Bhushan B. Handbook of Micro/Nano Tribology, Second Edition

Подождите немного. Документ загружается.

useful in tribology since you are often dealing with rough, inhomogeneous surfaces. We show a sample

SAM map in Figure 3.19.

Depth proﬁling is the process of sputter-eroding a sample by bombarding the surface with ions while

simultaneously obtaining AES or other spectra. This enables one to obtain the composition of reaction-

formed or deposited ﬁlms on a surface as a function of sputter time or depth. Consequently, AES has

many applications for studying tribological and other surfaces. Some examples will be given in subsequent

sections.

3.5.2 X-Ray Photoelectron Spectroscopy

The physical processes involved in XPS are diagrammed in Figure 3.20. XPS is a simpler process than

AES. An X-ray photon ionizes the inner level of an atom and in this case the emitted electron from the

ionization is itself detected, as opposed to AES where several levels are involved in the ﬁnal electron

production. The dispersion and detection methods are similar to AES.

FIGURE 3.19 Example scanning Auger microscopy results. Sample is silicon carbide ﬁber-reinforced titanium

aluminide matrix composite. Single element images as labeled, with higher concentrations represented as brighter

regions. (Courtesy of Darwin Boyd).

FIGURE 3.20 XPS transition diagram for an atom. (From Ferrante, J. (1993), in Surface Diagnostics in Tribology

(K. Miyoshi and Y. W. Chung, eds.), World Scientiﬁc, Singapore. With permission.)

Monochromatic, incoming X-ray photons are generated from an elemental target such as magnesium

or aluminum. Measurement of the energy distribution of emitted electrons from the sample permits the

identiﬁcation of the ionized levels by the simple relation

(3.15)

Since the ﬁnal energy is measured and the X-ray energy is known, one can determine the binding

energy and consequently the material. AES peaks are also present in the XPS spectrum. AES peaks can

be distinguished from the fact that the energies of the Auger electrons are ﬁxed because they depend on

a difference in energy levels, whereas the XPS electron energies depend on the energy of the incident

X ray. A sample XPS spectrum is shown in Figure 3.21 and a schematic diagram of the apparatus is shown

in Figure 3.22.

XPS can perform chemical as well as elemental analysis. As stated earlier, when an element is in a

compound, there is a shift in energy levels relative to the unreacted element. Unlike AES, where energy

level differences are detected, a chemical reaction results in an energy shift of the element XPS peak. For

example, the iron peak from Fe

is shifted by nearly 4 eV from an elemental iron peak. Not only are

the shifts simpler to interpret, but it is easier to detect peaks directly (as opposed to AES derivative mode

measurements) since the signal-to-background in XPS is greater than in AES. In addition, the mode of

operation in the dispersion step typically enables higher resolution. The surface sensitivity of XPS is

similar to AES because the energies of the emitted electrons are similar. Figure 3.23 shows some examples

of oxygen and sulfur peak shifts resulting from reactions with iron and chromium for wear scars on a

steel pin run with dibenzyl-disulﬁde as the lubricant additive (Wheeler, 1978).

3.5.3 Secondary Ion Mass Spectroscopy

The physical process involved in SIMS differs from both AES and XPS in that both the excitation source

and detected quantity are ions. Rather than illuminate the sample surface with either electrons (AES) or

photons (XPS), ions are used to bombard the sample surface and knock off (sputter) surface particles.

The dispersion phase analyzes the emitted particle masses, instead of energy analyzing the emitted

electrons as in AES or XPS. Although using sputtering implies an erosion of the sample surface, a

FIGURE 3.21 Example XPS spectrum. (From Ferrante, J. (1993), in Surface Diagnostics in Tribology (K. Miyoshi

and Y. W. Chung, eds.), World Scientiﬁc, Singapore. With permission.)

EEE

final xray binding energy

=−

compensating advantage for SIMS is extreme sensitivity. Under advantageous conditions, as few as

atoms per cm

(ppb) have been detected (Gnaser, 1997), with more typical sensitivities for most

elements in the ppm range (Wilson et al., 1989). A comprehensive discussion of the SIMS technique has

been published by Benninghoven et al. (1987).

