOTOPES by such a treatment of the mass spectra of the product is strengthened by the material balance on the isotopic species. Thus, in a typical case, 39-6 mm. hydrogen atoms as ethylene and 37'7 mm. deuterium atoms as deuterium gas were introduced into the reaction vessel. The product gas of hydrogen, deuterium, the various deuterated ethylenes and ethanes gave on mass spectrographic analysis and resolution of the data, 38-6 mm. hydrogen atoms and 38.8 mm. deuterium atoms, FIG. 3.-The formation of ethanes during the course of the reaction between ethylene and deuterium. ('1 c2H6- (11) CzH5D-+ (111) C,H,D2-0 (V) C2H2D4--8 (VI) CzHD5-Q (IV) c,H,D,--o (VII) CzD6-U.The results obtained for the interaction of ethylene with two volumes of deuterium on a nickel wire at goo C are presented in Fig. 2 and 3. It is seen that the exchange reaction proceeds more rapidly than the addi- tion reaction, that the concentration of light ethylene decreases exponenti- ally, while that of the substituted ethylenes goes through a maximum before it finally reaches zero. The surprising result is the discovery of completely light ethane from the reaction of ethylene and deuterium. The possibility of the light hydrogen coming from residues left on the catalyst by previous runs is minimized by the pre-treatment of the catalyst overnight with deuterium a t 300° C, and by the excellent material balances obtained in each run on the hydrogen and the deuterium. These results suggest that the ethylene hydrogenation is effected by the hydrogen that was previously on another ethylene molecule, and only indirectly by the hydrogen molecule. This point is being further investigated and the mechanism formulation is withheld until the kinetic investiga- tions, which are being vigorously pursued, are completed. We wish to express our appreciation to Dr. Richard Dodson of the Brookhaven National Laboratory for his interest in these problems. Work presented has been carried out in part under the auspices of the U.S. Atomic Energy Commission. Chemistry Department, Chemistry Department, Brookhaven National Laboratory, Princeton University, Upton, Princeton, Long Island, New Jersey, New York, U.S.A. U.S.A.

ISSN:0366-9033    

DOI:10.1039/DF9500800352

出版商:RSC    

年代:1950

数据来源: RSC

摘要:

