Wear 205 (1997) 144-152
Tribological properties of carbon- and nitrogen-implanted Si(100) P. Kodali *, M. Hawley, K.C. Walter, K. Hubbard, N. Yu, J.R. Tesmer, T.E. Levine, M. Nastasi Materials Scienceand TechnologyDivision. MS K764. Los AlamosNational laboratory, Los Alamos, NM 87544. USA
Received 10 April 1996;accepled I I July 1996
Abstract The knowledge of tribological properties of silicon is important due to its potential application for micromechanical devices. The present work focuses on the friction and wear studies of single crystal silicon (100) implanted with carbon and nitrogen at 60 keV. These tests were performed using a pin-on-disc (POD) apparatus with a 52100 pin, under contact stress of 225 MPa and at 35% relative humidity. Changes in surface roughnessbefore and after the wear tests were studied using atomic force microscopy (AFM). Transmission electron microscopy (TEM) investigations and ion-channeling measurements revealed amorphous phases for all the implanted species at all doses. Although the surface hardness measurements performed by nano-indentation indicated no ;.mprovementin surface hardness for samples implanted with carbon, the tribological studies indicated an improvement in wear and frictional properties. We relate this improved wear behavior to the formation of amorphous SiC. Keywords: Implantation;Frictioct;WeaK,Microstmctm'c;Surfacehardness;Channeling
1. Introduction For the lust three decades silicon has been the dominant material in micro-electronic and micro-mechanical devices. There are several reports aimed at understanding the friction and wear behavior of this material both at a microscopic and macroscopic level. These studies include surface modification by implantation of a single gas species [ 1-6]. Ion implantation is an effective and widely used technique to improve the tribological properties of metals, ceramics and polymers . Ion implantation is useful in modifying the tribological properties due to (a) control over the composition of the modified layer; (b) strong adhesion of the ionimplanted layer to the subslrate. Gupta and Bhushan  studied the effect of carbon and nitrogen implantation in single cry_sial and polycrystallin¢ siJicon as a function of dose and energy. Their studies indicated a change in the chemistry of the near surface region which improved the tribological properties of silicon. By implanting argon ions into silicon, J. Lekki et al.  illustrated that no chemical changes are required in the modified layer ,o improve the friction and wear properties. They suggested that the defects created during the implantation propagate below the surface during the wear process and help reduce ~ction. Our current investigation differs from previous work by energy of the ions * Con~pondingauthor.Tel.: 505 665 3036;fax: 505 665 2992. 0043-1648/97/$17.00© 1997ElsevierScienceS.A.All rightsreserved PII $0043-1648 (96) 07290-0
implanted and contact stresses. Our objective was m understand the role of mixed implants in silicon. We studied the tribologicai properties of silicon implanted with single implants of carbon and nitrogen as well as a combination of the two.
2. Experimental details Samples of single crystal silicon with an orientation of (100) were implanted with ions of carbon and nitrogen with an energy of 60 keV at room temperature. Typical doses used were 4X 10t7 C + ions cm -2, 4X 1017 N + ions em -2, 8 X 10~ N + ions cm -2, 8 X 1017 C + ions c m - 2 and 4 x 1017 C + ions c m - 2 + 4 X 10 I'p N + ion:, i;m -2. To minimize surface contamination a cold trap was used during implantation to maintain a low background pressure of about 2 × 10 -6 Tort. Rutherford backscattering spectrometry (RBS) was used to determine the retained dose, changes in composition and location of the peak dopant concentration after implantation. The crystallinity and depth of the damaged layer were examined by ion channeling. A 2 MeV He + ion beam was used for both channeling and RBS experiments. Roughness of the samples before and after implantation was examined by atomic force microscopy (AFM). These measurements were carried out using a Nanoscope-IH AFM. The microscope was
P. Kodali e! a L I Wear 20~
operated in contact and tapping mode with silicon nitride and silicon tips, respectively. Root mean square (RMS) roughness measurements were used to calculate the plasticity index, a parameter which describes the deformation behavior of asperities. The plasticity index, q~, is given by
where o'p = composite standard deviation ofth¢ aslg'.rity peak l~ights= ( ~ + ~ ) u 2 : 8=composite mean radius of cur-
vature= l/(/]l +~z); H=hafdness of the softer material; E = [ ( I -v~)lE, +(l-~,Z)lEz] -~. The additional vmables used above ~l a~l ~ arc, the radius curvature of the pin and the disc respectively; or, the RMS roughness; E, Young's modulus; and v. Poisson's ratio.
