A

Va

b

a

ARR1A

KWTSW

1

1

firgeca

lt(ibicti

0d

Wear 267 (2009) 168–176

Contents lists available at ScienceDirect

Wear

journa l homepage: www.e lsev ier .com/ locate /wear

model for wet silicon carbide tribo-corrosion

. Pressera,∗, K.G. Nickela, O. Krummhauerb, A. Kailerb

Eberhard-Karls-Universität Tübingen, Institute for Geoscience, Wilhelmstraße 56, D-72074 Tübingen, GermanyFraunhofer-Institut für Werkstoffmechanik IWM, Wöhlerstraße 11, D-79108 Freiburg, Germany

r t i c l e i n f o

rticle history:eceived 26 September 2008eceived in revised form4 November 2008ccepted 14 November 2008

eywords:

a b s t r a c t

Silicon carbide sustains chemical and mechanical deterioration during tribological exposure under waterlubrication. In particular, tribochemical wear leads to the formation of a thin (tens to hundreds of nanome-tre) layer composed of nanoscale SiC wear debris embedded in a silica-like matrix (SiOxHy) with possiblysome minor oxycarbidic content. The SiC wear particles are plastically deformed and rounded as a resultof mechanical tribolapping. Below that layer, subsurface damage builds up in the form of dislocations,ruptures and shear cracks. As a result of plastic deformation (similar to indentation plasticity) SiC single

earribologyilicon carbideet-lubrication

crystals within that transition zone are transformed into mosaic crystals with smaller domains due to slipplane gliding. Comparing results for static hydrothermal conditions (using a hydrothermal diamond anvilcell) and tribotest for the mild and severe wear regime (i.e., with and without external cooling) we deriveda qualitative wear-model of wet silicon carbide tribo-corrosion. While mechanical contact yields path-ways for water inflow and generally disrupts the structural integrity of SiC grains, hydrothermal reactionsof trapped water and subsequent pressure relief leads to a mechanism of dissolution and reprecipitation.

bserv

The latter produces the o

. Introduction

.1. General aspects of wet SiC tribo-corrosion

Sintered silicon carbide (SiC) is a common ceramic materialor extreme environments. Owing to SiC’s pronounced chemicalnertness and high-temperature stability no significant oxidationeaction is observed below 900 ◦C under ambient pressure in oxy-en or air. SiC is also known for its excellent wear resistance and,specially in case of sliding wear in aqueous media, a low frictionoefficient. Therefore, silicon carbide ceramics are often employeds face seals.

Advanced ceramics like SiC are generally brittle, which oftenimits application due to potential premature device failure. Plas-ic deformation of SiC, observed in indentation experiments [1–4]indentation plasticity), has frequently been reported for tribolog-cal exposure [1,5,6], but not for static compression experimentselow ≈500 ◦C [7–12]. In general, tribochemical wear of SiC ceram-

cs is subject to the following main deterioration mechanisms: (1)hemical reactions, (2) plastic deformation, (3) micro-crack forma-ion, and (4) micro-abrasion [5,6]. Each of these mechanisms isnfluenced by a large number of factors such as contact geome-

∗ Corresponding author. Tel.: +49 7071 29 76804; fax: +49 7071 29 3060.E-mail address: volker.presser@uni-tuebingen.de (V. Presser).

043-1648/$ – see front matter © 2009 Elsevier B.V. All rights reserved.oi:10.1016/j.wear.2008.11.032

ed amount of SiOxHy which acts as an adhesive for the SiC wear debris.© 2009 Elsevier B.V. All rights reserved.

try, surface roughness, grain size, sliding speed, mode and amountof applied load, surface temperature/pressure, exposure duration,composition of the lubricant and so on [13].

Erickson et al. [6] differentiate between mild and severe wearregimes. The first was not reported to cause any formation of achemically influenced tribolayer on top of the mechanically dete-riorated SiC matrix. However, the role of such a tribochemicallayer is most important for experiments conducted under water-lubrication [14]. Here, SiC shows a higher wear resistance and alower friction coefficient than alumina, due to more pronouncedsurface smoothening [5,15,16]. This was explained in terms of tri-bochemical wear producing a thin oxygen-rich layer comprisingrounded SiC wear particles. Soft hydroxides cause SiC surfaces toshow a self-lubricating behaviour [17]. Smoothened surfaces, inturn, are known to show the effect of hydrodynamic lubrication andthis, combined with tribolayer formation, explains the low frictioncoefficient.

Fig. 1 shows schematically the simple model for the formation ofa reaction layer for sliding ceramics under water lubrication as pro-posed by Xu and Kato [18] based on the work of Quinn [19–21]. Inthe contact zone between the two sliding SiC surfaces high localtemperature and pressure conditions are prevailing on asperity

peaks. Consequently, SiC reacts with water to form a silica or silica-like layer, which delaminates as a result of repeated mechanicalcontact. With time the sliding surfaces are, therefore, smoothenedby a combined mechanism of wear debris formation and peakabrasion. Instead of a no-wear situation, as expected for perfect

V. Presser et al. / Wear 267 (2009) 168–176 169

layer s

hd

t>

S

S

O

S

FtrC2p

S

Fafbw

twrm[

S

Scp≈esirwaoucut

Fig. 1. Formation of a tribochemical

ydrodynamic lubrication on smooth surfaces, bulk abrasion andelamination dominate during the initial run-in phase.