GENERAL DISCUSSION 357 GENERAL DISCUSSION Mr. J. A. Allen (Bristol) said : It is customary' in practice to distinguish between chemisorption and physical adsorption in the first layer of an adsorbed gas by measurements of heat of adsorption. In recent work we have found interesting effects on the electrical conductivity of evapor- ated copper films arising from the adsorption of various gases at - 183" C ; these are relevant to this Discussion since they appear to provide a useful alternative method of distinguishing between the two types of adsorption. The films were deposited on a glass substrate at - 183" C, annealed at 100" C and then cooled to -183" C, being maintained during this cycle under a pressure of 7 x 10-* mm. Hg. In this way it was possible to prepare reproducible clean films which were free of spontaneous re- sistance changes, which possessed reversible resistance-temperature be- haviour below roo" C and which had positive tempxature coefficients.Films of a nominal thickness of 400 A (calculated from the estimated amount of copper in the film, assuming a density of 8.92 g. ~ m . - ~ ) which had a nominal resistivity at - 183" C of about 13 times that of bulk copper were generally used. It was found that the admission of oxygen, carbon monoxide, and nitric oxide to such films at - 183" C caused an immediate increase in FIG. I . resistance which rose to a limiting value in a few minutes. Nitrogen and hydrogen had no measurable effect while ethylene produced a small initial increase followed by a gradual rise. The magnitude of these effects are shown in Fig.I, where R and R' are respectively the resistances of the film at -r83" C before and after the gas was admitted. The limiting value for the resistance change was reached if the pressure of the gas was Z I O - ~ mm. Hg; a subsequent reduction in pressure to I O - ~ mm. Hg did not appreciably affect the increase in resistance. Further, the admission of hydrogen after oxygen, carbon monoxide or ethylene did not affect the increases in resistance which had already taken place and the presence of nitrogen or hydrogen before oxygen or carbon monoxide were admitted only slightly decreased the rate at which the increases in resistance took place. The presence of argon at a pressure of I O - ~ mm. Hg before the ethylene was admitted prevented the slow rise in resistance shown by the second part of the ethylene curve.358 GENERAL DISCUSSION Although it is difficult to account for the magnitude of the resistance changes observed, the gases which cause marked increases in resistance are those which would be expected to be chemisorbed, this involving definite band formation with the metal.Nitrogen, hydrogen and argon which do not cause any increase in resistance are only physically adsorbed on copper at - 1 8 3 ~ C. A full account of this work together with the reasons which lead to this conclusion will be published elsewhere, but it seems worthwhile to emphasize the differences in behaviour between ethylene and hydrogen which may be useful in formulating the mechanism of the hydrogenation of ethylene. Mr.A. S . Porter (Imperial College, London) (partly communicated) : The need for accurate corrections for thermomolecular flow in adsorption work at low pressures is not always recognized. Neglect of these pre- cautions, when an adsorbent is at a temperature different from that of the pressme gauge may lead to significant errors in calculating (a) the equilibrium pressure over an adsorbent; and (b) the amount of gas re- maining in the gas phase and, therefore, of the amount adsorbed. Consider a tube of diameter d cm. connecting two parts of a system at absolute temperatures I', and T,, where the pressures of a gas at equi- librium are respectively p1 and p,. Knudsen 1 has shown that, in general, p, 9 p a , unless h, the mean free path of the gas molecules, is negligibly small compared with d.Except when 10 > d / h > I, the relationships between p1 and p, are given by eqn. (I) and (3) 1 below. When d / h > 10, where C is Sutherland's constant, qo and p o are respectively the viscosity and density of the gas at oo C and 1p0 bar pressure, K , and k , are constants, and p is measured in p bars. Eqn. (I) may be integrated, neglecting as did Knudsen, the variation of the term (I + C / T ) with temperature to give m + 1 - Tl + Pl m - I where m = (I + 0.178 k,k,)k, and Here (I + C/T)' is the mean value of I + C / T over the temperature When d/A < I, range T, to T , ; and p 1 and p , are in mm. Hg. I dT - (3) -. dP - 3k p 8'c+7h'T' * ' where k varies only very slowly with pressure and may be equated to 4 / 3 