¢ = ( E/ ~ C-C~J /3)
Fig. I. Alomic force microscopy images: (a) virgin; (b) 4 × 10I~C* ions cm-2: (c) 4 x 10L7N* ions cm-2; (d) 4 x 10t~ C + ions cm-2+ 4 × 10s7N* ions cm-2; (e) 8 × 1017C + i o n cm-2; (f) 8 × 10I~ N+ions ¢m-z.
P. Kodaliet al. / Wear205 (1997) 144-152
The subscripts I and 2 indicate the values for the disc and the pin respectively. These values are useful in evaluating the nature of the contact between the pin and the disc. The plasticity index, ~p, describes the transition from elastic to plastic deformation of surface asperities. The deformation is predominantly elastic if the value of cp is <0.6 and plastic if ~,> !.0. The deformation is elasto-plastic for 0 . 6 < ~p< 1.0. The values of the constants used in the above equation are v I =0.35; v2=O.2g; Et = 180 GPa; E 2 = 2 0 6 GPa. We have taken the RMS roughness of the surface equivalent as the standard deviation of the asperity heights . Hardness measurements were performed using a commercially available nano-indenter. Hardness indents were made using a Berkevitch trimigalar pyramid diamond indenter tip. This setup measures the load on the indenter tip as a function of displacement. Hardness is given by H=L(d)/A(d)
where L is measured load, A is the projected area of the indent, and d is the displacement from the surface. Multiple indents were made on each sample and the data were averaged. The data were corrected for elasticity. Transmission electron microscopy was used to examine the changes in the microstructure after surface modification. Cross-sectional samples were prepared and examined using a Jeol-2000FX microscope. Pin-on-disc wear tests were performed without lubrication at room temperature in air with 35% relative humidity and under a maximum contact stress of 225 MPa. A 6 mm diameter sphere of 52100 chrome steel is used as a pin. In this setup, the pin is mounted on a weighted arm which rests on a flat rotatin~ disc. The pin was polished with a 0.25 p~m diamond paste and examined for major processing defects using an optical microscope. Prior to each wear test, both pin and disc were degreased by ultrasonic cleaning in acetone followed by isopropyl alcohol. A sliding speed of 9.4 mm s - ~ was employed. Wear volume measurements were made using a profilometer to determine the wear track profile perpendicular to the sliding direction, ignoring the ridges of
material on the side of the wear track. The morphology of the wear tracks was examined by AFM.
3. Results and discussion
3. I. Roughness characterization Fig. I ( a ) - F i g . l ( f ) show roughness measurements obtained by AFM before and after implantation. Lighter regions correspond to a higher degree of elevation in these AFM figures. The RMS roughness value of the pin before implantation was found to be 3 nm. After the wear tests no damage was observed optically on the pin. The contact radius was about 30 ttm before the test and this value varies with time. To make a measurement on a spherical surface for its RMS roughness is difficult. The pin chosen is harder than the material against which it is slid. In this case the disc is going to wear more than the pin. Hence as an upper boundary, we have assumed the RMS roughness of the pin m be close m the R_MS roughness of the disc in our calculatious. Table 1 summarizes the values of the measured RMS roughness for the virgin and implanted samples before and after wear tests. These values of roughness were used to evaluate the plasticity index. Ttte values of the plasticity index between the pin and the disc before the friction and wear test were less than 0.6. This suggests that the contact between them is elastic for all the samples [ 1 ]. The calculated values of cp, the plasticity index, for after the wear test for 150 cycles is < 0 . 6 for all the samples, with the exception of the sample treated to a dose of 8 X l0 t? N + ions cm -2. The sample implanted with 8 X 10 t~ N + ions cm -2 had a plasticity index of 0.75 after a wear test for 150 cycles, indicating the deformation is elastoplastic at a lower number of cycles. The plasticity index value for carbon-implanted samples and the sample treated to a dose of 4 x 1 0 t7 N + ions cm -2 was greater than 0.6 but less than unity as the test continued for higher number cycles. The nature of the deformation changed from elastic to elastoplastic for these samples. In the unimplanted ease and the sample implanted with 8 × l0 t? N + ions cm -2 the plasticity
Table 1 Roughnessvaluesof the disc beforeand after wear tests Treatment
Virgin 4x 10~C + ions cm -2 4 × l0 t7 N* ions cm -2 4 x l 0 v~ C ° ions cm-2 + 4 × 10I~N ÷ ions crn-z 8 × 10I~N* ions cm - 2 8 × 1017C ~ ions cm-2
For a scan area 2.5 X2.5 p.Am RMS roughness
of the 150 cycles weartrack (nm)
RMS roughness of the 660 cycles wear track (nm)
0.14 0.21 0A2 0.91 2.4 0.15
20.6 26.8 2.0 11.2 79.7 8.8
62 ' 31.1 49 ' 21 44.5" 18.1
• Scanningwidth is less than the width of the wearu'ack. See text for discussion.
