XPS characterization of nitrogen implanted silicon carbide

Nuclear

Instrumenls

and Methods

in Physics Kescsrch

HO5

Nuclear Instruments & Methods in Physics Research St,r 1,011 B

(1992) X52-356

North-Holland

XPS characterization

of nitrogen

implanted

silicon carbide

A study ha\ been made of the surface hardness and chemical structure of nitrogen implanted ion5 were carried

out with doses ranging from 5~

made the near-surface

photoelectron

moved

the surface

concentration.

to I x IO’”

area soften. Tht: depth profile and chemical

by XPS (X-ray toward

10”

spectroscopy)

combined

as the dose increased.

The Si 2p spectra implied

The

the formation

with Ar binding

’

Ionx/cm2.

Microhardness

bonding states of clementa sputtering.

energies

The depth

rclatcd

SK’. Implantations in implanted

of the maximum

to Si?p

of I.50 keV N:

tests showed that N implantations Sic were inve
and C Is chnngetl

concentration

according

to the N

of Si tC, N;:

1. Introduction

2. Experimental

Ion implantation is a powerful method for modifying the near-surface properties of metals, ceramics, polymers and semiconductors. Recently. it has bc,cn found that ion implantation in ceramics results in significant changes in mechanical properties of their surface layers such as hardness and wear resistance [I ,2]. From many results it has been concluded that the change in structure induced by ion implantation plays an important role in the improvement of surface mcchanical properties. For example. Sic was implanted with Ar or Ag ions, and the tribological propcrtics have been investigated in conjunction with the structure of ion implanted layers. The experimental results showed the reduction of hardness and friction coefficient was due to the amorphization of Sic or/and the formation of amorphous carbon [3]. McHarguc et al. [4] also demonstrated that implantation of 62 kcV Nf to fluences of I x 10" N ‘/cm’ produced an amorphous surface that had a reduced hardness. In case of N-implanted Sic’, Nastasi et al. proposed that the reduction of the friction coefficient appears to be correlated with the thermodynamic tendency to form “nitride-like” bonds [S], although the bonds were not dctcctcd directly. In this report, the change in surface hardness was measured by a Knoop hardness tester, and the indcntation shape was observed by SEM. The atomic ratio and chemical bonding states of Si, C, N, and 0 atoms in N-implanted SiC as a function of depth were invcstigatcd by XPS measurements combined with Art sputtering.

The substrates were polycrystalline sintercd silicon carbide (Sic) sheets containing 0.3 wt.% boron and 3 wt.% carbon as a sintcring aid. The size of specimens was 25 mm X 25 mm X 2 mm. and the surfaces were polished to a mirror finish. Nitrogen molecule ion implantations into Sic wcrc performed with dohcs ranging from 5 x IO’” to I x IO’” NT ions/cm’ at an energy of IS0 kcV and a pressure of 2 x IO ” Torr. The beam current density was 1.3 to I.8 LA/cm’. The implantations were performed with a RIKEN 200 kV low current implantcr. Microhardncss tests were performed with a Knoop indenter at atmospheric room temperature using a Matuzawasciki microhardness tester, model DMH-21s. The load range used was from 5 to SO gf using ;1 dwell time of 15 s. The hardness was evaluated by averaging twelve experimental values. In order to examine the brittleness, the shape of Vickers indentation of I kgf marked on specimens was examined by JEOL JSM84OA scanning electron microscopy (SEM). X-ray photoelectron spectra (XPS) were obtained to identify and quantify the clcments and chemical bonding states in the nitrogen implanted Sic. A VG Escalab MK 11 spectrometer employing MgK,, radiation (1253.h cV) was used for the XPS studies. Standardization was achieved using Au4f,,,: (84.0 cV) and/or C Is spectrum (285.0 cV) of a characteristic component. The depth profile of elements was obtained by XPS spectra combined with sputtering of 5 keV Ar+ ion rate is approxi0 beam. The sputtering matcly IS A/min. Various mathematical operations

OlhX-583X/92/$05.00

0

1992 - Elsevier

Science Publishers B.V. All rights reserved

such as smoothing. convolutions wcrc ware package.

background removal and peak dcavailable with a VG analysis soft-

of the Sic surface was improved bv nitrogen implantation for doses of 5 x 10” ions/cm’ or more. 3.2. DqXh distribrf tion

3. Results

and discussion

3.1. Microhurdtzuss

Fig. 1 shows the Knoop hardness for unimpl~~ntcd and implanted Sic as a function of applied load. It is found that N-implantation results in the reduction of hardness, and SC surface layers arc more softened as the N dose increases. The apparent surface hardness is depcndcni on the applied load. At a load of 5 gf, since the indent~lti(~n depth for a dose of 1 X 10” ions/cm’ is 0.4 pm and is 2-3 times the average projected range of implanted N ions (- 150 nm), the hardness will be measured in implanted layers plus a substrate contribution. At the load of 5tl gf, the indentation breaks through implanted layers and mostly measure the bulk. In the cast of N implantation of the highest dose and the applied load of SO gf, the hardness is half of that for pristine SK’, although the indentation depth is six times as deep as the implantation thickness. Fig. 2 illustrates the SEM micrographs of surface morphology and indentation shape produced by the diamond Vickcrs-stylus. It is found that the Sic surface was damaged at the dose of 2.5 x 10” NT ions/cm’, and the appearance of damaged area changed dramatically beyond the dose of 5 x IO” ions/cm’. Below 5 x 10” ions/cm’, the cdgc of indentations is surrounded by cracks, and occasional iatcral cracks broke out into the surface. Above 5 x 10” ions,/cm”, no cracks could be seen. Thus. N implantation produced a soft layer on the SiC near-surface [5]. The brittleness

