Mechanical Engineers Reference Book 12E Episode 10 pot

Wear and surface treatment 9/85 the excellent abrasion resistance of PVD titanium nitride under the light load that was used.. It is useful to use a similar concept for fretting in ord

Wear and surface treatment 9/81

as tungsten, molybdenum and vanadium While there is a tendency to think that electrolytic deposits are mainly for corrosion resistance? decorative purposes or electrical/ electronics uses, there are many engineering/tribological applications for electroplates Hard and soft plates are used, depending on the particular function required To resist abrasive wear, adhesive pick-up and corrosion, hard chro- mium plates6' are ideal Porous or intentionally cracked chromium plates are used for oil retention, as in automotive cylinder liners The hardness is 8%-900 Ha/ and coatings thickness is typically 20-100 pm There are several propriet- ary variations on the plating techniques; some baths giving ultra-hard (1110 HV) deposits and others giving a dense, crack-free, but rather softer (700 13%') coating Electrolytic coatings tend to build up on outside corners and sharp edges (because of a concentration in the current density) and to give reduced thickness in holes and on inside corners

Soft plates of tin are used to facilitate 'running in', prevent fretting and confer corrosion resistance Plates of silver, lead, cadmium, tin and antimony are used in heavy-duty sleeve bearings, particularly in aircraft power units Nickel plate can

be deposited from a wide range of solutions and is used to

minimize abrasive wear in cases such as sliding contacts on hydraulic rams Some care should be taken in the selection of electrolytic nickel, particularly with respect to the counter- face, because of its tendencies to gall Nickel is a good undercoat plate for hard chromium However because the shock resistance of chromium is poor, it is prudent to make the bulk of the layer nickel, and give a relatively thin top-coat plate of chromium

Electroless nickel plates.68 autocatalytically depositing nickel-phosphorus (Ni-P) or nickel-boron (Ni-B), have many useful tribological applications In the case of the Ni-P deposits, a hardness of about 500 HV is obtained but can be thermally aged to a hardness in excess of 1000 HV This is achieved after one hour at 400°C by the precipitation of nickel phosphides Such hardness is not retained at high tempera- tures but the Ni-B deposits are superior in this respect The range of applications of nickel plates can be increased by incorporating fine dispersions of wear-resistant particles in the plating solution (NiC, S i c or A1203) Such coatings are particularly effective in high-temperature wear situations Electroless nickel is also available with the addition of PTFE This duplex coating is less hard but has excellent nonstick properties

In general, electroless nickel is not as abrasion resistant as hard chromium plate but, because it is not an electrolytic process, it does perfectly repiicate the component surface without build-up on edges or corners It is normally applied in thickness between 10 and 50 pm and it has excellent corrosion resistance, particularly in the 'as-plated' condition Le straight from the plating bath

Both electroless nickel and hard chromium plate are com- monly applied to ferrous materials but they are now also used increasingly for components made in aluminium Plating baths can usually accommodate components up to a metre or more

in size

These coatings are mainly used as a base for paint but they are effective in the presence of a lubricant to ease the deep drawing of steel, and to decrease the wear and fretting of sliding parts, particularly during 'running-in' pro- cesses The identity of the phosphating process is often concealed under a proprietary treatment In general, they are based on dilute phosphoric acid solutions of iron manganese and zinc phosphates either separately or in a combination Accelerators are added to shorten the process time to just a few minutes (in the temperature range 43-72°C) The simplest

textiles and filled plastics It can also be applied to shafts,

gears or plain bearings in equipment handling abrasive slur-

ries If there is an element of corrosion then the substrate itself

needs t'o be resistant Alternatively, TiN can be applied over

the top of another coating (for instance, electroless nickel) It

can be applied to carburized surfaces with only minimal

temperling, but if it is to be applied to a nitrided or nitrocarbu-

iized material, the compound layer must be removed first

There are now PVD chambers that can coat components up

to 3 m long and processing time and costs are being contin-

uously reduced PYD is a semi-'line-of-sight' process and so

holes and re-entrant areas will receive less coating Typically,

at the equivalent of one diameter deep down a hole, the

coating thickness will be 30% of that outside the hole The

coatings replicate the underlying surface texture without

build-up on edges or comers

Finally, PVD is a high-technology process It is best suited

for high-quality components; bright, clean, free from oxide,

burrs and contaminants

Chernic8uE vapour depo~ition~"~~ Chemical vapour deposition

(CVD) is, in many ways, a competitor of physical vapour

deposition The process is used to deposit metals and ceramics

by the decomposition of a reactive gas at the surface of

components placed within the chamber For instance, titanium

carbide is produced by a reaction between titanium tetrachlo-

ride and methane, or titanium nitride can be produced by

replacing the methane with nitrogen or by using hydrogen and

ammonia

Deposition temperatures are very high, usually above

800°C Thus there are potential problems with softening and

distortion of substrates In fact, the CVD process is most

commonly used to coat tungsten carbide tooling, particularly

the indexable inserts used for high-speed turning The cobalt-

cemented carbide is an ideal substrate because it has a similar

coefficient of thermal expansion to the T i c or TiN coatings

Also, it does not suffer any volume change during cooling and

gives go'od support

There: can be problems of decarburization of substrates

during deposition and it is common to apply duplex or

multiple coatings (for instance, TiN on T i c or A1203 on Tic)

to produce graded properties The coating thickness is

typically 1-5 pm It is possible to apply the process to high-

speed or tool steel components but these must be reheat-

treated to retrieve the substrate hardness This will produce

distortion and, in many cases, will bring an unacceptable loss

in precision However, for tools with non-critical dimensions,

CVD coatings can bring the same benefits to life as those

applied by PVD Equally, because the coatings have high

hardness, they can be used to resist low-stress abrasive wear

The syslem is well suited to handling large numbers of small

items, the parts being simply jigged on trays in the furnace

CVD coatings are also used to protect against corrosion and

corrosivse wear For instance, chromizing (referred to in

Section 9.8.3.2), evein though it is a pack process, can be

considered as CVD because it occurs by the decomposition of

a gas

Finally, it is now possible to combine the PVD and CVD

principles in the form of plasma-assisted chemical vapour

deposition (PACVD) bringing the flexibility of the chemical

process at a much lower temperature (less than 300°C)

However as yet, deposition rates are low and applications are

mainly in thin (submicron) coatings for electronics applica-

tions

Eectrolytic and electroless coatings Over 30 metals can rea-

dily be deposited from aqueous solutiod6 but they do not

include alkali or alkali-earth metals and refractory metals such

Tribology

phosphate coating consists of grey to black crystals of

Fe3(PO4)* and FeH Pod Zinc and manganese phosphates

produce more complex layers which absorb lubricant more

readily, and are effective in reducing adhesive processes such

as galling, pick-up and scuffing In addition to phosphating,

there are many chemical conversion coatings which involve

dipping components in solutions to develop specific com-

pounds Treatments such as chromating, used on non-ferrous

alloys to prevent corrosion, will hold lubricants, and provide a

base for bonded lubricant coatings

Sprayed coatings70 In spraying techniques, powders are

heated to a semi-molten state and deposited at high velocities

onto the component surface Coating thicknesses vary from

about 0.05 to 1.00 mm The techniques can be divided broadly

into flame gun, arc, plasma-arc and detonation gun processes

The merits of any one technique over another need to be

assessed with reference to the particular job in hand Ob-

viously the selection must be on a cost-effective basis taking

due account of the integrity required for a particular duty

One of the difficulties with these processes is to assess the

substrate bond integrity, porosity and general coating qualities

on a production basis Suppliers are well aware of this, and the

usual approach is to design their coating methods with care, so

that tight control of the process variables is maintained by

following the set procedure at every stage

Electric arc spraying is used for metal deposition for wear

resistance, corrosion resistance or reclamation The coating

material is fed as two wires and an electric arc is struck

between them to cause melting The molten metal is then

propelled onto the substrate by compressed air

In flame spraying, the source of heat is a burning gas, such

as acetylene, and the coating material is fed into the gun either

as a wire or a powder It is a relatively cheap process and gives

a high deposition rate but, in general, the bond strength is

lower and the porosity is higher than that achieved by the

electric arc process The spray-fuse process takes the tech-

nique a step further by first spraying and then fusing with a

second heat source, such as flame, torch or by induction This

is the basic technique used for the nickel- or cobalt-based

Stellite-type alloys and gives excellent resistance to corrosion,

erosion, abrasion and fretting Alloys can have hardnesses up

to 900 HV with near-zero porosity There will, of course, be

substrate distortion and the parts will require finish grinding

Coating thicknesses are typically up to 0.5 mm

I-

The plasma spray process makes use of an ionized gas (usually argon or nitrogen) to produce much higher tempera- tures than those created during flame spraying This allows deposition of higher melting point materials such as metal oxides or metaVceramic mixtures Such coatings have relat- ively good substrate adhesion and porosity levels are usually in the range 2 4 % (Figure 9.71) A typical coating thickness would be 0.1 mm and the process finds a wide range of applications in resisting abrasive wear Substrate heating is minimal

There are several variations on the plasma arc technique The process can be conducted in a partially evacuated chamber (Low Pressure Plasma Spraying - LPPS), giving reduced porosity, reduced oxidation of the coating material and, because the substrate reaches a higher temperature, better adhesion The same advantages can be gained by shrouding the arc in a non-oxidizing gas (Inert Atmosphere Plasma Spraying) and both these processes are now used increasingly to deposit the nickel-cobalt-chromium-alumin-

ium-yttrium (MCrA1Y)-type alloys which are used extensively

in the aircraft industry to resist high-temperature oxidation erosion and fretting A third variation is the Transferred Plasma-Arc process in which a secondary electric current is established between the arc and the workpiece This promotes substrate heating and surface melting and gives more dense and more adherent coatings Deposition rates can be very high and the technique is typically used to deposit thick (up to

10 mm) abrasion- and erosion-resistant coatings for use in applications such as mining and agriculture The substrate must be electrically conducting and be able to withstand some thermal distortion

The ultimate spray techniques are those producing the highest particle temperatures and velocities These are achieved in the proprietary techniques such as Detonation Gun, Jet Coat and Mach Stream As their names suggest, they are high-velocity techniques based on combustion of high- octane fuels Porosity is very low (less than half of 1%) and substrate adhesion is excellent They are used to deposit tough, abrasion- and erosion-resistant coatings such as chro- mium carbide or tungsten carbidekobalt cermets Thickness is typically 0.1 mm and substrate heating is minimal

For all spray techniques, a correct substrate preparation is essential The surface should be clean, free from scale, flash and burrs and should be pre-roughened by a grit-blasting procedure Attention should be paid to the required coating

Plasma- sprayed coating

I-

Substrate

t-i

20 pm

Wear and surface treatment 9/83 compress the apparent differences between coatings and it will

be difficult for users to predict the effects in their own applications, where completely different abrasives are Iikely to

be present In addition, a designer or user has to be sure that wear data were obtained under relevant conditions If high loads are present, it is no use reiying on data produced under Iow-load conditions Thin coatings or treatments even though they might be very hard, can be crushed into the substrate and torn away without any benefit

For other wear situations the idea of wear rates is equally doubtful The designer will rely on data or experience which shows, for instance, that certain coatings are better in fretting situations than others, or that some treatments are effective in corrosive applications Equally, in situations involving severe wear problems such as seizure, galling and scuffing wear rate has no meaning Material is transferred, torn and deformed; surfaces weld and seize The objective is to eliminate the problem, not merely to reduce it Wear data, therefore, are most commonly published for relatively simple situations of adhesive or abrasive wear The range of wear machines is extensive and some, together with examples of wear data, are described below

Adhesive wear data are produced in many test geometries

In most cases, the geometry is chosen because it allows a quick and easy measure of wear, rather than in any attempt to simulate a particular application The most common wear machine is probably a ‘pin-on-disk’ arrangement, with the wear on the pin being measured from its weight loss or reduction in length and that of the disk from profilometry of the worn groove Figure 9.72 illustrates the typical mild/severe wear transition that takes place with a normalized medium- carbon steel rubbing dry against itself If the disk is through- hardened, the severe wear regime is suppressed and a mild

distribution so that the spraying procedure can be optimized

and, for mass production, automated and computer con-

trolled

Laser dioying and ciaddi~tg~’.’~ These are relatively new

processtes For cladding, the powdered material is biown

directly into the laser-generated melt pool Laser alloying is a

similar process, except that the energy is increased to produce

more substrate melting and complete surface alloying with the

powdered material The reaction area is shielded with an inert

gas A particular advantage of such techniques is that a

specific area can be treated, thus minimizing component

distortion

Corrosion-resistant surfaces and metallic glasses are being

produced but more development is required before the pro-

cesses have wide industrial application One of the problems is

that of controlling the depth of heating, particularly when

thin-layer fusion zones are required

Welding and roll-cladding” These processes involve relat-

ively thick layers, typically up to 2 or 3 cm Welding can be

used to good effect in tribological situations where high-stress

abrasive wear is the problem, such as coating digger teeth,

tank tracks and on ore-handling equipment For instance,

some of the Stellite-type coatings are applied by weld deposi-

tion

Cladcling is usually associated with corrosion or mild wear

problems that are encountered in the chemical, food process-

ing or printing industries The two processes, roll-cladding and

weld cladding, are complementary; hard abrasion-resistant

materials are difficult to fabricate and are best deposited by

welding, while the more corrosion-resistant materials are

based on ductile austenites and are amenable to roll-cladding

and forming

9.8.4 ~ ~ i b Q ~ Q g ~ ~ a 1 dlata

Friction and wear data are available for surface treatments and

coatings Designers can sometimes find them in the open

literature or in the advertising information which accompanies

the various products However, before they can make use of

such data, they needi to consider how the wear tests were

performed Le the type of laboratory wear machine that was

used, and relate the conditions to those present En their own

applications The object of this sub-section is to acquaint the

designer with the wear data that are available and to assist in

interpretation and application

As described in Section 9.1, wear is expressed in terms of

the rate of material volume removal, usually as a function of

applied loading and rubbing distance For metal-on-metal

rubbing under dry conditions, the rate of wear is often

proportional to load and distance, and it is just possible that

data could be related to a designer’s specific application

However, it is more likely that there will be some further

influenoe on the rubbing situation (for instance, a process fluid

or lubricant) and that absolute wear data cannot be obtained

The situation in abrasive wear is even more complex Wear

rate is a function of so many factors (for example, hardness,

impact velocity and particle shape and size) that there is no

prospect of users finding data which have any absolute mean-

ing to their applications Relative wear data that rank a range

of treatments in order of wear resistance can usually be found

but, even then, the rainge of wear will depend greatly on the

aggressiveness of the abrasive For instance, two sets of wear

tests, one using Sic (2500 HV) as thie abrasive, the other using

Si@ (800 HV), may both produce ?he same ranking for a

number of coatings or treatments but the relative wear rates

will be very different The harder abrasive will tend to

(Severe wear)

(Mild wear)

I

1000 Sliding speed (cm s-’ )

Figure 9.72 Pin-on-disk tests for normalized 0.4%C steel, showing transition in wear

G = boronizing

H = carburizing

c = low-temperature K = plasma n i t L carbonitriding carburizing

Figure 9.73

0.4%C steel

Pin-on-disk tests on various surface treatments on

wear rate of about 1 X lo-' cm3 cm-' kg-' (volume per

distance travelled per unit load) is maintained Figure 9.73

shows that most of the thermochemical diffusion treatments

have the same effect, with only the sulphur-based process

failing to give mild wear This provides an excellent example

of the difficulties inherent in wear testing The sulphur-based

treatment is aimed specifically at eliminating scuffing, i.e

under lubricated conditions, and a different test is required to

highlight its advantages

Figure 9.74 shows similar wear data for an aluminium alloy,

demonstrating the reduction in wear achieved by anodizing

(particularly when PTFE is incorporated into the layer) and by

the Zinal electrolytic deposition process In this case, the wear

rates are about a factor of 10 higher than those found with

medium-carbon steel Another test geometry used for produc-

ing adhesive wear data is 'crossed-cylinders' Here the contact

stress is high and certainly unsuitable for evaluating thin

Wear and surface treatment 9/85

the excellent abrasion resistance of PVD titanium nitride under the light load that was used However, if high loads or large abrasive particles are employed the results would be completely changed, with only the thicker surface treatments surviving

