Raman_spectroscopy_and_mechanical_properties_of_multilayer_tetrahedral_amorphous_carbon_films1

Raman spectroscopy and mechanical properties of multilayer tetrahedral amorphous carbon ?lms

Sai Wang,Jiaqi Zhu ?,Jiazhi Wang,Xunbo Yin,Xiao Han

Center for Composition Materials,Harbin Institute of Technology,Yikuang Street 2,Nangang District,Harbin,150080,China

a b s t r a c t

a r t i c l e i n f o Available online 13January 2011Keywords:

Amorphous materials Thin ?lms Multilayer

Vapor deposition

Mechanical properties

In order to take the tetrahedral amorphous carbon (ta-C)?lms as the high acoustic impedence layer in a Bragg re ?ector isolating acoustic wave from the substrate in solidly mounted resonator,the multilayer ?lms consisting of sp2-rich layers and sp3-rich layers were deposited from a ?ltered cathodic vacuum arc by adjusting the substrate bias.The microstructure of the ?lms was evaluated using a visible Raman spectroscopy.The stress was calculated according to the changed curvature of the coated and bare substrate.The hardness,modulus and scratching were measured using a nanoindenter.It has been shown that the multilayer structure maintaining high tetrahedral content,high hardness and high elastic modulus is still characterized with lower intrinsic stress and better adhesion.

?2011Elsevier B.V.All rights reserved.

1.Introduction

The expanding demand of the modern electrical communication systems has accelerated the progress of new type acoustic wave devices used in higher frequency ranges,especially solidly mounted ?lm bulk acoustic resonator (SMR-FBAR)[1].This kind of device consist of a piezoelectric thin ?lm deposited on several pairs of alternately laminated low and high acoustic impedence quarter-wave layers on a silicon substrate.Heavy metal,for example W,Mo,et al.,is typically applied as the high acoustic impedence layer in a Bragg re ?ector isolating acoustic wave from the substrate [2,3].Therefore,the device would unavoidablely suffer the parasitic capacitance problem.If tetrahedral amorphous carbon (ta-C)?lms with high hardness,high elastic and high electrical resistivity deposited from a ?ltered cathodic vacuum arc (FCVA)were used as the high acoustic impedence layer,the device Q-factor would be distinctly improved [4].According to the central frequency of SMR-FBAR,the thickness of the high acoustic impedence layer should be usually achieved between 500nm and 1000nm.However,the high intrinsic compres-sive stress of the ta-C ?lms limits the effective adhesive thickness with the sp3-rich bonding content.Many groups have taken efforts to depress the stress level of the ta-C ?lms by means of the different processes,such as incorporating doped elements [5],post-thermal annealing [6],and multilayer structure designing [7].Wherein,the dopant atoms could act as the defects or impurities and annealing could lead to the tendency of graphitization [8].The multilayer structure composed of the alternate “soft ”(sp2-rich)layer and “hard ”

(sp3-rich)layer can signi ?cantly decrease the compressive stress [9].In this paper,the multilayer structure with the different tetrahedral component will be investigated to maintain the high elastic modulus and the excellent adhesion for the application of SMR-FBAR.

2.Experimental 2.1.Sample preparation

The ?lms were prepared at room temperature using the ?ltered cathodic vacuum arc technology,whose details have been described elsewhere [10].Before deposition,the substrates were precleaned with acetone in an ultrasonic bath for ?fteen minutes and then were etched for ?ve minutes by a Kaufman ion gun.The argon ?ow rate was ?xed at 8sccm by a mass ?ow controller and the impinging energy of Ar +was accelerated to about 1100eV.The base pressure in the chamber was below 2.0×10?6Torr and the pressure rose during deposition due to outgassing of the carbon target.During deposition,the arc current was 60A and the ?ltering magnetic ?eld intensity was 40mT.To ensure uniformity,the substrate disc holder rotated at a rate of 33r/min during deposition and etching.The impinging energy of the species was controlled by changing the substrate bias.The other substrates were bare p-type (100)polished c-Si wafers.The sequential multilayers were deposited at a substrates bias alternating between ?1000(layer A )and ?80V (layer B ).The deposition started by forming a thin layer A (~50nm)on substrates surface and followed by a thick layer B (~260nm)deposition .This process was repeated twice to prepare two bilayers (ABAB )with a total thickness of ~600nm.The single-layer ?lms with the same thickness were prepared at the bias of ?1000V (?lm A )and ?80V (?lm B ),respectively.

Thin Solid Films 519(2011)4906–4909

?Corresponding author.Tel./fax:+8645186417970.E-mail address:zhujq@https://www.sodocs.net/doc/066518020.html, (J.

