- Email: [email protected]
Elastic properties of silicon dioxide films deposited by chemical vapour deposition from tetraethylorthosilicate
ELSEVIER
Thin Solid Films 296 (1997) 102-105
Elastic properties of silicon dioxide films deposited by chemical vapour deposition from tetraethylorthosilicate G. Carlotti a,., L. Doucet
b,c, M.
Dupeux °
a Dipartimento di Fisica, Unit& INFM, Universitg~ di Perugia, Via Pascoli, 06100 Perugia, Italy b SGS THOMSON Microelectronics, 850 rue Jean Monnet, B.P. 16, 38921 Crolles, France c Laboratoire de Thermodynamique et Physico-Chimie MetalIurgiques, ENSEEG, BP75, 38402 Saint Martin d'Heres, France
Abstract Dielectric silicate glass films have been obtained from tetraethylorthosilicate using both Iow-pressure chemical vapour deposition and plasma-enhanced chemical vapour deposition. The elastic properties of these films, which are deposited on ( 100)-Si wafers, have been studied by means of the Brillouin light scattering technique. The phase velocity of both the surface Rayleigh wave and the longitudinal wave in the film material have been measured and the two independent elastic constants cll and c~ evaluated. This permitted us to derive the values of Young's modulus and Poisson's ratio which are useful quantities for the modelling of the elastic properties of multilevel structures used in electronics. Moreover, the substrate curvature method has been exploited in the case of the PECVD film, to study the evolution of the stress during subsequent thermal cycles. The values of the thermal expansion coefficient and of the water diffusion coefficient have thus been determined. © 1997 Elsevier Science S.A. Keywords: Elastic properties; Silicon dioxide films; Chemical vapour deposition; Tetraethylorthosilicate
1. Introduction Silicon dioxide films are extensively exploited as protective, insulating and planarization layers in the technology of integrated circuits. These films are usually deposited at a temperature in the range of 400-700 °C, in a multilayer stack on a thick substrate, by means of different techniques, such as atmospheric-pressure chemical vapour deposition ( A P C V D ) , low-pressure chemical vapour deposition (LPCVD) or plasma-enhanced chemical vapour deposition (PECVD) [1]. Recently, tetraethylorthosilicate (TEOS) has become more and more popular as a precursor material, because it enables one to obtain high quality films, with planarization properties (fluence) better than those obtained using silane (Sill4) as a precursor [2,3]. Knowledge of the elastic properties of the deposited films and of their dependence on the growth conditions is important in view of an optimization of the electronic devices. As a matter of fact, large thermomechanical stresses can occur in multilevel dielectric-metal stacks, affecting the reliability of the devices [4]. Moreover, knowledge of E and ~, is necessary in order to estimate the value of the thermal expansion coefficient a * Corresponding author. Tel: +39 75 5853067; fax: +39 75 44666; email: [email protected]. 0040-6090 / 97 / $17.00 © 1997 Elsevier Science S.A. All rights reserved PIIS0040-6090(96) 09346-7
from the stress versus temperature behaviour and to distinguish between the intrinsic and thermal components of the measured stress. However, a direct and selective measurement of E and v in films with thicknesses in the micron range is out of the reach of most conventional techniques. In a previous work we studied the elastic properties of silicate glass films obtained using silane (SiI--I4) gas as a precursor [5]. In the present research we extend the investigation to films obtained from liquid TEOS by means of both PECVD and LPCVD techniques. The experimental investigation has been carried out using the Brillouin light scattering (BS) technique [6], which enabled us to determine the two independent elastic constants and thereafter the values of E and v for the different films. In addition, the residual stress has been measured at room temperature by the substrate curvature method and the evolution of the stress during thermal cycles in the PECVD film has been analyzed and its thermal expansion coefficient estimated.