The lesson on wear data is, therefore, very clear The

designer should treat it with caution, making sure that it

involves the appropriate wear mechanism and that the test conditions have some relevance to the particular application

coatings Tests may also be quoted from Falex ‘pin-and-jaw’,

or ‘four-ball’ machines These evaluate scuffing or seizure

resistance and a ‘failure-load’ rather than a wear rate is the

usual quoted result A ‘disk-on-disk‘ geometry (sometimes

tested with a combination of rolling and sliding) gives a line

contact and tests can lead to fatigue pitting as well as adhesive

wear

The irange of machines used for performing abrasive wear

tests is more limited Data are produced by rubbing samples

against abrasive paper disks (sometimes in a spiral to ensure

that fresh abrasive is acting throughout the test) It may also

be produced from a grit-blasting machine (erosion) or from a

rubber wheel tester In each case the result is likely to be

expressed as a weight loss or penetration depth per unit time;

Le simple relative data An example of rubber wheel test data

for abrasion by Si02 is shown in Table 9.22 It demonstrates

Table 9.22 Abrasive wear rates for several coatings and

substrates: rubber wheel test

after 100 revs

at 130 N load (mm3) ( x ~ o - ~ )

9.8.5 Selection philosophy

As stated earlier, the process of selecting the wear-resistant surface should be started at the design stage If due considera- tion is given to the environmental factors, and if the user processes are considered at the onset, then preliminary selec- tion will be relatively easy (Figure 9.75)

The first stage will be to satisfy the mechanical engineering

demands for the component Ideally, the lowest cost compo-

nent which will meet their demands is required This usually means that, when manufacturing a component, it must be made with the cheapest material compatible with its design requirements, and be fabricated with the minimum number of low-cost operations During this stage, the designer will be considering all the mechanical and environmental require- ments; for example, there may be a need for corrosion resistance, good fatigue properties or resistance against creep

resistant surface treatments must be employed It is important

that the type of wear is carefully identified; abrasive (two- body, three-body, high or low stress), adhesive, erosive,

DECIDE WHETHER Use wear

availnbie OR COATINGS A R E

REQUl RED

t

CONSIDER L O C A L FACTORS

Selection of wear-resistant surface -

Tribology

fretting, chemical or fatigue Also, it is necessary to obtain

data on likely rubbing conditions, such as sliding speed,

contact pressures, load cycles, hardness and type of any

abrasives and the presence of any corrosive medium

If wear data are available and those data are relevant to the

load, sliding speed, environment and counterface material,

then this will assist in the selection process However, the

designer should look carefully at any wear data found in the

trade literature supplied by the surface treatment specialist, or

in published papers and journals, and ensure that they were

obtained under a relevant wear regime It is totally

inappropriate to assume that wear data obtained under, say,

abrasive conditions, would have any meaning to an application

involving another wear process such as adhesion, fatigue or

corrosion Obviously, each design and component cannot be

subjected to tribological testing, but there are wear and

performance characteristics of classes of materials and surface

treatments which allow a primary selection

By this stage of the selection process, the engineer should

have a reasonably short list of materials and surface treat-

ments that could be used, and should have considered possible

manufacturing routes and worked out some detailed require-

ments; for instance, depth of surface treatment (determined

by the loading conditions), hardness, core properties, whether

the total surface area of the component needs treating, the

required wear life, number of components to be produced,

etc However, a number of options for surface treatment will

remain and should be reconsidered when assessing the ‘local’

factors The final selection must be based on what is practic-

able, available and economic (Figure 9.76)

First there will be the question of availability Usually, the

user will like to sub-contract any surface treatment fairly

locally The designer will have to ensure that the selected

process is available and that the processor has the necessary

equipment, working skill and quality control to provide the

treatment reliably and reproducibly at the volume of produc-

tion anticipated

Second, the designer must consider the geometry of the

component Complex geometries, incorporating a number of

section thicknesses, are likely to distort when treated at high

temperature This may limit the choice to low-temperature

treatments The heat-treatment history of the component is

important Specific core properties may have been produced

by hardening and tempering so that any subsequent surface

treatment must be applied at a temperature below that of the

final temper If this is not so, the core may be softened and the

coating/substrate combination may then have insufficient

load-carrying capacity to withstand the service contact

stresses Depending on the areas to be treated, the geometry

may also exclude all ‘line-of-sight’ plating or coating pro-

cesses Obvious examples of this are the ceramic and cermet

spray processes such as plasma arc and detonation gun Ion

implantation is also a line-of-sight process and physical vapour

deposition techniques have only limited ability to penetrate

re-entrant surfaces or holes

In some applications the replication of the surface shape

may be important The electrolytic plating processes tend to

deposit thicker coatings on sharp corners and peaks while

leaving valleys with little or no coverage In contrast, electro-

less coatings tend to replicate the shape perfectly and may be

preferred On a finer scale the designer may need to preserve

a particular surface finish, perhaps applied to the component

for reasons of controlling friction or to provide specific optical

properties, and so may be limited to ion implantation, physical

vapour deposition or electroless coatings

In all cases the designer will be considering the dimensional

requirements and making allowances for any final machining

or grinding after treatment Low-temperature processes that

A V A I L A B I L I T Y OF TREATMENT

FACl LlTlES

COMPONENT GEOMETRY PROCESS TEMPERATURE

D ISTO RTl ON

COMPONENT MASS COMPONENT SIZE HANDLING FACILITIES FOR LARGE NUMBERS (JIGGING ETC)

PRACTICAL COATING THICKNESS

I

LABOUR SUITABILITY TOXICITY ETC

ECONOMICS OF THE PROCESS

Figure 9.76 Selection of wear-resistant surfaces - local factors

require no post-treatment finishing operations and retain dimensions after treatment are particularly attractive

Third, the question of component mass, size and numbers must be considered These are obvious points to look for but it

is surprising how easy it is to overlook such facts as: the length

of the component you want to nitride is a few inches longer than any salt bath available in the country, a component requiring a coating applied by physical vapour deposition turns out to be too heavy to be handled inside a vacuum chamber, or a particular heat treatment service does an excellent job on 10 components but is totally impracticable to deal with 100 000, because the logistics and cost of jigging would be prohibitive The designer should always seek advice from the contractor before finalizing on the surface treatment

A fourth point to consider is the practical coating or treatment thickness Although the wear performance of a particular treatment may be excellent it is advisable to check whether the particular process is capable of giving the desired depth of treatment to withstand the design pressure through- out its anticipated service life Some of the processes, particu- larly ion plate deposits, have excellent hardness, low friction, good corrosion and anti-galling characteristics but it is imprac- tical to deposit more than a few microns onto a substrate

Similarly, some thermochemical diffusion treatments, electrolytic/electroless plates and sprayed surfaces cannot be

Wear 2nd suriace t~eatment 9/87 and, with good documentation, faults can be traced back through the process and quickly rectified

developed to adequate depths for highly stressed surfaces For

this reason, the design engineer must consider not only the

surface but also sub-surface stressing on a particular compo-

nent

If the coating or surface treatment technique is to be

brought ‘in-house’ i.e out of the hands of sub-contractors, it

is important to consider all the implications When introducing

new skills new processes, high technology, processes requir-

ing the use of chemicals, salt baths, toxic materials or radia-

tion, all the technical and practical aspects must be appre-

ciated It is equally important to consider how they will be

received at shopfloor level

Finally, it is importan: to consider the economics Surface

treatments can provide tribological and environment-

compatible surfaces on relatively cheap substrates, and unless

the designer has a knowledge of the range of treatments

available it will be difficult to select the most cost-effective

solution

9.8.6 Quality control

A designer or engineer who is proposing the use of a surface

coating or treatment on a critical component requires confi-

dence, not only that he or she has chosen the correct solution

to the problem, but confidence in the integrity and reliability

of the treatment process Certification of a product should of

course, be the duty of the supplier, but it is up to both the user

and the surface treater to agree what surface properties are

important and how to specify them This should lead to a

formal quality-control procedure with full documentation

The surface properties to be specified depend on the specific

application However, jt is likely that, for modification tech-

niques such as case hardening, nitriding, induction hardening,

etc the two key properties will be hardness and case depth

For coatings, the thickness and hardness will again be impor-

tant but the bond strength to the substrate will be critical No

matter how hard or thick it is, if the coating falls off, it cannot

do its jobs For sprayed coatings, porosity may be an issue (for

instance, allowing a corrosive medium access to the substrate)

and, in :some applications, surface finish may need to he

specified

A further question to be answered is whether the quality-

control tests should be carried out on the actual component (in

which case they may need to be nom-destructive) or whether

they should Re performed on a test coupon ( a small, flat plate

sample) that is treated at the same time The danger in using a

coupon is that it may not experience the same treatment

conditions as the component For instance, it is nearly always

easier to (effectively treat a flat plate than to deal with a sample

with complex geometry The ideal solution is to include with

the batch some components (or representative sections) that

are expendable and can be used for destructive quality-control

tests The number of samples and the frequency of checks

would, again, be a matter of agreement between user and

supplier It should be decided on the hasis of batch sizes and

on the likely consequences to the end product of faulty surface

treatment being undetected

The detection of mistakes is, of course, the very purpose

behind quality-control )tests and some ideas for test procedures

are given below However, the real key to control is in quality

assurance, i.e the close definition and monitoring of the

treatment procedure itself The key treatment parameters

should be identified and the user should demand full docu-

mentation of each treatment cycle This way, the product is

reliable and there is less need for costly and time-consuming

monitoring of the final items There should always he some

quality-control testing, the unknown can sometimes occur

9.8.6.1 Hardness and case-depth sf treated surfaces

The only reliable way of determining the hardness and depth

of hardening is to prepare a polished cross-section through the treated sample and perform a series of hardness tests from the surface into the hulk Ohviousiy, this is destructive and must

be performed either on a tab sample or on sacrificed compo- nents

For case-hardened components the depth of hardening is defined as the distance from the surface to a plane at which the hardness is 550 HV The measurements must be made using a load of 1 kg The same method is applied to induction and flame-hardened surfaces but this time the convention is to define the limiting hardness at 400 HV This definition also applies to nitrided surfaces, but only for conventional nitriding steels For other nitrided steels (for instance, a high-alloy stainless steel) the specification would have to he agreed between the parties concerned All these conventions apply to steels with a case depth of more than 0.3 mm and are covered

by I S 0 Standards 2639,3754 (1976) and 4970 (1979) For cases less than that value it will be necessary for parties to draw up their specification In particular, for very shallow treatments (for instance, the surface ‘compound‘ layers produced by nitrocarhurizing processes) it may be necessary to use very light hardness loads, perhaps as low as 0.1 kg This demands expert metallographic preparation and careful measurements, taking account of the wide statistical spread in low-load hardness values For ultra-thin diffusion layers, such as those produced by ion-implantation, detection of the hardened surface is possible only by an ultra micro-hardness tech-

n i q ~ e ’ ~ However, this is really an academic tool and not suitable for quality control

9.8.6.2 Hardness of coatings

In the case of coatings, the whole layer usually has a constant hardness, so the idea of a ‘case-depth’ is inappropriate For dense coatings such as electrolytic chromium plate or electro- less nickel the measurement of hardness on a polished section

is straightforward, usually being performed at a load of 1 kg For sprayed coatings the results may be affected by porosity, the material collapsing into sub-surface voids and giving low hardness values In this case, hardness would be quoted alongside porosity (discussed below)

With very thick coatings, the hardness can be measured directly on the surface, provided that the surface finish is good and the load is not so high that the coating collapses into the substrate For very thin coatings, only 1 or 2 pm thick (typical

of coatings applied by physical vapour deposition), no direct method is practicable, The hardness can be predicted by performing a series of tests into the surface at different loads and using the analysis described by Thomas.74

9.8.6.3 Coating thickness

Obviously, a micrograph prepared at a known magnification from a polished section provides a positive record of coating thickness However, this is a measure at one single point and it

is often important to map the coating thickness over the contours of the components

For coatings in the thickness range up to 30 pm, X-ray fluo-

r e s ~ e n c e ~ ~ gives excellent results (provided there is atomic number contrast between the coating and substrate) It now supersedes the traditional method using electrons - Beta- backscattering It is particularly effective for measuring the

9/88 Tribology

thickness of coatings such as titanium nitride or zirconium

nitride applied by either physical or chemical vapour deposi-

tion

For thicker coatings, other techniques are available These

can be based on the use of eddy currents, ultrasonics, thermal

waves etc

9.8.6.4 Coating porosity

Porosity is important in most sprayed coatings and the best

method of measurement is to prepare micrographs of polished

sections Porosity can then be assessed by a visual comparison

with ‘standard’ photographs The major coating companies

have established such standards, usually covering a range from

0.25% to 10% porosity An absolute measure of porosity can

be made from a micrograph by making an area of line tracing

and determining either the area or length of the voids as a

ratio of the total (for a homogeneous structure volume, area

and line porosity are equal) It is important that the polished

section is prepared with care, so that the polishing procedure

itself does not pluck out material and create ‘false’ porosity

This is a particular problem with the harder plasma-sprayed

ceramics and causes frequent disputes between users and

coating contractors The polishing procedure should have

been worked out previously and it should then be included in

the overall quality-control specification

A second possibility is to measure porosity by vacuum

impregnating the coatings with a low-viscosity fluid and mea-

suring the weight taken up This relies on accurate knowledge

of the volume of coating (Le the area and the average

thickness) on the pores being interconnected and on all the

pores being filled (For very fine pores, capillary forces may

prohibit complete filling.) These are significant points but, if

the process can be perfected, it does have the merit of

determining a volume porosity In contrast, a polished section

gives only a single point value

9.8.6.5 Coating adhesion

The adhesion is the key property and the most difficult to

quantify In fact, measurements are usually qualitative rather

than quantitative

To produce an absolute measure of bond strength (in units

of forcehnit area required to detach the coating) it is possible

to use a tensile-type test This relies on gluing a peg to the

coating with a high-strength epoxy and measuring the tensile

force required to pull off the coating In practice, the glue

usually has a lower bond strength than the coatinghnterface or

the coating fails cohesively Additionally, the difficulties in

setting up and pulling the pins accurately at right angles to the

surface mean that a statistical approach, with multiple tests, is

required

For thin coatings, such as those applied by physical or

chemical vapour deposition, a scratch test technique” is

available This uses a spherical diamond which is dragged

across the surface at a steadily increasing load The point of

coating detachment is detected by measuring the friction or by

monitoring the acoustic emission, and a ‘critical’ load is

assigned as a measure of adhesion Such a test usually requires

a flat plate tab sample

For thicker coatings there are a number of alternatives

Samples can be bent in a closely defined way and the bent area

inspected for flaking Alternatively, they can be thermally

shocked, thermally cycled, impacted or subjected to high-load

indentations (for instance, using a Vickers hardness tester with

a pyramidal diamond indentor) In such tests, it is usual for

coatings to crack, but without flaking from the substrate The

procedure should be worked out between the user and con-

tractor and designed so that an unsatisfactory coating is highlighted It may actually be necessary for the coater to deliberately produce coatings with a poor bond so that a relevant procedure can be developed It is also preferable for the test to bear some relation to the actual application so that, for instance, components subjected to bending in service would be given a bend-type adhesion test