Zhu).

0040-6090/$–see front matter ?2011Elsevier B.V.All rights reserved.doi:

10.1016/j.tsf.2011.01.051

Contents lists available at ScienceDirect

Thin Solid Films

j o u r n a l h o m e p a g e :w w w.e l s ev i e r.c o m /l o c a t e /ts f

2.2.Film characterization

Hardness and Young's modulus were examined by MTS XP-type nano-indenter with Berkvovich three-sided pyramid diamond indenter, which stepped continuously into the substrate from the surface.The measurement precision of load and displacement was75nN and0.1nm respectively.Each sample was measured four times in different locations at room temperature and relative humidity30%.The scratch resistance of the?lms was determined by a nanoscratch test.The scratch with a linear loading from0to180mN was repeated three times in a length of500μm at a speed of5μm/s.The?lm thickness and the radius of curvature of the coated substrates were measured using a commercial Taylor Hobson surface pro?lometer.

The microstructure was characterized in turn by visible Raman spectroscopy.A Labram HR-800Raman spectrometer with a458nm line had been calibrated with single crystal silicon before measure-ments.In order to avoid damage to the microstructure of the?lms and clearly acquire the characteristic signals,the input power was kept at25mW,the measurement step was1.4cm?1in the range from800 to2000cm?1,and the sampling duration was set to100s.

3.Results and discussion

3.1.Raman evaluation

Raman spectroscopy is a standard nondestructive tool to obtain the detailed bonding structure of amorphous carbons[11].Though typical visible Raman spectra are sensitive to sp2bonds of amorphous carbon?lms,the bonding characters of sp3could be indicated[12]. Fig.1shows the Raman spectra between1000cm?1and1800cm?1 for the?lm A,?lm B and the multilayer?lm.The spectra display a remarkably good signal-to-noise ratio.Every spectrum consists of an asymmetric broad peak in the range of1100to1700cm?1,consisting of two major features:the G peak and the D peak.The peak of approximately1560cm?1refers to G peak derived from the bond stretching of all pairs of sp2bonds in both rings and chains.The D peak shoulder emerges around1360cm?1which is attributed to the symmetry breathing mode of benzene rings[13].The spectra were ?tted with a Breit–Wigner–Fano(BWF)line shape for the G-peak and a Lorentzian line for the D-peak.The ratio of peak heights I D/I G can be used to estimate the sp3bonding content of amorphous carbon?lms [13].A small I D/I G ratio of0.09–0.2will correspond to a high sp3content (up to70%)of the?lm B and the multilayer?lm.The D peak shoulder of the spectra of Film A was clearly shown in Fig.1,which implied that Film A contained more sp2hybridization bonding.Therefore,the substrate bias obviously in?uences the microstructure of ta-C?lms and the thick multilayer can retain a high average content of tetrahedral bonds.

3.2.Stress measurement

According to Stoney's equationσ=E s

61?νs

eT

t2s

t f

1

R

?1

R0

,the compressive stressσof the?lms was calculated by the comparison of the curvatures of the bare silicon substrate(R0)and the substrate coated with the?lms(R)[14].Wherein,E s,t s,νs are respectively Young's modulus(180GPa),thickness(0.50mm)and Poisson ratio(0.26)of the silicon substrate and t f is the thickness of the?lms.

The calculated stress values of Film A,Film B and the multilayer ?lm are plotted in Fig.2.The sp2-rich?lm has a small compressive stress of0.8GPa.The relaxing stress is attributed to the thermal spike induced by the higher energetic bombard in the local area[15].The formation of ta-C?lm is generally described by the subplantation model,which considers that incident energetic ions penetrating the surface create a compressive stress above the Berman–Simon line and cause the local preferential bonding of sp3hybridization sites[15,16]. Therefore,Film B is characterized with the most tetrahedral content consorted with the largest compressive stress.It has been shown that the alternate multilayer possesses a clear less compressive stress of about3.5GPa.

The decrease of stress in the multilayer could be ascribed to the additional dense interface areas between the sp3-rich and sp2-rich layers,which could induce the bending of the substrate and depress the interface energy and strain energy.This bending might engender macroscopic tensile stress to compensate the intrinsic compressive stress[17].Another reasonable explanation is that the sp2-rich layer being susceptible to a high amount of plastic deformation could act as a suffering layer to relieve the intrinsic stress of the sp3-rich layer.