2. Experimental The TEOS films here analyzed were deposited on (100)Si wafers, by PECVD at 400 °C and by LPCVD at 700 °C. The total thickness is 1 ~m for the PECVD film and 3 ~m
103
G. Carlotti et al. / T h i n Solid Films 296 (1997) 102-105
for the LPCVD film, while the refractive index and the mass density of both films are n = 1.45 and p -- 2.2 g cm - 3. The BS technique relies on detection of the small amount of light inelastically scattered by thermal phonons naturally present in the structure under investigation. Spectra were taken in air, at room temperature, using a 250 mW p-polarized light beam (single mode of the 514.5 nm line of an Ar + laser). The incident light was focused onto the surface of the specimen and the back-scattered light collected by means of a lens with f number 2 and a focal length of 50 ram. The frequency analysis was performed using a Sandercock-type, 3 + 3 pass, tandem Fabry-Perot interferometer [6,7], characterized by a finesse of about 100 and a contrast ratio higher than 5 × 10 :°. The sampling time per spectrum was typically 2h. The backscattering interaction geometry was used, so that the wavevector of the surface phonons involved into the scattering process is [7]: Qsurf = 2kisin(0), where k i is the light wavevector and 0 the angle of incidence. In the case of bulk phonons, their wavevector is: Qbum= 2nkl within n the refractive index of the medium. One can therefore directly derive the connection between the frequency shift f of Brillouin peaks and the phase velocity v of the corresponding acoustic mode, as follows: u= v=
2~f =
- -
Qs~f ki sin( O)
for surface modes,
2vf = 7rf for bulk modes Qbulk nkl
( 1)
We have also determined the stress in the dielectric films, using the Flexus apparatus based upon the substrate curvature method described in Ref. [8]. The uniform biaxial stress is deduced by use of the Stoney's formula, applied to the case of a thin film on a thick circular substrate:
°'=6(i-v,) h---~R, R]+
6=700
PECVD TEOS h=l !Jm
~[ /l
R...a
¢'3 k..
W
v
>. 4--' CO
rOD
LPCVD TEOS
~LA
E
-40
I
I
I
-20
0
20
40
Frequency (GHz) Fig. 1. Briltouin spectra relative to the two TEOS films investigated, for an angle of incidence 0 = 70 °.
since the film thickness h is much larger than the photon wavelength A (A ~ 274 nm), a well defined peak owing to bulk-like longitudinal acoustic (LA) waves is observed at about 33 GHz. This peak presents a broadened lineshape in the case of the PECVD film because of the reduced thickness of this film with respect to the LPCVD one. Measurement of the phase velocities uR and VLA,corresponding to the R and LA peaks, respectively, directly yielded the value of the elastic constants, provided that Cll=P/)LA [2], while c44 was subsequently determined on fitting the calculated value of vR [9] to the experimental value. The values of the measured phase velocities are reported in Table 1, together with the values obtained for the two elastic constants. In addition, we have reported in Table 1 the values of the Poisson's ratio u and the Young's modulus E, which are related to c:: and c ~ by the following classical expressions:
(2)
where Es = 130 GPa, Us=0.28 and hs=0.725 m m are the Young's modulus, Poisson's ratio and thickness of the sub-
strate, respectively, h is the film thickness, and Rs and Rs+ f the wafer radii of curvature before and after film deposition.
p=
2C44
--
C I1
2 ( C 4 4 - - C11)
E=c::
( C l i -- 2C44) 2 CI 1 - -
3. Results and discussion
3.1. Determination of the elastic constants The amorphous dielectric films we are concerned with are elastically isotropic media, so that there are only two independent elastic constants, namely c:t and c44. The values of these two constants have been determined by a careful analysis of Brillouin spectra. Fig. 1 shows two spectra from the two different TEOS films, for an angle of incidence 0 = 70 °. The peak corresponding to the surface Rayleigh wave (R) in the film is seen in the spectra at about 11 GHz. In addition,
C44
Table 1 shows that the elastic parameters of our films are appreciably different from those expected for fused silica. In particular, the Young's modulus is lower than in bulk silica, while the Poisson's ratio is higher. The values of the Young's modulus are also slightly lower than those measured in the past in thermally grown oxide films. As for the influence of the deposition conditions, from a comparison between the results of the PECVD and the LPCVD TEOS films we can see that the latter oxide is slightly softer than the former one. This may be owing to the use of plasma, since the continuous bombardment during film growth may result in a more compact structure [ 13].
104
G. Carlotti et al. / Thin Solid Films 296 (1997) 102-105
Table 1 Experimental values of the phase velocity of the Rayleigh and longitudinal wave, together with the obtained values of the elastic constants, the Young's modulus and the Poisson's ratio. For the sake of comparison we have reported in the last two rows some available literature data for thermally grown oxide and bulk silica Sample
vg ( m s -~)
PECVD TEOS LPCVD TEOS Thermally grown SiOa films Bulk SiO~ ~
3130 + 30 3050 4- 30 . . .
vt.A ( m s -~)
cta (GPa)
5890 + 50 5840 _+50 . . . .