9.8.6.6 Surface finish

If there is a need to preserve a particular surface finish, the quality control is a matter of a ‘before’ and ‘after’ texture measurement using a stylus profilometer In most cases, it will

be sufficient to determine the most common texture para- meter, R,, the ‘centre-line-average’ However, some coatings, even though they may generally replicate the underlying surface, have a subtle texture of their own This is true of some

of the evaporated physical vapour deposition coatings and also some electrolytic coatings In that case, it may also be necessary to monitor other surface finish parameters, particu- larly those relating to the ‘shape’ of the surface (skewness) and the sharpness of peaks and valleys (rms slope or average wavelength)

9.8.7 Closure

Surface engineering is the obvious solution to many of today’s engineering and wear problems It allows optimization of the surface and the substrate in a cost-effective way To be most effective, the principle of surface engineering must be adressed at the design stage of a component, with the designer making full use of available data on the coating or treatment properties, including friction and wear Then, having made a selection, the designer should be entitled to a good quality and reliable treatment service and it is the duty of the coating contractor to provide a certified product

The science of surface engineering is expanding quickly There are increasing possibilities of combining treatments and coatings to produce even more specialized properties Coat- ings might be ion-implanted, either after or during their deposition They might be laser-glazed to further increase hardness or diffusion One coating might be applied to another to produce a combination of properties (e.g abrasion resistance and corrosion protection) The time may come when expensive alloying elements can be added exclusively to the surface of a cheap component (perhaps by ion-plating) so that it can then be effectively nitrided Designers and en- gineers should be kept abreast of developments and be continuously aware of possibilities for improving component life, reliability and economics

Acknowledgements

The authors would like to thank Mr M Farrow of the Surface Science Division, Northern Research Laboratories, UKAEA, for his invaluable technical assistance in assembling this infor- mation

9.9 Fretting

9.9.1 Introduction

Tomlin~on’~ first investigated the phenomenon of €retting in

1927 after observations of red rusting of the grips of fatigue- testing machines He coined the term ‘fretting corrosion’, by which name it is commonly known, and carried out the first

Fretting 9189

quantitative s t ~ d y ’ ~ Ne considered that damage of the sur-

faces by fretting was initiated by mechanical wear produced by

sliding of one surface on another and that the corrosion

observed with base metals in air was a consequence of the

wear He stated that ‘although the presence of oxidation

products shows that chemical action accompanies fretting the

process is nevertheless certainly not one of corrosion as

ordinarily understood’

The distinction between fretting wear and ordinary wear is

that fretting generally occurs at contact surfaces that are

intended to be fixed in relation to one another but which

actually undergo minute alternating relative movement A

classical #example is the damage to the races of wheel bearings

of automobiles during shipment by rail or ship, initially

ascribed to brinelling caused by impact as a result of vibration

but now known to be caused by slip between ball and race

(and called ‘false brinelling’)

When fretting occurs on base metal in air, the wear debris

always consists of oxides of the metal so that the most common

symptom of fretting is the red-brown mud (comprising essen-

tialiy red iron oxide mixed with oil or grease) shown as a

patchwork over contacting steel surfaces

Frettin.g damage occurs when two loaded surfaces in contact

undergo relative oscillatory tangential movement (known as

‘slip’) as a result of vibration or stressing Amplitudes of

relative moment are small and often difficult to measure or

even to predict by analysis It is a deterioration process often

ignored or not understood by designers As a consequence,

many fretting problems often come to light only late in the

product development or, even worse, when the product i s out

in service In such cases it is often too late to allow anything

but minor redesign, and the only option then available is to

employ some type of palliative (which is rarely a Bong-lasting

solution), when fundamental changes in design concept can be

the only successful solution Possible situations in which

relative movement and hence fretting can occur, either inten-

tionally or otherwise, are legion More common ones are

flexible couplings, press fits of bearing raceways and hubs on

axles, riveted aircraft structures, screws in surgical implants,

steel ropes, heat exchangers business machines and electrical

contacts

Reports of the occurrence of fretting have increased over

the years, in part due to the increased demands placed on

materials by virtue of higher power densities and stresses in

modern machinery Fretting and fretting fatigue have become

increasing problems in the aerospace industry, particularly

with the use of exotic alloys based on titanium and aluminium

A number of excellent review papers are available in the

literature which rovide summaries of the state of knowledge

at various times!^'

Although fretting damage by itself is seldom sufficient to

cause failure directly, the irregularities produced in the surface

cause loss of dimensional accuracy and wear products can

cause excessive friction and seizure of closely fitting parts

A number of different terms have been used in the litera-

ture, including fretting, fretting wear, fretting corrosion, frett-

ing fatigue, false brinelling, rubbing fretting, impact fretting

and impact-slide fretting It is worth providing some defini-

tions to distinguish between these various terms Although

fretting was initially described as fretting corrosion, it is better

to use the term ‘fretting’ as a general title to cover a number of

aspects of the phenomenon In the first instance, fretting was

observed with materials such as low-carbon steels, which at

room temperature and in the presence of oxygen and water

vapour, produce copious amounts of finely divided red oxide

identified as haematite (Fe2Q3) and fretting was understood to

be closely associated with corrosion on the material surface

The oxidized particles are extremely hard and become em-

bedded in the contact surfaces producing a ied staining which even mild abrasive cleaning will not remove This staining can often be used to diagnose the occurrence of fretting It has often been stated that the oxidized debris was responsible for abrasive wear of the surfaces or even that it alleviated the damage by acting as a miniature ball bearing However, fretting is known to occur in non-oxidizing environments such

as high-purity inert gases where there is no oxidized debris and corrosion effects are minimal

It is generally accepted that fretting processes are caused by high-frequency relative movement of contacting surfaces even with slip amplitudes of less than 1 pm The upper transition between fretting and oscillatory wear is not well defined, but somewhere between 100 and 200 p m is the upper limit at which the wear process assumes the characteristics of unidirec- tional or reciprocating sliding wear

Fretting wear is best used to describe processes in which ultimate failure of the component occurs by loss of the material

surface leading to fracture, loss of function or pressure bound-

ary rupture (applies to heat exchanger tubes, pressure vessels) In such instances, the component geometry is such as

to allow escape of the wear debris with continuing penetration

of the component thickness This i s to be contrasted with the more common fretting processes in which the geometry traps the debris, restricts access of the environment and produces little loss of section Seizure of closely fitting parts such as bearing raceways, press fits, machine tool slideways and gear couplings is a characteristic mode of failure for fretting wear

Fretting fatigue is a consequence of fretting damage to components subjected to cyclic stressing in which the fretting scar acts as a fatigue crack initiator It usually occurs where fretting wear is minimal since either the production of copious amounts of wear debris interfere with the fretting process or the wear front progresses at such a rate so as to obliterate the propagating crack The elimination of excessive wear of a component can often lead to fretting fatigue problems, much

to the consternation of the machine designer, developer or operator A number of papers and publications provide re- views of this extensive topic.85,86

False brinelling is a specific term used to describe the damage to rolling element bearings particularly where the bearing has been subjected to vibration from neighbouring machinery during a period of inoperation The damage does resemble a brinelling indentation and can lead to rough running and premature fatigue failure during subsequent operation of the bearing.87

Although the classical forms of fretting involve loaded contacting surfaces, it is now recognized that fretting can occur between surfaces that undergo separation and repeated con- tact Thus, the term rubbing fretting should be used to differentiate between loaded non-separating surfaces and those which experience periodic impact and slide, for which the terms impact fretting or impact-slide fretting are used It should be noted that oblique collision of elastic bodies yields small-scale sliding under the Hertzian contact forces and milliseconds contact duration which can cause significant

fretting wear damage It is largely due to such occurrences of

impact fretting problems in nuclear power plant that this topic has received much attention in recent

9.9.2 Source of relative movement

9.9.2.1 Forcelstress exciied

Since it can be stated that wear cannot arise from surfaces which are not subject to relative slip and that the most successful method of overcoming fretting is to eliminate the fretting movement, it is worth examining how the relative

9/90 Tribology

movement between surfaces is generated An important dis-

tinction can be made between systems in which the relative

movement comes from an alternating force or stress applied to

elements of a machine, and ones in which the movement is

caused by defined displacements This distinction must be

made, as the nature of any palliative is critically dependent

upon the source of relative movement

In force- or stress-excited systems the degree of slip at a

surface is a non-linear function of the applied force or stress It

is influenced mainly by damping, which is a function of normal

load, static and dynamic friction coefficients and a number of

other parameters The success of a palliative such as a soft

interfacial layer, which changes not only the fretting wear

resistance but also the friction coefficient, cannot be guaran-

teed if perhaps the vibration amplitudes increase as a result of

the use of such a layer

Forces or stresses can arise from a large number of sources,

including acoustic noise, eddy or vortex shedding (flow-

induced vibration), pressure pulsations, mechanically trans-

mitted vibrations, aerodynamic loading and electrical noise

As part of any solution to a fretting problem, consideration

must be given to the vibration source and, if possible, some

attempt made to reduce it

In many mechanical systems subject to force or stress

excitation, the fretting wear rate has a maximum at some level

of load between the contacting surfaces (Figure 9.77) For

low-contact forces, the wear rate is low because the surfaces

are lightly loaded Suppose one decided that an increase in

load was necessary to attempt to reduce the fretting damage

It is possible for the wear rate to increase as the increased load

has little effect upon the amplitude of slip However, as one

passes the maximum, a further increase in load results in a

reduction in wear rate as the increased load damps out

vibrations and significantly reduces the relative slip The

effectiveness of changing load in this type of system critically

depends therefore on where the machine is currently operat-

ing on this curve and in part explains the contradictory results

obtained with some palliatives Force- or stress-excited

systems probably comprise 90% of the systems in which

fretting is observed and in general, the movements arc not

intentional

9.9.2.2 Amplitude driven

There are some mechanical components which are subjected

to a fixed amplitude of movement irrespective of loading (for example, misaligned gear couplings where the amplitude of movement is related to the degree of angular misalignment and not to the transmitted torque) Other examples are in business machines such as impact printers where the print head is designed to move through a defined travel A charac- teristic of these systems is that movements are generally intentional and little can be done to reduce or limit the relative slip at the surface without affecting the function of the machine

Where debris can escape, loss of fit can result, with continu- ing and perhaps accelerating fretting wear (particularly if slip amplitudes increase as loads reduce and wear progresses) The presence of oxide can affect the performance of electrical contacts or create safety hazards such as the pyrophoric oxide produced by the fretting of stacked aluminium ammunition boxes subject to vibration in ships' holds

Under some circumstances the production of debris can lead

to a significant reduction in fretting wear rate if, for example, the oxide can be compacted into a wear-resistant layer, such as occurs in steels used in high-temperature carbon dioxide for nuclear reactor heat exchangers; while in other situations, corrosion is accelerated by the continual removal of oxide (e.g the passivating chromium oxide on austenitic stainless steels), particularly in aqueous environments

c

5.9.3.2 Nature of debris

On steel surfaces fretted in air at temperatures less than

150°C the debris is largely hexagonal alpha-Fe203, a reddish- coloured oxide which is nonmagnetic and often contains metallic iron On aluminium the debris is black alumina containing about 25% metallic aluminium There is increasing evidence to suggest that the fretting debris is platelike in nature and these observations have led to proposed theories of

a delamination process along the lines suggested by S U ~ ~ ~ In

some instances s herical particles have been observed in the

to 500"C.96 fretting of silver 9 P and mild steel in argon at room temperature

I

Contact load

Figure 9.77 Variation of wear rate with load for force-induced

vibration systems

9.5.3.3 Mechanism of fretting wear

It was originally thought that three processes were involved in fretting:

Fretting 9/92

'specific wear rate', which relates the volume of material worn

to the load and total sliding distance The specific wear rate, k ,

can be calculated as follows:

1 The production of loose metallic particles by local welding

and adhesion which were subsequently oxidized;

2 Abrasion of the surfaces by hard oxide debris;

3 Continual scraping off and regrowth of oxide films

On this basis, an accelerating wear process would be expected

as more debris is produced, which is contrary to most observa-

tions Indeed, debris may have a beneficial effect As theories

of fretting have developed a review by Hurricks" considered

three stages to be involved:

I Adhesion and metal transfer;

2 Production of wear debris by mechanicochemical action;

3 Steady-state production of debris by fatigue action rather

and this is now largely accepted to be the most plausible

explanation of fretting processes Adhesion and welding in the

early stages, with steel and noble metals causing material to be

rzised above the level of the original surface, followed by

smearing and, finally, metal removal by a delamination pro-

cess Suh'sY4 delamination theory of wear envisages the coale-

scence of subsurface voids produced by dislocation pile-ups at

obstacles such as inclusions under the action of the alternating

shear stress It is possible that slip amplitude effects (discussed

later) may be explained by a critical amplitude above which

dislocations do not return to their original position, thus

accelerating fatigue crack initiation and growth Delamination

theory explains the platelike appearance of the debris and

subsurface cracks have been observed in sections through

fretted surfaces Observations made by Sproles and Du-

quette ,97 Jahanmir98 and Waterhousey9 also suggest that a

form of delamination wear also takes place in fretting The

delamination model proposes that fatigue cracks nucleate

beneath the surface The cracks then propagate parallel to the

surface, until instability or a material flaw forces the crack to

the surface to produce a flake wear particle Sproles and

Duquette describe how delamination occurs in multiple layers

over the fretting contact, resulting in a flaky metallic scale-like

covering lover the surface The observations were made under

fretting fatigue conditions, Le in the presence of a bulk

alternating stress; the surface stress level was high

The current theory therefore suggests that initial damage

arises from adhesion and welding at points of real contact,

resulting in material being raised above the original surface

The extent and severity of this stage depends upon the

reactivity of the metal and the corrosivity of the environment

Raised material is then smeared out and the surface is

removed by a delaminatior, process to produce platelike

debris which are essentially metallic but covered with oxide

Comminution of debris by grinding between the surfaces may

take place with further oxidation as the particles are reduced

in size Complete oxidation of very fine debris may ultimately

occur The compaction of debris in the contact zone allows

continued transmission of the alternating shear stresses Work

hardening and work softening can have adverse effects upon

the fatigue properties and accelerate the process

For cyclically stressed surfaces, propagating fatigue cracks

may be initiated in the early stages from boundaries between

slip and no-slip regions Although wear rates in the initial

stage are comparable with adhesive wear, in most cases the

wear rate falls significantly as steady-state conditions are

established For further discussions on fretting theories, see

The normal units for specific wear rate are m3N-Im-' It is

useful to use a similar concept for fretting in order to express the specific wear rate as a function of the normal load, frequency and slip amplitude The relationship becomes:

Volume of material removed

k = 2 X load x frequency X cycles X p-t-p slip (9.8) where the slip amplitude is the peak-to-peak value The specific wear rate can therefore be used to express results of tests to evaluate fretting wear, from which calculations of wear behaviour under different conditions of time, load and slip amplitude may be made

