Surface & Coatings Technology 191 (2005) 17 – 24

www.elsevier.com/locate/surfcoat

Effect of nitrogen flow rate on structure and properties of nanocrystalline

TiN thin films produced by unbalanced magnetron sputtering
Jia-Hong Huang*, Kiang-Wee Lau, Ge-Ping Yu
Department of Engineering and System Science, National Tsing Hua University, Hsinchu 300, Taiwan
Received 19 September 2003; accepted in revised form 8 July 2004
Available online 1 September 2004

Abstract

Nanocrystalline TiN thin films were deposited on (100) silicon wafers using an unbalanced magnetron (UBM) sputtering system. The
structure and properties of the TiN thin films were studied under single-variable experimental condition by varying nitrogen flow rate ranging
from 0.25 to 1.75 sccm. The grain size of the films was determined by X-ray diffraction (XRD), and the size was less than 7 nm. The images
of transmission electron microscopy (TEM), scanning electron microscopy (SEM) and atomic force microscopy (AFM) also showed a
nanometer-size grain structure of the TiN thin films. The N/Ti ratio increased (N/Ti=0.4–1.1) while the deposition rate decreased with
increasing nitrogen flow rate. The preferred orientation changed from (111) to (200), then back to (111) as the nitrogen flow rate increased.
There was no significant variation in the film hardness with nitrogen flow rate or preferred orientation. It is noted that the film hardness was
high, ranging from 23.4 to 27.6 GPa, even for the film thickness down to 140 nm. The resistivity decreased with increasing packing factor.
Ti/TiN nano-composite was found in the films with N/Ti ratio below 0.6.
D 2004 Elsevier B.V All right reserved.

Keywords: TiN; Nanocrystalline; Unbalanced magnetron sputtering; Nitrogen flow rate

1. Introduction To improve the properties of TiN thin film, many studies

have been performed to understand the relationships
Owing to their extreme hardness, high thermal and between the processing parameters, the film structure and
chemical stability, and low electrical resistivity, titanium properties [1–4]. It is known from our previous research [5–
nitride (TiN) has been widely applied as a coating material, 7] that the mechanical properties of TiN film are strongly
ranging from hard and protective coatings on mechanical affected by its preferred orientation, residual stress, packing
tools, decorative coatings, to the diffusion barrier in factor, and grain size. In addition, the resistivity of TiN thin
microelectronic industry. In the past decade, due to the films is significantly correlated to the packing factor [7].
development of micro-electromechanical systems (MEMS) However, most of the relationships were obtained from the
and precision machinery, thin coatings, less than 0.5 Am film thickness larger than 0.5 Am or even larger than 1 Am.
thick, with high wear resistance, high hardness, and special Few studies were carried out on the properties of thin
surface properties are in great demand. Due to the well- polycrystalline TiN films with a thickness close to 200 nm.
established application history, TiN thin films are good It is known that most of the PVD coatings possess a
candidates to meet the requirements. columnar structure; with an increasing columnar width as
the film thickness increases. Consequently, as the film
becomes thinner, the columnar width or grain size is getting
smaller. Our previous study [7] also showed that as the film
* Corresponding author. Tel.: +886 35715131x4274; fax: +886
thickness is close to 200 nm, the packing factor, preferred
35720724. orientation and hardness of TiN film are distinctly different
E-mail address: jhhuang@ess.nthu.edu.tw (J.-H. Huang). from those of the films thicker than 500 nm. Therefore it is a
0257-8972/$ - see front matter D 2004 Elsevier B.V All right reserved.
doi:10.1016/j.surfcoat.2004.07.066
18 J.-H. Huang et al. / Surface & Coatings Technology 191 (2005) 17–24