.
3410
76 _+ 1.5 75 4- 1.5 . . .
.
5970
78.5
c,,4 (GPa)
E (GPa)
~
E/(1-v)
25.5 4- 0.5 24.2 4- 0.5
64 + 2 61 + 2 66 " 70 t, 73
0.25 + 0.01 0.26 4- 0.01
85 4- 4 82 4- 4
0.17
88
31.2
(GPa)
From Ref. l I0]. b From Ref. [ 11 ]. c From Ref. [ 12].
25
25 a) First cycle
b) Second cycle
0
o • ,j
~,-2~
m i•••
~D
•
•
~-25
aua a e a o o ~ "~
••
oDm=nlwI
~==
aDo
a D
O0
~m•
i O' ,••
wm
• ,~.:::)..'."" 2 /
c0_50
-50
aD=
-75
-75
0
5'0 100
1;0
2;0
250 300
3;0
400 450
Temperature (°C)
0
5;
1;0
1;0
200 2;0 300 3&O
4;0
450
Temperature (°C)
Fig. 2. Dependence of the biaxial stress of the PECVD-TEOS film on the temperature, during two subsequent cycles. Filled squares, heating; triangles, I h stabilization at 400 °C; open squares, cooling down.
3.2. Measurement of the stress
sion coefficient o~ of the film in the stabilized state, by the equation [2]:
Mechanical stress includes two different components: the intrinsic stress, connected with the material properties and growth process, and the thermal stress, owing to the different thermal expansion coefficients of the film and of the substrate. We have studied the evolution of the total residual stress of the PECVD TEOS, by the substrate curvature method, during two subsequent thermal cycles from room temperature to about 400 °C. It can be seen in Fig. 2(a) that in the first cycle the stress evolution is quite linear during both heating and cooling, but with different slopes. This difference in slopes can be attributed to a change in the intrinsic stress, caused by desorption of water or hydrogeneous species as well as by a change of the elastic and thermal constants [ 14]. We also notice that the film is almost stress-free at 400 °C, which corresponds to the deposition temperature, indicating that both thermal and intrinsic stress vanish for such temperature. The final stress is more compressive than the initial one because of the internal modifications, such as the reconfiguration of the chemical bonds (Si-OH and Si-O stretching peaks in infrared absorption spectra are modified) [3]. During the subsequent annealing (Fig. 2 ( b ) ) there is an almost complete reversibility, with no hysteresis, indicating that the film has become stable. Measurement of the stress versus temperature slope enabled us to determine the thermal expan-
do" E d--f= ( % - c~) ( 1 _ u)
(3)
where % and c~are the temperature expansion coefficients of substrate and film, respectively. We obtained the value c~=2.61 × 10 .6 °C -I, using the values o r e and ~, given in Table 1, the slope do-/dT measured during annealing, and a , = 3 . 6 × 10-6 °C - I [15]. Finally, we have followed the evolution of the stress versus time after the second annealing, with the specimen stored in white room conditions. The stress decreases continuously because of water adsorption, from the initial value of - 60 MPa to about - 62.7 MPa after 200 rain. By a best fit procedure of the experimental stress behaviour to the Fick's diffusion equation [ 16] we obtained the value of the water diffusion coefficient D = 8 × 10-13 cm a s - 1, which is in the usual range of magnitude for silicate glass films [ 16]. In conclusion, we have exploited Brillouin spectroscopy to investigate the elastic properties of both LPCVD and PECVD grown silicate glass films used in electronics. Measurement of the elastic stiffness constants of the films enabled us to determine the Poisson' s ratio u and the Young' s modulus E. The results obtained show that the elastic properties of our silicate glass films are different from those of bulk vitreous
G. Carlotti et al. / Thin Solid Films 296 (1997) 102-105
silica and thermally grown oxides. These properties also depend, to a minor extent, on the preparation technique.
References [I] G. Rojas, R. Gomarasca, L. Zanotti, A. Borghesi, A. Sassella, O.
