Contents

    

Mechanism of Interaction and Degradation of Perfluoropolyethers with a DLC Coating in Thin-Film Magnetic Rigid Disks: A Critical Review

Czeslaw Kajdas¹ and Bharat Bhushan²

¹ Warsaw University of Technology, Institute of Chemistry in Plock and Central, Laboratory of Petroleum in Warsaw, Poland
² Computer Microtribology and Contamination Laboratory, Department of Mechanical Engineering, The Ohio State University, Columbus, Ohio 43210-1107

Abstract

This paper reviews, discusses, and analyzes the literature on the interaction and degradation mechanisms of perfluoropolyether lubricants with carbon protective overcoats used for magnetic media. Emphasis is placed on the degradation process of perfluoropolyether lubricants under sliding conditions. Detailed comments on various degradation mechanisms are presented. Particular stress is put on Z-DOL lubricant degradation mechanisms. It is believed that the dominant degradation mechanism is the low-energy electron initiated degradation mechanism.
* — Published in : J. Info. Storage Proc. Syst, Vol., 1, 1999

1. INTRODUCTION

The perfluoropolyether (PFPE) class of lubricants has been widely used for numerous special tribological applications because of their superb oxidative and thermal stability properties. The most sophisticated applications include magnetic recording disks for computer data storage (Bhushan, 1996). In computer industry applications, mostly for topical lubrication to reduce the wear of rigid disks, commonly used lubricants include Fomblin Z and Demnum, as well as their polar derivatives containing various reactive end groups, e.g., hydroxyl (Fomblin Z-DOL, Demnum SA), and piperonyl (Fomblin AM 2001; Bhushan, 1994).
The areal storage density, the number of bits stored per unit area, has been increasing at a rate of greater than 40% per year since 1990 (Simonds, 1995). This trend relates to a decrease in the flying height or spacing between the read-write head and the surface of the magnetic storage disk. Currently, the magnetic medium is protected with a hard overcoat, which is mostly a hydrogenated amorphous carbon film; however, nitrogenated carbon has been considered as well.
The primary goal of this review paper is to provide a better understanding of the PFPE lubricant interaction with carbon overcoats, under both static and sliding conditions. The secondary aim of the work is to elaborate and discuss in detail a model of polar lubricant-substrate interaction based on the hydrogen bonding. Emphasis is also placed on PFPE degradation mechanisms showing that low-energy electrons may play an important part in the PFPE degradation process.

2. PFPE LUBRICANTS

Perfluoroalhylpolyether (PFPE) lubricants are a group of polymers composed entirely of carbon, fluorine, and oxygen. Development of PFPE's as lubricants was first disclosed over 30 years ago (Gumprecht, 1966). Since that time, the following four major types of PFPE lubricants (Del Pesco, 1994), that is,
I CF3—O—(CF2—CF2—O)m—(CF2—O)n—CF3
II CF3—O—(CF—CF2—O)m—(CF2—O)n—CF3
║
CF3
III CF3—CF2—CF2—O—(CF2—CF2—O)m—CF2—CF3
IV CF3—CF2—CF2—O—(CF—CF2—O)m—CF2—CF3
║
CF3
have become commercially available. By brand names they are known as Fomblin Z (I), Fomblin Y (II), Demnum (III), and Krytox (IV). They are colorless, odorless, compatible with most other materials, liquid over a wide temperature range, and completely inert to most chemical agents, including oxygen. The linear PFPE structures (I and III) show less change of viscosity with temperature and pressure when compared with nonlinear PFPE's (structures II and IV). Demnum (III) and Krytox (IV) contain one type of ether group and are homopolymers, whereas Fomblin Z (I) and Fomblin Y (II) contain two types of ether groups. PFPE lubricants are superior candidates to be used under severe and starved operating conditions, particularly at high speeds and elevated temperatures, as well as under heavy loads. For example, they are being developed as lubricants for special applications in such severe environments as in aerospace engines and satellite instruments (Jones and Snyder, 1980; Jones, 1995). Another specific application of PFPE lubricants relates to gas turbine engine oils (Gschwender et al., 1992). PFPEs are the current lubricants choice for magnetic recording media; however, constantly increasing recording densities and faster data transfer rates in the magnetic recording industry-apart from rising smoother thin-film disk surfaces and lower flying heights-require improved PFPE lubricants.
The best examples of the improved PFPE lubricants for magnetic recording media are Z-DOL and Demnum SA. Z-DOL includes two hydroxyl groups combined with the PFPE molecule by means of a —CH2 bridge. Demnum SA includes the —CH2—OH functional group only at one end of the PFPE molecule. PFPE lubricants without functional groups are nonpolar. Those lubricants with functional groups are becoming popular because the electrons of the functional group, for example, in Z-DOL, are not shared equally between the oxygen atom and hydrogen atom. This is because the oxygen atom draws electrons in a bond toward itself and thereby obtains a partial negative charge. To obtain balance, the hydrogen atom has a partial positive charge. Accordingly, the polarity of a compound relates to the magnitude of the separation of charge. As they have both positive and negative poles they are called dipole; that is, they have a dipole moment. Dipole moment is particularly important from the point of view of weak intermolecular bonds (van der Waals forces) in liquids. These forces include the following types of molecular interactions:
  1. dipole-dipole,
  2. dipole-induced dipole, and
  3. induced dipole-induced dipole.
The dipole-dipole molecular interaction clearly relates to the hydrogen bonding (hydrogen bridge formation) that can be found in both water and alcohols. Thus, hydrogen bonding seems to be of particular importance for the interaction of polar PFPE lubricants (mostly Z-DOL and Demnum SA) with a diamondlike carbon (DLC) coating and Al2O3-TiC head slider substrate in the interface. This especially relates to a humid environment because the PFPE hydroxyl (OH) group can easily form hydrogen bonds to neighboring water molecules. Such compounds, like water, are said to be hydrophilic. Accordingly, PFPE lubricants without specific functional groups including the -OH group are hydrophobic.
In the case of Z-DOL, a hydrogen atom, covalently bonded with an oxygen atom in the HO bond, has a partial positive charge on the end of the bond. This hydrogen atom can be easily attached to the negative charge of other molecules. Similarly, the lone pairs of electrons in the oxygen atom can be attached to positive charges of other molecules. A hydroxyl group (OH), as a part of an alcohol group in the Z-DOL molecule or water molecule structure, can form hydrogen bonds through its hydrogen atom and the unshared pairs of electrons on the oxygen atoms of neighboring molecules, resulting in an intrinsically polar molecule. The most recent work clearly demonstrated the importance of hydrogen bonding in the tribological performance of Z-DOL and X-1P lubricants at a head-disk interface (Zhao et al., 1999a). Major characteristics of these specific polymers along with other most important PFPE lubricants are shown in Table 1. Other functional groups of PFPE lubricants include carboxyl, ester, or piperonyl entity.

Table 1 Chemical Structure and Selected Properties of PFPE Lubricants

LubricantFormulaEnd GroupsMolecular
Weight
(Daltons)		Density
(x10³ kg/m³ at 20°C)		Kinematic
Viscosity
(mm²/s)
Z-03CF3—O—(CF2—CF2—O)m—(CF2—O)n—CF3—CF336001.8230 @ 20°C
Z-15CF3—O—(CF2—CF2—O)m—(CF2—O)n—CF3—CF391001.84150 @ 20°C
Z-DOLaHO—CH2—CF2—O—(CF2—CF2—O)m—(CF2—O)n—CF2—CH2—OH ; (m/n ~ 2/3)—OH20001.8180 @ 20°C
34 @ 40°C
SAaF—(CF2—CF2—CF2—O)m—CF2—CF2—CH2—OH—OH36001.8775 @ 40°C
YR
CF3
║
CF3—O—(CF—CF2—O)m—(CF2—O)n—CF3
║
CF3		—CF368001.911600 @ 20°C
a — Functional or polar lubricants

The characteristic feature of all these functional groups is that they are polar. In other words, the polar groups determine the interaction level between a lubricant and various carbon surfaces. Usually, the magnitude of such interaction is evaluated by the heat of adsorption. Pertinent data are collected in Table 2 (from Yanagisawa, 1994), which shows heats of adsorption for PFPE with no functional group, ester, piperonyl, and hydroxyl groups to graphite, CVD carbon, sputtered carbon, amorphous carbon, and diamond. Table 2 clearly demonstrates the highest heat of adsorption for hydrophilic groups. This is the reason why Z-DOL and Demnum SA are currently the most-used lubricants to lubricate carbon overcoated disks. From the viewpoint of the functional group type, the magnitude of heat of adsorption is in the following order:

none < ester < piperonyl < hydroxyl

showing that the adsorption energy between functional groups and carbons increases with the increasing hydrophilic affinity of these groups. In contrast, the hydrophilic affinity of lubricants is also of importance for interaction of the environmental humidity with lubricant molecules. This relates to the hydrogen bonding of polar molecules in which there is a partial separation of charge to give positive and negative poles. It is of note at this point that hydrogen bonds may also be somehow separated from other examples of van der Waals forces because they are stronger than other intermolecular bonds. The strength of the hydrogen bond in water has been estimated to be between 10 and 12 kJ/mol (Bonder and Pardue, 1989). For comparison, the bond between planes of carbon atoms in a graphite structure is estimated to have a dissociation enthalpy of 17 kJ/mol.

