WEAR

ELSEVIER
Wear 181-183 (1995) 571-580

Tribological behavior of silicon nitride materials under unlubricated sliding between 22 C and 1000 C
A. Skopp, M. Woydt, K.-H. Habig
Federal Institute for Matetials Research and Testing @AM), 29 April 1994; accepted Unter den Eichen (97, 12200 Berlin, Gemany 20 October 1994 Received

Abstract Friction and wear of 13 different materials with silicon nitride (S&NJ matrix were investigated under unlubricated conditions. Tests were performed in laboratory air at sliding velocities between 0.03 and 5 m s--r and at temperatures in the range of 22 to 1000 C. These materials were tested in the same test rig by means of the pin-on-disc configuration with self-mated couples to evaluate tribological influences of composition, structure and other properties, which were varied to reach friction coefficients below 0.2 and wear coefficients below 1 X 10e6 mm3 N-r m-r. For monolithic silicon nitride materials, high friction coefficients between 0.4 and 0.9 and wear coefficients between 5 X 10e6 and 1 x lo-* mm3 N- m-r were measured. By adding boron nitride, the friction coefficient could be reduced to 0.1 at room temperature. Simultaneously, after running-in, the wear rate was also reduced by a factor of five in comparison to monolithic silicon nitride. Through the addition of titanium nitride, the wear was reduced by a factor of five at room temperature and even by two orders of magnitude at a high temperature of 800 C. All materials and worn surfaces as well as wear debris were extensively characterized by means of SEM, plus energydispersive X-ray, X-ray diffraction, small-spot electron spectroscopy for chemical analysis, electron-probe microanalysis, Auger electron spectroscopy, Fourier transform infrared spectroscopy and transmission electron microscopy before and after the tribological tests. Combined with theoretical estimations on mechanical and thermal stresses, which are based on calculations of hot-spot (flash) and bulk temperatures, wear processes and wear models are proposed. The monolithic silicon nitrides investigated reveal a mechanically dominated wear mechanism with fatigue between the grains and the amorphous grain boundaries. Tribochemical reaction layers (SiO,, SiO,N,, Si3N,.Y,03) did not tend to decrease friction and/or wear in a wide range of test conditions. Only at higher temperatures above 400 C and higher sliding velocities (above 0.3 m SK at 1000 C and above 1 m s-r at 400 C) was a more pronounced wear reduction caused by the tribochemical formation of protective layers and their plastic deformation. The addition of titanium nitride (TiN) or boron nitride (BN) led to microstructures which were mainly free from glassy grain-boundary phases. More important, these silicon nitride composites are able to form relatively soft lubricious oxides like TiOz_, and H,BO, or an intrinsically solid lubricant like BN.H,O on a hard substrate, so that then a tribochemical wear mechanism dominates. Considering tribological and surface analysis results as well as theoretical calculations, wear maps are given for the various monolithic silicon nitride materials and for the composites Si,N,-TiN and S&N,-BN. They give indications for the further development of ceramic materials.
Keywords: Silicon nitride; Wear maps; High temperature; Tribology; T-maps; Ceramics composites

1. Introduction

For moderately stressed silicon nitride sliding pairs (low mean contact pressure and/or low velocities) in humid atmospheres up to temperatures of about 800 C (in humidity-saturated gases), low friction and wear coefficients have been measured [l-8]. In dry and normal atmospheres and for more highly stressed sliding couples

of silicon nitride materials, only high friction coefficients and high wear coefficients under unlubricated conditions have been published [1,2,6-121. To compare the tribological results of ceramics in a wide range of testing conditions, wear maps have been proposed by different authors [12-181. However, nearly all of the numerous existing silicon materials have been tribologically characterized in different test rigs under differing testing

0043-1648/95/$09.50 0 1995 Elsevier SSDI 0043-1648(94)07082-2

Science

S.A. All rights

resewed

572

A. Skopp et al. / Wear 181-183

(1995) 571-580

conditions [l-8,16-17]. This hinders the comparison of the tribological behavior and the evaluation of influences from fabrication, composition and structure. Under severely stressed unlubricated sliding conditions, three silicon materials have been tested by Iwasa and Toibana [9] and by Childs and Mimaroglu [lo]. They were tested against themselves in the same test rig to investigate mainly the effects of composition and processing route. In a small range of testing conditions, the results of Iwasa and Toibana [9] indicate that the composition has an influence on the wear. The lowest wear was measured for S&N, with 5 wt.% La,OJ5 wt.% Y,O,, increasing for Si,N, with 5 wt.% YZ045 wt.% A&O,, and was highest for S&N, with 5 wt.% MgO. The tests by Childs and Mimaroglu [lo] indicate a lower wear coefficient for the rotating specimen, when silicon nitride with a lower glass-phase content (addition of 2% MgO instead 4% MgO) was used. For eight silicon nitride materials with different compositions (varied additives and additive contents between 4 and 20 wt.%), different production routes and different microstructures, tribological tests were performed under unlubricated and highly stressed sliding conditions [11,12]. All materials were investigated in one test rig with the pin-on-disk configuration. Ambient atmosphere was air with relative humidities between 20% and 50% (at room temperature). For sliding velocities between 0.03 and 5 m s-l and temperatures between 22 and 1000 C, the influences of composition and structure on the tribological behavior were small. For these materials the friction coefficients were high, in the region between 0.4 and 0.9, and the wear coefficients were higher than 5~10-~ mm3 N- m-l, increasing with ambient temperature to values up to lo- mm3 N-l m-l. The interrelationships of mechanical and chemical aspects were extensively investigated by various methods of surface analysis, and by theoretical estimations [11,12,19,20]. To overcome the unsatisfactory tribological sliding behavior of conventional silicon nitride materials, secondary phases (which should have self-lubricious capabilities with low friction and high wear resistance) were added. Secondary phases should also give a higher resistance against mechanical wear and, if possible, better mechanical properties. Ti-based phases like TiN, with the ability to form TiOa_x films [21-231, and BN, which can build up intrinsic self-lubricating films (which are expected to show a beneficial tribological behavior [24,25]), were chosen as proper candidates to be added to S&N,. Recent results with TiO,_, phases (e.g. Magntli phases of Ti,O,_, with n =4-10) show that the films are not stable against changes of temperature and atmospheric pressure [23] and that the measured low friction values are not reproducible. Tests have shown that the tribological behavior of hexagonal BN and that of materials