Kitaoka et al. [5,22] proposed the following tribochemical reac-ions for SiC based on the detection of H2 and, at temperatures300 ◦C, CO2:

iC + 2H2O<120 ◦C−→ SiO2 + C + 2H2 (1)

iC + 4H2O>300 ◦C−→ SiO2 + CO2 + 4H2 (2)

r, more generally:

iC + nH2O → SiO2 + COn−2 + nH2 (4 ≥ n ≥ 2) (3)

rom static hydrothermal experiments on SiC it was reportedhat SiC becomes unstable shortly before the H2O:SiC molaratio reaches 2:1 while carbon destabilises shortly thereafter [23].orrespondingly, at intermediate temperatures (100–300 ◦C and00 MPa) reaction (1) and a methane forming reaction becomeossible [24]:

iC + 2H2O100−300 ◦C, 200 MPa−→ SiO2 + CH4 (4)

or wet wear of SiC we can assume a high H2O:SiC molar ratio withn abundance of lubricating water. However, this may not be trueor the entire sliding surface and trapped water, as will be discussedelow, can become a key factor for understanding tribochemicalear.

In the tribochemical equations listed solid silica is always a reac-ion product of silicon carbide and high-temperature, high-pressureater along with gases (COx, CH4, H2). Increasing the H2O:SiC molar

atio facilitates the formation of a pure carbon layer on hydrother-ally treated silicon carbide surfaces due to high silica solubility

23–26]:

iO2 + nH2O → SiO2 · nH2O (5)

olubility of silica strongly depends on the chosen p-T and pHonditions and is as low as several nm per year below 100 ◦C atH 7 [8,27,28]. At 285 ◦C and 100 MPa, a silica solubility rate of800 nm/h can be expected in distilled water [29] and there areven higher solubility rates for more extreme p-T regions [30]. Con-idering that reported tribolayers on SiC range from �m to nmn thickness, bulk silica formation rivals such high silica solutionates. A corresponding loss of silica due to (partial) SiO2 solutionas discussed and reported, for example, by Andersson et al. [31]

nd Presser et al. [32]. However, the common presence of a solid

xygen-rich, silica-like tribolayer renders the solution rate to besually lower than that of silica formation. In fact, no single-phasearbon layer has ever been found on SiC after tribological exposurender water lubrication. Solution and reprecipitation of silica dueo local effects will be addressed in more detail below.

chematically after Xu and Kato [18].

1.2. On the p-T conditions

As mentioned above, the p-T conditions present on SiC slidingsurfaces must strongly determine the tribochemical behaviour. Toinvestigate the p-T influence on, for example, silica formation, staticexperiments using autoclaves [23,33] or the hydrothermal diamondanvil cell [32,34] can be used. However, to conciliate these exper-iments with tribological investigations it is essential to know theactual p-T range on the surface commonly assessed by numeri-cal methods. Such approaches use boundary conditions (thermaldiffusivity, velocity, etc.) and the input data from profilometry orthe surface topology data [35–38], because real contact areas areseveral orders of magnitude smaller than the geometric contactareas. Because of surface roughness, the pressure is applied onlyon asperities and, therefore, local contact stresses may exceed theyield stress of the material. Following the rough contact modelsof Greenwood [39], asperities can be approximated as ellipsoidsand the Hertzian contact theorem is applied to estimate pressures.The heat formation and temperature rise in frictional microcontactscan subsequently be calculated using the contact flash temperaturemodels of Kuhlmann-Wilsdorf [36,37].

Own calculations gave high flash temperatures of up to or morethan 1000 ◦C in sliding contacts of rough surfaces and correspond-ing pressures up to several GPa, but those are expected to prevailonly for short periods of time and in small areas. Therefore, in themodel after Xu and Kato [18] shown in Fig. 1, tribochemical dete-rioration is expected to happen predominantly on asperities. Thegood thermal conductivity of SiC counteracts high temperaturesand thus several hundred centigrade are likely to be present as flashtemperatures on the sliding SiC surfaces at best.

However, longer lasting high pressure conditions are envis-aged by mechanical deterioration processes, which entrap waterin near surface cracks and pores. This way – similar to the opera-tion conditions in a hydrothermal diamond anvil cell – increasingtemperatures will cause a corresponding increase in pressure.Temperatures of 500 ◦C, for example, will induce an isochoric,isostatically acting pressure of several hundred MPa. Rise (=dissolu-tion) and fall (=reprecipitation) of p-T conditions allow transport ofsilica and oxygen input deep into the ceramic body along pathwayssuch as cracks.

Static experiments performed using comparable p-T rangesstudied the hydrothermal oxidation of SiC. Only some olderstudies [40,41] reported silica formation around SiC grains. Theoverall observation, however, made for the temperature rangeof 220–800 ◦C and up to several hundred MPa was an activeoxidation mechanism for SiC fibres, single crystals and powder[5,22,33,42,43]. Active corrosion starts at grain boundaries and

secondary phases like yttrium–aluminium–garnet or boron car-bide are preferentially attacked [33,43]. Silica formation in thesecases is limited to a mechanism of dissolution and reprecipita-tion [32,34]. However, reported corrosion rates vary strongly, whichis partially caused by non-comparable experimental conditions

1 ear 26

ase

woTrotams

2

(oa2tt

70 V. Presser et al. / W

nd different choices of material (SSiC: different polytype andintering aid composition; single-crystal vs. polycrystalline bodytc.).