P. Kodali et aL I Wear 205 (1997) 144-152
Table 2 Rcsulcsof the ion-heamanalysis Sample
Virgin 4 × 10~VC ÷ ions c m - = 4 X 10 t'/N + ions c m -2 4 × 10*vC ÷ ionscm -~ + 4 × 101~N * ions c m -2
8 × 1017N÷ ions cm- 2 8 × I0 '7 C + ions cm-2
Peak concen;~aUon • (atomic fraction)
~ oflhe peak ~tomic concemrafion" (rim)
Pull width half n'mximum (FWHM) • (rim)
Wklth of the ionmodified l ~ r o, (nan)
160 160 .t80,2
140.6 150.6 140.0
300 290 350
• Calculated from the R B S data by fitting to a Ganssian distribution. b Includes implantation and radiation damage and is calculated from channeling data. Table 3 Comparison o f the width o f the ion-modified layer from ion-bantu amd T E M rr,eastwernents
Width of ion-medifiedlayer (nm) (channelingmeasutemems)
Width of ion-modifiedlayer (nm) (TEM nzzwmnnants)
Virgin 4 x 1017 C + ions cm- 2
4 × 10 I"/N + ions cm -2 4 × 10 I"/C ÷ iofl$ cm - 2 + 4 X 1017 N ÷ ions c m -2 8 X 1017N ÷ ions cm -z 8 × 10'~ C + ions cm-~
index value exceeded 1.0 indicating the deformation behavior is predominantly plastic.
3.2. Crystallinity and microstructural characterization Fig. 2(a) shows the channeling spectra of the virgin sample and the sample implanted with 8X 1017 C + ions cm -2. Fig. 2 ( b ) shows the RBS spectra of the sample implanted with 8 × 10 ~ C + ions c m - 2 along with its simulated specw,L The peak concentration, location of the peak concentration and concentration full-width half maximum were calculated from the RBS data. The channeling data indicated the yield was the same for the channeled and random directions, suggesting the modified layer was amorphous. The width of the ion-modified layer, including implantation and radiation damage, was calculated from the channeling data. The width of the ion-modified layer and the peak concentration was observed to increase with dose. The results are summarized in Table 2. Additional investigation of the crystallinity of the ion-modified layer was done by transmission electron microscopy (TEM) studies. These indicated that the ion-modified layer was amorphous at all the doses. The dark field images of the samples implanted with 8 X 1017C + ions c m - 2 and the mixed implant (4 × 10 I~ C + ions c m - 2 + 4 × 10 I~ N + ions cm -2) are shown in Fig. 3 (a) and Fig. 3 ( b ). The peak concentration of the implanted species (C and N) measured by RBS in the mixed implant sample exceeds the 57 at.% required for the formation of stoichiometric silicon nitride. A density change in this mixed implant sample is observed at 100 nm. This change corresponds to the projected range of nitrogen ions in silicon carbide. RBS studies indicate a peak atomic concen-
3o -o ,,~m "02O
•_ _ ~+'++':'+'+~-+-+.'-+~. "-~"i+.~l ~ +'+-+"+-+": < o~ ,.o
Channel 0.4 i
•-~ +.5 et to ,q
at m . . c / m - c / I
~nnel Fig. 2. (top) Channeling specffa o f the virgin samp/e and ~ i m p l o r e d with 8 × 1017 C + ions cm-±; (bottom) Rutherford back scatteringspecua
of the sampleimplantedwith 8× 1017C+ ionsem-2.