containing BN depends on the ambient atmosphere [24,25]. Low friction and low wear have been measured in humid air. Gardos [25] has stated that low friction is due to the unique role of water vapor in the friction of lamellar compounds. Erdemir et al. [26,27] have ascribed very low friction coefficients to the formation of self-lubricating boric acid films on boron-containing surfaces. The use of these phases in composites, which could have some stabilizing effect on the desired phases, has been shown only in a few publications [28,29]. Recently, experimental results with S&N,-TiN under dry oscillating sliding conditions were published by Imada et al. [28] and show that the wear coefficient of Si,N, was distinctly reduced by the addition of TiN. The wear then decreased with increasing humidity and decreasing load. Tribological tests with Si3N,-BN (2 to 30 wt.% BN) sliding pairs have been performed by Iwasa and Kakiuchi [29] and reveal a drop in the wear coefficient by an order of magnitude to low6 mm3 N- m -l, when more than 20 wt.% BN was added to Si,N+ Simultaneously, a friction-reducing effect was not observed, with friction coefficients above 0.5. Under severe (highly stressed) dry sliding conditions, friction and/or wear of silicon nitride-based materials has been reduced over a wide range of speeds and temperatures by the addition of titanium nitride (TiN) or hexagonal boron nitride (BN) [30,31]. The surfaces and reaction layers have been investigated by various methods of surface analysis [12,19,20,30]. On the basis of some of these results, as well as on the calculation of contact temperatures and thermal stresses [12], this work reveals in a comprehensive form the mechanisms of wear for self-mated Si,N, materials, giving wear maps and indications for the development of silicon nitride materials with an improved tribological behavior.

2. Experimental details 2.1. Materials Eight commercially available silicon nitride materials were tribologically characterized: SSi,N, (ND200), HIPRBSi3N_, (411), HIPS&N, (FX-950), GPSSi3N, (SN73), GPSSi,N_, (ESK), GPSRBSi,N, (ESK), HPSi,N, (ESK) and HIPS&N, (ESK), where S=sintered, HP = hot-pressed, HIP = hot-isostatically-pressed, RB = reaction-bonded, GPS = gas-pressure-sintered. Beside the production process, the amount, distribution and type of additives as well as the structure of these polycrystalline materials were varied over a broad range. Material characterization - chemical composition, structure and physical properties - is given elsewhere [11,12,19,20].

A. Skopp et al. I Wear 181483

(1995) 571-580

573

In addition, S&N, with 30 wt.% TiN and Si,N4-BN with 5, 10 and 20 wt.% BN were tribologically characterized. Analyses with TEM show that, in contrast to the polycrystalline silicon nitride materials, these materials are mainly free of glass phases. 2.2. Test rig and experimental procedure

volumetric wear coefficient is the sum of the volumetric wear coefficients of the stationary and the rotating specimens.

3. Results

and discussion

3.1. Experimental results All materials were investigated in the same hightemperature tribometer in a pin-on-disk configuration [32]. The schematic drawing and the arrangement of the specimens are shown in Fig. 1. The stationary specimen with a curved surface is pressed against the plane surface of a rotating disc (rotating specimen). With increasing wear, the contact pressure decreases according to the test geometry, with a defined point contact at the beginning, so that statements concerning the influence of the mean contact pressure are also possible. During the tests, up to a sliding distance of about 1000 m, sliding velocity (0.03, 0.1, 0.3, 1 and 3 m s-l), normal load (10 N) and temperature (22, 400, 800 and 1000 C) were kept constant. At room temperature the tribological tests were performed in laboratory air with relative humidities between 20 and 50%. Friction force, total linear wear and furnace temperature were continuously measured. The following definitions are used: Friction coefficient f= friction force FJnormal load Volumetric wear coefficient k= wear volume WV/ (normal load F,Xsliding distance s) The differential volumetric wear coefficient k (= WV/ (FNAs)) was calculated at a sliding distance of 1000 m, when after running-in the wear rate was stationary. The wear volume was calculated from the wear scar dimensions and profilometric measurements. In the literature, the wear factor is often used (here synonymous with the volumetric wear coefficient). The total Figs. 2-4 summarize the results of previous work [11,12]. At higher temperatures of 400, 800 and 1000 C (Fig. 2 shows examples for 800 C) for all conventional silicon nitride materials tested, the total volumetric wear coefficient drops with increasing sliding velocity by up to two orders of magnitude, whereas at 22 C the total volumetric wear coefficient increases at sliding velocities above 0.5 m s- by a factor of five to ten [11,12]. At sliding velocities between 1 m s- and 3 m s-l, only for GPSSi,N, (SN73) with additions of CeO, and ZrO, were a lower wear rate of 5 X10p6
Solid state frlctlon SIZN~ I SIGNS FN = IO N; rel humldlty at 22C. 20 - 40 %

lo+

10-6L
10.

,,I.