We conclude from static experiments that during tribochemicalear, “hydrothermal” conditions (hundreds of MPa and hundreds

f ◦C) can be expected, which should result in active SiC corrosion.his, of course, qualitatively contradicts the detection of SiOx as aeaction product in tribolayers raising questions about the kineticsf silica formation and dissolution. Unfortunately, it is impossibleo probe the actual reaction zone with respect to pressure, temper-ture and pH-value, so we can only rely on calculations and indirectethods to evaluate what really happens during tribological expo-

ure.

. Experimental procedure

For hydrothermal runs undoped 6H–SiC single crystal platesSiCrystal AG, Erlangen, Germany) were ground to a thickness

f 100 �m, polished with 1 �m diamond paste (of Rz ≈ 500 nmnd Ra ≈ 10 nm) and cut into small bodies of approximately40 �m × 240 �m × 100 �m. Each sample was carefully cleaned inridestilled water and acetone for 30 min using an ultrasonic batho remove remnants of the polishing paste [44].

Fig. 2. (a) SE mode SEM images and (b) AFM scan of EKasic® F

7 (2009) 168–176

Hydrothermal experiments were performed with a hydrother-mal diamond anvil apparatus (HDAC) after Basset et al. (Refs.[45,46]). Pure tridestilled water was used as a medium and thepressure was calculated by equations of state (EOS) of pure waterand the determination of the homogenisation temperature [47,48].More details on HDAC operations can be found in Ref. [49]. Sam-ples were heated with 20 K min−1 up to 500 ◦C. This temperature,which corresponds to an isochoric pressure of 500–770 MPa, wasmaintained at ±1 K for 5 h before cooling to 300 K with the samegradient.

For tribological tests EKasic® F (ESK Ceramics GmbH & Co. KG,Kempten, Germany) was used. This material has a fine grain size ofabout 5 �m. It consists of ≈80 wt% 6H–SiC, 20 wt% 4H–SiC + 15R–SiCand additions of boron carbide as sintering aid. The sliding surfaceswere lapped to values of Rz = 2.79 �m and Ra = 0.79 �m as deter-mined by a stylus profilometer (Hommel T 8000s, Hommel GmbH,Köln, Germany).

Tribological tests were conducted in a sliding ring tribometer

with deionised water as lubricant. The setup is described in detail inRefs. [35,44]. During the experiment the lubricant was applied witha pressure of 0.1 MPa between the friction couple. Test runs wereexecuted at 1.5 MPa with a sliding velocity of 6 m s−1. We performedexperiments with and without external cooling to investigate mild

wear tracks from experiments without external cooling.

ear 267 (2009) 168–176 171

at2

Em

sa

p(t(Z2wcetT

3

3

aaR5Sa

saor(fe

tabiacR

3

mictTwrsr

wp

between elevated parts, as seen on the right part of Fig. 3c, are filledwith larger SiC particles forming a porous layer with up to 1.5 �mthickness. The mechanical deterioration is expressed in crack anddislocation formation (cf. Fig. 3a). Latter is also shown with a highermagnification in Fig. 4.

V. Presser et al. / W

nd severe wear regimes. For experiments without external cooling,he abort criterion in case of scuffing was a measured torque ofN m.

AFM measurements were performed using a TopoMetrixxplorer® (TopoMetrix Corporation, Santa Clara, USA) in contactode with a tripod scanner (500 × 500 pixel data resolution).For SE mode SEM images we used a LEO VP 1450 electron micro-

cope (Carl Zeiss AG, Oberkochen, Germany) operating at 15 kVcceleration voltage.

Sampling of an EKasic® F sample for TEM analysis was accom-lished by using focused ion beam (FIB) milling. FIB preparationcf. Ref. [50]) was conducted under ultra-high vacuum condi-ions in an oil-free vacuum system using a FEI FIB200 instrumentFEI Company, Eindhoven, The Netherlands) at the Helmholtz-entrum Potsdam. As a result TEM-ready foils of approximately0 �m × 10 �m × 0.15 �m were obtained. The FIB-cut foils thenere placed on a perforated carbon grid on a copper mesh. TEM was

arried out in a FEI F20 X-Twin instrument operated at 200 kV andquipped with a FEG electron source. Electron energy-loss spec-roscopy (EELS) data was acquired with a Gatan imaging filter (GIFridiem; Gatan GmbH, München, Germany).

. Results and discussion

.1. HDAC experiments

Hydrothermal treatment of single-crystal SiC yields a bulkctive oxidation mechanism causing enhanced surface roughnessnd pronounced pitting. The surface roughness increases fromz ≈ 500 nm and Ra ≈ 10 nm to Rz ≈ 1900 nm and Ra ≈ 22 nm after ah HDAC experiment at 500 ◦C. This observation for single-crystaliC is in agreement with the finding of active corrosion for CVD SiCt 500 ◦C in supercritical water by Barringer et al. [33].