P. Kodali et aL I Wear205 (1997) 144-152
tration of the nitrogen ions is approximately 0.50 in the sample implanted with 4 x !0 I~ N + ions cm-:. When the dose was increased to 8 x 10 ~7N + ions c m - 2 the concentration of nitrogen ions changes to 0.70. This peak concenlration of ions is greater than the amount of nitrogen (57 at.%) required to form stoichiometric silicon nitride. Fig. 3(c) shows the TEM dark field image of the sample implanted with 8 × 10 ~7 N + ions cm -2. These results suggest that the low density region is due to nitrogen ions trapped in the form of bubbles. Fig. 3(d) shows a typical selected area diffraction pattern of the modified layers. The diffraction patterns have diffuse rings, indicating the amorphous nature of the modified layers. The width of the damaged layer measured by TEM studies was compared with the width of the ion-modified layer determined by channeling data in Table 3. There is a small discrepancy between TEM results and ion-channeling measurements. Density changes that occur due to implantation that have not been taken into account in the channeling-based calculations may account for this discrepancy.
3.3. Surface hardness measurements
Fig. 4(a) and Fig. 4(b) show the results of the nano-hardness meas,~ements. The data are presented in units relative to the unimplanted silicon. The samples implanted with carbon did not show any change in the surface hardness. The sample implanted with 4 × 10 ~7 N + ions cm-2 exhibited a slight increase in hardness, which at the lower dose of nitrogen is due to the alloying of nitrogen ions with silicon. A decrease in hardness was observed when the dose was increased to 8 × 10 ~7 N + ions cm-2. TEM studies revealed a decrease in the density of the sample and possible formation of nitrogen bubbles at this dose. The decrease in hardness is related to the loss of density in the sample. The trend in the surface hardness of the sample implanted with 4 X I0 l~ C + ions cm- 2 and 4 × 10 ~v N + ions c m - ~"strongly matches the sample implanted with 4X 10~ N + ions cm -2. This trend indicates that the hardness behavior in a mixed implant is governed by nitrogen ions. A noticeable increase in hardness is observed for the sample implanted with 8 × 1017 C + ions
(b) Fig. 3. TEM dark fieldimages:(a) 8× tO" C+ ionscm-Z; (b) 4× 10I~C+ ionscm-2+ 4× l017N+ ionscm-2; (c) 8× 10la N+ ionscrn-2; (d) selected area diffractionpatternof the modifiedlayers.
P. godali et al. / Wear 205 (1997) 144-152
Fig.3 (cantOned) cm - 2. We speculate that carbon remaining after formation of silicon carbide can be in the amorphous graphitic form. Bhushan and Koinkar  reported no change in hardness up to a dose of 2.5 × 10'~ C + ions cm -2 at 100 keV. Burnett and Page  reported a critical dose of 4 × 10'~ N + ions cm -~ in silicon showed a surface softening. Our results are consistent with these previous investigations.
3.4. Tribological studies Friction data of the unimplanted samples and implanted samples are shown in Fig. 5(a) and Pig. 5(b). The carbonimplanted samples show a low coefficient of friction for 280 cycles. After 280 cycles the coefficient of friction of the carbon-implanted sample changes to the value of the unimplanted sample. The sample implanted with 4 × 1017N + ions c m - 2 showed a slight decrease in the coefficient of friction. However, the sample implanted with 8 × 10 '~ N + ions cm -2 showed an increase in the coefficient of friction. The sample implanted with the mixed implant ( 4 × 1 0 '~ C + ions
c m - Z + 4 X 10'~ N + ions cm -2) showed alow coefficient of friction for a low number of cycles. These data suggest that carbon is responsible for the low coefficient of friction. The increase in the coefficient of friction in the nitrogenimplanted sample is due to density changes and a rougher surface of the sample. Friction is governed by the magnitude of the subsurface fatigue caused by the sliding between the disc and pin. The modified layer for N-implanted and Cimplanted samples is different in chemical composition. The response of this chemically modified layer with the slider is different hence the different trends in friction. The wear factor, defined as the wear volume per unit load per unit sliding distance, after 150 and 660 cycles for all the implanted samples as well as the virgin sample, is shown in Fig. 6(a) and Fig. 6(b), respectively. A reduction in wear factor was observed at 150 cycles for all the implanted samples..'!1~e lowest wear factor was exhibited by the sample implanted with 8X 10'~ C + ions cm -2. As tl~ wear test continued for a longer number of cycles, the wear factor of the sample implanted with 8 x 10t~ N + ions c m - 2 exceeded
P. Kodali et al.
. . . .