( ..I

,.,J
100 10%

lo-

Sliding velocity v [m/s] --)

Fig. 2. Total volumetric wear coefficient for unlubricated self-mated sliding pairs of conventional polycrystalline Si3N4 materials as a function of velocity at 22 C and 800 C.

SOlId Si3N, F,-

state frlctlon
/ Si,N, - 22 C 10 N :T

6 lo+1 10-z

10-I Slld;ng velocity

10o

. I

10

iii

2;

Im/s) -

Fig. 1. High-temperature

tribometer and test configuration.

Fig. 3. Critical frictional power density for the transition between a high wear rate and a lower wear rate of Si,N4 composites for unlubricated sliding.

574

A. Skopp et al. ! Wear 181483

(1995) 571-580

(EPMA), Auger electron spectroscopy @ES), smallspot electron spectroscopy for chemical analysis (ESCA/ XPS) and transmission electron microscopy (TEM) with convergent beam electron diffraction (CBED) and energy-dispersive spectroscopy (EDS). The results are published elsewhere [11,12,19,20]. The wear maps presented later are based on the results shown in the figures and on the results from wear surface analysis, as well as from hot-spot temperature calculations. 3.2. Calculation of temperatures on sliding surfaces
The temperature calculations of Ashby and co-workers [33,34] and Kuhhnann-Wilsdorf [35,36] were used in this work to estimate the bulk and the flash temperatures (hot spot at the real asperity contacts) at the interface of rubbing surfaces. They are based on theoretical models developed earlier by Blok [37], Archard [38] and Jaeger [39,40]. The surface temperatures in the tribocontact can be estimated from the frictional power density, given as the product of sliding speed, load per mean contact area and friction coefficient. Furthermore, the total surface temperature at the microcontacts is the sum of the ambient temperature, the bulk temperature and the flash temperature. The calculations consider the temperature dependence of thermomechanical properties of Si,N,. Influences due to wear particles and tribochemical reaction layers were neglected in the models used. Results from experiments with infrared thermography [41] for sapphire/Si,N, sliding pairs were introduced in the calculations to tune the T-MAPS 4.0 diagram [33,34] and for comparison, as has been described in Ref. [12]. It should be said that hardness and thermal properties of sapphire and Si,N, are similar, and that the tribological test conditions are comparable with those in the experiments of Griffioen et al. [41]. Therefore, the IR measurements give an indication of the surface temperatures in Si,NdSi,N, sliding contacts and of the number of asperities under the conditions of this work. Griffioen et al. [41] measured, for 8.9 N and 1.53 m s-l, hot-spot temperatures up to 1900 C. The average number of hot spots was 7 f3. The thermographically measured flash temperatures show a wide spread, probably because the asperity size lies in the range of 1 to 30 pm. Simultaneously, the flash temperature increases with the asperity size (under the assumption that the total area of the microcontacts is constant). Therefore, the temperature map has to be tuned due to experimental results. With the properties additionally given in the T-MAPS diagram, good agreement of measured and calculated temperatures has been obtained. Only at higher sliding speeds with high flash temperatures do the differences increase.

0.03 01 03

3 m/s

0.03 0.1 0.3

3m/s

Shdlng veloaty v

Fig. 4. Differential volumetric wear coefficient of the stationary specimen for the composites Si3N4-TiN and S&N,-BN in comparison to conventionally S&N., materials at 22 C and 800 C.

mm3 N-l m-l and a lower friction coefficient of 0.2 measured. It should be noted that under all test conditions the wear volume of the rotating specimen was higher than that of the stationary specimen [11,12]. The total volumetric wear coefficients were above lop5 mm3 N- m-l and increased with increasing temperature. The wear rate of silicon-nitride-based composites decreases with increasing sliding distance, whereas the wear rate of polycrystalline silicon nitrides is nearly independent of sliding distance (as has been shown earlier [11,12]). From the linear wear vs. sliding distance curves, a critical wear volume (at the end of runningin) can be determined for the composites. From this value, the critical mean contact pressure as well as the critical frictional power density can be calculated [12]. This critical frictional power density at the transition between high and low wear rates is shown in Fig. 3 as a function of sliding velocity for composites with BN or TiN. After a sliding distance of 1000 m the differential volumetric wear coefficients of conventional Si,N, materials and those of the composites S&N,-TiN and Si3N,-BN at 22 C and 800 C are compared in Fig. 4. The wear rate of silicon nitrides is distinctly reduced (up to orders of magnitude) when secondary phases are added to the silicon nitride matrix. This was observed over a wide range of sliding velocities at lower and at higher temperatures. The surfaces and the wear debris were investigated by means of scanning electron microscopy (SEM) plus energy-dispersive X-ray (EDX), small spot X-ray diffraction (XRD), Fourier transform infrared micro spectroscopy (micro FT-IR), electron probe micro analysis

A. Skopp et al. / Wear 181-183

(1995) 571-580

575

Table 1 Calculation of the flash temperature for one microcontact as function of sliding velocity for temperature-dependent and temperature-independent properties: Model Kuhlmann-Wilsdorf [35,36] cf= 0.5, N ,.,x= 1, T,=22 C). Calculation program, Bartelt [42]. Calculations for constant physical properties (AT,), temperature-dependent hardness (AT,) and temperature-dependent properties (ATJ 2, (m/s) AT, (C) 0.03 0.1 0.3 1.0 1.5 3.0 41 137 345 1007 1392 2396 x1 AT2 (C) 36 119 342 852 1030 1320 AT, (C) 31 131 453 1114 1303 2590 n at 1 atm. AT, (C) 56 182 600 1420 1992 p 3261 a AT, (C) 56 176 522 961 1350 1764 AT, (C) 59 211 732 1055 1480 1945 a AT, (C) 25 81 237 713 998 1750 AT2 (C) AT3 (C) SSi3N4/SSi3N4 (ND 200) HIP-Si~N,/HIP-S&N4 (FX-950) S&N,-TiN/Si,N,-TIN (ESK)