3000 ---7

0’ 0

5

Load

Fig. 1. Knoop a function

ions/cm’,

hardness

c>f applied

25

10

(gf)

fclr unimplant~d load: (0)

1

50

and implanted

unimplanted,

(Cl)

Sic as 5X 10”

(A ) 2.5 X IO” ions/cm’, ( v ) 5 X 10” ions/cm’. (+) I X 10” ions/cm’.

Fig. 3 shows the atomic concentration of Si, C, N and 0 as a function of sputtering time for SiC N-implanted to various doses. Oxygen enhanccmcnt in a near-surface layer was observed in c(~rnparis~)n with the depth distribution for pristine Sic, nc>t shown in tho figure; the thickness of the region with enhancedoxygen increased as the dose increased. The oxygen invasion during ion implantation may bc caused by knock-on doping from a surface oxide layer, as reported in the case of N,i implantation in aluminum [h] and Cu + implantati~)n Fn aluminum nitride [7]. The depth profile of N atoms implanted in Sic shows a nearly Gaussian-like distribution in the case of lower doses. The depth of the maximum N conccntration moved toward the surface as the dose increased. This phcnomcnon can he cxplaincd by sputtering during ion imp~antati~)n. The nitrogen migrati~)n into the deeper region seen for the spccimcns implanted with doses above 2.5 x 1O’7 ions/cm’. is considered to be caused by cnhanccd diffusion and/or grain boundary diffusion. It is found that atomic concentrations of silicon and carbon dccrcasc as that of nitrogen increases, and the reduction of carbon conccntrati~~n is remarkable, cspccially at the highest dose. This phcnomcnon can bc explained by the migration of defects induced by ion implantation and the “preferential migration” of atoms. In this case, it is considered that C atoms, since they are smaller than Si atoms, move easily through defects; the nligr~ltion of C atoms results in the reduction of the carbon concentrati~)n.

Fig. 4 shows the peak shift of Si2p spectra and C Is spectra with the dose of 5 x IO”’ and 1 x 10” ions/cm’ as a function of sputtering time. For a dose of 5 x IO’” ions/cm’, spectrum ( 1) in fig. 4a shows that the Si 2p spectrum on the surface is much wider than spectra U-(h) for the subsurface regions. Also spectrum (I ) is composed of Si-0 (103.5 eV) and Si-C (100.2 cV) peaks. Appearance of the Si-0 peak indicates that the surface is oxidized. The Si-C peak becomes the main component for spectra (2).-(6) for Ar ’ sputtering times of 5 to 200 min. At the intermediate sputtering times, spectrum (3) or (4) in fig. 4a, the Si2p peak shifts to the higher Si-C binding energy with the increase in nitrogen concentration. The C Is spectrum on the surface. as shown at (1) in fig. 4a, is composed of C-Si (283.5 cV1, C-C (2X5.0 cV) and C-O (286.5 eV). The C-Si peak hccomes main component as the Arf sputtering is repeated (spectra (2)-(h)). Broadening toward V. oXlDES/CERAMICS/C’ARBlnES

a high energy side of C Is spectra. as shown by spectrum (3) or (4) in fig. 4a, occurs as the nitrogen conccntration increases. This means that the C-C peak and/or C-N peak (CA 2Xh.S eV) appears.

Fig. 2. SEM

In

peak detected

the

cast

of the

corresponding

dose to

on the surface,

4b. It is compoxx~

of Si-0

of

oxide

I X 10”

formation

ions/cm‘, was

the clearly

in spcctrum (1) in fig. and Si-N (102.0 eV) peaks.

as seen

micrographs of surface morphology and indentation shape (I kgf) for unimplanted and implanted unimplanted, (h) 5 x 10 ” ions/cm’. (c) 2.5 X 10” ions/cm’. Cd) 5 X 10” ions/cm’, (e) I X 10” ions/cm’.

Sic:

(a)

ng

Sputter Fig. 3. Depth

prot’iles of Si. C. N and 0

for the SIC specimen? implanted

ions/cm’.

(c) 5 X IO” lon5/cm’.

1

-1

1 30

100

’ 0.2

‘00

13I n d

Fig. 4.

Peak shifts

planted

SiC: (a)

Spectra (I).

,‘Hii

4t 1 I I ij

c

r f: i >~1

.‘il

(h)

2fio

II

I

of SiZp spectra and Cls SX IO “’ ions/cm’,

Time(mln)

spectra in N-im-

I X IO”

ions/cm’.