. . . .
Wear 205 (1997) 144.-152
. . . .
o ~z:o t' c" ~ a , / ~ = ' +
e z l O tv C * t r a m / e r a ~ 4 x l O tv C * l e ~ t / e m | + 4 z l O I v l ~
--'~A/l~lO~v ~ 0.8
i e m l / e m II
...'.,i o.z •
. . . .
4xlO It N ~ ~ i e
. ."."~'~- ~/ • e ~ * ?/~t'~ ...t
t .... 200
i .... 300
n .... 400
i .... 500
o .... 600
. . . .
o ~10 ~" ~¢' tmm/~an"
e~:o~ ~ iota/era"
+ 4~1#' & Imp/=== + 4x10xvk'* imut/e.m8
" - . - e z t ¢ ' t~ tern/re' 0.8 , ....-.4xn~' & tern/era* + ~ t O " I¢ lem/em'
. ~, . . . - ~'~" -"::~"""._;'~'~.~ ~.~.~~.~.'~.-'.'~ ~ .
IP 1.5 I::
._._exto1~ ¢, io~l=ns ...... 4 x t O I1 e • ~ i e |
"" ";~ ""~'~ ";~"
°'4 ~ ~ ' ~ ' ~ , ' t / t , , ~ t j ~ , l , , .
, IO0 ,
. . . .
. . . .
. . . .
Depth ( r i m )
.... n .... t .... 200 300 400
Fig, 4, Relative hardness as a function of d~pth: (a) carbon implanted s a m p l e s ; (b) nitrogen implanted samples.
Fig. 5. Coefficient of friction as a function of cycles: (top) carbon implanted samples; (bottom) nitrogen implanted samples.
the virgin sample. This increase in wear factor may be attributed to the decreased density in the material at this dose. The samples implanted at other doses exhibited a similar trend in wear behavior at low and high numbers of cycles. The wear factor for the sample implanted with both carbon and nitrogen is less than the wear factor for the sample implanted with 4 × 10 ~7 N + ions cm -2 and greater than the sample treated to a dose of 4 x ! 0:7 C + ions cm- 2. This result suggests that carbon had an influence on improving wear properties of the mixed implant sample. The wear factor is reduced by half for the sample implanted with 8 x 10:7 C + ions cm- z. This sample has also shown a low coefficient of friction. This behavior suggests that there is a correlation between friction and wear properties. In our studies, hardness does not seem to dictate the wear properties. This improved behavior of carbonimplanted samples is due to the amorphous ion-modified layer and the chemical compatibility of this modified surface with its sliding counterface 52100 steel. The morphology of the wear tracks after 660 cycles of sliding as studied by AFM is shown in Fig. 7(a)-Fig. 7(c). Similar wear morphologies for 160 and 660 cycles were observed. No measurable wear was observed on the slider. There was a transfer of wear debris to the slider. However, this wear debris is loose and is easily removed by blowing gas. The parallel grooves mm~ing along the wear track sug-
gast an abrasive wear mechanism (Fig. 7(a)-Fig. 7(c)). Loose wear debris seen inside the track indicates third-body wear. The severity of the abrasive wear scales with the amount of plastic deformation which can be visualized by the plasticity index value. Abrasive wear was dominant in the sample treated with a dose of 8 X 10:7 N ÷ ions cm - 2, whereas adhesive wear was dominant in carbon-implanted samples. Abrasive and adhesive wear mechanisms seem to play an equal role in the unimplanted sample, Fig. 7(a). However, in the case of samples implanted with carbon, the adhesive wear mechanism dominated over the abrasive wear mechanism, Fig. 7(b) and Fig. 7(c). We believe that the susceptibility of the surface to form an oxide layer at the humidity that we have investigated is an important factor in deciding which wear mechanism plays an important role. Our data suggest that the nitrogen-implanted samples have poor resistance to the formation of an oxide layer as compared to the carbonimplanted sample. The plasticity index expression is derived for static contact. The current experimental investigation corresponds to repeated sliding of the pin on the disc. In addition to the changes in the surfaces due to wear, the changes on contact radius as the test progresses may contribute to the discrepancy in predicted and observed nature of the deformation at a lower number of cycles.