[2] [3] [4] [5]
Ottaviani, L. Moro and P. Lazzari, J. Vac. Sci. Technol., B70 (I992) 633; S. Rojas, L. Zanotti, A. Borghesi, A. Sassella and G.U. Pignatel, Jr. Vac. Sci. Technol., B l l (1993) 2081. G. Rojas, A. Modelli, A. Borghesi and B. Pivac, J. Vac. Sci. Technol., B8 (1990) 1177. K. Ramkumar and N. Saxena, J. Electrochem. Soc., 139 (1992) 1437. P.A. Flinn, MRS Bull., XX, November 1995, p. 70. L. Doucet, G. Carlotti and G. Socino, Mater. Res. Soc. Symp. Proc., 356 (1995) 215.
105
[6] F. Nizzoli and J.R. Sandercock, in G.K. Horton and A.A. Maradudin (eds.), Dynamical Properties of Solids, voh 6, North-Holland, Amsterdam, 1990, p. 307. [7] D. Fioretto, G. Carlotti, L. Palmieri, G. Socino and L. Verdini, Phys. Rev., B47 (i993) 15St286. [8] P.A. Flinn, D.S. Gardner and W.D. Nix, IEEE Trans. Electron Dev., ED-34 (1987) 689; see also P.A. Flinn, in Mater. Res. Soc. Proc., 130 (MRS, Pittsburgh, 1989) p. 41. [9] G.W. FarnelI and E.L. Adler, in W.P. Mason and R.N. Thurston (eds.), PhysicaIAcoustics, vol. 9, Academic Press, New York, 1972, p. 35. [ 10] R.J. Jaccodine and W.A. Schlegel, J. Appl. Phys., 37 (1966) 2429. [11] R.J. Charles,./. Appl. Phys., 31 (1960) 741. [ i2] B.A. Auld, Acoustic Fields and Waves in Solids, vol. I, WiIey, New York, 1973, p. 368. [ 13] M. Staedtmueller, J. Electrochem. Soc., 139 (1992) 3669. [ 14] R.E. Jones and M.L. Basehore, Appl. Phys. Lett., 50 (1987) 725. [ 15] L. Doucet, unpublished results. [ 16] E.J. Mc Inemey and P.A. Flinn, IEEE/Proc. IRPS, 264 (1982).
Thin Solid Films 296 (1997) 102-105
Elastic properties of silicon dioxide films deposited by chemical vapour deposition from tetraethylorthosilicate G. Carlotti a,., L. Doucet
b,c, M.
Dupeux °
a Dipartimento di Fisica, Unit& INFM, Universitg~ di Perugia, Via Pascoli, 06100 Perugia, Italy b SGS THOMSON Microelectronics, 850 rue Jean Monnet, B.P. 16, 38921 Crolles, France c Laboratoire de Thermodynamique et Physico-Chimie MetalIurgiques, ENSEEG, BP75, 38402 Saint Martin d'Heres, France
Abstract Dielectric silicate glass films have been obtained from tetraethylorthosilicate using both Iow-pressure chemical vapour deposition and plasma-enhanced chemical vapour deposition. The elastic properties of these films, which are deposited on ( 100)-Si wafers, have been studied by means of the Brillouin light scattering technique. The phase velocity of both the surface Rayleigh wave and the longitudinal wave in the film material have been measured and the two independent elastic constants cll and c~ evaluated. This permitted us to derive the values of Young's modulus and Poisson's ratio which are useful quantities for the modelling of the elastic properties of multilevel structures used in electronics. Moreover, the substrate curvature method has been exploited in the case of the PECVD film, to study the evolution of the stress during subsequent thermal cycles. The values of the thermal expansion coefficient and of the water diffusion coefficient have thus been determined. © 1997 Elsevier Science S.A. Keywords: Elastic properties; Silicon dioxide films; Chemical vapour deposition; Tetraethylorthosilicate
1. Introduction Silicon dioxide films are extensively exploited as protective, insulating and planarization layers in the technology of integrated circuits. These films are usually deposited at a temperature in the range of 400-700 °C, in a multilayer stack on a thick substrate, by means of different techniques, such as atmospheric-pressure chemical vapour deposition ( A P C V D ) , low-pressure chemical vapour deposition (LPCVD) or plasma-enhanced chemical vapour deposition (PECVD) [1]. Recently, tetraethylorthosilicate (TEOS) has become more and more popular as a precursor material, because it enables one to obtain high quality films, with planarization properties (fluence) better than those obtained using silane (Sill4) as a precursor [2,3]. Knowledge of the elastic properties of the deposited films and of their dependence on the growth conditions is important in view of an optimization of the electronic devices. As a matter of fact, large thermomechanical stresses can occur in multilevel dielectric-metal stacks, affecting the reliability of the devices [4]. Moreover, knowledge of E and ~, is necessary in order to estimate the value of the thermal expansion coefficient a * Corresponding author. Tel: +39 75 5853067; fax: +39 75 44666; email: [email protected]. 