_ -

a T,,,, higher than sublimation temperature

Lines of constant bulk and flash temperatures calculated using the software T-MAPS 4.0 of Ashby and co-workers [33,34] are shown in Fig. 5 as a function of mean contact pressure and sliding velocity. They reveal the strong influence of sliding velocity on the flash temperature, which exceeds the oxidation temperature of S&N, at nearly 1 m s-l. Ishigaki et al. [6] have found an increasing wear rate above 0.2 m s- and a phase transition from a-Si,N4 to P-Si3N, at a sliding velocity of 1.5 m s-l. In air this phase transition occurs normally at 1550 C, so that this is a further experimental indication for tribologically induced high temperatures. According to the model of Kuhlmann-Wilsdorf [35,36], the flash temperatures can be calculated as a function of the number and size of microcontacts. Based upon this model, a computer program [42] was used to calculate the bulk and the flash temperatures. First, the calculations were performed with constant physical

Fig. 5. Temperature map for unlubricated self-mated sliding pairs of SSiSNJSSiSN, (ND200); for comparison, experimentally detected temperature data and temperatures calculated with the model of Kuhlmann-Wilsdorf [35,36] are additionally given.

properties (at ambient temperature), then with a temperature-dependent hardness and, finally, with temperature-dependent properties (hardness, thermal conductivity, specific heat, thermal diffusivity, Young s modulus) [12]. The temperature dependence of these values was mainly available up to a temperature of 800 C, some up to 1900 C. Table 1 shows the calculated flash temperatures under the condition of only one microcontact for constant properties and temperaturedependent properties (giving quite different flash temperatures, especially at higher sliding velocities). The outcome of the temperature calculations with the model of Kuhlmann-Wilsdorf [35,36] and that of the temperature maps constructed according to the TMAPS 4.0 program of Ashby and co-workers [33,34] are close, if in both models only the temperature dependence of the hardness (which decreases between 22 C and 1000 C by nearly a factor of two) is considered. When additionally the temperature dependence of the thermal conductivity (which decreases between 22 and 1000 C, also by nearly a factor of two) is used, then the calculated temperatures above 0.3 m s- are higher than those computed with the program of Ashby et al. In this model the temperature dependence of the thermal properties, like that of the thermal conductivity, is not considered. The thermal stresses calculated on the basis of corresponding flash temperatures [12] are shown as a function of sliding velocity in Fig. 6. Above sliding velocities of 0.5 m s- , the thermal stresses in the vicinity of the microcontact can exceed the bending strength of S&N,. The simulation of thermal stresses with the laser pulse method [43,44] on surfaces revealed, that - above critical values of the laser power density in the range of 0.8 to 2 kJ slR cm- [43] - it comes to spontaneous thermal cracking and decomposition. A comparison of these critical values with tribologically induced frictional power densities (defined as the frictional power (NR =fiFN) divided by the contact area A and a factor of two for the distribution of the

SUO!J!pUOa 11"lapun SaSSaIJS a@VI

01 %U~MO

'8 '%A U!

UMOqS ~j1eayewaqas s! s~e$ta$euI '@IS auglewhQod 103 IapOUI IeaM ay,L *[z~~~] SUO@al aaeJXIS .rak?l JO uopeu~u~e~apay~o~ paIaae~nsaq~MoIaqsyaelaasaq~J0 uope%ola aqL .aae;llns aql MoIaq pue uo aseqd sse@ pue PNE!s-d uaawaq aaeJ;llaluj aq Je an%ge3 hepunoq -u;reSl qli~ tuyueqaaux IeaM pawuyuop L~~ea~ueqaatuow -1aql e Lq IC~u~etu pasnea s! (sayqaolarl r$$q pue arns

-sard~ae~uoaueauu qS!q) suoypuoa %u;rpgspatea!.rqnIun alahas lapun s.yed 2?tupqs *NE~s~NQs 30 JeaM ay,T, N ?S/N !S 1X-%

Uo. yc?p~ -xooqg pm uopemo3ap apseId sIaLeI apyed naM -8-a 'waga hepuoaas qvM sws!ueqaaur Lq paauangu! s! loveqaq 1ea!%o1oqy aql - salrsodum *N~!s 103 IeaM [wrtuaqaoq!.I$ pue sp+~awu *N%S auqIemtmL~od leuopuamoa 103 an@3 hepunoq-ugl8 pur! Suyyaem aug~ewhxav.q qlyi IeaM IeaIueqaatu - smsyueqaaw .maM Oetu~d asaql saprsaq uo!stqauoa UI *L %g r.q 3, 0()8 pm zz JO samwadwa~ $e ~~~eay?waqas uMoqs ale stuyueqaaw JeaM weuyopald aql N~-P@s pue N!L-PN%S sa)!sodwoD aql .103 pue ~NE~s~Q!s .IO~
ql+ saysoduw pue sapy!u UOXJS 103 swsytqaaw .u?aM p+uaqa pue jearueqaawowlaq~ pyueqaam 4eqJ uMoqs sehi 11 sassaw p2urraql pw Ieatmqaaur 30 suoyepyea ql!M uopeuyqum u;r pm sqnsal asaql uo pasea [()z ~~I zJ T~]a.IaqMasIa paqsgqnd uaaq amq sqnsal papelap aqL esanbyqaal s!sL~eue aae3ms sno!.m 30 sueaw dq paw%~saAu~ alahi saae3ms UIOM aql 30 suyueqaaw .waM pue sa%ueqa jeyuaqa Bo~oqdro~

