摘要:

VACUUM TECHNIQUE Techniques of Physical Measurement B. G. Gowenlock, Ph.D., D.Sc., F.R.I.C. and C. A. F. Johnson, B.Sc., Ph.D. Department of Chemistry, Heriot- W a t t University, Edinburgh I Pressure measurement Liquid manometers, 1 12 The McLeod gauge, 115 Mechanical manometers, 1 17 The viscosity gauge, 119 The Pirani gauge, 120 Ionization gauges, 123 Vacuum pumps . . .. . . .. . . * . .. . . 108 . . . . . . . . . . .. . . 112 Construction of vacuum apparatus: standard devices . . . . . . 124 Circulation of gases in vacuum systems . . . . . . . . . . 130 Ultra-high vacua .. . . .. .. . . .. .. 132 Possible developments . . . . .. . . .. . . . . 133 Vacuum techniques are common in research laboratories and it can be claimed that no student should now be allowed to complete an honours degree course in chemistry without becoming conversant with the basic details of vacuum systems and the associated techniques of the manipulation of gases and volatile materials. The manipulation and investigation of gases has been of importance in the development of the science of chemistry.Scheele collected gases in bladders, Priestley used the pneumatic trough, Bunsen applied spectroscopic techniques to gas analysis and Rayleigh, Ramsay, Travers and their co-workers made use of the pumps of Sprengel and Topler for their researches on the Group 0 gases. Parallel developments in physics based upon the application of vacuum techniques included low-pressure discharge tubes, X-rays, the demonstration of the existence of the electron, the mass spectro- graph, thermionic valves, tungsten-filament lamps and the Dewar vacuum flask.Fifty years ago, it would have been unusual to find a wide range of chemists using vacuum techniques in a research laboratory, although the basic prin- ciples upon which modern vacuum technology is based were known and devices for establishing high vacua had been constructed. The majority of the techniques for pressure measurement had been developed and the requisite taps, waxes, greases and low temperature refrigerants could also be obtained. The rapid development of vacuum technology through its commercial and industrial importance led to an increased availability of equipment that was rapidly adapted to chemical purposes. Chemists of a variety of interests made use of and extended vacuum techniques. We note one example from the Gowenlock and Johnson 107 development of gas kinetics, where Bodenstein and Dux1 constructed a greaseless valve of platinum and platinum-iridium and a spiral silica gauge for pressure measurement in their investigation of the photochemically induced reaction of hydrogen and chlorine.Preparative inorganic chemistry required vacuum techniques in order to permit the preparation of gases that were reactive to air, water and also tap greases and manometer liquids. Vacuum techniques were employed in the isolation and study of the noble gases2 and extensions of each method to chemical synthesis are to be found in Moser3 and in the detailed account given by Stock4 of the grease-free high vacuum techniques developed for his pioneering studies of boranes and silanes.Organic chemists exploited the technique in the distillation of compounds under reduced pressure, particularly where distillation at higher pressures led to decomposition. The present uses of vacuum techniques in chemical laboratories are (a) preparative, (b) physico-chemical e.g. gas kinetic, spectroscopic, vapour pressure, adsorption measurements. The principles of the techniques are common to all the uses though the degree of vacuum required may vary considerably. VACUUM PUMPS The most frequently used method of producing a high vacuum is a rotary oil pump. Such pumps are used either alone or in conjunction with a mercury or oil vapour pump.The basic principle of the rotary oil pump is due to Gaede (1907). For the purposes of this review it is not necessary to discuss the mode of operation of these pumps. It should be noted that pumps are made to certain basic specifications which define (a) the ultimate pressure, (b) the pump displacement. The ultimate pressure of a single stage rotary oil pump in good working condition is of the order 10-2-5 x 10-3 torr (1.33-0.67 N m-2) provided that gas ballasting is not being used. For a two stage pump, in which two pumps are mounted in series with a common driving shaft, ultimate pressures of the order 10-3-10-4 torr are obtained. The speed of a pump is defined as the volume of gas pumped from the system in unit time at a specified pressure (usually the pump inlet).The pump displacement is quoted as the speed of the pump at atmospheric pressure. A typical example of pumping speed/pressure characteristics is given in Fig. 1. The size of pump necessary for a particular application can be obtained from the equation (1) where T is the pump down-time, V is the volume of the vessel, S is the pump speed and PI and PZ are the original and final pressures respectively. From Fig. 1 it is obvious that S is not constant over the operating pressure range of the pump and equation (1) may be replaced by: T t X S v ___ .. .. where t is the time read from Fig. 2 for the given pressure. It is thus a simple matter to calculate the time to reach a certain pressure in a given system with a RIC Reviews 108 80 .e 50 - ? W v) 20 I O - ~ Without ballast - ,/ stage pump 3 0 2 4 6 8 10 12 14 16 18 20 Time (min) pump of known speed S.It is obvious that the units for S, V and T must be consistent. It is also necessary to remember that the curve in Fig. 2 makes no allowance for the impedance of the connecting pipelines (in particular, tap dimensions), nor is there an allowance made for outgassing of the vessel and leakage into it. Pump speeds of commercially available rotary pumps lie within the range 20-8000 1 min-1. Most chemical laboratories where general vacuum facilities are required, use pumps in the 20-100 1 min-1 range. In general the faster the pump speed the greater is the cost of the pump.Gowenlock and Johnson I lo-' - - - - - - - - - - - - - 100 10 With ballast ---- I Fig. I (above). A typical example of pumping speed/pressure charac- teristics. Fig. 2 (left). A typical example of time/pressure characteristics. Pressure (torr) - - - - 2 109 Mention has already been made of gas ballasting of rotary oil pumps. This technique permits the addition of controlled quantities of air to the exhaust chamber of the pump and allows the sweeping out of vapours from the pump together with the permanent atmospheric gases. This technique allows contin- uous pumping of condensable vapours without deterioration of the oil and can also allow partial removal of gases that have dissolved in the oil.This makes it particularly useful for distillation under reduced pressure where there is often the risk of contamination of the pump oil. There is the disadvantage that the ultimate pressure obtainable is considerably increased. Gas ballast facilities are incorporated into the great majority of commercially available pumps. Alternatives to gas ballast are the provision of chemical or physical means of removing the vapours before reaching the pump. Chemical means are con- ventionally provided by a trap containing a tray of phosphorus pentoxide, whereas a cooled (liquid nitrogen or solid COz) trap constitutes the usual physical means. In practice, direct use of the oil pump is confined to some distillations under reduced pressure, and for all uses where there is need for a higher vacuum than 10-3 torr it is necessary to use the rotary oil pump as a backing pump for a mercury or oil diffusion pump.Both types of diffusion pumps are available commercially although it is common practice for mercury diffusion pumps to be constructed by a glassblower to a local approved specification (see e.g. Sanderson,5 Robertson et aZ.6). It is necessary to know the ‘cut-off’ pressure of diffusion pumps, i.e. the pressure above which the pumping action ceases. This varies with the internal dimensions of the pump and for a mercury pump generally lies in the range 40-0.3 torr, where oil pumps have lower cut- off pressures (0.2-0.5 torr). All diffusion pumps require cooling chambers where the stream of vapour meets the pump wall.Such cooling is provided by a small water condenser jacket incorporated into the pump design or by efficient air cooling in the case of some oil pumps. The oils generally used for diffusion pumps are either silicones or hydrocarbon oils. The higher initial cost of the silicones is offset by their longer life, a consequence of their higher thermal and oxidative stability. Mercury pumps are required where no contamination by organic vapours can be permitted. The ultimate vacua of well designed diffusion pumps is 10-6-10-7 torr. The pumping speed of vapour diffusion pumps rises with decreasing pressure to a maximum, provided that the backing vacuum and the pumping rate of the rotary oil pump are both adequate. For typical small laboratory pumps (maximum speed 100 1 s-l) the maximum pumping rate occurs at pressures of 10-3 torr and below.It is possible to improve pumpicg speeds in the range 10-1-10-3 torr by incorporating a vapour booster pump; this is relatively uncommon in most simple vacuum systems where the extra cost has to be balanced against other advantages gained. Figure 3 demonstrates a preferred arrangement of the basic units of com- bined systems of rotary oil and diffusion pumps. The use of thick pressure tubing at point A helps to render the main vacuum line vibration free. Tap TI permits the admission of air to the rotary oil pump after shut down. In the absence of a non-return valve failure to admit air would lead to ‘suck-back’ of oil into the vacuum system.Trap 1 may be replaced by a phosphorus pentoxide RIC Reviews 110 f ' 1 I Fig. 3. Preferred arrangement of basic units of combined systems of rotary oil and diffusion pumps. trap. Tap T2 provides a by-pass for the diffusion pump so that direct evacua- tion of the apparatus to torr is possible before bringing the diffusion pump into operation by means of taps T3 and T4 with TZ closed. Trap 2 prevents diffusion of oil or mercury vapour into the main vacuum line. The methods for producing high vacua outlined above are not the only techniques available but merely those which are in most frequent use. Other pumps which have more specialist applications are : (i) the Roots pump, a high speed mechanical pump which must have a suitable backing pump so that it may operate at maximum efficiency,7 namely at a pressure of about 5 x 10-2 torr for a compression ratio of 10; (ii) sorption pumps, where the pumping action is based upon the sorbing activity at liquid nitrogen temperatures of activated materials such as zeolites or charcoal ; (iii) ion pumps, based upon the pumping action of hot and cold cathode ion gauges; (iv) getter pumps, based upon the highly reactive character of finely dispersed clean surfaces which results in both chemical and physical removal of gases from an enclosed volume; (v) combinations of ion and getter pumps such as sputter-ion and evapor- ion pumps.General details of means of producing high vacua and their application to typical chemical purposes may be obtained from standard texts.8-l3 Go wenlock and Jolinson 111 PRESSURE MEASUREMENT In the operation of equipment designed to enable studies to be undertaken at reduced pressure, the choice of a suitable pressure-indicating instrument and the correct interpretation of its reading is of great importance.Of the instru- ments to be described here, the liquid and mechanical manometers give abso- lute values, in that they measure the pressure directly as a force at a solid or liquid surface. The sensitivities of the mechanical manometers remain constant for all types of gases and vapows. The other types of gauge to be described are indirect in that they depend on the measurement of some secondary phenomenon, such as thermal conductivity or ionization of the gas, which is itself pressure dependent.Thus the pressure indicated will be depen- dent on the composition of the gas and also on the temperature. These gauges must be calibrated against an absolute pressure measuring instrument, and the calibration will hold only for the gas used in the calibration procedure. Detailed accounts of the many instruments that have been used can be found in standard texts such as those of Leck14 and Dushman.lo Liquid manometers The pressure gauge which finds wide application in the range 1-760 torr is the simple glass U-tube containing mercury. The minimum pressure change that can be detected with this manometer is about 0.1 torr, although, unless special precautions are taken, the absolute accuracy may be much worse.The sensitivity of the U-tube manometer may be increased about 15-fold by replacing the mercury with a low vapour pressure oil such as di-n-butyl phthalate, Apiezon oil A or silicone diffusion pump oil. The decreased density of these oils enables pressure differences of the order of torr to be detected. The main sources of error in this instrument are: (i) liquid sticking to the glass giving a variable capillary depression or elevation ; (ii) irregular light refraction through the glass; (iii) for the oil manometer, a difference in oil composition and hence density in the two limbs; (iv) vapours may dissolve in the oil; (v) a difference in temperature and hence density in the two limbs.The errors due to (i) and (ii) may be greatly reduced by using wide-bore thin-walled tubing, although the high cost of mercury could be the main limitation of the tube diameter. When high accuracy is required the whole manometer is enclosed in a thermostat to limit temperature differences. A further disadvantage of the oil manometer is due to the long drainage times of the relatively viscous oils. After an appreciable change in pressure, both levels rise slowly over many minutes as the liquid drains from the freshly exposed wall. This can prevent pressure readings being taken even though the pressure is constant. If there is a pressure increase, the draining arm is on the high pressure side and the thin film becomes saturated with gas at the high pressure.When the pressure falls, this film produces a degassing layer at the meniscus of the liquid, which may give rise to spuriously high readings. RIC Reviews 112 Vacuum Gas pressure Oi I height Oil height Heater Fig. 4. McAllan’s modification of an oil manometer in which a cistern is added. McAllanl5 describes an oil manometer in which these effects are reduced, and which permits efficient degassing when required. The simple U-tube is modified by the addition of a cistern to the high pressure side of the mano- meter (Fig. 4). If the apparatus is constructed so that the cistern has a diameter 10 times greater than that of the arms, then the cistern has a surface area 50 times that of the combined surface of both arms in a conventional U-tube.Thus the major movement of oil occurs in the arm that is maintained under high vacuum. When there is a pressure increase there is about 1/5 of the usual area draining into a surface which is over 50 times the area of a normal U-tube. There is thus an improvement in drainage error of about 250 times. When there is a pressure decrease there is drainage from about twice the normal surface, but into the cistern, giving an improvement by a factor of about 25. Capillarity effects are avoided by measuring the levels in the narrow tubes and not in the cistern. Any outgassing that may occur will be into the large volume of the cistern, and so the effect of spuriously high readings is greatly reduced. The connexion from the vacuum arm to the far side of the cistern facilitates degassing when necessary.The 25 W heater wound on the vertical part of the connexion causes rapid convection, the heated liquid rising to the large surface of the cistern and degassing rapidly. Cold liquid is swept from the cistern and arms through the connecting tube to the heater. This method also eliminates the possibility of bubbles of gas forming in the narrow U-tube, as is common in the normal type of U-tube. Use of the heater enables oil saturated with air to be degassed completely within 45 min. In some vacuum work the oil manometer must be rejected due to, for example, use of gases of very high solubility. In these cases many attempts have been made to increase the sensitivity of U-tube mercury manometers.Gowenlock and Johnson 113 Fig. 5. Rayleigh (left) and Schrader-Ryder (right) manometers. These are all limited by the tendency of the mercury to stick to glass, and a sensitivity of about 10-3 torr is the best that is to be expected. An example of an accurate mercury manometer is the one constructed by Lord Rayleigh.lG The two limbs are made wide to avoid trouble with sticking, and each is fitted with glass points just touching the mercury. When a pressure is applied to one limb the mercury is displaced, and the movement is com- pensated by tilting the whole manometer in a vertical plane. A mirror is attached to the manometer and an optical-lever enables pressures up to 1.5 torr to be measured with an accuracy of 10-3 torr. An alternative arrange- mentl7 to measure the movement of the mercury is to have a mirror and float on the surface of the mercury.The manometer is then attached to a rigid support. These manometers are shown in Fig. 5. A simple but sensitive mercury manometer is described by Gerard.l* Essentially the movement of mercury is measured by means of two micrometer screw heads. The device is shown in Fig. 6. The micrometer heads are jointed to the U-tube by means of an epoxy resin, and the micrometer spindles are made vacuum tight by means of pairs of O-rings. Care is needed to ensure that the two spindles are accurately parallel and vertical if the device is intended for absolute work. A wide bore U-tube (25 mm) is necessary to ensure that capillary depression is essentially the same in both limbs.The effects of vibration can be a serious disadvantage, and a vibration-free support is necessary. In manometers of this type contact of the mercury surface with the measur- ing device is often determined electrically, but Gerard finds that, aided by a jeweller’s eyeglass, contact may be found very accurately simply by observing the coincidence of the point on the micrometer spindle with its reflection in the mercury surface. Readings were found to be consistent to 10-2 torr. A sensitive mercury manometer that finds wide application and is simple to construct is the mercury-oil magnifying manometer. Its basic constructionlg is shown in Fig. 7. The magnification factor is simply the ratio of the area of cross-section of the wide bore tubing to that of the narrow tubing.The range RIC Reviews 114 6 Fig. 6 (left). Dual micrometer manometer. The position of contact is determined by observing the coincidence of the micrometer point with its reflection in the mercury surface. Fig. 7 (right). Mercury-oil magnifying manometer. of operation is about 1-20 torr, depending on the quantity of oil employed. The arm containing the oil, which is maintained under vacuum, can be made vertical if space is a criterion for the size of the manometer, but it should be remembered that the error due to change in head of oil will then become more important. Correction for this error involves only a knowledge of the density of the oil used and is simple to apply.In this design the U-bend serves as a safety trap in the event of blowing over of oil, and the bulb ensures safe degassing of the oil. TI2e McLeod gauge The mercury manometer can be used to measure pressures between 10-6 and torr by compressing the gas by a known factor and making the actual pressure measurement on the compressed gas. The true pressure may then be calculated by making use of Boyle’s law. It should always be remembered that this gauge depends for its applicability on the validity of Boyle’s law, and hence will give true pressure measurements only for the ideal gases. For the more easily compressible gases it is possible to calculate the correction that should be applied (see, for example, page 78 of ref. 9). If the pressure of the compressed gas is less than 100 torr, then the correction is usually less than 1 per cent.A gauge was designed using this principle by McLeod in 1874, and it is still widely used, with only minor modifications. The general construction of the gauge is shown in Fig. 8. It consists of a compression bulb with capillary in which the pressure of gas is measured by comparison with the level in the immediately adjacent capillary of similar bore. In order to obtain a well- defined cut-off, an internal seal is inserted as is shown in the diagram. The end of the compression bulb ends in a drawn-out tube to prevent violent surges of mercury in the bulb. A water pump is suitable for lowering the mercury, the mercury being raised by the admission of air to the reservoir. Full construc- tional details of the glass-blowing operations6 and calibration procedure9 are 115 Gowenlock and Johnson *O M - 1 - U 8 6 Fig. 8 (left).McLeod gauge. Fig. 9 (right). Scales for McLeod gauge. 9 available. Calibration is greatly simplified if precision-bore capillary tubing is employed. There are two methods of reading a McLeod gauge: (a) quadratic scale; (b) linear scale. Figure 9 illustrates these methods. In method (a) the mercury in the open limb B is brought to the level of the closed capillary A. If the difference between the two mercury levels is h cm it may easily be verified from Boyle’s law that the pressure of the gas is given by P =T 1 Onr2h2 torr where r is the radius of the capillary and VO the volume of the bulb.The volume of compressed gas hnr2 is negligible compared with VO. In the linear scale method (b) the mercury in the open limb A is brought always to the same fixed level (MM’). Thus the compression ratio is made the same for all pressures. In this case 1 Oh’honr2 torr where VO > honr2 and the compression ratio is Vo/honr2. B p = vo The quadratic scale is to be preferred for general use as it has a higher sensitivity at low pressures and a wider working range. The amplification and range of the McLeod gauge depends on the physical dimensions of the instrument. If the capillary has an area of cross-section of 10-2 cm2 and a length of 20 cm, with a compression bulb of volume 200 cm3, the useful range will be 5 x 10-6 to 2 x 10-1 torr. It should be noted that mercury vapour from the gauge will enter the vacuum system at low pressures.This is eliminated by the insertion of a liquid nitrogen trap between the gauge and the apparatus. If a trap is used, then the inlet and outlet arms should be of equal diameter to prevent thermal effusion effects, and the gauge will measure only the pressure of the non-condensable gases present in the system. Modifications of the McLeod gauge include the insertion of either a RIC Reviews 116 Fig. 10. Vacustat gauge. thermistorz0 or a small Pirani gaugez1 at the top of the closed capillary, and measuring the pressure according to one of the methods described later. Sensitivities of 10-7 torr are claimed for these methods.Multi-range McLeod gauges are constructed in a variety of patterns. One method22 uses a number of compression bulbs, which allows a wider range of compression ratios than a single bulb. The gauge due to Roman1123 contains two cut-off points, allowing double compression. For the measurement of moderately low pressures (> 10-3 torr) compact gauges are available. The ‘vacustat’ (Edwards High Vacuum Ltd) is main- tained normally in a horizontal position: on turning through 90” about the ground joint, gas is cut off from contact with the main apparatus and a McLeod measurement is made. The operating range of this instrument (Fig. 10) is 10 to 10-2 torr. Mechanical manometers A variety of mechanical manometers, which depend upon the elastic deforma- tion of a sensitive solid element with changes in gas pressure, are described in the literature.With a solid member replacing the liquid there is no possibility of vapours entering the apparatus. The sensitivity is the same for all gases and vapours, and some mechanical manometers may be adapted to give a con- tinuous indication of pressure. It should be noted that all these gauges may be subject to hysteresis effects, which may become more important if the gauge is operated at elevated temperatures. If the gauge is taken above its recommended maximum temperature, or is subjected to pressure overloading, then perma- nent distortion may occur. Bourdon or spoon gauge. This type of gauge and its modifications depend for their action on the movement of a thin diaphragm of soft glass, Pyrex or silica, under the applied pressure.It is extremely useful in being direct reading and may be sealed directly to the reaction system. It may also be used differentially, but the sensitivity is not more than 2 x 10-3 torr and torr is a typical value. A convenient form of glass gauge is shown in Fig. 11, which consists of a flattened bulb to which there is attached a long pointer to magnify the motion of the bulb. The whole assembly is enclosed in a glass envelope, and the motion of the pointer is followed by means of a low power microscope. Owing to imperfections and to size of glass envelope, the magnification cannot be made greater than that capable of detecting a movement of 0.01 mm.Gowenlock and Johnson 117 I2 II Fig. I I (left). Bourdon gauge. Fig. 12 (right). Spiral gauge. In some designs the movement of the pointer is caused to tilt a suitably positioned mirror. The more sensitive the gauge, then the more prone it is to vibration. If necessary the gauge may be inverted, the connexion to the body taken from the side, and the tip of the pointer immersed in a low vapour pressure oil to damp excessive vibration. One disadvantage of this gauge is the fragility of the flattened bulb. In some models there is incorporated a safety stop to prevent excessive movement of the bulb. It is possible to adapt the Bourdon gauge so that it may provide a continuous record of pressure. For example, in Cox and Andon’sZ4 gauge the pointer and reference point each carry a nickel plate, immersed in silicone oil, the two plates acting as a variable capacitance.The change in capacitance may be measured by means of a capacitance bridge. Machin25 mounts a small section of ferrite rod onto the pointer, and this rod acts as the core of a linear variable differential transformer. The transformer is supplied with 10 V at 500 Hz from a small oscillator. The output from each secondary winding is rectified and then smoothed with a choke/capacitance filter. The difference in output from the two secondaries is then displayed on a pen recorder. The output, at an impedance of 50 kQ, is of the order of 10 mV torr-l. Ashmore and Tyler26 use an optical system whereby they mount razor blades on the fixed and movable arms to form a slit.This slit controls the intensity of light falling on a photocell, the output from which is amplified and recorded, and varies almost linearly with pressure over the range 0-100 torr. The spiral gauge. The spiral gauge, which is more robust than the Bourdon gauge, was first devised by Bodenstein and Dux,l whose model consisted of a long thin spiral of flattened silica tubing carrying a mirror at the end to detect the degree of ‘unwinding’ of the spiral. The basic design of the spiral gauge is shown in Fig. 12. The methods available for detecting the movement of the pointer are similar to those for the Bourdon gauge. RIC Reviews 118 Membrane manometers. The use of a membrane as the pressure responsive element is common for the measurement of low pressures.For vacuum applications, highly flexible membranes constructed of rubber or synthetic polymers are unsuitable owing to their high permeability and degassing characteristics. Thus membranes are constructed usually of thin glass or metal. It is common to use these gauges as null indicating instruments, as in order to obtain high sensitivity the membrane has low mechanical strength. Hysteresis effects may also become important with large membrane move- ments. Although mechanical and optical methods can be used to detect the movement of the diaphragm, most membrane manometers rely on electrical effects to facilitate the measurements. Methods used have included (a) varia- tion of mutual inductance, (b) dependence of electrical resistance OI? the strain in wires, (c) variation of capacitance.These methods are reviewed by Neubert27 and the physical form that the sensing element may possess is also discussed. Arrangement in the form of a capacitance gauge produces the most sensitive and fast-responding diaphragm gauge. The diaphragm and a fixed plate form a capacitor, wherein movement of the diaphragm results in a capacitance change. This change in capacitance is detected either by means of a capacitance bridge or by measuring the change in frequency of an r.f. oscillator, the capacitance manometer functioning as a variable capacitance across a tuned circuit. Commercial instruments are available ready-calibrated, incorporating high sensitivity, together with the capability of withstanding a pressure differential of one atmosphere.For example a capacitance manometer with a full scale reading of 1 torr will detect a pressure change of 5 x torr, while one with a full-scale reading of 1000 torr will readily measure a pressure change of 0.5 torr. The pressure range of any instrument may be extended by using it as a null- reading instrument, by adjusting a separately measured air pressure applied to exactly balance the sensing element to obtain a zero reading. This method, however, may be cumbersome and is not continuously indicating. It may also be difficult experimentally if the pressure to be measured is not constant. A direct-reading null method has been described by Drawin,28 in which the position of the diaphragm is maintained virtually constant by the application of an electrostatic force in opposition to the deflecting force.The diaphragm displacement is detected as a change in capacitance, which is converted to a d.c. polarizing voltage applied to the same sensing elements so as to keep the capacitance change to a minimum. The d.c. polarizing voltage is thus a measure of the applied pressure. A typical working pressure range is to 30 torr. The viscosity gauge The viscosity of a gas becomes pressure dependent when the mean free path of the molecules becomes comparable with the dimensions of the apparatus, and this effect has been used for the measurement of low pressures in a number of ways.Gauges of this type are among some of the earliest to be used for the measurement of low pressures, but have not been widely used due to (a) difficulty in obtaining a continuous indication of pressure and (b) the length of time required to obtain readings. Gowenlock and Johizson 119 Perhaps the simplest gauge is that consisting of a single quartz fibre of diameter about 0.2 mm, mounted upon a rigid support. The fibre is set vibrating by means of a gentle tap, although it is often difficult to persuade the fibre to vibrate in one plane. The amplitude of the vibration is observed by a microscope with an eyepiece with a graticule. It can be shown (page 94, ref. 9) that ti1 = A + B\/IMp to where t , is the time required for the amplitude of vibration to fall to one half of its initial value.B is a characteristic constant for the gas and M is the molecular weight. This gauge is useful within the pressure range 10-1 torr. Descriptions of more sophisticated types of vibrating fibre mano- meter are to be found in ref. 9. More recently viscosity gauges have been designed to provide a continuous indication of pressure. For example a metallic ribbon is made to vibrate in a magnetic field, causing the development of an alternating voltage. The ampli- tude of this is compared with a reference voltage, and a current is fed back to the ribbon to maintain its vibration at a constant amplitude. The value of the feedback current is then related to the pressure in the system. This type of gauge may be unsuitable for some chemical applications due to the presence of the magnet in the vacuum system.A development, due to Dimeff,29 is to replace the ribbon by a diaphragm between two fixed plates. The diaphragm is made to vibrate at its resonant frequency by the application of electrostatic forces between the diaphragm and one of the plates. The displacement is measured by the capacitance change between the diaphragm and the other plate. As in the previous example, a current is fed back to the electrostatic plate to maintain the vibration at con- stant amplitude. The gauge is suitable for pressure measurements within the range 10-5-200 torr. A linear high-vacuum gauge which is absolute if the molecular weight of the gas is known is used by Harbour and Lord.30 This consists of a magnetically- suspended rotating polished-steel sphere, which decelerates because of gaseous drag, this deceleration being directly proportional to the pressure. Although the associated control system is rather complex for general application, the instrument may be employed for the calibration of ionization or other gauges in the range 10-6-10-3torr7 especially if the McLeod gauge is unsuitable due to the use of condensable vapours.The overall range of the gauge may be altered by the use of spheres of varying diameter, smaller spheres being used at the higher pressures. The molecular mean free path should be at least 10 times greater than the diameter of the sphere. Tlie Pirani gauge At low pressures the rate of heat transfer through a gas is a function of pres- sure.Thus a heated wire suspended in a gas will lose a proportion of its energy by the impact of cold gas molecules upon its surface. This principle was first suggested as a means of measuring low pressures by Pirani31 in 1906. The pressure range of Pirani and other thermal conductivity gauges is similar to that of viscosity gauges. As the thermal conductivity gauge is a direct reading RIG' Reviews 