The SIMS technique typically used in surface studies gives partial monolayer sensitivity using small

incident ion currents (“static” SIMS). Higher ion beam currents, often rastered, give species information

as a function of sputter depth (“dynamic” SIMS or SIMS depth proﬁling). SIMS instrumentation can be

roughly categorized by the type of ion detector used, e.g., quadrupole, magnetic sector, or time-of-ﬂight,

with their inherent differences in sensitivity and lateral and mass resolution. As well, the incident angle,

energy, and type (e.g., noble gas, cesium, or oxygen) of the primary ion sputtering beam employed can

greatly affect the magnitude and character of the secondary ion yield.

SIMS has several complexities. SIMS only detects secondary ions, rather than all of the sputtered

species, which can lead to difﬁculty in quantiﬁcation. Large molecules on the surface such as hydrocarbon

lubricants or typical additives can exhibit complex patterns of possible fragments. A knowledge of the

adsorbate and cracking patterns is often needed for interpretation. As well, multiply ionized fragments

or simply different species may overlap in the spectra, having nearly identical charge-to-mass ratios. As

a simple example, carbon monoxide (CO) and diatomic nitrogen (N

) overlap, requiring examination

of other mass fragments to distinguish between the two. As with depth proﬁling for either AES or XPS,

depth resolution “smearing” can occur either due to ion beam mixing of near-surface species or due to

the development of surface topography after long times under the ion beam. Despite these potential

limitations, SIMS should remain the technique of choice for many low detection limit, high surface

sensitivity studies (Zalm, 1995).

3.5.4 Infrared Spectroscopy

IRS is particularly useful in detecting lubricant ﬁlms on surfaces. It can provide binding and chemical

information for adsorbed large molecules. It has an additional advantage in that it is nondestructive.

Incident electrons in AES can cause desorption and decomposition even for aluminum oxide, and can

be very destructive for polymers. Similarly, the emitted electrons can cause destruction of some ﬁlms for

both AES and XPS. In IRS, the specimen is illuminated with infrared light of well-deﬁned energy. If the

FIGURE 3.22 Schematic diagram of XPS apparatus. (From Ferrante, J. (1993), in Surface Diagnostics in Tribology

(K. Miyoshi and Y. W. Chung, eds.), World Scientiﬁc, Singapore. With permission.)

energy of the incident light corresponds to a transition between vibrational energy levels in the specimen,

the light can be absorbed. When compared to the reference light beam that has not passed through a

sample, the infrared light interacting with the sample will appear at reduced intensity at these vibrational

excitation energies. The dispersion step is similar to dispersion in photon spectroscopy in that a grating

or prism is used to isolate the wavelengths of interest. A variation of IRS which has advantages in sensitivity

and resolution is called Fourier transform infrared spectroscopy (FTIR). In FTIR, the incident beams

are passed through a Michelson interferometer in which one of the paths is modulated by moving a

FIGURE 3.23 Sulfur 2p and oxygen 1s XPS peaks from unworn steel surfaces and wear scars run in mineral oil

with 1% dibenzyl disulﬁde. (From Wheeler, D. R. (1978), Wear 47, 243–254. With permission.)

mirror. As before, one of the modulated beams is passed through the sample and impinges directly on

the detector, which is a heat-sensitive device. When nonmonochromatic radiation is used, the Fourier

transform of the spatially modulated beam contains all of the information in one signal, as opposed to

the dispersion method where each beam must be analyzed separately. A schematic diagram of the

equipment is shown in Figure 3.24.

IRS is surface sensitive and has been used in a diverse range of analytical science applications (McKelvy

et al., 1996), which we will not address. A primary interest in tribology is the detection of hydrocarbon

or additive ﬁlms on surfaces. AES and XPS, for contrast, would primarily be useful in detecting elemental

species and are limited to use in ultrahigh vacuum. IRS can be used in air as well as vacuum. The surface

sensitivity of IRS can be enhanced by multiple reﬂections in the surface ﬁlms and by using grazing

incidence angles. Orientation of adsorbed molecules can be obtained by examining the polarization

dependence of the spectrum. There are selection rules for what materials can be detected depending on

whether the molecule has a dipole moment and the orientation of the molecule on a metal surface.

Sample spectra for micron thick and less than 200 Å thick Krytox ﬁlms on a metal substrate are shown

in Figure 3.25 (Herrera-Fierro, 1993).

