河中石兽原文Surface and Coatings Technology122(1999)219–224
www.elvier.nl/locate/surfcoat Residual stress characterization of diamond-like carbon coatings by
an X-ray diffraction method
Sam Zhang a,*,Hong Xie a,Xianting Zeng a,Peter Hing b
a Gintic Institute of Manufacturing Technology,71Nanyang Dri v e,Singapore638705,Singapore
b School of Applied Science,Nanyang Technological Uni v ersity,Nanyang A v enue,Singapore639798,Singapore
Received4February1999;accepted in revid form12June1999
Abstract
This paper prents residual stress measurements of amorphous diamond-like carbon(DLC)coatings obtained by studying the stress conditions of the substrate surface layer immediately adjacent to the coating via X-ray diffraction(XRD)with a thin film attachment.In such a t-up,the incidence angle a at which the primary beam strikes the specimen isfixed at a glancing
angle(2°in our experiments)relative to the sample surface while the detector rotates to collect the diffracted X-rays.The amorphous carbon coatings were deposited on single-crystal silicon wafers and on polycrystalline KBr substrates in an unbalanced magnetron sputtering system.The effects of substrate material and deposition parameters on the internal stress of the coatings
are discusd in detail.XRD with thinfilm attachment provides a new and more preci way to determine the residual stress in amorphous coatings.Increasing the relative nitrogenflow reduces the compressive stress level of the hydrogenated amorphous carbon coatings.Under the experimental conditions studied,higher substrate bias power and sputter power densities both incread
the compressive stress level.©1999Elvier Science S.A.All rights rerved.
Keywords:Coatings;Diamond-like carbon;Residual stress;X-ray diffraction
1.Introduction method[12].In all of the methods,stress are mea-
sured through the measurement of strain,and the strain Diamond-like carbon(DLC)coatings have a wide is measured by different‘strain gauges’.In XRD,the range of applications becau of their uful
properties‘strain gauge’is the d-spacing of a ries of planes:the including high hardness,optical transparency,low residual stress cau a change of the spacing of crystal coefficient of friction,chemical inertness and high electri-planes,reflected as the shift of the diffraction peak to cal resistivity.Residual stress are inevitably introduced higher or lower angle depending upon the nature of the in the coating during the deposition process.Control of stress(compressive or tensile).Measuring the peak shift residual stress is very important becau highly stresd or the lattice parameter change enables measurements coatings can show poor adhesion[1].In a sputtering of residual stress,as reported by Valvoda and col-process,the residual stress are generated mainly as a leagues[4],Perry and co-workers[5–8],Rickerby et al. result of the bombardment and differences between the[9]and Fischer and Oettel[10],in the analysis of TiN, thermal and elastic properties of the coating and the ZrN and other crystalline thinfilms.For amorphous substrate.Residual stress in hard coatings affect their coatings,however,becau there are no sharp diffraction adhesion strength,microhardness and wear resistance peaks,investigations of residual stress are usually [2,3].conducted by using the curvature method[2,13].
Residual stress can be measured in a number of Recently,Kondrashov and colleagues[14]ud the ways:X-ray diffraction(XRD)[4–10],acoustic-wave two-crystal method to determine the curvature(
and thus detection[11],curvature measurement by a lar profi-the residual stress)of an amorphous DLC-coated lometer[3]and the electrical resistance or capacitance
sample by comparing the substrate’s diffraction peak
with that of the reference substrate.In this paper,we
prent stress measurements of amorphous DLC coat-*Corresponding author.Tel.:+65-793-8577;fax:+65-792-2779.
E-mail address:v.sg(S.Zhang)ings via studying the surface layer stress conditions of
0257-8972/99/$–e front matter©1999Elvier Science S.A.All rights rerved.