120 Fig. 13. Pirani gauge. instrument and construction is simple, it has found wide application in low pressure work. Its main disadvantage is that it is not an absolute pressure gauge and must be calibrated for each gas measured. Commercial gauges as normally supplied are calibrated in air.For the derivation of the relation between pressure of gas and the heat loss from a thin hot wire a standard text, such as the monograph by Leck,l4 should be consulted. The construction of a simple Pirani gauge is similar to a small electric lamp. When constructing the gauge it is often simpler to make it as shown in Fig. 13, but without the wire. The tube can then be cut at the dotted line, and the filament spot-welded to the nickel supporting leads and then the whole assembly resealed. A suitable filament to use is of 0.01 mm platinum or nickel. The resistance of a filament about 5 cm long will then be of the order of 100 I2 at room temperature. Gauges of high sensitivity are sometimes operated immersed in liquid nitrogen, and in this case the tube carrying the wire connexions should be filled with paraffin wax, or some similar material to prevent condensation of atmospheric moisture in the tube, which could cause a short-circuit to develop.Unless the filament is very long, the use of a spring to prevent the filament sagging when at its operating temperature is unnecessary. If long filaments are to be employed, helical winding may be introduced. The sensitivity of a helical filament will be somewhere between that of a straight filament of the same wire and that of a thick filament of diameter equal to that of the helix. The shielding effect of each turn on its neighbours is inversely proportional to the pitch, and becomes important when the pitch is less than 10 wire diameters.Pirani gauges are used in Wheatstone bridge circuits. There are three modes of operation : (a) constant filament temperature, in which the resistance (and hence temperature) is maintained constant and the energy input required to do this is observed as a function of the pressure, (b) constant voltage, where the change in current is observed as a function of pressure, (c) constant current, where the change in current is measured. Gowenlock and Johnson 121 9 If the sensitivity required is not more than about 5 x 10-3 torr then a very simple Wheatstone bridge circuit, with a small accumulator as the voltage source, is all that is required. One of the arms of the bridge is made variable so that a suitable range for the gauge may be selected. Calibration is most easily effected by means of a McLeod gauge.If a liquid nitrogen trap is interposed between the Pirani and McLeod gauges to eliminate the presence of mercury vapour, it is important to ensure that the inlet and outlet tubes of the trap are of equal bore to prevent erroneous calibration due to thermal effusion. The ultimate sensitivity of the Pirani gauge is difficult to estimate as it is very dependent both on the stability of the bridge supply voltage and the ambient temperature. When high sensitivity is required, most workers prefer an electronically stabilized bridge supply, and in some cases the gauge is im- mersed in liquid nitrogen. It is possible to reduce the effect of voltage and temperature fluctuations by using two identical gauge elements, which are placed in adjacent arms of the bridge.The extra element is pumped out and sealed off at a low pressure. Use of a getter will maintain the sealed off element at high vacuum, by removing any gases that become desorbed while the gauge is in use. The two elements must be placed close together so as to be subjected to the same ambient temperature changes. The effect of temperature fluctuations of the gauge envelope on the tempe- ture of the filament itself depends on the gas pressure. Thus compensation is only truly effective when the pressures in the two elements are the same. In practice compensation is effective at pressures of less than about 10-1 torr if the compensating element is sealed off at high vacuum.Small bead thermistors can be used as the sensing elements in Pirani gauges. DushmanlO describes a gauge that employs a sealed-off gettered thermistor as a temperature compensator to the other thermistor element. Thermistor gauges can be operated in either constant current or constant voltage bridges, and can be made very compact. The pressure range of the thermistor gauge is similar to that of the hot wire gauge, i.e. 10-4-10 torr under normal condi- tions. Both types of gauge can be used at higher pressures but the response becomes excessively non-linear and sensitivity decreases. A thermistor gauge has been described32 recently which can be used at high pressures. This is achieved by surrounding the thermistor bead with finely powdered glass.Packing the gauge with a powder of about 40 pm particle size raises the useful upper pressure limit from 0.5 torr to 20 torr, and it is suggested that use of a finer particle size (1 pm) would enable the gauge to be used at pressures up to one atmosphere. The lower pressure limit of the gauge occurs when the thermal conductivity of the gas approaches that of the powder, and will be dependent on particle size. RIC Reviews An alternative to measurement of the wire temperature by a Wheatstone bridge arrangement is the attachment of a thermocouple to the hot wire. In general, the sensitivity of this arrangement is a little less than that of a con- ventional Pirani gauge, and the construction of the gauge is complicated by the introduction of the thermocouple. The thermocouple may be spot-welded or silver-soldered to the wire, or the joint may be made with a silicate cement.This last method will also electrically insulate the thermocouple from the hot- wire. These elements are available commercially. More extensive details of 122 construction are to be found in Leck’s m0nograph.1~ The thermocouple gauge would appear to have little if any advantage over the conventional Pirani gauge. Ionization gauges When a gas at low pressure is subjected to the passage of a stream of electrons the number of positive ions produced by electron impact is directly propor- tional to the concentration of molecules present. At low pressure the relation between ionization and concentration is linear, but this linearity ceases when the ion concentration is sufficient to change the current and energy of the electron beam.The positive ion current can thus be used to measure the gas pressure. There are three types of ionization gauge and these are all available com- mercially. The useful operating ranges are : (a) hot-cathode thermionic gauge lop2 to torr; (b) cold-cathode (Penning) gauge 10-2 to 10-7 torr; (c) radioactive ionization (Alphatron) gauge 103 to 10-3 torr. The response of these gauges depends on the nature of the gas being measured, and so calibra- tion against an absolute pressure gauge is necessary. Commercial instruments are calibrated in air, although some manufacturers publish conversion graphs for some of the more common gases.Leck’s monograph14 should be consulted for details of gauge construction and electronic control circuits that are necessary for the operation of an ionization gauge. It should be noted that in the hot-cathode gauge the hot filament may be chemically attacked, or may decompose organic vapours present. The ion-current pressure graph in the cold-cathode gauge may have discontinuities and sudden current increases may occur. These sudden dis- continuities are caused presumably by changes in equilibrium conditions of the glow discharge, and are neither predictable nor stable. With conventional hot-cathode gauges a lower pressure limit is set by the presence of a background current corresponding to a pressure of about lops torr.This spurious current is caused largely by photo-emission of electrons from the anode. This zero error is reduced in the Bayard-Alpert gauge33 to about 10-lO torr. This gauge, which can easily be degassed, has the conven- tional electrode configuration reversed, with a fine wire central anode sur- rounded by the grid in the form of a helical coil. The hairpin filament is external to the grid. This gauge is of importance in the measurement of ultra- high vacua. A very sensitive ionization gauge, which may come to be widely used in ultra-high vacuum studies, is the ‘minimass’ partial pressure gauge (AEI Ltd and Vacuum Generators Ltd). This gauge is essentially a very small low resolution mass spectrometer which may be jointed directly to the vacuum system.This gauge will detect pressures of 10-l1 torr and causes very little system disturbance, having a low degassing rate and virtually no pumping effect. Resolution is sufficient to give unit mass separation up to mass 28 and has a maximum mass range of 240. The information about the residual gases present in the system is extremely useful when working at ultra-high vacuum. When used to measure pressure, normal ionization gauges can themselves affect the system. The hot-cathode type of gauge may degas for an appreciable Gowenlock and Johnson 123 time after it is switched on, and for accurate measurement should be thoroughly degassed beforehand. The ionization gauge will also upset the pressure distribution in the system by acting as a pump and removing gas from the system.Gas is removed in two general ways: (i) adsorption of ionized molecules on the electrodes and the gauge envelope, (ii) reaction of active gases with the hot filament. The effect due to (i) normally ceases when the surfaces are saturated, but (ii) may continue indefinitely until all the tungsten has been consumed, oxygen at high pressure being particularly bad in this respect. The pumping rate of the Penning gauge is very much greater than for the hot-cathode gauge because of the greater electron density and energy. Saturation takes place for most gases but the pumping action continues for a much longer period, for nitrogen14 up to 24 h. The Penning gauge is robust and cheap and is very useful as a vacuum indicator but due to its pumping action and the discontinuities in its calibration curve it compares unfavourably with the hot-cathode gauge for accurate measurements.When pressure gauges are used to indicate the residual pressure in a high- vacuum system the position chosen for the gauge is important. Thus a pressure of 10-5 torr in a section of the apparatus remote from the pumping system is more meaningful than a pressure of 10-6 torr immediately adjacent to the diffusion pump cold trap. Care should always be exercised when interpreting absolute pressure readings. CONSTRUCTION OF VACUUM APPARATUS : STANDARD DEVICES Mention has already been made of vacuum taps and traps for removing volatile materials from vacuum systems.It is immediately apparent on reading the wide literature relating to vacuum techniques that the scope for ingenuity in constructing and adapting vacuum systems is almost unlimited. The main fabrication materials are metals, glasses, refractories and polymers, the first two of which are employed as tubing for the essential connecting lines of a vacuum apparatus. The choice of metal tubing is necessitated in those cases where mechanical strength and resistance to shock and vibration are para- mount. In a few cases metal systems have greater corrosion resistance than glass, but in general glass is preferred due to ease of cleaning, construction and modification. Copper and brass tubing are reasonably cheap and may readily be brazed and soldered.Mild steel and stainless steel are in regular use for vacuum valves, baffles and traps and also for diffusion pumps. In a metal system it will frequently be necessary to connect tubing together and where permanent joints are not permissible, demountable unions will be required. For such unions gasket seals are commonly employed. An O-ring of an elastomer such as Viton A or Neoprene is used which fits accurately into a groove (see Fig. 14) in the metal flange, so that when a second flange is bolted on distortion of the O-ring with complete filling of the groove occurs, resulting in a vacuum-tight seal. For temperatures higher than 200 "C, soft metal gaskets (e.g. gold wire) are used with flat ungrooved flanges. Various compression screw unions are also available.Where metal tubing is joined to glass, then it RIC Reviews 124 Fig. 14. Schematic diagram of elastomeric O-ring showing fit in groove in the metal flange of a gasket seal. is possible to use either metal-to-glass seals and cone-and-socket joints in addition to compression screw unions with appropriate gaskets. A wide range of metal vacuum valves is produced commercially and these are in frequent use either for controlling rates of addition of volatile materials to vacuum systems or for on-off action. The operating mechanisms are based upon screw or cam operation and vacuum sealed either by elastomers or by metal bellows or both. Such valves are often used in systems that are otherwise of glass construction. Glass vacuum systems preponderate in chemical laboratories.Where apparatus must be demountable, standard ground glass joints (cone and socket, ball and socket, ground flange) are used in conjunction with an appropriate high vacuum grease or wax. Glass vacuum taps are available in a variety of designs. For both taps and joints the vapour pressure of the grease I/ I I Neoprene "iton Or d i ap h rag rn 11 I I Pyrex glass 16 15 Fig. 15 (left), Fig. 16 (right). Schematic diagrams t o illustrate principles of greaseless taps. A, PTFE; 5, stainless steel fixing plate; C, flange, D and E, Viton A rings. Gowenlock and Johnson 125 Liquid air _ _ _ _ ~ _ _ _ - - _ - y{ l r Pump 17 Fig. 17 (left). High-efficiency trap. Fig. 18 (centre). Fog removal trap.Fig. 19 (right).Volman temperature gradient trap. limits the vacuum available in the apparatus. A wide range of vacuum greases and waxes is available and there is little difficulty in finding one of suitable vapour pressure characteristics. Some hydrocarbon-based greases have the advantage of being continuously usable for long periods of time without deterioration, but may have the disadvantage (under British conditions) of requiring a room temperature of 25 "C for smooth turning. This problem can be overcome with the aid of an electric hair drier. In recent years several types of greaseless taps have been manufactured, and these find extensive use where greasefree systems are necessary on chemical grounds. Some workers also prefer freedom from regular regreasing operations.The main types in regular use are based either upon controlled movement of an elastomer plunger on to a glass tube or upon movement of a PTFE plunger so as to produce a seal over an orifice. Figures 15 and 16 illustrate these principles. Traps for vacuum systems are often constructed as in Fig. 3 (p. 11 l), other designs9 are shown in Figs 17-19, the latter pair being particularly recom- mended for fog removal. It is sometimes necessary to use a trap cooled to temperatures lower than -196 "C (the boiling point of liquid nitrogen) and for these purposes a 'pumped out' trap (see Fig. 20) is employed. The Ward- 126 Heated to 200 OC Apparatus Cooled in liquid N, space fi =)j= - Imm annular 19 18 20 21 Fig.20 (left). Pumped-out liquid-air trap. Fig. 21 (right). LeRoy still. RIC Reviews -Icm i, ~ A Fig. 22. Kimber and Napier's high efficiency liquid-nitrogen-cooled trap.34 -@- I Section A-A 24 23 Fig. 23 (left). Schematic illustration of set-up for permitting motion within a vacuum system. Fig. 24 (right). Preparation of small amounts of oxygen in a windlass apparatus. Gowenlock and Johnson 127 S N . . . . . . . . . . . Fig. 25. All-glass needle valve. Le Roy still trap (see Fig. 21) is used for fractional distillation at low tempera- tures. The outer jacket C is cooled by an external cooling bath and can be evacuated to a desired pressure through V. A heating coil is wound on to the surface B up to point D. In addition several thermocouples are also first attached to B and insulated from the heating coil.The heating wires and thermocouples pass out through W and are sealed with a vacuum cement. A high efficiency liquid-nitrogen-cooled trap has been designed by Kimber and Napier34 (see Fig. 22) which makes use of a narrow annular space. Such a trap should make it possible quantitatively to remove low concentrations of condensable gases from large excesses of permanent gases. It is frequently necessary to allow for motion within a vacuum system. It is simple to see from Fig. 23 that this may be obtained using the principles of 0- ring seals as in Fig. 14. Another simple technique for vertical movement is the windlass principle using a greased joint as in Fig. 24 (used for preparation of oxygen).Two techniques which offer simplicity of construction are (a) electro- magnetic or magnetically operated devices, (b) devices operated by columns of mercury. Examples of electromagnetic valves are the all-glass needle valve (Fig. 25) where the needle N slides into the capillary C, and the greaseless taps (Fig. 26) using either cone-and-socket, or ball joints. Extensions to the opening of break-seals and combination of electromagnetic operations with mercury sealing of joints are immediately obvious. Mercury column devices are used in many varieties of cut-off valve. The operating principle is simple, mercury being allowed to flow into a U-tube, where it prevents the flow of gas through the apparatus. The mercury column is restricted in its movement either by sintered glass discs, which are permeable to gas but not mercury, or by float valves, which produce an effective ground glass seal (see Figs 27-28).Mention has already been made of the greaseless valve developed by 7 I" v- Fig. 26. Electromagnetic valves. C joint r RIC Reviews 128 tered glass 28 Fig. 27 (left), Fig. 28 (right). Schematic diagrams of mercury column devices used in cut-off valves. Bodenstein and Dux :l the limited movement possible in this case was depen- dent upon compression of a glass bellows. Commercially available vacuum valves are often of the needle valve design using metal bellows. Operation of such a valve offers advantages over the types with conventional packing glands.A relatively recent valve due to Snelleman35 is an O-ring sealed valve which uses a capillary of variable length. On moving the cylinder by means of the screw head, the effective length of the spirally grooved capillary in the mantle is altered (see Fig. 29). Other useful designs of O-ring operated valves are those designed by Pratt and P ~ r n e l l , ~ ~ and Cundell, Hay and Lemeunier37 for removing small samples of materials from reaction vessels operating at low pressures, with facility for subsequent connexion of the valve to carrier gas flows at high pressure for gas chromatographic analysis. Another feature that it is often necessary to incorporate into high vacuum systems is the controlled leak. This can, of course, be carried out by using needle valves in conjunction with a very fine capillary, thereby allowing a gas to enter the main apparatus with a very high degree of control.Another acceptable technique employs leakage of a gas through a porcelain rod. This porous material can function as an adjustable leak by using a mercury column to vary the area of the porcelain exposed to the gas which is to be leaked into the vacuum system. This technique, which was originated by Wansborough- Fig. 29. Partly exploded view of a capillary valve. The groove in the cylinder is indicated by G. 129 Gowenlock and Johnson Main apparatus C ___t I E t U C .- d 0 2 $ L Sample system c - Calibrated voli Jme - .7 f 31 30 Fig. 30 (left). A simple dosing system.Fig. 31 (right). A dosing system based on the Topler Pump- Jones,38 has recently been adapted to mercury free operation by Askins.39 A gas leak based on the differential expansion of glass held at two widely different temperatures was devised by Harrison.40 Simple dosing systems find regular usage in high vacuum systems. The simplest pattern employs a calibrated volume located between stopcocks the volumes of which are also known. Figure 30 is self-explanatory. More sophisti- cated devices employ the Topler pump principle. A known amount of gas can be admitted into a tube and then cut off at B by a mercury column. When the mercury in it is lowered the gas can then enter the apparatus through the sinter C (see Fig. 31). This device is due to Young.41 A modification to allow the transfer of millilitre quantities which uses a magnetically operated cup sealer has been devised by Pritchard.42 CIRCULATION OF GASES IN VACUUM SYSTEMS It is often necessary to circulate gases in closed vacuum systems, e.g.when reactions are to be studied in flow systems or when a considerable quantity of product is required in a synthesis and a simple batch technique is inapplicable. Attention has been drawn9 to the variety of pumps that are available for this purpose and the regular publication of improved designs of pumps indicates that there is yet scope for inventiveness. Melville and Gowenlock's classifica- tiong is followed in this section. Mercury diffusion pumps which possess a high cut-off pressure (i.e.15-40 torr) are particularly useful for circulating gases at moderate flow rates at pressures below the cut off pressure of the pump. It is necessary to remove mercury vapour from the stream of flowing gases, and high efficiency traps (see, e.g. Fig. 18) are used for this purpose. It is also necessary to ensure that the gases which enter the pump do not react with the mercury vapour. Pumps of this design have been in use for over 40 years and significant advances in design have not been reported. Electromagnetically operated pumps are, on the other hand, available in a variety of patterns based upon either horizontal or vertical pistons. The RZC Reviews 130 operating principle is the movement of a glass piston which incorporates a soft iron core.The piston moves either back and forth or up and down within a glass tube, a close fit of the piston being essential. Uni-directional flow of gas is ensured by the incorporation of glass check valves into the connecting tubing of the pump, two valves being inserted into each of the two lines. The most recent improvement to these pumps is the replacement of the multi- solenoid operation of the piston by a ring magnet surrounding the pump cylinder. The magnet movement is produced using a mechanical disc-and-arm linkage, the disc being keyed on to the output shaft of a vertically mounted variable-speed drive consisting of a gearbox and motor. The resultant possible variation of disc speed produces a stroke rate of the piston varying from five to 100 strokes per minute. Duncan and Lawson43 have produced a horizontal plunger pump of this design (see Fig.32) and a vertical pattern has been designed by Gustavsson.44 The flow rates available lie within the range 0.25-8 1 min-l and the flow rate is a linear function of (a) the stroke length (pressure difference and stroke rate being constant) and (b) the stroke rate (pressure difference and stroke length being constant). These recent develop- ments have extended the usefulness of this type of circulating pump. Rotating piston pumps, in which a high-speed rotating magnetic field is used to rotate a glass fan containing an internally sealed bar magnet, have not received much attention as gas circulating pumps. The published data45 indicate that these pumps are relatively inefficient at pressures lower than 400 torr and in general their use is confined to the stirring of mixtures of two or more gases to obtain complete mixing.Mercury piston pumps may be used, where there are no problems arising from the presence of liquid and/or gaseous mercury in the main vacuum apparatus. A recent design due to H i l l i ~ ~ ~ employs an oscillating mercury column with the flow controlled by electrically operated glass ball-and-socket valves. The wide operating range of this pump (760 to less than 0.1 torr) is a useful feature and this is enhanced by its ability to operate against considerable back pressures, such as are encountered in tightly packed reaction vessels, continuously for several days.The pumping rates are lower than for the glass piston pumps previously mentioned lying in the range 0-500 ml min-1. E D Fig. 32. Magnetically coupled gas circulating pump. A, piston cylinder; B, rod magnet; C, glass piston; D, ring magnet; E, B29 stopper; F, glass ball valves; G, ground-glass valve seats. Gowenlock and Johnson 131 Attention has been drawn9 to the fact that it is possible to use two auto- matically operated Topler pumps coupled together as circulating pumps. In this connexion it is pertinent to note that several designs of automatic Topler pumps have been published recently.46-50 These designs extend the previous range of automatic operation.9 Such pumps may also be used for evacuation and simple gas transfer operations.ULTRA-HIGH VACUA Evacuation to and operation at pressures of 10-10-10-7 torr is necessary in the study of for example surface phenomena, emission characteristics and in the far ultra-violet spectroscopy of evaporated metal films. Most organic sealing materials have vapour pressures greater than 10-7 torr, and are excluded from use at ultra-high vacuum. The whole vacuum system is constructed from metals, glass and ceramics of low vapour pressure (metals such as zinc, lead and brass are unsuitable). The complete system is mounted within a demountable furnace so that it may be heated to 400-500°C to facilitate degassing, the effect of which is reduced by having the surface area of the system as low as possible. High-vacuum greased and greaseless taps are unsuitable for use at ultra- high vacuum due to vapour pressure and porosity.Thus valves are constructed entirely of metal, and vacuum-tight seals are produced by driving a soft metal (e.g. copper) pad against a stainless-steel knife edge. In this way conductances of 10-14 1 s-1 or less are obtained when the valve is closed. Joints are con- structed using a variety of metals and types of seal. Metal-glass housekeeper seals and metal-ceramic seals may be used. Demountable seals employ compression-ring seals with aluminium or gold wire gaskets. Pressure measurement in ultra-high vacuum systems is accomplished by ionization gauges, which are described in an earlier section (p. 123). Gauges must be degassed before use at 400-500 "C, followed by the use of electron bombardment to heat the electrodes to red heat.Pumping effects are reduced by running the gauge at reduced electron emission and, if the gauge is not situated within the vacuum vessel itself, short wide-bore connexions should be used. An efficient rotary oil pump or a combination of rotary pump and diffusion pump is necessary to back the ultra-high vacuum pump. Pressures below torr can be attained with diffusion pumps, sputter-ion pumps or cryopumps, all of which are available commercially. The speed of the pump should be at least 1 1 s-1 for every 100 cm2 of surface area in the system, in order to pump faster than the system will degas. Diffusion pumps are fitted with very efficient baffles and cold traps to prevent oil vapour backstreaming into the vacuum system.To maintain high pumping speeds all connecting tubing is of wide bore. RIC Reviews The sputter-ion pump operates on a principle similar to that of the Penning vacuum gauge. Electrons, produced by thermionic emission from a filament, are attracted towards the anode. Due to the presence of an intense magnetic field, produced by an external magnet, the electrons describe a spiral path much greater than the direct distance between the filament and anode. The electrons collide with gas molecules, producing ions which bombard the titanium cathodes and eject titanium atoms. These are deposited largely on 132 the anodes, producing a continuously replenished gettering surface.Active gases are pumped by being converted into stable compounds of titanium when they come into contact with the getter surface, while the inert gases are pumped by ion burial at the titanium cathodes. In general diffusion pumps are to be preferred for large gas throughputs and when the system is repeatedly cycled from atmosphere to loplo torr, whereas ion pumps are used with apparatus of limited volume. In ‘ultra-clean’ systems of limited volume the roughing vacuum pumps may be dispensed with and sorption pumps used to back the sputter-ion pump. Cryopumping is a clean pumping process in which gases and vapours are condensed onto a liquid-helium-cooled surface. Very high pumping speeds of the order of 1000 1 s-1 are attained as the pump is sited within the vacuum vessel, and the gas thus has the minimum of restrictions upon its flow to the pump.An auxiliary ion or diffusion pump is necessary to pump helium or hydrogen, which are not condensable. Further information on the techniques of ultra-high-vacuum may be found in ref. 51 and useful details are included in manufacturers’ catalogues. POSSIBLE DEVELOPMENTS Sanderson concluded his book5 with a suggested series of 10 desirable develop- ments in basic vacuum techniques and apparatus. It is interesting to note that five of these developments remain to be put into effect. These are therefore detailed below with the hope that this will catalyse those qualities of inventive- ness that have characterized the development of vacuum techniques.(1) A lubricant that combines all of the following properties: very low vapour pressure, excellent workability over a long period of time and with regular turning, chemical inertness, non-solvent properties and flow resistance. (2) A simple, cheap, direct, continuously reading manometer which is com- pletely independent of vapour composition, with a range from to 760 torr. (3) A means for maintaining a complete apparatus at a temperature much higher than that of the room. (4) A chemically inert, non-solvent liquid of lower density than mercury. (5) An easily worked material combining the good properties of glass with those of greater strength, greater flexibility and greater inertness to chemical attack. ACKNOWLEDGEMENTS The authors wish to thank the following for permission to reproduce copyright material: Macmillan and Co.Ltd for Figs. 5, 7-13, 17-21,24-26,31; Edwards High Vacuum Ltd for Figs. 1 and 2; Chemistry and Industry for Fig. 22 ; Jencons (Scientific) Ltd for Fig. 16; Journal of Scientific Instruments for Figs. 4,6,29, 32; G . Springham and Co. Ltd for Fig. 15. REFERENCES 1 M. Bodenstein and W. Dux, 2. phys. Chem., 1913, 85,297. 2 M. W. Travers, The Experimental Study of Gases. London: Macmillan, 1901. 3 H. Moser, Die Reindarstellung von Gasen. Stuttgart: Enke, 1920. 133 Gowenlock and Johnson 4 A. Stock, Hydrides of Boron and Silicon, chapter 30. Ithaca: Cornell University Press, 1933. 5 R. T. Sanderson, Vacuum Manipulation of Volatile Compounds. New York: Wiley, 1948.6 A. J. B. Robertson, D. J. Fabian, A. J. Crocker and J. Dewing, Laboratory Glass-working for Scientists. London : Butterworths, 1957. 7 P. L. Read, Vacuum, 1963, 13,271. 8 R. E. Dodd and P. L. Robinson, Experimental Inorganic Chemistry, New York: Elsevier, 1954. 9 H. Melville and B. G. Gowenlock, Experimental Methods in Gas Reactions. London: Macmillan, 1964. 10 S. Dushman, Scientific Foundations of Vacuum Technique- second edn, J. M. Lafferty (Ed.). New York: Wiley, 1962. 11 K. Diels and R. Jaeckel, Leybold Vacuum Handbook, second edn. London: Pergamon, 1965. 12 B. D. Power, High Vacuum Pumping Equipment. London: Chapman and Hall, 1966. 13 L. Ward and J. P. Bunn, Introduction to the Theory and Practice of High Vacuum Technology. London : Butterworths, 1967. 14 J. H. Leck, Pressure Measurement in Vacuum Systems, second edn. London: Chapman and Hall, 1964. 15 J. V. McAllan, J. scient. Instrum., 1965, 42, 290. 16 Lord Rayleigh, Phil. Trans. R. SOC. A , 1901, 196, 205. 17 J. E. Schrader and H. M. Ryder, Phys. Rev., 1919, 13, 321. 18 V. B. Gerard, J. scient. Instrum., 1966, 43, 194. 19 L. Batt and B. G. Gowenlock, Trans. Faraday SOC., 1960, 56, 682. 20 R. S. Bradley, J. scient. Instrum., 1954, 31, 129. 21 A. H. Pfund, Phys. Rev., 1921, 18, 78. 22 R. F. Strickland-Constable Trans. Faraday SOC., 1947, 43, 769. 23 M. P. Romann, Vide, 1948, 3, 522. 24 J. D. Cox and R. J. L. Andon, Trans. Faraday SOC., 1957, 53, 1074; 1958,54, 1622. 25 W. D. Machin, Can. J. Chem., 1967, 45, 1904. 26 P. G. Ashmore and B. J. Tyler, personal communication, 1959. 27 H. K. P. Neubert, Instrument Transducers. London: OUP, 1963. 28 H. W. Drawin, Adv. Vac. Sci. Techqs, 1958, 1,274. 29 J. Dimeff, J. W. Lane and G. W. Goon, Rev. scient. Instrum., 1962, 33, 804. 30 P. J. Harbour and R. G. Lord, J. scient. Instrum., 1965, 42, 105. 31 M. Pirani, Verh. dt. phys. Ges., 1906, 8, 686. 32 M. Green and M. J. Lee, J. scient. Instrum., 1966, 43, 948. 33 R. T. Bayard and D. Alpert, Rev. scient. Instrum., 1950, 21, 571. 34 G. M. Kimber and D. H. Napier, Chemy Ind., 1965, 725. 35 W. Snelleman. J. scient. Instrum.. 1966. 43. 832. 36 G. L. Pratt and J. H. Purnell, Aialyt. Chem., 1960, 32, 1213. 37 R. B. Cundall, K. Hay and P. W. Lemeunier, J. scient. Instrum., 1966, 43, 652. 38 0. H. Wansborough-Jones, Proc. R. SOC. A., 1930, 127, 530. 39 P. J. Askins, J. scient. Instrum., 1966, 43, 267. 40 E. R. Harrison, J. scient. Instrum., 1953, 30, 170. 41 D. M. Young, Rev. scient. Instrum., 1953, 24, 79. 42 J. Pritchard, J. scient. Instrum., 1966, 43, 262. 43 S. Duncan and F. Lawson, J. scient. Instrum., 1967, 44, 388. 44 J. E. Gustavsson, J. scient. Instrum., 1967, 44, 860. 45 H. M. PapCe, A. Montefinale and T. W. Zawidzki, J. scient. Instrum., 1962, 39, 394. 46 A. J. Davis and G. J. Ogilvie, J. scient. Instrum., 1966, 43, 116. 47 I. McAlpine and H. Sutcliffe, Chemy Ind., 1967, 1170. 48 M. Bufalim and J. E. Todd, J. chem. Educ., 1967,44,425. 49 D. A. Rice and J. Roach, J. scient. Instrum., 1967, 44, 473. 50 K. G. McLaren and W. T. Williams, J. scient. Instrum., 1967, 44, 1033. 51 D. Alpert, Handbuch der Physik, Vol. 12. Berlin: Springer, 1958. R IC Reviews 134