Recently, cyclic phosphazenes have been developed as advanced lubricant additives for rigid thin

Table 2 Heat of Adsorption for PFPE's with Various Functional Groups to Various Carbon Surfaces

Adsorbent (Carbon)Functional Group (PEPE)Heat of Adsorption (mJ/m²)
GraphiteNo functional0.00
Carbon
CVD
No functional0.01
Sputtered
No functional-
Amorphous
No functional0.05
DiamondNo functional0.07
GraphiteEster0.06
Carbon
CVD
Ester0.15
Sputtered
Ester0.32
Amorphous
Ester-
DiamondEster0.55
GraphitePiperonyla0.44
Carbon
CVD
Piperonyla1.5
Sputtered
Piperonyla-
Amorphous
Piperonyla2.62
DiamondPiperonyla2.57
GraphiteHydroxyl3.56
Carbon
CVD
Hydroxyl5.64
Sputtered
Hydroxyl14.60
Amorphous
Hydroxyl19.14
DiamondHydroxyl38.47
a — Piperonyl is :

film magnetic media. As phophazenes have totally different chemical structure from PFPE they will be discussed separately in another review paper. It is only to note at this point that pertluoropolyether triazene and monophosphatriazene have been considered as
Fig. 1 Physical processes associated with surface deformation during boundary friction (Kajdas, 1997).

lubricant and additive for space application (Jones and Snyder, 1980; Jones, 1995) and for gas turbine engine oils (Gschwender et al., 1992), respectively. The characteristic feature for these two compounds is the presence of the six member ring including apart from phosphorus atoms 3 nitrogen atoms or 3 nitrogen atoms and one phosphorus atom (monophosphazene). The ring of the X-1P molecule is composed of nitrogen and phosphorus atoms. Such structures adsorb well on various substrates. The most recent papers (Zhao and Bhushan, 1999; Zhao et al., 1999a; Kasai, 1999) describe not only X-1P's role as PFPE lubricant in disk files, but also discuss its lubrication mechanism.

3. CARBON OVERCOAT AND ITS INTERACTION WITH PFPE'S

3.1. Introduction

Head-disk interface interactions for a given tribological system are controlled by the chemistry of both head and disk surfaces as well as by the lubricant chemistry. In a magnetic rigid-disk drive, recording or playback is accomplished by relative motion between a magnetic medium against a stationary read-write magnetic-head slider. There is a physical contact between the slider and medium during starting and stopping. To reduce the wear generated during the contact-start-stop (CSS) operation, hard overcoats and liquid lubricant films are typically applied to the disk surface. In modern high-end disk drives, the head-to-medium separation is of the order of 25 nm, and surfaces have a roughness of the order of 1.5 nm rms. The need for increasing high recording densities requires that surfaces should be as smooth as possible and flying heights be as low as possible. In contrast, smoother surfaces lead to increase in adhesive interaction between the mating surfaces controlling both friction and interface temperatures. It should also be emphasized that closer flying heights lead to occasional rubbing of high asperities and increased wear, accompanied with various physical processes. Usually, the application of mechanical energy associated with friction releases a great number of physical processes (Fig. 1), which can be the cause of tribochemical reactions of solids with lubricants molecules. According to Kajdas (1987, 1994), the most important factor governing tribochemical reactions under boundary friction is associated with the action of low-energy electrons with lubricating oil components. This is the basis of the negative-ion-radical action mechanism (NIRAM).
In order to find out if electrons are emitted at head-disk sliding contacts in magnetic recording media, electron emission phenomena were investigated together with the friction and wear in a frictional system of an alumina ball sliding on amorphous and hydrogenated amorphous carbon films in a dry air atmosphere (Nakayama and Ikeda, 1996). The researchers found that electrons are emitted during wear of both amorphous and hydrogenated amorphous carbon films. The transition from low to high electron emission intensity was caused by the increase in wear of the carbon films.
The general model of NIRAM assumes the creation of two types of activated sites on friction surfaces, i.e., thermally activated sites and sites activated by a low-energy electron (exoelectron) emission process. The situation is illustrated in Fig. 2. Bearing in mind that thin films of perfluoropolyethers can be attached to a variety of substrates, such as amorphous carbon, silica, and gold, by illumination of the substrate with 185-nm UV light (Saperstein and Lin, 1990; Vurens et al., 1992), it is possible to hypothesize that lubricant reactions with head-disk sliding contacts might not only be initiated by temperature and/or catalysis, but
Fig. 2 Surface activation during friction (Kajdas, 1994).
also by low-energy electrons emitted from the carbon substrate under the boundary friction conditions. The hypothesis can be strengthened by the fact that the electron bonding is some 3 to 4 orders of magnitude more efficient than the UV bonding process (Vurens et al., 1992). Application of the NIRAM approach to the head-disk sliding contact tribochemistry will be described in a later subsection.
Currently, carbon films lubricated with PFPE are widely used as the protective coating for thin-film magnetic storage disks. Therefore, before the work on PFPE degradation-reaction mechanisms is reviewed; carbon overcoats will be discussed.

3.2. Carbon Overcoat

3.2.1. Chemical

Carbon is known to occur in various elemental forms. There are three allotropic crystalline forms (diamond, graphite, and fullerenes) and a number of amorphous (noncrystalline) forms. The amorphous carbon form encompasses carbon black, charcoal, and coke. The properties of diamond are particularly noteworthy. It is among the least volatile substances known, and its melting temperature exceeds 3550 °C. This specific allotropic carbon is also the hardest substance known, and it expands less on heating than any other material. These diamond properties are a logical consequence of the fact that carbon atom forms covalent bonds (sp³) to four neighboring carbon atoms arranged toward the corners of tetrahedron. The strength of the C-C bonds plus their arrangement in space provide the unusual properties of diamond. Diamond is an excellent insulator, opposite to graphite. The specific feature of amorphous carbons relates to the fact that they usually have unpaired electrons (dangling bonds).
For rigid-disk overcoats, mostly hydrogenated carbon of various amorphous forms a-C:H are used. Hydrogenated carbon films generally have higher resistivity and therefore are expected to be more corrosion resistant to galvanic corrosion. Hydrogen content significantly influences various properties of the carbon films. Bhushan (1996, 1999a) reviewed DLC film applications for magnetic media, considering their physical nature, chemical structure, and tribological characteristics. It has also been emphasized that many carbons deposited by sputtering and other deposition techniques for nondisk applications have been found to be possibly more diamondlike; however, sputtered carbon for thin-film disks tend to be less diamondlike (primary amorphous), with few graphitic and diamond crystallites. Lee, Smallen, et al. (1993), investigating hydrogenated carbon films, found that the carbon bonding character changes from sp² to sp³ with an increase of hydrogen content. Dillon et al. (1984) revealed that the number or size of the carbon microstructure decreases with an increase in the hydrogen content. Most interestingly, the amorphous carbon film structures change from graphitelike structures with more sp² bonding when there is no hydrogen content, via diamondlike structures with more sp³ bonding when there is intermediate hydrogen content, to polymerlike structures with —CH2 bond when there is excessive hydrogen content (Dillon et al., 1984; Cho et al., 1990). These structural changes also cause the change in the internal stress of the hydrogenated carbon films.
Perry and Somorjai (1995), investigating the adsorption of water and small perfluorinated compounds on hydrogenated amorphous carbon films (a-C:H), found a relationship of hydrogen content on the surface free energy. Changes in hydrogen content of the films were observed to modify the a-C:H surface free energy and to directly influence the desorption temperature of submonolayer coverages of both water and perfluorinated species. They concluded that hydrogenated carbon films exhibiting the highest surface free energy provided a greater attractive interaction for the model lubricants and may provide greater stability of thin lubricant films on these surfaces.
Carbon surface chemistry of magnetic disks is of particular importance from the viewpoint of lubricant interaction with the disk or head surface. A variety of functional groups has been found to exist on carbon surfaces (Silva et al., 1992). Yanagisawa (1994), using X-ray photoelectron spectroscopy, determined many oxygen-containing groups along with the conjugated group on the sputtered carbon surface. Table 3 (Yanagisawa, 1994) summarizes the ratio of these groups to total carbon. From the measured functional groups and ESR spectrometry measurement results, which provided evidence for the existence of dangling bonds on the amorphous carbon, Yanagisawa (1994) proposed an amorphous carbon surface model (Fig. 3). The model particularly emphasizes the presence of dangling bonds (three dots shown in Fig. 3); five oxygen-containing functional groups (hydroxyl, carbonyl, carboxyl, oxide, and ester) are included by only one group each.
Fig. 3 Diagram for the Z-DOL molecule adsorbed on the carbon surface at 20 °C (Yanagisawa, 1994; Bhushan and Zhao, 1999).