3.3.Hardness and modulus

Nanoindentation is the most widely used technique for measuring the mechanical properties of thin?lms.Fig.3shows a plot of the hardness(H)and Young's modulus(E)as a function of the indenting depth for the multilayer ta-C?lm.The hardness reaches a stable value (43GPa)at a depth of approximately100nm.As the indenting depth increases,the hardness decreases obviously because of the soft substrate.The elastic modulus of the multilayer?lm is about400GPa. For hard?lm on soft substrate,the indenter deforms not only the?lm but also the substrate during dynamic loading[18].The measured hardness and elastic modulus are the composite for?lm and substrate. The intrinsic H and E of the?lm,being derived by?nite element analysis,should be more than the measured values.The in?uence of

Fig.1.Raman spectra between1000cm?1and1800cm?1for?lm A,?lm B and

multilayer ta-C?lm.The arrows denote the positions of the D and G peaks.The spectra are?tted by BWF line shape.Fig.2.The calculated compressive stress by Stoney's formula for?lm A,?lm B and multilayer ta-C?lm with a same thickness of~600nm.

4907

S.Wang et al./Thin Solid Films519(2011)4906–4909

substrate can be observed at the contact depth exceeding 10%–20%of the ?lm thickness.

Fig.4shows the columns of hardness and Young's modulus for Film A ,Film B and the multilayer ?lm.The sp2-rich Film A has a relatively low hardness (24GPa)and Young's modulus (235GPa).Nevertheless the hardness of the multilayer ?lm is comparable to that of sp 3-rich Film B ,even the elastic modulus with a small enhance-ment.As the enhancements of hardness and elastic modulus in multilayers have reported [19],the multilayers could be hardened by coherency strains,which are caused as the neighboring sublayers stretch to accommodate the small difference in their lattice constants.The dense interfaces between the sublayers could also contribute to the increased hardness.The ratio of H 3/E 2can reveal the plastic resistance of amorphous carbon ?lms under contact events [20].The H 3/E 2ratio of 0.5–0.6means a high elastic behaviour of Film B and the multilayer ?lm,whereas Film A has a relatively weak plastic resistance.A great amount of plastic deformation occurring in the soft sublayer A seems to be the essential for the stress relaxation of the multilayer ?lm.

3.4.Scratching

The nanoscratch test can re ?ect the adhesion of the ?lms deposited on the substrate.Fig.5shows the surface pro ?le and the coef ?cient of friction (C f )as a function of the scratch distance for the alternate multilayer ta-C ?lm on the Si substrate.The surface pro ?le consists of pre-scan,scratching and post-scan curves.Respectively,

the pre-scan and the post-scan under a very low load (0.1mN)are used to pro ?le the unscratched and damaged surface.The negative depth corresponds to the scratch stylus being pushed into the surface,and the positive depth indicates the outward blistering of the surface or the accumulation of chips [21].

As seen in Fig.5,the pro ?le curves are divided into three regimes by two dotted lines.Initially,the stress induced by a low normal load (b 63mN)stays small,and the deformation is only elastic.The coef ?cient of friction is kept at a stable value of 0.08.With the normal load ramping,the plastic deformation is revealed by a slightly negative depth of the post-scan curve.The coef ?cient of friction (C f )begins to increase with a low linear slope.The cracking and delamination of the ?lm is implied by a sudden increase of C f ,and the post-scan curve ?uctuates within a narrow range.The corresponding normal load is named critical load L c1(128.7±0.6mN)which induces stress exceeding the cohesive strength of the ?lm [22].Another abrupt change of C f and the scratching curve presents for the critical load L c2(140.8±0.7mN)which means the onset of ?lm collapse from the substrate.In this case,the scratch stylus has deeply indented into the substrate and the induced stress exceeds the interfacial adhesive strength of ta-C/Si interface.After L c2,the chip-piles induce the drastic ?uctuations of the post-scan curve.

Nanoscratch testing can determine the scratch resistance (L c1)which can characterize the cohesion strength of ?lms,as well as the critical load L c2which is related to the adhesion strength between ?lm and substrate [22].The critical loads L c1and L c2of the ?lm A ,?lm B and the multilayer ?lm are plotted in Fig.6.The soft Film A with a low L c1has a relatively bad scratch resistance.However,a good adhesion of the ?lm on Si substrate can be implied by the high L c2values.The higher critical loads of the multilayer ?lm indicate the greater scratch resistance and adhesion strength than the single-layer ?lms.The reasonable explanation could be due to the presence of sublayer interfaces which store a part of total elastic energy during scratching.Accordingly,the elastic energy stored at the interface between ?lm and substrate can be reduced correspondingly and the external contact stress between ?lm and substrate induced by the same loading is weakened [23].