01 8uypeaI sagsuap .IaMod .Ias?q pm$~m aq$ my1 laMoI qmux 0sIe pm speluo:, leaI aqj u! ueql .IaMojqanw a.n? sayfsuap lahtod p~1oya~~3aq$ pue sarnssald aql Bale laeluoa ueam aqj lo3 pue - y10M srql 30 srsal aql 30 weIs aql It2 eale laeluoa uerzyaq aq4 uI QT.] papadxa aq 01 seq 8ug~aq pampur L~le@oIoqy 01 anp %ugaem letmaql *_ s w ~ 0 ueq$ laq%q spaads Bmpgs 103 spewmo.yru 30 kJrIIg!A aql U' Jeql SMOqS seaJe z/r~awg mw.~o:, 30 loor aql Lq paqdrqnw (sa;rpoq %.uyw~oa OMJ aqi uaawaq Blaua leuuaq~

~x~uo:, a9130

'3.0011 w pue 3, zz v P~E!~ JO q@uarIs %u!puaq 01

uospedurm ~I!M&aolaa %;ulp!~JOuoymy e se ssaws IeuuaqL

9+$d

[S/W) AAl!aolaA 6lJlpll~

I
08S-1LS(~661) E81-181 J= M I 7 la ddoys V

,oi

3.22 = 1 IN 01 = Nzi

(OOZ ON) 'N'!SS 'N'!SS LlO!)OlJ) a)elSp!lOS

I
91s

A. Skopp et al. I Wear 181-183

(1995) 571-580

577

Sohd state fnction QN4 I &N4, FM = IO N

(a)

600

-mlcrocracktng - delamlnatlon

\ \

surface fatigue - mlcrocracking - delaminatton - tnbooxidatlon plastIcally deformed surfaces

lb1

Fig. 8. Wear models: (a) mechanically dominated wear mechanism with intergranular fracture and delamination for Si3N4/Si3N4; (b) tribochemically dominated wear mechanism with the formation of stable triboreaction layers or films on Si3N4-TiN and S&N,-BN.

b.03

0.1

Slldlng veloctty

1 0.3 v [m/s] -----&

FN = 10 N

t;

the wear volume at the rotating disk is between fivefold and about two orders of magnitude higher than the wear volume of the stationary specimen [11,12]. A homogeneous amorphous CVD S&N, coating (without glassy grain boundaries) on S&N, reduced the wear rate of the rotating disk by a factor of five for sliding speeds below 0.3 m s- at room temperature. At higher velocities thermal cracking, caused by high thermal stresses and the low thermal conductivity of this coating, probably occurs. Then the wear rate was as high as, or higher than, for uncoated polycrystalline Si,N,. Tribochemically formed phases such as SiOz, SiOH and SiO,N,,, which were also determined in the wear scar as well as in/on wear particles with various surface analysis techniques [11,12,20], could not reduce the wear rate sufficiently. After removing loosely adhering wear particles and loose layers in an ultrasonic bath, investigations with SEM indicate also microfracture on the surface below these layers. Recently performed TEM analysis of cross-sections reveals also intercrystalline microfracture below the worn surface [20]. It should be considered that wear debris influences the tribological behavior, especially for the Si,N&N, pairs. The wear debris often agglomerates and attaches to one or both of the sliding surfaces and forms more or less dense layers. These layers were plastically deformed, broken away and sometimes transferred to the counterbody [11,12]. At higher temperatures thicker layers, built up by multiple passes, were found on the stationary specimen, protecting it against wear. Removing these adhering particles during the tribological tests led to a drop of the friction coefficient to 0.2-0.3 for a short time. 3.3.2. Si,N, composites with TiN or with BN TEM investigations of the structure of these materials did not show any amorphous grain-boundary phase. Additionally, in Si,N,-TiN the residual stresses can be reduced by plastic deformation of TiN grains, which is indicated by dislocations. Therefore, both aspects reduce thermomechanical wear, and tribochemically formed reaction layers like Ti02_, and H3B0,, which

Si3N4-TIN I S&N,-TiN;

tribooxldatlon
k t,
??

lO~"mm'/Nm

f > 0.5

400 surface fatigue -microcracking L,-5 1O %m /N. f,O5

""OW
1

? 3 2 % E F

200 -

Shding velocity

v [m/s] -

%N4-BN

I SI,N.,-BN,

F~ = IO N

surface fatigue - microcracking - delamlnatlon

0.03

01 1 0,3 Slidmg velocity v [m/s] p

Fig. 9. Dominating wear mechanisms with friction coefficients and differential volumetric wear coefficients for self-mated pairs of S&N,, S&N,-TiN and S&N,-BN as functions of sliding velocity and ambient temperature (schematic wear map).

were determined with various surface analysis methods [12,20], become predominant. This is shown in Figs. 8 and 9. Whereas the layers on S&N,-TiN reduce the wear for all testing conditions, the layers on Si3N4-BN fail at temperatures above 150 C owing to the thermal decomposition of H,BO, at this temperature [26,27,46]. 3.4. Wear maps Wear maps with the dominating wear mechanisms, and the corresponding friction coefficients and wear