3.5.5 Thermal Desorption

We mention brieﬂy at this point another useful tool for examining the behavior of adsorbates on surfaces,

thermal desorption spectroscopy (TDS). We give only a brief description and for a more complete

treatment refer the readers to Zangwill (1988). The methods described so far give little information

concerning binding energies of adsorbates to surfaces, a topic of importance when choosing stable

lubricants or additives. Many species adsorb strongly but do not react chemically, e.g., by forming an

oxide. When the surface is heated they can be removed intact or in some decomposed form. A simple

view would be that the binding energy would resemble the curves presented in Figure 3.12, and the

adsorbates could be removed by giving them sufﬁcient energy to overcome the energy well depth. This

would be accomplished by heating a sample following adsorption and then either observing the surface

FIGURE 3.24 Optical arrangement for a FTIR spectrometer. (From Ferrante, J. (1993), in Surface Diagnostics in

Tribology (K. Miyoshi and Y. W. Chung, eds.), World Scientiﬁc, Singapore. With permission.)

coverage via AES or monitoring pressure increases in the vacuum system. There can be a variety of

bonding states depending on the structure of the surface; e.g., at a step edge one would expect a different

binding energy from a surface site. Although the real situation may be quite complex, we describe the

simplest case, a single bonding state, and no decomposition. The rate of desorption can be described by

(Meyers et al., 1996)

(3.16)

where θ is the coverage, ν is the pre-exponential frequency factor, E

is the desorption energy, T is the

temperature, and k is the Boltzmann constant. By heating to various temperatures and performing an

Arrhenius plot, one can extract E

and determine the strength of bonding to a surface. This technique

will be seen to be useful in studying monolayer and submonolayer adsorbed ﬁlm effects on friction

between otherwise clean metal surfaces.

FIGURE 3.25 IR spectrum of a krytox ﬁlm on a 440C bearing showing characteristic absorption corresponding to

certain functional groups: (a) micron-thick ﬁlm; (b) less than 200-Å-thick ﬁlm, note the increased sensitivity and

the bend from absorbed water from 1400–2000 cm

–1

. (Courtesy of Herrera-Fierro, 1993).

r d dt E kT

== −

()

θθνexp

3.6 Surface Effects in Tribology

In this section we will deal with a number of issues, analyzing evidence ﬁrst for effects at the atomic

monolayer or submonolayer level in tribology. We will examine monolayer effects both from a funda-

mental standpoint as well as from a more practical viewpoint. This will not be a comprehensive review

of the literature, but in keeping with the objectives of this chapter, again will be didactic in nature. The

references selected, however, will have references to the relevant literature. In this chapter we are not

concerned with the effects of lubricants other than their effects of changing shear strength or adhesion

at the interface. Consequently, we are interested in issues involving boundary lubrication and interfacial

properties (Ferrante and Pepper, 1989; Gellman 1992; Carpick and Salmeron, 1997).

There are a number of issues to address which emphasize the difﬁculties involved in answering

fundamental questions concerning bonding (Ferrante and Pepper, 1989). One clear difﬁculty is the fact

that one cannot observe the interface during the interaction. It is necessary to infer what happened at

the interface by examining the states of the surfaces before and after interaction to the extent analytical

techniques can determine. There are often situations where the locus of failure is not the interface, e.g.,

shear or adhesive failure can occur in the bulk of one of the materials rather than at the interface, or

both effects can occur depending on the region in contact. There are uncertainties regarding the mea-

surement of forces, although some recent efforts nicely relate lateral force sensitivity to normal force

sensitivity for a commercial scanning probe microscope (SPM) cantilever beam, i.e., for a single asperity

contact (Ogletree et al., 1996). Clearly, there are elastic effects in any measuring apparatus and in the

materials involved that make measurement of the force distribution at the interface difﬁcult. Materials

can change mechanical properties as a result of the forces applied. For example, such properties as

hardness, ductility, defect formation, plasticity, strain hardening, and creep must be considered. Surface

properties may be altered just by contact with the counterface material (Carpick et al., 1996). Generally,

even the true area of contact is not known in macroscopic studies due to the fact that asperities determine

the contact area on most practical materials.

There is a great deal yet to be learned concerning the basic interfacial properties in tribology. This is

in part due to the complexities involved. As an example of such complexities, in Table 3.1 we show some

results of Buckley and Pepper (1971) for metallic transfer of dissimilar metals in sliding contact performed

using a pin-on-disk apparatus in an ultrahigh vacuum system with AES analysis in the wear track.

Both pin and disk specimens were ion sputter cleaned. As we can see all metals transferred to tungsten,

and cobalt transferred in all cases. However, iron and nickel did not transfer to tantalum, molybdenum,

or niobium. The surprising result is that the softer metals did not transfer to the harder in all cases.