0040-6090 / 97 / $17.00 © 1997 Elsevier Science S.A. All rights reserved PIIS0040-6090(96) 09346-7
from the stress versus temperature behaviour and to distinguish between the intrinsic and thermal components of the measured stress. However, a direct and selective measurement of E and v in films with thicknesses in the micron range is out of the reach of most conventional techniques. In a previous work we studied the elastic properties of silicate glass films obtained using silane (SiI--I4) gas as a precursor [5]. In the present research we extend the investigation to films obtained from liquid TEOS by means of both PECVD and LPCVD techniques. The experimental investigation has been carried out using the Brillouin light scattering (BS) technique [6], which enabled us to determine the two independent elastic constants and thereafter the values of E and v for the different films. In addition, the residual stress has been measured at room temperature by the substrate curvature method and the evolution of the stress during thermal cycles in the PECVD film has been analyzed and its thermal expansion coefficient estimated.
2. Experimental The TEOS films here analyzed were deposited on (100)Si wafers, by PECVD at 400 °C and by LPCVD at 700 °C. The total thickness is 1 ~m for the PECVD film and 3 ~m
103
G. Carlotti et al. / T h i n Solid Films 296 (1997) 102-105
for the LPCVD film, while the refractive index and the mass density of both films are n = 1.45 and p -- 2.2 g cm - 3. The BS technique relies on detection of the small amount of light inelastically scattered by thermal phonons naturally present in the structure under investigation. Spectra were taken in air, at room temperature, using a 250 mW p-polarized light beam (single mode of the 514.5 nm line of an Ar + laser). The incident light was focused onto the surface of the specimen and the back-scattered light collected by means of a lens with f number 2 and a focal length of 50 ram. The frequency analysis was performed using a Sandercock-type, 3 + 3 pass, tandem Fabry-Perot interferometer [6,7], characterized by a finesse of about 100 and a contrast ratio higher than 5 × 10 :°. The sampling time per spectrum was typically 2h. The backscattering interaction geometry was used, so that the wavevector of the surface phonons involved into the scattering process is [7]: Qsurf = 2kisin(0), where k i is the light wavevector and 0 the angle of incidence. In the case of bulk phonons, their wavevector is: Qbum= 2nkl within n the refractive index of the medium. One can therefore directly derive the connection between the frequency shift f of Brillouin peaks and the phase velocity v of the corresponding acoustic mode, as follows: u= v=
2~f =
- -
Qs~f ki sin( O)
for surface modes,
2vf = 7rf for bulk modes Qbulk nkl
( 1)
We have also determined the stress in the dielectric films, using the Flexus apparatus based upon the substrate curvature method described in Ref. [8]. The uniform biaxial stress is deduced by use of the Stoney's formula, applied to the case of a thin film on a thick circular substrate:
°'=6(i-v,) h---~R, R]+
6=700
PECVD TEOS h=l !Jm
~[ /l
R...a
¢'3 k..
W
v
>. 4--' CO
rOD
LPCVD TEOS
~LA
E
-40
I
I
I
-20
0
20
40
Frequency (GHz) Fig. 1. Briltouin spectra relative to the two TEOS films investigated, for an angle of incidence 0 = 70 °.