578

A. Skopp et al. I Wear 181483

(1995) 571-580

coefficients as a function of temperature and sliding velocity are shown schematically in Fig. 9 for S&N, as well as for the composites S&N,-TiN and S&N,-BN. Mechanical wear mechanisms (abrasive wear, grainboundary fatigue, delamination) with intercrystalline cracking and, secondarily, wear particles determine the wear rate of the eight various conventional silicon nitrides. For higher sliding velocities above 0.5 m s-l, high surface temperatures with thermal stresses and thermal cracking can increase the wear rate at room temperature, with the exception of GPSSi,N, (SN73) (Fig. 2) [12]. At higher ambient temperatures the wear rate drops by an order of magnitude above a critical sliding velocity; the critical sliding speed decreases with increasing ambient temperature [11,12] and is nearly independent of the type of silicon nitride used in the material pairing [12]. Distinct plastically deformed surface regions and tribochemically formed reaction layers were found in this low-wear region. Probably the plastic deformation of the tribochemically formed layer and/or of a layer of compacted wear debris and the plastic deformation of the surface regions reduce the mechanical stresses at the contact and lead to the decreasing wear rate. (The thermomechanical wear model of Ting and Winer [45] considers the superposition of thermal and isothermal stresses and the temperature-dependent material properties, and they discuss plastic deformation on the surface as a wear mechanism. A wear criteria equation, which is governed by three non-dimensional parameters, is used to predict the transition between mild and severe wear for hertzian contact. The critical temperature for the transition between mild and severe thermomechanical wear was also calculated.) In any case, the wear rate for the applied test conditions was above 10e5 mm3 N-l m-l, and the friction coefficient was in the range of 0.4 to 0.9, so that conventional silicon nitride materials are not appropriate for most applications under unlubricated conditions. Tribochemically dominated wear mechanisms determine the tribological behavior of the silicon-nitridebased composites with additions of TiN or BN. Stable surface layers of Ti02_, (mainly TiO, in the rutile structure) on Si,N,-TiN and thin films of H,BO, or BN. (H,O) on Si,N,-BN were determined [12]. Simultaneously, according to the nearly glass-free grain boundaries, the wear resistance of these composites against mechanically caused grainboundary fatigue or mechanical wear is increased. For the stability of tribochemical films, the contact pressure is important. There exist critical mean contact pressures and critical frictional power densities. Below these values [12,30] the wear rate drops by orders of magnitude. The critical mean contact pressures and the critical frictional power densities for these composites are dependent on the sliding velocity. At room

temperature the critical mean contact pressures lie in the range between 3 and 14 MPa. The critical frictional power densities increase with sliding velocity and are in the range between 0.3 and 4 W mm- for Si,N,-TiN and between 0.01 and 10 W mmw2 for Si,N,-BN.

4. Conclusions Considering published results of tribological tests and surface analysis on various polycrystalline silicon nitride materials and on the silicon-nitride-based materials S&N,-TiN and S&N,-BN [11,12] and the results of this work, especially the new calculations on friction-induced surface temperature increase, wear mechanisms were elucidated, wear models deduced and schematic wear maps constructed. Wear of Si,NdSi,N, sliding pairs is determined by mechanically, thermomechanically and/or tribochemitally dominated wear mechanisms, depending on composition (addition of secondary phases), structure (grainboundary properties, amount and distribution) and test conditions. Monolithic silicon nitride materials reveal high friction and wear for relatively highly stressed sliding components between 22 and 1000 C. Their tribological behavior is primarily determined by mechanical and thermomechanical wear mechanisms. Mechanisms with secondary effects, such as the formation of wear particle layers, plastic deformation and tribo-oxidation, could not reduce the friction and wear significantly. Therefore, utilization of this material under severe dry sliding conditions seems to be very limited. Additions of TiN or BN to the S&N, matrix can reduce the wear and/or the friction significantly. Besides the optimized mechanical properties of Si,N, composites without any glass phase, the distinct improvement of the tribological behavior was due to the (tribochemical) formation of stable triboreaction layers. The following points for the development of materials with silicon nitride matrix with improved tribological sliding behavior can be deduced from earlier tribological tests, surface analysis and new theoretical calculations: (1) To increase the wear resistance of conventional Si3N4 materials (against mechanical and thermomechanical wear), ?? the content of glass phase must be avoided and/or reduced (if the glass phase is necessary for low-cost sintering, a recrystallisation of the secondary phase and/or an accommodation of the thermomechanical properties of the Si,N, crystal and the phases between them should be beneficial), and ?? the thermal ditfusivity and thermal conductivity (reduce thermal stresses; hinder thermal cracking) must be increased.

A. Skopp et al. / Wear 181-183

(1995) 571-580 S. Danyluk reaction and M. McNallan, on friction in reactive 3033-3039. Influence and wear

579 of triof silicon L

(2) To reduce wear and/or friction, a secondary phase

[31 D.S. Park,

bochemical nitride Amer. silicon

must be added to the silicon nitride matrix, which ?? forms stable, wear-resistant and lubricious oxides or intrinsic solid films, or ?? is plastically deformable (like TiN particles) to reduce internal stresses in the material. The combination of points (1) and (2) led to silicon nitride-based materials with friction coefficients of 0.1 and differential wear coefficients of about 1 X lob7 mm3 N-l m-l, which h ave been measured under severely stressed sliding conditions (high velocities, high loads, high temperatures). The better understanding of the tribological effects from the addition of phases like BN and TiN to silicon-nitride-based materials and from their combination is an important basis for improving the tribological performance of these materials. The aim is to get mechanically and thermally resistant materials. Additionally, third phases should build up layers reducing friction and wear in a wide range of stressing conditions.

products temperatures

at elevated Cerum. Sot., nitride

environments,

75 (11) (1992)

[41 Y. Enomoto,

Y. Kimura and K. Okada, Wearing behaviour of in plane contact, in Proc. 50 Years of Tribo@~

Gong, Tribology - Friction, Lubrication and Wear: Fifr Years On, Vol. 1, Institution of Mechanical Engineers, London, 1987, pp. 173-178.