over this range.Integration of eqn. (3) gives explicitly as d273 where b I= --, A. = the mean free path of the gas molecules at N.T.P., A0760 and p1 and 6% are again in mm. Hg. Knudsen, Ann. Physik, 19ro,31,205. Knudsen, ibid., 1g10,33,1435.GENERAL DISCUSSION 359 P J P , = ( W 7 - 2 P - (5) When d/A + I integration of eqn. (3) gives Fig. 2 shows the application of these equations to hydrogen, where d = 1-18 cm., Tl = 78" K and T , = 298" K. I t is reasonable to assume that eqn. (4) becomes progressively less accurate as d / A increases in the range 10 > d/h > I, and the dotted curve, linking the two calculated curves, is drawn on this assumption. The experimental points show that eqn. ( 2 ) and (4) and the above assumption are satisfactory for practical purposes.It may be noted that Dushman's suggestion to use eqn. (5) when d/h = 0.1 introduces an error of approx. 8 yo with the values of T , and T , used here. FIG. ~.--Thermomolecular flow of hydrogen in 1.18 cm. diameter tube, where T , = 78K and T2 = 298K. The effect of thermomolecular flow on (a) above is apparent from the graph, but the significance for (b) depends on the ratio of the quantity. of gas adsorbed to that remaining in the gas phase. This is illustrated by reference to Allen and Mitchell's results at - 183" C. Neglecting, for the present, the breakable gas seal, their results may be re-calculated as follows. Mr. Allen gave the final pressure as I O - ~ mm. ; presumably this is the pressure in the bulb, calculated by eqn. ( 5 ) from a Pirani gauge reading of 1.80 x I O - ~ mm. For oxygen ( A , = IO-j cm.), eqn.(4) would give the pressure in the bulb as 1-57 x I O - ~ mm. when T , = goo K, T? = 300° K and d, from their diagram, is I cm. Calculated in this manner, column 3 of their Table I would read 0.3, 0.5, 0-7, 1.0, 1.6, 7-3, 9-9, 9-2, 14.2, 21.4, ( x 1015 molecules/cm.2) in place of. their values of 1.1, 1.3, 1.5, 1.8, 2-4, 8.1, 10.7, 10.0, 15.0, 22.2 ( x 1 0 1 ~ molecules/cm. *) . While their ingenious use of the narrow gas seal undoubtedly reduces the errors due to the use of eqn. (5), it cannot be regarded as entirely satisfactory, owing to its short length (1-1-5 cm., from the diagram) and the fact that only one end is attached to the outer glass tube. Both these facts make it unlikely that the upper end of the narrow tube is near to room temperature, which is necessary if it is to be effective.Experimental calibrations might overcome the difficulty, but the precise level of the liquid oxygen bath, and the point of breakage of the seal would then be highly critical. It is evident that the significance of thermomolecular flow must be assessed experimentally, or by detailed calculation, for any particular case. The above considerations show the desirability of using tubing 3 Dushman, Scienti3c Foundations of Vacuum Technique (Wiley, New York, 1949). P. 65.360 GENERAL DISCUSSION of uniform diameter throughout a temperature gradient, so that the prob- lem is amenable to accurate calculation. While the initial calculations are laborious, the graphical presentation shown in Fig.z permits the necessary corrections to be read off rapidly and accurately. Mr. J. A. Allen and Dr. J. W. Mitchell (Bristol) (communicated) : With regard to Mr. Porter's criticism of our use of the simple Knudsen correction for the thermomolecular flow, we did in fact carry out calibra- tion experiments with blank bulbs. These experiments showed that the simple correction was strictly applicable with our apparatus for oxygen at temperatures down to - 78°C. At - 183°C deviations from the simple correction value were observed and the results given at this temperature may be too large by 0.3 x 1015 molecules per cm., for pressures in the neighbourhood of I O - ~ mm. Hg. The method used for the calibration did not allow one to distinguish between the error associ- ated with the Knudsen correction itself and the increase in the amount of oxygen adsorbed on the glass surface which might occur in the tem- perature range between - 78" C and - 183" C.Similar calibration measurements carried out with hydrogen at these pressures showed that the possible error at - 183" C was not greater than 0.1 x I O ' ~ molecules per cm.2 which suggests that there may be an increased adsorption of oxygen on a glass surface when the temperature is reduced from - 78" C to - 183°C. It may be pointed out that even if the results at - 183" C are too lsrge by 0.3 x 1015 molecules per crn.,, the interpretation is not seriously affected because this value is of the same order as the spread of the results between different experiments made under the