Chemical reaction(s) are also evidenced by higher dehomogeni-ation temperatures, that is, during cooling the fluid separates intoliquid and a gaseous phase at much higher temperatures than

bserved during initial heating (e.g., at 310 ◦C instead of 80 ◦C). Theise is attributed to the formation of a gaseous reaction producthere: carbonaceous species). A graphitised surface layer was notound by Raman spectroscopy, which is sensitive to carbon layersven on the nm scale.

Silica was observed which was probably formed by precipi-ation. At least some of the silica present (partially containingluminium and iron as remnants from sample machining) muste seen as a quenching product when the small amount of water

n the HDAC sample chamber evaporates during sample extractiont ambient temperatures. More details on static SiC hydrothermalorrosion using a Bassett-type diamond anvil cell can be found inef. [32].

.2. Tribological exposure experiments

Typical images from worn samples in SEM, AFM, STEM and TEMode are shown in Figs. 2 and 3. While tribological exposure leads

n general to a morphological smoothening, samples from severeonditions without external cooling had both smoothened wearracks (plateaus) and lower areas of increased surface roughness.he inset in Fig. 2a shows the rough part next to the smoothedear track in a higher magnification. Note the plastically deformed

emains of a former wear track (Fig. 2a). Also, in Fig. 2a the two

ampling sites from which the FIB foils were taken (Fig. 3b and c,espectively) are marked.

Fig. 3 compares areas subjected to mild and severe wear regimesith the aid of STEM/TEM studies, which were carried out on FIBrepared cross-sections. For tribological experiments conducted at

Fig. 3. Electron microphotographs of the superficial area of EKasic® F wear tracks(a) with and (b, c) without external cooling. White spots in 3c are due to Ga-contamination. For discussion see text.

1.5 MPa with external cooling a thin tribolayer of ≈35 nm above a500–800 nm thick region of severely mechanically deformed siliconcarbide was found (Fig. 3a). This region of mechanic deformationshows plastic deformation, dislocation formation and pronouncedcracks, which are partially filled with material. Morphological andstructural differences between smooth wear tracks and rough parts(Fig. 3b and c, respectively) were more pronounced in samples fromtests without cooling water.

Figs. 3b and c show that the tribolayer’s thickness onsmoothened yet elevated areas is ≈100 nm, while on rough areasa total thickness of ∼500 nm is observed. Even deeper “valleys”

Fig. 4. TEM dark field image of the superficial area of EKasic® F wear tracks withexternal showing mechanical deterioration in form of crack and dislocation forma-tion.

172 V. Presser et al. / Wear 267 (2009) 168–176

F s correr

wlwtTzb

cbpittpas

tsir[b

ig. 5. Electron microphotographs of the tribolayer formed on SiC. Oxygen-rich areaedeposited) cracks (5b).

Fig. 5 depicts EELS mappings and TEM images of the smoothear track without external cooling (Fig. 5a) and the thin tribo-

ayer observed in mild wear experiments with external cooling byater (Fig. 5b). The oxygen distribution maps shows that nanome-

re scale SiC wear debris is contained within an oxygen-rich matrix.he same oxidation product can be found in cracks and shearones in the mechanically deteriorated area within the ceramicody.

Further EELS analysis of the tribolayer revealed the presence ofarbon and silicon (Fig. 6). Oxygen seems to be restricted to theoundaries of nm-sized grains of SiC. Nonetheless the oxygen-richhase seems not to be a uniform grain-boundary phase and we

nterpret it rather to be an adhesive, holding the angular wear debrisogether. The nm-scaled microstructure makes it very difficulto determine the exact chemical composition of the oxygen-richhase, but EELS mapping clearly indicates that all oxygen-rich areaslso show a strong silicon signal, which is weaker than the siliconignal from the SiC matrix.

From GDOES analysis [44] we know that the tribolayer con-ains significant amounts of hydrogen. Hence, an overall SiOxHy

toichiometry seems feasible. The weak carbon signal makes itmpossible to exclude the presence of oxycarbides, which wereeported to be possible reaction products in tribochemical wear51], as part of the assemblage. Some larger regions are free of car-on down to the detection limit and, therefore, we conclude that

spond to the tribolayer (in which SiC wear particles are rubbed in; 5a) and (partially

the overall composition of the oxygen-rich regions is close to (wet)silica.

Differences between mild and severe deterioration can clearlybe made comparing the EELS results and TEM images from exper-iments with and without external cooling. However, unlike themodel proposed by Erickson (Ref. [6]), chemical changes play animportant role in the “mild wear” regime, too. Mechanical dete-rioration dominates the first �m below the surface, but a thintribolayer (<50 nm) is present and oxygen input evidently takesplace along cracks several hundreds of nanometre deep inside theceramic body. These cracks, therefore, provide a pathway networkfor water ingress into the ceramic body as long as they are con-nected to the surface.

Without external cooling we found crystalline silica in someparts of the reaction layer (Fig. 7) by selected area diffraction (SAD).Comparing our results with Ref. [52], the reflections with d = 3.27 Åcould be attributed to �-quartz-(0 1 1), those with d = 2.06 Å to �-quartz-(2 2 0).