PII:S0257-8972(99)00298-4
220S.Zhang et al./Surface and Coatings Technology 122(1999)219–224
Fig.1.Reflection geometry of XRD with thin film attachment.In measurements,the incidence angle a is fixed at a small value (up to 9°).
the substrate using XRD with a thin film attachment.form Talysurf ries).Raman spectra of the as-deposited Although u of XRD to study the stress of crystalline DLC coatings were obtained with a Rennishaw bulk materials or crystalline films is an established Ramanscope with He–Ne lar radiation of 632.8nm as technique,measurement of the residual stress of amor-the excitation source.
phous coatings with the XRD method is a new venture.The amorphous carbon coatings were deposited on single-crystal silicon wafers and on polycrystalline KBr 2.3.Residual stress determination
substrates.The e ffects of substrate material and depos-ition parameters on the internal stress of the coatings XRD spectra were taken using a Rigaku X-ray are discusd in detail.
di ffractometer (RINT 2000ries,model D /max-2200)with a thin film attachment.A Cu K a 1
X-ray source
was ud at 30kV and 40mA.The XRD di ffraction 2.Experimental geometry of the thin film attachment is shown schemati-cally in Fig.1.The XRD instrumental error is 0.001°2.1.Coating deposition
2h .The incidence angle a ,which can be t from 0.1°to 10°,was fixed at 2°during scanning.For coatings of Diamond-like carbon coatings about 1m m in thick-constant thickness,decreasing the incidence angle a ness containing hydrogen and nitrogen (a-C:H and increas the di ffracted X-ray intensity,K ,significantly a-C:N)were prepared on single-crystal silicon wafers (cf.Fig.2).
(20mm ×20mm)and on KBr polycrystalline pellets In the commonly ud Bragg–Brentano method,(B 13mm)presd from KBr powders.The depositions which operates in the h –2h scan mode,the residual
were carried out in an unbalanced magnetron sputtering system by sputtering of solid graphite targets in an argon plus hydrogen and /or nitrogen atmosphere.The ba pressure in the sputtering chamber was below
6.5×10−3Pa,and the working pressure was 1.3Pa during deposition.The substrates were placed in a rotary sample holder facing the target about 85mm above the sample.Radio-frequency (RF)bias was applied during the deposition.The bias,target current and the ratio of gas flow rate of H 2to N 2
in deposition were varied and
the influence on the residual stress assd.2.2.Coating characterization
A Jeol JEM 5410scanning electron microscope (SEM)was ud to obrve the cross-ction of the DLC coatings to determine the microstructure and growth mechanism.Coating thickness was determined by measuring the coating step produced by masking Fig.2.With decreasing incidence angle a ,the X-ray di ffraction inten-sity increas significantly [15].
using a lar stylus profilometer (Rank Taylor Hobson
221
S.Zhang et al./Surface and Coatings Technology 122(1999)219–224amorphous hump rather than sharp peaks are prent in the di ffraction spectra.Hence,direct measurement of the residual stress in the amorphous coating itlf is di fficult.However,one can always measure peak posi-tion and the d -spacing change in the substrate adjacent to the coating /substrate interface.As the a ffected depth is usually shallow in the ca of thin coatings,the traditional XRD method fails to detect the change.XRD with a thin film attachment,however,makes such detection possible since the X-ray penetration depth is very small,sometimes no more than a few hundred angstroms,depending on the incidence angle and the absorption constant of the materials,etc.For
instance,in the ca of TiN coatings [10],as the incidence angle Fig.3.Orientation of the di ffraction plane to the sample surface.
varies from 2to 10°,the X-ray penetration depth varies from 0.59to 1.72m m,as obtained through non-destruc-stress is calculated quantitatively via [16]:
tive measurements under Cu K a 1
radiation.The inci-dence angle a at which the primary beam strikes the specimen can be adjusted to change the irradiation s =−
可鲁
E
n A
d n −d 0d
海事管理B
,(1)depth.Thus,this method was employed in this study.In our experiment,the incidence angle a is fixed at 2°where E ,n ,d n and d 0
are,respectively,the Young’s
to maximize the di ffraction intensity and minimize the modulus,Poisson’s ratio,d -spacing of the di ffraction interference from the bulk.优秀家风故事
plane parallel to the surface of the coating under stress and the d -spacing of the same ries of planes in the abnce of stress.