challenge to produce a TiN film with a film thickness about bias of 1000 V for 10 min to remove the surface oxide
200 nm and with comparable properties as that thicker than layer. The argon pressure was fixed at 0.8 Pa (6
103 Torr)
500 nm. To reach this goal, further improvement of the and the current density was 0.15 mA/cm2 during pre-
deposition process should be carried out, and better under- sputtering. The target-to-substrate distance was 20 cm. After
standing of the structure and properties of TiN thin film is pre-sputtering, high purity working gas and reactive gas
necessary. were introduced, using mass flow controllers to regulate
DC reactive unbalanced magnetron (UBM) sputtering both gas flows. Argon (99.9995%) gas flow rate was fixed
system is one of the popular coating systems to deposit high at 15 sccm with a maximum flow rate of 50 sccm. Nitrogen
quality films of pure metals, metal nitrides, metal oxides, gas (99.9995%) flow rate was varied from 0.25 to 1.75 sccm
and metal carbides [8]. Therefore, UBM sputtering system with a maximum gas flow rate controlled at 5 sccm. The
was chosen in this study to deposit thin TiN films with mass flow controllers were precisely calibrated and the
thickness close to 200 nm. Using Taguchi experiment accuracy of the gas flow rates was within 2%. The total gas
design method, we have found the optimum conditions for pressure was fixed at 0.133 Pa (1
103 Torr). The DC
the deposition of TiN thin films [9] by UBM sputtering. power supply of the Ti target was operated at 0.9 A using
Based on the conditions, single-variable experiment by constant-current mode; in other words, the target current
varying nitrogen partial pressure was performed in this density was 2.8 mA/cm2. The substrate ion current density,
study to further narrow down the window of process simultaneously monitored during the coating process, was
parameters. The purpose of this study was to investigate higher than 3 mA/cm2. During deposition, a negative
the effect of nitrogen flow rates on the structure, compo- substrate bias voltage of 50 V was applied and the
sition and mechanical properties of nanocrystalline TiN thin deposition temperature was maintained at 400 8C, moni-
film. tored using a thermocouple near the substrate. The
deposition duration was 40 min for all samples. The coating
conditions were adopted from our previous Taguchi
2. Experimental procedures experimental study [9]. The deposition conditions for the
seven samples are listed in Table 1.
The deposition of TiN films was carried out using a DC The crystal structure and grain size of the TiN films were
unbalanced magnetron (UBM) sputtering (STS-400) sys- determined by X-ray diffraction (XRD). Cu Ka line at
tem. The deposition chamber was connected with a 0.15405 nm was used as the source for diffraction. The XRD
pumping system consisting of a turbomolecular pump, with analysis was also utilized for the evaluation of preferred
a pumping speed of 500 l/min, and a rotary pump as orientation. The strength of the preferred orientation is
forepump. A rectangular titanium target (99.9%) with defined by a texture coefficient=I(hkl)/[I(111)+I(200)],
dimensions 248
130
10 mm3 was used. The substrate where I is the integrated intensity of the corresponding peak
material was Si (100) wafer with dimensions of 45
45
0.7 and (hkl) represents (111) or (200). The full width at half
mm3. Prior to the coating process, the Si specimens were maximum (FWHM) and the position of (111) or (200) peak
undergone ultrasonic cleaning progressively in acetone and were used to calculate the grain size by Scherrer equation
ethanol and then dried in a vacuum dryer. [10]. The microstructure of the deposited TiN film was
Before deposition, the substrates were gradually pre- observed using both field-emission-gun scanning electron
heated to a temperature of 300 8C. Meanwhile, the chamber microscopy (FEG-SEM) operated at 5 kV, and cross-
was evacuated to 8
104 Pa (6
106 Torr) to avoid sectional transmission electron microscopy (TEM) operated
contamination during deposition process. Prior to deposi- at 200 kV. The samples for SEM observations were prepared
tion, the substrate was pre-sputtered by argon discharge at a by cleaving the TiN deposited Si specimens. Since TiN is an