It is well known [53] that the presence of quartz is a good indica-tor for a moderate hydrothermal regime. At atmospheric pressure

only cristobalite forms and it does not transform into modifica-tions other than the low-T-form of cristobalite. In fact, amorphoussilica transforms even under hydrothermal conditions first to cristo-balite and then to quartz [54] and the synthesis of silica powder inhydrothermal conditions gives cristobalite and/or quartz [55,56].

V. Presser et al. / Wear 267 (2009) 168–176 173

F . A sech .

Tcr(ad2rtctt

3

bbcSr

FFs

ig. 6. EELS mappings of EKasic® F tribologically exposed without external coolingigh carbon concentration on top of the tribolayer is due to previous carbon coating

his is also known from hydrothermal oxidation of SiC, whereristobalite is formed first and quartz appears later during longeruns at several hundred centigrade [23]. Under very high pressures>2 GPa) other modifications of silica form (coesite/stishovite) andt high temperatures (>750 ◦C, depending on the pressure) quartzisappears. Hence, the p-T-regime must have been limited to about50–750 ◦C at pressures of several hundred MPa. Although this is aelatively large p-T range, the detection of crystalline silica proveshat (1) the tribolayer consists of silica as a major phase, (2) the p-Tonditions were hydrothermal, and (3) these hydrothermal condi-ions must have been sustained for an extended period of time nearhe surface.

.3. Qualitative wear-model of wet silicon carbide tribo-corrosion

Our first ideas on how hydrothermal tribo-corrosion works

ased on static experiments and tribological tests [32] may nowecome refined, because the TEM studies allow to describe theomposition and structural characteristics of tribolayers formed oniC during sliding contact under water lubrication. The findings ofounded, lapped and mechanically deformed SiC wear debris glued

ig. 7. Selected-area diffraction (SAD) performed in the highlighted (dashed) area inig. 6 (without external cooling; rough part) evidencing the presence of crystallineilica (�-quartz).

tion from the rough part shown in Fig. 3c with some large SiC debris particles. The

together by a silica-like phase and re-deposition of the latter incracks and depressions leads us to the following qualitative modelwhich is illustrated schematically in Fig. 8.

As boundary conditions we assume a sliding contact withwater as lubricant for two SiC faces, which have a surface rough-ness smaller than the average grain size. Otherwise, pronouncedgrain flaking would result. Both surfaces are “real” surfaces, wheremechanical contact is starting as asperity interaction:

Step 1 When in contact without ideal hydrodynamic lubrica-tion, asperity peaks collide, which causes large, but locallyextremely restricted rises in pressure and temperature. Thecollision is not to be viewed as a Hertzian contact but ratheras an impact event. Consequently, not only dislocations areinflicted to the ceramic body—SiC will structurally be shat-tered and disassembled into smaller crystallites and weardebris. Some of these particles will be entirely removed fromthe tribointerface by water. Other particles derived fromerosional wear will accumulate elsewhere on the sliding sur-face. Strained silicon carbide is more susceptible to chemicalattack of hydrothermal water. Hence both reaction gas andsilica start to form preferentially on crack surfaces and otherhighly stressed points.

Step 2 Repeated mechanical contact between asperity peaks nowinflicts more severe mechanical deterioration of the SiCgrains. Large ruptures and shear cracks enable water to dif-fuse, to flow or to become pressed into these newly formedopen spaces. Also, the SiC single crystals out of which theceramic body is made locally transform into mosaic crystalsdue to slip plane gliding. The most important process is thatthe cracks formed at one event may be sealed off during thenext collision event.

The direct mechanical contact of asperity peaks inducesonly temporarily high pressures and the energy of theimpact will be transformed to deformation, cracking andthermal energy. A single event will raise the local tempera-ture only to see a rapid decrease afterwards, because energyis lost to the lubricating water and into the material by heatconduction (SiC has a good thermal conductivity). As themechanical contact is often repeated, a dynamic interfacetemperature gradient is established over time. Any mea-surement of the interface’s temperature, for example, byintroducing a thermocouple a few millimetres below theactual surface, will thus record temperatures from this gra-

dient and calculations of surface temperatures must involveheat transfer models.

Step 3 Energy of repeated impacts will not be easily modelled,because the material forming a tribolayer changes bound-ary conditions by the lowering of thermal conductivity with

174 V. Presser et al. / Wear 267 (2009) 168–176

odel o

Fig. 8. Schematic illustration of our m

time and the decreasing roughness will also tend to lowerthe energy production with time. Nonetheless, the repe-tition of contacts will tend to raise the temperature to adynamic equilibrium value.

The mechanical opening and closing of cracks will be sup-ported by the increased production of further wear debris,which is smeared and pressed into any surface opening.Only now there is a situation, where water is separated andcontained from the rest of the lubricating medium. Watertrapped this way shows isochoric behaviour upon heat-ing, that is, the dynamic temperature regime causes trulyhydrothermal conditions with accordingly induced pres-sure, which is kept high. Under these conditions we expecthydrothermal reactions to take place, for example:

SiC + (n + 2)H2O → Si(OH)4 + COn−2 + nH2 (6)

In a dynamic situation newly formed cracks can alsoreopen trapped water reservoirs. Then dissolved silica repre-

cipitates due to pressure relief and, accumulates to build upthe oxygen-rich component of the tribolayer, most likely awet silica-rich phase (SiOxHy) with or without oxycarbides(SixCyOz). A minor amount of oxygen and/or hydrogen maybe stored in strained/amorphised silicon carbide [32].

f wet silicon carbide tribo-corrosion.