Table 3 Functional Group to Total Carbon Ratio on Hydrogenated DLC Surfaces and Their ICN's

Functional GroupFormula% Total CarbonICN
Ether—C—O—C—15.720
Carbonyl═C═O6.065
Ester—COOR3.560
Carboxyl—COOH0.5150
Hydroxyl—COH0.3100

Fig. 4 Simplified diagram of the chemical structure of a hydrogenated carbon overcoat surface.

The ratio of oxygen-containing functional groups presented in Fig. 3 does not relate to their percentage of total carbon. Even considering inorganic character numbers (ICN), as discussed by Yanagisawa (1994), the most abundant functional groups would relate to the ether group (15.7% of total carbon with ICN = 20), and the carbonyl group (6.0% of total carbon and ICN = 65). The presence of other oxygenated groups (—COOR, —COOH, and —OH) is much less significant, although their ICN's are high, particularly for the carboxyl group (ICN =150). The inorganic character number shows the relative magnitude of association between function groups.

Taking into consideration all the above discussed data, we propose the use of a simplified carbon surface diagram (Fig. 4) which reflects the real content of oxygenated functional groups as presented in Table 3. The simplified diagram shows only one combined group, representing both the ester group and the carboxyl group. As the carbonyl group cannot be combined with a five-valence carbon (see Fig. 3), in Fig. 4 it is presented in the form of an aldehyde group, which is substantiated by the fact that mostly carbon overcoats are hydrogenated. For comparison, Fig. 5 shows an idealized graphite surface layer plane with various functional groups located at the periphery of the plane (Janzen, 1982).
Carbon overcoats lubricated with PFPE are used as the protective coating for magnetic storage disks and metal evaporated (ME) tapes. The thickness of the protective overcoat is 10-15 nm for disks and 8-10 nm for tapes (see Fig. 6). The addition of a limited amount of hydrogen in the amorphous carbon overcoat improves the CSS durability of thin-film disks lubricated with PFPE's.
Fig. 5 Aromatic layer plane with functional side groups (Janzen, 1982).
  1. Liquid lubricant 1-2 nm
  2. Diamond like carbon overcoat 8-1Onm
  3. Magnetic coating 25-75 nm
  4. Al-Mg/1Oµm Ni-P. Glass or Glass-ceramic 0.78-1.3 mm
  1. Liquid lubricant 2-3 nm
  2. Diamond like carbon overcoat 8-IOnm
  3. Magnetic coating 100-130 nm
  4. Polymer film with particulates 10-25nm
  5. Base film 6-10 µm
  6. Back coating 0.3-0.5 µm
  7. Liquid lubricant 8-10 nm
Fig. 6 Construction of a thin-film magnetic rigid disk in comparison with a ME tape (Bhushan and Zhao, 1999).

Hydrogen incorporated in the amorphous carbon changes its chemical and physical properties. For example, electrical resistance and hardness of the carbon film vary with the amount of hydrogen. A hydrogenated carbon surface chemistry differs significantly from a nonhydrogenated one (Wang et al., 1996). The same is true for the triboemission process (Nakayama, Bou-Said, et al., 1997). However, it is not due to dangling bonds. Yanagisawa(1994) discussed the question why dangling bonds are not saturated by hydrogen, referring to papers by Wada et al. (1980). The authors of the two latter papers emphasized that dangling bonds in carbon are relatively stable. Dangling bonds are not easy to bond to hydrogen because double bonds in sp² carbon would be saturated first with hydrogen (Cho et al., 1990). According to Jansen et al. (1985), even if the carbon is hydrogenated, the density of unpaired
Fig. 7 Concentration of carbonyl and hydroxyl end groups of a-C:H films (Wang et al., 1996).
electrons (dangling bonds) in the hydrogenated carbon decreases only from 1018 to 1016 cm -3.

Interestingly, Wang et al. (1996) found that the concentration of the carboxyl groups on carbon surface decreases with increasing hydrogen content, whereas that of the hydroxyl groups exhibits an increasing tendency dependence on the hydrogen content increase. This specific relationship is depicted in Fig. 7.
Another characteristic feature of hydrogenated carbons concerns triboelectromagnetic phenomena. These phenomena include the microplasma formation of sliding contacts, the triboemission of electrons, ions, and photons, and tribocharging (Nakayama, Bou-Said, et al., 1997; Nakayama, Yamanaka, et al., 1997). As already mentioned in the preceding subsection, Nakayama and Ikeda (1996) first investigated the emission of triboelectrons, using the frictional system of alumina sliding on hydrogenated carbon films with different hydrogen content. Alumina simulated the Al2O3-TiC head slider. They presented evidence that the intense electron emissions were controlled by the hydrogen content.
In the most recent work by Nakayama, Bou-Said, et al. (1997), triboemission of negatively charged particles and positively charged particles, tribocharging, and friction coefficient were measured simultaneously by using a tribosystem with diamond sliding on hydrogenated carbon films in ambient air. The hydrogen content of the carbon films differed from 0 to 43 at. %. Figure 8 clearly demonstrates that all the carbon films tested emitted both negatively and positively charged particles under the sliding conditions used. A typical pin-on-disk apparatus was used in which the diamond stylus was a cone having an included angle of 90° and a tip radius of 300 µm. Sliding experiments were conducted under an applied normal force of 200 mN; the wear track diameter was 16.5 mm and the sliding velocity was 5.2 cm/s. Tests were carried out in ambient air that had a relative humidity of 52-53%. Interestingly, at low hydrogen content (below 15 at. %), no tribophotons were observed.

Fig. 8 Relation between charge intensity and hydrogen content in carbon film (Nakayama, Bou-Said, et al., 1997).

Fig. 9 Friction coefficient during the sliding of a diamond stylus against hydrogenated carbon film, as function of hydrogen content in carbon film
(normal force = 200 mN, sliding speed =5.2 cm/s; Nakayama, Bou-Said, et al., 1997).

The emission intensities of the negatively charged particles and positively charged particles are low in that region of hydrogen content. Figure 8 also manifests that the charge intensity of negative particles is greater than that of positive particles, particularly in the low hydrogen content region. In the region of hydrogen contents between 15 at. % and 37 at. %, the emission intensity of the charged particles increased very rapidly, accompanied by the photon emission. It is of note at this point that the friction coefficient also increased from a low to a higher value in the region of hydrogen content from 15 to 27 at. %. The relationship of the friction coefficient values and hydrogen content in the carbon films tested is presented in Fig. 9. From these results, Nakayama, Bou-Said, et al. (1997) concluded that a microplasma state is formed at the frictional contacts of diamonds sliding on hydrogenated carbon films.

Other research by Wang et al. (1995) studied the effect of hydrogen content on the tribology of the head-disk interface. Among other things, they found that the durability of the film increased as the hydrogen content was increased from 12 to 36 at. %.