Table 1
Summary of the experimental results
Nitrogen Thickness Substrate (111) Deposition Grain Lattice N/Ti Hardness Packing Resistivity Residual
flow rate (nm) ion current Texture rate size parameter ratio (GPa) factor (AV-cm) stress
(sccm) density coefficient (nm/min) (nm) (nm) (GPa)
(mA/cm2)
0.25 253 3.4 1a 6.3 4a 0.4201 0.4 23.4 0.62 216 –
0.50 244 3.6 0.63a 6.1 7a 0.4223 0.5 26.8 0.68 142 10.2
0.75 224 3.2 0.22 5.6 7 0.4259 0.9 25.8 0.79 98 8.8
1.00 200 3.4 0.26 5.0 7 0.4240 1.0 27.6 0.80 88 9.2
1.25 192 3.4 0.51 4.8 7 0.4255 1.0 26.4 0.80 90 9.4
1.50 184 3.1 0.78 4.6 7 0.4217 1.1 24.8 0.68 119 8.9
1.75 140 3.8 0.76 3.5 6 0.4240 1.1 24.6 0.78 123 8.1
Deposition condition: DC power 340 W, PAr+P N2=0.133 Pa, bias 50 V, temperature 400 8C, deposition time 40 min.
a
Peak was de-convoluted.
 J.-H. Huang et al. / Surface & Coatings Technology 191 (2005) 17–24 19

electric conductive material, no further pretreatment was specimens are less than 10 nm. The AFM image also
done on the specimens. The cross-section of the specimen supports the result. It can be seen that the surface grain
was examined. The TiN film thickness was also measured size is in nanometer scale, as shown in Fig. 1, where at
from the SEM observations. least hundreds of grains within the scanning area of
The N/Ti ratios and packing factors were obtained by 100
100 nm2. The cross-sectional SEM images, as shown
Rutherford backscattering spectrometry (RBS) using RUMP in Fig. 2, reveal that there is no columnar structure in the
simulation code. The N/Ti ratio and packing factor were TiN films.
obtained from the RBS results. The packing factor is defined It is known that high mobility of adatoms may lead to the
as formation of coarse grains in the TiN film [12], due to the
fact that the bulk and surface diffusion rates are enhanced
SAD
1015 and adatoms can migrate for a larger distance. Since the
Packing factor ¼
DN0 t=A deposition was performed at 400 8C, coarse grains were
expected to form at such a high deposition temperature.
where SAD is the simulated atomic packing density (#/cm2) However, the high deposition temperature did not facilitate
from RBS, D is the density of TiN (g/cm3), N 0 is Avogadro the formation of coarse-grain films; instead, the processing
number 6.02
1023, t is the film thickness (cm), and A is the produced nanocrystalline TiN films. Evidently, other factors
molecular weight of TiN (g/mol). The surface compositions may dominate the process.
were determined using X-ray photoelectron spectroscopy The energy delivery and momentum transfer to the
(XPS). The photoelectron spectra were excited by soft X-ray growing film by the incident ions and atoms should be
using Mg anode at 1253.6 V. To eliminate the surface oxide, considered for understanding the formation of nano-
the specimen was sputtered with 3 keV Ar+ ions for 2.5 min. crystalline films. To estimate the incident ion and atom
The spectrum was first subtracted its background using energies, mean free path of the particles in the deposition
Shirley’s method and then fitted with Gaussian function chamber needs to be first taken into account. The mean
using least-square method. free path of particle can be estimated by the following
The residual stresses of the TiN films were determined equations [13],
using the modified XRD sin2w method [11]. The Cu Ka
1 P
line at 0.15418 nm was used as the source for diffraction. X- k¼ and n ¼
ray was incident at a grazing angle of 28 to increase the np
d 2 kT
diffraction volume of the thin film specimen. For the TiN where k is the mean free path, d is the collision diameter
films, (220) peak was used for the residual stress measure- of argon (d=1.76 2), n is number of particle per m3, P is
ment because it provided sufficient intensity for precise the working pressure ( P=0.133 Pa), k is the Boltzmann
determination of the peak position. constant (k=1.38
1023 J/K) and T is the chamber
To avoid the substrate effect, hardness of the TiN film temperature (T=623 K). The calculation shows that the
was measured using nanoindentation with a Berkovich mean free path of the sputtered particles is close to 66
diamond indenter head. The testing load was carefully cm, while the substrate to target distance in our system
selected to meet the requirement that the contact depth was 20 cm. Therefore, the mean free path is larger than
should be about one-tenth of film thickness. To evaluate
the creep effect, the load was first increased in 20 s to the
desired load and held for 5 s; then, unloaded in 20 s. A
standard quartz sample was used to calibrate the nano-
indenter and the accuracy of the hardness was within 10%.
For each sample, 10 measurements were performed and
the average value was reported. The precision of each
value was within 10%. The resistivity of the TiN film was
measured using a four-point-probe. The surface roughness
of the TiN film was determined by an atomic force
microscope (AFM).