since the film thickness h is much larger than the photon wavelength A (A ~ 274 nm), a well defined peak owing to bulk-like longitudinal acoustic (LA) waves is observed at about 33 GHz. This peak presents a broadened lineshape in the case of the PECVD film because of the reduced thickness of this film with respect to the LPCVD one. Measurement of the phase velocities uR and VLA,corresponding to the R and LA peaks, respectively, directly yielded the value of the elastic constants, provided that Cll=P/)LA [2], while c44 was subsequently determined on fitting the calculated value of vR [9] to the experimental value. The values of the measured phase velocities are reported in Table 1, together with the values obtained for the two elastic constants. In addition, we have reported in Table 1 the values of the Poisson's ratio u and the Young's modulus E, which are related to c:: and c ~ by the following classical expressions:
(2)
where Es = 130 GPa, Us=0.28 and hs=0.725 m m are the Young's modulus, Poisson's ratio and thickness of the sub-
strate, respectively, h is the film thickness, and Rs and Rs+ f the wafer radii of curvature before and after film deposition.
p=
2C44
--
C I1
2 ( C 4 4 - - C11)
E=c::
( C l i -- 2C44) 2 CI 1 - -
3. Results and discussion
3.1. Determination of the elastic constants The amorphous dielectric films we are concerned with are elastically isotropic media, so that there are only two independent elastic constants, namely c:t and c44. The values of these two constants have been determined by a careful analysis of Brillouin spectra. Fig. 1 shows two spectra from the two different TEOS films, for an angle of incidence 0 = 70 °. The peak corresponding to the surface Rayleigh wave (R) in the film is seen in the spectra at about 11 GHz. In addition,
C44
Table 1 shows that the elastic parameters of our films are appreciably different from those expected for fused silica. In particular, the Young's modulus is lower than in bulk silica, while the Poisson's ratio is higher. The values of the Young's modulus are also slightly lower than those measured in the past in thermally grown oxide films. As for the influence of the deposition conditions, from a comparison between the results of the PECVD and the LPCVD TEOS films we can see that the latter oxide is slightly softer than the former one. This may be owing to the use of plasma, since the continuous bombardment during film growth may result in a more compact structure [ 13].
104
G. Carlotti et al. / Thin Solid Films 296 (1997) 102-105
Table 1 Experimental values of the phase velocity of the Rayleigh and longitudinal wave, together with the obtained values of the elastic constants, the Young's modulus and the Poisson's ratio. For the sake of comparison we have reported in the last two rows some available literature data for thermally grown oxide and bulk silica Sample
vg ( m s -~)
PECVD TEOS LPCVD TEOS Thermally grown SiOa films Bulk SiO~ ~
3130 + 30 3050 4- 30 . . .
vt.A ( m s -~)
cta (GPa)
5890 + 50 5840 _+50 . . . .
.
3410
76 _+ 1.5 75 4- 1.5 . . .
.
5970
78.5
c,,4 (GPa)
E (GPa)
~
E/(1-v)
25.5 4- 0.5 24.2 4- 0.5
64 + 2 61 + 2 66 " 70 t, 73
0.25 + 0.01 0.26 4- 0.01
85 4- 4 82 4- 4
0.17
88
31.2
(GPa)
From Ref. l I0]. b From Ref. [ 11 ]. c From Ref. [ 12].
25
25 a) First cycle
b) Second cycle
0
o • ,j
~,-2~
m i•••
~D
•
•
~-25
aua a e a o o ~ "~
••
oDm=nlwI
~==
aDo
a D
O0
~m•
i O' ,••
wm
• ,~.:::)..'."" 2 /
c0_50
-50
aD=
-75
-75
0
5'0 100
1;0
2;0
250 300
3;0
400 450
Temperature (°C)
0
5;
1;0
1;0
200 2;0 300 3&O
4;0
450
Temperature (°C)
Fig. 2. Dependence of the biaxial stress of the PECVD-TEOS film on the temperature, during two subsequent cycles. Filled squares, heating; triangles, I h stabilization at 400 °C; open squares, cooling down.