9.9.4 Parameters influencing fretting

There are a significant number of parameters which can affect the fretting damage sustained by a sliding interface Among these are:

e Time or number of cycles

0 Normal load

e Slip frequency and surface velocity

0 Contact geometry Materials properties (hardness, work hardening)

0 Environment (humidity, temperature, medium, oxygen Slip amplitude

potential) Lubrication

9.9.4.1 Time or number of cycles

Figure 9.78 presents a schematic representation of the pro-

gression of wear with time or number of cycles An initial

9.9.3.4 Concept of specific wear rate

Archard'" has expressed the wear behaviour of materials

ucder unidirectional or reciprocating sliding with the term

Number of cycles Figure 9.78 Schematic representation of debris production as a

9/92 Tribology

high wear is followed by a decrease in the volumetric removal

rate As oxidized debris is formed, the pattern for mild steel

falls to that of curves B or D, with the latter typical of low slip

amplitude Curves A or C typify soft materials with a hard

abrasive oxide which can dominate the material wear be-

haviour in the second stages of wear

When expressed in terms of change in specific wear rate

with time, the wear rate shows a decrease with time (Figure

9.79) for mild steel fretted in air at 20°C using the results of

Ming-Feng and Uhlig,'02 Ohmae and T s ~ k i z o e ' ~ ~ and Ming-

Feng and Rightmire.ln6 The specific wear rate reduces by

more than two orders of magnitude when tests are taken from

a few hours to a few hundred hours This demonstrates the

necessity to extend any laboratory to realistic timescales in

order to reach an equilibrium value of specific wear rate

9.9.4.2 Slip amplitude

There is general agreement that above an amplitude of

100 pm the volume of material removed is directly propor-

tional to the slip amplitude In other words, the wear volume

is directly proportional to the load and the sliding distance and

specific wear rates correspond typically to those of sliding

wear At low slip amplitudes, the evidence suggests that

damage is much lower Some investigators claim that there is

no measurable damage below 100 p m and elastic movement

takes up all the displacement betwen the surfaces Stowers and

R a b i n o w i c ~ ' ~ ~ claim difficulty in measuring the actual move-

ment and comment upon the possibilities of lost motion in the

test apparatus However, Tomlinson'* found damage down to

2 nm, and data on mild steel at 20"C, compared by Lewis and

Didsbury'"' to other ~ ~ r k ehave shown an increase r ~ ~ ~ ~ ~ ~ ~ ~ , ~ ~ ~

in specific wear rate of over two orders of magnitude over the slip amplitude range 50-100 p m (Figure 9.80) This observa- tion has an important effect in any attempt to cure a fretting problem since a twofold reduction of slip amplitude from 100

to 50 pm can lead to over a hundredfold reduction in wear volume

The occurrence of fretting damage even at very low levels of slip means therefore that any machine element at which relative movement can occur will see some level of damage A lower limit of damage of 0.75 pm has been suggested by Kennedy et a1."' so that it very difficult to reduce slip to a

level at which no damage will occur The exact threshold is, to some extent, dependent upon the properties of the materials

in use Microslip can occur in many supposedly fixed or rigid connections (e.g press or shrink fits of hubs or bearing raceways) At very low levels of slip, fretting fatigue becomes

a potential problem at the boundary between slip and no-slip regions

It should be noted that slip amplitude effects are not necessarily a feature of all materials and are dependent upon the environmental conditions Slip amplitude effects for mild steel disappear with increasing temperature, which is probably more a reflection of the increasing influence of protective oxides (to be discussed later)

9.9.4.3 Normal load and surface pressure

Most worker^'^".'^^ report a constant specific wear rate as a function of load, providing the movement is forced and takes place over the entire contact area (Figure 9.81) If, however,

+ Ohmae and Tsukizoe

Test duration (cycles) Effect of test duration on specific wear rate for mild steel under rubbing fretting conditions

71 Lewis (1978)

X Feng and Uhlil + Ohmae and Tsukizoe

Tribology

the increased load results in a decrease in the slip amplitude

then the rate at which fretting occurs can be reduced A

reduction in contact area to increase the pressure has often

been used to overcome fretting problems, but one should be

careful that the increased pressure does not exacerbate frett-

ing fatigue problems since the increased pressure increases the

stress concentration at the slip boundary

9.9.4.4 Slip frequency and surface velocity

Slip frequency may affect the rate of fretting if the process is

removal and regrowth of oxide If the oxide grows according

to a logarithmic law, then frequency effects are probably

negligible above 17 Hz, because only that portion of the

oxidation curve at short exposure times is in effect

Figure 9.82 shows results from Ming-Feng and Uhlig"' for

mild steel fretted in air over the frequency range 1-60 Hz

There is little influence of frequency indicated, while Soder-

berg et al."' suggest little effect at high slip amplitudes for a

frequency range of 10-1000 Hz and recommend frequency as

a useful parameter for accelerating fretting wear tests

It is interesting to note that greater surface damage has been

reported for very low frequencies (in which surface velocity is

typically 1 mm/day) and oxide growth and removal effects

dominate This has been particularly important in electrical

contacts which can be subject to low slip at very low frequency

Feng and Uhlig

X 2 3 0 p 67800cycles

f 9Op 457800cycles

0 S o p 67800cycles

0 l o p 67800cycles

9.9.4.5 Hardness and surface finish

Observations would generally suggest that increased hardness gives lower fretting damage It is believed that combinations

of similar metals of differing hardness give good fretting resistance, though wear in general is thought to be alleviated

by combinations of materials which have low mutual alloying tendencies This view is partially supported by Sakman and Rightmire,'"' who produced a ranking of fretting wear res- istance (given in Table 9.23)

Opinions on the effect of surface finish are quite contradic- tory in the literature Some investigators believe that a smooth

Table 9.23 Fretting resistance of various materials in dry air

Low

Steel on steel Nickel on steel Aluminium on steel AI-Si alloy on steel

Sb plate on steel Tin on steel

Frequency (Hz)

Figure 9.82 Effect of frequency on specific wear rate for mild steel under rubbing fretting conditions: results of Feng and Uhlig

Fretting 9/95 surface is more susceptible to fretting damage while others

consider that super-finishing can prevent excessive damage

The weight of evidence points to the superiority of a rough

surface for lubricated contacts the surface irregularities pro-

viding a reservoir for lubricant

9.9.4.6 Contact geometry

The geometry of contact is an important variable in fretting In

general, large-area nominally flat or conforming contacts tend

to trap the debris produced and give an over-deepened pitted

surface during fretting Conforming contacts are poor from

the point of view of lubricant access since the sliding move-

ment is not sufficient to entrain lubricant Some surface

treatments which are used to alleviate fretting problems are

most effective with conforming contacts because of their low

thickness Conforming contacts are most prone to seizure In

contrast, non-conforming contacts allow debris to escape

9.9.4.7 Environment

In an inert atmosphere, such as nitrogen or vacuum, the

fretting wear of steel reduces but the incidence of material

transfer 'between the surfaces increases The initial phase of

adhesion and welding is preserved rather than the transition to

oxidative wear appearing Little loose wear debris tends to be

produced Humidity also reduces the amount of fretting

damage as the hydrated oxides of iron are less hard than their

dehydrated counterparts lo'

Since fretting involves both the mechanical properties of a

material and its reactivity with the environment, it is to be

expected that changes in temperature would have consider-

able effect on fretting wear The frettin behaviour of mild

steel has been investigated by H ~ r r i c k s ' ~ ~ ~ " ~ and his results

are given in Figure 9.83 A significant reduction in specific

wear rate is observed at 15@200"C due to the greater thick-

ness and adherence of the oxide magnetite Fe304) formed

This behaviour has also been found by Lewis'" with mild steel

in air u~p to 500"C, in which the specific wear rate for

high-amplitude (100 p n ) fretting shows a considerable reduc-

tion with increasing temperature with a lesser reduction for

low slip (10 pm)

I: would appear that the formation of a compacted oxide

layer on mild steel increasingly reduces the fretting wear rate

by providing a protective wear-resistant film These so-called

'glazes' are observed at high temperatures on corrosion-

resistant materials such as austenitic stainless steels and nickel-

based alioys similar to those formed in reciprocating slid-

In aquieous environments a passivating film is relied upon to

limit corrosion of the metal surface Disruption of the film by

fretting can produce large changes in electrode potential for

the baser metals Fretting can therefore show its effects more

by continuous disruption of the passivating film Experiments

in which the potential of the fretting surfaces is kept constant

show a linear relationship between corrosion current and slip

amplitude Calculations show that the bulk of the material

removed is the result of mechanical action rather than chem-

ical d i ~ s o l u t i o n ~ ~

Corrosion reactions can be controlled by imposing a catho-

dic potential on the system (or on metals which display

passivation, an anodic potential) It has been observed that

cathodic protection can @ve a significant improvement in

fretting fatigue behaviour

ing.'15,"6

Rubbing fretting of mild steel in air

Hurricks results at 750 p amplitude

as well as thickening the oil or grease by mixing with it (e.g the thick red 'cocoa' observed in many lubricated fretted contacts) Shear-susceptible greases (Le those whose viscosity falls with shearing) appear to be the most effective and the literature generally recommends greases of low base viscosity

to improve the flow into the fretting zone The selection of greases for lock coil steel ropes is quite different, however, with high viscosity being most suitable

Lubricants containing molybdenum disulfide have frequently have recommended for the alleviation of fretting However, in line with other work for unidirectional sliding, the effective- ness of MoSz is much reduced in the presence of a liquid because it cannot effectively attach to the metal surface when there are other surface-active agents competing (e.g anti- wear additives) Considerable promise, however, has been shown in dry-bonded MoSz films for the control of fretting."'

9.9.5 Theoretical considerations

A classic example of microslip in fretting contacts is given in the elastic contact of two spheres or a sphere on flat where the surfaces are subject to an alternating tangential force Mind- lin118 showed that for a sphere pressed against a plane by a normal force, N , and subjected to a tangential force, T , the shear traction is unbounded towards the edge of the contact (see Figure 9.84) In reality this distribution is not possible

9/96 Tribology

Normal force I

Figure 9.84 Pressure and shear stress distributions at elastic contact of two spheres

and a relief mechanism must operate This mechanism is

provided by micro-slip, which occurs over an annular area at

the edge of the contact As the tangential force is increased,

the annular area of micro-slip grows until:

T = p N

where p i s the coefficient of friction At this point the whole of

the two surfaces are in relative motion, i.e gross sliding takes

place

Mindlin's analysis illustrates how the localized concentra-

tion of shear traction gives rise to fretting Elimination of the

shear concentration is therefore a fundamental means of

avoiding fretting Furthermore, the designer has the potential

to predict fretting early in the design process By carrying out

a stress analysis of the contact interface, areas of potential

fretting can be identified where the following inequality holds:

9- ' Pull

where T is the shear stress at the surface and vn is the in-plane

stress normal to the surface

The shear traction also has an important effect on the

generated contact stress field which is now considered impor-

tant in the mechanism of fretting fatigue Without a shear

traction (zero friction coefficient) only a small tensile stress is

generated at the edge of a hemispherical contact In such

cases, failure is likely beneath the surface where the material is

subjected to the maximum yield stress As the friction coeffi-

cient increases the point of maximum yield stress moves

towards the surface, reaching it at a friction coefficient of 0.33

In the fretting of steels, friction coefficients approaching

unity are not uncommon Figure 9.85 illustrates the rapid

increase in the tensile stress at the trailing edge of contact as

the friction coefficient is increased Figure 9.86 shows the

damage developed between a steel sphere and a flat as a result

Fretting

+

( 3 ) - = 0.40

PS

9/98 Tribology

of fretting under increasing values of tangential force.”’

Figure 9.86(1) shows a very slight outer ring for a value of

$/ps = 0.16 where $ = T/N (the shear coefficient) and

pLs = static friction coefficient The damage extends inwards

as the area of microslip grows until the entire surface is

subjected to gross slip when the value of $/pS reaches 1 (Figure

9.86(2))

It is at the boundary between the regions of slip and no-slip

that fretting fatigue cracks can initiate and grow, though

fretting wear in the outer area of slip causes a redistribution of

the contact pressure and often fatigue effects are halted as the

wear zone progresses inwards

9.9.6 Fretting wear evaluation

There are a plethora of testing facilities described in the

literature, each developed to study a particular aspect of

fretting or fretting fatigue Confusion over the effects of slip

amplitude has arisen in the past, largely because some inves-

tigators have ignored or underestimated the lost motion that

inevitably occurs Modern fretting apparatus are generally

designed with very stiff drive mechanisms in order to ensure

that relative slip is occurring at the contact interface and often

take great care to measure the actual slip close to the

interface

There are no standard fretting test machines, each industry

developing a design to suit its needs There is no standard test

geometry, with test specimen designs varying from hemisphere-

on-flat to crossed-cylinders, cylinder-on-flat and flat-on-flat

In many instances, actual component geometries have been

used The particular difficulties in fretting machine design to

ensure area contact with flat-on-flat specimens lead to the use

of hydrostatic bearings to allow a self-aligning feature A

number of organizations have attempted to create a standard

for fretting evaluation, including the American Society for the

Testing of Materials.’”

9.9.7 Preventative measures - some palliatives

9.9.7.1 Some lines of attack

There are a number of methods which may be used to reduce

or eliminate fretting damage:

0 Prevent the relative movement by a change in design or

modifying the source of vibration

0 Provide a treatment to the surfaces to reduce or eliminate

welding and adhesion processes, remembering that all

materials are subject to fretting to a greater or lesser extent

0 Encourage movement to allow access of lubricants, usually

a difficult line of attack in view of the high contact press-

ures, the generation of wear debris and low relative velocity

of sliding

A number of observations can also be made which can assist in

solving a fretting problem:

No wear takes place between surfaces subject to vibration

unless its amplitude is sufficient to produce slip

0 Damage is produced on any solid surface, whether metallic

or non-metallic, if slip takes place

Fretting wear of base metals in the absence of oxygen gives

less rapid accumulation of debris and less intense damage

than in air or oxidizing atmospheres

No liquid lubricant entirely prevents fretting but some can

produce a considerable improvement

Additionally, one must make the distinction between the

fretting of components (such as shrink fits, press fits, bolted

flanges, etc.) which are not intended to undergo relative

motion and those (such as flexible couplings, bearings, uni- versal joints) which undergo intentional oscillatory movement

9.9.7.2 Prevention of gross slip

In order to overcome fretting damage in components not intended to slip, gross slip can be reduced or eliminated by:

0 Increasing the load or closeness of fit;

0 Increasing the coefficient of friction (e.g with grit blasting

0 Interposing a layer having high elastic strain limit (e.g followed by lead plating”’);

rubber or polymers)

9.9.7.3 Design analysis to prevent or reduce fretting damage

Fundamentally, the best solution to a fretting problem is to

‘design it out’ if at all possible by using all-welded or unit construction Design modifications may be necessary to reduce or eliminate the source of vibration (e.g isolation from local excitation sources) Bearing in mind that slip occurs at an interface because the shear force is greater than the opposing friction force, one can reduce the shear stress or increase the friction force This reduces to the problem of eliminating the concentration of shear stress over the surface The friction force may be increased by reducing the apparent area of contact while keeping the load constant, thus increasing the pressure Similarly, some platings such as cadmium can increase the friction coefficient by promoting seizure

In situations where the concentration of shear traction occurs at the edge of the contact, a simple design modification may be all that is required For example, on a clamped joint, undercutting at the edge is an effective means of reducing the shear concentration (Figure 9.87) Another simple example is

a press-fitted hub on a shaft in which stepping down the shaft diameter significantly reduces the shear concentration (Figure 9.88) Unfortunately, simple geometries such as these are the exception In the majority of real engineering assemblies while geometric modifications are usually possible, it is not often clear how effective the modifications will be in eliminat- ing the shear concentration In these cases the answer is to carry out investigations of the joint by a numerical stress analysis technique such as finite-element analysis

Several commercially available finite element codes have special elements for modelling Coulomb friction which will provide detailed information about the behaviour of a poten- tial fretting interface However, since sliding is a non-linear element in the problem, these programs are usually expensive

in computing time Fortunately, in the majority of cases the main interest is in predicting whether fretting will occur or not This simplifies the analysis and allows the interface to be

Figure 9.87 Undercutting at clamped joints to avoid stress concentration

Fretting 9/99

For a displacement controlled problem the approach to alleviate fretting wear or fretting fatigue is to try to reduce the normal load and the friction coefficient With fretting fatigue this palliative is clearly aimed at reducing the contact stress levels In a fretting wear problem the palliative should also be effective since it has been found that the wear volume is proportional to the applied load.104

In a force controlled fretting wear probiem the require- ments of a palliative are completely the reverse Increasing normal load and friction coefficient can act as effective palliatives by lowering slip Slip has been shown to be one of the most im ortant fretting parameters Work by Ohmae and Didsbury’08 all suggest that the specific wear rate decreases with decreasing slip amplitude Figure 9.80 summarizes the results of these investigations into the effects of slip amplitude The data from the individual experiments suggest that the specific wear rate (volume lost per unit load per unit sliding distance) varies as a function of the slip amplitude raised to a power ranging between 2 and 4 Clearly, if fretting cannot be

entirely prevented even a modest reduction in slip may be sufficient to reduce the level of wear down to acceptable proportions