3. Results and discussion

3.1. Nano-grain size

The grain size was estimated from the FWHM of (111)

or (200) XRD peaks using Scherrer equation [10]. The Fig. 1. The AFM image for the TiN specimen deposited at 1.25 sccm
results, as listed in Table 1, show that the grain sizes for all nitrogen flow rate at 100
100 nm2 scan size.
20 J.-H. Huang et al. / Surface & Coatings Technology 191 (2005) 17–24

Fig. 2. (a, b) The cross-sectional SEM pictures for the TiN films deposited at 1.5 sccm (a) and 0.25 sccm (b).

the distance between target and substrate. This implies volume (assume 1
1014
thickness nm3), and R is the N/
that the effect of nitrogen flow rate on the mean free (N+Ti) ratio from RBS. The results show that N ranges from
path can be neglected. In other words, the particles 7.7
1017 to 1.1
1018. It is obvious that nitrogen atoms in
directly impinge on the growing film without colliding the TiN films do not significantly vary with increasing
with other particles. nitrogen flow rate. This indicates that the variation N/Ti
The energy delivered to growing film per unit volume S E ratio is due to the change of titanium content in the TiN thin
is defined as S E=(V bV p)
(I b/D) [14], where V b is the film. Therefore, the supply of Ti atoms is progressively
negative substrate bias, V p is the plasma potential (V pbbV b, decreased as the nitrogen flow rate increases. This is
V p can be neglected), I b is the substrate ion current density supported by the evidence that the deposition rate decreases
and D is the deposition rate. The energy is estimated to be with increasing nitrogen flow rate (Table 1). Since the
ranged from 16 to 34 MJ/cm3, which is much higher than deposition rate is closely related to the number of Ti atoms
the previously reported value about 5 MJ/cm3 [14]. The arrives at the substrate surface, lower deposition rate
high energy particles not only deliver energy but also represents that fewer Ti atoms reach the substrate and
damage or re-sputter the film lattice by momentum transfer. thereby increasing the N/Ti ratio.
Thus, many heterogeneous nucleation sites can be created
during the deposition process. Since too many grains are 3.3. Preferred orientation
nucleated simultaneously on the substrate, the small grains
are difficult to grow without encountering neighboring Fig. 4 shows the (111) and (200) texture coefficients with
grains. The increase of nucleation sites and the decrease of respect to nitrogen flow rate. The texture coefficient is
grain-growth territory apparently facilitate the formation of defined as I(hkl)/(I(111)+I(200)), where (hkl) represents
nanocrystalline TiN film. (111) or (200), and I(111)/(I(111)+I(200))=0.42 corresponds
to the random orientation of TiN from the data in the Joint
3.2. Compositions Committee of Powder Diffraction Standard (JCPDS) [15]. It

Fig. 3 shows that the N/Ti ratio of the TiN films with
respect to the nitrogen flow rate. It can be seen that the N/Ti
ratio steeply increases from 0.4 to 1.0 when the nitrogen
flow rate increases from 0.25 to 1.00 sccm. Beyond this
range, the N/Ti ratio slowly increases to 1.1 as the nitrogen
flow rate increases from 1.00 to 1.75 sccm. Obviously, N/Ti
ratios are sensitive to the nitrogen flow rates.
It is interesting to understand that the increase of N/Ti
ratio is due to increasing N atoms or decreasing Ti atoms.
The number of nitrogen atoms, N, in the TiN film can be
calculated as below
n
N¼ VR
a30