3.2.2. Interaction Mechanisms of a Carbon Overcoat with PFPE Lubricants

There is a wide variety of issues, combined with a possibility of obtaining adequate lubrication and protection of the disk surface. One is the PFPE lubricant thickness and its bondability with the carbon surface. A second is the chemical stability of the lubricant film under sliding conditions. Another issue is the ability of the lubricant to flow across the surface. While it must have an interaction with the DLC (a-C:H) overcoat that is strong enough to prevent loss during spinning, it must at the same time be able to flow into small regions of the surface that are depleted of lubricant as a result of contact with the head. Unhydrogenated amorphous carbon has a small energy gap and thus has a relatively high electrical conductivity. The small energy gap results possibly because of increased presence of sp² bonds. Resonance hybridization leads to delocalization of the electrons within a plane. Because electrons are free to move through the plane, they can conduct an electric current. Such material usually develops dangling bonds. Kaplan et al. (1985) found, however, that the hydrogenated amorphous carbon continues to have a loss energy gap of 1.2 eV in comparison with the direct gap of purely sp² bonded carbon (7.3 eV). This led to the conclusion that the presence of sp² bonding states requires the hydrogen not only to passify the dangling bond but also to saturate the double graphite bonds (sp²). Thus, for a-C:H films, it is of importance to know to what extent hydrogen passifies the dangling bonds and how these control the carbon surface activity and electrical conductivity. This issue also interrelates with electron emission intensity during sliding, as presented in Fig. 8. It is also pointed out that the electrical conductivity and hardness have been found to depend on the carbon preparation conditions (Enke, 1981; Cho et al., 1990). Amorphous hydrogenated carbon films of high electrical resistivity were studied by Miller and McKenzie (1983), using electron spin resonance both as prepared and after vacuum heat treatment. They found that all the carbon films investigated had a high spin density in the range from 10²¹ to 10²² spins/kg. Dangling bonds in a-C:H overcoats should be of particular importance from the viewpoint of the carbon overcoat interaction with PFPE lubricants. Most interestingly, for a-C:N materials, Torng and Silvnrtsen (1990) found that the energy gap is increased from 1.1 to 1.4 eV when the nitrogen partial pressure increased from 2.5 mTorr to 10 mTorr. These values are very close to the energy gap value (1.2 eV) for hydrogenated amorphous carbon (Kaplan et al., 1985). At the moment we have not found any information on triboemission from the a-C:N substrate; however, comparing the energy gap data for both a-C:H and a-C:N materials, we can speculate that nitrogenated carbon also emits electrons under boundary friction conditions, either in contact with diamond, alumina, or any DLC substrate. As with a 2.5-mTorr N2 partial pressure in the plasma, the nitrogen content in the carbon is already quite high (25 at. %; Torng and Silvertsen, 1990); we can assume that lower nitrogen content a-C:N substrates emit fewer electrons than a-C:H substrates. Considering the interaction of lubricants with surfaces of hydrogenated carbon films, we must bear in mind what types of active sites exist on the surface (see Figs. 3 and 4) and how they can interact with the lubricant molecule. Thus, the most important factor controlling this interaction will relate to the chemistry of the PFPE end groups. Accordingly, lubricants with polar end groups such as hydroxyl, carboyl, or piperonyl can strongly interact with the DLC surface, forming bonded films. To achieve necessary wear protection, the magnetic medium is covered with a film (10-20 nm; see Fig. 6 ) overcoat of a wear-resistant amorphous hydrogenated carbon, and it is coated with a molecularly thin film (1-2 nm) of a perfluoropolyether lubricant.
Perry and Somorjai (1995) investigated the interaction of low-molecular model lubricants with the surfaces of a-C:H carbon films. To evaluate the interaction of model perfluorinated lubricants with thin a-C:H films, they employed thermal desorption spectroscopy (TDS). The following compounds were tested:
  1. perfluorodiethyl ether,
  2. perfluoropentane,
  3. perfluoro-octane, 2,2,2-trifluoroethanol, and 1,1, 7-H-perfluoroheptanol.
They found that all the tested molecules, modeling both the backbone and end groups, reversibly adsorbed with little chemical reaction with the a-C:H surface. Lee, Zube, et al. (1993) postulated that end-group chemistry dominates the adsorption process of the lubricant and provides an anchor to the surface. Perry and Somorjai (1995) have not found any evidence for chemical bonding of the alcohol end group to the carbon surface. In contrast, they emphasized that studies of the desorption of monodispersed fractions of Z-DOL lubricant demonstrated that desorption begins in the temperature range of decomposition, suggesting that physisorption energies are sufficient in bonding the lubricant to the surface at temperatures less than 377° C. Increases in desorption energies were observed with increasing chain length and classified as purely van der Waals interactions. Incorporation of alcoholic end groups promoted hydrogen bonding to the surface and produced about a 20 kJ/mol increase in desorption energy relative to that of a perfluorinated alkane of similar chain Length. Ether linkages within the model lubricant promoted little increase, as fluorine substituents effectively screen the oxygen. Waltman et al. (1998) showed that the evaporation of Z-DOL lubricant from the disk has an activation energy of 5.4 kcal/mole; Z-DOL bonding to the disk is relatively easy with an activation energy of 3.6 kcal/mole. Evidence for Z-DOL lying flat 3n the carbon surface is obtained from the thickness dependence of the dispersive component of the surface energy for both mobile and bonded Z-DOL. The major interactions leading to this conformation are between the oxidized sites on the carbon and the hydroxyl end groups of the lubricant. They also indicated that the model reaction between Z-DOL and the carboxylic acid terminated PFPE lubricant (Z-DIAC) raises the possibility that esterification reaction to form a chemisorbed lubricant may occur at elevated temperature.
Hara et al. (1991) investigated the bonding feature among a carbon overcoat film, atomic level contaminations, and perfluoropolyether lubricant with thermal energy (activation energy) as a parameter. They found that thermal energy significantly affects the bonding feature of the lubricant; however, no explanation of the mechanism was provided. The emphasis was made on the fact that removal of the contamination reduced the coefficient of friction and increased media durability. Z-DOL containing hydroxyl groups strongly interacts with dangling bonds and functional groups on the carbon surface, apart from the overcoat interaction with the lubricant ether groups. Additionally, hydroxyl groups of the Z-DOL molecule can form hydrogen bonding with oxygen and hydrogen atoms of the carbon surface film functional groups. The most recent work (Zhao et al., 1999a) presents interaction details of Z-DOL lubricant and adsorbed water film on the head-disk interface. It is also emphasized that heating enhances the bonded Z-DOL film significantly. It has also been shown that prebaking at 150 °C doubles the bonded film of Z-DOL lubricant (Zhao et al., 1999b). This effect is accounted for in terms of the water molecules' removal from the carbon overcoat, thereby exposing more dangling bonds and the surface functional groups to an interaction with the lubricant molecules. A schematic presentation of this effect is depicted in Fig. 10 (Zhao et al., 1999b).

4. DEGRADATION MECHANISMS OF PFPE LUBRICANTS

4.1. Introduction

A PFPE lubricant deposited on thin hard carbon coatings under static conditions interacts with the surfaces mostly through physical adsorption forces. The physically adsorbed lubricant film can, however, also be chemically bonded by a specific treatment of the deposited film. The known approaches to produce chemically bonded films encompass exposure of the lubricated disk to specific radiation modes, such as
  1. low-energy X-ray (Heideman and Wirth, 1984),
  2. nitrogen plasma (Homola et al., 1990),
  3. high-energy ion beam (Lee, 1991), and
  4. low-energy electrons and far ultraviolet (Vurens et al., 1992).
The effect of heating on Z-DOL bonding on carbon overcoats was discussed in subsection 3.2.2. It should be emphasized at this point, however, that the length of the thermal treatment and the temperature for PFPE lubricants with polar end groups are important factors in determining the bonding level (adsorption strength) of lubricants to the disk surface (Zhao and Bhushan, 1996).
During sliding both physically adsorbed and chemisorbed PFPE lubricant molecules undergo degradation. The degradation process is very complex. Although there are a number of papers aiming at accounting for the PFPE degradation mechanism (e.g., see Zhao et al., 1999c), it is still not yet fully understood. Broad, detailed, and sophisticated research resulted in the development of several approaches. They include
  1. thermal decomposition processes,
  2. catalytic degradation processes,
  3. electron mediated degradation, and
  4. mechanical degradation by a shearing process.
The following subsections will discuss all these approaches to the PFPE degradation mechanism.

4.2. Thermal Decomposition

PFPE's as a class of fluid lubricants exhibit excellent thermal and oxidative stability. Notwithstanding the fact that they are thermally stable up to 350 °C (Helmick and Jones, 1990), there is clear evidence that their actual thermal stability is adversely affected by the presence of metals. Jones et al. (1983) demonstrated the effects of steel and some titanium alloys and additives on the thermal stability of linear perfluoroalkyl ether fluids. The presence of metal alloys drastically increases the rate of degradation. Two inhibitors, a perfluoroalkyl ether substituted monophospha-s-triazine and a perfluorophenylphosphine, were found to be highly effective in degradation reduction at 288 °C, but they had only limited effectiveness at 316 °C. Results of another work (Zehe and Faut, 1989) showed that complete degradation of Fomblin Z takes place at 185 °C in the presence of iron oxide (Fe2O3).
• Dangling bond; — Covalent bond; « Hydrogen bond interaction; (x) Functional group
    Fig. 10.
  1. Adsorption of water molecules on a DLC surface exposed to ambient environment. R is an alkyl group, e.g., —CH3 group;
  2. adsorption of Z-DOL molecules on a DLC surface exposed to ambient environment;
  3. DLC surface after prebaking at 150 °C; water molecules are desorbed; and
  4. bonding of Z-DOL molecules on a DLC surface after postbaking at 150 °C for 2 h.

Interestingly, such degradation was not observed with the Krytox lubricant. The complete degradation of Fomblin Z was ascribed to the presence of acetal groups (—O—CF2—O—) in the polymeric chain, as suggested earlier by Jones et al. (1983).

Under sliding conditions, particularly at high sliding speeds, the temperature on hard carbon coatings can be sufficiently high to initiate PFPE thermal degradation (Bhushan, 1992). For instance, Huu et al. (1997), investigating friction and wear of hard carbon coatings at sliding speeds in the range 30-35 m/s in air, presented evidence that the temperature increase produced by friction in the contact zone was high, about 285 °C. The transition from the sp³ to sp² phase occurs at this temperature. As the transition consumes a large part of the friction energy in the contact, apart from creating a thin graphite layer on the track, it is also reasonable to suggest that in such a specific situation the thermal degradation of PFPE lubricant can take place. A schematic presentation of this most simple free-radical PFPE degradation mechanism is depicted in Fig. 11.
Fig. 11. Thermal degradation process of PFPE lubricants (temp. >350 °C); the free radicals can react with PFPE molecules and oxygen molecules, recombine with other free radicals, or decompose.

