Journal of
AD£~ AND COMPO~S ELSEVIER
Journal of Alloys and Compounds 219 (1995) 83-87
Structure of thin films of titanium nitride V. Valvoda Faculty of Mathematics and Physics, Charles University, Ke Karlovu 5, 121 16 Prague 2, Czech Republic
Abstract T h e micro- and picostructures of T i N films deposited by magnetron sputtering are compared with their properties and deposition conditions. T h e transition from porous to compact films and the changes in their microhardness, lattice parameters and microstrains are controlled by the t e m p e r a t u r e of deposition, gas pressure and energy of ion bombardment. T h e extended crystallographic anisotropy of inhomogeneous lattice deformations is a new p h e n o m e n o n in which thin polycrystalline films differ from bulk stress-free materials. Keywords: Structure determination; Thin films; Titanium nitride
1. Introduction
Titanium nitride thin films and surface coatings are widely used in contemporary microelectronics as diffusion barriers, in the automobile and glass industries as reflecting materials and in jewellery as golden-colouted surface-finishing paints. The main field of their applications is in the machine industry as refractory materials and hard and wear-resistant coatings on machine tools and mechanical parts. The detailed structure and bonding character of their cubic compounds with the sodium-chloride-type arrangement of atoms were thoroughly investigated in Ref. [1]. The aim of this paper is to compare some of the most important structural features of polycrystalline films in the system Ti-N deposited by magnetron sputtering on steel substrates with their properties and deposition conditions.
2. Micro- and picostructures
Thornton [3] in terms of the so-called microstructural zone model (Fig. 1). X-Ray diffraction (XRD) analysis can also give information about some microstructural aspects of thin films, e.g. size of coherently diffracting blocks or preferred orientation of grains. On the other hand, XRD also gives information on phase composition, lattice parameters, residual stresses and microstrains in thin films, which are structural features with small differences of the order of a few picometres in individual films. These structural aspects of thin films can be thus regarded as their picostructure. Mainly these pico-
111~.':7.=' ."" " .':.':..:"-'- '~.",'~',,~'~,.-~::;~1 t [I .~°
1
Despite the known atomic structure of compounds in the Ti-N system, the properties of thin films made from these materials depend strongly on their real structure. The thickness of polycrystalline TiN coatings is usually a few micrometres and features such as grain size, grain shape and grain packing observed by crosssectional transmission electron microscopy (XTEM) represent the microstructure of the films [2]. Basic changes in microstructure as a function of substrate temperature and gas pressure have been described by 0925-8388/95/$09.50 © 1995 Elsevier Science S.A. All rights reserved SSDI 0925-8388(94)05012-0
I ="-'-"=,;~ " '" " ",:"::"."-~=,~fl I J I I I I ~09 • .. :=..:..g" : Zone i ::'." 08
lUllllllfl
.
Y,,;',', flliJ
'
o.6
0,~ Argon pressure
/T m 1
Fig. 1. Thornton's zone model of microstructure of thin films as a function of normalized substrate temperature (Tm is the melting point) and argon gas pressure in magnetron.
F.. Valvoda I Journal of Alloys and Compounds 219 (1995) 83-87
84
structural aspects will be summarized in the following sections.