ISSN:0035-8940    

DOI:10.1039/RR9680100107

出版商:RSC    

年代:1968

数据来源: RSC

摘要:

THE CHEMISTRY OF TRIBOLOGY Friction, Lubrication and Wear Geoffrey W. Rowe, M.A., Ph.D., D.Sc. Department of lndustrial Metallurgy, University of Birmingham, Birmingham I5 Introductory Survey Fundamentals of Surfaces. . . . .. . . . . . . Adsorption and clean surfaces, 140 Surface structure and energy, 142 Physical adsorption of gas, 143 Chemisorption, 146 Surface topography, 148 .. .. .. I. Greases . . ,. .. .. .. .. . . . . . . . . . . . . . . . . Preparation and Properties of Mineral Oil Lubricants Production of mineral oils, 152 Chemical additives for lubricating oils, 155 Greases, Synthetic Fluids and Solid Lubricants Preparation of greases, 163 Types of greases, 165 11. Synthetic fluids . . . . . . .. . . .. .. Friction of Dry Solids and of Lubricating Layers Introduction to rolling and sliding friction, 173 Friction of a van der Waals solid, 174 Friction of covalent solids, 178 Friction of ionic solids, 179 Friction of organic polymers, 179 Friction of layer-lattice crystals, 182 Friction of metals, 186 Thin-film lubrication of metals, 188 .. .. Mechanical and Chemical Interactions in Wear Processes . . The three stages of wear, 194 Steady wear conditions, 195 Influence of oxygen and water vapour on wear and metal transfer, 198 Influence of lubricant films on wear and metal transfer, 200 Corrosive aspects of wear, 200 This survey of the role of chemistry in the varied processes of friction, lubrica- tion and wear is written for professional and student chemists.It is hoped that it may arouse an interest in this branch of their subject; and particularly that it may indicate the considerable contribution still to be made by chemists in tribology, especially by exploring the details and exploiting the potentialities of surface chemical action, which appears so frequently in tribology. A deliberate attempt has been made to indicate the areas in which the author believes that fruitful research can be performed. Rowe . . 136 . . 1 40 .. 151 .. 162 . . 162 . . 168 .. 173 . . 194 135 No claim is made to originality of subject matter, though the author accepts responsibility for the presentation. He hopes that the text will be easily read and may prove stimulating to those having little acquaintance with tribology.To simplify the presentation, the usual practice of giving detailed references has been abandoned, but the author wishes to express his obvious debt to all fellow workers in tribology and particularly to thank Dr E. R. Braithwaite, Dr J. P. G. Farr, Mr I. S. Morton and Mr T. G. Rowe who read the script and made helpful comments. Some books which are recommended for further reading are cited in the bibliography. These provide comprehensive lists of research papers. INTRODUCTORY SURVEY The word tribology is new, but the subject has been important throughout recorded history. It covers the whole field of science and technology associated with sliding, and there is evidence of the solution of tribological problems in the ancient civilizations of Egypt and the East.Tribochemistry and tribo- physics are considered as subdivisions of the subject, dealing explicitly with chemical and physical phenomena associated with sliding surfaces. It is however important to recognize that most of the characteristics of friction, lubrication and wear are complex, and their understanding requires some discussion of chemistry, physics, metallurgy and mechanical engineering. We shall be concerned here primarily with the role of chemistry in tribology but the subject cannot satisfactorily be divided into independent academic compartments. Let us for example consider one of the simplest and most perfect tribological devices, a journal bearing operating at its designed speed and load.Such a bearing consists of a solid cylindrical journal, which may be of any size from a few thousandths of an inch up to several feet in diameter, rotating within a hollow cylindrical bearing of slightly larger internal diameter. When the journal is at rest its centre will be vertically in line with, but slightly below, the centre of the bearing. The space between the two is filled with lubricating oil of medium viscosity. As the journal is rotated this oil will be subjected to viscous drag and forces will develop which create a pressure in the narrow wedge of oil beneath the journal. Eventually, when full speed is reached, this pressure will be sufficient to support the journal and the weight of the shaft and other loads which it may carry.The two surfaces will then be completely separated by a film of lubricant and the bearing is said to be operating under hydrodynamic conditions. There will be no wear of the moving parts and the friction will be determined solely by the viscous properties of the lubricant under these conditions. The coefficient of friction (frictional resistance divided by load) is usually very low, of the order of p = 0.002, and may be lower still in aero- dynamic bearings where the lubricating fluid is air. Engineers consequently strive to maintain this ideal state of hydrodynamic lubrication wherever possible. Considerable success has been achieved in recent years in designing such bearings, making use of hydrodynamic theory. Chemistry plays no part in these calculations, except in so far as the molecular structure of the lubricant RIC Reviews 136 determines the viscosity and the dependence of viscosity on pressure and temperature. Data sheets now exist with which mechanical engineers can design journal bearings to operate under specified load, speed and temperature conditions with lubricants of known viscosity characteristics.Bearings cannot always operate under the conditions prescribed in the design. We have already mentioned that the lubricant film is not established until the journal attains sufficient speed. At low speed the film of lubricant will be very thin indeed, and may be completely excluded from some local areas of the contacting surfaces. This is especially important with aerodynamic bear- ings, but the problem occurs in all bearings during starting and stopping, and also when the bearing is subjected to pulsating or shock load.In the regions of intimate contact the molecular fields of force become very important. As we shall see later (pp. 139 and 141), if two uncontaminated metal surfaces are brought into contact the adhesive forces between them will produce high friction and may lead to transfer of fragments of one surface to the other during sliding. In some machines the forces developed may be sufficient to bring the machine to a violent stop. Even if such seizure is avoided, the transfer of metal eventually results in serious damage to the surfaces, and loss of the accurate dimensions required for hydrodynamic operation.It is therefore imperative to reduce the surface forces, which immediately brings the problem into the realm of surface chemistry and physics. Suitable polar additives are incorporated in the lubricant to adsorb on, and preferably react with, the metal surfaces. The most effective materials for this purpose are fatty acids which, as we shall see, can greatly reduce friction and wear even when present as only a single molecular layer on each of the metal surfaces. This regime, known as boundary lubrication, is consequently of very great importance in bearings. In recent years the development of a new theory dealing with elastohydrodynamics has shown that hydrodynamic lubrication can be retained at much lower speeds than had previously been supposed, because the sliding members themselves deform elastically and thereby encourage the formation of an oil film between the surfaces.This can be very valuable, especially when non-polar lubricants are used, but the load-carrying capacity is considerably improved even in this regime when polar additives are used. In many applications it is not possible to establish elastohydro- dynamic lubrication, for example in a door hinge where the components are at rest for a long period followed by a slow-speed movement of short duration, which then changes direction. All these conditions tend to prevent the forma- tion of a film by viscous action. For this and many other familiar purposes, reliance is placed entirely upon boundary lubricants.As we shall see, the action of typical boundary lubricants such as the fatty acids depends essentially upon the formation of a solid film of reaction product, usually a metallic soap. Cupric laurate, (CllH&O&Cu, is a good example. The formation of such compounds, by reaction with metals via their atmo- spheric oxides, is believed to be fairly well understood. In some circumstances, for example in wire drawing and other metalworking operations, fresh metal surface is generated continuously and must be lubricated within the time of passage through the drawing die. At a speed of 3000ft/min this may be appreciably less than a millisecond. Little attention has been given to lubricant Rowe 137 10 chemistry at high speeds.As the demand for increasing production speeds continues, this may become a fruitful field for study. The problem is still more severe in metal cutting and very little is known about the detailed action of lubricants in metal cutting. Although metallic soaps are known to be very effective lubricants, their use is restricted to conditions in which the interfacial temperature is unlikely to rise above about 150 "C. It is found that breakdown of lubrication occurs at temperatures approximately corresponding to the softening temperature of the compound. Copper laurate softens at about 100 "C and sodium stearate, one of the most stable boundary lubricants, at about 250 "C. Such temperatures can be generated under apparently moderate frictional conditions because considerable energy is involved in the plastic deformation of small regions at the interface.The heat flow and temperature gradients can be calculated by the principles of physics, and spectacular improvements in performance of, for example, the mechanical seals used in chemical engineering can be obtained by suitable design for good heat transfer. To provide lubrication at temperatures in excess of 150 "C, it is common to use oil additives known as extreme-pressure (e.p.) compounds. These are usually organic compounds containing chlorine or sulphur, which are stable at room temperature but decompose when the temperature rises. The chlorine or sulphur then reacts with the substrate metal to form a solid chloride or sulphide film. The chloride shows lower friction on steels, but breaks down at lower temperatures. In the absence of oxygen, ferric chloride will lubricate up to about 350 "C and iron sulphide up to about 700 "C.Frequently oxygen does not have ready access to the local areas of contact between the sliders, but these are likely to be exposed on emergence from the interfacial zone. It has been suggested that the organic chlorides can react through the formation of HCl, and indeed dilute HCl is an effective lubricant for steel at elevated temperatures. It is however corrosive and the chloride film formed with HC1 on steel is readily hydrolysed and leads to rusting. Extreme-pressure Zubrication is thus an application of controlled corrosion and great care must be exercised to obtain the chemical attack only at the points where it is essential. One important situation in which accelerated corrosion is desirable arises in the running-in of an automobile engine.The surfaces as prepared by machining and grinding are not perfectly smooth and the mating surfaces do not exactly conform. A lubricant of relatively high reactivity is therefore used and the engine is run gently. The high spots on the surfaces will be subject to most heating and so will react most strongly with the lubricant. The inorganic films formed will be weaker than the metal, and will be relatively quickly worn away, exposing more metal for chemical attack. In this way the mechanical and chemical actions will combine to wear down the protruding regions of the surface and to improve the conformity and smoothness of the interface. It is clearly undesirable to continue the corrosive action too long, so the lubricant is replaced by a less reactive one after a short while.Nevertheless, mechano- chemical interaction is important in all lubrication and wear. Finally, when all has been done to improve the performance of a bearing by attention to mechanical design, physical properties, and chemistry of the metal-lubricant combination, there remains the possibility of metallic contact RZC Reviews 138 occurring at local areas, either because the temperature is too high for the lubricant film, or a film has failed to form. Minute protrusions or asperities are always present on real surfaces.They may be only a few microns across, but can penetrate very thin films of lubricant and weld onto the opposing surface. Considerable heat is generated in the deformation of such asperity junctions and local temperatures in excess of 1000 "C can be recorded in high- speed unlubricated sliding. If the speed is high enough the surface of a metal may even melt, despite the good thermal conductivity of the metal. This is the province of metallurgy, and much has been done to develop suitable bearing alloys, usually based on lead, tin or bronze, which will allow sliding to con- tinue even when the lubrication is impaired. It is found that some combinations of metals and alloys are less prone than others to seizure and transfer of metal across the interface.Some metals, including stainless steel and titanium, are very difficult to lubricate, probably because of the tenacious and unreactive surface films which they form in air. It is now recognized that the oxides and other contaminant films formed on metals in air play a vital part in reducing friction and wear. If these films are completely removed in a high vacuum, metal surfaces show very high friction and there is extensive metal transfer and damage to the surfaces. Even when this does not occur the oxides can seriously influence wear. A smooth, continuous oxide film will effectively reduce wear but if it is fractured the oxide fragments, being very hard, will act as an abrasive. They will tend to cut the opposing surface.In some circumstances, particularly in fretting, which is a form of wear with oscillatory motion of very small amplitude, oxide is sequentially formed on clean metal and abraded away, by earlier oxide debris, to produce more fresh metal for oxidative attack. Atmospheric water, sulphur and organic compounds also strongly influence the films formed on ordinary surfaces and thereby affect the friction. Chemistry, and particularly surface chemistry, is thus of major direct significance in friction, lubrication and wear. It is also of great indirect importance in the production and control of lubricants. The refining of crude oil to form the basic paraffinic and naphthenic lubricating oils is a large-scale chemical engineering process involving, as we shall see, a variety of solvents, surfactants and catalysts.Quality control of these fluids is mainly concerned with chemical and physical techniques. The highly purified mineral oils are, however, of relatively little use until various additives have been incorporated to improve temperature stability, oxidation resistance and other properties. For some purposes, for example in automobile engines, it may be desirable or essential to add compounds which will adsorb on carbon and other debris to prevent agglomeration. Most of these additives, as well as the specifically friction-reducing compounds, depend essentially upon chemical principles. In this survey we consider in some detail the chemistry involved in the preparation and utilization of lubricating fluids, including greases and the newer synthetic fluids.We also consider the characteristic frictional behaviour of materials of different types and of various lubricant films, attempting to emphasize the areas in which chemical research is most needed. Finally we discuss briefly the subject of wear, which despite its economic importance has received far too little attention from research scientists. We should, however, Rowe 139 recognize that in this survey we are dealing with only a portion of the whole interdisciplinary science and technology of tribology, which is slowly being recognized as a subject of considerable academic interest as well as economic importance. FUNDAMENTALS OF SURFACES The chemistry of lubrication is predominantly surface chemistry, not only for the obvious reason that sliding is an interfacial phenomenon.We shall see that most lubricants depend upon chemical reaction with the sliding members to form surface films, but even the inert petroleum oils require treatment with catalysts and other surface-active solids during refining. Lubricating emulsions are stabilized by surfactants, and many additives used for oxidation resistance, corrosion prevention, dispersion and detergency act by surface adsorption. Adsorption and clean surfaces The potential energy of an ion at the surface of a solid in a vacuum is greater than that of a similar ion within the solid body, because an equilibrium interaction is established between all the ions in a three-dimensional crystal, and this cannot be maintained at the surface. There is a smaller number of neighbouring ions at the surface, so each ion is less strongly bound and an excess free energy is stored.If a gas is admitted to the vacuum, the free energy can be lowered by adsorption. This occurs spontaneously, and the adsorbed particles are either adsorbed at localized sites or retain a mobility over the surface. In either condition there is less freedom of motion than in the gaseous state, and the entropy of the system is reduced by adsorption. Since both free energy AG and entropy A S decrease, the heat content AM decreases according to the basic thermodynamic equation : AH = TAS Surface adsorption is consequently exothermic. In single crystals the atomic arrangement at the surface will differ according to the particular face exposed, and adsorption characteristics will consequently also depend upon crystallo- graphic orientation.There has been considerable work of great interest and aesthetic appeal in recent years using field-ion microscopy, in which details of adsorption on an atomic scale have been observed. These have so far been quite unrelated to tribology and will not further concern us here beyond noting the fruitful field for research lying totally unexplored. The magnitude of the heat of adsorption will obviously depend upon the respective materials. In some instances it may be comparable with the heat of reaction. Chemisorption will then occur, which for most lubrication purposes we can regard as chemical reaction.This will be considered more fully in the next section. Frequently, however, the heat of adsorption is small, comparable to the heat of condensation of the gas. The adsorption is then described as physical. A convenient distinction, widely used in tribology, is that the adsorbate, which is physically bonded by van der Waals forces, retains its molecular identity and can be removed simply by pumping off the surrounding RIC Reviews 1 40 gas. In chemisorption there is usually some electron transfer between the solid and the adsorbate, and the adsorbed molecules are often dissociated into independent atoms. Chemisorbed material can be removed only if energy is supplied, usually by heating to high temperature. The heats of adsorption are respectively of the order of 0.1 eV (-J 1.60 x 10-19 .?) per molecule and 1-10 eV per molecule for physical and chemical adsorption.Direct friction experiments with metals in a vacuum chamber show that because of their high surface energy, there is strong attraction between de- nuded surfaces. The frictional stress developed in sliding one over another can reach a very high value, comparable to the bulk shear strength of the metal. Such conditions are however difficult to achieve, and prolonged cleaning is necessary in the highest available vacuum. Two methods have been used effectively. One is to heat the metals for about an hour at a temperature just below that at which evaporation becomes rapid.The other is bombardment with energetic argon ions, followed by annealing with electron bombardment to remove the surface strains produced. Low pressures of gas, such as oxygen at torr (- 0.133 N m-2), can cause a large reduction in the friction between such denuded surfaces. If the chamber is re-evacuated without heating the specimens, it is found that there is usually a reversible component attri- butable to physical adsorption but the major reduction in friction is due to chemisorption. Experiments of this type performed with copper and low- pressure oxygen show that the chemisorbed oxygen may survive until the temperature of the metal is raised to over 900 "C in a high vacuum. Even without deliberate admission of gases, it is very difficult to maintain metallic surfaces in a clean state after they have been denuded.The rate of adsorption of residual gas from the vacuum chamber depends initially only on the collision rate, which can be calculated by kinetic theory. For example, we may expect 1023 molecular impacts on each square centimetre every second at atmospheric pressure. There are about 1015 adsorption sites on each square centimetre of a metallic surface, so if we assume that each molecule or atom reaching the surface is captured, a monolayer of adsorbate may be expected to form in one second at a pressure of 10-8 atmospheres (10-5 torr). To maintain a surface with less than one tenth of a monolayer for several minutes the pressure should be less than 10-8 torr. Although this simple type of calculation is reasonably valid when there is little adsorption, immediately after the exposure of the surface, the situation soon develops more complexity. As the surface becomes covered, the number of available sites is reduced so that there is a continuous reduction in the accommodation coefficient, which is the proportion of the total number of atoms, colliding with a surface, that remains adsorbed.The adsorbed atoms are usually highly mobile on the surface, though they cannot easily escape. In some circumstances there are preferred sites giving a greater reduction in free energy, so that the adsorbate tends to be localized. In these areas there is a tendency for a second molecular layer to build up, and this may also happen elsewhere on the surface when the monolayer is nearing completion, especially if there is restricted mobility.Porous surfaces will provide much higher levels of adsorption by capillary condensation at high vapour pressures. This may be important in lubrication 141 Rowe because it provides reservoirs of lubricant, but we shall first consider more idealized conditions, assuming smooth surfaces and low gas pressures. It is convenient to consider surface structure and physical adsorption first. Surface structure and energy There is reason to suppose that for a homogeneous solid surface there would normally be monolayer adsorption if no chemisorption occurred. The inter- atomic forces in a solid are electrical in character but increase rapidly as the separation decreases.The relationships between attractive force and distance for primary valence bonds have not yet been properly developed, but we are here concerned more with secondary molecular forces, which are better known. An equilibrium separation between two atoms is established and can be characterized by a general equation of the type V = Ar-m - Br-n V is the potential energy of a pair of spherical atoms when their separation is 2r, and the four constants depend on the species of atoms. The first term represents the energy of repulsion and the second term the energy of attraction. It is generally agreed that the exponent n is appreciably greater than 2, but m is very much greater than n. The potential is frequently represented diagram- matically, showing the familiar potential well which gives the equilibrium separation.It can be deduced from arguments based on the work of London and of van der Waals that n should be equal to 6. Lennard-Jones suggested that the value of m is not very critical but may be taken as 12. This Lennard-Jones potential function is a convenient quantitative starting point for many theoretical studies : v = 4 E ((p)12 - (s)"i The quantity E represents the depth of the 'energy trough' or 'potential well' in erg/mol corresponding to the energy of vaporization or sublimation at absolute zero, while CT has the dimensions of length and represents half the atomic separation at V = 0, sometimes considered a 'collision radius'. For crystals the function is modified somewhat.Thus the potential for a hexagonal close-packed lattice (h.c.p.) is a simple multiple of that for the single pair : where n is the coordination number, or number of nearest neighbours, which is 12 for h.c.p. It is possible in this way to allow also for the influence of the next-nearest neighbours at r', beyond the 12 nearest neighbours of the h.c.p. atom. A calculation from the geometrical value of Y' shows that these modify the potential by only about 3 per cent. The second molecular layer underlying the surface of a condensed phase (either liquid or solid) is thus virtually at the same energy level as a layer deep within the bulk. The boundary or transition layer of the condensed phase is consequently one molecule thick.There is R E Reviews 142 experimental evidence from low angle X-ray diffraction and electron scattering which supports this conclusion. We can further see that since the potential depends directly on the co- ordination number, which is 6 for the surface ions and 12 for ions in the bulk solid of an h.c.p. crystal, about 50 per cent of the latent energy of vaporization is required to transport the ion from the bulk to the surface layer. In most other crystals this requires between 20 and 40 per cent. It is possible to develop these arguments further to obtain an assessment of the surface tension y, which is numerically equal to the work of formation of unit area of surface; since this can be regarded as the work necessary to transfer lla molecules from the bulk to the surface, a being the cross-sectional area for a molecule in the plane of the surface.A relationship can be deduced on this basis: where AEvap is the molar internal energy of vaporization, M is the molecular weight p is the density of the condensed phase N is Avogadro’s number K is a constant whose value lies between 0.5 and 1.0. This formula gives a fair prediction of the surface tension of liquids. There will however be a superimposed thermal energy at ordinary temperatures, which will make the surface distribution of molecules less dense and so the boundary tension will be reduced. A special situation must be recognized with unsymmetric molecules, which are often important in lubrication. The lowest energy state is achieved when the surface layer is so orientated that the high-intensity areas of the surface molecules are directed towards the condensed phase.This can be understood by considering the transport of a molecule from the bulk to the surface. There is therefore a strong tendency for asymmetric molecules to be preferentially orientated at a phase boundary, and the boundary tension will be correspond- ingly lower than would be expected in the absence of orientation. The homo- logous series of aliphatic acids shows this tendency. Acetic acid shows fairly high surface tension, which is nevertheless quite low relative to the cohesive energy density of the liquid. As the molecular weight increases, the surface tensions of the liquids drop rapidly to values comparable to those of the corresponding saturated hydrocarbons.Molecular orientation of surface layers and of adsorbed layers is of major importance in lubrication. There are however many complications in dealing with the surfaces of solids, to which we shall return later (pp. 145 and 147). Let us next consider the equilibrium of an adsorbed layer. Physical adsorption of gas For monolayer adsorption Langmuir derived an equation on the assumption that an equilibrium would be established between molecules approaching and Rowe 143 leaving the surface at a gas pressure p: bP vm - 1’ - ~~ 1 + bp where v is the volume of gas adsorbed at equilibrium pressure Y, is the volume required to produce a full nionolayer b is a constant, the adsorption coefficient.At very low pressures the volume adsorbed is thus proportional to the pressure, as is frequently observed, but the value of b depends upon the amount adsorbed: b = CleEI/R* where El is the average heat of adsorption c1 is a term dependent upon the entropy of adsorption R is the gas constant, and T the absolute temperature. As the surface becomes covered, both El and c1 will vary. Because surfaces are usually not uniform, the adsorption will occur first on the high-energy sites, progressively lowering the heat of adsorption, but this is counterbalanced to some extent by the attraction between the adsorbed molecules themselves, which increases the heat of adsorption. This interaction of the adsorbate molecules is of considerable importance in boundary lubrication and is believed to contribute largely to the durability of lubricant monolayers.In an elementary way the adsorbate can be regarded as a two-dimensional ideal gas, at least at low pressures. The equation of state of this gas can be written n-A = RT where n- is the two-dimensional pressure A is the area covered per mole of adsorbate. The relationship between n- and the gas pressure p of the superincumbent gas can be found from thermodynamic considerations. At constant temperature, Gibbs showed that the relevant equation could be written in the form Thus r, the number of moles of gas adsorbed on unit surface area is given by combining these equations I’ = klp.As the pressure of gas is increased it is more appropriate to use an equation of state analogous to the van der Wads equation for a three-dimensional gas. Thus we may write (rr + 2) ( A - bz) = RT. Combination of this equation with the Gibbs differential relationship leads to more complex equations which are handled by computer, after introducing RZC Reviews 144 various assumptions about the surface heterogeneity and heats of adsorption. The results are then in reasonable accordance with specific experiments. There is however evidence that in some systems, with condensable vapours, multilayer adsorption can occur at temperatures below the bulk critical temperature. Brunauer, Emmett and Teller assumed that it was possible for a second monolayer to adsorb on top of the first, and that this process could continue.Each layer was assumed to obey a Langmuir equation. The heat of adsorption for the second and subsequent layers was assumed to be the heat of condensation of the vapour, but the heat of adsorption of the first layer would in general be significantly higher. They derived an equation V _ ~ _ vm - (1 - x)(l CX - X +..>; = PIPS The relative pressure p/ps is here considered, p s being the saturation pressure of the vapour. The constant c is given by c = C2e(Ei-EL)/RT where c2 is a constant depending upon the entropy of adsorption El is the average heat of adsorption of the first layer EL is the heat of condensation of the vapour. In the usual form, this BET equation is: Many solid/vapour systems follow this equation when the relative pressure is low and it has been widely used for determination of specific surface area of solids.Experiments are commonly performed with nitrogen as the adsorbing molecule at a temperature of - 183 "C. A linear plot of p/v(ps -p) against p/ps is obtained. The slope and intercept then give the appropriate values of c and Vm. The volume of a full monolayer on a known surface is first determined, and from this, assuming spherical molecules in hexagonal close packing, the area of a single adsorbed molecule; for example the area of an adsorbed nitrogen molecule at - 195.8 "C (the boiling point) has been found to be 16.2 square angstroms (0.162 square nm). Consequently the adsorption of nitrogen can be used to measure an unknown surface area.The Langmuir equation can thus be considered as a special case of the BET equation, applying to a single molecular layer adsorption. Other theories have been proposed and there has been considerable discussion about the nature and even the existence of long-range forces at a solid surface. As in all surface studies, the results obtained are critically influenced by small amounts of impurity. For example, it has been shown by direct weighing with a micro- balance that, on thoroughly outgassed platinum surfaces, hydrocarbons adsorb as monolayers. Water vapour adsorbs in only two molecular layers even at pressures near saturation. When the same surfaces are cleaned in a conven- tional manner enough water to form as much as 20 molecular layers may be adsorbed.It is concluded that even with condensable vapours monolayer adsorption is observable on uncontaminated surfaces, as would be expected 145 Rowe from the above consideration of the Lennard-Jones potential function. However, the surfaces normally encountered in the laboratory are always contaminated and consequently heterogeneous. In particular they may well contain hygroscopic impurities which locally increase the adsorption and condensation. The platinum surfaces mentioned above as showing adsorption equal to the weight of 20 monolayers appeared to have only one layer when examined by a polarized light technique which was unaffected by the high local concentrations of water.Local chemical reaction will of course also increase the adsorption. There is scope for development of adsorption thermodynamics in conjunc- tion with experiments on well-characterized surfaces. It is clear that the nature of both the adsorbate and the adsorbent determines the interaction energy in monolayer adsorption and it is probable that this is the doininant feature of true physical adsorption. In general there are three energy terms to be considered ; the non-polar interaction considered above is ubiquitous and usually contributes the major term. In addition there may be interaction between the electrostatic field of the surface and the dipole moment of the adsorbed molecule, and also polarization of the adsorbate by the surface.It may be hoped that studies of this subject in association with field-ion micro- scopy will be fruitful, especially when the less refractory metals can be studied. Such work is likely also to provide information about surface migration of adsorbed molecules, which may be important in lubrication; at present this aspect is almost entirely neglected though surface self-diffusion has been studied in metals. There is further evidence that external polarization can greatly influence adsorption and also cause significant changes in friction. The possibilities of developing this into practical lubrication improvement have also been totally ignored. Chem isorp t ion We have seen that physical adsorption will occur spontaneously on all surfaces.Chemisorption may involve an energy barrier, so that activation energy must be supplied. In such a system only physical adsorption will occur at low temperatures; but as the temperature is raised chemisorption will occur at an increasingly rapid rate, until eventually an equilibrium is established between the rate of arrival of molecules at the surface and the rate of desorp- tion. When the adsorbate forms a stable chemical compound with the substrate it is likely that rapid chemisorption will occur. Most metals react readily with oxygen, and it is thus to be expected that any metal surface exposed to air will quickly form a chemisorbed layer of oxygen. The activation energy is so low that adsorption appears spontaneous at room temperature. As we have seen in considering the nature of surfaces, this is likely to reduce the surface energy very considerably, and the oxygen layer will in general be only one molecule thick because the forces fall off very rapidly with distance (probably l/r6 for single atoms, but l/r4 for extended surfaces). In chemisorption however this is a temporary situation.Metal ions will diffuse into the surface layer and react with the adsorbed ions. At first a non-stoichiometric compound will be formed but as further diffusion occurs the natural oxide will build up. This will have a RIC Reviews 146 distorted structure in the first layers, determined by the epitaxial growth, but rapidly assume its normal crystal form, as shown by low-energy electron diffraction studies.It appears that more complicated situations may also occur. For example, if oxygen is chemisorbed on ordinary iron and sub- sequently desorbed by heating to about 1000 "C in avacuum, analysis of the desorbed gas shows mainly CO, arising from reaction with the carbon in solid solution in the iron. There is also evidence that hydrogen adsorbed on nickel is in the form of two separate H atoms which can migrate independently on the surface. Such actions are probably involved in catalysis which, as we have seen, is of considerable importance in the production of lubricating oils. The whole subject of catalysis is of great interest and may have much wider significance in tribology than has been generally recognized. Obviously the general chemical properties are important but it appears that many other factors are involved. The geometric spacing of the surface atoms is believed by some workers to influence the hydrogenation of aromatic hydrocarbons, some crystal orienta- tions being more favourably matched to the benzene ring.It is however possible to explain many of the same results on the basis of the electronic rather than atomic structure, especially correlating with the number of vacancies in the d-orbitals for each atom. Alloying can alter this, and some evidence has been obtained in favour of the electronic theory. One of the most promising tools for the study of catalysis and chemisorption is the infrared spectrograph. Quite unexpected radicals can be detected after the adsorption of organic substances on metals.Even an inert gas such as pentane is apparently dissociated on a nickel-silica catalyst. Recently there has been an interest in the catalytic properties of semiconductors and there may be a correlation with electrical conductivity. As many of the chemisorbed oxide layers, which are formed naturally on metals, exhibit semiconductor properties, this approach may be relevant to the adsorption of organic molecules on previously exposed metal surfaces. It will be clear from this brief discussion of adsorption that little enough is known about the chemical properties of pure surfaces, and that a very great deal has to be learnt before reliable interpretations can be given for the adsorption on practical pieces of machinery. These are heterogeneous in all senses ; chemical, metallurgical, topographical, crystallographical, mechanical and thermal.In addition it is probable that in the critical regions of close approach, equilibrium is never established. The performance may depend to a very large extent on time of exposure to a potential adsorbate in the lubricant, and on the accessibility of the lubricant. Mechanical and chemical interactions. To offset to some extent this apparently intractable problem of building up a theory of surface chemistry in lubrication, there is interesting evidence that tribological investigations may elucidate some aspects of chemisorption and catalysis. In most sliding processes plastic deformation occurs and there is extensive movement of dislocations at and below the surface.High-energy sites are thus created at the surface, and recent work suggests that there can be preferential chemical action at these sites. One of the most severe operations in metal forming is cutting, which involves very Rowe 147 large plastic strain over a highly localized region. It has been shown that there is greatly enhanced chemisorption in this region. A large area of highly- deformed virgin surface is also produced during cutting and this exhibits high chemical activity. There is evidence that both direct and catalytic reactions can be accelerated and even initiated by the mechanochemical activation associated with cutting a metal under the appropriate fluid.Work along these lines, together with the work now being done on stress-corrosion, may feed back valuable information about the influences of active defect centres, dislocations, vacancies and stacking faults, on interfacial activity. There are several special surface-chemical phenomena which remain to be integrated with surface theory, or may themselves prove to be helpful in developing theory. The Russell effect has long been known; hydrogen peroxide can be detected when a surface is freshly exposed in the presence of moist air. More recently, Kramer found that electrons of appreciable energy were liberated when a fresh surface was exposed to light or other energy source. These exo-electrons are probably associated with 0 2 - and OH- radicals.Both these effects, which may be closely interrelated, are thought to be associated with local deformation energy. Indeed it is possible to produce an autographic film blackening with the radiation from deformed zinc crystals, showing the regions of slip. Finally, there is a considerable body of evidence from Russia, formerly largely ignored by others, but now gaining acceptance in the USA, concerning the Rehbinder effect. Fundamentally this deals with the influence of surfact- ants on the mechanical properties of metals. In the presence of a surface- active compound such as oleic acid, the tensile yield strength of a single crystal is significantly reduced. One explanation is that the adsorbate, by lowering the surface energy field, facilitates the emergence of dislocations and thus allows the metal to flow plastically at a lower stress level.The reverse effect of strengthening a single crystal by forming an oxide skin is well known. The relatively strong oxide forms a dislocation barrier. A different explanation is that the adsorbate electrically influences the current carriers (defect centres in oxides) which in turn interact with the dislocations and so modify the mechanical properties. There is some evidence that the effect is large enough to be significant in surface stressing even of polycrystalline metals, where the grain boundaries, and the impurities contained in them, already contribute major hindrance to the passage of dislocations. It seems probable that these effects have an important influence on the sliding of metals, and it is to be hoped that there will eventually be a syn- thesis of the predominantly Russian interpretation of boundary lubrication with the low shear-strength adsorbate film theory widely accepted in Europe and America and the more recent defect centre work.Surface topography Although the topography of a surface is not a chemical feature, it is possibly related to the surface adsorption characteristics and is certainly of major importance in lubrication theory and practice. All considerations of friction, lubrication and wear must take into account the geometric irregularity of surfaces. Casual inspection affirms that no surface is completely smooth and RIC Reviews 148 the more detailed the inspection the more obvious it becomes that the surface asperities, as they are usually called, must influence interfacial movement.The smoothest surfaces ever produced are those obtained by cleaving mica. It can be shown by optical interference that these may be flat, over an area of several square millimetres, to a single molecular level. Studies of such surfaces nevertheless lend support to the general statement, for it is found that the perfectly smooth surfaces do not obey the normal laws of friction. There are many methods available for recording or measuring surface topography. Simple interference microscopy conveniently provides contour lines at height intervals equal to one-half the wavelength of the light used, often A _N 500 nm.Improved resolution can be obtained by a multiple-beam interference technique which provides the same contours but greatly sharpens the bright lines. With care it is then possible to detect changes in level of 2 nm or less, and by a refinement using white light and observing colour tints the limit may be reduced to 0.3 or 0.4nm. Attempts have even been made to measure the thickness of an adsorbed monolayer of water by this method. Usually this degree of refinement is not necessary and some more con- venient technique can be used. A diamond stylus with a fine point, about 1 pm radius of curvature, is often used industrially. This is moved steadily across the surface being examined and the vertical displacements are amplified electronic- ally by 10 000 or even 40 000 times.The horizontal movement is also ampli- fied, by 20 or 100 times, and the results can be plotted automatically by commercial equipment, or they can be averaged to give a single value ofpeak- to-valley height or centre-line average (c.1.a.). The latter is favoured in the UK, and represents the average departure of the surface profile from a mean line. Some typical values for common mechanically-produced surfaces are given in Table 1. Table I: Surface Roughness of Metal Surfaces Mechanical preparation method Lathe t u rned Fine ground Pol is hed 0.5-5 0.1-0.2 0.05-0. I Even