The presence of oxygen in the environment will lead to the thermal oxidative degradation process in which a very reactive peroxide radical can be generated.

4.3. Catalytic Degradation

As a-alumina exists as the structural element in the construction of a head slider for a magnetic disk drive, the degradation mechanism of PFPE's by a-alumina was examined (Kasai et al., 1991). The results obtained led to a followup experiment of the degradation process in the presence of AlCl3, and at higher temperatures (Kasai et al., 1991; Kasai, 1992). Although a-alumina is a typical Lewis acid, A1C13 is a stronger Lewis acid. According to Kasai (1999), the above-mentioned studies conclusively demonstrated that the degradation of PFPE's in the presence of Lewis acids (metal oxides or halides) is singularly dominated by the intramolecular disporportionation reaction,
R1—CF2—O—CF2—O—CF2—R2

→ R1—CF3 + FC(O)—R2

where R1 is the left part of Z-DOL molecule from the above-shown acetal sector, and R2 is the right portion of Z-DOL molecule.

This reaction occurs most readily at acetal sector (O—CF2—O), but, given sufficient activation energy or in the presence of a stronger Lewis acid, it also occurs at other linkages. It was also found that, when the reaction occurred at the terminal linkage, the fluorine transfer was always from the terminal group into the internal sector (Kasai, 1999).
The degradation mechanism of a Z-type lubricant in the presence of alumina has been characterized as follows (Kasai et al., 1991):
  1. it is catalyzed by A1F3 formed during the induction period;
  2. it uniquely involves the acetal moiety (—O—CF2—O) of the polymer chains;
  3. it generates methoxy end groups (—O—CF3), and
  4. it results in fragmentation of polymer chains but does not involve an unzipping process.
Figure 12 shows schematically the reaction mechanism of the catalyzed degradation mechanism of a Z-type lubricant. The reaction sequence is initiated by a bidentate linkage between an acidic aluminum on A1F3 and the two oxygen atoms of an acetal unit, as presented in the upper part of Fig. 12.
Fig. 12. Interaction between a Lewis acid site on AlF3 and an acetal sector of Z-lube (Kasai et al., 1991).
The partial positive charge developed at the acetal carbon induces a fluorine atom transfer from the adjacent CF2 unit leading to chain scission with transformation of the acetal unit into a methoxy (—O—CF3) end group, and the adjacent unit into either a fluoroformate F—CO—O—CF2 end group or an acylfluoride F—CO—CF2 end group from the adjacent unit, depending upon whether the adjacent unit was originally a methylene oxide unit or an ethylene oxide unit (Kasai et al., 1991).

The catalytic degradation mechanism of PFPE lubricants is very well elaborated and clearly presented in the papers discussed above. Illustrative weight loss of all major PFPE lubricant types during heating in the presence of various catalysts, depicted in Fig. 13, provides excellent information concerning the thermocatalytic stability of these lubricants. The degradation was induced by placing 5 g of lubricant and 1 wt. % of Al2O3 (or AlCl3) in a test tube, and immersing the tube in an oil bath maintained at a desired temperature. For each lubricant, the weight loss was measured after the heat treatment for a given period of time (Kasai, 1999). Looking at the modeled degradation processes presented in Fig. 13, from the viewpoint of the head-disk lubricated situation, we see that Fig. 13 (a) seems to be the most relevant to the real world. Under these test conditions, Al2O3/200°C all the investigated lubricants but the Z-type lubricant are very stable. In contrast, even the least stable lubricant starts to degrade after 50 min of heating. At this point the question arises whether such a situation can be attained under CSS sliding and flying conditions. We do not think so, as the test conditions for their experiments are extreme and cannot be believed to be achieved at a head-disk interface. However, note that Kasai (1999) reported some similarities between PFPE lubricant fragments released from the head-disk interface with volatile products generated from Z-type lubricants undergoing an alumina-catalyzed degradation process at 200 °C and head-disk interface studies conducted in UHV.
    
    Fig. 13. Weight loss observed on PFPE lubricants when heated in contact with
  1. Al2O3 at 200 °C,
  2. AlCl3 at 200 °C, and
  3. AlCl3 at 250 °C (Kasai, 1999).

    
Fig. 14. Schematic presentation of the catalyzed PFPE degradation process.
The result of this comparison was concluded as follows: "What is most significantly revealed here is that the lubricant degradation in the disk environment is also singularly dominated by the Lewis acid catalyzed disproportionation reaction, and no other reaction of consequence is operative" (Kasai, 1999). Although the considered similarities are very interesting and might provide some evidence for the alumina-catalyzed degradation mechanism of a Z-type lubricant, we will also show that another mechanism accounts for all the UHV-generated lubricant fragments (see subsection 5.2). Figure 14 schematically summarizes the catalytic degradation process of PFPE lubricants.
Recently, Fukuchi (1999) investigated the chemical degradation of major PFPE lubricant fluids, such as Z-DOL, Fomblin AM-2001, Z-25, Demnum SP-3, and Demnum SA, with bivalent metal ions. These lubricants were allowed to react with a variety of metal sulfates under identical heating conditions, and the resulting fluoride ion concentrations were measured. The respective reaction mechanisms involve an ether cleavage that is caused by the coordination of metal ions with oxygen atoms in the main chain of the PFPE. Each lubricant with a given sulfate was heated at 250 °C in an oven. Heating times were 30, 60, 90, or 150 min. In terms of the ability to degrade a lubricant, metal ions were evaluated in sequence from Cu+2 > Ni+2 > Co+2 > Fe+2 > Mn+2 > Mg+2. This trend was found to be explained by the ionization potentials of the metal atoms. For an identical metal ion, the extent of degradation was somewhat higher with Fomblin than with Demnum lubricants.

4.4. Electron-Induced Degradation Mechanism

It is well known that a wide variety of materials, after mechanical treatment, become electron emitters. The term "exoelectron emission" relates to low-energy electrons and originated from the investigation of the emission in freshly formed metals, which were accounted for as a consequence of an exothermal transformation process of the surface (Kramer, 1950). Figure 1 demonstrates that exoelectron emission occurs when a material's surface is disturbed by plastic deformation, abrasion, fatigue cracking, or phase transformation. The energy of exoelectrons is considered to be very low, of the order of 1-3 eV (Chaikin, 1967; Dickinson et al., 1997). Wei et al. (1998) reported that the energy of triboelectrons ranges form 0.5 to 80 eV, depending on the materials and test conditions. However, there is no reference given either to who measured the energy of these electrons or to the apparatus or technique used. There is also no information related to the mentioned materials and test conditions. Grunberg (1958) reviewed the exoelectron emission phenomena. Kajdas (1994) discussed the importance of low-energy electrons for the initiation of tribochemical reactions, proceeding according to the NIRAM approach. The principal thesis of the model is that lubricant molecules, for example, alcohols (Kajdas, 1987), form anions that are then chemisorbed on the positively charged areas of rubbing surfaces. Vurens et al. (1992) examined the bonding of the thin PFPE films by low-energy electrons and UV. They provided evidence that the bonding takes place through an interaction of the perfluoropolyether molecule, with a low-energy photoelectron emitted by excitation of the substrate by the UV photons, and demonstrated that for a Demnum S200 film on a diamondlike carbon substrate bombarded with low-energy electrons, the most abundant fragment ions comprise the following negative species: F¯, C3F5O2¯, and C3F7O¯. Therefore, the electron mediated degradation mechanism of perfluoropolyether lubricants seems to be of particular importance.
There is no reference that directly refers to triboemission of the head-disk interface. However, Nakayama, Yamanaka, et al. (1997) simulated the head-disk interface by using the friction system of alumina sliding on a-C:H films and found that both negatively charged particles and positively charged particles are emitted under friction in ambient air. Matsunuma et al. (1996), using empirical molecular orbital (MO), simulation, showed that an electron attachment to a PFPE model leaving a methylene oxide segment reduced the bond order of the C—O bond. Thus the cleavage of the weakest C—O bond of the anion radical produced an anion and a neutral radical easily. This is in line with the NIRAM approach (Kajdas, 1987; Kajdas, 1994). Previously, this approach was also successfully applied in accounting for the vinyl tribopolymerization process in both metal-on-metal systems (Kajdas et al., 1993) and ceramic-on-ceramic systems (Furey et al., 1997) Therefore, it is reasonable to suggest that anionic intermediates play an important role in the electron-induced degradation process of PFPE lubricants. A schematic presentation of the low-energy electron-induced PFPE degradation process is shown in Fig. 15.

4.5. Mechanical Degradation

The application of an adequate shear stress results in the ordering of the PFPE chain and a corresponding
Fig. 15. Schematic showing the electron-induced PFPE degradation process:
  1. free radicals can recombine with other radicals, react with PFPE molecules, decompose, or interact with low-energy electrons to form other anions;
  2. anions react with the positively charged sites of the surface.