Pepper explained these results in terms of the mechanical properties of the materials. Tungsten which is

the hardest of the materials ﬁts the expected pattern. However, since nickel and iron strain harden, transfer

TABLE 3.1 Metallic Transfer for Dissimilar

Metals in Sliding Contact

Transfer of Metal

Disk Rider from Rider to Disk

Tungsten Iron Yes

Nickel Yes

Cobalt Yes

Tantalum Iron No

Nickel No

Cobalt Yes

Molybdenum Iron No

Nickel No

Cobalt Yes

Niobium Iron No

Nickel No

Cobalt Yes

and deformation are minimized. Cobalt, which has a hexagonal close-packed structure, has easy slip

planes and thus transferred in all cases. Thus, simple explanations based on cohesive and interfacial

energies can be misleading if mechanical properties are not taken into account.

We give a second example by Pepper (1974) demonstrating the care necessary in performing studies

of polymer ﬁlms transferred from polymer pins sliding on a S-Monel disk in ultrahigh vacuum.

Figure 3.26 shows the AES spectra for polytetraﬂuoroethylene (PTFE), polyvinyl chloride (PVC) and

polychlorotriﬂuoroethylene (PCTFE) pins sliding on S-Monel. The PTFE spectrum shows large ﬂuorine

and carbon peaks and large attenuation of the metal peaks (care had to be taken due to electron

bombardment desorption of the ﬂuorine). The friction coefﬁcient was low and smooth suggesting slip.

The combined results indicated that PTFE strands were transferring to the metal surface consistent with

the models of Pooley and Tabor (1972). For PVC the AES spectrum shows a large chlorine peak and

small attenuation of the metal peaks suggesting decomposition and chlorine adsorption rather than

polymer transfer. The friction coefﬁcient, although reduced, remained large and exhibited some stick

slip. For PCTFE the spectrum shows chlorine, carbon, and intermediate attenuation of the metal peaks,

but with stability under electron bombardment suggesting the possibility of both decomposition and

some polymer transfer. The friction coefﬁcient was high with stick slip. Consequently, it is difﬁcult to

anticipate what is happening, again demonstrating the need for more extensive surface characterization.

3.6.1 Monolayer Effects in Adhesion and Friction

The bulk of the discussion in this section will be based on the recent high-quality experiments of Gellman

and collaborators (Gellman, 1992; McFadden and Gellman, 1995a,b), while we acknowledge the pio-

neering contributions of Buckley and the members of his group (Buckley, 1981). Gellman has performed

a number of adhesion and friction experiments on single crystals in contact, both for clean and adsorbate

covered interfaces, namely, Cu(111)–Cu(111) and Ni(100)–Ni(100). The apparatus for the friction and

adhesion experiments is shown in Figure 3.27 (Gellman, 1992).

The contacting crystals had a slight curvature in order to prevent contact at the edges of the samples

where large concentrations of steps were expected, and to ensure point contact. The vacuum system was

also vibration isolated which is very important for reasons to be discussed below. Normally, the experi-

ments were repeated for ten different points. The normals of the crystals were aligned by lasers. The

apparatus provided the ability to sputter clean both contacting surfaces, and LEED and AES could be

performed on the surfaces in order to guarantee the crystallinity and to measure contamination and

determine concentration of adsorbates. The AES measurements were performed with LEED optics in

the old retarding ﬁeld analyzer mode. Following sputter cleaning, the samples were annealed and analyzed

with LEED in order to verify the crystal structure and with AES to verify cleanliness.

On Cu(111) McFadden and Gellman (1995a) examined the effects of sulfur adsorption, since sulfur

is a common antiwear additive. It is often used along with phosphorous in order to prevent metal to

metal contact when the lubricant breaks down. Both surfaces were dosed with sulfur using hydrogen

sulﬁde, which decomposes upon heating, desorbing the hydrogen and leaving sulfur behind on the

surface. At saturation, sulfur forms an ordered superlattice which has a × R19 degree LEED

pattern. This corresponds to a close-packed monolayer of sulﬁde ions (S

2–

) on top of the Cu(111) lattice

with an absolute coverage of 0.43 monolayers relative to the Cu(111) substrate. The results of the adhesion

experiments are shown in Figure 3.28.

The adhesion coefﬁcient, µ

= F

, which is the ratio of pull-off force to normal force was found

to be 0.69 ± 0.21. As little as 0.05 monolayers (11% of saturation) gave a substantial percentage reduction.