the most carefully prepared surfaces are thus very rough on a molecular scale. The c.1.a. is about one quarter of the peak-to-valley height which is thus about 200 nm even for a polished surface.Although there has been great emphasis in the past on the height variation of surfaces, it has also been recognized that in friction and wear the amount of contact is important. As we shall see later, the highest peaks will deform until the yield stress of the metal is nowhere exceeded. During the actual sliding process the surfaces will consist of more or less well-defined plateaux, at least if one slider is much harder than the other, as is usual in bearings. The surface topography is consequently sometimes described by an Abbott- Firestone type of graph which shows the percentage bearing area as a func- tion of height above some datum. This is produced by taking imaginary 1 49 Rowe horizontal cuts at sequential vertical levels and assessing the ratio of solid to space at each level.In some circumstances a surface could follow such an idealized configuration change by a process of abrasive wear, but more generally the shape would be considerably influenced by plastic flow, so that the solid is deformed as well as removed. Recently programmes have been written for digital computers which allow not only bearing area but also peak height, trapped volume, peak frequency and various other parameters to be evaluated from a numerical surface profilometer output. The significance of these data to tribology is believed to be considerable, but there has as yet been very little interpretation of them. There appears to be a large field for study here, relating changes in surface topography to viscosity and compressibility of the lubricant and possibly to the Rehbinder effect of chemical surfactants. There seems to be little doubt that over the regions where the surfaces approach to within about 10 nm their chemical interaction with the lubricant is very important.A recent development in surface profilometry allows the topography to be recorded at a magnification of 1000 000 times. The edge of a monolayer deposit of gold can be detected with such an instrument. Although this equipment is becoming highly sophisticated and may be of great importance in tribology, it should be recognized that the record from a stylus moving linearly represents essentially a two-dimensional section.There is likely to be further development in faster scanning to provide three-dimensional information. Advanced forms of interference microscope will also probably be valuable for contour determination in this field. A further two-dimensional presentation which has been widely used in research is provided by cutting the specimen at a small angle (5" or 10") to the surface and preparing this section metallographically. Such taper sections have been used to study metallurgical changes occurring during sliding as a result of local temperature changes. They can also be used to examine certain specialized tribochemical effects, such as nitrogen absorption from the atmo- sphere at a sliding surface, which can lead to drastic metallurgical changes because of the short high-temperature flashes.It has been reported that nitrogen martensite can be formed in this way with a hardness of over 900 kg mm-2, from a carbon steel whose hardness is only 200. Steels can also be hardened by carbon produced by decomposition of organic lubricants. Conversely, stainless steel may form chromium carbide which seriously weakens the structure, causing cracks. Adsorption can be used to measure surface area. This method is commonly used for determination of the surface area of powders, the results being calculated with the help of the BET equation and a knowledge of the equiva- lent area of an adsorbed molecule of nitrogen or other vapour. In tribology this method is valuable, for example, in measuring areas of dispersion particles, usually with an electromagnetic sorption balance, though various other techniques are used.For measurement of the surface area of large objects it becomes inaccurate. Even when the specimen is light enough to be weighed in comparison with the film of adsorbate, the conclusions may be seriously in error, for example if isolated hygroscopic impurities are present. For sliding surfaces generally, in laboratory experiments or in engineering RIC Reviews 150 practice, average measurements of surface area are of little value. The mean slope of the surface asperities is seldom greater than a few degrees on a well- prepared specimen, and the total surface area is consequently unlikely to exceed the projected area by more than 10 per cent.Nevertheless, these small asperities are often of paramount importance in friction, lubrication and wear. The area of intimate contact between two sliding surfaces is frequently less than one-thousandth of the projected area of the whole contact. Detailed study of the surface topography is therefore very important and great care must be exercised in any attempt to correlate theories of solid surface structure or of physical and chemical adsorption with friction measurements. Yet, partly for this reason, there is great scope for careful research with this objective. PREPARATION AND PROPERTIES OF MINERAL OIL LUBRICANTS Herodotus, who lived from 484 to 425 BC, described an ancient method of producing bitumen and a lighter oil from petroleum, but there is evidence of deliberate use of lubricants from the earliest historical times.The first lubri- cants (other than water) were probably animal fats, which still play an important part in lubrication today. They were used widely in situations requiring retention of the lubricant over long periods. A chariot dating from the 15th century BC has been found with fat still on the axle. The advantage of a pad of lubricant in such inaccessible places as the trunnions of church bells was soon appreciated, but it was realized that for lubrication over short periods it was convenient to use fluids which could easily be poured onto the surfaces. Suitable liquids could be obtained from a variety of plants, especially from seeds.A great many of these vegetable oils and animal fats are still in use in modern industry though mainly as additives to mineral oils. Palm oil is considered the best lubricant for cold rolling of thin steel sheet, and tallow is excellent for drawing large steel bars to reduce their diameter and improve the surface finish. For many years castor oil was used almost exclusively in air- craft engines, though this was at least partly due to its resistance to dilution by petrol. The first main departure from ancient traditions in lubrication is associated with the invention of the steam engine during the 18th century. This introduced rapidly increasing demands for higher speeds, loads and temperatures, which have continued to rise and to dominate lubricant development ever since.Steam-engine users first turned to mineral oils, which were known to have better thermal and oxidation stability than the animal and vegetable oils, but these properties could not be fully exploited until the refining process had been sufficiently developed to produce a pure hydrocarbon oil. Mineral oils are now the most widely used lubricating fluids. The chemical industry has thus been a keystone in modern lubrication technology. More recently, especially since the outbreak of the World War 11, the rapid advances in all branches of technology have created demands for lubricants to operate under extreme conditions not even considered a few decades ago. To a large extent these requirements are being met by a revolution in the lubricant industry, greatly assisted by developments in polymer chemistry, and whole new ranges of synthetic lubricants are now being produced.These and the wide variety of Rowe 151 solid lubricants for even more exotic conditions of extraterrestrial space and nuclear radiation depend for their existence entirely upon chemical expertise. These very recent developments are considered later (pp. 168-173). In this section we are concerned with the production of highly refined mineral oils, which are still commercially by far the most important group, and the treatments by which they are converted into general-purpose or specialized lubricants. Production of mineral oils The basic natural product from which nearly all commercially important lubricants are developed is crude oil.Crude oils are complex mixtures contain- ing compounds which have an almost continuous range of boiling points, from petroleum gas to bitumen. They are however predominantly hydro- carbon fluids with molecular weights between about 200 and 1000, the composition of which varies according to the geographical source. Pennsyl- vania crude oil, for example, is mainly paraffinic while Venezuelan crude is mainly naphthenic. Much of the lubricating oil used in Britain is derived from Middle-East crude which contains both major types. In general, oils with high paraffinic content tend to show less viscosity reduction with increase in temperature but are somewhat more expensive than the corresponding naphthenics. Current Middle-East production of crude oil is about 400 million tons/a, which represents nearly one quarter of the world production.Distillation. The first important stage in a refinery after the crude oil arrives by tanker, some of which now deliver 100 000 tons at a time, is distillation at atmospheric pressure. This removes all the more volatile constituents such as kerosenes and motor fuels, which are condensed and sold separately after further processing. The remaining oil is generally known as atmospheric residue. It is not possible to heat this residue beyond about 350 "C because it would then begin to decompose, or crack. Further distillation is therefore carried out under vacuum at moderate temperatures.The distillate from this second process is usually divided into several fractions, according to require- ments, which can subsequently be blended to produce lubricating oils with the desired viscosity characteristics. The residual fluid, known as vacuum residue, can be used as a basis for high-viscosity lubricating oil. Further purification of both distillate and residue is however necessary in the preparation of lubricants with satisfactory thermal properties. PuriJication. Vacuum residue is first purified by removing asphaltic material. The oil and wax components are dissolved in a suitable solvent, usually liquid propane, leaving the asphaltic residue. The propane is recovered for re-use, leaving a product known as bright stock which is a widely-used lubricant base.All the vacuum distillates, as well as the bright stock, are likely to contain aromatic hydrocarbons which are less stable than the aliphatics and are liable to form an insoluble sludge, or to oxidize to varnishes during use. These are therefore removed, preferably by solvent extraction. Phenol is a suitable solvent which can be stripped easily and used again. As well as possessing high-temperature stability, most modern lubricating RIC Reviews 152 oils are expected to function satisfactorily at temperatures below 0 "C. It is therefore necessary to remove long-chain paraffinic compounds which solidify at or slightly below ordinary atmospheric temperatures. Such paraffin waxes are normally soluble in the oils but may precipitate when the temperature falls, increasing the apparent viscosity of the fluid.They are removed by treating the oil with mixed solvents and chilling the product to freeze out the waxes. Again, of course, the solvents are reclaimed for further use. The final stage of purification is usually a treatment with finely-powdered clay to neutralize the remaining acidic constituents. This also improves the colour of the oil. The stability of the oil may be further improved by mild hydrogenation over a catalyst. Before being released for use the oils are subjected to various tests. A recent development for special purposes is known as super-refining. This greatly improves the low-temperature properties, so that for example the viscosity at -55 "C may be only one-half of that shown by the normally refined oil.Quality control. Different oil companies and different consumers may specify certain tests which are related to final performance or serve to maintain a uniform quality. Most of the tests are empirical and some still in current use are archaic and can even be misleading. Although considerable efforts have been made to standardize and rationalize test procedures, their relationship to field performance is often obscure. The Institute of Petroleum annually publishes Standards for Petroleum and its Products, which are continually revised. The most important characteristic of a mineral oil for use as a lubricant is its viscosity, provided that the stability of the oil is satisfactory.Viscosity is defined as the stress required to shear unit thickness of fluid at unit velocity. It is usual to quote the kinematic viscosity v in centistokes. It is found from the ratio of the dynamic viscosity 7 (centipoise) to the density p (g/cm2). In SI units 7 is expressed in Newton seconds/(metre)2, (1 N s m-2 = 10 poise) and v in (metre)z/second (1 m2 s-1 = 1 0 6 centistrokes). The viscosity of mineral oils is therefore carefully measured as a control of quality. A standard- ized apparatus, usually of the capillary-flow type, is used at specified tempera- tures. Various other viscometers of empirical design have been widely used to measure viscosity in Redwood seconds, Engler degrees etc. but it is to be hoped that these will become obsolete.In addition to the viscosity at a given temperature it is usually important, for example in an automobile engine, to know the variation in viscosity of the fluid with temperature. This is very commonly quoted as the viscosity index (v.i.), which compares the decrease in viscosity over a given temperature range with that of two oils rated respectively at 100 and 0. Since it is desirable that viscosity should change as little as possible, the value 100 represents the reduction in viscosity with increasing temperature found for the best oil available when the scale was introduced. Thus higher v.i. indicates relatively smaller change in viscosity. It should be noted that no oils exhibit constant viscosity, but many modern oils with additives, and especially some synthetic oils, show much less variation than the original 100 oil.A supplementary scale (v.i. extended) has been introduced for these oils, but the system has little to recommend it. Not only is it more precise to state the kinematic viscosity at selected temperatures; there is Rowe 153 11 increasing recognition that variation of viscosity with both temperature and pressure is important in many applications. Two common physical tests which indicate the degree of wax removal from an oil are thepour-point test and the cloud test. The pour point is defined as the lowest temperature at which an oil will flow under standard test conditions. If the wax constituents have not been completely removed, the pour point will be abnormally high.The oil may then be unsuitable for use where low atmo- spheric temperatures are likely to be encountered. Obviously the pour point may be high because the oil has a high viscosity, so this test in isolation does not necessarily show inadequate dewaxing. A more definite indication is given by the cloud point which is the temperature at which a cloudy wax precipitate can be seen to form as the oil is cooled. Chemical refining is sometimes indicated by the carbon residue and ash content of the oil. The carbon residue is determined by weighing the carbon produced by heating a specified quantity of oil and burning the vapour under controlled conditions. If the organic material is completely burnt there may be an inorganic ash residue. For a well-refined oil this is usually negligible, so the test is useful for detecting impurities.It should, however, be recognized that some oils, particularly detergent motor oils, may show a relatively large ash content which is not at all detrimental to their performance. Other detergent additives are organic polymers which are ashless but contribute to the carbon residue. If any ash is formed it will normally be weighed with the carbon and thus increase the apparent carbon residue. Care must clearly be exercised in specifying these values for purposes other than routine control. For many purposes the acidity of an oil is a measure of the adequacy of refin- ing. This can be determined by titration and is usually recorded as the number of milligrams of potassium hydroxide required to neutralize the acids present in one gram of oil.For well-refined oils this neutralization value is usually negligibly small, less than 0.1 mg KOH/g. During use the acidity increases as a result of oxidation so the test may be used to indicate deterioration in service. Some detergent oils are naturally alkaline, and the degree of alkalinity can be expressed as mg KOH equivalent to the alkali present in one gram of oil. Certain physical tests can also be carried out. The refractive index n, density d and mean molecular weight M can easily and quickly be determined. From this so-called n-d-M analysis other parameters can be deduced, for example the viscosity/temperature/pressure variations. Service tests.Apart from the tests described above, which are mainly used for routine quality control, it may be desirable to conduct tests more directly related to the operating conditions. The distinction is not clear, since both viscosity and pour point can be regarded as service parameters. Flash-point and $re-point tests are used to indicate possible hazards. The flash point is the temperature at which sufficient vapour is formed to burn if ignited in air in a specified apparatus. At the fire point the liquid itself can be ignited. Both these temperatures are lower than the spontaneous ignition temperature and are therefore of greater practical significance. It is neverthe- less unwise to rely entirely on these data, since the flash point may be seriously lowered if the oil becomes contaminated with volatile compounds.RIC Reviews 154 Various corrosion tests are frequently performed and it is important to bear in mind the common possibility of electrolytic corrosion. Most lubricant applications involve the presence of dissimilar metals because, as we see in the next section (p. 162), dissimilar pairs are more resistant to lubricant failure. It is also probable that most lubricants will eventually become contaminated with moisture, leading to ionization. In some applications it may be necessary to control the moisture content of the lubricant or even the atmospheric humidity, but this is usually far too expensive to be attempted. All simulative tests are thus to be interpreted with caution.This becomes still more important when measurements of friction and wear are made. A wide variety of standard machines is available for evaluation of the coefficient of friction, breakdown load and speed, or amount of wear. These machines are normally used with lubricants containing various additives ; they are discussed later (pp. 164-165). We shall first consider the final stage of lubricant production, which is the incorporation of these additives to provide special properties. Chemical additives for lubricating oils In a plain bearing operating under the design conditions with clean oil in full supply, the load-bearing capacity and the frictional drag are both determined by the geometric dimensions, the speed and the fluid viscosity.In this ideal situation a full fluid film is maintained which supports the journal out of contact with the bearing bore, and there is no wear. To obtain maximum efficiency the lubricant would then be chosen to have the lowest viscosity compatible with load-carrying at the appropriate speed. Even air can be used if the speed is sufficiently high. Unfortunately all mechanisms must start and stop, so protection must be given to the surfaces which will then come into contact. In general, mineral oils are unsatisfactory when the interfacial film is very thin, so boundary or extreme-pressure additives need to be incorporated. Many mechanisms, particularly those which involve oscillation or continuous low-speed sliding, do not allow a full hydrodynamic film to form at all.These rely entirely on the additives which can provide low friction and low wear. It is perhaps not immediately apparent that the lubrication conditions are more severe for a local delivery-van than for a motorway express coach travelling at a high but fairly constant speed. The express coach will depend to a larger extent on the viscous properties of the oil, and it is important that the viscosity should not vary unduly with temperature. Excessive viscosity at low atmo- spheric temperatures will increase the drag and power loss, and may cause difficulty in starting the motor. Insufficient viscosity at high engine tempera- tures will reduce the load-carrying capacity of the bearings and may lead to seizure. In recent years there has been an increasing interest in v.i. improvers, which reduce the natural variation of viscosity with temperature, for both commercial and private vehicles.Ro we Apart from these additives which directly influence frictional behaviour, many others may be valuable in special circumstances. Thus it may be important to reduce corrosion of the machine or to prevent foaming or low- temperature gelling of the lubricant. The useful life of many lubricants is limited by oxidation, especially when exposed to air at high temperatures. In 155 addition, oils may become contaminated, particularly when used for internal- combustion engines, and it may be necessary to keep the engine and oil-ways clean by using suitable detergent and dispersant compounds.All these additives are essentially chemical products and may depend on surface chemical properties. The additives are discussed individually, but in practice it is very important to ensure that mixed additives are compatible, with no mutually deleterious effects. Viscosity index improvers. All mineral oils show a large decrease in viscosity with temperature. Even a light paraffinic oil of v.i. 100 can change in viscosity by a factor of five between 45 and 100 "C, and the variation is greater for more viscous oils. The cheaper naphthanic oils have much lower v.i. For example a naphthanic bright stock of kinematic viscosity 4000 CS at 45 "C has a viscosity less than 100 CS at 100 "C. Slight improvements can be obtained by additional refining, but this is prohibitively expensive.Additions of up to 10 per cent of v.i. improvers are now quite widely used. These are polymers of high molecular weight (10 000 Staudinger units or more), usually polyisobutylenes or poly- methacrylates such as the copolymer lauryl methacrylate : n Some nitrogenous copolymers have a dual role as v.i. improvers and dis- persants. These additives may be soluble in the mineral oil, but are more likely to be present as a colloidal dispersion. Because of their large molecular size, such polymers increase the apparent viscosity of the oil at room tempera- ture; but the effect of temperature on the additive is small, so there is a much greater proportionate increase in apparent viscosity at high temperatures.The v.i., measured on the extended scale, may be increased to 130 or more. A characteristic of lubricating oils with polymer additives is that they are thixotropic and do not exhibit Newtonian flow ; the apparent viscosity depends upon the rate of shear. When speed is increased the apparent viscosity falls, but this is a temporary effect which is immediately reversed when the rate of shear is subsequently reduced. It can be advantageous since the high rates of shear are generated at high speeds for which, as we have seen, the load-bearing capacity in a hydrodynamic bearing is improved and consequently high viscosity merely increases the frictional resistance. At higher rates of shear the long polymer molecules may be degraded into shorter components with a consequent irreversible reduction in both viscosity and viscosity index of the oil.Oxidation inhibitors. All mineral oils will react with oxygen when exposed to air at temperatures above about 100 "C. In automobile engines such tempera- tures can frequently be reached and the oxidation is also accelerated by the RIC Reviews 156 vigorous agitation of the oil and by catalytic action of the copper and iron present. Lead from the fuel may also appear in significant quantities in the sump. Oxidation inhibitors, also known as antioxidants, are therefore widely used, especially in engine oils, even though the oils may be quite stable to oxidation under simple conditions. The concentration of the inhibitor may be between 0.5 and 5 per cent.The initial oxidation products of paraffinic oils are high molecular weight compounds which are soluble in the oil but will cause a significant increase in its viscosity. Aromatic hydrocarbons, and particularly any animal or vegetable oil additives present in the final blend, are likely to polymerize to form a gum or a harder compound usually described as a varnish. This can cause the valves of the engine to stick, and may contribute to the formation of an insoluble sludge. Oxidation of the mineral oil can produce corrosive organic acids, which can attack the metallic engine parts, especially copper-lead bearings. It is considered probable that oxidation inhibitors function in two ways. Some adsorb onto metallic surfaces, thereby reducing the catalytic activity.Others decompose the organic peroxides which are formed as intermediate oxidation products. The former are of course also useful as corrosion inhibi- tors. Many of the commoii oxidation inhibitors contain sulphur or phos- phorus. Sulphurized alkenes, formed by direct attachment of elemental sulphur at an unsaturated bond, have been extensively used. The reaction product of sulphur with sperm oil, a stable fatty ester, was one of the earliest lubricant additives and is still used. Sulphur compounds with terpenes, for example sulphurized dipentene, are found to be effective : Various other complex organic compounds are also used. These are capable of decomposing on a metal surface at elevated temperatures to produce metallic sulphides, and can therefore also show extreme-pressure lubricating properties. The more reactive e.p.additives, including elemental sulphur dissolved in oil, are too corrosive for use as adsorbed oxidation inhibitors. Phosphorus is also too reactive if simply dissolved in the oil, and causes serious corrosion of non-ferrous metals. Oil-soluble organic phosphorus compounds such as tributyl phosphite are however effective antioxidants : CH3-(CH2)3-0 C H 3-( Tr i -n - b u t y I p h 0s phi t e C H &-0- \P / CH3-(CH2)3-0 / Organic compounds containing both sulphur and phosphorus have proved very effective as oxidation inhibitors in crankcase oils. The most common are the metallic dithiophosphates, which can be prepared by reaction of phos- phorus pentasulphide with an alcohol or phenol and subsequent reaction of' Rowe 157 the resulting acid with a metallic oxide, thus : 4 ROH + P2S5 + 2 [ R-o\P//S ] + H2S \SH R-O/ 0-R R-0 S S \,p/ + H2O \O-R R-0’ R-0 ‘SH \pH / \ S-Zn-S’ Zinc dialkyl and diary1 dithiophosphates are widely used. For less severe applications, such as those encountered with turbine and hydraulic oils, where the temperature is unlikely to exceed 100 “C, oil-soluble amines and phenols have proved effective.Dibutylmethylphenol is a common additive of this type. Corrosion inhibitors. It is convenient to consider corrosion nhibitors separately, with particular reference to the metals involved, though in general both oxidation-inhibiting mechanisms hinder corrosion. Reduction of surface reactivity by adsorption of surfactants is a recognized function of corrosion inhibitors, while another group of compounds will neutralize the acidic products of oxidation which are the corrosive agents in lubricating oils.Usually ferrous components are fairly resistant to corrosion by these acids but they are highly susceptible to rusting. It is practically impossible to prevent the ingress of atmospheric moisture, and again the problems are increased in automobile engines, because of the wide temperature fluctuations and of the presence of moisture in the combustion products. Numerous rust-preventive treatments have been devised for steel surfaces, such as the conversion coating by reaction with zinc manganese phosphates, but we are here concerned only with the oil additives used for corrosion protection. Barium and calcium sulphonates of high molecular weight are effective and widely used.They can be prepared by substitution from sodium sulphonates made by acid treatment of appropriate lubricating oil distillates. These salts are polar and readily adsorb on iron and other surfaces. They also form an emulsion with water, helping to prevent corrosive attack, but they slightly encourage oxidation of the oil and so require further addition of an antioxidant. Sodium sulphonates, often used together with fatty compounds, are useful for short-term applica- tions such as occur in metalworking processes and metal cutting, but are themselves susceptible to oxidation.A special form of lubricant-corrosion inhibition is required in metal cutting, where a major function of the fluid is to cool the cutting tool. Water has approximately twice the specific heat of mineral oils, better thermal conductivity and a higher latent heat of vaporiza- tion. As there is no possibility of hydrodynamic lubrication in metal cutting, high viscosity is of no importance. Water is therefore an excellent base fluid. With the addition of sodium nitrite as a rust inhibitor it is a good lubricant for grinding. The commonest of all cutting fluid groups is the so-called soluble RIC Reviews 158 oil, which is a colloidal suspension of oil in water, stabilized by addition of an emulsifying agent, a sulphonate.This fluid is not inherently corrosive, but may have rust-inhibitors included to increase protection and as a safeguard when the emulsion deteriorates. It should be recognized that surfactant corrosion and oxidation inhibitors will compete with other surfactants, such as those used to impart boundary and e.p. lubrication properties. Although some oxidation inhibitors may show mild e.p. properties, great care must usually be taken with mixed additives to ensure that the components are not mutually deleterious. The second type of common corrosion-inhibiting additive functions by neutralizing the organic acids formed by low-temperature oxidation. Various alkaline-earth compounds can be used, and in recent years it has been found convenient to use ‘overbased’ phenates and sulphonates which are otherwise included for their detergent properties, discussed in the next section. Detergents and dispersants.A large group of additives, again primarily de- veloped for automobile lubrication is now available for maintaining oil in a clean condition. It is convenient to distinguish two major types, though both function by adsorption onto solid particles. Detergents are then classified as the additives used to maintain particles of carbon and other materials in suspension in hot oil. If a detergent alone is used a ‘cold sludge’ may accumu- late when the oil is cold. In an internal combustion engine this usually contains soot, various oxidation products of the oil and its additives, metallic wear debris, water and probably lead from a high-octane petrol. If the latter is present the grey mass may be known as lead sludge.Additives capable of dispersing this cold sludge are known as dispersants, though there is not complete agreement about the terminology. The detergents used in lubrication technology are of course soluble in oil, and are to be distinguished from the familiar water-soluble domestic deter- gents. The action of lubricant detergents is to adsorb on the solid particles formed in the oil, lowering the surface energy at the interface between liquid and solid and thus preventing aggregation. Barium and calcium salts of petroleum sulphonic acids have already been mentioned as effective corrosion inhibitors because of their surface adsorption on metals.0 I I 6a Barium petroliurn sulphonate 0 I I Some additives themselves combine the functions of detergent and anti- 159 Ro we oxidant. Barium amyl phenol sulphide is an example: 0-Ba-0 I If the alkaline-earth phenates or sulphonates are heated over a catalyst with an excess of the metal base they can acquire much more than the quantity necessary to form a stoichiometric compound. Such ‘overbased’ detergents can provide a dual action by neutralizing also the organic acids produced by mild oxidation of the oil, yielding an oil-soluble salt instead of allowing degeneration to proceed further to give an insoluble resin. When discussing quality control tests, it was mentioned that the ash-content of an oil could be misleading.The ‘overbased’ detergent/antioxidants have a high inorganic content and since they may be present in quantities of several per cent in the oil, the residual ash will be considerable. In fact, many deter- gents and many corrosion inhibitors have a significant inorganic content. The ash content test is therefore meaningless in its normal form. A modified test can be used, determining sulphated ash, to reduce loss of volatile inorganic compounds. For some purposes a high ash content is thought to be undesirable. Ammonium-based corrosion inhibitors can then be used, and considerable effort has been expended on the production of ‘ashless’ detergents. These are polymeric materials which are usually somewhat less effective than the organometallic compounds mentioned, but can show useful dispersant proper- ties not possessed by these compounds.The specifically dispersant additives which have been developed in recent years are copolymers, which contribute a surface-active agent to adsorb strongly on the polar sludge particles, preventing agglomeration, and an oil- soluble hydrocarbon monomer or polymer of lower polarity. Nitrogen- containing methacrylate-ester monomers and various other organic nitrogen compounds can be used to provide the surface activity. These additives are naturally ashless. There is also considerable interest in additives which combine detergent and dispersant properties. Acrylated amines appear to have promise in this field, and there is little doubt that further progress will be made in the chemical development of multipurpose additives.Because the dispersants are long-chain polymers they tend to increase the viscosity of the oil and to improve the viscosity index when present in sufficient quantities. Other examples of multiple function have been mentioned above. Boundary and extreme-pressure additives. Practically all lubricating oils contain additives specifically intended to improve sliding when the hydro- dynamic contribution is weak or absent. These can conveniently be divided into two categories : Boundary additives to provide the lowest frictional resistance, and extreme-pressure additives to function at higher local or general temperatures and to prevent seizure.Both will reduce metallic transfer and wear. Because of their great importance in tribology, there has been extensive RIC Reviews 160 study of compounds suitable for these purposes. Their detailed action will be considered more fully later (pp. 189,192). The most effective boundary lubri- cants are fatty acids, which react with metallic surfaces to form solid soaps having good lateral cohesion but relatively low shear strength. A 1 per cent solution of stearic acid [CH~(CHZ)~&OZH] in a mineral oil can be considered typical, but many other compounds are used. These all fail to lubricate satisfactorily at temperatures exceeding about 150-200 "C. The e.p. compounds are then used. These are mainly organochloride or organosulphur compounds which decompose on a metal at high temperatures to form metallic chlorides and sulphides.Suitable long-chain chlorinated compounds can be formed by direct chlorination of paraffinic oils. Some commercial fluids contain 30-40 per cent by weight of chlorine and are considerably diluted in a base mineral oil for use. The maximum temperature at which the chloride film formed is stable on steels is about 350 "C. To withstand higher local temperatures a sulphur compound is used. Sulphurized fatty acids have proved effective for a range of applications. It is possible to use elemental sulphur dissolved in oil, for example in a cutting operation where corrosion is not a major factor, but for most purposes the e.p. active element should be compounded in a molecule which is then stable until the high temperature is reached.Considerable care is necessary in the choice of additive, particularly when used to lubricate non- ferrous alloy materials, since the effectiveness depends essentially upon con- trolled corrosion. Typical additives are : C H CH3 CH3 I HC I I (c H 2)16 I I I /' (CH2)7 C I 'OH (c H 2)14 H-C-H CI OH 0' 0' C I 'OH Stearic acid Boundary additive C / \o (7H2)7 Cetyl chloride Sulphurized oleic acid Extreme-pressure additives Minor additives. One of the important features of an industrial lubricant is that it should be easy to apply. As we see in the next section, the ability to be transferred by pumping is an essential characteristic of greases used in centralized systems.Most oils can easily be pumped, but some may produce undesirable quantities of foam when vigorously agitated in contact with air. Anti-foaming additives are therefore sometimes necessary, particularly in fluids used in recirculating lubricant systems for metal cutting. They are also commonly incorporated in oils transmitting hydraulic power, where any foam may have very serious consequences. Silicone fluids are capable of collapsing the foam, probably by reducing surface tension, and are effective in concentra- tions as low as 5 p.p.m. If lubricants are to be used in very low temperature environments it may be necessary to lower the pour point. Paraffinic oils are usually dewaxed to give a pour point at 10 O F (-12 "C) because further refrigeration becomes very expensive.The naphthenic oils often have much lower pour points, down to Rowe 161 -40 O F (-40 "C). The paraffinic oils can be improved by addition of pour- point depressants which, like many other additives, are surface-active materials. Very small quantities, often less than 0.2 per cent, are needed. It is believed that the additive, which may be a polyalkyl methacrylate, alkylated naphtha- lene, or alkylated phenol, does not prevent the formation of wax crystallites but adsorbs onto their surfaces, preventing agglomeration and gel formation, and thus maintaining relatively low viscosity. It should not be thought that gel formation is always to be avoided in lubricants. It can be valuable in metalworking lubrication and is an essential characteristic of greases, which are considered in the next section.High molecular weight polymers may be incorporated as stringiness additives, for example to reduce splashing. Some lubricants contain additives which bear no relationship to the actual lubrication. It may, for example, be necessary to add compounds which have germicidal or bactericidal action, particularly for metal-cutting fluids, which are exposed and may come intocontact with the hands and clothes of operators. Some oils are even slightly perfumed so that they are more pleasant to handle. There are still some properties which are important in lubrication but for which no additives have yet been found. In straight mineral oils these include thermal stability, demulsibility and air bubble release.GREASES, SYNTHETIC FLUIDS AND SOLID LUBRICANTS I. GREASES Most greases in current industrial use are based on petroleum oils, which are thickened by addition of a gelling agent. They are used extensively in rolling bearings and for many industrial and domestic purposes where it is necessary to retain a small quantity of lubricant at the sliding zone for long periods. Before 1940 practically all greases were thickened with calcium or sodium soaps, and there is evidence from the Egyptian chariot mentioned earlier (p. 15 1) that lime-based grease was already in use by 1400 BC. The unprecedented demands of the last two or three decades have begun to change the pattern. There has been a remarkable increase throughout the USA in the use of lithium-based greases which, as a group, are capable of withstanding higher temperatures. A small but expanding utilization of inorganic thickeners has developed, primarily for the still higher temperatures encountered in aircraft and spacecraft.This restricted, highly lucrative market has also fostered the development of synthetic greases based for example on diester and silicone fluids. The synthetic fluids themselves have appeared in the lubrication field only since 1940. Before the war, the main emphasis in lubricant development was, as we saw in the preceding section, on increasing degrees of refinement of natural mineral oils and on improvement of specific friction- and wear- reducing additives.The demands of civil and military aircraft and space vehicles have led to development of new groups of lubricating fluids made available by the simultaneous rapid expansion of polymer chemistry. These new lubricants are mainly characterized by remarkable resistance to extremes of temperature. They tend to have low pour points, good thermal and oxida- RIC Reviews 162 tive stability and fairly high viscosity-indices. They are often also chemically inert and fire resistant. Many are resistant to nuclear radiation. For ordinary applications these advantages may not be important, and most of the synthetic lubricants are very expensive. Solid lubricants too have been extensively developed for exotic applications. Their range of use is much greater than that of the synthetic fluids.Appropriate solid lubricants are used at bright red heat, in equipment cooled with liquid nitrogen, in pressure vessels developing 1000 atmospheres, and in space vacua. Some are used in highly corrosive liquids, including concentrated HzS04 at 40 "C, and in pumps for high pressure water at over 250 "C. Graphite is still the most common solid lubricant, though the low friction of natural graphite has been known and valued for centuries. The development of synthetic electric-furnace graphites around 1900 and the subsequent dis- covery of a method of maintaining fine graphite in dispersion provided a new family of colloidal lubricants. These and many other forms of solid lubricant have also been extensively developed with the impetus of aviation and space programmes.Unlike the synthetic fluids, which have so far found an industrial market mainly in the restricted field where fire resistance is paramount, solid lubricants of the layer-lattice and polymeric types have considerable commer- cial application. Preparation of greases Processing. The plant for manufacturing greases usually occupies a four-storey building. Large storage tanks for oil and molten fat are located on the top floor. The process is started by releasing a weighed quantity of fat, or fatty acid, into an autoclave on the floor below. An aqueous alkaline solution or suspension and a proportion of oil are added and the mixture is then saponified for about half an hour under pressure. For the newer lithium-based soaps the temperature may be as high as 200 "C.The freshly-formed molten soap is then allowed to flow down to a mixing kettle on the first floor where it is mixed with the required proportion of oil from the storage tank, and is slowly cooled. Special additives can be incorporated at this stage, and any free acid or alkali is neutralized in accordance with the titrations made using test samples taken at intervals during cooling. Finally the grease is pumped through filters, while still at about 70 "C, directly into storage containers where it is left undisturbed for 24 hours before testing. Some greases receive special treatment. For example the lithium hydroxy- stearate greases are held at an equilibrium temperature and stirred vigorously for a long period to produce soap crystals which have grown to an optimum 1ength:diameter ratio.Gelling agents other than soaps may need to be dispersed in a colloid mill or homogenizer, and a dispersing agent may have to be added. Clays, for example, are dispersed with a volatile agent such as methanol. Some manufacturers of greases omit the saponification stage and merely dissolve a commercial soap in heated oil. Continuous grease production methods have been developed, but they have found little application. Quality control. Many empirical tests have been produced for greases, some of which have been standardized by the Institute of Petroleum and other Rowe 163 bodies. It is now recognized that many of the older tests are not reliable since they may measure properties which have little influence on the quality of the grease.Even the best test procedures, which are suitable for quality control, may give results which are irrelevant or even misleading when applied to service performance. As for oil technology, a better knowledge of the detailed chemical and physical action of lubricants in actual use would assist in formulation of more satisfactory tests. The problem is perhaps more difficult with greases than with oils, because the greases do