Fig. 16. PFPE degradation process caused by shearing.

decrease in friction (Hirz et al., 1992). Under operating conditions, the removal of PFPE lubricant molecules from sliding and flying tracks, comprises its displacement and loss. The removal rate depends on film thickness, chemical structure, and molecular weight. Microscale friction and properties of molecularly thin liquid films most recently were discussed in detail by Bhushan (1999b). In flying and sliding, the lubricant removal rate from monolayer films is significantly slower than from multilayer films (Novotny and Baldwinson, 1991). The lubricant loss in sliding can be due to polymer scission (Buchholz and Wilson, 1986). Direct mechanical scission at shear rates of 108-1010 s -1; is possible (Novotny and Baldwinson, 1991; Bhushan, 1999b); however, other factors can contribute to the lubricant degradation as well. Mechanical scission of PFPE bonds leads to the formation of free radicals, similarly as in the thermal degradation process. Figure 16 provides a schematic presentation of the PFPE degradation process caused by high shear stresses.

5. DEGRADATION OF PFPE LUBRICANTS UNDER SLIDING CONDITIONS

5.1. General Information

Under disk-drive operation conditions, the head slider slides on the disk surface during start and stop operations. CSS operations lead to elastic and plastic deformation of asperities and some wear. The generated energy impacts the lubricant molecules, which are both physically and chemically adsorbed. The adsorbed and chemically bonded PFPE molecules undergo important changes caused by physical and chemical processes. Physical changes include the mechanical displacement of the molecules and their possible desorption or evaporation. Chemical changes are much more complex and can proceed according to different mechanisms, as discussed in Section 4. The common denominator of all these specific mechanisms is the lubricant degradation process leading to its loss, mostly in the form of gaseous species generated at the sliding contacts. Currently, for detection of gaseous species generated by sliding in a high vacuum chamber, the most widely used measurement technique relates to quadrupole mass spectrometry. Data concerning the detection of gaseous wear species from lubricated thin-film disks have been presented by Vurens et al. (1992), Strom et al. (1993), Novotny et al. (1994), Bhushan and Ruan (1994), Bhushan and Cheng (1997), and Li and Bhushan (1997).

5.2. Discussion on PFPE Lubricant Degradation Mechanisms

Novotny et al. (1994) applied high-resolution mass spectrometry to monitor in situ gaseous wear products generated during the sliding of lubricated carbon-carbon interfaces in a vacuum. Lubricant loss, alteration, and removal as precursors of the wear of solids were determined by scanning microelipsometry. The amount of altered lubricant that was left on the disk and slider surfaces was quantified by X-ray photelectron spectroscopy (XPS), and the chemical structure of altered lubricant was evaluated by time-of-flight mass spectrometry. It was believed that the dominating process, before the wear of solids occurs, is the tribochemical scission of fluorocarbon polymers. In other words, the processes occurring at slider-disk interfaces take place before a microscopic wear of solids, excluding local microscopic wear at asperities. Novotny et al. (1994) also compared the mass spectra obtained during sliding experiments with those obtained in thermal desorption and found that the spectrum during sliding at 0.5 m/s resembled the thermal spectrum at 300 °C rather than that at 90 °C. Thus, they speculated that the loss mechanism of lubricants could be of two types: either transient interfacial temperatures were high enough to evaporate lubricant from the surface, or polymer scission produced fragments small enough to be gaseous. It is of note at this point, however, that according to the work by Bair et al. (1991) and Bhushan (1992), local increases in temperature at contact spots are less than 100 °C for typical test conditions. From the mass spectrum of wear fragments and thermal and electron decomposition products of PFPE lubricants, Vurens et al. (1992) concluded that the degradation of PFPE lubricants in sliding is an electron mediated process, since most of the product fragments generated by low-energy electrons were comparable with that in sliding experiments.
Li and Bhushan (1997) studied the degradation process of Z-DOL and Fomblin AM2001 lubricants on the thin-films disks with an Al2O3-TiC slider under high vacuum. Gaseous products generated from the head-disk interfaces were detected and monitored as a function of sliding distance by using a quadrupole mass spectrometer. They found that the gaseous products generated from these two lubricants are quite similar. Decomposition of both lubricants was found to take place from the beginning of sliding, and the evolution rate of these gases decreased with the sliding time. Surface roughness of the disk has an effect on the sliding distance to failure, but little effect on the decomposition of lubricant.
Using the same setup, Bhushan and Cheng (1997) investigated three kinds of lubricants with nonpolar and polar ends and with and without thermal treatment, which were applied over a thin-film disk with DLC overcoat. They found that fluorocarbon fragments are generated from lubricants during a period of sliding with a low and stable coefficient of friction, followed by a sharp rise in frictional force and generation of gaseous products of DLC overcoat material. It was concluded that decomposition of PFPE lubricants begins from the onset of sliding, and the DLC overcoat starts to wear simultaneously from the beginning of the sliding for untreated lubricants but was well protected in the case of chemically bonded lubricants. Detailed degradation mechanisms of PFPE lubricants are the most complex and arguable issue. For example, Wei et al. (1998) applied chemical bonding theory to analyze the degradation process.
Most recent high-vacuum test results (Zhao and Bhushan, 1999) demonstrate that the degradation process of the 1.5-nm-thick Z-DOL lubricant film generates CFO, HCF2, and CF species during sliding. CF2O fragments have not been detected. Another extensive work (Zhao et al., 1999c) also includes the Z-DOL lubricant degradation process and clearly shows the importance of both CFO and CF2O species along with HCF2, CF3, and other fragments. The formed Z-DOL degradation fragments are accounted for in terms of the electron-induced degradation mechanism. For example, to account for the CF2H and CFCF3 Z-DOL degradation fragments detected in mass spectra under UHV sliding conditions, the following reactions are proposed (Zhao et al., 1999c):
—CF2—CF2—O—CF2—O—CF2— + e →

—CF2—CF2—O¯ + •CF2—O—CF2

A¯ (anion) R• -(free radical) (1)

The negative ion A¯ reacts with the DLC surface. The formed free radical R• can further react in various ways. Some reactions follow, which lead to the formation of two species HCF2and CF3 found in mass spectra:
R• + H• → HCF2—O—CF2—
Compound I
(2)

R• + Z-DOL → CF3—O—CF2— + [Z—DOL—F]•
Compound II
(3)

Compounds I and II, interacting with emitted low-energy electrons, generate the fragments detected in mass spectra in the following reactions:
HCF2—O—CF2— + e → •CF2H + ¯O—CF2—
(4)

CF3—O—CF2— + e → •CF3 + ¯O—CF2—
(5)

Negative ions formed in reactions (4) and (5) react with the DLC positively charged spots generated during sliding. There are several possibilities to produce CF and CF2 fragments during the Z-DOL degradation process. One possibility to generate CF2 species may relate to the decomposition of CF2CF2O species into CF2 and CF2O by friction shear as well as electron impact (Wei et al., 1998). CF2 may be further cleaved into the F atom and CF species.
The detailed Z-DOL lubricant degradation mechanism, as described by Zhao et al. (1999c), emphasizes the importance of the negative-ion-radical action mechanism for the PFPE lubricant degradation process. The experimental results of that study also show that the decomposition of the lubricant molecule is greatly affected by the mechanical shearing. The Z-DOL lubricant molecule presents a linear long chain structure, including mostly weaker C—O and C—C bonds; the lubricant molecules are easily subjected to cutting by the microasperities on the rubbing surfaces, which results in the decomposition of the lubricant in sliding process. No obvious effect of temperature on the decomposition of the lubricant was observed in the test described by Zhao et al. (1999c).
Step 1 Low-energy electron emission process (1-4 eV) during sliding
Step 2 Attachment of emitted electron to lubricant molecule
Z-DOL + e → —CF2—CH2O¯ + H•
(Reactive species)
Step 3 Chemical reactions with the surface
Bonded molecule
H• species may recombine and generate hydrogen molecules H2 or they can interact with the dangling bonds
Bonded hydrogen atom
    Fig. 17. Interaction of emitted electrons with Z-DOL lubricant molecules and subsequent interactions with the surface (Zhao et. al., 1999b).