PI

H. Tomizawa nitride

and T.E.

Fischer,

Friction

and wear

of silicon

at 150 C to 800 C, ASLE Truns. 29 (4) (1986) 481-1187.

161 H. IshigaJd,

I. Kawaguchi, M. Iwasa and Y. Toibana, Friction and wear of hot-pressed silicon nitride and other ceramics, in KC. Ludema (ed.), WearofMaterials, 1985 (Znt. Co& Vancouver, ASME, New York, 1985, pp. 13-21. Silicon nitride rolling contact conditions, ASLE Trans. 2.5 (4) and L.B. Sibley, for extreme operating

B.C.), bearings (1982)

171 B. Bhushan

417-428.

PI

M.G. Gee, C.S. Matharu, E.A. Almond and T.S. Eyre, The measurement of sliding friction and wear of ceramics at high temperature, Wear, 138 (1990) 169-187. 94 (3)

[91 M. Iwasa and Y. Toibana, Friction and wear of ceramics

measured by a pin-on-disk tester, J. Jpn. Ceram. Sot., (1986) 336-343.

[lOI T.H.C. Childs and A. Mimaroglu, Sliding friction and wear up

Acknowledgments
PI

to 600 C of high speed steels and silicon nitrides for gas turbine bearings, Wear, 162-164 (1993) 890-896. A. Skopp, M. Woydt and K.-H. Habig, Unlubricated sliding friction and wear of various S&N,-pairs between 22 C and 1000 C, Tribal. I&, 23 (3) (1990) 189-199. Verhalten von Siliziumnitridwerkzwischen 22 C und 1000 C, A. Skopp, Tribologisches BAM Research Germany).

This work was partially supported with financial contributions by. a project of BMFT (German Ministry of Research and Technology) (project no. 03TOO13A) and in the framework of a BRITE/EURAM project (contract no. RllB-183; proposal no. P-2148). This support is gratefully acknowledged. The materials GPSSi,N, (EKasinS), HPSi,N, (EKasinD), GPSRBSi,N,, HIPSi,N, and Si,N, with 30 wt.% TiN were provided by Elektroschmelzwerk Kempten GmbH, Germany; SSi,N, (ND200) b y Feldmtihle AG, Germany; HIPRBSi,N, (411) by Hoechst CeramTec AG, Germany; GPSSi3N4 (SN73) by NGK Insulators Ltd., Japan; HIPSi,N, (FX-950) by Tungalloyl Toshiba, Japan; and three Si,N,-BN (HIP) materials by HTM AG, Switzerland. Dr. Bartelt (BAM-5.23) is gratefully acknowledged for the development of a program to calculate hotspot temperatures according to the model of KuhlmannWilsdorf. Dr. Klaffke and Dr. Wasche are gratefully acknowledged for helpful discussions. We thank our colleagues at BAM for their support and the intensive cooperation. Mrs. Binkowski is gratefully acknowledged for metallography, photography and figure layout.

PI

stoffen bei Festkorpergleitreibung

Report no. 197, December 1993, ISSN 0938-

5533, ISBN 3-89429-400-o (Wirtschaftsverlag NW, Bremerhaven,

1131 M. Woydt, D. KJaIIke, K.-H. Habig and H. Czichos, Tribological

transition phenomena of ceramic materials, Wear, 136 (1990) 373-381.

[I41 S.C. Lim and M.F. Ashby, Wear mechanism maps, Acta Metall.,
35(l) (1987) l-24. and method development, Munro, Ceramic wear maps: J. Sot. Tribal. Lubric. Eng.,

P51 S.M. Hsu, D.S. Lim and R.G.

Concept 47 (1) (1989) 49-54.

WI

S.W. Lee, M.C. Shen and S.M. Hsu, Ceramic wear maps: Silicon nitride, Key Erg. Mat., 89-91 (1994) 751-756.

P71 X. Dong and S. Jahanmir, Wear transition diagram for silicon

nitride, Wear, 165 (1993) 169-180.

P81 H.C. Wang,

K. Kato

and N. Umehara,

Water

lubrication

of

93, Proc. 6th Int. Con& ceramics, in M. Korma (ed.), Eurotrib on Triboiogy, Budapest, 1993, Vol. 3, International Tribology Council, Budapest, 1993, pp. 149-155.

P91 A. Skopp and M. Woydt, Charakterisierung des tribologischen

Verhaltens von keramischen Gleitpaarungen mit modemer Oberllachenanalytik, Materialwissenschaft & Werkstoflechnik, 22 (8) (1991) 289-300.

PI

I. Diirfel, W. Gesatzke, W. Gsterle and A. Skopp, Tribological induced microstructural changes in subsurface layers of S&N,based materials, Key Eng. Mat., 89-91 (1994) 763-767. M.N. Gardos, The effect of anion vacancies on the tribological properties of rutile, TriboI. Trans., 32 (4) (1988) 427-436. substrates, in L.E. Pope, L.L. Fehrenbacher and W.O. Winer

References
[l] [2] K. Kato, Tribology of ceramics, Wear, 136 (1990) 117-133. T.E. Fischer and H. Tomizawa, Interaction of tribochemistry and microfracture in the friction and wear of silicon nitride, Wear, 205 (1985) 29-45.

PI

1221 M.N. Gardos, The tribooxidative behavior of rutile-forming

(eds.), New Materials Approach to Tribology l7aeory and Applications, MRS Symp. Proc., 140, Materials Research Society, Pittsburgh, PA, 1989, pp. 325-338.