3.Results and discussion
In XRD using the thin film attachment,however,the incidence angle is fixed at a glancing angle relative to 3.1.Cross-ctional image and XRD spectra
the sample surface while the detector rotates to collect the di ffracted X-ray.Thus for a single-crystal substrate, 3.1.1.Cross-ction image of the DLC Coatings
silicon (100)in this ca,the di ffracted planes will not Fig.4is a typical SEM cross-ctional image of the necessarily be parallel to the sample ,they amorphous carbon coating under study.As can be en will be at angle h −a .To account for this e ffect,the d -from the micrograph,the coating is about 1m m thick.spacing in Eq.(1)is replaced by l defined as:This is in good agreement with the thickness measured by the surface profilometer.Fig.5is a typical cross-l =
d
hkl cos(h −a )
,(2)
消得ctional view of the DLC coating on the KBr substrate.In this ca,‘column-like’growth is obrved even where d hkl
is the d -spacing for (hkl )planes,h and a are
though the coating is amorphous.
the di ffraction angle and the incidence angle (Fig.3).Rewriting Eq.(1)gives:s =
E n D l ,
(3)
where D l =(l −l 0)/l 0
;l reprents the distance of the
stresd hkl plane in the direction normal to the sample surface and l 0
is that for the same unstresd hkl plane.
In data treatment,since the actual thickness slightly deviated from the target of 1m m,the stress obtained from Eq.(3)is then normalized against coating thickness for a fairer comparison.
Note that stress measurement by the XRD method is achieved by measurement of the change in d -spacing characteristically revealed as changes in di ffraction
peak position.In other words,the samples under study must Fig.4.SEM cross-ctional image of a DL
C coating on a silicon wafer.be crystalline (to give a sharp di ffraction peak).Since The coating thickness shown agrees well with the thickness measured using the lar surface profilometer.
DLC coatings are basically amorphous in nature,an
222S.Zhang et al./Surface and Coatings Technology122(1999)219–224
Fig.8.XRD difraction peaks of the KBr polycrystalline substrate at
the immediate vicinity of the a-C coatings at different incidence angles
from0.2°to0°.
Fig.5.Typical cross-ctional micrograph of the DLC coating on a is an obvious peak shift towards higher2h with increas-KBr substrate.ing deposition target power density.To be sure that the
peak shift was not from the bombardment during sputter 3.1.2.XRD spectra cleaning of the silicon surface layer,XRD spectra before
The XRD spectrum of the Si(100)substrate before and after sputter cleaning were compared.Under the deposition is shown in Fig.6.The main diffraction peaks sputter cleaning conditions ud,no peak shift was are centred around2h=22°,55°and77°.obrved due to sputter cleaning.The temperature was After deposition of the coating,the diffraction below200°C during deposition.Although thermal mis-spectrum centred around the55°peak was collected match is one of the contributing factors,it should be a again under the same conditions and is shown in Fig.7.minor one becau of the low temperature involved. The55°peak(311)is chon becau of the higher Also,it was assumed that there are no significant change intensity and lack of amorphous interference from the in refractive index of the substrate before and after coating.It can be en in Fig.7that,after coating,there coating.
Thus,the peak shift was attributed to the stress
due to coating deposition.
The shift D2h shown in the plot is for the deposition
at higher target power density.This D2h reflects the
change of the d-spacing of the hkl plane at2h.This
change is ud to calculate the residual stress using
风雨无阻造句
Eq.(3).For polycrystalline KBr,no apparent change
of peak position is obrved with increasing incidence
angle from0.2°to9°,as shown in Fig.8.Since the
梦的解析作者
signal taken is from the substrate,not the coating(the
coating was amorphous),the diffraction intensity also
increas with increasing incidence angle.It was found
at different incidence angles that there is a slight decrea Fig.6.XRD spectrum of the Si(100)substrate before deposition of
of peak position:for(100)single-crystal silicon,2h a-C at a glancing angle of2°.
decreas by about0.8°while the incidence angle varies
from2°to9°.Therefore,to ensure a fair comparison,
the samples should be irradiated at the same incidence
angle before and after coating.In our experiments with
single-crystal silicon,afixed incidence angle of2°was
ud.In the ca of the KBr substrate,an incidence
angle of0.2°was ud becau the diffraction intensity
from the polycrystalline surface was strong enough for
accurate analysis at such a low incidence angle.