The saturation value at one monolayer was 0.26 ± 0.07. They also found that there was no dependence

of the adhesion coefﬁcient on contact time, separation time, temperature, or normal force. There would

be no expected dependence on normal force since higher loads would simply increase the contact area,

and the adhesion coefﬁcient, therefore, would be expected to track with the normal force. Buckley (1981)

reported adhesion coefﬁcients much greater than 1.0 for clean single crystals in contact. As mentioned

earlier, it is important to control vibrations in the experiments. Bowden and Tabor (1964) showed that

when clean metal surfaces were plastically loaded and then translated, the junction grew, i.e., the true

contact area increased until shear occurred. In all probability, the large adhesion coefﬁcients observed

by Buckley were caused by increased contact area from junction growth caused by vibrations. In any

event, there is strong bonding at clean metal interfaces and, as Gellman and others have shown, the

adhesion coefﬁcient can be decreased by submonolayer ﬁlms.

McFadden and Gellman (1995a) performed a somewhat different experiment on the Cu(111) surface

with regard to static friction measurements. The surfaces were exposed to laboratory atmosphere for

several days. The resulting AES spectrum is shown in Figure 3.29. As can be seen, the primary contam-

inants were sulfur and carbon. Although tenuous, they estimated the coverage to be between 10 and 15 Å

thick. They then executed a number of sputtering cycles followed by annealing with static friction

measurements after each cycle. The results are shown in Figure 3.30. The friction coefﬁcient, µ

, increased

with removal of contaminant to their clean surface value of 4.4 ± 1.3. Again these results imply sub-

monolayer effects by these contaminants.

Wheeler (1976) earlier performed static friction experiments with both chlorine and oxygen adsorbed

on polycrystalline surfaces of copper, iron, and steel. The measurements were performed with a pin-on-

disk apparatus, with AES used to monitor coverage. The results of these studies are shown in Figure 3.31.

Wheeler found that there were no differences between the effects of chlorine and oxygen on any of the

surfaces if adsorbate coverage was taken into account. Adsorption at partial monolayer coverages reduced

the coefﬁcient of friction in all cases. Wheeler found that he could get a good correlation with a junction

growth model where the surfaces were partially covered during translation. The values of the coefﬁcient

were of comparable magnitude to those of Gellman.

3.6.2 Atomic Effects Due to Adsorption of Hydrocarbons

Our primary objective in this and following sections is to explore the evidence for atomic effects on

friction. Again, we start by referring to recent publications by Gellman et al. (Gellman, 1992; McFadden

and Gellman, 1995a,b; Meyers et al., 1996) describing a number of friction experiments performed with

hydrocarbon adsorbates. First, we present the effects of ethanol adsorption on a sulfur-covered Ni(100)

surface. The sulfur was adsorbed as previously described and produced a c 2 × 2 structure on the nickel

surface. The sulfur overlayer was used since stable results could not be obtained with ethanol alone,

which Gellman found desorbed at very low temperatures. There were a number of different friction states

depending on ethanol coverage, shown in Figure 3.32 (Gellman, 1992).

These are the classic behaviors observed in friction (Buckley, 1981). The ﬁrst shown is “slip,” where

once a certain transverse force is applied there is simple sliding. The second is “stick-slip,” where rebonding

occurs and the simple slip process is repeated between rebonding events, with poor deﬁnition of a friction

coefﬁcient. This behavior probably results from junction growth with the adsorbate only effective in

certain regions, as observed by Wheeler (1976). Finally, “stick” is where the surfaces weld and it is similar

to shearing a bulk solid. The results of these studies are shown in Figure 3.33 where the key indicates

differences in behavior.

Unlike the results for Cu(111) the sulfur coating did not reduce µ

and gave stick behavior. Partial

coverages of ethanol gave erratic behavior typically exhibiting stick-slip. Finally at one monolayer the

ethanol effectively lubricated the surface giving the desired slip as well as low adhesion. Consequently, a

monolayer ﬁlm can effectively lubricate. Similar behavior was observed for 2,2,2-triﬂuoroethanol

adsorbed on clean Cu(111) surfaces (McFadden and Gellman, 1995b). Again stick-slip was observed for

coverages less than a monolayer and slip for coverages greater than or equal to one monolayer. Similarly,

high adhesion was observed for low coverages and low adhesion for higher coverages.

FIGURE 3.26 AES spectra from an S-Monel disk with transfer ﬁlms from different polymers. (From Pepper, S. V.

(1974), J. Appl. Phys. 45, 2947–2956. With permission.)