not even exhibit Newtonian viscosity. The most important physical test for a grease is the consistency orpenetrution test. A cone of standard dimensions is allowed to descend under its own weight into the grease at a specified temperature.The depth reached after five seconds is measured on a dial depth-indicator reading in 0.1 mm intervals. Since the result depends upon the amount of shearing which has been sustained by the grease before the test, it is usual to pre-work the test specimen in the penetro- meter vessel by forcing a perforated plate through it 60 times. This also serves to mix heterogeneous samples. The National Lubricating Grease Institute of the USA has classified greases according to this test in a scale which is accepted throughout the world but is not always used in quoting consistency. A variation of this test is one which determines the mechanical stability of the grease by measuring the penetration after prolonged working, usually 1000 strokes of the perforated plate.Unworked grease penetration results are unreliable because some shearing is inevitable in filling the test vessel. Although the penetration test is widely used, it has been criticized. Mechani- cal stability can be determined under conditions more closely relating to service by the roll stabiEity test. In the standard form laid down by the Ameri- can Society for Testing Materials a 5 kg roller rotates inside a cylinder revolving at 160 rev/min. The consistency is measured in a penetrometer after two or four hours working at ambient temperature. A good grease should not have softened considerably. A test of this type can also be used at elevated temperatures, and under other conditions for research purposes.A further characteristic of greases which may be important in ball bearings is known as cleurubility. Various greases having approximately the same consistency, as measured by the penetration test, but different physical appearance can be prepared from the same soap and oil by altering the processing technique. Some are smooth and shiny while others are rough and look dry. Detailed study with polarized-light microscopy reveals that the latter contain tangled fibres which are partly crystalline, and are probably aggregates of smaller fibres. These clear from the bearing quite rapidly and the grease exudes past the seals, a phenomenon known as ‘winding out’. This depletion of the reservoir of lubricant leads to early wear and failure.Smooth greases are less aggregated and are so resistant to winding out that the packed grease remains under pressure in the bearing, which consequently tends to run hot. Inter- mediate textures are therefore chosen for most practical purposes. RIC Reviews For very soft greases the penetration value depends on both the yield stress and the apparent viscosity because the cone descends at an appreciable speed. It may be preferable to measure the cone resistance value (c.r.v.) by recording the equilibrium depth to which a standard cone will sink under a specified 164 load. This determines an approximate yield stress and is analogous to the hardness tests employed in engineering and metallurgy, especially to hot hardness tests in which creep is significant.The apparent viscosity can then be found independently at known rates of shear, usually in a pumped capillary viscometer. The c.r.v. is important in measuring slumpability, which is the ease with which grease will slump in a drum and feed out to a pump. At high tempera- tures the slumpability should not increase significantly, or the grease will flow out of the bearings. The drop point is also useful for maintaining uniform quality but is un- related to lubricant service. The drop point is recorded as the lowest tempera- ture at which a drop will fall from a specified orifice in a container. Oxidation stability is the most important chemical characteristic tested. Thin layers of grease are spread on trays and heated to about 100 "C in a sealed bomb filled with oxygen under a pressure of 100 lb in-2.The degree of reaction which occurs during a specified time is measured at intervals up to 100 hours by recording the drop in pressure. A test of this type can give infor- mation about the effectiveness of antioxidant additives, but the conditions obviously differ from those encountered in service and so should be interpreted with care. The acidity or alkalinity of a grease is normally monitored during produc- tion by standard titration, and appropriate additions are made in the mixing kettle. The final grease will also usually be tested, but complications can arise. For example, a soap of a weak aluminium base may react with potassium hydroxide used for titration and so register spurious acidity.Some specifica- tions call for determination of inorganic ash content, as for oils, but the result is mainly dependent on the quantity of gelling agent, usually Na or Ca soap, and has no other significance. Various other properties are measured. In some applications heat stability may be important. This is recorded as the amount of oil separating from a known quantity of grease at a specified high temperature (120 "C). For calcium-based greases the water content is critical, as we shall see in the next section. To extract the water, the grease is reflux distilled with an excess of a petroleum spirit whose boiling point is close to 100 "C. The spirit acts as a carrier but the condensate separates and the volume of water is easily mea- sured.Most other greases should not contain appreciable quantities of water. Corrosion may be important since the greases are often in contact with iron and copper alloys for long periods of time. Accelerated corrosion tests are often specified. Types of greases Common greases. We have already noted that the commonest greases are lime- based. The cheapest, known as Sett grease, is made from lime and rosin oil or heavy fuel oil. This has long been used for rough service conditions. The more modern calcium-based greases are usually made from a medium viscosity petroleum oil with suitable fats and hydrated lime. To stabilize this type of grease it is found to be necessary to include a small proportion of water.The action of this water is not fully understood but electron micrographs have Rowe 165 shown that the presence of a hydroxyl radical produces a pronounced regular twist in the soap fibres, which is thought to increase their mechanical stability; though there has been a contrary suggestion that the similar twist seen in lithium hydroxystearate fibres is attributable to the method of production for electron microscopic examination. It may be that the water assists crystalliza- tion which, as we saw above, affects texture and durability. If calcium greases are used at temperatures above about 60 "C, water will slowly be lost and the grease will become unstable. Excess water can be tolerated because the calcium soaps are insoluble and a water-in-grease emulsion may be formed.This will not greatly affect the consistency, so the grease may be considered water-resistant, in contrast to the greases containing water-soluble sodium or potassium soaps which form thin grease-in-water emulsions and are liable to leak away. Calcium greases are thus more suitable for example in rolling mills which are cooled by water sprays. Sodium soaps are to be preferred when only small quantities of water are present since the grease-in-water emulsion is less corrosive than water droplets which may be expressed from moist calcium greases. The sodium soaps are also stable to higher temperatures, even up to 100 "C. The upper limit is set by phase transformations to less ordered condi- tions which occur well before the grease finally becomes a soap solution.Several such transformations with appreciable latent heats can be detected in some greases by differential thermal analysis (often referred to as DTA). This alone casts doubt upon the validity of such crude measurements as the drop point. The sodium soaps are now tending to be superseded by lithium soaps, particularly in the USA. These were introduced primarily for aircraft uses because they combine a high upper service-temperature, up to 145 "C, with good low-temperature performance when blended with low pour-point synthetic oils such as the diesters discussed later (pp. 169-170). Lithium soaps can also be compounded with silicone fluids. In addition to good thermal stability the lithium-based petroleum-oil greases are mechanically stable, particularly if lithium hydroxystearate is used, and their water resistance is almost as good as that of the calcium greases.There are many industrial applications, particularly as multipurpose greases, and it seems likely that lithium greases will be used widely in future. It should nevertheless be recog- nized that the general classification covers a very wide range of products, as with calcium greases, and not every lithium grease necessarily has the desired properties and range of utility. There is a tendency to emphasize the solid phase in a grease, but the liquid phase can also have a strong influence on the performance. In the USA the NLGI (National Lubricating Grease Institute) survey reveals that the sales of lithium greases now equals those of calcium greases.Aluminium-, bismuth- and strontium-based greases have been rendered virtually obsolete; at best they found only a small market in the USA. There may continue to be some demand for lead-based greases for special applications because the lead soaps are surfactants conferring some anti-corrosion and e.p. properties. High-temperature greases. Though there is no doubt about the place of soaps as gelling agents, the drive towards high temperature uses has produced RIC Re views 166 interest in other possible materials. The main demand has come from aircraft production and space research but there are possible commercial applications, especially for clay-thickened greases.In all the new high-temperature greases the gelling agent is insoluble in the liquid at all temperatures. Most interest has centred on the silicone fluids as the liquid phase. Carbon black, or silica powder obtained in the micron size range from sodium silicate can be used in these fluids at temperatures up to 200 "C. The silica is, however, hygroscopic, so if other fluids are used it is necessary to add a surface-active hydrophobic compound such as butyl alcohol to 'waterproof' the dispersion and prevent it breaking down. Greases composed of silica in silicone oil are resistant to nuclear radiation as well as being temperature stable, and are good electrical insulators. They also have low vapour pressure, but are very expensive. Silicone greases containing boron nitride, a lamellar compound somewhat resembling graphite except in colour, have been found effective for high temperature applications.Gelling agents of greater commercial interest are derived from mont- morillonite clays such as bentonite. These are hydrated magnesium aluminium silicates, which are easily obtainable and inexpensive but are stable over a wide temperature range. The natural clays are hydrophilic, like silica, but they can be reacted with a surface-active ammonium salt or an amide to produce an affinity for oils. It seems likely that clay-based petroleum greases will compete with the lithium-based greases for the multipurpose commercial market. Clays can also be dispersed in silicone fluids for high-temperature applications.They can, however, be rendered unstable by strong surface-active agents, such as the conventional e.p. additives. There will undoubtedly be considerable activity in the field of high tempera- ture greases for some time to come. There may also be scope for improving specific properties. Thus, for example, it is claimed that sodium octadecyl terephthalamate, formed by reacting NaOH with the methyl ester of n-octa- decyl terephthalamic acid, makes greases which are much more water resistant than those using sodium stearate; while at the same time the sensitivity of clay-based greases to additives is avoided. Grease additives. The properties of conventional greases can often be improved by additives, which fulfil the same functions as the additives in oils.Oxidation is the main limitation in most applications, and is more serious in greases than in oils because the soaps used as gelling agents tend to promote oxidation. Solvent-refined oils are more resistant to oxidation than the distillates and are thus preferred, since the life of the grease is more likely to be determined by the oxidation of the liquid phase than of the gelling agent. This is an important consideration to be borne in mind when selecting a grease, particularly because of the common practice, also adopted in the preceding section, of discussing greases according to the class of gelling agent. No commercial grease will have better oxidation resistance than that of the petroleum oil from which it was manufactured.The antioxidants used for oils can also be used for greases, provided that they do not interfere with the dispersion, as they may do with clays. Since the temperatures at which greases are used will be lower than those encountered with engine oils, the nitrogen-containing additives are suitable. Rowe 167 Phenylnaphthylamine is often effective, providing protection against oxidation during long periods of storage. More active agents such as phenothiazine may be necessary for protection under more severe conditions of agitation and higher temperatures. In the presence of copper alloys a further surface-active compound, often disalicylidene ethylene diamine, is used to form an adsorbed layer on the metal and so prevent catalysis. Corrosion is less likely to be troublesome with greases than with oils.Most greases are either naturally hydrophobic or can readily be converted, and they usually form a relatively impermeable coating on the bearing. To allow for condensation or the presence of moisture, corrosion inhibitors are however sometimes required. Oil-soluble lead soaps can be incorporated. To protect steel surfaces, micron-sized sodium nitrite crystals may be dispersed in the grease. These dissolve in water to form a passivating solution. Some greases used under severe conditions require fortifying by the addition of e.p. com- pounds. Mild e.p. action can be provided by lead soaps but chlorinated or sulphurized organic compounds may be used to provide greater protection from frictional damage to the surfaces.As with oils, sulphurized fatty acids are among the most effective e.p. additives. 11. SYNTHETIC FLUIDS The first synthetic lubricants were produced to supplement or replace mineral oils in countries where the natural products were not available. For economic or military reasons, this is still sometimes desirable. The usual production method is polymerization of alkenes with or without the use of catalysts such as AlC13 or BF3. Another technique is condensation of alkenes or chlorinated hydrocarbons with aromatic hydrocarbons. This has the commercial advantage of imparting a green fluorescent bloom that has no technological significance but is often associated with good lubricating oils.Under some circumstances, depending upon supply conditions, it may be worth while to dechlorinate the chlorinated hydrocarbons. Since all these fluids are intended as substitutes and have no specific advantages we shall not consider them further. In discussing modern developments in grease formulation we have men- tioned the impetus received from aircraft and space vehicle demands for developing lubricants to operate at higher temperatures. This has been equally important for new fluid lubricants, some of which have been referred to as the liquid phases in greases. The major requirements for these new liquids are good high-temperature stability in oxidizing atmospheres, small viscosity change with high temperature, low pour point for starting conditions, and general chemical inertness to metallic, elastomeric and polymeric construct- ional materials. In addition, many of these fluids are naturally resistant to fire and to nuclear radiation.For ordinary purposes they are usually much more expensive than conventional lubricants. Halogenated hydrocarbons. The initial choice of a synthetic lubricant is usually conditioned by the need for resistance to oxidation at high temperatures. It is clear that the fluid must then also have a high boiling point. In the conven- RIG' Reviews 168 tional oils the upper temperature limit is set by the stability of the carbon- carbon bond in the hydrocarbon chain. The oxidation behaviour is however determined by the carbon-hydrogen bond stability.Fluorination greatly increases the inertness. The fully fluorinated ethylene polymer, polytetra- fluoroethylene, is an outstanding example of an inert, stable compound. It also has most remarkable frictional properties to which we refer later (p. 18 1). It is however a solid and so does not further concern us here. Low molecular- weight polymers which are liquid at ordinary temperatures can be obtained with mixed chlorine and fluorine substitution. For example, the unsaturated monomer of chlorotrifluoroethylene can be solution-polymerized using free radicals and halogenated hydrocarbon-chain transfer-agents : Such fluids are several times as expensive as petroleum oils, but are used in atomic energy and spacecraft work. They have found some uses in chemical engineering because of their general inertness.Esters. The ester linkage C-0-R is much more stable to heat than the C-C bond, and forms the basis of the most important group of synthetic lubricants at the present time. Castor oil, the original aircraft engine lubricant, is a natural ester but it has poor thermal stability. It consists mainly of the triglyceride of ricinoleic acid and can be represented by: CH2.O.CO-R R=CH3*(CH2)5*CHOH*CH2. CH CH * 0. CO * R II CH.(CH2)7* CO2H CH2.O.CO.R I I engines, are synthetic diesters of the general type R’02C - R COzR’, obtained The modern lubricants, also used in aircraft particularly for gas-turbine by reacting an acid R-(COzH)2 with an alcohol R’OH. It is common to use a straight-chain dicarboxylic aliphatic acid such as sebacic acid (H02C * (CH2)g - COzH) with a branched-chain primary alcohol such as 5-methyl-5-ethyl-heptan- 1-01 : CH3 CH2 4 4 - 0 H I C H 3-L-(CH CH2 I CH3 Monobasic acids are also reacted with dihydric alcohols, and more complex esters formed from polyhydric alcohols - including neopentyl polyol esters Rowe 169 12 and polypentaerithyrtol esters -have found a use in more advanced gas turbines.Neopentyl glycol is an example, which remains stable up to 260 "C. It is commercially available because of its use in paints. Linear polyalkaline glycols were originally used as substitutes for castor oil but also have intrinsic value as fire-resistant fluids. Most of the synthetic diesters have good thermal stability, but some are little better than hydrocarbons when exposed to oxi- dizing conditions.They are however good solvents and can be considerably improved by antioxidant additives. Boundary and e.p. compounds can also be added without difficulty ; v.i. improvers are not usually needed. Similar stable ester-linkages can be used in phosphate, borate and silicate esters and in fluoroesters. The latter are used in gas turbines but have little general application because of their relatively high freezing point and low boiling point. The silicate esters may however be stable at temperatures exceeding 300 "C. The Si-0 bond energy is 25 kJ mol-1 compared with 20 kJ mol-1 for C-C. The general formula of silicate esters is 0 R' I I 0 I R " The substituent groups can be alkyl or aryl, derived from alcohols reacted with silicic acid.These compounds are relatively easily hydrolysed to silica. Tetraethyl orthosilicate is actually used to preserve masonry. The long-chain alkyl derivatives are more resistant to hydrolysis and are consequently preferred as lubricants. These also show the highest viscosity index, but their main application is in hydraulic transmission rather than in lubrication. They are non-corrosive, but do not exclude water, so rust-preventatives are usually added. Organosilicon fluoroesters, stable in air at 200 "C and in the absence of oxygen at 350" C, can be used as lubricants, sometimes with alkali-metal salts as antioxidants. OH I R-0-Si-0-R" Phosphate esters of the type: 0 .R R.O-P=O with, eg.R = I 0 . R have been used for many years as antiwear additives, the best known being tricresyl phosphate shown above. Some, such as alkyl polyethyleneoxyphos- phate esters have been reported to show good load-carrying capacity in a four- ball testing machine. They are now available in a wide range of viscosity and other properties but the main interest is in their good fire-resistant properties. RIC Reviews 170 When used as lubricants, they dissolve many seal and gasket materials and fairly easily hydrolyse to corrosive acids. At temperatures below about 100 "C they have good performance in both hydrodynamic and boundary conditions. Ethers. Various ethers have been suggested for aerospace uses.Polyphenyl ethers may be stable oxidatively to 260 "C and thermally to 400 "C. Metal chelates and other organometallic compounds can inhibit oxidation in these ethers though conventional antioxidants such as amines and phenols are ineffective. Some alkylated aromatic ethers are thermally stable to 450 "C and can withstand high radiation levels. They are used in atomic reactors and in gas turbines. Some inhibited aromatic ethers are effective as lubricants from -35 "C to 250 "C but are less resistant to radiation. bis-(m-phenoxybenzene) + 30 per cent bis(m,p-diphenoxybenzene) + 5 per Elaborate formulations have been proposed, for example 65 per cent cent bis(p-phenoxybenzene). This is recommended to be used with 1 per cent by weight of CC13C02H or with Zn, Bi or other salts of trichloropropionic acid, presumably to confer boundary lubricating properties since these are usually poorer in the absence of carboxylic compounds.Silicones. Silicone fluids are used for many applications. They are very inert and have very poor boundary lubricating properties unless halogenated. All the common additives reduce the stability of the silicones, but these fluids show very little change in viscosity with temperature and are thus good lubricants under hydrodynamic conditions. Most of the silicone lubricating fluids are semi-organic polymers with repeating Si-0 bonds along the chain : R The side groups are usually methyl or mixed methyl and phenyl. Chlorine may be substituted into the phenyl groups, producing greatly improved results in wear tests.The long molecules are flexible, without branching or cross-linking which would increase the dependence of viscosity on both temperature and pressure, and the viscosity varies less than that of any other synthetic lubricant. They remain liquid even with high molecular weights. For hydraulic systems this is advantageous, but a viscosity increase, and even solidification under pressure, can often be valuable in elastohydrodynamics and lubrication in metalforming. At high temperatures, 360 "C in the absence of air, thermal degradation reduces the molecular weight and the viscosity. In air the silicones can be used at 200 "C, where a slow oxidation increases the viscosity.Organo- selenium compounds, for example 5-methyl-benzo[2.1.3]selenadiazole, have been used as antioxidants conferring fair stability at temperatures up to 300 "C. At 150 "C they are stable indefinitely. They have a very low vapour pressure and are widely used as vacuum diffusion pump fluids. The flash points are generally high (280 "C), but the spontaneous ignition temperature may be below that of the phosphate esters. For low temperature use their exception- ally low pour point (-70 "C) may be significant. It has also been suggested Rowe 171 that addition of about 1 per cent polyethylsiloxane to dixylyl methane will improve the wear preventive properties, equalling those of a good mineral oil, but with greater oxidation resistance. A minor use of the silicones in lubrication has already been mentioned.A few parts per million, when finely dispersed, will effectively collapse foam. iMiscellaneous synthetic lubricants. Some fire-resistant hydraulic fluids have already been mentioned as lubricants. Others which are reputed to be suitable are trimeric and tetrameric phenyl polyfluoroalkyl phosphonitrilates. These have low pour points, about -50 "C, together with high compression ignition ratios and high spontaneous ignition temperatures. They impart good wear resistance and are compatible with rubbers, but they may hydrolyse unless carefully purified. High viscosity is sometimes necessary. Butadiene dioxide copolymerized with ethylene or propylene oxide using NaO catalyst has been proposed.Isopropyl 1 ,g-diphenylnonane has been reported to have excellent radiation resistance and to be suitable for use in ball bearings operating with liquid oxygen and hydrogen rocket propellants. Liquid metals can also be used as high-temperature non-organic lubricants. Eutectics with melting points as low as 150 "C are available, but usually this class of lubricant is considered only for much higher temperatures, 600 "C upwards, where the choice of other materials is restricted. Molten inorganic salts have potential uses. Some are already used in metalworking, for example alkali phosphates at about 400 "C, well above the useful temperature range of organic lubricants but below that at which conventional glasses soften. Aqueous lubricants.Numerous other possibilities of developing synthetic lubricants exist. Polyethylene oxide and water-soluble polyglycols have been proposed as constituents of metalworking lubricants. Insoluble polyglycols are likely to have uses in the future because they decompose without forming deposits on metal parts. They have been used successfully at very low tempera- tures, and also have high v.i. and fair thermal stability. Aqueous sodium nitrite solution provides a satisfactory non-corrosive coolant for grinding operations, and oil-in-water emulsions are widely used as metal-cutting fluids. One patented grinding fluid contains up to 9.5 per cent by weight of ethylene oxide, together with sodium oleate for boundary lubrication, sodium nitrite for corrosion inhibition, a germicide and an antifoam additive.Such fluids may have applications in other lubrication fields, particularly in metalworking, where removal of the lubricant without residue is an important factor. Lubrication in space environment. It is perhaps appropriate here to consider briefly the environment of space vehicles, since the extreme conditions encountered and the immense costs of space exploration have given great impetus to development of new lubricants. The gas pressure is of course very low, about 10-12 torr at 400 miles from the earth's surface. The kinetic temperature may be about 95 "C in the stratosphere and ionosphere but as high as 2000 "C in the exosphere, though the latter figure is not very significant since relatively few molecules are present ; the temperature of a body depends RIC Reviews 172 almost entirely on direct radiation. There is strong ultraviolet radiation once the vehicle has passed outside the filtering influence of the atmosphere, and high electron and proton intensity in the van Allen band.Solar flares increase all the radiations and also produce heavier ions. All these sources provide energy for chemical reactions that can degrade or decompose lubricants. For external lubrication, the vapour pressure requirement may be dominant. Vapour-diffusion pump oils have been used; for example, phenyl methyl silicone fluids and sebacates such as di-2-ethylhexyl sebacate. Diesters and polyphenyl esters can also be used. Somewhat surprisingly it has been found possible to use certain mineral oils of low volatility, though these may have vapour pressures of about 10-8 torr.The loss of fluid by evaporation can be substantially reduced by using labyrinth seals, or the fluid can be replaced by wicks or other devices. In this context it should be remembered that bearings usually become warm. At least one satellite camera failure has been attributed to condensation of volatilized bearing oil onto the lens. Eventually this re- evaporated and the lens became clear again. The advantage of the mineral oils is their better boundary lubricating ability, though this may be impaired by lack of oxygen. Boundary lubricating proper- ties can be imparted to the inert fluids by halogenation; thus trifluoropropyl methyl polysiloxane appears to be useful.A chlorinated methyl phenyl silicone is reported to give good rolling contact fatigue life at 180 "C. FRICTION OF DRY SOLIDS AND OF LUBRICATING LAYERS Introduction to rolling and sliding friction While there is little doubt that the conscious practice of lubrication can be traced back to at least 1500 BC, it seems likely that the first specifically tribo- logical invention was the wheel, in use in Sumeria before 3000 BC. It was indeed a most ingenious idea to convert the sliding resistance into a torque with a small lever arm and thus to reduce the amount of work expended in overcoming the friction, without any change other than the basic design. There is however a second feature which makes the wheel so useful.The resistance to rolling at the wheel periphery is negligibly small. This property has of course been utilized for a very long time, probably long before the wheel was invented, by placing rollers under heavy loads. Curiously, the civilizations of central America, though highly developed, never invented a wheel, but they appear to have been familiar with rollers. The principles of rolling friction have been clearly appreciated since the 15th century and the first improvement on the wheel was made in the 18th century by providing rolling-element bearings on the axle. It is however only recently that a satisfactory explanation of the resistance to motion in pure rolling has been given. The energy loss can be attributed to hysteresis in the solid bodies, rather than to any specifically interfacial characteristics.This explains the otherwise surprising fact that pure rolling friction is practically uninfluenced by lubri- cants. When tractive force is transmitted across a rolling interface, as is required with driving wheels, reliance is placed upon sliding friction stresses. For this reason, although rolling friction is of great practical importance it is not associated with chemistry and we shall not discuss it further, except to 173 Rowe mention the influence of chemical composition and structure on the mechani- cal hysteresis of rubbers. Rubbers with high hysteresis can be used on the tread of pneumatic tyres to provide increased resistance due to rolling when the sliding component is diminished by a lubricant, thus affording protection from skidding on wet roads.The sliding friction between dry surfaces is obviously useful in transmitting tractive forces, but most of the thought associated with dry sliding has been directed towards reducing the friction force, and especially the wear which usually accompanies high friction. This problem too was recognized by the technologists of ancient civilizations, who realized that some combinations of materials had desirable tribological properties. For example, a mural painting from Egypt clearly shows a stone bearing-cup being used with a bow drill. This principle is still very important and there is currently considerable interest in jewel and ceramic bearings for mechanisms operating in space environment.We have already seen how some of the widely-used modern lubricants and additives, such as the fatty oils, were also anticipated some thousands of years ago. Nevertheless, the processes of friction and the detailed actions of lubricants have been studied systematically for only a few decades. The majority of the research in this field has been related to metals and metallic alloys. These are, of course, the most widely used materials in machinery of all types. Friction of a van der Waals solid Experimental measurements have been made of the coefficient of sliding friction between two specimens of solidified krypton. The crystallographic bonding in this material is almost entirely attributable to van der Waals forces, with a low bond energy of about 4.8 kJ mol-1.As explained earlier (p. 140), the surface energy will be even less and it might be expected that the friction would also be low. On the contrary, the coefficient of friction at low temperatures (between 24 K and 70 K) is found to be about p = 0.7, very similar to that of rock salt which has been thoroughly cleaned in a vacuum, and somewhat greater than that of similarly denuded diamond. The explanation for this can be given in terms of a very simple experiment with a child’s glass marble pressed lightly into butter. In fact the mechanical properties of butter on a cold day are very similar to those of solid krypton at liquid nitrogen temperatures. When the marble is removed it will be seen that a permanent depression remains in the butter, which has thus been plastically deformed.In addition, though this is not observable by eye, the molecules in the vicinity of the indentation will have been forced more closely together while the load was maintained on the marble, increasing the force of repulsion between them, but as soon as it was removed they sprang back. The molecules near the centre where the applied stress was greatest would spring back more than those near the periphery. This elastic recovery consequently reduced the depth of the final impression, and also slightly modified its shape, leaving a somewhat greater radius of curvature than that of the marble. With rubber instead of butter the recovery would be complete.These two components of elastic and plastic deformation can be visualized RIC Reviews 174 on an atomic scale with reference to the atomic potential already discussed (p. 142). Any displacement of one atom relative to another by less than the effective radius of the potential well (ignoring small probabilities of escape) will raise the energy of the appropriate atoms, but the position is unstable and on release of the displacing force the atoms will revert to their equilibrium spacing. It is in fact possible to calculate the elastic modulus (stress/strain) on this basis. If however a tensile displacement is given which separates one atom from another beyond the position of the subsidiary energy maximum or hump, the atom cannot return.In principle it would be expected that this position would be unstable with respect to further separation and consequently fracture would occur. Under special circumstances, such as may occur in metallic and non-metallic whiskers, this can be observed. These whiskers have been of great interest in the past decade because of their very high strength. Usually, in bulk materials, the atoms will separate into new equilibrium positions relative to different atoms, at much lower stress levels. This is known to be essentially a shear process, analogous to moving one row of close-packed spheres over another. Each sphere in the row moves from one low-energy position to the adjacent one, without losing contact. Such a process requires little energy if for example one atom in a row is missing, and the atoms move along one at a time to fill in the holes.Obviously this process can be regarded as the vacancy moving one space at a time in the opposite direction. As shown in modern textbooks of metallurgy and solid-state physics, all common materials contain a high density of vacancies and dislocations in the crystallographic structure. Dislocations can be considered as extended vacancies, in which for example a whole row or plane is missing. They move with relatively low activation energy and cause plastic deformation to occur at a stress (energy) level much lower than that required in theory for elastic fracture. Because plastic flow is essentially a shear process it can be initiated whenever the appropriate shear conditions are established, whether the applied forces are tensile or com- pressive.The final size of the indentation made in our experiment with butter is thus a measure of the amount of plastic flow produced, and the extent of the recovery from the deformed state is a measure of its elastic properties. This type of test is very valuable, and is widely used in metallurgy to determine hardness. The hardness number, which is really a yield pressure expressed in kgmm-2, is calculated by dividing the applied load W by the projected area A of the indentation or, in a more common form of the test, by the curved surface area. Elastic recovery is usually ignored for metals. It will be clear that soft materials will deform more under a specified load and the area of contact between the sphere and the test block will be greater.We can now return to the friction experiment with solid krypton. Because the interatomic forces are relatively weak, and the resistance to plastic deformation is much lower still as a result of the crystallographic imperfec- tions, the hardness of krypton is low. Direct measurement has shown it to be of the order of 0.25 kg mm-2 at the temperature of liquid hydrogen, compared with crystalline NaCl, for example, which has a hardness of about 20 kg mm-2 at room temperature. Two krypton specimens placed in contact under load will deform plastically until the area of contact between them is large enough to Rowe 175 support the load Wat their mutual yield pressurep, which depends to some extent on the geometry but is roughly equal to the hardness.If for example the load is 25 g as in a typical friction experiment, the contact area will be 25 x 10-3/0.25, about 10-1mm-2, compared with about 10-3mm-2 in a comparable experiment with rock salt. As we saw earlier (p. 140), the surface free energy of a solid is considerable, and so when two surfaces are brought into intimate contact by the processes of plastic flow it is energetically ravour- able for the solids to coalesce and so to eliminate the interface. The force required to deform such a region subsequently in shear will be comparable with that required to shear a similar area of polycrystalline material. The shear stress, found by dividing the force F by the area of contact A , is usually denoted by s.Thus we may suppose that if the resistance to sliding arises from the necessity to shear the junction between the two surfaces, the frictional force F will be given by F = A.s = (W/p)S Because the shear strength and yield pressure are inherent properties and their ratio remains nearly constant, the frictional force is proportional to the normal load for many materials. This is one of the two classical laws of friction. The equation also shows that F is directly proportional to the area of intimate contact. Because all practical surfaces are rough (p. 139), this is usually very much less than the apparent geometrical area so the friction force is found to be independent of the latter.This is the other important law of friction. The coefficient of friction p is found by dividing the friction force by the load : It is known from the general theory of plasticity that the stress required to cause yielding in pure shear is approximately 0.6 times the stress to cause yielding by simple compression. For reasonably flat sliders the coefficient of friction would thus be expected to be about 0.6. The coefficient of friction for solid krypton approximates to this value, but as we shall see later this agree- ment is fortuitous, resulting from the interaction of two opposing effects. Nevertheless, the simple theory is sufficient to show why the coefficient of friction is practically unaffected by the change in plastic yield properties, which differ by a factor of 100 between a van der Waals solid, krypton, and an ionic solid, NaCl.The coefficient of friction depends upon the ratio of two mechanical properties but both of these depend upon the same basic atomic displacement process of shear within the material, so the ratio remains constant. RIC Reviews In general, perfect cohesion between two solids in contact will not be experienced. Two important factors are the topographical mismatching and chemical contamination. A perfect match can be obtained only in special circumstances. If a mica sheet is partially cleaved and the two leaves are released they may fit together perfectly. A strength equal to that of the original sample can then be measured if the experiment is performed rapidly.It has already been shown that clean surfaces in air become contaminated spon- taneously and at a high rate. Therefore, if the mica surfaces are left exposed to air for a short time the surface energy will be greatly reduced by adsorption 176 and the adsorbate will not be eliminated when the leaves come into contact again. The measured junction strength is then greatly reduced even for perfectly matching surfaces. In the krypton experiment the contact between the sliders is intimate be- cause the plastic deformation allows the surface topography to adjust to conformity, at least partially. At the low temperature of liquid hydrogen the rate of contamination will be low. This system is consequently suitable for analysis, though it is experimentally more difficult than common frictional studies.The frictional behaviour of krypton is also interesting at higher tempera- tures. As the temperature is raised the hardness diminishes but since the shear strength also diminishes, the coefficient of friction remains unchanged between 20 K and 70 K. If however the experiment is interrupted and the specimens are allowed to stand under load at an elevated temperature, the area of contact will slowly increase. This is the same phenomenon that is observed in creep of metals. If a weight is hung on a metal wire at a high temperature (above about half the melting point 8, measured in degrees absolute) the wire will slowly extend plastically. The process is attributable to self-diffusion and an activation energy can be calculated from the results.For solid krypton this amounts to about 0.25-0.5 kJ mol-1, as determined from the creep in hardness tests. Incidentally, a tensile creep-test on a metallic wire can be used to determine the surface tension of a single crystal. The creep rate is measured as a function of load at a given temperature near the melting point. A linear relationship is obtained, from a balance between the tendency of the wire to contract under surface tension and to extend under load. The load correspond- ing to zero creep-rate exactly balances the surface tension, and the results are in fair accordance with the theoretical predictions discussed earlier. In the interrupted type of friction experiment the area of contact becomes larger, but when sliding is resumed the shear yield stress remains the same as before the pause, so the frictional force is initially greater than before.After a short distance of sliding the conditions revert to normal. This behaviour can also be observed in metals at high temperatures and in some solid polymers at room temperature. As the temperature is still further raised, beyond about 8 = 0.9 Om, a large drop in friction is observed, attributable to surface melting. If a liquid film forms between the surfaces the interfacial shear strength is greatly reduced, though the area of contact is still determined by the substrate strength and remains unaffected by the melting, so the coefficient of friction falls.Values well below p = 0.1 have been recorded with krypton. Similar low coefficients of friction have been measured with ice near 0 "C. Surface melting has been shown to account for the low friction of skates and skis on ice and snow at moderately low temperatures. When the temperature is well below zero as in the polar regions of the world the frictional resistance of sledges is known to be greatly increased. It is possible to observe the same effect of surface melting on metals when the speed is high enough. Bowden has suggested that if speeds of 2000 mile h-1 could be maintained it would be possible to ski on a copper mountain. Practical use of this phenomenon is made in the copper driving bands of artillery projectives.Ro we 177 Friction of covalent solids The characteristics of these solids may be described by considering diamond, which has been extensively studied. The well-known crystal lattice of diamond is composed of carbon atoms, each of which is strongly bound by covalent bonds to four neighbours at the corners of a regular tetrahedron. The elastic modulus is very high, up to 108 N m-2, depending on direction. It is also difficult for dislocations to move even on the most favourably orientated planes of atoms, so the crystal tends to be brittle and shows practically no ductility, though there is evidence for very slight plastic deformation being possible even in diamond. It is the hardest natural substance known and may well be harder than its nearest rival, the artificial cubic modification of boron nitride known as borazon.It is much harder than tungsten carbide. The normal deformation of diamond and of other covalent solids is entirely elastic, up to the stress at which cracks occur. Consequently the frictional behaviour differs from that of a predominantly plastic solid such as krypton. If a friction experiment is performed with a rounded diamond slider on a fairly flat diamond face in the laboratory