580

A. Skopp et al. I Wear 181-183

(1995) 571-580

1231 M.N. Gardos, The effect of Magneli phases on the tribological properties of polycrystalline rutile, in M. Kozma (ed.), Eurotrib 93, Proc. 6th Int. Conf Vol. 3, International 201-206. 1241 G.W. Rowe, Some observations boron nitride and of graphite, on the frictional Wear, 3 (1960) behavior of 274-285. Tribology on Tribologv, Council, Budapest, 1993, Budapest, 1993, pp.

and M.N. Gardos, 1251 J.M. Martin, Th. Le Mogne, C. Chassagnette Friction of hexagonal boron nitride in various environments, Tribal. Trans., 35 (3) (1992) 462-472. F.A. Nichols, R.A. Erck and D.E. appli1990, Vol. acid films for tribological Nagoya, Tokyo,

WI

A. Erdemir, Busch, cations,

G.R.

Fenske,

Self-lubricating

boric

in Proc. Jpn. Int. TriboIogy Co@, Society of Tribologists,

3, Japanese

1990, pp. 1797-1802.

1271 A. Erdemir, G.R. Fenske, R.A. Erck, F.A. Nichols and D.E. Busch, Tribological properties of boric acid and boric-acidforming surfaces, Part I: Crystal of formation and boron 179-184. F. Honda nitride,J. and K. Nakajima, The accompanying friction and wear of silicon chemistry and mechanism of self-lubrication of boric acid,Lubric. Eng., 47(3) (1991) 168-173; of boric Lubti. surfaces,

Part II: Mechanisms acid films on boron Eng., 47 (3) (1991) 1281 Y. Imada, tribological reaction

and self-lubrication

oxide containing

K. Kanamura, titanium

nitride containing

Triboi., 114 (1992) 230-235.

1291 M. Iwasa and S. Kakiuchi, Mechanical and tribological properties of SisN4-BN composite ceramics, J. Jpn. Ceram. Sot., 93 (10) (1985) 661-665 (in Japanese). composite materials 1301 M. Woydt and A. Skopp, Ceramic-ceramic with improved friction and wear properties, Tribol. Intl., 25 (12) (1992) 61-70.

[311 R. Wlsche,
Bartz many, bometer (ed.),

D. KlatIke

and A. Skopp, nitride

Tribological

behavior in J. Ger-

of silicon nitride

with boron

as a solid lubricant, Esslingen, Habig, Ostfildem,

Tribology 2000, 8th hat. Koll. Esslingen, 14-16 Jan. Akademie and K.-H.

1992, Vol. 1, Technische

1321 M. Gienau,

1992, pp. 10.2-l-10.2-10. M. Woydt Hochtemperaturtribis loo0 fur Reibungsund Verschlei8untersuchungen 29 (7,8) (1987) 197-199.

C, Materiahm @mg,

1331 (a) M.F. Ashby, H.-S. Kong and J. Abulawi, T-MAPS 4.0: User Manual +Background Reading: - On surface temperatures at dry sliding surfaces, Cambridge University Engineering Dept., Cambridge, UK, 1992; (b) M.F. Ashby, H.-S. Kong and J. Abulawi, Temperature maps for frictional heating in dry sliding, TriboL Trans., 34 (1) (1991) 577-587. 1341 F.A. Nichols and M.F. Ashby, T-M4PS:A PCcode forcalcuiating average and local ( FLASH) temperatures on sliding interfaces, in Contact Problems and Surface Interactions in Manufacturing and Tribological Systems, Int. Conf, New Orleans, LA, 1993, Vol. 4, ASME, New York, 1993, pp. 75-86. Flash temperatures due to friction and 1351 D. Kuhlmann-Wilsdorf, Joule heat at asperity contacts, Wear, 105 (1985) 187-198. Demystifying flash temperatures, I. [361 D. Kuhlmann-Wilsdorf, Analytical expressions based on a simple model; II. First-order approximations for plastic contact spots, Mat. Sci. Eng., 93 (1987) 107-133. H. Blok, The flash temperature concept, Wear, 6 (1963) 483494. J.F. Archard, The temperature of rubbing surfaces, Wear, 2 ;;; (1958/59) 438-455. at [391 J.C. Jaeger, Moving sources of heat and the temperature sliding contacts, L Proc. R. Sot. New South WaZe.s, 76 (111) (1942) 203-224. [401 H.S. Carslaw and J.C. Jaeger, The Conduction of Heat in Solids, Oxford University Press, London, 2nd edn., 1959. surface t411 J.A. Griffioen, S. Bair and W.O. Winer, Infrared temperature measurements in a sliding ceramic-ceramic contact, in D. Dowson, C.M. Taylor, M. Godet and D. Berthe (eds.), Mechanisms and Surface Distress, Proc. 12th Leeds-Lyon Symp. on TriboZogy, Butterworth, Guildford, 1985, pp. 238-245. on temperature calculation [421 G. Bartelt, Personal communication and program development, BAM, Berlin, 1993. [431 R. Benz, A. Naoumidis and H. Nickel, Thermal shock testing of ceramics with pulsed laser irradiation, J. Nucl. Mat., I50 (1987) 128-139. WI K.-H. Tonshoff and S. Bartsch, Thermal load simulation by laser-beaming, Ceram. Sci. Eng. Proc., 9-10 (9) (1988) 1453-1462. thermal influences [451 B.-Y. Ting and W.O. Winer, Friction-induced in elastic contact between spherical asperities, J. TriboZ., 111 (1989) 315-322. [461 A. Erdemir, A review of the lubrication of ceramics with thin solid films, in S. Jahanmir (ed.), Friction and Wear of Ceramics, Marcel Dekker, New York, 1994, pp. 119-162.