002
110
004 ~-Fe. 211
3. Effect of nitrogen content
~-Fe
The effect of nitrogen content in Ti-N films is shown in Figs. 2--4. With increasing nitrogen content the structure of the films changes from a hexagonal solid solution of nitrogen in ~-Ti to the cubic ~ phase of TiN. The three sets of Ti-N films shown in the figures were deposited with different magnetron parameters (for more details see Ref. [4]) and all of them show a clear correlation of microhardness with residual stress,
30
40
50
60
70
80 2G
(a) ~.-Fe 110
(x-Fe
TiN 211 222 I
ot-Fe
I
AA T;N
t(~Jm) I 3
I x~ x
2
1
~~ 0
\'× ~o.o x,1
i i i
i i
I
I
I
o
I
0
_
01
~
0.2
03
HVzo(kgf/mm2)
4000
_
/ I ×,r
0
I
x
-
'/,
Ii
~ O
I
ct- Fe 2i2N ~200
I
l l %
O
×
i
, /
I
~'
II
;II
I
30 (c)
I
I
I
x 1% I x
xlsO
70
80 20
oeFe 110
~ I
t
L
c~- Fe 200 i N
T
TIN 311
T,N
I I[
I
40
50
60
70
80 20
(d) x
/ ' ,X~, -
X
x~X t
#O dO I1~O
o.O
\x ~x.'xI A"AA~
I
I
I AA
x
0 ~
-4
50
ct-Fe 211
I
B111(nm-I) (oo2) 01 I--
-2
TiN 200
50
/
30
0
40 TiN 111
1 J
I
'
I
i T~ TiN
A "A
I
/' 10
]
i 0
t
Ac~ (deg) t
(x-Fe 211
/ ,\ ,,XB , I V " | ~.-~/~"1 A ix
80 2 8
05 'IN2/AF
0 "0
I
70
110
111
.... _L__.
104
/i~C
60
cx-Fe
TiN
_~_u
50
(b)
i
II
40
30
\
\
, J, ;
;/\i o
"'K.'/
o-
Fig. 3. S e l e c t e d X-ray diffraction patterns o f T i - N coatings f r o m set C of Fig. 2 with various c o n t e n t s of nitrogen: (a) 0.5, (b) 27.8, (c) 38.0, (d) 49.3 at.%. T h e reflections o f a - F e originate f r o m the substrate. ( A c c o r d i n g to Ref. [4].)
preferred orientation of grains and line broadening, mainly caused by microstrains. In some cases the tetragonal interphase E-Ti2N was observed between transition from a-Ti solid solutions to &-TIN compounds; however, the hardest films contain no E-Ti2N phase but only the cubic &-TIN phase [5].
-6 Fig. 2. Thickness t, m i c r o h a r d n e s s HV, width Aw o f o r i e n t a t i o n distribution o f crystallites, integral line b r e a d t h B a n d stress o- as functions o f gas flow ratio N 2 / A r for three sets o f T i - N coatings. ( A c c o r d i n g to Ref. [4].)
4. Effect of layer thickness The general tendency of lattice parameters is to decrease with increasing layer thickness as illustrated
V. Valvoda I Journal of Alloys and Compounds 219 (1995) 83-87
d,,, (10"l°rn)
(oo21
I
I
I
',C
',B
',A
I
I
,, ,,
2.46
i x.~/'0
xX"
2.44
x
,
•" 2.42
61 ,
"
II
,
400
,
"2ZO
Ti N
I
.0"1
o
°{10-1°m)~
)
O0
85
I~
dll 1
.z~
_,n
427- e 3 3 1 ~
'' ."d~xzx ,(a= 4' 24~10 "~m °
2.40
"o
z~ £A
e420
~ _
9nn
2.36'
I
l
0.1
0.2
2.34 0
H 0.3
I 0.4
) 0.5
N2/Ar
Fig. 4. Interplanar spacings of (111) planes (or (002) planes in hexagonal layers). The vertical broken lines indicate the flow rate ratio at maximum microhardness for each set. The horizontal full line corresponds to the interplanar spacing of bulk stoichiometric TiN. (According to Ref. [4].)
I
O
+
( 1(3turn] 428
I +
I
I
[
4 27
/\;--:\
A~
4.26
~
~+
~
,
+~
"'--
•~+-++--+ +
I
4.24
I
I
0
1
2
'+~,,~ 111 I
cot 0 cos0
200 B
[nr,n -I]
./"
os deposited
425 ~424
13
anneated at: . A 500"C
Z I
I
I
I
I
I
2
4
6
8
10
12
I
OI
I
I
I
I
( 10"1° m )
A~
~
/ •
t[tu m)
o
oi
/
600"C
~ 7oo'c
/~/°o~°~"
O
/
o/
0.2 423 ~-
3
Fig. 6. Effect of substrate dissolution. Cohen-Wagner plots of lattice parameters of TiN: O, adherent film; + , free-standing film.