where n is the number of atoms in a TiN unit cell, a 0 is the

lattice parameter of TiN determined from GIXRD, V is the Fig. 3. The N/Ti ratio of the TiN films with respect to nitrogen flow rate.
 J.-H. Huang et al. / Surface & Coatings Technology 191 (2005) 17–24 21

but varied between (111) and (200). Therefore, the change

of preferred orientation is not purely from the reduction of
strain energy.
The ion bombardment at the substrate may increase the
residual stress and strain energy and thereby switching the
preferred orientation of the TiN films. In this study, the
substrate bias voltage is relatively low and kept constant at
50 V; as a result, the ion input energy or channeling effect
is not the major factors for the change of orientation
between (111) and (200). All the other deposition conditions
except nitrogen flow rate were kept constant, and the
experimental results show that the relations of substrate ion
current density and residual stress of the TiN films to the
nitrogen flow rates were not distinct. Apparently, other
factor should be considered.
Fig. 4. The texture coefficient with respect to nitrogen flow rate.
The change of preferred orientation may be related to the
stoichiometry of the TiN film. For the stoichiometric film,
can be seen that as the nitrogen flow rate is below 0.50 the (200) plane has strong bonding states of (pd)r and (pd)p ,
sccm, (111) is the preferred orientation, and (200) becomes while (111) plane only consists of Ti–Ti (dd)r bonding state
prevailed as the nitrogen flow rate increases from 0.75 to [17], as shown in Fig. 5. It is clearly that the fraction of Ti–
1.00 sccm, and (111) dominates again as the nitrogen flow N bonds are lower when N/Tib1 or N/TiN1 than that of the
rate is higher than 1.50 sccm. stoichiometric film. There are fewer Ti–N bonds in the non-
Pellege et al. [16] proposed that the change of the stoichiometric TiN films and hence (111) orientation is
preferred orientation of TiN film from (200) to (111) can be preferred. The number of Ti–N bonds increases when the N/
attributed to the competition between surface energy and Ti ratio is close to 1, which is in favor of the existence of
strain energy with increasing film thickness, and (111) (200) preferred orientation.
preferred orientation is the result of relieving strain energy.
As listed in Table 1, the residual stresses in the TiN films 3.4. Hardness
with different thickness, ranging from 8.1 to 10.1 GPa
(Table 1), vary in a small extent, about 10% from the As shown in Fig. 6, the variation of the TiN film
average value. Consequently, the effect of strain energy on hardness is not apparent for different deposition conditions.
the preferred orientation due to grown-in stress, which is The average film hardness is about 26 GPa, which is
associated with the film thickness, is not supposed to be comparable to most of the previous studies on TiN. Notice
significant. Previous study [3] also reported that high that the hardness maintains at a quite high value of 24.6
momentum particles may induce high stress on the growing GPa, even when the film thickness is down to 140 nm.
film at low total pressure, and the presence of (111) Previous studies [7,18,19] have reported that (111) texture
preferred orientation is to reduce the strain energy. Although coefficient is an important factor to the film hardness,
the residual stress of the TiN films is very high for all because h111i is the hardest orientation in TiN. This is
nitrogen flow rates, the preferred orientation is not at (111) derived from the fact that at h111i direction the Schmid

Fig. 5. Schematic diagram of orbital bonds of stoichiometric TiN (200) plane.

22 J.-H. Huang et al. / Surface & Coatings Technology 191 (2005) 17–24

Fig. 7. The resistivity versus packing factor of the TiN films.

Fig. 6. The hardness of the TiN films with respect to nitrogen flow rate.