Interestingly, the NIRAM approach can also be applied in accounting for experimental results concerning chemical bonding of PFPE lubricant films with DLC under sliding conditions. The most recent research (Zhao et al., 1999b) demonstrated how low-energy electrons may attach to the PFPE lubricant molecules and allow the chemical reactions between lubricant molecules and a DLC surface. Figure 17 depicts the steps through which lubricant Z-DOL molecules are chemically bonded to the DLC surface after sliding. Under sliding conditions, low-energy electrons are emitted at the interface (Step 1); these electrons then attach to lubricant molecules forming reactive species (Step 2). The generated negative ions ( —CF2—CH2O¯) react with the surface and form bonded molecules (Step 3). Hydrogen atoms (free radicals) may recombine and produce hydrogen molecules H2, or they can interact with dangling bonds on the surface. Figure 18 describes the steps through which lubricant Z-15 or Z-25 molecules are chemically bonded to the DLC surface after sliding. Step 1 is similar to that in Fig. 17. The low-energy electrons then attach to lubricant molecules and form the reactive species (Step 2). The produced negative-ion reactive species allow chemical reaction with the surface and form bonded molecules; free-radical species can react with the surface dangling bonds (Step 3a). Free radicals can also recombine, forming a PFPE molecule, or they can interact with Z-15 lubricant molecules, producing a smaller PFPE compound and generating another free radical [Z-15-F}•. The latter radical can recombine with •CF2- free radical, generating a higher molecular-weight lubricant compound. More research is needed, however, to better understand all these processes.
Step 1 Low-energy electron emission process (1-4 eV) during sliding
Step 2 Attachment of emitted electron to lubricant molecule
—CF2—O—CF2— + e → —CF2—O¯ +•CF2—
Step 3 Further Chemical reactions a) with the surface
Bonded molecule

Bonded molecule
b) between lubricant fragments
•CF2— + •CF2— → —F2C—CF2—

Z15 + •CF2— → [Z-15—F]• + CF3—

    Fig. 18. Interaction of emitted electrons with Z-15 or Z-25 lubricant molecules and subsequent interactions with the surface (Zhao et al., 1999b).

6. CONCLUDING REMARKS

This review paper is mostly focused on interaction and degradation mechanisms of PFPE lubricants with DLC protective overcoats for magnetic media. Major stress is put on the degradation process of these lubricants under sliding conditions. A detailed discussion of various degradation mechanisms of Z-DOL lubricant is provided, which includes thermal decomposition, catalytic degradation, electron-mediated degradation, and mechanical degradation processes. Comparing various degradation mechanisms, we suggest that the catalytic degradation mechanism proposed by other authors is not relevant, because kinetics is slow at asperity temperatures. Based on a wide variety of reviewed or discussed experimental data, it is reasonable to emphasize that anionic intermediates (negative ions and/or negative-ion-radical species) produced by low-energy electrons play an important part in both
  1. the electron-mediated degradation process of PFPE lubricants, and
  2. chemical bonding of PFPE lubricant films with DLC surfaces under sliding conditions.
This is in line with the NIRAM approach.

ACKNOWLEDGMENT

The authors acknowledge Dr. Zheming Zhao for his help in figure preparation and for fruitful discussions. Financial support for this research was provided by the industrial membership of the Computer Microtribology and Contamination Laboratory.