same initial conditions.We should, in conclusion, like to thank Mr. Porter for his valuable contribution to the discussion on the paper. Dr. B. M. W. Trapnell (Royal Institution, London) said : Mr. Allen and Dr. Mitchell have suggested that no oxidation of their copper films takes place between - 183'C and about room temperatures. Their main evidence for this is that the quantity of gas taken up at " equili- brium " does not vary much in this range, and that this quantity cor- responds to that expected for a chemisorbed monolayer on (111) faces, assuming a film roughness factor of unity. Now one would expect the chemisorption of CO, H, and N, to be nearly complete at - 183°C and at small pressures, and therefore the ratio of CO, H,, N, and 0, uptakes to be 2/1/1/1.Allen and Mitchell actually find about 2/1/1/4. The first is that 4 oxide layers are formed at these low temperatures, and that the film roughness factor is less than one and about 1/4, which is only understandable for a continuous surface if it is partly covered with initial contamination. The second is that CO, H, and N, only have one-quarter their expected covering power, which is hardly reasonable. It is better to conclude that low-temperature oxidation is taking place, and that Allen and Mitchell have only obtained a partially clean surface. This conclusion is supported by the work of Brunauer and Emmett, who found that 7 or 8 oxide layers are formed on iron with great rapidity at liquid-air temperatures, by the work of Beeck, Smith and Wheeler, who found similar phenomena, with nickel, and by my own work on tungsten, where again some low-temperature oxidation takes place.Mr. J. A. Allen and Dr. J. W. Mitchell (Bristol) (communicated) : We have stated that copper films do not oxidize below 240" K and not below room temperature as has been quoted by Dr. Trapnell. In our opinion, based on measurements of resistance changes accompanying adsorption on thin films, nitrogen and hydrogen are physically adsorbed on copper at low temperatures, and at the pressures used, the formation of complete physically adsorbed layers is unlikely. The number of molecules of oxygen and carbon monoxide taken up per cm., below 240" K is approximately the same for both gases as is shown in Fig. 3 This leaves us with two alternatives.GENERAL DISCUSSION 361 in which the number of molecules taken up per cm.2 is plotted against the temperature.Trapnell has assumed, without justification, that oxygen is chemi- sorbed as atoms and that one atom of oxygen and one molecule of carbon monoxide can each occupy one site on the surface. He has also argued that the surfaces of our copper films are initially contaminated to the m 0 X N 5 Y a n 5 X w " Y - - 8 Temperature *K FIG. 3. extent of at least three quarters of their area. We do not accept these arguments for the following reasons. (i) Work in this laboratory over a period of years has established a technique whereby beads of copper on tungsten filaments can be thoroughly outgassed. This has been checked by very sensitive contact potential measurements and the tech- nique used in outgassing the beads used for the present work has been proved to be entirely satisfactory.(ii) Reproducible results are unlikely to be obtained if contaminated films are used for the measurements. (iii) The technique used for depositing the films ensures beyond all reason- able doubt that, provided the beads are thoroughly outgassed, the free surfaces of the films will be uncontaminated. As far as we are aware, there is no evidence whatever to support the contention that oxygen is dissociated on copper below 240° K. If this were the case, it seems that oxidation should certainly' proceed, the oxide film growing very rapidly to approximately 20 in thickness. Our own work with aluminium has shown that, with this metal, oxidation does proceed in this manner even at - 183" C and the comparison of the behaviour of copper with that of aluminium leaves little doubt that an oxide layer is not formed on copper below 240" K.Fig. 3 shows that this is the critical temperature above which oxidation proceeds. It is suggested, and this may be supported by theoretical arguments, that oxidation of copper does not proceed below 240" K because the oxygen molecules are not dissociated on copper below that temperature. The equality' between the number of molecules of oxygen and of carbon monoxide adsorbed below that temperature is most simply accounted for by assuming that one molecule of oxygen and one molecule of carbon monoxide occupy the same number of adsorption sites on the surface.362 GENERAL DISCUSSION