In the alleviation of force controlled fretting fatigue, reduc- ing the shear traction is the primary objective Thus, reducing the friction coefficient and normal load are desirable palliat- ives providing the increase in wear can be tolerated

Tsukizoe,lo PHall~day,”~ . Halliday and Hirst’” and Lewis and

Figure 9.88 Stepped shaft to avoid stress concentration

modelled by simpler linear elastic elements Provided a suffi-

cient refinement of elements is used in modelling the joint, the

design can be established as satisfactory, if the surface shear

stress is less than the product of contact pressure and friction

coefficient (Le 7 < pun) everywhere across the joint If the

condition fails at any element node, then the extent of fretting

cannot be predicted, and the design should be considered as

prone to fretting

In certain cases, optimizing the design of an assembly on the

basis of other essential design criteria may mean that fretting

becomes inevitable For this situation, a palliative may be the

only option available to minimize the fretting damage In such

instances a detailed analysis of the fretting interface can be

useful to categorize the type of fretting problem This, in turn,

can assist with the selection of a suitable palliative

9.9.7.4 Selection of frettirzg palliatives

To aid with the selection of a palliative, fretting problems can

be classified into one of two categories determined by whether

the movement is force controlled or amplitude controlled To

explain these categories it is useful to consider again the

Mindlin roblem In the partial slip regime according to

Johnson,‘2 the ratio T/pN determines the level of micro-slip

The slip amplitude is therefore a function of the forces

involved, and it is proposed that the problem can defined in

the category ‘force controlled’ Under gross sliding, i.e

T = p N , clearly the forces involved play no part in determin-

ing the slip amplitude Fretting in this case is put in the

category ‘displacement controlled’ since the slip amplitude is

essentiallly fixed

In readity, defining in which category the problem falls is

not, in the majority of cases straightforward Fretting involv-

ing wear between components which are intended to move

(e.g oscillating bearings, ball joints, slideways, etc.) is, in

general, a displacement controlled problem and there is little

ambiguity However, in clamped assemblies where movement

occurs unintentionally, fretting alllows strains of different

magnitudes in the component mating surfaces If this ‘differ-

ential strain’ is largely unaffected by the applied loads, then

the problem effectively is one of displacement controlled

fretting If, however, the differential strain is substantially

modified by the applied loads then the problem should be

classed as one of force controlled fretting A finite-element

znalysis can be used to determine in which category the

Lubricants can obviously produce a large decrease in the coefficient of friction In force controlled situations this effect

is likely to result in unsatisfactorily high slip levels Also, with clamped assemblies the load transmission across the joint may

be modified and fasteners may become overloaded However, lubricants can prove very useful in displacement controlled fretting fatigue Application of a lubricant will Iower the level

of shear traction which will considerably reduce the contact stress level Also, in displacement controlled fretting wear, boundary lubrication can significantly improve performance However, unless continuously replenished, the self-cleaning action of fretting very quickly removes the boundary- lubricating film Gaining access to the inner regions of the fretting contact without separating the surfaces is difficult for

oils of high viscosity and virtually impossible with greases A

thin penetrating oil is therefore the best choice Vapour blasting or phosphating may also assist in providing a reservoir for the lubricant

9.9.7.6 Solid lubricants

Often the difficulty of containment precludes the use of an oii and an alternative in this situation is a solid lubricant such as molybdenum disulfide or zinc oxide Solid lubricants aie very effective at reducing fretting damage over a limited life.”’ If the problem involves high numbers of fretting cycles

(>5 X lo’), then solid lubricants such as described are un- likely to be satisfactory without re-application

Tribology

Phosphate treatment prior to application of the MoS2

coating can produce a significant improvement in fretting wear

resistance

9.9.7.7 Thermochemical treatments

These treatments involve the diffusion of carbon, nitrogen

and, less usually, chromium or boron The most common

thermochemical treatments are carburizing, carbonitriding,

nitrocarburizing and gas nitriding The factors involved in the

selection of a thermochemical treatment are too numerous to

go into in detail here, but it is possible to make generalizations

about this form of palliative

This category of treatments is designed to increase surface

hardness In general, the adhesive and abrasive resistance

improve with increased surface hardness The most effective

treatment against fretting wear is therefore likely to be the one

producing the largest increase in hardness However, the

effect of the treatment on the friction coefficient must not be

ignored For instance, on En32 steel, liquid nitrocarburizing

using the Cassel Sulfinuz process reduced the friction coeffi-

cient in slow linear sliding from 0.8 down to O.2.l" Clearly, for

displacement controlled fretting, reduced frictional traction

will enhance the effectiveness of the palliative However, in a

force controlled situation the reduction in friction may negate

the benefits

By virtue of the compressive stresses developed in the outer

surface layers the fatigue strength of steel is normally en-

hanced by thermochemical treatments As a general principle,

any process which increases the normal fatigue strength of the

steel, providing it is not accompanied by a significant increase

in friction coefficient, will improve the fretting fatigue perfor-

mance The process giving rise to the greatest increase in

normal fatigue strength is likely to provide the largest im-

provement in fretting performance

9.9.7.8 Surface coatings

In previous investigations of palliatives, both hard and soft

coatings have been employed to mitigate both fretting wear

and fretting fatigue However, it is probably true to say that

there are more conflicting opinions in the literature on the

performance of individual coatings than for any other class of

palliative The reasons for the conflicting opinions on coating

performance are not always clear However, sometimes a

successful coating performance can be dependent on the

characteristics of the test rig

For example, in a rig producing low-amplitude force con-

trolled fretting, such as that described by B ~ d i n s k i , ' ~ ~ coatings

having high coefficients of adhesion may bring about seizure

of the contacts A palliative which is given a favourable

assessment by such a rig may in fact only be effective in force

controlled situations where seizure can be brought about

Sikorskiilz6 has correlated high coefficients of adhesion with

soft coatings such as silver, indium and lead It is perhaps

therefore no surprise to find that these are precisely the

coatings which are said to be beneficial in fretting.lZ7 It has

also been suggested that soft metal coatings work by absorbing

the fretting movement.84 However, the amount of movement

which can be absorbed is likely to be so low that it does not

seem to the author a credible explanation

With thin soft coatings applied on a hard substrate, a good

palliative performance may be the result of a low coefficient of

friction which can occur Halling128 has proposed an explana-

tion for this low-friction phenomenon In general, however,

the poor durability of soft coatings leads the author to believe

that, with one or two special exceptions, soft coatings are

likely to be of little benefit as a fretting palliative

With electroplated chromium, Alyab'ev et found that the conditions of deposition and the coating thickness can influence the wear resistance In this work it was also found that the chromium deposition with the highest hardness had the best fretting wear resistance However, under fretting fatigue the high levels of residual stress present in the coating reduce the normal fatigue performance considerably; and the coating has little effect on the friction coefficient Therefore, although improvements in the fretting fatigue performance have been claimed for chromium plating,13' there is no reason why under dry conditions this treatment should be of any significant benefit in fretting fatigue

In general, because of the adverse effect on the normal fatigue limit, it is not considered that hard coatings are effective palliatives for fretting fatigue unless the coating significantly reduces the friction coefficient Electroless nickel impregnated with PTFE can reduce friction and therefore has the potential to give improved fretting fatigue performance While improved wear performance using electroless nickel has been reported by Gould et there is no reported work to test the effectiveness on fretting fatigue of a nickel coating impregnated with PTFE

Hard coatings deposited by spraying or welding can give high wear-resistant surfaces There are numerous coatings commercially available, and deposition can be by several methods Little is known about the effectiveness of the majority of these coatings as palliatives in fretting wear situations However, in work carried out at the National Centre of Tribology (NCT), various coatings applied by the Union Carbide detonation-gun process based upon bonded carbides have been tested At temperatures below 200°C in a carbon dioxide environment, the application of these coatings

to mild steel or stainless steel reduced the wear rate by up to

an order of magnitude

9.9.7.9 Polymeric materials

Replacing one of the fretting surfaces with a polymer material can be effective in reducing fretting damage to a steel surface However, the choice of polymer is all-important This is because hard abrasive oxide debris produced from a steel surface can become embedded in the polymer The polymer then becomes an effective carrier of the abrasive debris and the wear process becomes two-body abrasion One of the best polymer materials to use in a fretting wear problem is PTFE This is probably because PTFE works by forming a transfer film on the steel counterface Any abrasive debris is then encapsulated by the PTFE transfer film

Unfortunately, PTFE is a rather soft polymer that will move under load, and it is therefore not generally suitable for the high-load situations of fretting fatigue However, in commer- cial bearings additional support has been achieved by incorpo- rating PTFE into a sintered phosphor-bronze matrix Tests at NCT have also shown the benefits to be gained by the use of PTFE-based materials (see Table 9.24) The fretting wear of mild steel was considerably reduced by the use of a fabric-based PTFE weave counterface, the steel exhibiting an almost imperceptible level of damage and wear, while wear of the polymer was also low, Clearly, consideration should be given to these types of material for fretting problems in flat-on-flat geometries

PTFE can also be constrained and supported by incorporat- ing it into a hard electroless nickel coating The fretting resistance of the hard nickel is enhanced and certainly for low-c cle fretting wear this coating is a very effective palliat- the low friction, therefore the palliative may not be effective in force controlled situations Also, the reduction in fatigue ive.13 P In fretting fatigue the benefits are likely to come from

Fretting 9/101

9.9.7.12 Surface cold working by shotpeening

Shot peening is one of a number of treatments which can be used to produce surface cold working This type of treatment increases the hardness and induces a compressive stress in the surface layers Both of these features, as previously explained, can be advantageous in reducing fretting damage

In addition, shot peening produces surface roughening which may well have benefit in both fretting fatigue and fretting wear problems It has been suggested that rough surfaces are more resistant to fsetting fatigue.137 This may be because a rough surface has, to a limited extent, the same de-stressing effect as the notches described previously

A shot-peened surface will also retain lubricant better, each indentation acting like a small oil reservoir Therefore, in both displacement controlled fretting fatigue and wear, shot peen- ing may be expected to enhance the effectiveness of an oil lubricant

Table 9.24 Rubbing fretting wear of various materials versus mild

steel at 125 pm slip amplitude

Counterface Specific wear rate

Woven PTFE fibre/glass fibre + 6 8

PTFE flockiresinlNomex cloth 4-5

resin

For additional information on materials and manufacturers see reference 132

strengtlh associated with the hard nickel plating may be

detrimental to the fretting fatigue performance

9.9.7.10 Interfacial layer

It has been shown that micro-slip can be prevented when a

thin layer of flexible material (for example, rubber or Terylene)

is interposed between the fretting surfaces.133 With a iayer of

the right compliance and thickness, the shear stress concentra-

tion at the edge of the contact giving rise to the slip can be

eliminated from flat contacting surfaces With Hertzian con-

tacts because the contact pressure goes to zero a1 the edge, to

achieve the same results it is necessary to bond the layer to

both the surfaces

Althlough this palliative can be effective at preventing

relative sliding there are design limitations to its applicability

Clearly, the palliative can, in practice, only be used with

relatively small slip amplitudes, therefore it is not likely to be

suitable in large-amplitude displacement controlled situations

The increase in compliance produced by the palliative may

also be unacceptable to the overall performance of the design

The durability of the interfacial layer under the alternating

loading may also be a limiting factor

Notwithstanding these limitations, the ability of interfacial

layers to eliminate the shear concentration can prove very

effective in many frettin fatigue problems involving bolted or

interfacial layers for use in the aircraft industry and found the

palliative to be a success

riveted joints Sandifer' 9 has carried out practical testing of

9.9.7.11 De-stressing notches

This has been suggested as a palliative only to fretting fatigue

problems This palliative was investigated by Kreitner,'35 who

tested a flat fatigue specimen machined with closely spaced

de-stressing notches (0.4 mm deep) running laterally and

longitudinally over the contact area The tests showed that

with the de-stressing notches fretting was prevented from

influencing the fatigue strength of the specimen Moreover,

although there was a reduction in the fatigue limit caused by

the notches, the improvement in fretting fatigue strength was

still in excess of 100% The effectiveness of this palliative has

been c'onfirmed by Bramhall who also obtained a signifi-

cant increase in fretting fatigue strength using de-stressing

grooves cut in a lateral direction only

The advantage of this palliative is that the fatigue strength is

governed purely by the notch effect of the de-stressing

notches; it is therefore more predictable Also, the palliative is

applicable to both displacement and force controlled problems

9.9.8 Summary of palliatives

9.9.8.1 Palliatives for fretting fatigue

Palliatives for fretting fatigue roblems have been reviewed by Chivers and G ~ r d e l i e r ~ ~ ' ~ ~ ~ ' ~ ~ in an attempt to provide a rational approach to their selection They observed that a useful starting point is provided by the work of Nishioka and HirakawalN in which an equation is developed for the initia- tion of fretting fatigue cracks in flat fatigue specimens which are fretted by cyclindrical fretting pads The fretting pads were orientated such that the direction of slip was orthogonal to the cylinder axes They derived the expression:

mfwl = uwl - 2pP0 i 1 - exp ( - - 91 (9.9)

where mfw, is the alternating stress necessary to initiate fretting

fatigue cracks, a,+,, the alternating stress necessary to initiate cracks in the absence of fretting, p the coefficient of friction,

Po the peak Hertzian stress, s the relative slip and k a constant depending on the material and surface condition This equa- tion relates to the condition of partial slip in which relative movement takes place over only part of the contact area

When the full slip condition is reached and s becomes large,

equation (9.9) reduces to:

Vfwl = a w l - W O

An improvement in fretting fatigue behaviour can therefore

be achieved by raising the fatigue strength, a,,, of the base

material, reducing the relative slip, s, either to achieve partial

slip or to reduce the degree of partial slip present, lowering the contact pressure such that Po is reduced or lowering the coefficient of friction p

Excluding a change in base material, the problem is that the remaining options are interactive since a reduction either in contact pressure or in friction coefficient may result in an increase in the slip amplitude An important consideration is what form the driving force to produce the movement takes, that is, whether conditions are controlled by displacement or force

With displacement control, the amplitude of movement will

be constant regardless of the force required Considering

equations (9.9) and (9.10), it is apparent that for either partial

or full slip conditions, and with constant slip amplitude, reductions in either friction coefficient or contact force should result in an improvement in fretting fatigue performance With force control, the peak oscillating force to produce the movement is fixed in magnitude In some circumstances therefore a modest increase in either contact force or friction

(9.10)

Tribology

coefficient could have a large effect in inhibiting the slip that

occurs If, however, these increases should fail to achieve a

partial slip condition, equation (9.10) shows that a deteriora-

tion in fretting fatigue performance would be expected and

hence the result of a modification to the conditions of contact

is dependent on the slip regime These results are summarized

in Table 9.25 Additionally, Chivers and Gordelier reviewed

possible mechanisms of alleviation and these are summarized

in Table 9.26

9.9.8.2 Palliatives for fretting wear under force excitation

A similar distinction based upon the source of excitation may

be made for fretting wear There are a wide range of possible

palliatives for control of fretting wear caused by force or stress

excitation and these are summarized in Table 9.27

9.9.8.3 Palliativesfor fretting wear under amplitude

excitation

Palliatives for control of fretting wear under amplitude-driven

movements are summarized in Table 9.28

9.10 Surface topography 9.10.1 Effects of surface topography

All machining processes leave characteristic topographic fea- tures on the surfaces of components These features vary in amplitude, spacing and shape and can exert a significant effect

on the component’s function or aesthetic appearance The British Standard, BS 1134: 1988,141 identifies two com- ponents of surface topography which are generated by most common machining processes and classified according to their cause:

1 Roughness: ‘The irregularities in the surface texture that are inherent in the production process but excluding wavi- ness and errors of form.’

