Relocation 10-20 wm and as the stylus is only um invalidate the result wide this is sufficient to The problem was overcome by Williamson and Hunt (1968) who designed what they called a relocation table The table is bolted to the bed of the stylus instrument, and the specimen stage is kinematically located against it at three points and held in position pneumatically The stage can be lowered and removed, an experiment of some kind performed on the specimen, and the stage replaced on the table Relocation of the stylus then occurs to within the width of the original profile (Fig 2.13) It is necessary to raise the stylus during the return stroke of the pickup, and of course the specimen may not be removed from the stage during the course of the experiment The original device has been widely copied (Grieve et al 1970; Thomas et al 1971; Thomas 1972) Figure 2.14 shows a succession of relocated profiles of a surface which was initially rough turned (Thomas 1972) After each measurement the stage was transferred to the table of a surface grinder and a slightly deeper cut taken The progressive O disappearance of the peaks and the persistence of the valleys can be (a) Relocation table Fig 2.13 showing stage raised and held in position by pneumatic clamps during measurement followed in great detail The same relocation table has also been used successfully to study the effect of successive coats of paint on surface topography (King & Thomas 1978) 31 Stylus instruments ©) (b) Relocation ready for removal table Fig 2.13 lowered 2.6 Replication g Replication will be considered here although it is not strictly speakin other to nary prelimi a itself a technique of measurement but rather used techniques, principally optical and stylus methods Replication is sect with some optical methods to provide a transparent specimen (see Pg.gg.1A., Replication Fig 2.14 N © Yocated profiles of an initially rough turned surface ©) provide a conducting 3.1.2.) It is also used in electron microscopy to use with stylus instruspecimen from a non-conducting workpiece Its which are not easily ments is generally to obtain measurements on parts surfaces (Sawyer accessible, such as internal surfaces or underwater ht to the instrument, such 1953), or which cannot conveniently be broug ns 1948) It has also as the rollers from steel mills (Pearson & Hopki direct measurement been used with compliant surfaces in the belief that on et al 1967/68) would damage or misrepresent the surface (Daws measured in contact The principle is simply to place the surface to be hopefully faithfully with a liquid which will subsequently set to a solid, , what might be reproducing the detail of the original as a mirror image t Materials such as plaster of Paris and dental cemen termed a negative to use a polymerizing have been employed, but it is now customary liquid the features of The vital question is how closely the replica reproduces us causes The liquid the original Lack of fidelity may arise from vario 33 Stylus instruments may not wet the surface completely; certainly it will first be necessary to degrease the surface carefully If the surface is itself already wet, as in the case of articular cartilage, there may be problems of diffusion or even of chemical reaction during setting Portions of the replica may adhere to the surface as they are parted unless a relea se agent is used In Fig 2.15 Comparison of power spectra of original and replica for three different replicating materials Effect of the act of replication on the original surface is also shown (George 1979) C ` đ|JE = Wavelength in wm 1,000 100 I ais ay & any case the replica is a negative and a stylus instrument does not respond to a valley bottom in the same way as to a peak In the case of transparent replicas, optical techniques generally rely on detecting an optical path difference which is a function of refra ctive index Misinterpretation can occur here due to inhomogeneit y of the replica or to changes in refractive index due to temperature A rigid replica may not reproduce short wavelengths faithfully, while a flexible replica will certainly not be faithful to long wavelength s 10 | Wavelength | a) rT” 1,000 21§ Sg Pia E lễ L 1H3 _ +R ——- _ _ F— Strand glass resin B -1 | Ly 10 C) -1 100 Wavelengthin 1,000 100 ị g1 ele s Warwick Spatial frequency in cycles/mm —! a 10 T LE a) a ap 100 in wm chemicals polymaster | Ly, 10 100 Spatial frequency in cycles/ mm! um 10 - 1,000 I ® E15 Wavelength in um 100 | HN [o 25 3/5 2|8 3| Kế Qa So (RS St _LLrL~t/unn Parent Strand glass resin C —Ï | ! Ly 10 100 Spatial frequency in cycles/mm_| tal CT ett Lr surface before and acrulite replica -I Ị 10 Spatial frequency after i 100 ° in cycles/mm7! > Surface mapping Of course if an R, reading is all that is required the loss of short © wavelengths is irrelevant, and acceptable results can be obtained (Van Dam 1953) It may be significant, however, that very few of the authors who have investigated replication techniques have ever reported a detailed comparison between replica and original A published comparison of SEM micrographs does not inspire confidence (Andersson 1974) Of course with stylus instruments it is difficult to find corresponding pairs of profiles A case has been reported where this problem was overcome (Pearson & Hopkins 1948) and excellent fidelity was achieved, but the care taken in replication was far greater than most workers would be prepared to take, and involved