atmosphere, it is found that at 25 g load the coefficient of friction is very low, about p = 0.04. As the load is increased no significant change in p is observed, but at loads below about 10 g friction increases fairly rapidly. The explanation of this is that in elastic deformation the area of contact between two spheres is not directly propor- tional to the applied load Was in plastic deformation, but is proportional to the two-thirds power of the load, according to the classical Hertz theory. Thus, since the frictional force F is still assumed to be the product of the area of contact A and the mean shear stress s, the coefficient of friction will be p = - = - = - As A Cc w-113 w w w It is found that the friction is little influenced by lubricants and there is a widespread but fallacious belief that the friction of diamond is inherently low.This is commonly associated in a general way with the extreme hardness of diamond, about 2000 kg mm-2, but as we have seen it is the ratio of shear strength to hardness which determines p, not the absolute value of hardness.We shall find that there are some highly anisotropic solids, such as molyb- denum disulphide, which exhibit a low ratio of s top, but for most materials s and p are closely related, and diamond shows high values for both. Actually diamond is slightly anisotropic and directional dependence of p can be detected. The low coefficient of friction usually measured results from the reduction of surface fields by adsorption from the atmosphere. Then, of course, if the interfacial shear strength is no longer a property of the slider material but has a much lower value associated with the adsorbate, the friction will be low and a high hardness value will reduce p still further. It is possible to demonstrate this experimentally by outgassing the diamond surfaces, that is, heating the diamonds to a high temperature in a good vacuum to remove as much as possible of the chemisorbed material, and subsequently cooling to room temperature while maintaining the vacuum.The coefficient of friction between two such diamonds will be increased about 10-fold. The shear RIC Reviews 178 strength deduced from the frictional force and the measured contact under these conditions is found to be about 200 kg mm-2, compared with about 0.05 kg mm-2 for krypton. For most materials the outgassing treatment is conducted at as high a temperature as possible, preferably until there is appreciable loss by vaporization, but in the particular instance of diamond care has to be exercised because the structure is not a stable one.Eventually the surface and even the whole diamond will graphitize. About 700 "C is a suitable degassing temperature for diamond. If oxygen, water vapour or air is admitted to the denuded diamonds the friction is rapidly reduced to the usual low value. Most of the reduction is attributable to chemisorption but the friction rises slightly when the superincumbent gas is pumped away. Friction of ionic solids Sodium chloride may be chosen as representative of ionic solids. It combines elastic and plastic deformation properties, but is predominantly elastic under ordinary conditions. In a friction test, as we have seen, the local pressure may be very high. This suppresses the tendency to fracture when a tangential force is applied.Thus at the actual contact between sliders, NaCl behaves as a plastic solid but in the peripheral region where the hydrostatic pressure is low it deforms elastically. During sliding, cracks can be observed to form in the surface immediately behind the slider in the region where the pressure drops. If the shear strength is calculated from the force and contact area in a friction test it is found to be considerably greater than that of the bulk solid, but the discrepancy is small if the bulk strength is also measured under high pressure. Because of the plastic flow, the surfaces of two rock salt crystals can come into close conformity, and a significant adhesion can be detected by a pull normal to the interface, with clean specimens in vacuum.If a single crystal is cleaved and the two portions are placed in contact immediately, under a heavy load, adhesive strengths up to 0.4 kg mm-2 can be recorded. The reasons that such adhesion cannot be found with diamond are that the large plastic deformation required to ensure intimate contact is impossible, and the elastic recovery of diamond to its original shape tends to fracture in tension any interfacial bonds formed at the local contacts, especially near the periphery of the contact region. Most other ionic solids show frictional behaviour intermediate between the fully elastic type of diamond where the frictional force is proportional to W2/3, for a single spherical contact, and the elastic-plastic type of NaCl for which F x W.Even with the elastic solids it has been shown that if the surface geometry is more complex the frictional force becomes more nearly propor- tional to the load. Friction of organic polymers In recent years there has been a great expansion in the use of polymeric materials, in lubrication applications as in many other fields. The frictional properties of solid polymers are generally similar to those of the other materials we have discussed, but important differences arise from their specific properties, especially their viscoelastic deformation characteristics. The coefficient of friction of most polymers when sliding under dry Rowe 179 conditions lies between p = 0.4 and ,u = 0.6, whether sliding in homogeneous pairs or in contact with a metal. This is somewhat lower than the friction of rock salt, for example, but higher than that of diamond which has not been specially cleaned in vacuum.A notable exception is polytetrafluoroethylene, to which we shall refer in detail later because it possesses unique low-friction properties. Thermoplastic poyrners. Nylon is one of the commonly used thermoplastic material groups, Nylon 66 is formed by reaction between hexamethylenedia- mine and adipic acid with elimination of water, and is usually polymerized to a molecular weight exceeding 12 000. It has a good strength and can be used in bearings from which water is excluded. This polymer has the disadvantage that it fairly readily adsorbs water, which causes dimensional change, so in a high- tolerance bearing it may be preferable to use Nylon 11 which does not have this disadvantage, though it is weaker and has less abrasion resistance.Nylon 11 is formed by a similar reaction, but with only one type of molecule- 1 1 -aminoundecanoic acid : HzN * (CH2)lo. COzH + HzN * (CH2)io. COzH + 2nH20 force can be represented by: J. [-NH(CHz)lo* CO. NH(CH2)lo. CO-1, Many other polymers have similar frictional properties. Experiments have been carried out over a very wide range of loads with nylons and other polymers. It is found that the influence of load on frictional F = aW + bWn or more simply by F = kWn For a simple spherical slider with normal load the area of contact varies with the $ power of the load, as for elastic sliders such as diamond. As the com- plexity of the surface increases, the index n tends towards unity, also as men- tioned before.The friction force does not bear exactly the same relationship to load because the shear strength increases slightly with pressure. The results are also speed dependent because the viscoelastic deformation resembles creep, in that the area of contact under load increases with time of loading and con- versely decreases with increasing speed. Heating of the surface at higher speeds tends to soften the load-bearing region and so to produce larger areas of contact. It may also help to desorb surface contaminants and thus promote welding. Both these effects will increase the frictional resistance, but if the speed is very high it may be possible to produce surface melting, which, as with krypton, ice and other materials, gives very low coefficients of friction.At moderate speeds it is believed that the flexibility of the long molecules in the polymer chains is the dominant feature. Rigid cross-linked polymers show relatively small dependence of friction upon speed or temperature. Thermosetting polymers. The thermosetting polymers used in tribology are mainly phenolic resins bonded with woven cotton or other fibres. A major RIC Reviews 180 application for reinforced phenolic polymer bearings occurs in hot-rolling mills, where it is often essential to use bearings which will perform well in the presence of water. Such bearings give very low friction, with coefficients of about p = 0.006.They can also be used to replace lignum vitae in the stern- tube bearings of ships. Formerly this dense wax-filled natural wood was widely used for this and similar purposes, but more recently it has become common to use oil-lubricated white-metal bearings with a seal to prevent ingress of sea-water and thus to reduce corrosion in the bearing. An attractive feature of the reinforced resin bearings is that they are much more resistant to nuclear radiation than the thermoplastic materials and so can be used in nuclear power plant, needing no attention. Reinforced resins are also quite widely used in gears and numerous small sliding components. A second group of thermosetting resins used in bearings is based on epoxides.These are in general more expensive, but give good performance especially when filled with graphite or molybdenum disulphide. They may conveniently be sprayed onto metal backing strips. Polytetrafluoroethylene. The polymer PTFE has unique frictional properties. The coefficient of friction is found to be much lower than that of other polymers, characteristically p = 0.05-0.1, compared with about p = 0.4 for polyvinyl chloride and p N 0.7 for polyethylene. Moreover the friction is independent of temperature from the liquid hydrogen range up to almost 300 "C. It is also unaffected by the presence of contaminants and operates as well in high vacuum as in air. There has been much discussion of the basic explanation of this behaviour, but it is still not really clear.In general terms, the large negative F atoms will screen the positively-charged C atom at the core of the polymer and consequently the external field will be expected to be less than that of polyethylene, which is otherwise structurally similar. The surface forces will thus be low, but the ratio of surface to bulk strength would still be expected to be of the same order as for the other polymers, so the low co- efficient of friction remains unexplained. It has however recently been suggested that the bulk strength is significantly higher than this would imply, because PTFE normally exhibits a high degree of crystallinity and the well-orientated molecules provide mutual support and increase the resistance to deformation.It is possible that electrostatic effects may influence the friction. There is undoubtedly a considerable generation of charge when polymers are rubbed together, but it is usually considered that the energy involved is much smaller than that required for mechanical deformation. Here again there is considerable scope for basic investigations. It is remarkable that the substitution of chlorine for even one of the four fluorine atoms, producing the polymer known as Kel-F, removes the charac- teristic low-friction property of PTFE. The friction of this monochlorotri- fluoroethylene polymer is similar to that of polyethylene. Although PTFE has such attractive frictional properties these cannot immediately be utilized in a bearing because the polymer is mechanically weak in comparison with metals and has poor thermal conductivity.The bearing temperature would therefore rise rapidly and the polymer would soften further. Both these disabilities can be overcome by impregnating PTFE in a Rowe 181 porous metal matrix, usually made from sintered bronze powder. The viscosity of PTFE is very high above its transition point (327 "C) (despite the postulated low interaction forces of the molecules) so injection moulding is unsatisfactory. Bearings are usually made by pressing powdered PTFE into the surface of the porous metal and sintering at a temperature above the transition. Other forms of bearing are coated with PTFE but difficulty arises from the marked lack of adhesion between PTFE and other materials.Various techniques can be used, such as weaving PTFE fibres together with cotton or glass fibres, so that the polymer appears on one side and the cotton cloth can easily be attached by resin or other adhesive to a steel backing. Another method which has been used in the USA involves chemical attack by sodium dissolved in liquid ammonia. Surfaces so treated can form adhesive bonds, and it is interesting to notice that they also show high friction. It is also possible t o heat a PTFE surface to about 400 "C, which is believed to liberate free radicals that can then be copolymerized with some more reactive monomer. Friction of layer-lattice crystals There remains an important group of non-metallic solids the complex struc- tural bonding of which directly influences their friction.These are the layer- lattice compounds, often used as lubricants. The commonest layer-lattice lubricant is graphite but in recent years there has been increasing use of molybdenum disulphide. Other compounds resembling MoSz have been studied and some have found application, especially in aircraft and space vehicles. Both graphite and molybdenite occur naturally with a fair degree of purity, but it is usually necessary to refine the ores carefully to remove abrasive impurities. For lubrication purposes molyb- denum disulphide is commonly synthesized, and extensive use is also made of electrographite, produced by heating carbon with a carbonaceous binder at temperatures between 2000 "C and 3000 "C in an electric furnace.The full explanation of the well-known lubricating ability of these materials has not yet been discovered. It is however established that many of the so- called lamellar solids exhibit low friction. Superficially it appears that this is a crystallographic effect. The atoms arranged in the basal planes are strongly bonded but the atomic spacing between layers is much greater and the inter- layer forces are correspondingly weak. It can thus be anticipated that the basal planes will be resistant to penetration and fracture but will shear easily over one another. An analogy is drawn between these crystals and a pack of playing cards. This provides a simple interpretation, but the problem is in fact much more complex and it seems probable that the explanation will eventually be found to be primarily a chemical one, but associated with dislocation be- haviour.In recent years there has been lively controversy on this subject. The first remarkable observation contrary to the elementary concept is that graphite, usually considered the typical lamellar solid, does not show an inherently low friction. If two blocks of graphite are outgassed and slid together in high vacuum, they will wear away rapidly producing a fine dust, and the coefficient of friction will be much higher than usual, about p = 0.5, depending on the source, structure and preparation of the samples. At high speeds it is not even necessary to outgas the graphite. A striking experiment RIG' Reviews 182 can be performed with a rotating copper or aluminium slip-ring and a graphite brush in an evacuated bell jar.As the gas pressure is reduced the sliding mode will change suddenly, and the brush will wear out rapidly in a cloud of dust. After the experiment it will be found that the debris even contains small metallic turnings. Small traces of water vapour or other contaminants will prevent this catastrophic wear and maintain the friction at a low level. Clearly the lubricating characteristics of graphite are associated with adsorption. In contrast, molybdenum disulphide shows low friction over a wide temperature range in high vacuum, but is readily hydrolysed in moist air to Moo3 with production of H2S and SOz.When this occurs, the low-friction property is lost. Any theory of the lubricating action of lamellar solids must consider these marked differences between graphite and molybdenum disulphide, which are also found in various degrees with other lamellar solids. There have been several hypotheses for specific materials, but the most comprehensive interpre- tations fall into two groups. Both emphasize adsorption but one considers adsorption on the basal planes to be dominant while the other attributes the action of vapours mainly to adsorption and interaction at the edges of crystal- lites. To follow the arguments, which are still the subject of controversy, it is necessary to consider the structures in some detail. The characteristic spacing of the C-atoms in graphite is 140 pm in contiguous hexagonal rings, forming the lamellae.The bond energy is about 20 kJ mol-1. The rings lie in parallel planes separated by 340pm. The interlayer bond energy is difficult to specify with precision but has been assessed at values from 2.4 to 4.8 kJ mol-1, depending on the material studied. It is considerably greater than the typical van der Waals bond energy of about 1.2 kJ mol-1 and the reason for this is apparent from the electronic structure of graphite. All C-atoms have two s- and two p-electrons in the second shell, and in graphite one 2s-orbital is hybridized with the two 2p-orbitals to form three equivalent orbitals spaced mutually at 120 O in one plane. These overlap with the orbitals of adjacent C-atoms to form covalent a-bonds.The remaining 2s-electron from each C-atom forms part of a n-electron cloud between the layers that is believed to increase the interlayer bonding. This cloud is also responsible for the electrical conductivity of graphite. Molybdenum disulphide has a similar crystallographic structure, but it is a very poor electrical conductor. Each layer consists of a plane of Mo-atoms between two planes of S-atoms. The separation of the Mo-atoms, which are hexagonally disposed, averages 320 pm but the distance between one Mo-layer and the next is 620 pm. Rowe One hypothesis for the action of adsorbates on the friction of graphite is that the n-bonding is weakened, thereby reducing the shear strength, at least between the layers near the surface.MoS2, having naturally weak van der Waals bonding between the S-layers, shows low friction in the absence of adsorbates. In support of this theory it is shown that graphite can form inter- calation products, for example with bromine, which increase the interlayer spacing and seriously reduce the mechanical strength, even causing the material to crumble into powder. Direct measurements of the strength of graphite show a marked increase, by 50 per cent or more, after thorough out- gassing. Bend tests show strong anisotropy; the effect of vapours is most 183 pronounced parallel to the basal plane. The high friction of outgassed graphite is immediately reduced in low pressures (10-1 torr) of oxygen and other gases and vapours, primarily as a result of chemisorption.Boron nitride is also a lamellar solid, but with a third type of electronic structure. Both B- and N-atoms possess the coordination number 3 and each operates three valence electrons paired off in a-bonds. The B-atoms utilize completely their two s- and one p-electrons while the N-atoms contribute only their three p-electrons. The lone electron pair of the s-subgroup of nitrogen remains and, as the absence of colour shows, is relatively firmly bound. The only action of these electrons is to reinforce the van der Waals forces between layers. The friction of outgassed boron nitride is somewhat higher than that of outgassed graphite. Boron nitride does not form intercalation products with inorganic gases, and it is found that oxygen, hydrogen and nitrogen have only slight influence on the friction of a boron nitride film grown on a boron crystal.The N-atom is however a powerful donor forming organic complexes by disruption of the s electron pair. Thus it may be anticipated that the formation of such complexes will reduce the interlayer bonding and so lower the friction. from p N 0.5 to p = 0.2 by heptane or ethyl alcohol vapours. Experiments have shown that the friction of outgassed BN sliders is reduced The other major type of theory is based on the properties of crystallites of the lamellar solids rather than the detailed atomic bonding. The evidence in favour of this hypothesis comes from experiments with powdered material, usually on a metallic substrate, rather than with solid polycrystalline blocks.A crystallite will show distinctly different adsorption sites, on the basal plane and on the edges of the planes. The latter will be higher energy sites and will be covered preferentially when a limited quantity of adsorbate is present in vapour form. There will in general be three types of interaction between crystallites; these involve two edges, two faces, or an edge and a face. The facial interaction will be relatively weak whether adsorbate is present or not, whereas the energy associated with the edges will be large when the material is outgassed, and greatly reduced by adsorption. Quantitative measurements of the adsorption of water vapour on graphite show that amounts sufficient only to saturate the edge bonds will fully reduce the friction.Much greater quanti- ties of water vapour would be required to produce full intercalation between all layers. It is further pointed out in favour of the crystallite theory that the orientation of graphite during sliding is facilitated by an adsorbate. Diffrac- tion patterns show that if graphite is slid over a metal in air, the friction is low and the deposit is highly orientated with the basal planes parallel to the surface. In high vacuum there is much greater wear and the graphite track is completely disorientated. Many other lamellar solids of the MoS2 type show low inherent friction. One interesting compound is Ti12 which was first found to have lubricating proper- ties as a reaction film formed by heating clean titanium in iodine vapour.Ti12 is hygroscopic and is readily decomposed in moist air, but recent work has shown that iodine dissolved in a hydrophobic carrier such as butylbenzene is an effective e.p. lubricant for use with titanium, chromium steels, and other alloys notoriously difficult to lubricate. This is discussed more fully later (p. 193). Various sulphides, halides, selenides and tellurides with lamellar RIC Reviews 184 crystal structure have been examined for possible aerospace lubrication purposes. It is likely that some of these will be useful, and may also find applicationin high-temperature terrestrial devices, though they are all at present expensive. MoS2 provides low friction in vacuum at temperatures up to 900°C. Both graphite and boron nitride show inherently low friction in vacuum at temperatures in excess of about 1500 "C.The strength of graphite is also lower at these temperatures, despite reports to the contrary. It has been suggested that graphite is unique in showing a marked increase in strength with increasing temperature, up to 2000 "C. It seems likely however that the increase is due to removal of adsorbed material. If the graphite is outgassed for a long period at a temperature of about 2000 "C it shows a relatively high shear strength of about 1.5 kg mm-2 at least 50 per cent greater than under ordinary conditions. This is maintained at room temperature if contamination is prevented. A short exposure to air immediately reduces the strength.Usually it is not feasible to form these lamellar compounds directly on the surfaces to be used, and they are often applied as powders. An important feature is then the ability to form a coherent film on the substrate. It has been suggested that the dominant characteristic for this purpose is the ability of the lamellae to penetrate into the metal or other surface during sliding, so that they form barriers behind which the powder may accumulate. An appreciable resistance to deformation and fracture of the individual lamellae is con- sequently required. Talc, for example, is a relatively soft hydrated magnesium silicate corresponding to Mgs(OH)zSi4Olo and is unable to penetrate a steel surface. Talc, although inherently a low-friction material, will not lubricate steel but will lubricate gold, whose penetration hardness is about 30 kg mm-2 compared with 150 kg mm-2 for steel.Mica has a structure somewhat similar to that of talc, but usually has strong potassium bonding between the layers, with a representative composition K(OH)2A12(Si3Al)Olo. The shear strength of mica has been measured directly and is found to be 10 kg mm-2. The fric- tion of mica is normally high, but the potassium atoms can be progressively leached out, weakening the structure. Carefully controlled experiments of this type might be very useful in elucidating the general problem of the low-friction properties of lamellar solids. Solid lubricant dispersions. A convenient and widely-used method of bonding solid lubricants to a surface is to disperse the finely-divided material in a volatile solution of a synthetic polymer, which is then sprayed or painted onto the article.Thermosetting binders such as phenol-formaIdehyde resin are commonly used. After appropriate drying and curing these provide a durable low-friction film with graphite, polytetrafluoroethylene, molybdenum di- sulphide or other solid lubricants. Most commercial dispersions contain dispersoids ranging in size between 10-4 and 1 pm mean diameter. The surface area of dispersed graphite is often of the order of 100 m2 8-1. Some are true colloidal dispersions but the majority contain larger particles which remain stably suspended for limited periods even in highly viscous fluids.The solid lubricants are usually prepared by grinding and ball milling. Since large mechanical strains are produced during these processes, some areas are likely to have a high surface energy and precautions must be taken, for example to Ro we 185 13 prevent undue formation of Moo3 from MoS2. The smallest particles used industrially are prepared by precipitation. The solution of binder resin or other material must be selected so that it is easy to paint or spray the disper- sion, but the solvent must evaporate readily without allowing the lubricant particles to reaggregate. Most commercial dispersions are stabilized, whether they are intended to provide bonded films or to be used directly. Small quantities of a surfactant additive are used which, by adsorbing on the particles and reducing their surface energy, diminish their tendency to reaggregate.It is probable, however, that the situation is usually more complicated than this simple interpretation would suggest. The particles are, in general, likely to acquire charge by adsorption of ions, and the charged surfaces will then attract ions of opposite charge forming an electrical double layer. The stability of the dispersion is believed to be associated with the zeta potential between the adsorbed liquid and the remainder. In addition to the stabilizing action of surfactants in lubricant dispersions, other surfactants are used to control or modify the rheological and thixotropic properties of colloids. It is not immediately obvious for example that the mechanical yield stress of putty is dependent upon surface chemical action.Friction of metals Historically, metals have been studied much more extensively than the solids we have so far considered, but chemically and physically they are more com- plex and they exhibit simultaneously many of the features discussed separately above, especially when contaminated by the atmosphere or by lubricants. At ordinary temperatures metals behave as elastic and plastic solids, but if the temperature exceeds about half the melting point, in degrees absolute, the elastic contribution can usually be ignored. The metal is then relatively soft and deforms plastically with appreciable creep. Thus lead (m.p.600 K) exhibits 'high-temperature' behaviour at room temperature (20 "C, 293 K). Indium (m.p. 429 K) shows still less elasticity. Experiments with lead and indium show that their frictional behaviour is somewhat similar to that of krypton. As long ago as 1724 an experiment was performed which demon- strated appreciable adhesion, with a tensile strength of 150-300 kg mm-2, between two cleaned lead balls which had been placed in contact under a load, with a slight twisting motion. After the introduction of the asperity-welding theory, two centuries later, it was believed that this type of adhesive junction or weld was primarily responsible for the frictional resistance between two metallic sliders. The coefficient of friction would then be given, as already discussed in connexion with the van der Waals solids, by p = F/W= AslAp because the contact area A is determined at any given load W by the yield pressure p of the metal, and the shear strength s of the weld is presumed to be equal to that of the solid metal.Since this theory ignores elastic deformation, it would be expected to apply most accurately to lead and to indium. Numeri- cally the yield pressure p is known to be related to the shear strength s of a RIC Reviews 186 plastically-deforming body, p being approximately six times s. A coefficient of friction of about 0.15 would thus be expected. Experimentally, however, very high values up to five or even 10 are found with indium, even in the open air, accompanied by strong adhesion such that a pull equal to several times the original load may be required to separate the specimens in a direction normal to the interface.Very similar results can be found with outgassed platinum or nickel in high vacuum. The explanation has been given in terms of plasticity theory, which does not allow s andp to vary independently. It is the combina- tion of shear stress and normal compression which causes the metal to yield plastically, and the relationship can be formally written p2 + as2 = constant = p: When a load is first applied, s = 0 and the area of contact A0 is determined, as in the simple theory, by the yield pressure, now designatedpo. If a tangential force is applied, a finite shear stress s is introduced and consequently the pressure on the contact is reduced.Physically this happens by plastic flow occurring so that the two sliders sink together and the area of intimate contact increases to a value A , greater than Ao. The shear strength of the junction is correspondingly increased in the ratio A/& and the apparent coefficient of friction is similarly increased, assuming that the applied load remains constant. This junction growth can easily be seen with a clean lead or indium specimen on a clean glass plate, observing the contact through the glass. More sophisti- cated measurements show similar results with denuded metals, in good agree- ment with the theoretical predictions. This process of cold welding is in fact the basis of an industrial technique of roll bonding used to make, for example, aluminium heat-transfer panels by rolling two sheets together.These will, if suitably cleaned, weld together everywhere except in the regions previously painted with a lubricant film. The painted pattern can then be converted into a system of cylindrical pipes by applying air pressure to bulge the aluminium outwards in the unwelded regions. In the complete absence of lubricant or adsorbate during a friction test, the theory suggests an indefinite increase in the effective coefficient of friction. With the best available conditions of cleanliness, values of p up to 10 or more have been recorded, and with single crystals of copper even up to 100. Very slight adsorption quickly reduces these values.This can be accounted for theoretically by supposing that the junction growth is limited at a lower value of shear stress when an interfacial film is present. It has been shown by delicate measurement over distances of up to 20 pm that the initial growth under the action of tangential stress follows the same pattern whether lubricant is present or not, but the lubricant limits the growth, and sliding then occurs. Formally this can be expressed in the above equation by allowing s to increase only to the limiting value si. Thus at the moment when macroscopic sliding begins : p2 + as? = constant = p ; = as, 2 where Sm is the yield stress of the metal in the absence of normal stress. If it is assumed that si is some finite fraction p of Sm, the coefficient of friction can be Rowe 187 written : If p N 1, as for denuded metals, this predicts very high coefficients of friction, as found experimentally.Under more common conditions the surfaces are covered with adsorbed films which greatly reduce the shear strength at the interface, and even the most carefully cleaned surfaces in air rarely show coefficients of friction much above unity. Under these conditions no adhesion between the specimens can usually be detected because the elastic change of shape on removal of the load fractures the small welds which have been formed. The reality of these welds is nevertheless established by experiments in which one slider is made radioactive. Transfer of metal from one surface to the other is then easily observed.If the surfaces are well lubricated, the interfacial shear strength will be much lower. If /3 is less than about 0.2, as it will be for all lubricated sliding, the equation reduces to The frictional resistance is thus due entirely to the average, low shear strength of the interface, which need not contain any contribution from metallic junc- tions. In fact, radiotracer experiments show that under lubricated conditions the amount of metallic contact makes a negligible contribution to the friction, though it is clearly important in wear. It is curious that the experiments performed to examine the deficiency of the adhesion theory confirmed on the one hand the possibility of strong adhesion and led to a more complex formulation which eventually reduced to the simple formula for lubricated conditions, but on the other hand showed that in this form the friction was not dependent upon metallic adhesion.The determining mechanical parameters for most practical purposes are the yield pressure or elastic properties of the metal substrate and the shear strength of the lubricant layer. It is therefore important to consider these layers in some detail. Thin-JIm lubrication of metals It is apparent that the frictional force between two sliding bodies is primarily determined by the viscous properties of the lubricant when a thick film of fluid is present at the interface. For many years a clear distinction has been drawn between these conditions and the conditions envisaged in the preceding section where metallic interaction was the dominant feature.In recent years the transition between these regimes has been seen to cover a wide range, possibly even the majority of normal applications. On the one hand elasto- hydrodynamic theory now shows that hydrodynamic influences extend to much lower speeds and heavier loads than had previously been recognized; while on the other hand the development of the metallic friction theory just considered shows that the overall strength of the interfacial layers determines the friction, whether or not there is metallic contact. Elastohydrodynamic experiments suggest that with reactive lubricants the deduced viscosity is very RIC Reviews 188 much higher in a region within about 100 nm of the surface.It is not yet established whether or not this surface-zone anomaly can be associated with the earlier postulate that boundary lubricants form a solid reaction product at the surface of metals. We shall confine our attention to the latter. Boundary Zubrication. Any experiments intended to show boundary lubrication should be conducted at low speeds and under heavy loads, to avoid contribu- tions from hydrodynamic action. Typically speeds of about 0.01 cm s-1 and loads of about 2 kg have often been used with a hemispherical slider of 2-3 mm radius sliding over a flat plate. Considerable departures from these conditions can however be made without significantly affecting the conclusions. The most effective boundary lubricants are reactive organic fluids.Tests made with low-carbon steel sliders using various members of the paraffinic series as lubricants show intermittent motion and relatively high friction for all the paraffins which are liquid at room temperature. When the molecular weight exceeds about 300 the paraffins are solid at 20 "C and smooth sliding is found with a coefficient of friction p N 0.07, unaffected by further increase in molecular weight ( M ) . Alcohols exhibit very similar behaviour. For both series there is a marked transition at the melting point. If a long-chain com- pound is used and the specimens are heated, the friction becomes irregular and there is an increase in the surface damage as soon as the melting point is reached. The fatty acids also show this general trend, but with an important peculiarity.The coefficient of friction decreases with increasing molecular weight, but the smooth sliding condition with p N 0.08 is reached at about M 100, while the lubricant is still liquid. The transition temperature is also found to be well above the melting point. For example lauric acid (CllH23C02H) melts at 44 "C, but it continues to lubricate copper at tempera- tures up to 100 "C. Above this temperature both the friction and the metallic transfer increase, though the effect is reversible with temperature. Further study of the fatty acids shows that lauric and stearic acids remain effective at room temperature even when present in concentrations as low as 1 per cent solution in mineral oil.Some influence can be detected at still lower concentrations, down to 0.01 per cent. Detailed assessment of the amount of fatty acid necessary to provide low friction over a given surface area suggests that this is of the order of one or two molecular layers, if uniformly distributed. It is possible to examine this suggestion with precision, since a single molecular layer can be deposited on a surface from a Langmuir trough. A drop of stearic acid dissolved in ether is placed on water and allowed to spread. The film formed will be highly orientated with the polar hydroxyl group at the water surface, and it will spread until it is one molecule in thickness. It can then be picked up on a metal surface, in the same way that an electron microscope replica is picked up.The procedure can be repeated to give any desired number of molecular layers. A sliding experiment is subsequently conducted with a hemispherical slider passing over such a coated surface. The results show that even a single molecular layer will provide low friction, p N 0.1, but the film is easily worn away. With 50-100 layers the lubricant will last for many traversals. If a solid alcohol or ester is used it is found that even with a large molecule such as cholesterol a monomolecular film is not sufficient, again illustrating Rowe 189 the superior lubricating properties of the fatty acids. The durability of all such lubricant layers increases as the chain length is increased, but is practically unaffected by sliding speed within the range 0.01-1 cm s-l.It is apparent that if the reactive liquid is present the surface film can be repaired continuously and good lubrication will persist indefinitely. There is however evidence that surface films produced by reaction, for example copper stearate, are inherently more durable than layers deposited in solid form with a Langmuir trough. On the contrary, silicone fluids which do not react with the metal surface, and also prevent reaction with atmospheric oxygen or water vapour, are found to increase the friction above the values found in open air under nominally clean conditions. There is thus a strong suggestion that reaction between the lubricant and the metal is an important feature. This can be studied in more detail using radiotracers and electron diffraction.Reaction of boundary lubricants with metals. Sliding experiments show that lauric acid in paraffin solution gives low friction on Cu, Cd, Zn and Mg but the lubrication is no better than that of a paraffin of similar chain length, dodecane (C12H26), on Ni, Cr, Pt, Ag. On Fe and A1 the coefficient of friction is intermediate, but higher concentrations of fatty acid lubricate iron well. This classification correlates with the chemical reactivity of these metals with lauric acid. Direct evidence of chemisorption or reaction can be obtained by using radioactive metals. Thin foils are immersed in the solution of fatty acid for a short time, and washed with hot benzene to dissolve away any surface layer which may have formed.The benzene is then examined for radioactivity due to dissolved metallic soap. Cu, Zn and Cd show strong activity, having formed much more than a single molecular layer of reaction product. Pt, Au and Ag show no reaction. Similar experiments with alcohols show no reaction on any of the metals. There is slight reaction with esters, possibly attributable to hydrolysis. The reaction can also be followed by friction experiments in a vacuum chamber. If two copper sliders are thoroughly denuded by heating to high temperature in high vacuum they will subsequently seize if sliding is attempted at room temperature in the vacuum. The coefficient of friction is then rather meaningless, but values of 100 have been quoted.If a volatile liquid fatty acid such as hexanoic (CSHllCOzH) is suitably purified and degassed and exposed in the chamber, the vapour will be absorbed on the copper specimens. The friction is then reduced to about p = 1.2. If the excess vapour is pumped away this value is increased to about 2.0, suggesting that the major influence on friction arises from a chemisorbed film. A further drop in friction, to ,u N 0.8, occurs if oxygen is admitted with the vapour, and a subsequent reduction to p N 0.4 occurs if the system is left for a few hours. It is believed that a thicker film is established during this time by chemical reaction and diffusion of metallic ions, and this appears to be facilitated by the presence of water vapour. It is interesting to note that the friction of cleaned copper sliders shows about the same value (0.4) after the sliders have been exposed for some time in the laboratory atmosphere.Contamination of ordinary labora- tory specimens is consequently a very important factor in their frictional performance. The importance of atmospheric oxygen is also shown by other RIC Reviews 190 experiments in which a surface is scraped ahead of the slider, under a pool of degassed lubricant which excludes the air. The friction is then much higher than usual. The lubricant layer formed by reaction of copper, oxygen and a fatty acid is likely to be a copper soap. If a preformed copper laurate, for example, is smeared onto a copper surface the friction is the same as that of copper lubricated with lauric acid in air.Moreover, the low friction persists for both as the temperature is raised to 100 "C, as mentioned before, well above the melting point of lauric acid. Electron diffraction studies confirm this con- clusion. It is necessary to use low-energy electrons in order to obtain clear diffraction patterns from the surface layers, without the confusion from the substrate which would arise if more penetrating ,%radiation or X-rays were used. Layers of a solid paraffin show a characteristic rhombic pattern with the carbon chains perpendicular to the surface. Fatty acids show an interesting structural transition. The outer layers form into monoclinic crystals but the first molecular layer at the metal surface shows a different pattern, indicating adsorption of the polar groups at the interface with the long hydrocarbon chains normal to the surface.The outer layers become distorted by rubbing, but the monolayer is remarkably stable. It is found that the greatest degree of orientation of the monolayer is produced in films formed by reaction between the metal or oxide and the fatty acid. When the temperature of the substrate is raised, the electron diffraction pattern shows disorientation as the melting point of the fatty acid or of the reaction product is approached. The diffraction pattern of the metal does not however become clear until a considerably higher temperature is reached, showing that soap films remain adsorbed well above their melting points.Influence of carrierfluid. Fatty acids are not usually employed as lubricants in concentrated form. They can conveniently be dissolved in paraffinic oils, when a concentration of a few per cent will provide adequate lubrication. Not only is this less expensive, but the risk of corrosion is reduced. It has been found however that the performance is influenced by the carrier fluid. This may be assessed in a commonly used four-ball test, in which a 0.5 in diameter steel ball is rotated at high speeds while loaded in contact with a nest of three similar balls. As the load or speed is increased, a point will be reached at which the lubrication breaks down and the balls seize into a solid block. Alternatively the surfaces may be seriously damaged, or scuffed, without seizure occurring.It has been found that the maximum load-bearing capacity is exhibited when the carrier fluid has exactly the same chain length as the fatty acid. For example, palmitic acid (CIS) showed a scuffing load of 500 lb when dissolved in hexadecane, whereas fatty acids with either 14 or 18 C-atoms showed scuffing at about 400 lb load under the same conditions. Mechanical strength of a monomolecular soap layer. The electron microscope studies and friction experiments mentioned above suggest that the adsorbed films of fatty acids, and especially of reacted metallic soaps, are remarkably strong. Direct measurements of the shear strength of single molecular layers of calcium stearate have been made. The films were deposited from a Lang- Ro we 191 muir trough on carefully cleaved mica which was molecularly flat. The back surfaces of the mica had previously been partially silvered so that the area of contact between the two cleaved faces could be examined with multiple- beam interferometry.The interface between the two strips of mica, with the deposited soap film, was then subjected to normal and tangential forces. The tangential stress sufficient to cause macroscopic motion was recorded as the shear strength of the monolayer. This was found to be about 250 g mm-2 for calcium stearate. A similar value has been deduced from sliding experiments with a pointed crystal of sodium stearate on a clean platinum surface. The coefficient of friction in the latter experiment was about p = 0.3.If the same fatty acid is used as a lubricant film between two platinum sliders the friction is lower, p = 0.1, but far from being as low as the simple theory of friction would suggest. The yield stress p of platinum is about 40 kg mm-2, so if p P s/p we should expect 0.25/40 2 0.006. The difference from p fl 0.1 has been explained in a general way by supposing that the shear strength is greatly increased by the pressure. This is at least plausible, by analogy with the known strong dependence of viscosity of organic liquids on pressure, but little detailed work has been done on this. It seems likely that even at this low shear-stress level some junction growth occurs. An attempt has been made by Cameron to evaluate the shear strength of an adsorbed hydrocarbon layer theoretically.The long-chain molecules are assumed to be orientated perpendicular to the opposing surfaces. It is then possible, as shown by Miiller, to calculate the van der Waals attraction be- tween respective -CHz- groups and to sum these over the molecules. The repulsive forces can be calculated from the adjacent hydrogen atoms, the difference between these two values giving the equilibrium energy, which can be considered as existing at a potential well, as for single atoms (p. 142). This provides an upper limit to the force required to move each hydrocarbon chain from one equilibrium position to an adjacent one. A typical value of the shear strength derived on this basis is about lo4 N cm-2 or 10 kg mm-2, which is considerably higher than the measured value for calcium stearate and dem- onstrates the possibility of the frictional resistance being entirely attributable to molecular interaction in the lubricant.Extreme-pressure lubrication. We have seen that the upper temperature limit for the fatty acids and other boundary lubricants is not more than about 200 "C. For lubrication at higher temperatures it is usual to employ organic sulphur, chlorine or phosphorus compounds. These are generally known as 'extreme-pressure' or e.p. lubricants, because they were first used in hypoid gears in automobiles where the local pressure is high. I t is recognized however that the name is really inappropriate because the mean elastic pressure referred to is lower than the yield pressure occurring locally in all sliding contacts.The action of these compounds is now known to depend primarily on temperature. RIG' Reviews Extreme-pressure compounds have been most extensively studied with steels. It is found that an iron chloride film formed by direct reaction with chlorine gas will provide low friction in a vacuum chamber if it is allowed to build up to a thickness of about 100 nm so that its normal lamellar crystal structure is developed. The lubrication will persist at temperatures up to 192 about 300 "C. In the open laboratory a film of FeC13 deposited on a steel specimen will also provide low friction up to this temperature. Similar results are obtained if the surfaces are lubricated with stearyl chloride, even if it is used as only a 0.5 per cent solution in a paraffin.It can thus be concluded that the action of an e.p. compound is to produce a surface film, of iron chloride for example, which will withstand higher temperatures than the organic materials. The process is essentially one of controlled corrosion. Hydrochloric acid will form a satisfactory lubricating layer on steels, but is too corrosive. The compound used should remain stable in the temperature range over which boundary lubricants are effective, but it should decompose to liberate the inorganic ion when the temperature rises above this. Extreme-pressure additives are, therefore, commonly used in conjunction with boundary additives. Various compounds of chlorine can be used.Even inorganic com- pounds such as SeClz can provide e.p. lubrication, but the most widely used chlorine additive is commercial chlorinated paraffin. This is mainly composed of long-chain paraffins such as CH3.