2 Waviness: ‘That component of surface texture upon which roughness is superimposed This may result from machine

or work deflection, vibrations or release in machinery Three surface characteristics are illustrated by Figure 9.89

In addition to the above definitions a third surface shape component, that of ‘form’, is generally recognized as being the

Table 9.25 Changes in contact parameters to improve fretting fatigue behaviour

~

Crack growth rate

Depth wear rate < 1 Crack growth rate

Table 9.26 Possible mechanisms of alleviation for various palliatives

of fretting fatigue) Reduces amplitude of slip Movement taken up in elastic shear

of interfacial layer Prevention of adhesion Improved fretting wear resistance

(beware lower fretting fatigue limit) (beware lower fretting fatigue limit)

Improved fretting wear resistance

Improved fretting wear resistance

Table 9.28 Palliatives for fretting wear under amplitude excitation

Design changes

Reduce load

Reduce pressure

Increase surface area

Decrease and lubricate

Reduces wear rate Reduces wear rate Lower depth loss (beware decreased lubricant access) Increased amplitude of slip to allow better access of lubricant Prevention of adhesion Improved fretting wear resistance (acts as good lubricant reservoir)

Improved fretting wear resistance

Improved fretting wear resistance

( b ) ldaviness

- 1

Figure 9.89 Diagrammatic representation of a machined surface

illustrating the classification of topography by the separation of profile components

nominal shape of the component i.e the surface shape when roughness and waviness are neglected Normally, deviations from ideal form are referred to as ‘errors of form’ Figure 9.89 also shows that machining marks in surfaces may predomi- nantly fall along one direction This is frequently called the

‘lay’ of the surface

These definitions of surface features are arbitrary and, therefore, also ambiguous It is possible for surface features of entirely different scales, produced by different machining processes, to have the same classification The term ‘surface roughness’ is also frequently regarded to be synonomous with

‘surface texture’ and ‘surface tcpography’ and, as a result, could potentially be confused with the ‘roughness component’

of a surface structure Despite such difficulties, these terms are well established

The sizes of machining marks are usua!ly discussed in units

of millionths of a metre, i.e micrometres (abbreviation pm)

In mechanical engineering metrology, the size of features of interest generally ranges from a few hundred micrometres down to hundredths of a micrometre and ?he scale of these features is not easy to visualize Figure 9.90 attempts to place objects of this scale into perspective by relating their dimen- sions to physically important parameters of interest in the study of surface topography and the size of some items familiar from daily life

It appears that no surface can be regarded as being perfectly flat (Even though its imperfections may arise only as a result

of misalignment in the positions of atoms in its surface.) As a consequence, considerations relating to surface finish stretch across a broad band of subjects which relate to the functional performance and manufacture of engineering components The influence of surface topography on physical phenomena varies widely in scope and includes lubrication, electrical contact resistance, heat transfer, fluid flow, noise generation, the performance of optical components and the fidelity of thin-film coating processes The influence of surface topogra-

9/104 Tribology

Length (rnicrornetres)

100.0 + Thickness of loose leaf writing paper

+ Width of record groove

+ Thickness of cigarette paper

Length of yeast cell

- Thickness of a single strand of a

spiders web

c 1.0 + Thickness of metallic oxide films

I + - Range of wavelengths of visible light

4 Length of an oil molecule

L 0.1

Figure 9.90 List of typical dimensions to demonstrate the

perspective of features of significance in t h e surface topography

of engineering components

phy on manufacture is principally one in which 'qualitý is

traded off against cost and a number of these influences are

discussed below

9.10.1 I Liquid-film lubrication

Theoretical investigations into the influence of surface rough-

ness in lubrication has been a topic of considerable interest

over recent years Surface roughness is of importance because

fluid flow rate, the fundamental principle on which the

Reynolds equation is formulated, depends on the cube of film

thickness Flow rates will suffer perturbation as a result of

local changes in film thickness due to surface roughness

Several analyses of lubrication which involve surface rough-

ness have been described and the simplest of these studies

investigate the influence of longitudinal striated roughness on

hydrodynamic lubrication Michell'" examined the effect of

sinusoidal roughness in slider bearings while Dowson and

who mế^^ investigated the lubrication of cylindrical rollers

containing vee-shaped, square, and sinusoidal roughness

running along a circumferential direction A study which

examined the effects of non-dimensional roughness was con-

ducted by Christensen

The investigation of transverse roughness is a more difficult problem because an ađitional term arises in the Reynolds equation when the direction of motion of a surface does not coincide with the direction of its tangent D y ~ o n ' ~ ~ points out that some analyses of transverse roughness (ẹg Christensen and T ~ n d e r ' ~ ~ ) did not appreciate this and consequently their results are only applicable in situations where a plane surface moves over a stationary rough onẹ An accurate analysis of the effects of transverse roughness was given in a study of the lubrication of rough disks by Berthe and Godet.I4'

The inclusion of surface roughness effects in the solution of the Reynolds equation in EHL problems is uncommon, as the situation is very complicated even in the treatment of ideally smooth surfaces However, a perturbation analysis, applicable

in situations where the rms surface roughness is less than one third of the film thickness has been developed'48 and applied

in a study of finite bearing l ~ b r i c a t i o n ' ~ ~ The effects of surface roughness in hydrodynamic lubrica- tion and EHL can be summarized In general, for an infinitely long bearing the effect of transverse roughness is to increase load capacity while longitudinal roughness decreases load capacitỵ This occurs because transverse roughness ridges inhibit lubricant flow, generating an increase in pressure in the load-bearing zonẹ In contrast, longitudinal ridges provide flow channels for the lubricant to escape from the load-bearing zone, reducing its load capacitỵ In a finite bearing the situation is more complex Narrow bearings have the greatest side leakage flow and, therefore, support the lowest loads Longitudinal roughness tends to inhibit side leakage and, as a consequence, will tend to improve the load-carrying capacity

of 'short' bearings However, it should be noted that the net effect of roughness depends on the length-to-width ratio of the bearing as well as the nature of its roughness Load-carrying capacity is reduced when the bearing is wide and increased for very narrow configurations The friction force also depends on the bearing length-to-width ratio, with friction for longitudinal roughness being greater than that for transverse roughness The influence of surface roughness in mixed lubrication has also been considered Patir and Cheng'51s1s2 have derived a model which allows any three-dimensional roughness struc- ture to be analysed Mixed lubrication is a complex situation but in general, this model shows that as 'roughness' is increased, friction and load-carrying capacity of a surface is also increased Ađitionally, it has been of value in studying the power loss at the iston-ringkylinder interface in internal The majority of lubrication analyses employ the Reynolds equation, which assumes that the cross-film lubricant flow is negligible compared with the flow in other directions It should be noted that as surface roughness is increased this assumption becomes less valid Under such conditions the more complex Navier-Stokes equations should be used in principlẹ The lubrication of a rigid slider bearing has been considered using the Navier-Stokes equations and it has been demonstrated that the Reynolds equation progressively under- estimates the load capacity of such an arrangement as rough- ness is i n ~ r e a s e d ' ~ ~

Since surface roughness has an influence on friction it is not surprising that it also affects wear D a ~ s o n ' ~ ' found that the ratio, D , given by

D = 1lM was inversely proportional to the tendency for a bearing to begin pitting; a standard mode of fatigue failure experienced

by roller bearings and gear teeth The stresses generated by the repeated elastic deformation caused by the passage of a ball or roller result in this failure which manifests itself as a small 'pit' in a component where a fragment of metal has fallen combustion engines 1g

ality This would seem to support the validity of Archard's model, as it is usually assumed that the force of friction is proportional to the true area of contact between two bodies (Since friction is proportional to load, according to Amonton's law, it is reasonable to expect that the true area of contact is proportional to load.)

Subsequent models of surface contact, based mainly on statistical approaches, have introduced descriptions of surface topography which have varying levels of realism Many mo- dels assume surface asperities to have a Gaussian height distribution Experimental investigations imply that at least some surfaces produced by engineering processes do have near Gaussian height distributions, so such an assumption can

be j ~ s t i f i e d ' ~ ~ There are several versions Greenwood and Williamson168 constructed a model which used Gaussian and exponential functions to describe the height distribution of parabolic asperities prior to contact with a non-deformable plane Greenwood and trip^'^^ constructed a similar model which employed spherical asperities on both surfaces Whitehouse and ArchardI7' conducted an analysis which avoided stringent assumptions regarding the geometry of individual asperities by assuming instead that the surfaces had an exponentially decay- ing autocorrelation function Nayak17' applied the technique

of random process theory to describe engineering surfaces in a modification of the more complex analysis of random, mov- ing, ocean surfaces conducted by L o n g u e t - H i g g i n ~ ' ~ ~ , ~ ~ ~ This work enabled useful relationships between the statistics of surfaces and profiles to be derived and applied later in an analysis of plastic ~ 0 n t a c t l ~ ~

An academic debate exists regarding the nature of the deformation at asperity contacts Some of the authors of the surface models outlined above derive a dimensionless para- meter, the plasticity index, in an attempt to establish the proportion of plastic/elastic deformation which ocurs at asper- ity contacts These plasticity indices relate the elastic modulus, the material hardness and a measure of surface roughness in order to achieve this

The Greenwood and Williamson168 model considers the deformation of a rough surface of hardness, H , which has a Gaussian distribution of asperity peak heights with standard

deviation, u, and constant tip radius 0 , in contact with a plane

If the joint, elastic modulus of the contact is E the plasticity index, $, of the rough surface given by this model is

out (The mechanism of pitting in relation to surface rough-

ness is described in detail in reference 156.)

Scuffing is a second form of failure which is influenced by

surface roughness When brief inter-asperity contact or

asperity-particle contact occurs it causes friction which gen-

erates heat Scuffing is thought to arise when the speed and

pressure at the contact are increased so much that, at a critical

point, the heat generated cannot be dissipated in the sur-

rounding bulk material without melting at the contact An

expression to describe friction in gears which includes a

surface roughness term has been developed157 and it appears

that over a limited range of roughness friction can be reduced

by selecting an appropriate surface raughness for gears As a

consequence, resistance to scuffing is increased

The phase of wear which occurs in the early stages of sliding

is usually referred to as running-in The influence of surface

topography in running-in has been reported under both lubri-

cated conditionsL5* and dry sliding condition^.'^^ During the

runningin phase rough surfaces tend to become smoother, the

rate of wear depending on both the initial surface topography

and the applied load Other studies have found that once

runningin is complete the wear rate becomes independent of

the initial surface roughness16' and can be described by a

mechanism known as de1amination.l6' According to delamina-

tion theory, a material fatigued under sliding develops sub-

surface dislocations and voids which run parallel to the

surface When these faults coalesce, sheet-like wear particles

are released Under low loads the initial wear rate of a smooth

surface 1,s higher than that of a rough one because delamina-

tion commences relatively quickly compared to rough sur-

faces: where delamination is delayed until the asperities are

worn smooth, The opposite is true under high loads as rough

surface asperities are easily and quickly removed by adhesive

and abrasive processes under such conditions (see Halling162

for desciriptions of adhesive and abrasive wear mechanisms)

Bayer and S i r i ~ o ' ~ ~ have conducted tests into the influence

of anisotropy of surface roughness on wear Spherical speci-

mens with isotropic surface topography were positioned

against flat planes with a well-defined lay and caused to slide

under a fixed load in the boundary lubrication regime Using

sliding parallel and perpendicular to the lay of the anisotropic

plane Bayer and Sirico showed that as the coarseness of

surface increased, the sensitivity of wear to orientation be-

came gr'eater However, below a limiting roughness the in-

fluence exerted on wear by changes in roughness was a more

significant effect than the directional dependence

9.IO.I.2 Coniact mechanics

Surface topography exerts an influence on the processes which

occur at static junctions as well as ones where surfaces are in

relative motion Theoretical and experimental studies of static

contact are diverse and range from such subjects as the

analysis of the stiffriess of machine tool joints'64 to the

investigation of heat transfer and electrical resistance.l6j

The main aim of theoretical studies of static contact is to

predict tlhe real area of contact between the components which

form the interface so that related physical characteristics can

be derived During the last three decades several models of

various levels of geometric and statistical complexity have

been proposed to describe contact at interfaces

One of :he earliest csontact models was of a geometric nature

and was proposed by Archard.'66 Archard considered the

deformation of spherical asperities as they were loaded against

a non-deformable plane He showed that by progressively

covering each asperity with similar ones of smaller radius the

relationship between the true area of contact and the load

which it supported elastically approached direct proportion-

1 1 - v f 1 - v ;

where - = - +- Note that if El > E2 then

E E'= with E,, Ez and vl, v 2 being Young's Modulus (I - v2)

and Poisson's Ratio respectively for the interface materials 1 and 2

The index indicates if a surface is likely to suffer a significant proportion of plastic deformation of asperities at a nominal

pressure, P Surfaces which have a plasticity index greater

than unity will undergo plastic deformation at very low loads and can be expected to deform plastically in most practical applications If $ is less than 0.6, significant plastic deforma- tion will only arise under very high loads, which would not be encountered in routine applications Greenwood and William- son368 give an example in which a 'significant' proportion of plastic deformation is defined as plastic deformation over an area 2 2 % of the real area of contact They state that a nominal pressure of 2.0 x lo-* is required to cause significant

Tribology

plastic deformation if 1c, = 1 and that a nominal pressure of

4.3 x lo5 kg m-* is required to generate the same proportion

of deformation for a surface where t = 0.6 They also point

out that, in practice, the index 1c, can vary between 0.1 and

more than 100

Whitehouse and ArchardI7' derived a more general version

of the plasticity index from their surface model which does not

rely on the assumption that surface asperities have a constant

radius It has the form

where K is a numerical constant, H is the hardness of the

material, p* is a parameter (defined in Section 9.10.3.8, called

the 'correlation distance', and derived from the autocorrela-

tion function of the surface shape)

Onions and A r ~ h a r d ' ~ ~ found that if the constant, K , took

the value unity then the plasticity index, $*, could coinci-

dently, be interpreted in the same way as the index proposed

by Greenwood and Williamson This suggestion appears to be

generally accepted

The plasticity index is potentially a very valuable parameter

because it provides an assessment of the physical and geo-

metrical properties of a surface It has been suggested that it

could provide a useful guide to the condition of a surface

during the running-in process since the variation of the index

with time will show how quickly the surfaces involved

approach an elastic contact ~ 0 n d i t i o n l ~ ~

Differences between the redictions of these surface models

are discussed by Thomas." These differences arise principally

from variations in the form of the asperity models adopted,

illustrating that the character of a surface, as well as its scale of

roughness, has an effect on the nature of the contact between

the surfaces

P

9.10.1.3 Fluid flow

The topography of a surface influences the flow of fluids in

other situations in addition to lubrication This influence is

economically significant in a number of engineering applica-

tions, including sealing and the propulsion of ships and

aircraft

Many different types of sea! are used in engineering applica-

tions Establishing the desirable surface characteristics of an

efficient seal is not a trivial problem One approach has been

outlined in an investigation of the performance of lip

s e a l ~ " ~ ' ~ ~ This approach involved dividing the members of a

collection of lip seals into two categories: those which sealed

effectively and those which leaked The topography of all the

seals was characterized by measuring a number of commonly

used surface parameters, and some of these parameters

showed a loose correlation with the tendency of a seal to leak

By constructing a suitable algebraic function it was possible to

use all the measured parameters together to differentiate

between sealing and leaking topography much more effect-

ively, making it possible to identify the characteristics of an

ideally good and an ideally bad lip seal surface.'80

Fundamental work on fluid flow conducted by Nikuradse in

the 1930s has been recently reviewed Nikuradse investi-

gated the flow of fluids through pipes roughened by sand grain

coatings and demonstrated that the skin friction factor was a

function of the ratio of the grain size to the pipe diameter He

showed that in the turbulent flow regime of a rough pipe, the

logarithm of the friction factor is not inversely proportional to

the logarithm of the Reynolds number as it is in the case of an

ideal smooth pipe Instead, the linear dependence ends at

some point and the friction factor tends to a constant value

which is related to the surface roughness/pipe diameter ratio

as the Reynolds number increases

The work of Nikaradse has been criticized on the basis that

it does not simulate the topography generated by machining processes very effectively and is, therefore, of limited value

As a consequence, alternative methods of analysing fluid flow over rough surfaces in the form of model cracks have been investigated.'"