Mr.P. F. Tiley (Bristol) said : With reference to the adsorption of oxygen on metals, I wish to mention some results obtained recently at Bristol on the uptake of oxygen by reduced copper at 20' C. Successive admissions of oxygen were made to a sample of copper powder reduced with hydrogen at 150'C and then evacuated €or 16 hr. at I ~ o ' C . For each admission the integral heats of adsorption were measured over a period of 12 min. in the calorimeter previously described.4 The first two admissions gave a heat value of 108 kcal./mole, falling on the third admission to 83 kcal., and remaining constant (&- 3 kcal.) at this value until the 8th admission, when it dropped in two further admissions to 56 kcal. These figures should be compared with the heat of formation of Cu,O = 81.6 kcal./mole of 025 and heat of adsorption of 0, on Cu,O = 55 kcal./mole of O,.4 For the first 7 admissions the rate of uptake of oxygen was very rapid, and appeared to be controlled only by the rate of diffusion of the gas into the calorimeter, but at the 8th admission the rate showed a marked decrease.The total amount of oxygen taken up over the first 8 admissions corresponded to a thickness of cuprous oxide of approx. 15 A. The results largely confirm the work of Allen and Mitchell and, in addi- tion, suggest the rapid formation at room temperature of a normal Cu,O phase. It is felt that the initial high values of > IOO kcal./mole (which cannot be accepted without reservation until the work has been repeated) are relevant to the observations of Prof.Taylor on the posioning of active metal catalysts by small traces of oxygen. I should also like to draw attention to the fact that the heat of ad- sorption of oxygen on tungsten of 140 kcal.6 is very close to the heat of formation of WO,, which is approx. 130 kcal./mole of O,.' It is open to question whether the observed phenomena on the uptake of oxygen by a clean tungsten surface may not be interpreted in terms of the forma- tion of a thin oxide film due to the rapid mutual penetration of the surface metal atoms and oxygen, followed by adsorption of oxygen on this oxide. Dr. J. S . Anderson (Harwell) said : The results mentioned by Mr. Tiley (that the initial heat of reaction of oxygen with a copper film is greater than corresponds to the formation of Cu,O) suggest that we must always bear in mind that to get significant thermochemical measurements we must start from a properly defined state and finish up with a properly defined state thermodynamically.Neither condition is entirely fulfilled in reactions occurring on evaporated films. Until measurements have been made, for example, of the heat of solution of such films, their content of stored energy is quite uncertain. The formation of highly mobile films, when condensation takes place a t low temperatures, shows that a considerable degree of metastability may be involved. On metals such as copper or iron, the nature of the oxide formed is quite undefined. Thus ferrous oxide is thermodynamically unstable below about 600' C, and even within its existence range is not stable with the composition FeO : the wiistite phase has a range of composition from about FeO,.,, to FeO,.,,.Dr. Beeck has referred to the formation of oxide films about four layers deep by reaction of oxygen with evapor- ated films at low temperatures. The composition of such films (if essenti- ally FeO) could be inferred from the cell dimensions, as determinable by electron diffraction measurements, for example. It would be desirable for this sort of analysis to be done before attempting to draw too many theoretical conclusions from calorimetric and other data. Cuprous oxide is likewise a phase of potentially variable composition, and the composi- tion of the surface oxide on a copper film may or may not be Cu,O,.OOO.Garner, Gray and Stone, this Discussion. Randall, Nielsen and West, Ind. Eng. Chem., 1931, 23, 388. Roberts, Proc. Roy. SOC. A , 1935, 152, 473. Moore and Parr, J . Amer. Chem. Soc., 1924, 46, 2656 ; Int. Crit. Tables, 5, 192.GENERAL DISCUSSION 363 Both the heat of formation and the diffusion coefficient involved in the formation of “ polymolecular ” oxide films would be sensitive to the composition. Dr. D. D. Eley (Bristol) (communicated) : The observations in the paper of Allen and Mitchell (and also Beeck, et d8 on the stabilization of copper films by chemisorbed hydrogen prompt me to mention the results of some approximate calc~lations.