(1). (3), (4). (5) and (6) correspond to sputtering times of 0. 5, 30, 80. I50 and ?OO min.

with N,’

ions: (a) SX 10”’ ions/cm’.

(h) 2.5x

IO”

(d) IX IO’” ion
The Si-N peak bccomcs the main component aftcr the Ar ’ sputtering (spectra (2)-(4) in fig. 4b). In the deeper region, spectrum (6), the Si-C peak appears. The C Is spectrum on the surface, as shown by spcctrum (I) in fig. 4b, is composed of C-C and C-O (oxygen layer and contaminated layer) peaks. After Ar ’ sputtering for intermediate times (spectra Q-(4)), the C Is spectrum is mainly composed of C-C and C-N peaks. The C Is binding cncrgy of the C-N peak was similar to that of C-O peak. Since only low conccntrations of oxygen were detected by the depth profile, as seen in fig. 3d, this peak has been dcnotcd as the C-N peak. The C-5 peak appears in the deepest regions (spectra (5) and CO), fig. 4b).

The peak shape of Si 2p spectrum becomes wide in N-implanted layers, as seen in fig. 4. Considering that the change in peak shape is caused only by the peak shift, the Si,N, peak (102.2 cV) seems to appear. The peak dcconvolution for Si 2p spectrum was pcrformcd with curve fitting analysis, as shown in fig. 5. The Si2p spectra. however. could not bc dcrivcd with a Si,N, peak and a Sic peak by deciding the binding energy for 102.0 eV (Si,N,), 100.2 cV (Sic) and FWHM (2.0 i 0.2 cV). This suggests that the implanted layer is not composed of Sic and Si,N,. Considering the appearancc of the C-C bonding and C-N bonding. the imV. OXIDES/CERAMICS/CARBIDES

near-aurfacc Si-0

mcasurcmcnts

peak.

influcnccd

It

is clear

and

that

were

Glicon

influenced

binding

statcx

by arc

by nitrogen.

4. Summary The

effects

&formation have

of N+

Implantation

behavior

been

and

studicd.

on the near-surface

chemical

The

structure

following

results

of wcrc

SiC ob-

tained. (I)

The

nitrogen

brittlcncss

of

implantation

Sic

surfaces

bcyoncl

the

decrcasccl

dose

of

5 x

for 10”

ions/cm’. (2) planted

layer

associated The

could

with

bc compozcd

concentration

the Si?p

binding

which

is

cncrgy

concentration lit

(3)

bc&cn

nitrogen

outside

C-c‘

nitrogen

middle-state

Si ,C,.N..

relationship

nitrogen

of some

Si?p shown

shifts

in

fig.

to higher

incrcascs.

of gcncral

binding

The

cncrgy

6.

In

in

the

data

value

The

binding

SK

of

and C-N cncrgy

of

Si 2p

to that of Si,N,.

deconvolution

of Si 7p peak,

Si 3p may not

bc composed

points

SiC’ and Si,N,.

wcrc

bonding

middle-states

appear

as

the

incrcascs.

as the

gcncral,

values

anomalous

trend

and

bonding

dose

The for

shifted

From the

binding

of only

implanted

from

the result state

of the

the two states

layer

may

the

of peak for

include

the

Modification

of

Si ,C,.N..

References [I]

-s1c

P. Mazalldi

and (;.W.

Inaulntor5

(Elscvicr.

[2] C‘.J. MCI Iargue.

A

Surface-Modil’icd \owsky

S13N4-

‘I‘. Hioki

1-11 C.J.

1

-I.--U

101

1

Fig.h. Relationship gen atomic 1o’T ions/cm’,

Binding

Energy

betwcrn

Si 2p binding

concentration: (a)

5 x IO”

( ) 5x IO”’ wnh/cm’.

99

100

(v

i0nsjcm2, ) I X IO’”

(0)

3x

ions/cm’.

Hofcr

Rclation
eds. C’.J. Mel Iargue.

(Kluwcr.

in

R. Ktrs-

1089) p. 2.53. In\tr.

and Mcth.

Farlou.

anti II.

C.W.

Naramoto.

White.

J.M.

Mater.

Sci.

Williams. Eng.

60

N. Elliott.

.I.

(IOX.5)173. [5] M. Naata\i, Mater.

R. Kosaowaky.

Rca. 3 (I%+))

[7] S. Ohira and nitro-

Surface-Propertics

and .I. K; w,;mote, Nucl. (;.C.

Appleton

[(i] S.G. Robrrra

(eV) rncrgy

McHargue.

U.K.

Beam

19S7).

(I 9x9) ho?.

R37/2X a

in:

(‘eramia.

and W.O.

[3] A. Itoh,

Ion

Arnold.

Amsterdam,

(IOS7)

and

162.

and T.F.

J.P. flirovnrn

and

1127. Pap,

J. Mater.

M. Iwak~. Nucl.

Instr.

Sci. 71 ( and

Meth.

IOXh)357. UlO/20