Fluid flow over rough surfaces is of considerable importance

in ship hull design, and it is reported that between 80% and YO% of the total resistance of a ship to motion is due to skin friction lg3 In addition, relatively small changes in roughness

exert a significant effect Conn et nl.lg4 describe an early

investigation into ship hull roughness effects For one particu- lar ship they observed that different paint finishes could vary its frictional resistance by up to 5% while allowing the ship hull to foul could increase it up to nearly 50%

A discussion of ship trial results collected over many years indicates that, typically, a 1% increase in power is required to maintain ship speed for every 10 p m increase of a roughness parameter called the mean apparent amplitude (MAA).ls5 It

is also stated that typical hull deterioration rates are between

10 p m and 40 p m MAA per annum, with new ships having a hull roughness of approximately 130 p m MAA while that of old ships may exceed 1000 p m MAA

The efficiency of aircraft is also strongly influenced by skin friction However, methods being considered to reduce skin friction seem to be directed along lines which are not asso- ciated with studies into the effect of the surface topography of the wings/fuselage.186

9.10.1.4 Vibration

Machines with moving parts inevitably produce noise when in operation Noise can be generated as a result of the misalign- ment of components and the occurrence of mechanical reso- nances at particular speeds However, a contribution to the noise spectrum can also arise as a result of surface roughness Thomas'87 discusses three separate sources of noise arising as

a consequence of surface roughness:

Noise generated by the elastic deformation and release of form and waviness features during rolling This type of noise can be generated in EHL At low speeds it contri- butes to the low-frequency end of the noise spectrum and it

is unlikely to cause problems However, at higher speeds both the freqllency of the noise and its energy are increased and its effect may be significant

Shock noise caused by the elastic deformation and release

of asperities within the Hertzian contact zone This form of noise arises in both rolling and sliding contacts It is apparently the most dominant form of surface-generated noise and it may affect a broad range of frequencies, although its most severe influence is exerted between

300 Hz and 10 000 Hz

Shock noise arising from asperity collisions and debris collisions This form of noise arises in EHL where very thin films separate moving surfaces Under such conditions inconsistencies in surface roughness can cause inter- asperity contact or collision with entrained debris to gen- erate transient noise

Two principal conclusions may be drawn with regard to noise generated by surface roughness:

1 By identifying the operating conditions of a component it should be possible to determine a surface finish which maintains low levels of acoustic noise

2 Surface-generated noise may be valuable as a form of 'non-invasive' surface monitoring It has already been

ones and the difficulties which limit the miniaturization of electronic circuits are engineering problems

The surface roughness of component substrates appears to

be one consideration in this respect Photolithographic pro- cesses are used in the manufacture of integrated circuits Deviation from flatness in the substrate surface causes gaps between the substrate and the emulsion of the photoplate which allows stray light to enter, causing defective circuit geometry lg8 Similar problems are encountered when circuit films are deposited through an evaporation mask Deviation in flatness between the deposition mask and the substrate permit eva orated material to enter and thus destroy line geome- try!99 In addition to these effects, substrate roughness has been found to influence the resistance of tbin-film resistors2” and the charge-storage capabilities of thin-film capacitors.201 Roughness effects appear to become significant when the thickness of the film layer - typically, 0.14.01 ,am - is of the same order as the average roughness of the s u b ~ t r a t e ’ ~ ~ The design of magnetic recording media is another branch

of modern technology in which surface topography is of importance The quality of a magnetically recorded signal depends on head-to-media spacing and, therefore, on surface roughness Computer disk systems are one important appiica- tion of magnetic recording technology The readwrite heads which transfer data to and from hard disks must maintain close proximity to the disk without actually coming into contact and causing wear They operate as small air slider bearings located between 0.25 p m and 0.4 pm from the disk surface, whose roughness amplitude is generally an order of magnitude less than the aidfilm thickness White202 has anaiysed the perfor- mance of thin-film air slider bearings with two-sided rough- ness He suggests that, in practice, the character of the topography of the read/write head and the disk will not have

an important influence on its operation, and shows that only the amplitude of the surface roughness therefore needs to be considered in the design of such components

shown that surface-generated noise can be used to detect

wear.’@ It is therefore potentially useful for the diagnosis

of po,Lential failure in rolling and sliding contacts

9.10.1.5 Coating technology

The appearance and service performance of coated surfaces

depends strongly on the surface topography of the overlaid

substrate As a result, materials which are to be coated are

generally required to have a surface finish which is uniform

and of a high quality, i.e free from pits and scratches

Cold-rolled steei is a typical example of a material for which

a good-quality surface finish is essential to maintain the

aesthetic appearance of objects constructed from it It is

widely used for the body panels of cars where small blemishes

are visible even when sprayed Avoiding visible defects is

exceedingly difficult The human eye can detect pits only

0.95 pin deep in a ‘smooth’ surface, and such features are only

just within the range of routine snrface-measurement proced-

urcs Eliminating them by improvements in the manufacturing

method can be difficult and expensive Other applications in

which control of surface roughness plays an important part in

maintaining an acceptable aesthetic appearance include the

manufacture of card, paper and photographic film

As well as influencing aesthetic appearance, substrate

roughness can also affect the service performance of painted

materials Substrate roughness causes variability in the film

thickness of painted materials, which has an observable effect

on corrosion resistance and ease of cleaning.’sg In general, it

seems that the effect of painting a ‘rough’ surface is to make it

‘smoother’ whilst the consequence of painting a ‘smooth’

surface is to make it ‘ro~gher’.’’~

9.10.1.6 Optics

Optical components such as lenses, prisms and mirrors are

used in many different types of equipment This equipment is

used in a wide range of applications for both conventional

purposes (e.g microscopy, photography, astronomy) and

more ‘exotic’ ones (e,g laser beam collimation, the control of

synchrotron radiation and infrared surveillance) Common

applications also include the rapidly increasing use of o tical

components in communication device^,'^' transd~cers’~’ and

data-storage equipment 193

In genlxal, in order to achieve satisfactory performance,

exceedingly stringent tolerances on the roughness of optical

components must be attained Typically the standard devia-

tion of their roughness is in the range 0.001-0.01 p ~ n ” ~ It is

important to be able to characterize the surface topography of

optical components so that limitations in performance can be

estimated Beckmann and S p i z ~ i c h i n o ’ ~ ~ explain how a stat-

istical description of a component such as a lens or mirror can

be used to predict its scattering properties In addition, Mie

theory which requires a knowledge of the size, shape and

optical constants of the feature can be used to redict scatt-

ering from small isolated features on a surface” The effects

of scratches, digs and other surface defects which have dimen-

sions much greater than the wavelength of light can be

determined only by relating their dimensions to those of

standard scratches whose scattering characteristics are known

c?

9.10.1.7 Electronics and computer hardware

In recent years the size of integrated circuits has been reduced

considerably while the number of electronic elements in a

given area has approximately doubled each year Ultimately,

fundamental physical criteria will limit this progress 19’

However, for the moment, such factors are not constraining

9.10.1.8 Bio-engineering, pharmacy and hygiene

There are several naturally occurring circumstances in biology where surface topography is important (e.g the lubrication of synovial joints as described by Tandon and Rakesh203) However, such examples cannot strictly be regarded as engi- neering applications! More valid examples are to be found in the discipline of bio-engineering

The performance of temporary artificial blood vessels is one aspect of this subject The range of roughness in tubes used to bypass blood during heart operations varies from about 2 pm

up to 15 pm.204 One problem associated with these tubes is that they can damage the red blood cells flowing through them Stewart204 found that twenty times more red blood cells were destroyed by some tubes in comparison with others His investigation established a correlation between the scale of roughness of the bypass vessel and the damage to red blood cells passing through them In addition, surface character appeared to exert an effect: ‘smooth’ tubes were found to cause less damage than ‘rough’ ones

Another aspect of the influence of surface topography on performance is evident in the pharmaceuticals industry The manufacture of tablets is conducted in shaped dies and the efficiency and reliability of the tableting process is influenced

by the surface roughness of these dies Dies which have an unsuitable topography result in the incorrect formation of tablets or their fracture on removal

Surface topography also has an important bearing on hygiene Many surgical, catering and household implements are manufactured with high-quality surfaces because smooth,

91108 Tribology

polished surfaces possess relatively few sites which harbour

bacteria during cleaning

9.10.1.9 Considerations in production

As long ago as 1933 a pioneering paper published by Abbott

and FirestoneZo5 recognized some of the inadequacies of

simplistic schemes of surface characterization The fact that it

is not possible to fully specify the character and scale of a

surface with a small set of arithmetic indicators still remains a

serious problem for production/design engineers Often a

description of a surface is made by specifying a particular

machining process to identify the principal surface character

and accompanying this by statistical parameters which give

further detail

The topography of an engineering component is largely

determined by two factors The type of machining method

used controls its character, while the ‘coarseness’ of the

operation, and to a point the time spent on manufacture,

determines the roughness scale

BS 1134’”‘ indicates that in a number of finishing pro-

cesses there is a range of finish over which the cost of

production changes only minimally with reductions in surface

roughness It also points out that there is normally a trans-

itional finish beyond which the cost of producing a surface

with further reductions in roughness increases rapidly It is,

therefore sensible to manufacture components with the scale

roughness which still maintains acceptable performance In

addition the least costly process which produces a surface

character with acceptable functional performance should also

be selected

Selection of an appropriate manufacturing process is not a

straightforward decision because it involves a detailed consi-

deration of how the entire component is to be made Many

components are made by multi-pass operations in which a

coarse maching process is used to obtain the nominal compo-

nent shape while a fine is used to establish the desired surface

finish This might involve applying the same type of operation

more than once using different tool or machine settings or it

may mean that a combination of entirely different processes

are used to obtain the final surface The combination of

operations which gives rise to target specification most econo-

mically may not be obvious

Establishing the machining conditions for a finishing process

to obtain a specified topography is also not straightforward, as

many interacting factors are involved Under ideal cir-

cumstances the factors to be considered would involve only

the operational setting of the machine (e.g the geometric

characteristics of the cutting tool the work speed, the tool

feed rate and the type of cutting fluid used) Even under ideal

conditions it is only possible to calculate the theoretical

roughness developed in a machining operation for the simplest

types of process, Le single-point tool cutting Relationships of

this type for turned surfaces generated by tools with a range of

tip profiles have been reported by D i ~ k i n s o n ~ ” Relationships

for less traditional metal-shaping processes such as electro- discharge machining (EDM), electro-chemical machining (ECM) and ultra-sonic machining (USM) have also been deduced.”’ However, they cannot be applied directly as they involve a parameter called a ‘surface finish factor’, which depends on the material of the workpiece/tool and the ul- timate surface finish of the component In order to apply the equations it is necessary to conduct trial tests to determine the surface finish factor for specific situations (Some values of the surface finish factor for specific situations are given by Gha- briel et d.20’.)

Expressions which describe the relationships between oper- ating conditions and surface roughness for complex cutting operations such as grinding and milling are apparently not available Companies employ their own heuristic guidelines to estimate surface finish in these cases

In practice, it is not usually feasible to obtain the surface finish predicted by theoretical conditions Several factors combine to prevent this Probably the most significant factor is the accretion of cutting debris on the tip of the tool This accretion is in a dynamic state, continually breaking down and being replaced with new debris, causing the tool to have variable cutting characteristics In general, the roughness of the surface produced by the built-up edge of the tool increases

as the built-up edge grows larger Changes in machining conditions which tend to reduce chip-tool frictioniadhesion result in improved surface finish These include increased workpiece speed, use of different tool materials (e.g carbide tool tips) and the application of a good cutting lubricant A

detailed study of the effect of built-up edge on the surface roughness of a turned component is presented by Selvam and Radhakrishnan.*09

Many other factors contribute to non-ideal surface finish arising under practical machining conditions The most com- mon of these are: chatter vibration of the machine tool, inaccurate tool movement, defects in the composition of the workpiece and discontinuous chip formation in the machining

of brittle materials In addition to these problems it appears that where several types of machining are applied in the manufacture of a component the ‘shadows’ of earlier opera- tions sometimes remain evident in the topography of the finishing process.210

From these studies it is apparent that many problems are associated with the production aspects of surface roughness Some are concerned with achieving good functional perfor- mance but just as many are of an economic nature BS 1134*06

points out that ‘seeking too good a finish’ is a common error leading to waste in production time It also outlines two simple philosophies to guide designers in controlling surface texture, but it concedes that the practice to be adopted depends very strongly on the area of engineering concerned

The optimization of manufacturing operations is a highly complex subject, even when only the problems outlined above are considered However, in the real manufacturing environ- ment many other important variables also become involved (e.g the power consumed by the various machining processes,

Surface topography 9/109

Meter system

the effect of the type of material being worked, and the

quantities of raw material wasted in manufacture) Surface

finish is just one consideration that must be reconciled with

u n i t

9.10.2 Measurement

Surface structure exists in two directions: the vertical direc-

tion where it is characterized by height (or amplitude)

parameters, and the horizontal direction where it is character-

ized by spatial (or wavelength) parameters Any measurement

method should be able to record roughness variations in both

directions

In the vertical direction roughness amplitudes vary widely

Coarse machining operations can produce features several

hundred micrometres high However, in contrast, some sur-

faces mainufactured for special applications may contain per-

turbations of only molecular dimensions In the horizontal

plane, roughness variations arise on a scale which varies from

the dimensions of the specimen down to atomic diameters It

is clear therefore, that wide extremes of range need to be

encompaissed by any technique used to measure surface

topography

The relevance of surface topography to the functional

performance and production cost of a component over a very

wide range of applications has precipitated the development of

a plethora of instruments to record and parameterize the

structure of surfaces, and each of these devices has different

limitations and advantages In recent years development has

centred largely on instruments which measure surface

topography without contact with the specimen, i.e mainly

optical methods However, these devices are, as yet, not

particularly well established in terms of the number of applica-

tions in which they are used routinely This situation may

change in the future but, for the moment, more traditional

techniques (i.e stylus instruments and ‘mechanical’ compa-

rators) share the bulk of the workload

h p l i f i e r

9.10.2.1 Stylus instruments

For many years the most popular device used to measure

surface topography has been the stylus instrument Despite

some disadvantages, it has proved versatile and reliable in

both the manufacturing and research environments It is also

the instrument in terms of which all national standards are

defined.’”