Step 4 On the interface where mechanical contact repeatedly takesplace, smooth wear tracks will form over time. Wear debriscaught between the sliding surfaces will be subject to tri-bological lapping which inflicts plastic deformation, furtherreduces the average crystallite size, and makes the grainsmore roundish. This accounts for the nanometre scale SiCwear particles found in the triboscale. The silica-rich phaseout of which a significant part of the tribological layer iscomposed, originates mainly from reprecipitated silica andonly to a minor extent from SiC oxidized directly along thecontact zone.

Loose wear debris accumulates on suitable spots on the SiCsurface, in particular in any depressions, crevices etc., whichare protected from becoming washed out by the counterfaceof the wear partner. Reprecipitated SiOx is present in suchspots as well, even though these areas comprise a significant(open) pore volume and show poor densification. Nonethe-less, when filled up these sites become preferential areas forhydrothermal reactions and spallation.

Due to the dynamic nature of the conditions along thetribointerface temporarily fluctuating p-T values will beobserved. Pressure relief which is related to a drop in tem-perature causes not only reprecipitation of silica but alsodehomogenisation of the fluid. This phenomenon is known

ear 26

cosrttm

4

deuitbuwec

odbtwttanp

A

sRsM

R

[

[

[[

[

[

[

[[[

[

[

[

[

[

[[

[

[

[

[

[

[

[

[

[

[

[

[

[

[

V. Presser et al. / W

from HDAC experiments (Refs. [32,34]) and possibly con-tributes to bulk wear by cavitation further contributing tothe fatigue of the already disrupted tribolayer.

In addition to the processes described above a loss of parti-les to the lubricating medium occurs. A variation in the extentf the individual processes may be the only thing distinguishingevere and mild wear conditions as a tribochemically influencedeaction layer is present even for the mild wear regime. Evidently,he resulting thickness of the individual zones strongly depends onhe conditions chosen for tribological exposure (pressure, contact

ode, duration, external cooling, etc.).

. Conclusion

The wet tribo-corrosion behaviour of sintered SiC has beeniscussed in the light of HDAC and tribotests with and withoutxternal cooling. While active corrosion is the dominant processnder hydrothermal conditions, the formation of a silica-rich phase

s always observed during tribological exposure. Such phases havehe general composition SiOxHy and may locally be of oxycar-idic nature. SiOx is a product of dissolution and reprecipitationnder hydrothermal p-T conditions on sliding surfaces with trappedater. The presence of hydrothermal conditions is not limited to

xtremely short periods of time as evidenced by the formation ofrystalline quartz in experiments without external cooling.

For both mild and severe wear conditions we found the presencef this chemically modified triboscale above a layer of mechanicallyeteriorated silicon carbide. While the latter can reach up to 2 �mulk thickness the tribochemical layer is much thinner, usually inhe range of tens to hundreds of nanometre. Only in areas of severeear where large wear debris particles (up to hundreds of nanome-

re in diametre) accumulate in depressions a significantly largerribolayer thickness can be observed. However, these triboscalesre composed of randomly distributed SiC particles loosely con-ected by SiOxHy comprising a significant amount of partially openores.

cknowledgements

This work was supported by the Deutsche Forschungsgemein-chaft (DFG) through Ni299/12-1 and Ho 1149/92. We thank Dr.ichard Wirth (Helmholtz-Zentrum Potsdam) for the TEM mea-urements; Dr. Christoph Berthold, Dipl.-Min. Anselm Loges andargarete Schloßer are thanked for valuable discussion.

eferences

[1] V. Presser, C. Berthold, R. Wirth, K.G. Nickel, Structural characterisation of tribo-logically influenced silicon carbide ceramic surfaces,Current Opinion in SolidState & Materials Science, submitted for publication.

[2] A. Kailer, Y.G. Gogotsi, K.G. Nickel, Use of hardness indentation coupled withmicro-Raman spectroscopy in high-pressure materials research, MaterialsResearch Society Symposium Proceedings 499 (1998) 225–230.

[3] A.A. Wereszczak, K.E. Johanns, Spherical indentation of SiC, Ceramic Engineer-ing and Science Proceedings 27 (2007) 43–57.

[4] J. Lankford, D.L. Davidson, Indentation plasticity and microfracture in siliconcarbide, Journal of Materials Science 14 (1979) 1669–1675.

[5] S. Kitaoka, T. Tsuji, T. Katoh, Y. Yamaguchi, K. Kashiwagi, Tribological character-istics of SiC ceramics in high-temperature and high-pressure water, Journal ofthe American Ceramic Society 77 (1994) 1851–1856.

[6] L.C. Erickson, A. Blomberg, S. Hogmark, J. Bratthall, Tribological characterizationof alumina and silicon carbide under lubricated sliding, Tribology International26 (1993) 83–92.

[7] A. Mussi, J. Rabier, L. Thilly, J.L. Demenet, Plasticity and deformation microstruc-

ture of 4H–SiC below the brittle-to-ductile transition, Physica Status Solidi C 4(2007) 2929–2933.