~ It is well known that evapor- ated films show a degree of mobility of atoms at low temperatures even, e .g . Appleyard,10 Picard and Duffenbach. l1 To a first approximation, we might treat the film as made up of pure liquid drops with a surface tension of ca. 1000 dynes/cm. The surface tension will, of course, be the driving force behind aggregation and if a chemisorbed gas tends to lower the surface tension, it will tend to prevent sintering of the film. For hydrogen on tungsten, we know the heat of adsorption of hydrogen from 8 (surface covered) = o to I . We may accurately enough calculate the entropy of adsorption for any gas pressure. Thus, we obtain the result that in the presence of I O - ~ mm. pressure of H, gas the surface tension of a W film is lowered 580 dynes/cm, which may well suffice for the purpose of stabil- ization. Mr. P. R. Rowland (Guy’s Hospital, London) said : With reference t o the evaporated copper films of Dr.Allen and Dr. Mitchell I would like to ask them if they consider that there may’ be faces other than (111) present in films sintered at room temperature ? Stranski and Mahl l2 have carried out some calculations which indicate that in the equilibrium crystal forms of face centred cubic metals the faces which predominate are (111) and (001) if we take into account interactions between nearest neighbours only. (If interactions between more distant neighbours are included, the calculations indicate that more vicinal faces will also appear.) Dr. Trapnell has suggested that some molecules may cover more than one adsorption site each. If we consider cases where the effective rzdius of each adsorbed particle is greater than that of the metal atoms the number of sites covered per adsorbed particle depends on the geometry’ of the face concerned.l3 Mr. J. A. Allen and Dr. J. W. Mitchell (Bristol) (communicated) : We have no evidence as to which crystallographic planes are exposed in our evaporated copper films. The ( I I I ) face was used for calculating the number of adsorption sites simply because it is the face of minimum free energy for the close-packed lattice ; the value deduced gives the order of magnitude of the maximum number of adsorption sites on a plane copper surf ace. Prof. J. H. de Boer (Geleen and Delft) said : Dr. Mitchell has given a most excellent review of the various possibilities of obtaining semi-conductors. May I add two more systems ? There is, in the first place, the substance Fe,O, where the presence of ferrous and ferric ions in equivalent lattice positions causes an excellent semi-conductivity. Secondly, I would like to draw attention to the fact that very thin films of metals, obtained by sublimation in vacuo, do not conduct electricity in a purely metallic way.The temperature coefficient of the conductivity is abnormal and it is often found that the thin film behaves formally as a semi-conductor. It was on such very thin molybdenum films that about 12 years ago Dr. H. H. Kraak and I could demonstrate the van der Waals’ adsorption of oxygen at low temperatures and chemisorption at higher temperatures. The chemisorption resulted in a large decrease of conductivity. It demonstrated that thin films of metals which are properly annealed can * Reeck, Smith and Wheeler, Proc.Roy. SOC. A , 1940, 177, 62. lo Appleyard, Proc. Physic. Soc., 1937, 49, 118. l1 Picard and Duffenbach, J . AppZ. Physics, 1943, 14, 29. l2 Stranski and Mahl, 2. physzk. Chem. B, 1942, 51, 257, 319. l3 Stranski and Suhrmann, Ann. Physik, 1947, 4, 169. Eley, Nature, 1946, 158, 449.364 GENERAL DISCUSS ION serve as models for studying chemisorption in a similar way as the semi- conductors which are under discussion. Dr. D. D. Eley (Bristol) said : Dr. de B ~ e r has suggested that very thin metal films are semi-conductors, but I wonder if this may not be misleading. Is it not a question of electrons having to pass from one isol- ated crystallite to another, across gaps or along the glass ? This jumping process will be activated, and profoundly influenced by adsorbed gases.The effects described by Twigg l4 on silver deposited on glass wool ale probably of this type, and possibly also the effects of gases on the con- ductivity of films as described earlier in this Discussion by Allen and Mitchell . Prof. J. H. De Boer (GeEeen and Delfi) (communicated) : The view expressed by Dr. Eley is often found in literature, but does not, in my opinion, explain the behaviour of well-sintered thin metallic films. I would like to draw his attention to our work published in 1936 and 1937.1~ Gases, when physically adsorbed, give a slight increase of conductivity, when chemisorbed a larger decrease of conductivity may be found, provided the films have been carefully heated to a higher temperature previously.Films of metals