Fig.9compares the spectra for KBr polycrystalline
diffracted at the same incidence angle of0.2°before and
顶真的典型句子after coating.Aside from the obvious peak shift after
coating deposition,it is obrved that the intensity of Fig.7.XRD spectra from the specimen of silicon wafer coated with
DLC coating.the peaks at higher2h angles also increas after coating.
223
S.Zhang et al./Surface and Coatings Technology 122(1999)219–224
Fig.9.XRD spectra of the KBr polycrystalline substrate before and Fig.11.Variations of residual stress of the DLC thin coatings with after deposition of the a-C coating,irradiated at 0.2°incidence angle.
sputtering power at di fferent substrate bias.
From Fig.10,the compressive stress decreas by Moreover,the relative intensity of the peaks is also 200%when the relative flow rate of nitrogen gas is changed:before coating,I 3is higher than I 2
,but after
incread 100%.Reduction of stress in a-C:N coatings coating I 2became stronger than I 3
.The change in peak
as a result of nitrogen incorporation has been reported shape was also obrved for the silicon substrate after [3,17–19].Angus and Wang attributed the stress reduc-coating deposition (Fig.7).Although the mechanism is tion to over-constraining [18]of DLC.According to not clear yet,this may have resulted from the stress this model,Franceschini et al.[19]further attributed distribution and its e ffect on certain planes.
the decrea of internal stress to a reduction of the average coordination number and of the over-constrain-3.2.Influence of process parameters on residual stress ing in the a-C:N films as a result of the replacement of C–H with N–H bonds.With increasing nitrogen,the Fig.10shows the measured residual stress as a fraction of N–H bonds is incread while the unbound function of gas flow ratio.Just as stress measurement (dangling)hydrogen content is decread.The incread through measurement of curvature changes,the mea-fraction of N–H bonds contributes to the reduction of sured stress should be a good reprentation of the stress the stress according to the over-constraining model,in the coating.It can be clearly en that with the while a further decrea of the stress is caud by the relative increment of nitrogen flow,the residual stress reduced amount of unbound hydrogen.
decreas appreciably.This result agrees with that The residual stress resulting from the growth of reported by Grill and Patel [3]and Torng et al.[17].sputtered films originates primarily from the specific The influence of the substrate bias and sputtering power energy transfer into the substrate and the growing films.density on the residual stress is demonstrated in Fig.11.The applied bias power thus plays a big role.Increasing With incread sputtering power,the residual stress the bombardment leads to an increa in the compressive levels incread (becoming more negative).C
omparing stress.This is demonstrated in Fig.11which shows that,the data obtained under di fferent substrate bias,higher with increasing bias power,the compressive residual bias resulted in higher stress level (compare the solid stress is incread.In our earlier paper [20]it was and the open circles).
reported that increasing bias would result in an increa in residual stress and that,in turn,would decrea the adhesion.Fig.11provides direct stress measurement
confirmation of this.Also en from Fig.11is the e ffect of sputtering power:at the same bias power,higher sputtering power results in greater stress.The increa in target power density e ffectively increas the ion bombardment and thus leads to the development of higher compressive stress.A similar trend has also been reported in the ca of crystalline TiN coatings [6,10].In a certain sputtering power range,this may be best understood by relating the deposition ion energy (E )with the target power density (D w
)together with
bias voltage (V b
)and chamber pressure (p )through:
Fig.10.The decrea of compressive residual stress in the DLC thin coatings with increasing relative nitrogen flow.
E 3D w V b
/p 1/2.(4)