[8] V. Presser, K.G. Nickel, Silica on silicon carbide, Critical Reviews in Solid Stateand Material Sciences 33 (2008) 1–99.

[9] A.V. Samant, P. Pirouz, Activation parameters for dislocation glide in �-SiC,International Journal of Refractory Metals & Hard Materials 16 (1998) 277–289.

[

[

7 (2009) 168–176 175

10] A.V. Samant, X.L. Wei, P. Pirouz, An optical and transmission electronmicroscopy study of deformation-induced defects in 6H–SiC, PhilosophicalMagazine A 78 (1998) 737–746.

[11] J.L. Demenet, M. Zhang, H.M. Hobgood, P. Pirouz, Fragile–ductile transition ofsingle crystal 4H–SiC, Journal de Physique IV 106 (2003) 53–62.

12] J. Patten, W. Gao, K. Yasuto, Ductile regime nanomachining of single-crystalsilicon carbide, Journal of Manufacturing Science and Engineering 127 (2005)522–532.

13] S.M. Hsu, M. Shen, Wear prediction of ceramics, Wear 256 (2004) 867–878.14] J.K. Lancaster, Y.A.-H. Mashal, A.G. Atkins, The role of water in the wear of

ceramics, Journal of Physics D 25 (1992) A205–A211.15] S. Kitaoka, Y. Yamaguchi, Y. Takahashi, Tribological characteristics of �-alumina

in high-temperature water, Journal of the American Ceramic Society 75 (1992)3075–3080.

16] S. Sasaki, The effects of the surrounding atmosphere on the friction and wear ofalumina, zirconia, silicon carbide and silicon nitride, Wear 134 (1989) 185–200.

[17] H. Tomizawa, T.E. Fischer, Friction and wear of silicon nitride and silicon carbidein water: hydrodynamic lubrication at low sliding speed obtained by tribo-chemical wear, ASLE Transactions 30 (1987) 41–46.

18] J. Xu, K. Kato, Formation of tribochemical layer of ceramics sliding in water andits role for low friction, Wear 245 (2000) 61–75.

19] T.F.J. Quinn, Oxidational wear, Wear 18 (1971) 413–419.20] T.F.J. Quinn, Oxidational wear modelling. Part I, Wear 153 (1992) 179–200.21] T.F.J. Quinn, Oxidational wear modelling. Part II. The general theory of oxida-

tional wear, Wear 175 (1994) 199–208.22] S. Kitaoka, T. Tsuji, Y. Yamaguchi, K. Kashiwagi, Tribochemical wear theory of

non-oxide ceramics in high-temperature and high-pressure water, Wear 205(1997) 40–46.

23] T. Kraft, K.G. Nickel, Y.G. Gogotsi, Hydrothermal degradation of chemical vapourdeposited SiC fibres, Journal of Materials Science 33 (1998) 4357–4364.

24] Y.G. Gogotsi, M. Yoshimura, Degradation of SiC-based fibres in high-temperature, high-pressure water, Journal of Materials Science Letters 13(1994) 395–399.

25] T. Kraft, K.G. Nickel, Hydrothermal carbon coatings of alpha-SiC crystals, in:H. Dimigen (Ed.), Surface Engineering—EUROMAT, 99, Wiley-VCH, 1999, pp.306–311.

26] Y.G. Gogotsi, Hydrothermale Korrosion von SiC - Betrachtung der schädlichenund nützlichen Aspekte, in: R. Telle, P. Quirmbach (Eds.), Korrosion und Ver-schleiß von keramischen Werkstoffen, Deutsche Keramische Gesellschaft e.V.(DKG), Aachen, 1994, pp. 114–122.

27] R.K. Iler, The Chemistry of Silica, John Wiley & Sons, New York, 1979.28] G. Perera, R.H. Doremus, W. Lanford, Dissolution rates of silicate glasses in water

at pH 7, Journal of the American Ceramic Society 74 (1991) 1269–1274.29] S. Ito, M. Tomozawa, Stress corrosion of silica glass, Journal of the American

Ceramic Society 64 (1981) C160.30] G.C. Kennedy, A portion of the system silica–water, Economic Geology 45 (1950)

629–653.31] P. Andersson, J. Juhanko, A.-P. Nikkila, P. Lintula, Influence of topography on

the running-in of water-lubricated silicon carbide journal bearings, Wear 201(1995) 1–9.

32] V. Presser, O. Krummhauer, K.G. Nickel, A. Kailer, C. Berthold, C. Raisch, Tribo-logical and hydrothermal behaviour of silicon carbide under water lubrication,Wear (2008).

33] E. Barringer, Z. Faiztompkins, H. Feinroth, T. Allen, M. Lance, H. Meyer, L. Walker,E. Lara-Curzio, Corrosion of CVD silicon carbide in 500 ◦C supercritical water,Journal of the American Ceramic Society 90 (2007) 315–318.

34] V. Presser, K.G. Nickel, Hydrothermal oxidation of silicon carbide single crystalsusing a modified diamond anvil cell, in: 10th International Conference of theEuropean Ceramic Society, Berlin, 2007, pp. 191–195.