4.Conclusions

The alternate multilayer ta-C ?lm consisting of the sequential sp2-rich and sp3-rich sublayers can be deposited from a ?ltered cathodic vacuum arc to cater for the high acoustic impedence layer in the Bragg re ?ector of SMR-FBAR.The multilayer structure maintaining

high

Fig.3.The hardness and Young's modulus of multilayer ta-C ?lm on Si substrate as a function of indenting

depth.

Fig.4.Hardness and Young's modulus of ?lm A,?lm B and multilayer ta-C ?

lm.

Fig.5.Surface pro ?le and coef ?cient of friction (C f )as a function of scratch distance during scratching on multilayer ?lm.The dotted lines correspond to the positions of critical load L c1and L c2.

4908S.Wang et al./Thin Solid Films 519(2011)4906–4909

tetrahedral content,high hardness and high elastic modulus is still characterized with lower intrinsic stress and better adhesion.Acknowledgment

This work was supported by the National Natural Science Foundation of China (Grant no.51072039).In addition,the authors thank the Specialized Research Fund for the Doctoral Program of Higher Education in China (Grant no.200802131006)and the Youth Fund of Natural Science Foundation in Heilongjiang Province (Grant no.QC2009C56).

References

[1]R.Ruby,IEEE Ultrason.Symp.(2007)1029.

[2]G.G.Fattinger,IEEE Ultrason.Symp.(2008)762.[3]J.Kaitila,IEEE Ultrason.Symp.(2007)120.

[4]G.Liu,J.Q.Zhu,S.Wang,X.X.Lu,Y.P.Liu,S.Y.Huo,X.W.Yuan,J.Inorg.Mater.25(2010)1233.

[5] E.Dekempeneer,K.Van Acker,K.Vercammen,J.Meneve,D.Neerinck,S.Eu ?nger,W.Pappaert,M.Sercu,J.Smeets,Surf.Coat.Technol.669(2001)142.

[6]T.A.Friedmann,J.P.Sullivan,J.A.Knapp,D.R.Tallant,D.M.Follstaedt,D.L.Medlin,P.B.Mirkarimi,Appl.Phys.Lett.71(1997)3820.

[7] C.T.Ha Peter,D.R.Mckenzie,M.M.M.Bilek,E.D.Doyle,D.G.McCulloch,P.K.Chu,Surf.Coat.Technol.200(2006)6405.

[8]W.Zhang,A.Tanaka,K.Wazumi,Y.Koga,B.S.Xu,Diamond Relat.Mater.13(2004)2166.

[9]S.Logothetidis,M.Gioti,C.Charitides,P.Patsalas,Vacuum 53(1999)61.

[10]J.Q.Zhu,S.Wang,C.Z.Jiang,Y.Dang,S.Y.Du,Appl.Phys.A 97(1)(2009)195.[11]L.G.Cancado,A.Jorio,M.A.Pimenta,Phys.Rev.B 76(2007)0643034.

[12] B.K.Tay,X.Shi,H.S.Tan,H.S.Yang,Z.Sun,Surf.Coat.Technol.105(1998)155.[13] A.C.Ferrari,J.Robertson,Phys.Rev.B 61(2000)14095.

[14] B.K.Gupta,B.Bhushan,Thin Solid Films 270(1–2)(1995)391.

[15] D.R.Mckenzie,D.Muller,B.A.Pailthorpe,Phys.Rev.Lett.67(1991)773.[16]J.Robertson,Diamond Relat.Mater.2(1993)984.

[17]S.Logothetidis,M.Gioti,C.Charitides,Y.Panayiotatos,M.Handrea,W.Kautek,Diamond Relat.Mater.9(2000)756.

[18] C.Charitidis,S.Logothetidis,P.Douka,Diamond Relat.Mater.8(1999)558.

[19]S.Logothetidis,S.Kassavetis,C.Charitides,Y.Panayiotatos,https://www.sodocs.net/doc/066518020.html,skarakis,Carbon 42(2004)133.

[20] A.Leyland,A.Matthews,Wear 246(2000)1.

[21]L.Y.Huang,J.W.Zhao,K.W.Xu,J.Lu,Diamond Relat.Mater.11(2002)1454.[22]J.P.Quinn,P.L.emoine,P.Maguire,J.A.Mclaughlin,Diamond Relat.Mater.13(2004)1385.

[23]

J.Qi,https://www.sodocs.net/doc/066518020.html,i,C.S.Lee,I.Bello,S.T.Lee,J.B.Luo,S.Z.Wen,Diamond Relat.Mater.10(2001)

833.

Fig.6.Measured critical load L c1and L c2of Film A,Film B and multilayer ta-C ?lm based on nanoscratch test.

4909

S.Wang et al./Thin Solid Films 519(2011)4906–4909