among other precautions the irradiation of the replica by ultra-violet light in situ for an hour as it cured One series of careful comparisons made (Sayles et al 1979) has found replicas of an optical flat of negligible measured roughness to show roughnesses of between 0.03 wm and 0.13 um, and replicas of machined surfaces to disagree in roughness with the originals by up to 17 per cent In two other recent papers, Narayanaswamy et al (1979) and George (1979) have attempted to compare the power spectra of replica and original This approach shows dramatically the range of wavelengths over which the replication material is effective (Fig 2.15) 2.7 Surface mapping As will be clear in subsequent chapters of this book, many fundamental problems associated with surface roughness require quantitative information about the surface in three dimensions (if this is not a contradiction in terms), i.e the same sort and amount of information as the Ordnance Survey require to produce detailed maps of the Earth’s topography None of the instruments so far described is capable on its own of providing such information, and various more or less ingenious systems have therefore been devised for the purpose One such system (Veerman 1962) employed the phenomenon of photoemission from strained regions of a surface A light beam is scanned across the surface in a raster by a servo-driven mirror system and the excited photoelectrons are picked up by a detector and displayed on an oscilloscope whose time base is linked to the scan Recognizable ‘pictures’ of the surface are produced, but their topography is distorted The largest area that can be scanned is only 0.4 mm x 0.24 mm, and of course the surface must be damaged in some way to produce a picture at all A ‘topografiner’ has also been described (Young et al 1972) in which a servo-controlled, non-contacting field emission probe maintains a constant current between a conducting specimen and itself The motion of the probe is amplified and displayed, forming an isometric picture as the surface is scanned in a series of parallel traverses The technique’s high resolution compares well with both stylus and microscopic techniques It appears, however, to be restricted to very small specimens 35 Stylus instruments and has the added disadvantage of requiring conducting specimens in a high vacuum To demonstrate the instrument a specimen obtained by gold-plated replication was used Replication is known to create in- accuracies in the small-scale structure of the surface (see above) which unfortunately seems to be the most useful range of this instrument An optical profilometer working on the raster-scan principle has also been described (Kelly et al 1977), but as its vertical resolution is reported as only 50 ym it will not be further considered here The most widely used alternative to the stylus instrument is the scanning electron microscope (SEM), which creates images of the surface from a raster scan of an electron beam Its high resolution and depth of focus make it a useful instrument for the'study of relatively smooth surfaces: however, its use in assessing rough surfaces is more difficult (Whitehouse 1972) One technique of SEM analysis employs a modified detector which follows the line-of-sight properties of back- © scattered electrons (McAdams 1974) These electrons, which travel in straight-line trajectories, are thresholded according to direction and produce an electron optical sectioning of the surface along the critical trajectory direction Surface elevation is therefore recorded as varia- tions in detector position, as opposed to signal intensity in the standard SEM The technique has been demonstrated on single sectional profiles but it is not clear how, without a fixed datum, profiles can be combined to give a three-dimensional representation Point-by-point comparisons of secondary electron images and stylus instrument raster scans (Samuels et al 1974) show rather poor agreement A multiple-detector technique from the same laboratory (Lebiedzik & White 1975) has produced rather better pictorial results, and reasonable agreement between measurements of average roughness As the most widely used method for roughness measurement is the stylus instrument, one would correctly suppose mapping techniques based on the use of the stylus instrument to be well to the fore As the stylus only traces out a line, some kind of scanning technique is required to cover an area In principle a polar scan (Edmonds et al 1977) or even © a spiral scan (Mollenhauer 1973) could be employed, but in practice the majority of experimenters have used a raster scan (McAdams et al 1968; Wallach 1969; Grieve et al 1970, Guerrero & Black 1972; Deutsch et al 1973; Samuels et al 1974), following the pioneering work of Williamson (1967/68) The main problems and limitations of stylus systems for three-dimensional work are the establishment of an arbitrary flat datum plane; the small size of the surface amenable to analysis because of the vast data-storage and data-handling problem; and the physical difficulty of recording such data in a reasonably short period of time In the system operating in our own laboratory (Sayles & Thomas 1976a) the specimen is mounted on an air bearing By employing a linear air bearing, the short-wavelength structure of the mating bearing surfaces is effectively smoothed out The action is analogous to.that of a spectral window: the pressure distribution falls away