35] M. Zimmermann, Experimentelle Untersuchung und numerische Modellierungdes Gleitkontaktverhaltens von gesintertem Siliciumcarbid (SSiC), UniversitätKarlsruhe, Karlsruhe, 1998.

36] D. Kuhlmann-Wilsdorf, Demystifying flash temperatures. I. Analytical expres-sions based on a simple model, Materials Science and Engineering 91 (1987)107–118.

37] D. Kuhlmann-Wilsdorf, Demystifying flash temperatures. II. First-order approx-imation for plastic contact spots, Materials Science and Engineering 91 (1987)119–133.

38] B. Bhushan, Contact mechanics of rough surfaces in tribology: multiple asperitycontact, Tribology Letters 4 (1998) 1–35.

39] J.A. Greenwood, The area of contact between rough surfaces and flats, Transac-tions of the ASME 1 (1967) 81–91.

40] M. Yoshimura, J.-I. Kase, S. Somiya, Oxidation of SiC powder by high-temperature, high-pressure H2O, Journal of Materials Research 1 (1986)100–103.

41] M. Yoshimura, J. Kase, S. Somiya, Oxidation of Si3N4 and SiC by Hightemperature–high pressure water vapor, in: 2nd International Symposium onCeramic Materials and Components for Engines, Germany, 1986, pp. 529–536.

42] W.-J. Kim, H.S. Hwang, J.Y. Park, W.-S. Ryu, Corrosion behaviors of sintered andchemically vapor deposited silicon carbide ceramics in water at 360 ◦C, Journal

of Materials Science Letters 22 (2003) 581–584.

43] K.A. Schwetz, J. Hassler, Zur Beständigkeit von Hochleistungskeramiken gegenFlüssigkeitskorrosion, Ceramic Forum International 79 (2002) D9–D19.

44] O. Krummhauer, V. Presser, A. Kailer, K.G. Nickel, Comparison of tribologicaland corrosive behaviour of silicon carbide under water lubrication, in: (Eds.),48. Tribologie Fachtagung. Reibung, Schmierung und Verschleiß. Forschung und

1 ear 26

[

[

[

[

[

[

[

[

[

76 V. Presser et al. / W

praktische Anwendungen. Band 1., Gesellschaft für Tribologie, Göttingen, 2007,pp. 13/11–13/15.

45] W.A. Bassett, A.H. Shen, M. Bucknum, I.-M. Chou, Hydrothermal studies in anew diamond anvil cell up to 10 GPa and from −190 ◦C to 1200 ◦C, PAGEOPH141 (1993) 487–495.

46] W.A. Bassett, A.H. Shen, M. Bucknum, I.-M. Chou, A new diamond anvil cell forhydrothermal studies to 2.5 GPa and from −190 to 1200 ◦C, Review of ScientificInstruments 64 (1993) 2340–2345.

47] H.T. Haselton, I.-M. Chou Jr., A.H. Shen, W.A. Bassett, Techniques for determiningpressure in the hydrothermal diamond-anvil cell: behaviour and identifica-tion of ice polymorphs (I, II, V, VI), American Mineralogist 80 (1995) 1302–1306.

48] A.H. Shen, W.A. Bassett, I.-M. Chou, Hydrothermal studies in a diamond anvil

cell: pressure determination using the equation of state of H2O, in: Y. Syono,M.H. Manghnani (Eds.), High-Pressure Research: Application to Earth and Plan-etary Sciences, American Geophysical Union, Washington, DC, 1992, pp. 61–68.

49] V. Presser, M. Heiß, K.G. Nickel, EOS calculations for hydrothermal diamondanvil cell operation, Review of Scientific Instruments 78 (2008), 085104-085101-085104-085109.

[

[

7 (2009) 168–176

50] R. Wirth, Focused ion beam (FIB): a novel technology for advanced applicationof micro- and nanoanalysis in geosciences and applied mineralogy, EuropeanJournal of Mineralogy 16 (2004) 863–876.

[51] J.L. Lauer, Wear-induced changes in the Raman, infrared, and fluorescence spec-tra of silicon nitride surfaces, Applied Spectroscopy 50 (1996) 1378–1388.

52] M. Dusek, V. Petrícek, M. Wunschel, R.E. Dinnebier, S. van Smaalen, Refinementof modulated structures against X-ray powder diffraction data with JANA2000,Journal of Applied Crystallography 34 (2001) 398–404.

53] K. Byrappa, I. Yoshimura, M. Yoshimura, Handbook of Hydrothermal Tech-nology: A Technology for Crystal Growth and Materials Processing, NoyesPublications, Norwich, 2001.

54] A.S. Campbell, W.S. Fyfe, Hydroxyl-ion catalysis of the hydrothermal crystalliza-tion of amorphous silica; a possible high-temperature pH indicator, American

Mineralogist 45 (1960) 464–468.

55] Y. Zhu, K. Yanagisawa, A. Onda, K. Kajiyoshi, The preparation of nano-crystallized cristobalite under hydrothermal conditions, Journal of MaterialsScience 40 (2005) 3829–3831.

56] K. Yanagisawa, Y. Zhu, A. Onda, K. Kajiyoshi, Hydrothermal synthesis of mono-dispersed quartz powders, Journal of Materials Science 39 (2004) 2931–2934.