like silver on glass show, especially in the presence of oxygen, disturbing (re)crystallization phenomena. Dr. T. J. Gray (Rristol) said : In catalysis it is the interaction of gases with the surface region of the solid which is of major significance. In the course of catalytic processes, defects are created and destroyed. It follows, therefore, that in developing a technique foi- a study of cata- lysis by measurements of semi-conductivity those systems should be selected for which there is the optimum change in conductivity during the adsorption and desorption of gases commensurate with stability. At working temperatures for most oxide catalysts the surface zones are of greater significance than the bulk solid both for catalysis and conducivity so that the use of ‘‘ stabilized ” semi-conductors of the TiO, varieties would appear to be excluded as insensitive for the required measurements. To be satisfactory the system selected must show reproducible and reversible changes of semi-conductivity during the adsorption and de- sorption of various gases.Long-term stability is essential and the various activation energies determined from the kinetic and conductivity relation- ships must be reproducible between samples although the absolute activity of the samples may vary considerably. These conditions have been found to hold with the copper and manganese oxide systems but with zinc oxide the kinetic study shows some variation with history. Considering the postulated model for the adsorption of oxygen we have 0 2* 11 Site (A) 0, + 0 +- Oxide Oxygen is initially adsorbed at site A, dissociates and diffuses as atoms over the surfaces.At site B oxygen atoms interact with the surface giving rise to defects which modify the semi-conductivity. The tempelahre coefficient of conductivity leads to thz activation energy for the formation of a current carrier in the process (B). The activation energy derived from the rate of change of conductivity during oxygen adsorption at varying temperatures and pressures leads to the overall activation energy embracing adsorption, dissociation, diffusion and formation of current carrier. The rate of uptake of oxygen gives the activation energy for the initial adsorption process at site A either separately or including the dis- sociation process.The very large increase in rate of change of conduc- tivity by adsorbing atomic oxygen has indicated the relative significance l4 Twigg, Trans. Faraday SOC., 1946, 42, 657. 16Rec. trav. chzm., 1936, 55, 941, and 1937, 56, 1103.GENERAL DISCUSSION 365 of this stage while the non-equivalence of the two oxygen atoms from each molecule suggests that one enters into solution in the surface region without affecting conductivity. Apart from the question of the part played by defects in the catalytic process, it is clear that semi-conductivity measurements afford a satis- factory method of studying the kinetics of catalytic processes. Dr. A. F. H. Ward (Manchester) (communicated) : In connection with the measurement of heat of adsorption with small amounts of materials, it may be useful to draw attention to the microcalorimeter which I con- structed for this purpose.ls This functioned as an adiabatic calorimeter, not by varying the temperature of an adiabatic shield, but by removing a quantity of heat from the calorimeter vessel equal to the heat of ad- sorption. The removal of heat was effected by passing a small current through a series of thermocouples, with one set of junctions in the calori- meter, and producing a cooling by the Peltier effect. This current was adjusted in such a way that the temperature of the calorimeter remained constant, as indicated by another series of thermocouples. A heat evolu- tion of 0.0005 cal. could be detected and heat cou’d be extracted at rates up to about 0 - 2 cal./min. These figures could be made greater or less by suitable modifications to the electrical circuits. Dr. B. M. W. Trapnell (Royal Institution, London) said : Chemiscrption of cthylene takes place with dissociation of hydrogen and formation of a complex MC2H2. Deuteration of ethylene, which must proceed through deuteration of the acetylenic complex, will therefore only yield C2H,D2 and C,D4 unless intermolecular exchanges between complexes take place in the adsorbed layer. Now Turkevich, Bonner, Schissler and Irsa assume that deuteration of ethylene is governed solely by probability considerations, and that mono-, di-, tri- and tetra-substituted bodies are all possible products. The analysis of data on this basis may well give misleading conclusions. 16Ward, Proc. Camb. Phil. SOC., 1930, 26, 278.