smoothly and slowly with distance from the centre of each air jet Thus any short- Surface mapping wavelength structure will have little effect on the overall flow, and the subsequent change in height of the bearing will be only a fraction of the amplitude of the surface structure The automatic stepping mechanism (Fig 2.16) on the air bearing positions the specimen relative to the stylus prior to each traverse To avoid backlash a simple indexing system is employed Connected to the air-bearing lead screw, a modified 0.5 mm pitch micrometer spindle, is a 125-tooth ratchet wheel which is rotated one tooth by energizing a solenoid on a computer command One computer-generated pulse therefore represents a specimen movement of 4m normal to the stylus traverse Programming controls the number of pulses in each step, consequently any multiple of 4m can be obtained between traverses The width of the Talysurf stylus is about wm (Jungles and Whitehouse 1970); return a resolution greater than m is therefore unnecessary To the stylus after each traverse, a pneumatic ram, mounted externally to the Talysurf, operates the manual control on the front of the gearbox Air supply to the cylinder via a computer-controlled valve is restricted to allow a smooth return action of the stylus When building a map of a surface by parallel traversing it is essential to maintain a common origin for each profile Williamson (1967/68) aligned his profiles with a scratch on one of his coplanar flats In addition he cross-correlated adjacent profiles, a method also adopted by Fig 2.16 Mechanical components of threesystem measuring dimensional (Sayles 1976) The specimen, a tool bit, is mounted on the moving component of an air bearing The indexing mechanism is visible in the foreground The transparent Perspex plate carries the pneumatic actuator for the pickup return © 37 on ee TÔ NNớc Stylus instruments = Grieve et al (1970) This method is applicable only if the surface structure is random Should a predominant periodic component be present, representing long-crested asperities not parallel to the ` 3) traversing, then cross-correlation would tend to distort the seconda ry structure of the surface by aligning the main periodic structure A stylus relocation device is thus required which is reproducible to a fraction of the minimum sampling interval and whose operation is consistent over long periods of time Such a device must also operate in a mode compatible with the control systems available on the computer In the simple arrangement here employed the Talysurf pickup opens a compound switch by contacting its leaf spring, thus triggering the computer sampling routines through the contact sense facility To prevent contact bounce a thyristor is connected across the switch and is reset after each traverse by another switch connected in series The device is free from drift, a problem found with electro-optical systems, and accurate to within the minimum tolerance set by the leading edge of the current pulse received at the computer (Fig 2.17) This is about ms with the length of line used, and represents a maximum possible stylus relocation error of +1/8 the minimum sampling interval available at each gearbox speed The system is operated and controlled remotely by an IBM 1800 computer, which also converts the data to digital form and records it The operating sequence is to return the stylus to the start position of each traverse; to index the specimen by the required traverse spacing; and to record the requisite number of ordinates in each traverse This procedure is repeated over each traverse until the required area of surface is mapped To ensure the relative position of adjacent traverses, the relocation switch described earlier determines the position on each Fig 2.17 Two views of the letter ‘D’ from the Prince of Wales’ motto ‘Ich Dien’ on the reverse of a British 2p coin (a) By SEM (b) By the system shown in Fig 2.16 Note Cpt the stylus system has faithfully eproduced the small blemish left by the die below and to the right of the lower serif (Sayles & Thomas 1976) (a) Cc ES ee tS Stylus instruments O l Fig 2.18 Computer-plotted isometric view of an area 2.8 mm x 2.4 mm of a worn gear tooth — SSAC SSS ` Seen SS ES QE RA lạ `» ASSES ` ca? ES N > traverse where data acquisition should begin Over a traverse length of mm at the fastest traverse speed, with five indexing steps between each traverse, the cycle of events, including data storage, takes about 18 s An area of mm X mm has been represented by 400 traverse s of 406 ordinates per traverse, taking a total time of about 24 hours to map An example of the pictorial information obtained is shown in Fig 2.19 A commercial three-dimensional stylus measuring instrument checking the geometry of an integrated circuit (courtesy Gould Inc Instruments Division) ... conventional rough surfaces are almost invariably composite; that is, they contain a distribution of roughness scales from ? ?rough? ?? to ‘smooth’ Even C ng O Optical methods that time-honoured example of rough- surface... of an optical flat of negligible measured roughness to show roughnesses of between 0.03 wm and 0.13 um, and replicas of machined surfaces to disagree in roughness with the originals by up to 17... decomposition of the surface roughness The next texture class is random one-dimensional (1-D) roughness, which scatters light along a line passing through the specular spot If this roughness was literally