D
"i~ + / \
4.25-
/o5.,,.. D/ []
,o~"U1-×1×..×xlXX 8oo'c
/ 4.29
428
\
I
+~
0
rn',,
012
I
I
01.4
I I
0.6
I
II I I 01'8 sin 0
1"10
Fig. 7. Effect of annealing. Williamson-Hall plots of line broadening B; O is the Bragg angle [7].
427
426
200 " N k , ~ r - -
I +
4.25 0
I 1
I . 2
I 3
I 4
I 5
I 6 t(~umJ
Fig. 5. Lattice parameters calculated from various reflections as a function of layer thickness. Nitrogen contents: (a) 34.5+2.0, (b) 50.3+1.2 at.% in the whole sets. (According to Ref. [6].)
in Fig. 5. The two sets of TiN coatings shown differ in nitrogen content. All the films are under a compressive stress which gradually decreases in the topmost layers,
which can only be seen by X R D owing to absorption. Another interesting effect is the crystallographic anisotropy of the lattice deformation, with a(200)> a(111) in the substoichiometric nitride (Fig. 5(a)) and a ( 2 0 0 ) < a ( l l l ) in the stoichiometric one (Fig. 5(b)). It should be noted that all values of lattice parameter a of a reference cubic lattice were obtained in the cited experiment from spacings of lattice planes parallel to the sample surface. An expansion of the lattices in the perpendicular direction can be seen from Fig. 5 when taking into account the stress-free bulk value of
V. Valvoda /Journal of Alloys and Compounds 219 (1995) 8 3 ~ 7
86
the lattice parameter of TiN, 4.24 X 10-10 m. Almost complete relaxation of these lattice deformations and lattice defects by dissolving the substrate (Fig, 6) and by annealing the films at temperatures higher than 900 °C was found (Fig. 7).
0,429
Iorous compact
.hhh 0.427
E ~.~.425
i
. . . . . .
~hOC
5. Effect o f i o n b o m b a r d m e n t 0.423
Ion bombardment is an important factor influencing the nucleation and growth kinetics and therefore also the structure of thin films. The average energy Ep carried by bombarding ions per deposited atom in
F 0.421
0.429
,3000~orous;compact
260
460 660 Ep [ e V / a t o m ]
porous
',compact
I /~.----~ ~ , ~
hhh hh0 h00
E
H
~0.425
._~1soo,
1000
I I i e r
0.427
FE2ooo-
a60
l
F 0.423
> 1000"I-
G 0.421
260 4 porous I
460 e6o Ep [ e V / a t o m ]
860
160
1000 0,429
260 Ep [ e V / c t o m ]
porou~
360
compact
compoct
hhk
I
0.427
E c s ~7_~.42 ~-2
hh0 h00
o
~
0.423
I 1 I
'
-E
260 porous
460 e6o Ep [ e V / a t o m ]
860
1000
l.--.d
F
(1)
0.0
G
260
1go 260 Ep [ e V / a t o m ]
2go
300
Fig. 9. Lattice parameters as measured on reflections hhh, hO0 and
1.0
~,
160
hhO of TiN films of sets F-H.
compact
r.---n
0.5
H
I
0.421
-g
1.5
400
460 600 Ep [ e V / a t o m ]
800
iooo
Fig. 8. Vickers microhardness HI/, homogeneous stress o- and local strain e of TiN films of sets F-H.