ISSN:0366-9033    

DOI:10.1039/DF9500800357

出版商:RSC    

年代:1950

数据来源: RSC

摘要:

AUTHOR INDEX * Allen, J. A., 309, 357, 360, 363. Anderson, J. S., 238, 305, 362. Baldock, G. R., 27. Barrer, R. M., 88, 206. Beeck, O., 93, 118, 159, 193, 314. Bevan, J. M., 238. Blekkingh, J. J. A., 200. Boer, J . H. de, 93, 206, 208, 363, 364. Bonner, Francis, 352. Bradley, R. S., 94. Bremner, J . G. M., 79, 92. Bruijn, H. de, 69, 95, 300. Chapman, P. R., 258. Colburn, C. B., 39. Cole, W. A., 314. Coulson, C. A., 27. Couper, A., 172. Denbigh, K. G., 83. Dixon, J. K., 290, 305. Dowden, D. A., 184, 203, 204, 206, 208, 210, 296, 305. Eggleton, A. E. J., 92, 195. Eley, D. D., 34, 99, 172, 191, 199, 205, 302, 363, 364. Eucken, A., 128. Evans, Alwyn G., 302. Evans, U. R., 296. Everett, D. H., 86. Eyring, Henry, 39. Garner, W. E., 194, 211, 246, 298. Given, P. H., 301. Goodeve, Charles, 192.Gray, T. J., 246,250,331, 364. Griffith, R. H., 258, 299. Guinier, A., 344. Halsey, Jr., J. E., 54, 88. Hillier, James, 348. Huang, K., 18. Hubbell, Harry H., 348. Huttig, G. F., 215. Irsa, Peter, 352. Johnson, Marvin, F. L., 303. Kemball, C., 94. Kropa, E. L., 290. Laidler, Keith J., 47, 90. Lindars, P. R., 258. Los, J. M., 321. Luft, N. W., 306. Lyon, Lorraine, 222. Maxted, E. B., 135. May, D. R., 290. Melik, John S., 303. Mignolet, J. C. P., 105, 326. Miller, A. R., 57, 80, 87. Milliken, Jr., T. H., 279. Mills, G. A., 279. Mitchell, J. W., 307, 309, 360, 363. Moon, K. L., 135. Morrison, J. A., 321. Oblad, A. J., 279, 302. Overgage, E., 135. Porter, A. S., 203, 358. Reynolds, P. W., 184. Rideal, Eric K., 96, 114. Ries, Jr., Herman E., 303. Ritchie, A. W., 159. Rowland, P. R., 196, 209, 363. Saunders, K. W., 290. Savage, S. D., 250. Schissler , Donald, 352. Schroyer, F. K., 337. Schuit, G. C. A., 205, 299. Schwab, G.-M., 79, 88, 89, 91, 166, Selwood, P. W., 222, 306, 337. Steiner, H., 264. Stone, F. S., 194, 246,254. Sykes, K. W., 82. Tamele, M. W., 270. Taylor, H. S., 9. Thomas, N., 80, 82. Tiley, P. F., 201, 202, 254, 362. Tompkins, F. C., 85, 92, 201, 202, 203. Trapnell, B. M. W., 114, 191, 193, Turkevich, John, 348,352. Twigg, G. H., 89, 90, 152. Ubbelohde, A. R., 203, 204. Uri, N., 207. Ward, A. F. H., 95. 365. Weiss, J., 302. Wheeler, A., 314. Wicke, E., 199. Winter, E. R. S., 231, 300. Wright, P. A., 194. Wyllie, G., 18, 82, 91. Young, D. M., 84, 85. Zawadzki, J., 140. Zwietering, P., 196. Zwolinski, Bruno J., 39. 198, 205, 207, 298. 360, 365. * The references in heavy type indicate papers submitted for discussion.

ISSN:0366-9033    

DOI:10.1039/DF9500800366

出版商:RSC    

年代:1950

数据来源: RSC