(CH&s wCH2C1, but the commercial material contains an excess of dissolved chlorine which is thought to be labile. It is important to recognize that moisture may adversely affect these lubricants. HCl may be formed in the bulk lubricant if it is exposed to water vapour or water, with consequent risk of corrosion. The chloride films themselves are easily hydrolysed and may lead to rusting, as well as losing their lubricating effectiveness. Sulphide films are less sensitive to moisture and are stable at higher temperatures.The reaction film of sulphide formed on iron exposed to HzS at an elevated temperature will continue to lubricate in the absence of air at temperatures up to about 750 "C, but the friction is considerably higher than that of the chloride film, probably because FeS does not have a lamellar structure. Sulphide films can be formed with ammonium polysulphide, or with flowers of sulphur dissolved in a paraffinic oil. Sulphurized oleic acid is a commonly used additive combining e.p. and boundary properties. Sulphur compounds are used in lubricating oils when local high temperatures are anticipated and where the general operating conditions are too severe for chlorinated oils. For the former purpose they are usually compounded with chlorinated and boundary additives to provide low friction over the lower temperature range.It is also believed that the chlorine assists in the reaction between the metal and the sulphur. The same compounds are used on other metals, for example the copper alloys. There are, however, complications ; long-chain paraffinic sulphides and some other compounds such as dithiocyanates will provide e.p. action on steel but not on copper. Mercaptans will react with copper and provide good lubrication over a wide temperature range, but are unsuitable for general use. Staining, which may be inevitable if reaction is to occur, can be undesirable in many applications where the appearance of copper or aluminium alloys is important. For aluminium even the fatty acid reaction in boundary lubrication may spoil the surface.It seems likely that a new family of e.p. compounds based on iodine will become available. Vacuum-chamber experiments showed some years ago that an effective lubricating film of Ti12 could be formed by heating titanium in Rowe 193 iodine vapour. This compound has a lamellar structure similar to that of MoS2 and shows low friction at temperatures up to 250 "C. Because of its inert surface oxide, which will not react with common boundary lubricant additives, titanium is otherwise very difficult to lubricate. Fatty acids and chlorinated oils have little more effect than paraffins. Recently it has been found that iodine dissolved in organic fluids can provide e.p. lubrication for titanium. Butyl- benzene is reported to be a particularly effective solvent because it is hydro- phobic and prevents hydrolysis of the Ti12 which is otherwise easily decom- posed.It is reported that simple solutions of iodine in paraffinic oil can also lubricate titanium, stainless steel and even glass. There may be a fruitful field for development of these fluids if health hazards can be avoided. MECHANICAL AND CHEMICAL INTERACTIONS IN WEAR PROCESSES Wear is of great commercial importance; it has been estimated that losses due to wear amount to many millions of pounds sterling each year in Britain alone. The commonest form of wear is probably mechanical abrasion, but in most circumstances physical and chemical properties also contribute, in a complex manner. The study of wear is complicated not only by the interaction of many factors, but also by the extremely small quantities of wear which are normally pro- duced.It has often been said that the change in critical dimensions between a new engine and one which has been completely worn out may be no more than a few thousandths of an inch. Direct wear tests on operating equipment are thus expensive and time-consuming. It is also very difficult to maintain satis- factorily controlled conditions throughout such tests. Even in laboratory experiments with greatly accelerated rates of wear it is notoriously difficult to obtain reproducibility. The most reliable results can be produced with specimens which have been carefully run-in over a long period to establish steady conditions.Most detailed conclusions therefore relate to materials in this state. This is realistic, since most machinery is run-in under fairly close control, yet it must be recognized that transient effects of vibration, extraneous abrasives, atmospheric humidity and other variables may temporarily increase wear rates by large factors and may even destroy the carefully-established equilibrium conditions. It should also be recognized that simplified laboratory experiments, and even those conducted with sophisticated test rigs, always limit the number of variables. Indeed, simple wear-test machines are used more extensively by the manufacturers of lubricants, for control and development purposes, than by the manufacturers of machinery. In this section we attempt to describe briefly the main modern conclusions about wear generally, and then to show the significance of chemical factors in the wear process.This leads to a brief consideration of corrosive wear, stress corrosion and fatigue. The three stages of wear It is found in many instances that the operating life of a machine, for example an automobile engine, is characterized by three fairly clearly defined phases. When the machine is first commissioned there is a relatively rapid rate of removal of material from the surfaces. The major irregularities left by the RIC Reviews 194 cutting, grinding or other fabrication process are worn away and the surfaces come into better local and general conformity.The rate of removal of material decreases progressively during this period and eventually reaches a low steady value when the machine is fully run-in. In the absence of adventitious circum- stances this low rate will persist with little alteration for a very long time, in fact for the useful life of the machine. The wear rate will finally increase again, for a variety of possible reasons. Steady wear may have increased the mechani- cal clearance between the sliding members to such a value that a bearing, for example, is loose and allows percussive loading or vibration. The wear debris, which is usually strain-hardened or oxidized and consequently abrasive, may also have accumulated in sufficient quantities to aggravate the wear.Often local corrosion leads to pitting and to stress concentrations which initiate the catastrophic stage of wear. Even in the absence of these causes, long repetition of cyclic stressing may produce surface or subsurface fatigue failures, so that small fragments of the surface break away. This is particularly likely to occur in gear teeth and in ball bearings where high local loading occurs for short periods at frequent intervals. Any of these conditions, once established, will tend to become rapidly worse. The machine will then need to be renovated or replaced. Earlier we mentioned the co-operative chemical and mechanical actions of formation and removal of surface films with e.p. additives, which are beneficial in the running-in stage. The steady state of wear is discussed later in this section. To prolong the life of a machine it is clearly desirable to delay the onset of the final stage.This is mainly the province of the mechanical designer, who endeavours to avoid high stress levels, vibration, and shock loading ; to exclude abrasive materials and to provide adequate lubrication. Proper attention to the chemistry of the overall system can however be very important. All the materials must be chosen to avoid corrosion due to direct chemical or electrochemical attack, and the lubricants must have no deleterious effects even if they decompose or oxidize in use. Particular care must be taken if, because of relatively high temperatures, it is necessary to use chlorinated or sulphurized lubricants, which might lead to stress-corrosion failure as well as direct acidic attack of the metals.The chemical nature of the lubricant has, of course, a major positive influence in reducing the tangential stress at a sliding surface and in preventing metallic transfer which can otherwise easily become cumulative and lead to early failure. Steady wear conditions Much of the available fundamental information on wear is derived from experiments with simple apparatus of ‘pin and ring’ type, though many other varieties of apparatus have been used. The pin is a cylindrical test specimen with a flat or hemispherical end, held against the periphery or face of a rotat- ing ring or solid disc. The loss of material from the pin is usually measured by simple weighing, or by optical assessment of the wear scar diameter if a hemi- sphere is used.In more refined experiments it is possible to detect much smaller quantities of wear by spectrographic, microchemical or radiotracer techniques. The latter can reveal not only the total loss of material but also the amount transferred from one surface to the other. Rowe 195 It has been shown in this way that metal transfer can be a very important feature of wear. Using a radioactive pin of 60 Cu : 40 Zn brass, for example, it is found that the amount of pin material transferred to a steel ring increases steadily at first but soon reaches a steady level. If at this stage the brass pin is replaced by an inactive but otherwise identical one, the amount of active brass on the ring decreases with time, but none is transferred back to the pin.It is therefore concluded that this wear process occurs continuously in two sequential steps. The pin material is first transferred to the ring, mainly by metallic adhesion. The thin layer of discrete metal-transfer particles then oxidizes, under normal atmospheric conditions, and becomes detached. Electron microscopy and electron diffraction examinations of the debris show that it consists mainly of fine iron oxide (wFez03) particles, many of which are only about 10 nm across. If an experiment of this type is performed in an atmosphere of argon, the transfer is enhanced but the film cannot oxidize. The debris is then mainly metallic and consists of much larger fragments.It seems probable that these fragments become detached by the formation of cracks due to surface fatigue. This ‘severe’ type of wear, in contrast to the more common ‘mild’ type with an oxidative process, can also occur in air if the applied load is sufficiently large. It is of course very damaging and would rapidly lead to failure in practice. Even the ‘mild’ wear in an unlubricated experimental condition such as this is very much more rapid than could be tolerated in machinery operating in a normal way. It is however believed that transfer with subsequent chemical conversion and detachment is a feature of many common wear processes. Most experimental and practical measurements of wear suggest that during the second, or steady-state, stage of wear the volume of material removed is very nearly proportional to the distance of sliding.This is often regarded as one of the laws of wear. It is valid for a remarkably wide range of materials and conditions, and in fact underlies the general concept of steady-state wear. It is convenient to define a wear rate, to which we have already referred several times, as the volume (or mass) of material removed in unit distance of sliding. The numerical value of the wear rate for different combinations of materials varies over a much wider range than the coefficient of friction. Thus, under dry conditions two mild steel sliders may give a wear rate equal to 2000, in units of 10-lo cm3/cm, whereas two dry sliders of sintered 94 WC : 6 Co may give only 0.03, in the same units.In the presence of a lubricant the wear of the mild steel may be reduced by a factor of 1000 but that of the carbide may even increase, as we shall see. The corresponding range of coefficients of friction is from 1.5 for the steel sliders to 0.2 for the carbide sliders without lubricant, reducing to about 0.08 for both with a good boundary lubricant. The influence of small traces of contaminant is much greater on wear than on friction, and even small surface irregularities can significantly increase the wear. This can be examined in detail with a pin and ring apparatus in which a spiral track is followed so that the ring surface is traversed only once. It is convenient to use a hemispherical WC/Co slider which has been highly pol- ished and etched after being irradiated in a neutron flux. Either l87W or 6OCo radioisotopes can then be detected in the wear debris. If such a slider is traversed once over a well-prepared copper plate, the transfer is uniform and RIC Reviews 196 the transferred material adheres to the plate.Subsequent traversals of the same path obscure the detail because the first fragments act as an abrasive lap. The wear rate is increased in a second traversal, and loose debris is produced. Abrasive particles of silicon carbide or alumina embedded during metallo- graphic preparation of the copper similarly increase the wear. It has however been shown using radioactive 28A1, despite its short half-life of 2-3 min, that normal metallographic preparation technique removes the residual grit from each grinding paper at about the time that the visible scratches are removed by the next in the sequence. Careful preparation, finishing with 4/0 grade Sic paper, if performed under a flood of methanol or benzene, will produce a copper surface which is free from damaging debris or scratches.On such a copper plate the transference from the radioactive WC/Co slider can be detected by exposing an X-ray film for a period of up to 24h and then developing the image. Within the detection limit imposed by the grain size of the film, which must be rather coarse (about 10 pm) to achieve adequate film speed, the deposit of tungsten carbide will be found to be uniform over areas of more than 1 cm2.The wear rate will however vary from one part of the plate to another, and this variation can be associated with the drying pattern of the benzene or other solvent. To obtain uniform and reproducible results it is necessary to produce a chemically uniform surface on the copper, as well as removing the mechanical scratches. A preparation technique has been developed which has given reliable results over a number of years. A metal plate is first abraded unidirectionally under light pressure with grade 0 emery paper under a flood of methanol. When the surface is uniformly scratched, the plate is thoroughly washed with methanol and then abraded on grade 2/0 at right angles to the original direc- tion until none of the first set of scratches remain.This procedure is repeated with 3/0 and finally 4/0 emery papers. After the final washing the plate is leached with a continuous flood of 1 : 1 solution of concentrated HCl in distilled deionized water, gradually changing without exposure of the plate to a flood of purified water alone. It is then dried rapidly in clean warm air. The method has been proved by the following experiment. Immediately after the final abrasion and methanol wash, one third only of the plate was leached with 1 : 1 HCI and another third with 1 : 3 NH40H, both separately followed with purified water and dried. The hydrochloric acid dissolved any copper oxide on the surface and detached the organic residue left by the sol- vent, without attacking the copper itself, whereas the ammonium hydroxide reacted with the metal.The central zone was untreated, so that there were three distinct zones on a single plate, separated by boundary regions. In one of these boundaries there was a residue of CuClz, which is a soft solid of lamellar structure, and in the other a residue of Cu(OH)2, which is a mild abrasive. A polished radioactive tungsten carbide slider traversed over this plate in a spiral showed increased wear in the hydroxide boundary and de- creased wear in the chloride boundary, but the wear rate in the leached and water-washed zones was uniform and practically identical whether HCl or NH40H had been used. The surface film in each of these zones was thus attributable to the action of water and air only.It is probable that it was a Rowe 197 chemisorbed hydroxyl which was rapidly converted by the air to a thin film of copper oxide. A remarkably sharp change in the wear rate occurred at each boundary region, reaching full intensity on each side within a distance of 0.2 mm, though the time taken to traverse this distance was only about 3 ms. This illustrates the extreme sensitivity of wear to the local environment, and the care necessary to obtain reproducibility. In comparison with this standardized treatment, the film formed by exposure to benzene, possibly a copper phenate, shows a reduction in wear by a factor of about 10. Care must therefore be exercised in interpreting friction and wear data obtained with conventionally degreased surfaces.Even electrolytic degreasing may leave a chloride or similar residue. For carbon and alloy steels a solution of 2 per cent nitric acid in alcohol (commonly used by metallurgists for etching, and known as Nital) may be used instead of ammonia. This may be preferable to the use of HC1 since any inadvertent residue will increase the wear locally and may be more readily detected. Influence of oxygen and water vapour on wear and metal transfer The experiment described above illustrates the important influence of chemically-formed films on the wear of metals. The results obtained in high vacuum (pp. 187-188) suggest that in the absence of all contaminants the friction of metals is very high and is accompanied by extensive interfacial welding and damage.There have been various suggestions that certain combi- nations of metals will not readily adhere; for example those which will not form a solid solution. This appears to be refuted by the observation that iron and silver, which are well known to be mutually insoluble (0.0005 per cent) can be pressure-welded between heavy rollers at room temperature without difficulty. A further possibility is that metals will not adhere well to B sub- group elements. The explanation proposed is that the B sub-group metals display covalent bonding which makes them relatively weak and brittle. Such bonds as may be formed with other metals are hence supposed to fracture readily. This general problem of observable adhesion between metals and between metals and other materials is of great interest, and possibly of theoretical and practical importance, but results are very difficult to obtain because of the necessity for providing accurate location, movement and force measurement for specimens under the highest possible degree of surface cleanness.As we saw earlier (p. 141), it is necessary to maintain a vacuum at RIC Reviews 10-9 torr or better if the surfaces are to remain sensibly uncontaminated for several minutes. In practice, studies of this type are now conducted under ultrahigh vacuum conditions at about 10-11 torr, though much of the early work on adhesion was of necessity performed at 10-6 to 10-7 torr. The speci- mens can conveniently be cleaned by argon-ion bombardment. Interesting results relating to the crystallographic dependence of adhesion are now appearing.For example lanthanum shows a marked decrease in adhesion to a second lanthanum specimen at a temperature below which the crystal struc- ture changes from body-centred cubic to hexagonal close-packed. Slight traces of oxygen or other atmospheric contaminants will reduce the friction of many and probably all metals considerably and produce a spectacular reduction in 198 surface damage. An associated transition in wear character can occur between ‘severe’ and ‘mild’ wear even in open air, as we have seen earlier in this chapter. A tenacious oxide skin is thus advantageous. Some oxides are sufficiently soft at high temperatures to provide a positive lubricating action.CuO, for example, is softer than copper at temperatures above about 600 “C and can be used to lubricate copper in extrusion processes. Most oxides are however harder than the metals from which they are formed, and if broken during the sliding process will provide abrasive particles that increase the wear. More subtle distinctions can be drawn on the basis of mechanical properties; it has for example been shown that a-Fe304 is a better lubricant than Fez03. PbO has been found to be a good lubricant for use at high temperatures in a space environment. A striking example of the influence of oxygen in a more industrial sliding process can be seen in metal cutting. If the atmospheric oxygen is removed from an enclosure surrounding the cutting tool and workpiece, the cutting forces are approximately doubled for mild steel (0.1 per cent C), and there is considerable tearing of the surface. Even at a pressure of 0.2 torr, normal cutting performance can be restored at speeds up to 160 ftlmin, with almost complete prevention of build-up of metal at the tool face.In contrast, it requires less energy to cut a magnesium alloy in a vacuum than in air, and a good surface is produced in both conditions. The tentative explanation given is that an oxide film is formed on the free metal surface, away from the cutting tool, and acts as a barrier impeding the emergence of the dislocations associa- ted with the plastic flow of the metal. These consequently pile up beneath the surface, and plastic deformation becomes increasingly difficult.The film thus strengthens the material in the critical region of shearing. Recent studies of metal flow show that there is an analogy between some cutting processes and the ‘ploughing’ action of a simple slider heavily loaded on soft metal. It is consequently suggested that the balance between these competing effects of oxygen may be relevant to some wear problems. Very little work has been done in this area, though there is evidence in cutting that chemically active lubricants such as chlorine compounds can influence the performance by attack both at the tool face and on the free surface. We refer to this again when discussing corrosive wear. We have already seen that oxygen can be important in boundary lubrication by assisting in the formation of metallic soap films.There is evidence that some lubricants become more effective when they contain dissolved oxygen. Both the friction and the wear are reduced, but it must be recognized that oxidation of the lubricant itself is almost always undesirable. Rowe Water vapour is almost as effective as oxygen in reducing the friction of clean sliders in a vacuum chamber. It may however promote corrosive attack and thus increase wear considerably. The formation of acids in condensation products of automobile engines, for example, can be a serious cause of wear. From some recent laboratory sliding tests it appears that the rate of loss of material from a slider increases as the relative humidity of the surrounding atmosphere is increased, up to about 85 per cent, but above this level the wear either remains sensibly constant or decreases.The decrease may be associated with the formation of relatively thick, liquid, films. As we saw earlier 199 hygroscopic impurities can enhance the formation of such condensed adsorb- ate films. Influence of lubricant Jilms on wear and metal transfer The most obvious and important function of a lubricant is to reduce the amount of wear. If a lubricant film of sufficient thickness is present, due for example to hydrodynamic action, the surfaces will be completely separated and the wear reduced to zero. As we have seen, this situation is highly desir- able, but it is not dependent upon chemistry, except insofar as the molecular structure of the lubricant influences its viscous properties.Usually, boundary lubricants will reduce metallic contact between sliders and prevent the growth of adhesive junctions under the influence of plastic flow. Experiments with a radioactive copper slider show that boundary lubricants can reduce the metallic transfer by a factor of 10 or even 100. Other experiments have shown on the contrary that the presence of a boundary lubricant may actually in- crease the wear. Thus, during unlubricated sliding the initial wear rate of a sintered WC 6 per cent Co specimen on stainless steel was found to be 2000 x 10-10 g m-1. In the presence of air at 70 per cent humidity, lubrication with neat oleic acid increased this to 10 500 x 10-10 g m-l.The reason is that the oleic acid corrosively attacked the cobalt phase. To separate the wear contributions, the specimen had been irradiated for a long time to produce both 187W and 6OCo isotopes. In the dry condition the wear rate of the WC was approximately 16 times as great as that of the Co. This is equal to the weight ratio in the sintered material and is attributable to unselective abrasive wear. The lubricant reduced the wear of the carbide phase from 1900 x 10-1O g m-1 to 400 x 10-10 g m-1, but increased the removal of the cobalt phase so that the ratio of wear rates of WC and Co was changed from 16 : 1 to 1 : 25. A cobalt oleate was formed which provided lubrication but was rapidly removed during sliding.A similar experiment with an active chlorinated lubricant showed chemical attack of both the carbide and the cobalt, in- creasing the total wear rate to about 25 000 x 10-lO g m-1. It is therefore important to exercise care in the use of active lubricants and to consider the rate of chemical attack when highly active compounds are used, even if diluted. Frequently the rate of attack may be greatly enhanced by the sliding process itself. Both boundary lubrication and e.p. action depend upon con- trolled corrosion of freshly-exposed surfaces. This feature can however be turned to advantage, for example in the grinding of steel specimens on silicon carbide abrasive papers. If a small quantity of fatty acid is applied to the paper it will be found that the rate of removal is considerably increased and the preparation time correspondingly shortened.Corrosive aspects of wear Corrosive wear even by lubricants can be significant, but if decomposition occurs, with formation for example of dilute HC1, the wear rate may be much further increased, and may even determine the useful life of a component. In general, chloride films are beneficial in providing low friction, but they are much more easily worn away than the hard tenacious films of oxide which RIC Reviews 200 would otherwise form on a metal. When the chloride is removed from a surface during sliding, nascent metal is exposed and the high surface energy promotes rapid chemisorption of further chlorine. The surface layer is then removed as sliding proceeds and a continuous cycle is established.The rate of chemical attack may thus be much greater than it would be under static coizditions with the reaction rate controlled by diffusion through the layer of reaction product. Direct experiments with radioactive WC/Co, as described above, showed that the rate of formation of a chloride film from a highly chlorinated lubricant was increased from 100 x 10-lo g in two minutes to 100 000 x 10-10 g in two minutes, by sliding under moderately severe conditions. The ratio of Co to W in the product was approximately 4 : 1 over this time. Eventually, of course, the relative wear rates must adjust to the composition ratio 6 : 94. The high initial rate of removal is due to chemical attack of the cobalt which is present as a continuous matrix phase, containing some dissolved WC, between the relatively massive grains of tungsten carbide (1 pm diagonal platelets).Detailed studies suggest that the attack proceeds most rapidly in the interface between the grains and the matrix. The grains are thereby loosened and can fairly easily be detached mechanically before they are themselves fully con- verted to chloride. The overall rate of removal of both phases is thus rapid. If however the grains are closely packed, as they will be in a well-prepared commercial material, the reaction between the cobalt phase and the lubricant will be retarded by the narrow channels becoming packed with reaction product. The observed wear rate will then be dominated either by abrasive removal of material or by chemical action on the carbide grains.It is generally recognized that the best wear resistance of these composite bonded carbides is obtained with low Co content and fine WC grains. Many examples of corrosive wear can be found in practice, in addition to the obvious problems arising in marine and other hostile environments. Automobile engines are particularly susceptible to corrosion and corrosive wear of piston rings and cylinder liners. Formerly this could become serious in private cars as a result of condensation and solution of acidic combustion products on the cylinder walls, but the more rapid attainment of working temperature in modern cars has reduced this problem.Nevertheless, striking improvements have been claimed when a sodium-potassium eutectic has been used to dry the circulated oil. Steam and gas turbines present special corrosion problems and numerous seals must operate with low friction and wear under highly corrosive conditions. Many difficulties of this type arise in chemical engineering where rotating seals must be provided for pumps handling sul- phuric acid, superheated water and a host of other materials. The usual solution is to provide accurately flat graphite, ceramic or polymer/glass sealing faces and to allow an infinitesimal leakage of the fluid itself to act as a lubri- cant. Provided that the leakage is no more than can be dispersed by volatiliza- tion at the outer edge of the seal this system proves highly satisfactory over periods of years.The seals will of course wear but this can be allowed for in the design by using annular faces transverse to the axis of rotation and maintaining a positive pressure between them by a spring. We have already seen that corrosive wear involving the production of abrasive oxides can be very damaging. A special example of this process that Rowe 20 1 14 deserves mention is fretting corrosion, which occurs when there is relative motion of small amplitude between two members, often at riveted or bolted steel joints if they have worked slightly loose and are subjected to vibration. It is usually characterized by the removal of minute fragments of steel which are subsequently converted by moist air to a fine red powder of a-Fe203.Fretting corrosion is easily recognized by the accumulation of this powder, commonly known as cocoa, which it closely resembles. The oxide is, however, abrasive and can lead to further wear, since it is most likely to be formed in situations where there is n3 flow of lubricant to wash it away. The best remedy is to ensure that the joint is tight enough to prevent any movement other than an elastic distortion, but if this is impossible the damage can be minimized by applying a solid lubricant and excluding air and moisture. A different form of corrosive wear can often be seen in ball and roller bearings which have failed in service. Such bearings may have been badly fitted or subjected to excessive vibration.Under these conditions the surface of the race will experience repeated impacts by the balls, and even in the pres- ence of a lubricant may become indented. The impressions formed by the balls resemble those produced in a standard Brine11 hardness test and the mechani- cal damage is consequently known as Brinelling. In the presence of corrosive agents, such as carbon dioxide in damp air, small corrosion spots are likely to form, even under less severe conditions, where the protective oxide skin is continually broken by the hammering of the balls. This false Brinelling is a common cause of trouble in machinery containing rolling-element bearings which have been assembled and maintained without proper care. It should not be concluded that all chemical attack under sliding conditions is damaging. Apart from the beneficial surfactant lubricants of the boundary and e.p. type, it is possible to improve wear resistance of surfaces by friction- induced carburizing or nitriding of steels. A hard smooth slider is passed over the surface in the appropriate atmosphere and the high local temperatures generated allow C or N diffusion into the surface. The rapid quenching by thermal conduction to the bulk material can produce very hard and wear- resistant carbon or nitrogen martensite. Care should of course be taken to avoid the production of nitrided debris which can act as an abrasive. Experiments have been performed recently which suggest that there may be a relatively new field for study of reactions with metals undergoing plastic deformation, especially during metal cutting. Very severe plastic strain is produced in the cutting operation. The pile-up of crystallographic dislocations beneath the surface of the deformation zone has been mentioned when discussing the influence of oxygen on wear, earlier in this chapter. This pro- duces a high concentration of elastic energy which can be released by dissolu- tion or reaction of the metal. Such sites have exceptionally high chemical reactivity. In addition, the whole region is subjected to a rapid and large rise in temperature, though this is not important at low speeds. Reactions may occur during cutting with relatively stable lubricants and other fluids. Remark- able results have been obtained with Cc14 while cutting steel or copper. Radiotracers show positive reaction at the interface between a cutting tool and the chip it produces, but also at the free surface at the base of the chip away from the tool. The latter reaction seems to be concentrated at the high-energy RIC Reviews 202 site where the shear zone reaches the surface. It is suggested that by reacting to form a chloride instead of the stronger oxide normally formed, the barrier to emergence of dislocations is reduced. The effective shear strength of the metal is thus reduced and less force is required to cut the metal, even though the lubricant is not applied between the rubbing surfaces. In some processes, where the fluid cannot gain access to the interface, this may be a significant function of a reactive lubricant. A more extensive study has been made of stress corrosion, which occurs during elastic deformation, at much lower stress levels. Many metals, and particularly alloys, are prone to this type of damage, in static structures as well as in sliding components. It is thought that stress corrosion usually starts by slight selective attack by the fluid, which may otherwise have negligible corrosive properties. The local areas of heterogeneity are often grain bound- aries subjected to stress. As the corrosion proceeds, the stress concentration will increase and a crack may develop. This exposes more metal to attack so that the action can continue. The damage is enhanced if a brittle phase is present. Intercrystalline stress-corrosion of this type is found in brasses exposed to moist ammoniacal atmospheres, and in many other instances, but steels are relatively immune under normal service conditions. In some materials such as Mg alloys the crack may propagate along certain crystallographic planes across the grains. Even the residual stresses produced by cold working can cause stress-corrosion of a-brass (70 Cu : 30 Zn) in moist ammonia. This season cracking is not usually a direct tribological problem but is sufficiently common in drawn brass to be the subject of routine testing. In tribology, stress corrosion is probably most important in enhancing wear by surface fatigue. Repeated stress cycles, for example in rolling-element bearings and in gears, are likely to initiate fatigue cracks at or just below the surface, especially in case-hardened materials. In the presence of a mildly corrosive agent, such as an e.p. compound, the crack, once formed, will propagate more rapidly. Surface fatigue is believed to be a common cause of wear damage under conditions which are sufficiently well-controlled to make abrasive wear negligible. The study of chemical attack during metal cutting has led to development of a novel reaction technique. During cutting, large areas of nascent metal are exposed. These can be used for catalytic reactions and also for continuous production of organometallic compounds. For example, synthesis of phenyl- magnesium bromide is rather dangerous when carried out as a conventional Grignard reaction and it can be performed only with small batches. It has been shown that by cutting solid magnesium chips under the surface of a mixture of ethyl ether and phenyl bromide continuous production is possible. The nascent metal reacts spontaneously with the phenyl bromide and the product dissolves in the ether. It is desirable to conduct the process under nitrogen to avoid the presence of oxygen but no other precautions are necessary. Such mechanochemical activation processes appear to be potenti- ally useful for a variety of purposes. Rowe 203 BIBLIOGRAPHY G. Barr, Monograph on Viscometry. Oxford: OUP, 1931. F. T. Barwell, Lubrication of Bearings. London : Butterworths, 1956. C. J. Boner, Manufacture and Application of Lubricating Greases. New York: Reinhold, 1954. F. P. Bowden and D. Tabor, The Friction and Lubrication of Solids. Part I. 1950, Part 11, 1964. London: OUP. E. R. Braithwaite, Solid Lubricants and Surfaces. Oxford : Pergamon, 1964. E. R. Braithwaite (Ed.), Lubrication and Lubricants. Amsterdam : Elsevier, 1967. A. Cameron (Ed.), Principles of Lubrication. London : Longmans, 1966. D. Dowson and G . R. Higginson, Elastohydrodynamic Lubrication. Oxford : Pergamon, 1966. c. W. Georgi, Motor Oils and Engine Lubrication. New York: Reinhold, 1950. N. S. Grassman and J. W. Powell, Gas Lubrication Bearings. London : Butterworths, 1964. R. C. Gunderson and A. W. Hart, Synthetic Lubricants. New York: Reinhold, 1962. W. B. Hardy, Collected Works. Cambridge: CUP, 1936. R. E. Hatton, Introduction to Hydraulic Fluids. London: Chapman and Hall, 1962. R. J. Hengstebuk, Petroleum Processing Principles and Applications. New York : McGraw- Hill, 1959. M. D. Hersey and R. F. Hopkins, Viscosity of Lubricants under Pressure. New York: Rept Am. SOC. mech. Engrs, 1954. T. P. Hilditch, Chemical Constitution of Natural Fats. London: Chapman and Hall, 1940. A. Jobling and J. E. Roberts, Rheology, Theory and Applications. New York: Academic Press, 1957. W. D. Jones, Fundamental Principles of Powder Metallurgy. London: Arnold, 1960. E. N. Klemgard, Lubricating Greases-their Manufacture and Use. New York : Reinhold, 1937. 1. V. Kragelskii, Friction and Wear (trans. from Russian). London: Butterworths, 1965. W. I. Lichtman, P. A. Rehbinder and G. W. Karpenko, The Influence of Surface-Actjve Materials upon the Deformation of Metals. Part I, London: HMSO 1961 ; Part 11, Berlin: Akademie-Verlag. (in German), 1964. See also A. R. C. Westwood, D. L. Goldheim and R. G. Lye, Phil. Mag. 1968, 17, 951. c. Lipson, Wear Considerations in Design. London: Prentice Hall, 1966. A. G. M. Michell, Lubrication. London: Blackie, 1950. H. Mykura, Solid Surfaces and Interfaces. London: Routledge and Kegan Paul, 1966. A. Palmgren, Ball and Roller Bearing Engineering. Philadelphia: SKF Industries, 1946. L. Pauling, Nature of the Chemical Bond. Ithaca: Cornell University Press, 1948. E. Rabinowicz, Friction and Wear of Materials. New York: John Wiley, 1965. G. W. Rowe, Introduction to the Principles of Metalworking. London: Arnold, 1965. M. C. Shaw, Metal Cutting Princ@les, third edn. Cambridge, Mass: MIT Press, 1954. L. R. Underwood, The Rolling of Metals. London: Chapman & Hall, 1952. K. van Ness and H. A. van Westen, Aspects of the Constitution of Mineral Oils. Amsterdam: Elsevier, 1951. C . G. Verver and van der Have, Petroleum and its Products, London: Pitman, 1957. ASTM Symposium on Properties of Surfaces. Philadelphia : ASTM, 1962. ASTM Stand., Part 18, Petroleum Products. Philadelphia, 1964. Inst. Petrol. Standards for Petroleum and its Products (Annual). London: Institute of Petroleum. Statistical Review o j the World Oil Industry, London: B.P, 1964. Various: The Control of Quality in Working Operations, Monograph No. 16. London: Institute of Metals, 1954. Friction, Wear and Lubrication-Terms and Dejinitions, Paris : OECD, 1968. RIG' Reviews 204