REFERENCES

  1. Bair, S., Green, L., and Bhushan, B. (1991), "Measurement of Asperity Temperatures of a Read/Write Head Slider Bearing in Hard Magnetic Recording Disks," ASME J. Tribol. 113, 547-554.
  2. Bhushan, B. (1992), "Magnetic Head-Media Interface Temperatures. Part 3: Applications to Rigid Disks," ASME J. Tribol., 114, 420-430.
  3. Bhushan, B. (1994), "Tribology of Magnetic Storage Systems," in CRC Handbook of Lubrication and Tribology (E. R. Booser, ed.), Vol. III, pp. 325-371, CRC Press, Boca Raton, Florida.
  4. Bhushan, B. (1996), Tribology and Mechanics of Magnetic Storage Devices, 2nd ed., Springer-Verlag, New York.
  5. Bhushan, B. (1999a), "Chemical, Mechanical and Tribological Characterization of Ultra-Thin and Hard Amorphous Carbon Coatings as Thin as 3.5 nm: Recent Developments," Diamond and Related Materials, 8, 1985-2015.
  6. Bhushan, B. (1999b), Principles and Applications of Tribology. Chap. 11, Wiley, New York (Micro/Nanotribology).
  7. Bhushan, B. and Cheng, Y. (1997), "Wear and Degradation Mechanisms of Magnetic Thin Film Rigid Disks with Different Lubricants Using Mass Spectrometry," J. Appl. Phys., 81, 5390-5392.
  8. Bhushan, B. andRuan,). (1994), "Tribological Performances ol Thin Film Amorphous Carbon Overcoats for Magnetic Recording Disks in Various Environments," Surface Coating Technol., 68/69, 644-650.
  9. Bhushan, B. and Zhao, Z. (1999), "Macroscale and Microscale Tribological Studies of Moleculary Thick Boundary Layers of Perfluoropolyether Lubricants for Magnetic Thin-Film Rigid Disks." J. Info. Storage Proc. Syst., 1, 1-21.
  10. Bonder, G. M. and Pardue, H. L. (1989), Chemistry-An Experimental Science, Wiley, New York.
  11. Buchholz, F. L. and Wilson, L. R. (1986), "High Shear Rheology and Shear Degradation of Aqueous Polymer Solutions," J. Appl. Poly. Sci., 32, 5399-5413.
  12. Chaikin, S. W. (1967), "On Frictional Polymer," Wear-10, 49-60.
  13. Cho, N. H., Krishnan, K. M., Veirs, D. K., Ruhin, M. D.. Hopper, C. B., Bhushan, B., and Bogy, D. B. (1990), "Chemical Structure and Physical Properties of Diamond-Like Amorphous Carbon Films Prepared by Magnetron Sputtering," J. Mat. Res., 5, 2543-2554.
  14. Del Pesco, T. W. (1994), "Perfluoroalkylpolyethers" in CRS Handbook of Lubricant and Tribology (E. R. Booser, ed.), Vol. Ill pp. 287-303, CRC Press, Inc. Boca Raton, Florida.
  15. Dickinson, J. T, Scudiero, L., Yasuda, K., Kim, M-W., and Langford, S. C. (1997), "Dynamic Tribological Probes: Particle Emission and Transient Electric Measurements," Tribol. Lett., 3, 55-67.
  16. Dillon, R. 0., Woolman, J. A., and Katkanant, V. (1984), "Use of Raman Scattering to Investigate Disorder and Crystallite Formation in As-Deposited and Annealed Carbon Films," Phys. Rev. B29, 3482-3489.
  17. Enke, K. (1981), "Some New Results on the Fabrication of and the Mechanical, Electrical and Optical Properties of i-Carbon Layers." Thin Solid Films 80, 227-234.
  18. Fukuchi, T. (1999), "Degradation of Perfluoropolyether Fluids with Metal Ions," ASME J. Tribol. 121, 348-351.
  19. Furey, M. J., Kajdas, C., Kempinski, R., and Tripathy, B. S. (1997), "Tribochemistry of Vinyl Monomers Under Boundary Lubrication of Alumina-on-Alumina System," Lubr. Sci. 10, 3-25.
  20. Grunberg, L. (1958), "A Survey of Exoelectron Emission Phenomena," Br. J. Appl. Phys. 9, 85-93.
  21. Gschwender, L. ).. Snyder, C. E. and Fultz, G. W. (1992). "Soluble Additives for Perfluoropolyalkylether Liquid Lubricants." Lub. Eng. 49, 702-708.
  22. Gumprecht, W. H. (1966), "A New Class of High-Temperature Fluids," ASLE Trans. 9, 24-30.
  23. Hara, H., Kubota, M., Enolo, I., and Ohmae, N. (1991), "Effect of Lubricant Durability of Thin Film Disk Media," Adv. Info. Storage Syst. 2, 587-592.
  24. Heideman, R. and Wirth, M. (1984), "Transforming the Lubricant on a Magnetic Disk into a Solid Fluorine Compound," IBM Tecb. Disclosure Bull. 27, 3199-3205.
  25. Helmick, L. S. and Jones, W. R. (1990), "Deformation of the Thermal Stability of Perfluoroalkylethers," NASA memo TM-102493, NASA, Washington, D.C.
  26. Hirz, S. I., Homola, A. M., Hadziiannou, G., and Frank, C. W. (1992), "Effect of Substrate on Shearing Properties of Ultrathin Polymer Films," Langmuir, 8, 328-333.
  27. Homola, A. M., Lin, L.}. and Saperstein, D. D. (1990), "Process for Bonding Lubricant to a Thin Magnetic Recording Disk," U.S. Patent 4,960,609, October 2.
  28. Huu, T. Le, Zaidi, H. and Paulmier, D. (1997), "Friction and Wear Properties of Hard Carbon Coatings at High Sliding Speed," Wear 203-204, 442-446.
  29. Jansen, F., Machonkin, M., Kaplan, S., and Mark, S. (1985), "The Effects of Hydrogenation of the Properties of Ion Beam SputterDeposited Amorphous Carbon," J. Vac. Sci. Technol. A3, 605-609.
  30. Janzen J. (1982), "Physichemical Characterization of Carbon Black," Chem. Technol. 55, 669-706.
  31. Jones, W. R., Jr. (1995), "Properties of Perfluoropolyethers for Space Applications," Tribol. Trans. 38, 557-564.
  32. Jones, W. R., Paciorek, K. J. L., Ito, T. I., and Kratzer, R. H. (1983), "Thermal Oxidative Degradation Reactions of Linear Perfluoroalkyl Ethers," Ind. Eng. Chem. Prod. Res. Dev. 22, 167-170.
  33. Jones, W. R., Jr. and Snyder, C. E., Jr. (1980), "Boundary Lubrication, Thermal and Oxidative Stability of A Fluorinated Polyether and Perfluoropolyether Triazene," ASLE Trans. 23, 253-261.
  34. Kajdas, C. (1987), "About an Anionic-Radical Concept of The Lubrication Mechanism of Alcohols," Wear 116, 167-180.
  35. Kajdas, C. (1994), "Importance of Ionic Reactive Intermediates For Lubricant Component Reactions With Friction Surfaces," Lubr. Sci. 6, 203-228.
  36. Kajdas, C. (1997), "Industrial Lubricants," in: Chemistry and Technology of Lubricants R. M. Mortier and S. T. Orszulik (eds.), Chap. VIII, Blackie Academic & Professional, London.
  37. Kajdas, C., Lafleche, P. M., Furey, M. J., Hellgeth, J. W., and Ward, T.C. (1993), "A Study of Tribopolymerization under Fretting Contact Conditions," Lubr. Sci. 6, 51-89.
  38. Kaplan, S., Jansen, F., and Machonkin, M. (1985), "Characterization of Amorphous Carbon-Hydrogen Films by Solid-State Nuclear Magnetic Resonance," Appl. Phys. Lett. 47, 750-753.
  39. Kasai, P. H. (1992). "Perfluoropolyethers: "Intramolecular Disproportionation," Macromolecules 25, pp. 6791-6799.
  40. Kasai, P. H. (1999), "Degradation of Perfluoropolyethers and Role of X-1P Additives in Disk Files," J. Info. Storage Proc. Syst. 1, 22-31.
  41. Kasai, P. H., Tang, W. T, and Wheeler, P. (1991), "Degradation of Perfluoropolyethers Catalyzed by Alumina Oxide," Appl. Surf. Sci. 51, 201-211.
  42. Kramer, J. (1950), Der metallische Zustand, Vandenhock and Ruprecht, Goettingen, Germany.
  43. Lee, T. D. (1991), "Enhanced Chemical Bonding Between Lubricants and Magnetic Coatings by High Energy Irradiation and its Application to Tape Durability," J. Vac. Sci. Technol. A9, 1287-1292.
  44. Lee, J. K., Smallen, M,, Enguero, J., Lee, H. J., and Chao, A. (1993). "The Effect of Chemical and Surface Properties of Hydrogenerated Carbon Overcoats on the Tribological Performance of Rigid Magnetic Disks," IEEE Trans. Magnet. 29, 276-281.
  45. Lee, H. J., Zube, D., Hollar, D., Lee, J. K., Chao, A., and Smalinn, M. (1993), "Enhanced Tribological Performance of Rigid Disk by Using Chemically Bonded Lubricant," J. Vac. Sci. Technol. A13, 711-714.
  46. Li, Y. and Bhushan., B. (1997), "Degradation of Perfluoropolyether Lubricants on Magnetic Recording Disks During Sliding," Proc. Inst. Mech. Eng. Part J, 211, 279-288.
  47. Matsunuma, S., Miura, T., and Kataoka, H. (1996), "Theoretical Simulations of Electron-Induced Degradation of Perfluoropolyether," Tribol. Trans. 39, 380-385.
  48. Miller, D. J. and McKenzie, D. R. (1983), "Electron Spin Resonance Study of Amorphous Hydrogenated Carbon Films," Thin Solid Films 108, 257-264.
  49. Nakayama, K., Bou-Said, B., and Ikeda, H. (1997), "Tribo-Electromagnetic Phenomena of Hydrogenated Carbon Films: Tribo-Electrons, -Ions, -Photons, and -Charging," ASME ]. Tribol. 119, 764-768.
  50. Nakayama, K., Yamanaka, K., Ikeda, H., and Sato, T. (1997), "Friction, Wear, and Triboelectron Emission of Hydrogenated Amorphous Carbon Films," Tribol. Trans. 40, 507-513.
  51. Nakayama, K. and Ikeda, H. (1996), "Triboemission Characteristics of Electrons During Wear of Amorphous Carbon and Hydrogeneted Amorphous Carbon Films in A Dry Air Atmosphere," Wear 198, 71-76.
  52. Novotny, V. J. and Baldwinson, M. A. (1991), "Lubricant Dynamics in Sliding and Flying," J. Appl. Phys. 70, 5647-5652.
  53. Novotny, V. J., Pan, X., and Bhatia, C. S. (1994), "Tribochemistry at Lubricant Interfaces," J. Vac. Sci. Technol. A12, 2879-2886.
  54. Perry, S. S. and Somorjai, G. A. (1995), "The Interaction of Short-Chain Model Lubricants with the Surfaces of Hydrogenated Amorphous Carbon Films," Tribol. Lett. 1, 47-58.
  55. Saperstein, D. and Lin, L. (1990), "Improved Surface Adhesion and Coverage of Perfluoropolyether Lubricants Following for-UV Irradation," Langmuir 6, 1522-1524.
  56. Silva, S. R. P., Amaratunga, A. J., and Constantinou, C. P. (1992), "Optical Properties of Amorphous Carbon/Diamond Thin Films." J. Appl. Phys. 72, 1149-1153.
  57. Simonds, J. L. (1995), "Magnetoelectronics Today and Tomorrow," Phys. Today April, 6-32.
  58. Strom, B. D., Bogy, D. B., Walmsley, R. G., and Brandt, J. (1993), "Gaseous Wear Products from Lubricated Thin Disk," IEEE Trann. Magnet MAG-29, 253-258.
  59. Torng., C. J. and Silvertsen, J. M. (1990), "Structure and Bonding Studies of the C:N Thin Films Produced by rf Sputtering Method." J. Mat. Res. 5, 2490-2496.
  60. Vurens, G. H., Gudeman, C. S., Lin, L. J., and Foster, J. S. (1992), "Mechanism of Ultraviolet and Electron Bonding of PFPEs," Langmuir 8, 1165-1169.
  61. Wada, N., Gaczi, P. J., and Solin, S. A. (1980), "Diamond-like Threefold Coordinated Amorphous Carbon," J. Non-Cryst. Solid 35, 543-548.
  62. Waltman, R. J., Pocker, D. J., and Tyndall, G. W. (1998), "Studies on the Interactions between Z-DOL Perfluoropolyether Lubricant and The Carbon Overcoat of Rigid Magnetic Media," Tribol. Lett. 4, 267-275.
  63. Wang, R., Meeks, S. W., White, R. L., and Weresin, W. E. (1995). "The Effect of Hydrogen in Carbon Overcoats on The Tribology of The Head-Disk Interface," IEEE Trans. Magnet. 31, 2919-2921.
  64. Wang, R. H., White, R. L., Meeks, S. W., Min, B. G., Kellock, A., Homola, A., and Yoon, D. (1996), "Interaction of Perfluoropolyether with Hydrogenated Carbon," IEEE Trans. Magnet. 32, 3777-3779.
  65. Wei, J., Fong, W., Bogy, D. B., and Bhatia, C. S. (1998), "The Decomposition Mechanisms of a Perfluoro-polyether at the Head/Disk Interface of Hard Disk Drives," Tribol. Lett. 5, 203-209.
  66. Yanagisawa, M. (1994), "Adsorption of Perfluoro-Polyethers on Carbon Surfaces," in Tribology and Mechanics of Magnetic Storage Systems, Vol. 9, SP-36, 25-32, STLE, Park Ridge, Illinois.
  67. Zehe, M. J. and Faut, 0. D. (1989), "Acidic Attack of Perfluorinated Alkyl Ether Lubricant Molecules by Metal Oxide Surface," NASA Memo TM-101962, NASA, Washington, D.C.
  68. Zhao, Z. and Bhushan, B. (1996), "Effect of Bonded Lubricant Films on the Tribological Performance of Magnetic Thin-Film Rigid Disks," Wear 202, 50-59.
  69. Zhao, Z. and Bhushan, B. (1999), "Tribological Performance of PFPE and X-1P Lubricants at Head-Disk Interface. Part I. Experimental Results," Tribol. Lett. 6, 129-139.
  70. Zhao, Z., Bhushan, B., and Kajdas, C. (1999a), "Tribological Performance of PFPE and X-1P Lubricants at Head-Disk Interface. Part II. Mechanisms," Tribol. Lett. 6, 141-148.
  71. Zhao, Z., Bhushan, B., and Kajdas, C. (1999b), "Effects of Thermal Treatment and Sliding on Chemical Bonding of PFPE Lubricant Films with DLC Surfaces," J. Info. Storage Proc. Syst. 1, 259-264.
  72. Zhao, X., Bhushan, B., and Kajdas, C. (1999c), "Lubrication Studies of Head-Disk Interfaces in a Controlled Environment: Part II. Degradation Mechanisms of PFPE Lubricants," Proc. Inst. Mech. Eng., Part J (in press).