P. Kodali et oL I Wear 205 (1997) 144-152 fh~
profiles of the implanted species used in the current investigation.
Fig. 6. Wearfactoras a functionof different speciesand dof,es: (a) 150 cycles; (b) 660 cycles.
Implantation of carbon and niu'ogen at 60 keV at doses of 4 X 10~7and 8 × 10t~ ions cm -2 in single crystal silicon produced an amorphous layer. The thickness of the damaged layer increased with dose. The depth of the damaged layers at different doses measured by ion-beam techniques match with those obtained by TEM. The plasticity index evaluated from these roughness measmen~nts indicated the nature of the deformation between the sliding surfaces. Our investigations soggest that the chemically modified surface in the carbon-implanted samples is responsible for the improved behavior in friction and wear properties at both doses (4X10 l? C + iONSem -2 and 8X1017 C + iop~ cm-2). In samples implanted with dual species (C and N) the hardness behavior was governed by nitrogen implant and wear behavior by both carbon and nitrogen implants. Although the data suggest that friction and wear behavior in the current investigation are related, no mathematical relationship between them was derived. Our results also suggest that hardnessdoes not dictate the wear behavior. In spite of a chemically modified surface, the nitrogen-implanted samples did not show any improvement in tribological properties at the contact stresses that we have investigated. Our studies demonstrate that modifying the surface chemistry does not necessarily improve the tribological parameters,
The authors acknowledge the technical help of and discussion with C. Maggiore, M. Hollander and C. Evans. All ionbeam work was performed at Los Alamos Ion Beam Materials Laboratory. The authors would like to acknowledge the referees for helpful suggestions.
0 50 I00 pm Fig. 7. Morphologyofthe wearIracksmeasmedbyatomicforcemicroscopy for scanageaof 125× 125 mm~:(a) virgin; (b) 4 × I0~ C~ ionscm-2, (c) 8)< 10~C÷ ionscm-~.
Studies of carbon implantation in silicon by Gupta et al. [5 ] suggest that the tribological properties can be improved in comparison to our observations by increasing the contact stress. We speculate that the difference in our results may be due to low contact stress, sliding speed and the different depth
 J.A. Greenwoodarid J.B.P. Williamson.Proc. R. Soc. Lomb. A295 (1966) 300, [2l L Lekki,Z. Statchura,N. l~ikschas, B. Cleft, M. Cholewaand G.J.F. Legge.J. Mater. Res., 9 (1994) 91--95.  PJ. Bumenand T.F. Page,./.Mater.Res.. 9 (1984) 91.  B.K.Guptaand B. Bheshan.Surf. Coat. Teclmol.. 68/69 (1994) 564570.  B.ICGupta,J. Chevallierand B. Bhushan.Trans. ASM~ 115 ( 1993~ 392-399.  T. Miyamoto,S. Miyakeand R. Kaneko,Wear. 162-164 (1993) 733738.  B. Bhushanand B.K. Gupta, Handbook of Tribology, McGraw-Hill, NewYork, 1991.  B. Bhushanand V. Koinkar,J. AppL Phys.. 75 (1994) 5741-5746.
205 (1997) 144452 191 R.D. Am&. P.B. Davies. J. Halling and T.L. Whomes. Tribdogy. Sptinger-V&g. prinred in Hong Kong. 1993.
Biographies Padma Kodali is a doctoral candidate in Materials Science and Engineering at the University of Maryland. Her doctoral work is focused on mechanical and tribological properties of ion-beam processed surfaces, under supervision of Dr. Michael Nastasi at Los Alamos National Laboratory
(LANL). Dr. A. Christou is her thesis adviser at the university. Marilyn E. Hawley is a technical staff member at LANL. She received her Ph.D. from The Johns Hopkins University. Her work is focused on two primary targets: ( 1) the development of new scanning probe techniques and (2) the application of scanning tunneling and atomic force microscopies (STM and APM). and other related scanning proximity probe characterization techniques to a wide range of applications and environments, including ultra-high vacuum, liquid cell, and magnetic and electrical fields.