3.2. Measurement of the stress
sion coefficient o~ of the film in the stabilized state, by the equation [2]:
Mechanical stress includes two different components: the intrinsic stress, connected with the material properties and growth process, and the thermal stress, owing to the different thermal expansion coefficients of the film and of the substrate. We have studied the evolution of the total residual stress of the PECVD TEOS, by the substrate curvature method, during two subsequent thermal cycles from room temperature to about 400 °C. It can be seen in Fig. 2(a) that in the first cycle the stress evolution is quite linear during both heating and cooling, but with different slopes. This difference in slopes can be attributed to a change in the intrinsic stress, caused by desorption of water or hydrogeneous species as well as by a change of the elastic and thermal constants [ 14]. We also notice that the film is almost stress-free at 400 °C, which corresponds to the deposition temperature, indicating that both thermal and intrinsic stress vanish for such temperature. The final stress is more compressive than the initial one because of the internal modifications, such as the reconfiguration of the chemical bonds (Si-OH and Si-O stretching peaks in infrared absorption spectra are modified) [3]. During the subsequent annealing (Fig. 2 ( b ) ) there is an almost complete reversibility, with no hysteresis, indicating that the film has become stable. Measurement of the stress versus temperature slope enabled us to determine the thermal expan-
do" E d--f= ( % - c~) ( 1 _ u)
(3)
where % and c~are the temperature expansion coefficients of substrate and film, respectively. We obtained the value c~=2.61 × 10 .6 °C -I, using the values o r e and ~, given in Table 1, the slope do-/dT measured during annealing, and a , = 3 . 6 × 10-6 °C - I [15]. Finally, we have followed the evolution of the stress versus time after the second annealing, with the specimen stored in white room conditions. The stress decreases continuously because of water adsorption, from the initial value of - 60 MPa to about - 62.7 MPa after 200 rain. By a best fit procedure of the experimental stress behaviour to the Fick's diffusion equation [ 16] we obtained the value of the water diffusion coefficient D = 8 × 10-13 cm a s - 1, which is in the usual range of magnitude for silicate glass films [ 16]. In conclusion, we have exploited Brillouin spectroscopy to investigate the elastic properties of both LPCVD and PECVD grown silicate glass films used in electronics. Measurement of the elastic stiffness constants of the films enabled us to determine the Poisson' s ratio u and the Young' s modulus E. The results obtained show that the elastic properties of our silicate glass films are different from those of bulk vitreous
G. Carlotti et al. / Thin Solid Films 296 (1997) 102-105
silica and thermally grown oxides. These properties also depend, to a minor extent, on the preparation technique.
References [I] G. Rojas, R. Gomarasca, L. Zanotti, A. Borghesi, A. Sassella, O.
[2] [3] [4] [5]
Ottaviani, L. Moro and P. Lazzari, J. Vac. Sci. Technol., B70 (I992) 633; S. Rojas, L. Zanotti, A. Borghesi, A. Sassella and G.U. Pignatel, Jr. Vac. Sci. Technol., B l l (1993) 2081. G. Rojas, A. Modelli, A. Borghesi and B. Pivac, J. Vac. Sci. Technol., B8 (1990) 1177. K. Ramkumar and N. Saxena, J. Electrochem. Soc., 139 (1992) 1437. P.A. Flinn, MRS Bull., XX, November 1995, p. 70. L. Doucet, G. Carlotti and G. Socino, Mater. Res. Soc. Symp. Proc., 356 (1995) 215.
105
[6] F. Nizzoli and J.R. Sandercock, in G.K. Horton and A.A. Maradudin (eds.), Dynamical Properties of Solids, voh 6, North-Holland, Amsterdam, 1990, p. 307. [7] D. Fioretto, G. Carlotti, L. Palmieri, G. Socino and L. Verdini, Phys. Rev., B47 (i993) 15St286. [8] P.A. Flinn, D.S. Gardner and W.D. Nix, IEEE Trans. Electron Dev., ED-34 (1987) 689; see also P.A. Flinn, in Mater. Res. Soc. Proc., 130 (MRS, Pittsburgh, 1989) p. 41. [9] G.W. FarnelI and E.L. Adler, in W.P. Mason and R.N. Thurston (eds.), PhysicaIAcoustics, vol. 9, Academic Press, New York, 1972, p. 35. [ 10] R.J. Jaccodine and W.A. Schlegel, J. Appl. Phys., 37 (1966) 2429. [11] R.J. Charles,./. Appl. Phys., 31 (1960) 741. [ i2] B.A. Auld, Acoustic Fields and Waves in Solids, vol. I, WiIey, New York, 1973, p. 368. [ 13] M. Staedtmueller, J. Electrochem. Soc., 139 (1992) 3669. [ 14] R.E. Jones and M.L. Basehore, Appl. Phys. Lett., 50 (1987) 725. [ 15] L. Doucet, unpublished results. [ 16] E.J. Mc Inemey and P.A. Flinn, IEEE/Proc. IRPS, 264 (1982).