magnetron sputtering is proportional to the ratio Usis/ a~, where Us is the substrate bias voltage, i~ is the substrate ion current density and ad is the deposition rate [8]. The sputtering system with the unbalanced magnetron enables us to decouple the ion current density and the deposition rate. Thus only one deposition parameter defining the energy E o can be varied, with the other parameters kept approximately constant. Results of an investigation of three sets of TiN films deposited at a pressure of 5 Pa with various parameter combinations are shown in Figs. 8 and 9. All the films show a change in microstructure from porous to compact
V. Valvoda /Journal of Alloys and Compounds 219 (1995) 83-87
at approximately Ep = 150 eV atom- 1 (this corresponds to a transition from zone I to zone T in the Thornton zone model shown in Fig. 1). Significant changes in Vickers microhardness HV, macrostress tr, microstrain e (e = ( M / d ) , where d is the lattice spacing) and lattice parameter values can also be seen at this threshold value of energy delivered by bombarding particles. The parameter Ep can thus be useful for the description of the transition from zone I to zone T, in addition to the deposition temperature and gas pressure. The crystallographic anisotropy of lattice parameters calculated from lattice spacings in the three principal directions of the cubic lattice (Fig. 9) shows similar features as already observed in films with different nitrogen content (see Fig. 5). The general tendency is a greater normal expansion of (hhh)-oriented grains in compact films and of (h00)-oriented grains in porous films. This picostructural inhomogeneity, together with similar effects observed in line broadening, is caused by a combination of elastic anisotropy, oriented lattice defects such as stacking faults and dislocation loops [9] and entrapped nitrogen or gas atoms located in interstitial positions [10]. However, an unambiguous explanation of these phenomena does not exist till now.
6. Discussion and conclusions
TiN coatings belong nowadays to the most thoroughly studied polycrystalline films owing to their successful industrial applications. Structural analyses by transmission electron microscopy and X-ray diffraction have revealed empirical rules controlling the development of their micro- and picostructures as a function of deposition conditions, which in the case of magnetron sputtering are mainly given by the temperature of deposition, gas pressure and energy of ion bombardment
87
during the deposition process. The properties of the films, especially their microhardness, are closely connected to their chemical composition, microstructure, thickness and subtle picostructural modifications with respect to the stress-free polycrystalline bulk samples. The existence of one principal direction of deposition of the atoms (approximately the normal direction to the sample surface), thermodynamically unstable deposition conditions, permanent ion bombardment of growing films and generally large residual stresses conserved in the films lead to the creation of an inhomogeneous lattice deformation which is dependent on the crystallographic orientation of grains with respect to the sample surface. Such effects in bulk materials are observed to a much smaller extent and their studies in TiN films have enriched our knowledge of the behaviour of grains in polycrystalline materials prepared at and exposed to extreme conditions by the adhesion to the substrate.
References [1] A. Dunand, H.D. Flack and K. Yvon, Phys. Rev. B, 23 (1985) 2299. [2] J.E. Sundgren, Thin Solid Films, 128 (1985) 21. [3] J.A. Thornton, Ann. Rev. Mater. Sci., 7 (1977) 239. [4] V. Valvoda, R. Ku~.el, R. (~ern3) and J. Musil, Thin Solid Films, 156 (1988) 53-63. [5] V. Poulek, J. Musil, V. Valvoda and R. (VSern~,,J. Phys. D: Appl. Phys., 21 (1988) 1657-1658. [6] V. Valvoda, R. (~ern~,, R. Ku~el, J. Musil and V. Poulek, Thin Solid Films, 158 (1988) 225-232. [7] V. Valvoda, R. t~ern~,, R. Ku~el, L. Dobifi~ov,5, J. Musil, V. Poulek and J. Vysko~il, Thin Solid Films, 170 (1989) 201. [8] V. Poulek, J. Musil, R. (~ern# and R. Ku~el, Thin Solid Films, 170 (1989) L55. [9] R. Kuiel, R. ~ern#, V. Valvoda, M. Blomberg and M. Merisalo, Thin Solid Films, 247 (1994) 64. [10] D. Rafaja, R. Ku~.el and V. Valvoda, Proc. 3rd Fur. Powder Diffraction Conf., Vienna, September 1993, in press.