S t y l u s

t r a n s d u c e r

Construction and opefation A typical commercial stylus

instrument comprises five basic components: a stylus trans-

ducer, an amplifier, a chart recorder, a traverse unit and a

meter system as illustrated by the block diagram in Figure

9.91 (a)

A comimon form of stylus transducer, which employs a

linear variable differential transformer (LVDT), is manufac-

tured by Rank Taylor Hobson It consists of a beam which is

pivoted on two knife edges and carries the stylus at one end

with a ferrite block at the other (Figure 9.91(b)) The ferrite

block is located between two coils As the stylus is drawn over

the irregularities of a surface by the traverse unit the stylus is

displaced, causing the ferrite block to move between the coils

The coils form part of an inductance bridge circuit which is

balanced ,when the stylus is in a neutral position When the

stylus cha.nges position it causes a change in the mutual

inductance in the coils modulating a high-frequency carrier

signal in proportion to the displacement of the stylus The

relative change in the phase of the carrier signal indicates the

(a) Principal components of a commercial stylus

direction of displacement of the stylus The carrier signal is amplified and demodulated to yield a signal representing a surface profile which may be output to a chart recorder or used to evaluate some parametric assessment of surface rough- ness given on a meter or visual display unit

In addition to LVDTs, several other forms of displacement transducer are in use (e.g optical interferometers, variable

capacitor^'^^)

Range and resolution LVDT transducers usually have a maximum vertical range of between 0.5 mm and 1.0 mm The vertical resolution of stylus instruments depends mainly on the level of ambient vibration in the vicinity of the instrument but

it is also influenced by electrical noise inherent in the amp- lifier Divisions of 0.01 p m are common on the output of chart recorders, suggesting that this level of accuracy can be approached under controlled conditions

The horizontal range of a stylus instrument is determined by the distance which the stylus is able to traverse The horizontal resolution depends on the shape and dimensions of the stylus tip, and it is a limit which cannot be defined e ~ a c t l y ” ~

9/110

Datum arrangements The output of a stylus transducer de-

pends on the difference in height between the stylus and a

reference datum To exclude the form of a component from

measurements of topography it is necessary to generate this

datum by causing the transducer to follow a path parallel to

the nominal shape of the component This can be achieved in

two ways:

1 Using an independent datum as illustrated in Figure

9.92(a) Here an accurately flat or curved shape which

corresponds to the nominal shape of the specimen is used

to constrain the vertical position of the transducer as it

traverses the specimen

2 Using a skid datum which is attached to the transducer

The skid rests on the specimen and follows its form as the

transducer is drawn along by the traverse unit in the

manner illustrated by Figure 9.92(b)

When the independent datum is used, all departures from

the nominal shape are recorded However, it is a time-

consuming method of measurement as an extensive setting-up

procedure is often required to align the specimen and datum

Use of the skid datum circumvents this difficulty but surface

profiles recorded using this datum may not be accurate A

specimen surface containing features with wavelengths greater

than the length of the skid surface profile will be subjected to a

high-pass filtering operating as the skid changes its vertical

position when crossing these features Isolated peaks in a

surface can also generate spurious artefact

S u p p o r t column

S a d d l e s u p p o r t i n g t r a n s d u c e r , S t y l u s

Figure 9.92 Datum arrangements for nominally flat specimens

(a) Use of the independent datum: (b) use of the skid datum

A number of experimental and theoretical investigations have been conducted to examine the extent of profile distor- tion by the skid datum The main conclusions are:

1 Use of the skid datum, in appropriate circumstances, only has a small influence on the parametric characteristics of a profi~e.~"

2 When parametric influences are observed the peak para- meters of a profile are those most significantly affected.216

3 Inaccuracies caused by use of a skid are greater for surfaces with prominent directional characteristic^.^^'

Stylus effects Early publications re ort that gramophone styli were used as the transducer probe!18 These have since been replaced by specially designed styli which are smaller and tipped with diamond to improve resistance to wear Two types

of stylus shape are commonly used: a conical stylus with a spherical tip and a pyramidal stylus with a truncated, nominally flat tip The spherical tip of a conical stylus usually has a radius of less than 10 pm The flat area of a pyramidal stylus is typically a rectangle of dimension 8 p m X 4 pm, the longer side being normal to the direction of traverse The profile recorded by the stylus tip is never a perfect assessment of the shape of the specimen This arises as a consequence of two factors; the finite size/geometry of the stylus and the load it applies to the specimen surface:

Geometricisize considerations

The shape recorded by a stylus as it traverses a surface is the locus of a fixed point on the stylus and is known as the 'effective pr~file'.'~' As a consequence of its shape and finite size the profile recorded by the stylus can differ from the true surface shape; i.e its shape can be distorted and, in some circumstances, its amplitude may be attenuated In general, the shape and finite dimensions of the stylus cause the radius of curvature of peaks to be slightly enhanced and the width of valleys to be slightly reduced (Figure 9.93) Attenuation of high-frequency contributions to the profile shape also occurs This arises because the stylus is unable to enter valleys which are narrower than its own length This is

a progressive effect which begins when the radius of curva- ture of a profile valley is smaller than the effective radius of curvature of the stylus

In most stylus instruments a spring is used to apply a small force, approximately 0.001 N, to the stylus to ensure that it

always remains in contact with the specimen As this load is

supported by a very small area at the stylus tip, high pressures are generated at the contact with the specimen

and temporary or permanent deformation can ensue A

number of investigators have estimated the magnitude and form of deformation caused by styli However, theoretical estimates of deformation due to stylus pressure tend to vary widely depending on the assumptions made in the calcula- tion Many physical and chemical factors such as variation

in hardness, oxidation level and fatigue influence the result

as well as parameters which relate to the stylus geometry

Additional problems can also arise as the bulk properties of

a material do not necessarily give an acceptable model of its microscopic behaviour (e.g microscopic hardness has been found to be a function of load219)

It appears to be accepted that the errors caused by the effects described above are negligible under most cir- cumstances However, care should be taken when features of

interest have the same order of magnitude of width as the

stylus tip and, additionally, when the specimen has a low bulk hardness value Methods for estimating errors due to the shape and load of styli are given in reference 214

e Physical interactions

Figure 9.913 Distortion of a sinusoidal profile when traversed by a spherical stylus

Digital data-acquisition systems The development of flexible

digital syijtems has permitted the value of stylus instruments in

experimental studies in surface topography to be extended

considerably Digitized data are commonly acquired by samp-

ling the amplifier ouiput from stylus instruments In the

earliest systems, analogue data were converted into digital

form and stored on punched paper or magnetic tape and then

taken to a remote computer for subse uent analysis Such an

More recently, therc has been an increase in the use of

on-line systems Equipment typically used in on-line data

acquisition is ihstrated in Figure 9.94 The principal compo-

nents are a commercial stylus instrument which is linked to a

computer through an analogue-to-digital converter (ADC)

and an interface which usually includes a filter to set a

approach was adopted by Williamson 9220

high-frequency cut-off point in the data To generate the output signal the transducer gearbox may be operated manually or automatically by the computer via a series of electric relays Collected data are normally stored on-disk or

in memory for subsequent processing rather than being ana-

lysed in real time A number of on-line data systems have been

d e s ~ r i b e d * ~ ~ - ~ ~ ~

One common criticism of stylus instruments is that they are

normally restricted to supplying only profile data However, stylus instruments can be adapted to allow an areal record of a

surface to be obtained, rather than a simple profile This is

achieved by recording several parallel profiles, each displaced laterally from the previous one by a short distance If all the profiles are referenced to a common origin they form a

raster-scan record of the surface A raster scan recorded using

Stylus instrument

7

Timebase t o control sampling

Chart recorder disk systems and

4-

converter

Visual display u n i t

9/112 Tribology

the author’s equipment is illustrated in Figure 9.95 Areal

measurements are attracting increasing interest and devices

which perform raster measurements have been described in

several p ~ b l i c a t i o n s * ~ ~ - * ~ ~

9.10.2.2 Comparative techniques

A wide range of instruments which use comparative tech-

niques have been developed Generally, their principle of

operation is based upon a physical phenomenon which shows

some dependence on surface roughness Sometimes these

devices are designed to give a parametric assessment of

surface roughness, although the reliable operation of this type

of instrument is often limited to a given class of surface

topography

Comparative devices have several advantages; they are

relatively inexpensive, they are quick to apply, and they do

not require extensive special training to operate

Friction tests Probably the most well-known comparative

measurement technique is the tactile test In this test the finger

is drawn across a specimen and its ‘feel’ (Le the frictional

resistance) is then compared to that produced by a set of

calibrated samples manufactured by the same process The

sample whose ‘feel’ most closely resembles that of the speci-

men is noted

It has been found that the tactile test is significantly more

reliable than visual inspection as a method for assessing

Vertical scale

1.31 p m

surface roughness However, the range of accurate compari-

son appears to depend on the character of the surface under inspection.”x

Electrostatic techniques The distance separating two plane metal surfaces can be estimated by using them to form a capacitor If they are separated by an insulating layer of air of

thickness t,, the capacitance, C, of the combination is given by

KE,A

C = -

t a

where A is the area of the smaller plane, e, is the permittivity

of air and K is a constant Measuring the capacitance of the

combination using a bridge circuit permits t, to be calculated

In principle, by forming a capacitor using a free electrode and

a rough surface it is possible to obtain some assessment of surface roughness by measuring the capacitance of the ar-

rangement and calculating t,, which will then be some average

measure of separation (Note that it is not the distance of the mean plane of the surface from the probe.)

A number of transducers of this form have been described

in publications2*Y~”0 and the most reliable one has a flexible electrode coated with a dielectric The flexible electrode allows the transducer to conform to the surface

shape to remove the effect of waviness and form errors The

dielectric layer serves to make the transducer less sensitive to contact pressure Capacitance measured roughness, t,, shows

Surface topography g i l l 3 The finite size of the stylus and the load it applies to the specimen can give rise to measurement errors and speci- men damage

Its operation is relatively slow It cannot be used as an

‘in-process’ measurement technique (e.g to monitor the performance of numerically controlled tools)

some correlation with a parameter called ‘depth of surface

smoothness’, defined by reference 231

Pneumatic methods Surface roughness can be measured by a

method called air gauging, which assesses the flow of air

through a gap between a specimen surface and an open-ended

nozzle placed facing downwards onto it The nozzle is

norma1l:y connected to a source of air at a constant pressure,

P ; via an intermediate chamber The flow of air intq this

chambeir is regulated by a valve whose open area is a ; its flow

out is controlled by the area, A , of the gap between the nozzle

and the surface asperities which support it Escape of air from

the nozzle causes the pressure in the intermediate chamber to

fall to p It can be shown that, over a certain range, the

relationship between p i p and a / A is linear, allowing the

method to be used to assess surface r o ~ g h n e s s ’ ~ ~

Experimental measurements made by air gauging have been

found to exhibit a linear relationshi with an average rough-

ness of R, = 0.1 p m to Ra = 5 pmr?’ The extent of correla-

tion of back pressure with other rou hness parameters has also

been investigated by experiment.23‘

A novel approach to air-gauging techniques has been

describemd by Tanner.’35 Rather than measuring pressure diff-

erences, a null method based on a pneumatic analogue of a

Wheatstone bridge was used to measure pressure in the

nozzle Several stages of development resulted in a compact

device able to give an electrical signalireadout proportional to

the avemge surface r o n ~ g h n e s s ~ ~ ~ ~ ~ *

Air-gauging methods provide a simple, inexpensive, port-

able, quick and robust way of assessing surface roughness well

suited to use in the quality control of surfaces on the shopfloor

and the technique appears to find particular favour in the

paper industry British Standards have been compiled in an

attempt to develop a common approach in its use in this

applicatiion (e.g reference 239) Table 9.29 summarizes the

performance of three comparative techniques used to assess

surface roughness

9.10.2.3 Optical methods

Although the stylus instrument is currently very widely used in

the measurement of surface roughness, it suffers several

significant disadvantages:

1 It normally provides information only for a profile section

of a surface

Table 9.20 A summary of the performance of parametric

techniques of surface rodghness measurement

Technique Usable R , Notes

range

Friction tests 0.025-3.2 p m Accuracy depends on

Electrost.atic 1.5-10 p m Data not available on

measurements Ref 230 wider range measurements

Pneumatic 0.16-5.69 pm Range apparently

methods Ref 233 extendable if required

Ref 228 surface type

1 Normally areal data can be obtained easily

2 There is no contact between the specimen and the instru- ment, so no surface damage is caused during measurement

3 Measurements can be performed quickly

4 Measurements can be made on any type of material (Le they are not restricted to observations of electrically con- ductive materials as required by electron microscopy)

5 The specimen does not need to be in a vacuum

6 Complex specimen preparation is not required

Techniques which measure surface roughness using optical phenomena can be divided into two broad categories: non- parametric techniques, which are &le to record the actual topographic structure of a specimen, and parametric tech- niques, which are only able to assess the general characteris- tics of an area of surface (e.g its rms roughness)

Non-parametric instruments A number of non-parametric

techniques are available and can be divided into three catego- ries by their principle of operation:

1 Light sectioning

2 Interference microscopy

3 Focus feedback methods

Light sectioning is a non-destructive analogue of the process

of taper sectioning whereby an optical microscope is used to examine a section cut through a specimen surface at a shallow angle to magnify height variations.z4o Light (or optical) sec- tioning is sometimes referred to as the Schmaitz technique, after its inventor Figure 9.96 illustrates the principle of light sectioning The surface is illuminated by a light beam colli- mated by a narrow slit This is then viewed from the side, usually by means of a microscope Topographic features in the surface, illuminated by the beam: appear as profile sections Illumination and viewing angles of 45” are normally adopted, leading to a magnification of the height of features in the profile by a factor of (m

The practice of illuminating the specimen from an angle introduces distortion into the observed image The features visible in it do not represent a true cross section of the surface because the valleys are displaced laterally from the peaks by a small distance This can be overcome by ilfuminating the specimen from a normal angle, although this reduces the

vertical magnification

Optical sectioning is suitable for examining surfaces whose roughness range is between 2 pm and 200 p n The vertical resolution of the technique is about 0.5 pm?

Until recently, because of the difficulty of extracting quanti- tative information, optical sectioning has only been used to assess the profile range, R,, and for quantitative examination

of surface features This difficulty has been overcome by digitally recording an image delineated by a narrow laser beam using a television camera This image is then transferred to a computer for storage, processing and di~play.”’~~~’

Figure 9.96 Principle of the light-sectioning method (Based o n

Figure 128, Dagnall, H Exploring Surface Texture and reproduced

by permission of Rank Taylor Hobson)

In interference microscopy, if two slightly inclined glass

plates are illuminated by a coherent monochromatic light

source, a series of parallel light and dark bands will be visible

when the arrangement is viewed from above (see Figure 9.97)

The dark bands arise as a result of the destructive interference

of light wave fronts The distance between neighbouring

fringes is Ai2, where A is the wavelength of the illumination In

principle, interference techniques can be used t o examine

surface topography by replacing the lower glass plate with a

reflective specimen and the upper one with some form of

reference plane nominally aligned with the specimen Inter-

ference between light beams from these surfaces will generate

a contour pattern of the surface irregularities of the specimen

Interference microscopy is suitable for measuring surface

irregularities with low slopes and roughness ranges less than

1 pm (In areas of high slope, surface contours blend together

and cannot be interpreted.) Several types of interference

microscopes exist and their design and capabilities are re-

viewed by references 244-246 The two main categories of

interferometer are the double- and the multiple-beam types

Typically, the double-beam interferometer has a horizontal

resolution of 1.0 p m and a vertical resolution of 0.01 pm The

corresponding figures for the multiple-beam device are

2.5 p m and 0.001 pm.247 (In multiple-beam interferometry

the beam is displayed sideways at each of the 50-100 reflec-

tions, causing a reduction in horizontal resolution.)

Until recent years interferometric measurements have been

restricted in their use to the examination of specific features,

such as the height of steps and the depth of grooves, because

interferograms could not be readily transformed into an

electronic signal which could be submitted for further analysis

Figure 9.97 Interference fringes caused by an air wedge (wedge

angle exaggerated) At X, the path difference, PD, is zero but a dark fringe is observed This is caused by the 180-degree phase change which occurs when light is reflected at a dense medium

The phase change is equivalent to a PD of A/2 At the m t h dark

fringe where the wedge thickness is t, the effective PD, mA, is

to obtain accurate surface statistics for use in verifying light- scattering theory.246

Interferometric measurements of even higher accuracy can

be obtained using phase detecting systems in which two polarized beams of slightly different frequency are focused onto a point on the specimen surface and a stationary ref- erence The beat frequency of interfering return beams is directly proportional to any change in height of the specimen surface The ‘interrogating’ beam can then be swept around the surface to obtain hei ht measurements at specific points Circular252 and linear253J54 scan patterns have been used in surface examination These systems are able to examine small