ISSN:0035-8940    

DOI:10.1039/RR9680100135

出版商:RSC    

年代:1968

数据来源: RSC

摘要:

CHEMICAL EDUCATION : PROBLEMS OF INNOVATION H. F. Halliwell, B.Sc., F.R.I.C. School of Chemical Sciences, University of East Anglia, Norwich, Norfolk, NOR 85 C A review usually implies the question: where have we got to? Science educa- tion, in fact education in general and certainly chemical education is in a state of reorganization; there has been an upsurge of detailed exhortation and advice and a rush to follow new banners. The more pertinent question is, therefore, where are we going? An even still more pertinent question is on what basis and on what principles can we judge the wisdom or folly of pro- posed targets. At this stage of development, therefore, it is to be expected that a review will raise more questions than it answers. Chemical education has still many of the characteristics of a cottage industry.In fact, the very phrase ‘chemical education’ suggests a parochial attitude. Education through chemistry, to change the emphasis, is part of education through science, which is but one aspect of the overall education of an individual. With the establish- ment in many countries, and the development in many others, of a policy of education for all children, the fundamental question is: what are we trying to do both with, and for the oncoming generation? The words w+th and for indicate the two areas of responsibility: to the well-being of the community, and to the well-being of an individual. This article, therefore, attempts to indicate a general pattern of innovation in chemical education so that in administration, argument, action or selection, there is a framework for judgement.Much that will be said here is not new. What is recent, however, is the markedly increasing awareness of the urgency to think, before we go much further, along some such lines as those which follow. ANALYSIS OF INNOVATION Thinking about the processes involved has, over the last two decades, crystal- lized into what might be called the interacting four A’s of curriculum reform (Fig. 1). Seen in this perspective any curriculum reform (and therefore that of chemical education at any level) is a continuous and not a once-for-all process. It will be a continuous process, partly because appropriate assessment will indicate whether the suggested lines of action are effective or as effective as alternative lines, partly because changes in the political outlook of a country may require changes in attitudes on the part of its population and, therefore, a change in the specification of aims, partly because the machinery of commu- nication develops new techniques, and partly because more is known about the process of learning.If these factors are themselves slow to alter, as on the whole they have been up to the middle of this century, then the rate of desir- able adjustment can be so slow as to make the establishment of its machinery of apparently little importance. If, however, any or all of these factors change ,yctlli well 205 planning and i rn pl erne n tat ion of specification of use of valid and reliable AIMS- in the light of feedback -ASSESSMENT Fig.I. The processes of curriculum reform rapidly as they have been tending to do in the last two decades, not to set up machinery for considered adjustment might lead to ill-advised expediency. It is for this reason that developing countries, which were confronted with the need to enlarge the scope of their science education, have been recently urged to establish a permanent, but not static, steering committee.1 However, a recent directory of reform groups concerned with science education lists 30 countries in which revision of science curricula is taking place2 and the sub-set of developing countries is seen not to be identical with the sub-set of non- technologically advanced countries.Indeed, as might be expected, it is the countries who have had the longer period for reflection and the greater resources for research that have been actively involved in analysis and implementation of reform. The establishment of such advisory bodies as the British Committee on Chemical Education jointly by the Royal Society and the Royal Institute of Chemistry in England, the Committee on the Teaching of Chemistry in the Scottish Universities at the Department of Chemistry, University of Dundee, the Advisory Council on Chemical Education by the American Chemical Society, an Education Committee by IUPAC, another by the Chemical Society of Japan and yet another under the Directorate for Scientific Affairs by OECD is an example of action in the fields of chemical education itself following more widespread awareness of the need for the pooling and consideration of ideas and intentions.Articles in such journals as School Science Review in the UK or the Journal of Chemical Education in the USA show that advice and exhortation to look in wider perspective and more critically at what is current practice are by no means recent phenomena. What is recent is the proliferation of nostrums, the greater and wider influence of persuasion, the increased opportunity for exchange of enthusiasms and apprehensions, and a more widely felt sense of 206 RIC Reviews urgency. What is wanted in a rapidly changing situation like this is not only an analysis of the various aspects of curriculum reform so that strategy and tactics can be planned with more confidence in their outcome, but also the clarification of the basis and principles of judgement in these areas, so that wisdom has at least the edge on folly in a matter which affects the future well- being of a community.The simple analysis of reform into four areas of interacting activity referred to above has already shown promise for meeting these needs. We follow this analysis further. Aims Certain broad patterns of change can be detected in the newer proposals. One such pattern, a change of perspective common at least implicitly to many schemes, is a swing from the transmission of a static past with a stress on acquisition of knowledge, to a preparation for a changing future with a stress on informed adaptability. Both these educational perspectives require a corpus of information: in the former it is extensive and tends to be the main if not exclusive goal for the majority; in the latter it is less extensive and is sub- servient to the main objective, namely an ability to scan such a corpus imaginatively and critically in order to cope with that which is novel.Another change, in a way an important outcome of the former, is that chemistry, in its own right, along with physics and biology is an important vehicle of education for that great majority who are not going to be chemists, and not an irrelevant even if interesting peep at the prerogative of back-room boys. This change, new perhaps only in its more vigorous implementation, has had two results. In a way it has inverted the structure of purpose in a curri- culum.In most established systems of education, that of the elite was agreed and the not-so-elite were confronted with a watered-down version. This seems to be changing to a scheme which is designed to serve a dual purpose, i.e. to be suitable for the majority (specialist and non-specialist alike) and at the same time to form an adequate basis for that important minority whose particular interests will be catered for at a more advanced stage. A second result is that it has raised sharply the question of the pertinence of much of the detail of a chemistry course. Of what value is the knowledge of the preparation of a pure dry sulphur dioxide specimen to a boy who becomes a bus driver in Bolton or a shopkeeper in Singapore or a business man in Los Angeles ? Questioning the pertinence does not deny that an acceptable positive answer exists; it only stresses that in the process of reappraisal pertinence to the individual and to the community are important criteria of selection and rejection, which must be satisfied and-in these days of assessment by pupils and students-seen to be satisfied.Yet another change is based on a reappraisal of what is suitable for a developing and immature mind. Pupils of 13 and 14 years are increasingly being confronted with ideas and facts once regarded as appropriate to under- graduate study, and chemistry is not alone in this : modern mathematics offers even more startling examples.In chemistry electronic structure, enthalpy and free energy changes, entropy, kinetics and mechanisms of reactions, pH notation and radioactivity are increasingly appearing in studies intended for Halliwell 207 15 or 16 year old pupils in many parts of the world. Bruner from his studies of children learning mathematics points to the feasibility of such programmes when he writes : ‘We begin with the hypothesis that any subject can be taught effectively in some intellectually honest form to any child at any stage of development. I t is a bold hypothesis and an essential one in thinking about the nature of a curriculum. No evidence exists to contradict it; considerable evidence is being amassed that supports it’.3 There is undoubtedly a growing opinion that children‘s ability is under-rated.It is likewise true that the pupils are living in an age when commonplace and uninformed talk involves ideas not thought of in their teachers’ youth-a fact that was startlingly brought home to the reviewer on hearing the following skipping rhyme being chanted by eight year old girls in a playground of a primary school in Newcastle-under- Lyme: ‘Ashes to ashes, dust to dust If the atoms don’t get us the hydrogen must.’ The skill to translate the sophisticated into a form suitable for the immature has been shown by individuals for a long time. The Royal Institution has been one notable centre for such. There has been a brilliant line from Faraday’s History of a Candle through Tyndall’s Heat; a mode of motion and, in recent years, the many examples of Bragg’s skill to Porter’s BBC talks, ‘The Laws of Disorder : an introduction to chemical change and thermodynamics’.* There have also been innumerable, unheard of, but personally remembered, teachers all over the world who have this skill.I t is noticeable, however, that what has been done for some time is acceptable : it is using this skill in new areas that is sometimes suspect. Joule’s work raised considerable doubts and discussion in the Royal Society in mid-19th century yet it is now an area of study accepted as suitable for all moderately bright 15 year old pupils, but the introduction of entropy or wave-mechanics, relativity or indeterminancy, polarity of mole- cules or i.r.spectroscopy engenders resistance. The question still remains : What is educationally desirable? What are we trying to do with and for the oncoming generation? Probably the most important pattern of change to be seen in the new proposals is the growing insistence on specification of aims and objectives in terms of final pupil behaviour, and the close link of these objectives with the procedure of assessment. The focus of attention is on what we want pupils to be able to do-intellectually and manipulatively-and this focusing arises from the claim that only by what people do (through speaking, writing or performing operations) can we judge whether our aims have been achieved. This approach to the specification of aims is not new: it is over 30 years since it was suggested by Tyler.5 What is new, again, is its widespread adoption and the impact it has had on curriculum reform. Its adoption is encouraged by the claim that it offers a basis for the analysis and classification of many of our problems.On the other hand it has met with opposition chiefly on the grounds that it leads to emphasis on trivialities, that it leads to a dehumanizing use of judgement, that it is unacceptable to plan how pupils should behave, and that the important goals of education cannot be defined- much less measured. In these early days no doubt this argument will continue, but also there is no doubt that the principle of stating educational objectives in operational terms, elaborated and vitalized by the outcome of the delibera- RIC Reviews 208 tions sparked off by the American Psychological Association in Boston in 1948 and disseminated by Bloom and his colleagues,6 has given guidance to many, including chief examiners of public examinations.728 This guidance has come from clearer statements on the one hand about what the inclusion of chemistry in a curriculum is supposed to do, and on the other about what an examination is supposed to find out. It is reasonable to suggest that these two should at least be consistent if not fairly closely matched. When this is accomplished there is then available a basis for selection and rejection of proposed class-room and lecture-room action. Aims and assessment are closely linked.Thus it is fairly frequently said that we want pupils to understand chemistry, not just to memorize what they have been told. But as Bloom points out, it is not until we know what it is that a boy ‘who understands’ can do, which a boy who does not understand cannot, or what it is that the former does not do, which the latter does, that we know how to detect ‘understanding’ and therefore have a clear idea of what we are trying to achieve. If a teacher says that his aim is to show pupils that organic chemis- try can be seen as a coherent whole by looking at it from the point of view of mechanism then it is clear what he intends to do-to explain a point of view: the pupils do not come into it. But if as a result of his putting forward a point of view he wants the pupils to be able to do something they could not do before, it would be well if this intended operational result were clearly stated.Both efficacy in terms of communication and worthiness in terms of educa- tional policy can then be scrutinized. Let us look at what this principle of reform has begun to do and can do in chemical education. We shall then be in a better position to accept or reject it, although we must also bear in mind that its application is still in its infancy. The stating of a general principle is one matter: its consistent and wise use is quite another. Politicians, administrators and educationalists, those responsible for educational policies of a country or at least of influencing them, have on the whole stated their aims in general terms.These will have to be translated into behavioural terms appropriate for a stated group in a stated social environ- ment, and the translation checked for its validity. This i s a difficult exercise. As yet it has only been seriously attempted in a few cases, because much attention has been given by innovators to the reappraisal of the structure of their subject (the medium of education) rather than to critical analysis of its impact on the younger generation (the purpose of education). There is an important distinction, on the one hand, between reappraisals of the structure, emphases and content of a course and the analyses of its sociological impact on the other. The former try to bring the subject into line with late 20th century thinking but fail to answer, or only peripherally and implicitly deal with, the question raised by Nyholm in the Marchon lecture: ‘Education in science-for whom and for what p u r p ~ s e ? ’ ~ They are more profitably considered under the heading of action.The aims of science teaching in some countries have been stated in near behavioural terms. The spectrum is broad as we can see from a few examples. Writing in 1959 of science in Soviet schools, Yantsov says, ‘It is the function of the secondary school to train the younger generation for active and con- Halliwell 209 scious participation in the social life of the country and in the production of material goods.’lO At the Conference on the Teaching of Science in Schools, organized by the Commonwealth Education Liaison Committee in 1963, the Minister of Agriculture and Natural Resources, Western Nigeria, said ; ‘Science should be taught in developing countries in such a way as to make it serve as an instrument for removing those evils (over-population, poverty, malnutrition and chronic ill-health)’, and went on to say that science teaching ‘should aim at stimulating interest in agricultural pursuits, inculcating the dignity of manual labour and familiarizing young people with the technical equipment employed in agriculture’.ll In Ceylon, the National Commission on Education in 1961 stated that the intention of the new programmes in science teaching was to make the population scientifically literate.At home we find in 1960 a major lead from the Ministry of Education saying: ‘There is little need to plead for more attention to science; everything around us demands it. . . . (If scientific knowledge is limited in the main to specialists) we risk becoming a well-fed rabble ordered about by experts, subjects of a new despotism. We cannot do without specialists; if we are not to become their slaves, we must see to it that scientific knowledge and the ability to reason scientifically are widespread. . . . Our future community will need not only more specialists able to maintain the machines of the automatic age, but also more citizens capable of imaginative and creative thinking within the context of science.’12 The professional association of most of those teaching science in the UK has also given a lead: ‘The effects of science on human life and thought have become so great, and are potentially so much greater, that those who have no understanding of them, and of the science which has produced them, cannot be considered properly educated or truly cultured and therefore are unable to participate fully in the life of their time.Present “scientific illiteracy” is, in part, due to a lack of factual knowledge, but is much more the result of a lack of under- standing of the basic nature and aims of science. . . . . . . Science should be recognized-and taught-as a major human activity which explores the realm of human experience, maps it methodically but also imaginatively, and by disciplined speculation, creates a coherent system of knowledge.’l3 The analysis advocated here is already indicating that there are three different areas in which to exercise one’s judgement.Do we agree with the general intentions ? This is an area of political-sociological value-judgement. Are the actions planned to achieve these aims ? This is a matter of professional pedagogic judgement. Are the actions taken efficacious? This is no longer a matter of subjective opinion. Progress in devising valid and reliable instru- ments of assessment is increasingly making the answer a matter of objective testing. We are not concerned here with agreement with or disagreement about political-sociological aims, but with the possibility of a more rational pro- cedure of reform through application of the principle of stating these aims RIC Reviews 210 operationally in terms of final pupil-student behaviour. Let us take for brief dissection a proposal where the principle was applied.14 (1) The starting point was the two statements given in the preceding paragraph.These were first briefly restated from a learner’s angle: ‘pupils should gain an understanding that lasts throughout their lives of what it means to approach a problem scientifically. They should be taught to be aware of what scientists are doing and can do’. (2) What are the characteristics of this aim ? It seemed that there were four : the learning should have a lasting effect; it should develop judgement as to whether a problem is a scientific one or not; it should develop the know-how, personal understanding and judgement necessary for tackling a problem scientifically; it should develop an awareness that science, as Richardson pointed is both a process and a product and one product is power over the environment-power to do good or to do evil.(3) It is still necessary to translate these characteristics further into pupil behaviour. Judgement involves an ability to sort phenomena into examples and non-examples of what is being assessed, an ability in the face of a novel situation to scan quickly an acquired corpus of information for that which is pertinent, and the imagination to suggest a plan of action. Such considerations finally led to the following requirements : (i) after finishing formal education a pupil should continue to show the abilities listed below ; (ii) during the course a pupil should show satisfaction at being involved; (iii) at the end of the course a pupil should show an ability to discriminate between problems to which a scientific approach is appropriate and those to which it is not, recall information and experience, devise schemes for solving practical problems, use and classify given information, apply previous knowledge to new situations, interpret information with evidence of judgement and assessment, report and comment selectively and effectively, relate the subject to other school subjects and to life in society.Clearly, objective ( i ) can only be reliably checked by a careful follow-up.Objective (ii) can only be detected, subjectively at present, by the teacher concerned. The others are all abilities which can now be assessed with increasing validity and reliability. In what way does this brief dissection indicate the possibility of a more rational procedure of reform? It has yet to be shown whether what has been said helps to clarify what should be done. None the less, by temporarily isolating this part of the problem it has made a more manageable argument possible. It is a helpful step to follow Kerrl6 and distinguish the aims from the objectives, using the former word for the general target and the latter for the statement of operational criteria. We can then ask the authors of the aims whether the translation into objectives is acceptable or not (we might even suggest that both aims and objectives, or at least some indication of the latter, are promulgated at the same time).If the answer is no, then there is doubtful profit in discussion until the authors have performed an analysis which they find acceptable. It is of great importance to note that in teaching, as in all human interaction, there are many aspects too subtle to be starkly stated in words. On the whole these are in the realm of action rather than intention. Halliwell 21 1 Teaching is an art. Nevertheless, there are also areas of education, seen as a great national industry, where logical thinking and purposive planning should be predominant. Perhaps ability to discriminate is pertinent to educationalists as well as to pupils.We leave this question of aims to turn to the action required to achieve them. Here the play becomes very open and varied, involves many more participants, but is still not without guiding principles now that objectives have been stated. ACTION This area of reform involves not only pupils and teachers, but also admini- strators, professional chemists, the designers, manufacturers and suppliers of equipment and audio-visual aids, architects, authors and publishing firms, those responsible for the training of future teachers, examiners, and those who have to accept the products of an educational system. Vast though the area is it can be broken down into sub-areas and one such analysis sees four important subjects raised : (i) the merits of a concurrent or subsequent structure in science education ; (ii) the place of chemistry in science education; (iii) adequate allowance for local and national variations of needs; (iv) selection and rejection of suggested experience for pupils and students.The first perhaps seems a matter of little concern to us in the UK: our system of concurrency of treatment of the sciences seems well established- even if the coverage is still often limited. Yet it is not the only scheme and that common in the high schools of the USA tends to concentrate work in a specific science into a single year. One can detect a mutual influence now altering these two schemes. It seems to be brought about by the breakdown of compart- mentalism in the sciences as well as by the realization of virtues in the other party’s programme.Whether a final synthesis takes place whereby there is a subsequent structure consisting of a fairly long period of integrated science followed by a shorter period of separated sciences, which is in turn followed by another period of integrated science followed by separated studies, is a matter of extrapolation with all its advantages and dangers. The growth of science in the primary sch001,~7J* the resurgence of general ~ c i e n c e * ~ J ~ and the emergence of combined physical sciences for the 16-18 age groupz1 seem, however, to be pointing that way. The question, however, is a sharper one for developing countries since they are likely to have previously come under the influence of one scheme or the other and may be looking at the possibility of a change, being at the same time saddled with texts written specifically for one of the schemes.Whatever plan is adopted two principles must be borne in mind. The first is that the pupil or student often meets the same ideas and the same phenomena in his separated subjects but often has to do something quite different with them. The unit of mass is the kilogram-the student of mech- anics is concerned with the movement of bodies, and whether the body consists of iron or copper is usually irrelevant in deciding what he has to do with his XIC Reviews 212 information. To the shopkeeper or to a person making up a solution the unit of mass is the kilogram (or soon will be for the former) because he needs to achieve reproducibility in repetition.For the chemist the amount of a substance is expressed as a mole and weighing is used to count out comparable numbers of particles. A chemist is often much closer in purpose to a bank cashier who by weighing compares the number of pennies in a bag with the number of pennies in 2 standard bag: the latent heat of evaporation of materials is of more significance to him in kcallmole than in kcallgram. Sometimes the difference in purpose in different sciences leads to a reciprocal perspective. In traditional physics a pupil pays much attention to mass per unit volume: in chemistry he is more interested in the volume per mole because the roominess of stuff is significant from the point of view of structure and he wants to compare volumes.Again in physics he is concerned with electrical resistance, in chemistry with electrical conductance. If a pupil is to use ideas with discrimination it would seem reasonable for him to use them in different contexts. This of course is looking-from the pupil’s operational point of view-at the frequent exhortation to see that certain concepts, i.e. structure, energy, electric charge, rates of change and conservation, are common to many sciences. The move towards a closer link between science subjects comes not only from the way the sciences themselves are changing. The second principle that must be borne in mind (whether it is a concurrent or a subsequent structure of science teaching which is chosen) is that a pupil’s ability to use ideas, his mode of using them, and his rate of operational assimilation of new patterns, vary with the length and the nature of his experience.There is nothing new there, unless it is the stress on using rather than knowing, but a critical application of the principle is none the less a safeguard in the otherwise very desirable exchange of schemes between countries. Nothing new, but it sometimes shows an unexpected twist especially in a pupil’s use of words. Words summarize and help to sharpen experience. ‘Element’, or ‘equilibrium’, for example, will be used differently by 13 or 14 year olds and 18 year olds and this fact will influence our attitude to ‘defini- tions’.An eight year old will know whether to apply the label ‘plant’ or the label ‘animal’ to a dog or a geranium, but could not ‘define’ a plant. An 18 year old may feel confident of achieving this. An undergraduate or research worker would find himself in a similar position to the eight year old, but for different reasons. The recent disquiet about whether we know when to use the word ‘dead’ underlines this fact that the problem of categories and definitions is very much with us. This growing up through confidence in certainty to a confidence in the face of uncertainty is a growth pattern about which we need to know more. Nevertheless, attention to operational objectives helps us. A pupil first using the words ‘rate of a reaction’ will not ‘understand’ them as we do, in the sense that he cannot do as much with the idea as we after many more years of experience can.To expect otherwise would not be reasonable. The question is: does he need to use the idea within the context of his immediate problem, and will he continue to use it in a planned but more and more varied and detailed way. Understanding develops by use. Ours did. Sometimes it is forgotten that theirs must. What has been said up to now is common to education through any science. Halliwell 213 The second question raised in this analysis touches us more specifically: the place of chemistry in science education. Nyholm’s question, for whom and for what purpose, must be asked both in the context of chemistry and in the context of the ability of the pupils.For the UK, educational statistics show that for every 100 children likely to be learning science at the age of 14, only 15 will take 0-level chemistry one or two years later, only six will go on to take A-level chemistry and only one goes on with formal learning of chemistry after leaving school. If chemistry in school is merely undertaken to provide a means of selection to go on to chemistry in higher education, it seems a waste of youthful endeavour for a large majority. If it is designed to encourage more of the high-ability group to go on with chemistry, present trends suggest it is not successful. In fact there is no evidence that chemists either at the technician or research level could not be trained equally effectively if they started almost from scratch after they left school-as indeed happens in architecture, engineering and law.That chemistry is a traditional science subject, that it is the pride and joy of many teachers, these alone will not justify its place in science education in a changing and critical educational world. Already a sharp difference is showing between the attention given to and the time spent on chemistry in schemes for children of average and below-average ability and schemes for the average and above-average. This is seen in a UK context by comparing the small fraction of time given to chemistry in the Scheme for Young School Leaverslg and the time suggested for those likely to stay 0 1 1 .~ ~ 9 ~ ~ Two problems arise: what is the coverage appropriate to these two groups; what is the justification of the great increase in chemistry for the second group over that for the first? Plans for coverage are not left so open to personal exhortation if it is realized that the crux of the question is not the extent of coverage, but the appropriate- ness of the selected areas for achievement of operational aims. Both extremes of treatment stress that the areas selected are intended to bring about an awareness of the ‘alchemy of stuff’ (an awareness presumably detected by the ability to change selected raw materials into desired products), and an aware- ness of man’s technological mastery over raw materials (presumably detected by a competence to talk, with some practical know-how, about the importance of selected areas of chemical industry-including food production-to society).The difference between the extremes lies in the introduction of the more able to the atom-molecule-ion theory as a unifying ‘explanatory’ concept. The educational justification for inclusion of chemistry is thus twofold : for all to differing degrees it can establish, through personal commitment with the transformation of material, a personal involvement with a major basis of society in a way not possible in any other field of experience; to the more capable it uniquely offers an area for the development of critical, integrating speculation about the observable in terms of the unobservable. Physics, chem- istry and biology form a complimentary but ill-defined and merging trio.The first, at school level, is predominantly concerned with patterns of general behaviour, the second with patterns of specific behaviour, the third with patterns of individual and fluctuating behaviour. Chemistry as a school subject has been attacked on the grounds that it calls for no judgement, that it is merely a catalogue of irrelevant facts and fancies of minimal or no interest to all but a RIC Reviews 214 strange minority. It can be so-even the stimulating explorations of one generation can be inert ideas for another as Whitehead pointed out-it can be so, but it need not be so. If attention is focused on the objective aims in terms of skills and attitudes to be acquired by pupils we have criteria for selection, and perhaps a basis for wiser judgement on our part.The third question raised earlier was that of adequate provision for varia- tion of treatment within a community. This is not just a question of difference of treatment of the subject for the very capable, the capable and the not so capable, but of uniformity of treatment for a given ability range. Absence of uniformity is noticeable in the extreme age groups-the under 12 year old pupils and the undergraduates. Both are free from the restraints of a centralized external examination geared to uniformity of coverage. This is not so at the present time with the able 15-18 age group-particularly in European countries and those influenced by them. The effects of examinations on chemistry curricula up to the level of university entrance has been the theme of a recent report by the IUPAC Committee on the Teaching of Chemistry22 although it does not take an IUPAC report to convince many teachers in the UK of the restraining influence of an examination syllabus.The question is not whether these syllabuses have or have not a restraining influence, but whether the restraining influence is acceptable or not. How is that to be decided? An examination is an instrument for detection and measurement; if it is shown that it is a valid and a reliable instrument for assessing the extent to which approved aims and objectives have been achieved, its restraint is likely to be acceptable. A major change in perspective in the last decade has been from a pattern in which action is geared to examinations (traditionally respectable, but never- theless often of unstated purpose and usually of unproven validity and reliability) to a pattern in which both action and examinations are geared to the same set of objectives.We shall return to the question of assessment. The point here is that if the objectives stipulate a framework of operational objectives within which there is flexibility of coverage and the exurninations are geared to this, then uniformity or non-uniformity of coverage is very much more a matter of tactical decision for individual teachers. The last question raised in this analysis of action was that of selection and rejection of suggested experience for pupils.Change and tactical freedom are bound to bring discomfort to the established and uncertain-and that means all of us in some respect-and although the emergence of a guiding procedure has been indicated yet there are factors and pressure groups which influence action in the area where there is choice: the class-room. These factors influencing immediate action need to be looked at. The present climate of opinion Although it is limited in its sampling, the Schools Council report23 on science for the more able and older pupil summarizes a great deal of contemporary professional thought. In particular the paper by the Master of Marlborough on four major patterns which are emerging and that by V. J. Long on the problem of breadth versus depth of study are pertinent to our immediate concern.But there are opinions other than those of educationalists: those of Halliwell 21 5 the consumers-industry, commerce and universities-and those of the products (‘consumed’ perhaps would be an unhappy choice of word)-the students. The former are asking more and more for boys and girls who have a lively critical mind and can cope and look forward to coping with problems in an industrial setting. There is much talk about improving ‘the poor image of industry’ which seems to be an outcome of our present system. The opinion of students at the moment seems more focused on unacceptable administration and the lack of intellectual stimulation in learning than on questions of pur- pose, but if one of the aims of education is to develop a lively critical in- dependent mind, then the need to have to cope with lively critical independent action is to be expected.It was early pointed out in the USA that one of the results of new schemes of chemical education would be dissatisfaction with what might be met at a later stage. The pattern which is emerging is a weaning from a ‘we-they’ to much more of an ‘us’ situation, and this is unlikely to be possible unless the freedom in tactical decisions referred to above is a part of the action to achieve the aims. The Chemical Bond Approach proposalsz4 in the USA and the Nuffield Chemistry approach in the UK14 have begun to make this freedom possible. This will be so for any project that suggests an approach to be used rather than a rigid course to be followed and suggests optional ways of rounding off the work.The need for examinations which have the same objectives is obvious. Availability and suitability of resources The more affluent countries are short of manpower but comparatively rich in educational material resources. The converse often holds for the less affluent countries. However, both groups are in the process of introducing changes and whether resources, human or material, are available or planned, the common problem is the appropriateness and quality as well as availability. The USA, the UK to a lesser extent, and to a still lesser extent other countries with a well established educational system have spent considerable sums of money bringing existing teachers up-to-date and giving them a feel for new approaches as well as swelling their ranks.The supply of appropriate and necessary equipment is a different matter: some countries are bedevilled by lack of money, others ensnared by an abundance of gadgetry, and yet others bogged down by an encumbrance of outmoded-even archaic-apparatus in schools which gives a restricting, although erroneous, impression of being well equipped. Some guidance as to what is wanted will be got from consideration of the purpose of practical work. A valuable account of an inquiry into the nature and purpose of such work in school science teaching in England and Wales was published in 1964.z5 However, this was general in nature, and for the efficient supply of equipment the precise objectives of each particular piece of practical work need to be recognized.In this area much thought, investiga- tion and planning are still needed, but it should be noted that some countries have already set up centralized units to do this. It is of interest to note that Ceylon has set up a pilot production-unit for apparatus as part of a teacher training and research centre for science teaching. A reappraisal of the use of national resources is emerging. The focus is beginning to change from fitting a pupil into a scheme, from looking round RIC Reviews 216 hopefully to see what is available, to formulating what effect practical work is supposed to have on the pupil in his particular socio-economic context and designing equipment, whether simple or complicated, to bring about this effect.It is equally important that the place in which the work is done be designed for effective working out of the objectives. There are schools and universities in many parts of the world where the laboratories show little change of plan from those set up in German universities in the middle of the last century. Design of an educational work room must be controlled by purpose as much as is the design of an industrial workshop or a domestic kitchen-and purposes are changing at the same time as materials and equip- ment are changing. Attention is drawn to the recent publication of suggestions and plans for laboratories in the UK.26 Other injuences on education But there are three other factors which are having a particularly powerful influence on class-room and lecture-room activity : the change of perspectives and emphases in late 20th century chemistry itself; the diversification and development of aids to learning based on the considerable amount of research carried out on the learning process particularly over the last two decades; and the changes in objectives and techniques and, therefore, in the demands of examinations. Much has been said about the explosion of knowledge in science over the last quarter of a century and even more about the dismal outlook as the pro- cess goes on.Its significance for chemical education is that it forces the issue of selection. Less perhaps has been said about the changes in perspectives and emphases in science, and it is these changes which are having a marked influence on the necessary selection and the adopted approach.The change from saying, ‘It is known that. . .’ to ‘It looks as i f . . .’ when dealing with human interpretations, the change from an intellectually arrogant certainty about de rerum natura to an attitude of operational expediency in the use of man-made and changing conceptual models, the change from an acceptance of hard categorization to an awareness of merging sub-sets of phenomena, the change of focus from exposition of truth to acquisition of mastery, these changes of perspective are becoming more and more noticeable in students’ and pupils’ texts.Campbell’s new text for college students,27 the proposals of the Chemical Bond Approach Project24 and of the Nuffield Proposals in Chemistry,14 Kauzmann’sZ8 treatment of the kinetic theory of gases for under- graduates are poles apart from corresponding texts written even three or four decades ago. If science is seen as a human activity born of curiosity, developing through the use of critical imagination, corrigible and asymptotically tentative in its achievements, but amorally powerful in its application, then selection of areas for pupils’ experience must be geared in these directions. A pattern of treatment which seems to be emerging is that education through chemistry involves exercise in answering five questions : What material changes are going on and why should I bother with them? Do they fit into a wider pattern and does it help to think about them in terms of atoms, molecules, ions and giant structures? HaIliweIl 217 15 How fast is the change and can it be adjusted? How far does the change go and can it be controlled? What energy changes also take place and do they throw any light on what is happening ? Pertinence and the organization of experience for judgement and discrimination are taking the place of ‘preparation and properties’.Whereas the latter often was second-hand book knowledge, the former demands much more personal experience of materials. Paradoxically, an attempt has been made29 to con- struct a chemistry syllabus without mentioning a specific chemical : it is written in terms of intellectual and manipulative skill which should be acquired by the end of the course.However, a pupil cannot know or show that he has these skills unless he has exercised them on materials and on ideas-his own and other people’s. A recent review of the way man uses different types of intellec- tual models for different purposes in the natural sciences, theology, psychology and the social sciences is particularly commended to edu~ators.3~ The May 1968 issue of Education in Chemistry was concerned with audio- visual aids. A great deal has been published in the last five years on this topic, varying from unhelpful gadgetry to fundamental insight. As a learning influence we have had personal experience as long as man has existed, the teacher since prehistoric times, the printed book for 400 years, the film for half a century, the audio-video tape for a comparatively brief period.It is not surprising that we have much to do before we know how or when to use them all to an effective end. That films can have a considerable impact, that they can communicate an idea and stimulate thinking we know-‘Entropy’ made by Professor Porter and Professor Campbell’s ‘Crystals and their structures’ are two outstanding examples-but we still need to know how best to follow them up so that the ideas are usable and become used. The waste of time and loss of power in using a lecture for the transmission of factual material is acknowledged. The possibility of using video tapes for individual learning is known.But how to weave practical work, personal communication and indivi- dual use of learning aids into a profitable programmed experience for the individuals of a large class is still comparatively unexplored. One outcome is certain: more time spent in new techniques means less time spent on tradi- tional ones and coverage of subject is bound to diminish. ASSESSMENT There are three areas where assessment is necessary and they involve the use of different techniques. On a national scale, society as well as those responsible for policy will want to know whether the overall aims of an educational policy are being achieved and whether, and what kind of, adjustments are necessary. In the area of science teaching the existence of the Dainton Report31 is evi- dence that all is not well: from its content it is also clear that we are largely ignorant of the factors causing the malaise, and we have to try to bring about an adjustment while working in the dark.During the development of new schemes, critical assessment is needed to judge the suitability of a proposed treatment for a given age group or the relative advisability of different treatments. If the proposal to introduce say RIC Reviews 218 three-dimensional structure, or the difference between changes in enthalpy and free energy, or the use of partial molar quantities, fails with a group of speci- fied age and ability, the reasons can be diverse. It may be that the topic was badly positioned in a wider programme, or that the actual programming of the topic was at fault, or that the teacher was inadequately helped to carry out the programme as intended.The Chemical Bond Approach Project, the Chemical Education Material Study and the Nuffield Foundation Science Teaching Project were all three developed with considerable feedback from teachers and pupils under class-room conditions. This was judged adequate as a first approximation, but at the stage of adjustment and revision the time has come for a more critical study of specific cases of learning situations. Most progress has been made in techniques for measuring the extent to which individual pupils have achieved the objectives of a course. A series of publi~ations3~ on new examination techniques has been issued by the Schools Council in the UK.Mathews8 makes an important point that good examina- tion questions reflect good teaching situations, and examples of new types of questions from England, USA, Australia and Japan21 show that this is now an established approach. A primary question is what is the examination supposed to find out-whether it is a school examination for average pupils or a final degree examination for undergraduates. The IUPAC report,22 already referred to, detects at least eight purposes. These are: ( i ) to test achievement; (ii) to indicate a student’s potentiality in the subject; (iii) to select for further education; (iv) to select for employment; (v) to select for monetary or scholastic award; (vi) to test the effectiveness of teaching; (vii) to guide teaching methods; (viii) to guide pupils and diagnose their difficulty.It is questionable whether any one instrument can measure all these. For some years noyv the College Entrance Examination Board in New York has separated its examination into two distinct parts, achievement tests and scholastic aptitude tests; but more investigation is needed into the analysis of demands and responses in examinations for those who are not going on to higher education. If the purpose of the examination is to measure the extent to which indivi- duals have achieved the stated objectives the situation is clear. The designing of the instrument and the checking of its validity and reliability is nevertheless a complex and expensive operation if there are more objectives than just the ability to recall.Multiple-choice questions have been considerably developed in recent years and the early reaction that they only measured the trivial is now far from justified. They are difficult and therefore expensive to produce; their validity and reliability can be pre-tested and adjustments made if they are faulty; marking them is quick and objective. They are, however, restricting in that they limit a candidate’s response. The tendency now is to use them in their simple form for part of the examination, to devise programmed questions for another and to use open questions for a third part. Hudson has recently made Halliwell 219 a lively and stimulating survey of the way teenagers react to confrontation with a question.33 His analysis of pupils into a spectrum ranging from out- standing convergent thinkers to outstanding divergent thinkers cannot but be a warning to those planning programmes: it also underlines the need for examination demands complementary to straight multiple-choice questions.As yet we have only an incomplete picture of the anatomy of innovation in chemical education, and but tentative knowledge of the integrated functioning of its parts. The interactions which arise and the great range of expertise required in planning and achiving innovation in a complex society are formid- able. However, it is clear that there is at least one guiding principle, which is applicable to the whole gamut of organized education but which nevertheless provides considerable opportunity for individuality of treatment in the class-room. REFERENCES UNESCO, 1967.1 H. F. Halliwell, Strategy and Tactics in Educational Reform in Science, Sc/WS/ll Paris : 2 Report of the International Clearing House on Science and Mathematics Curricular Developments, Science Teaching Centre, University of Maryland, 1967. 3 J. S. Bruner, The Process of Education, 33, Harvard University Press, 1960. 4 G. Porter, The Laws oj’Disorder: an introduction to chemical change and thermodynamics, London: BBC, 1967. 5 R. W. Tyler, Educ. Res. Bull., 1933, 13, 197. 6 B. S. Bloom (ed.) et al., Taxonomy of Educational Objectives, Part I Cognitive Domain, 1956; Part I1 Afiective Domain, 1964; London: Longmans.7 University of London, General Cevt$cate of Education Examinations, Nufield Science Teaching Project Chemistry Ordinary-Level: Teacher’s Booklet and Candidate’s Booklet, Schools Examination Department, University of London, 1967. 8 J. C. Mathews, Educ. Chem. 1967, 4, 2. HMSO, 1960. 9 R. S. Nyholm, Marchon Lecture, University of Newcastle-upon-Tyne, 1964. 10 A. I. Yantsov, ‘Achievements in Modern Science and Technology and the Soviet School’, Supplement to the Higher Education Journal, 1959, 17, 11 S. D. Onabamiro, School Science Teaching, A Report of an Expert Conference held at the University of Ceylon, Peradeniya, December 1963, 50, London: HMSO, 1964. 12 Science in Secondary Schools, Ministry of Education Pamphlet No. 38, 1, London: 13 A Policy Statement issued by The Science Masters’ Association and The Association of Women Science Teachers, 4, London : Murray, 1961. 14 Nufield Chemistry: Introduction and Guide, Nuffield Foundation, London : Longmans/ Penguin, 1966. 15 J. S. Richardson, ‘The Consequences of Contemporary Advances in Scientific Know- ledge on Secondary School Science Teaching’, Supplement to the Higher Education Journal, 1959, 11. 16 J. F. Kerr, The Problem of Curriculum Reform, 1 1 , Leicester University Press, 1967. 17 Science Curriculum Improvement Study, University of California, Berkeley. 18 Nufield Junior Science, London : Collins, 1967. 19 Science for the Young School Leaver, Schools Council Working Paper No. 1, London : HMSO, 1965. 20 Time Space Matter, Secondary School Science Project. New Jersey: Princeton University. 21 ‘The Nuffield Foundation A-level Physical Science Course’, Educ. Sci., 1967, 29. 22 The Efect of Examinations in determining the Chemistry Curriculum up to the Level of University Entrance, Report of the IUPAC Committee on the Teaching of Chemistry, London: Butterworths, 1966. 23 Science in the Sixth Form, Schools Council Working Paper No. 4, London: HMSO, 1966. 24 Chemical Systems, Chemical Bond Approach Project, London: McGraw-Hill, 1964. 25 J. F. Kerr, Practical Work in School Science, Leicester University Press, 1964. 26 Designing for Science, Building Bulletin No. 39, Department of Education and Science, London : HMSO. RIC Reviews 220 27 J. A. Campbell, Chemical Systems: Dynamics, Energetics, Structure, San Francisco : Freeman, 1966. 28 W. Kauzmann, Kinetic Theory of Gases, New York: Benjamin, 1966. 29 Ref. 14, p. 17. 30 I. T. Ramsey, Models and Mystery, London: OUP, 1964. 31 Inquiry into the Flow of Candidates in Science and Technology into Higher Education, Chapter 7, London: HMSO, 1968. 32 Examinations Bulletins, 1965 onwards, London: HMSO. 33 L. Hudson, Contrary Imaginations, London : Penguin, 1967. 22 1 Halliwell

ISSN:0035-8940    

DOI:10.1039/RR9680100205

出版商:RSC    

年代:1968

数据来源: RSC

摘要:

Index to Volume I, 1968 . . . . . . 205 1 Chemical Education : Problems of Innovation Chemistry and the Consumer . . . . . . . . . . . . Chemistry of Tribology, The Gowenlock, B. G. and C. A . F. Johnson, Techniques of Physical 205 39 . . . . . . .. . . . . . . .. 135 39 . . Measurement : Vacuum Technique . . . . . . . . . . 107 Hallam, H. E., Infrared and Raman Spectra of Inorganic Compounds Halliwell, H. F., Chemical Education: Problems of Innovation Infrared and Raman Spectra of Inorganic Compounds Ives, D. J. G. and T. H. Lemon, Structure and Properties of Water . . Johnson, C. A . F.-see Gowenlock, B. G. Lemon, T. H.-see Ives, D. J. G. Roberts, Eirlys, Chemistry and the Consumer . . Rowe, Geoflrey W., The Chemistry of Tribology Structure and Properties of Water 62 1 13 Vacuum Technique . . .. .. Transition Metal Ions in Biological Processes, Role of . . . . .. 135 . . . . .. . . . . 62 . . . . . . .. .. . . .. . . .. 107 Williams, R. J. P., Role of Transition Metal Ions in Biological Pro- cesses . . . . . . . . . . .. . . .. .. 13 223

ISSN:0035-8940    

DOI:10.1039/RR9680100223

出版商:RSC    

年代:1968

数据来源: RSC