[7]. Since low packing factor represents more lattice defects

factor is zero for all slip systems. However, the experimental
inside the film, it is reasonable that resistivity increases with
results show that the variation of film hardness is not agreed
decreasing packing factor. It is also evident that the
with that of (111) texture coefficient.
resistivity is related to the stoichiometry of the TiN film.
The relatively high hardness of the thin TiN film may be
From Table 1, the lowest resistivity associated with
from the nanocrystalline structure. Recent study reported
stoichiometric TiN film. Again, the non-stoichiometric
that for the nanocrystalline material, the deformation
TiN films contain more lattice defects and consequently
mechanisms might be different from normal bulk material.
increasing resistivity.
Instead of dislocation slip via slip systems, the nanocrystal-
The effect of grain boundary on resistivity of the TiN
line material may be deformed by grain boundary rotation,
films needs to be noted. It is expected that nanocrystal-
or grain boundary sliding [20]. If this is the case, the Schmid
line materials have higher electrical resistivity due to
factor, or geometrical strengthening, may be no longer the
extremely high ratio of grain boundary area. In fact, the
controlling factor for the hardness because dislocation slip
results show that the resistivity increases about five times,
does not provide plastic deformation. This could explain the
compared with TiN single crystal film (18 AV cm [21]),
result that film hardness is not dependent on the (111)
due to the increase of grain boundary. This suggests that
preferred orientation.
the effect of grain boundary on the resistivity of the TiN
Table 1 lists the residual stress of the TiN films, ranging
films is significant. On the other hand, with similar grain
from 8.1 to 10.2 GPa. Compared with the residual stress
size, about 7 nm, the resistivity of the films may change
of the TiN films produced by HCD-IP in our previous study,
ranging from 2 to 6 GPa [6], the residual stress in the
present study is much higher. Since the origin of the grown-
in residual stress is due to ion bombard induced lattice
defects, the high residual stress implies high defect density.
The high defect density may be also contributed to the high
film hardness.

3.5. Electrical resistivity

The electrical conduction of TiN is from metallic origin

[21]. From previous studies [7,22], it is known that lattice
defects may increase the resistivity of TiN film due to
scattering of electrons by the lattice defects. In this study,
the resistivity of the films is ranged from 88 to 216 AV cm,
which is much higher than the TiN single-crystal film, 18
AV cm [21]. This indicates that the nanocrystalline TiN
films contain much higher defect density than the single-
crystal film.
Fig. 7 shows that resistivity decreases with increasing
packing factor. This is consistent with our previous result Fig. 8. XPS spectra of Ti 2p at different nitrogen flow rates.
 J.-H. Huang et al. / Surface & Coatings Technology 191 (2005) 17–24 23

nitrogen flow rate. It can be seen that nano-TiN grains are

embedded in the Ti matrix, which confirms that the film is a
nano-composite of Ti and TiN.
The resistivity is relatively high for both specimens
deposited at 0.25 and 0.50 sccm nitrogen flow rates. This
may be due to higher lattice defect density in the composite
structure, and thereby increasing the scattering of conduc-
tion electrons.

4. Conclusions

(1) Nanocrystalline TiN thin films were successfully

produced using unbalanced magnetron (UBM) sputter-
ing system at different nitrogen flow rates. The grain
size is less than 7 nm determined by X-ray diffraction
using Scherrer equation.
(2) From the RBS results, the N/Ti ratio, ranging from
0.4 to 1.1, increases with increasing nitrogen flow
rate. The increase of N/Ti ratio is mainly from the
Fig. 9. TEM picture for the TiN specimen deposited at 0.25 sccm nitrogen
decrease of Ti in the film, which is related to the
flow rate. Notice that the nano-TiN grains are embedded in the Ti matrix.
decreasing deposition rate with increasing nitrogen
flow rate.
(3) The hardness of nanocrystalline TiN films is quite
three times due to the change of packing factor. high, ranging from 23.4 to 27.6 GPa, even for the film
However, since the variation of packing factor, ranging thickness down to 140 nm. The high hardness may be
from 0.62 to 0.8, is not very distinct, it is inequitable to due to high defect densities and small grain size.
compare the relative significance with that of grain (4) The hardness and the residual stress of the nanocrystal-
boundary. line TiN film are not related to the nitrogen flow rate.
(5) The resistivity decreases with increasing packing
3.6. Nano-composite of Ti–TiN factor.
(6) Ti/TiN nano-composite is found in the films with N/Ti
As shown in Fig. 3, the N/Ti ratios of the TiN films range ratio below 0.6.
from 0.4 to 1.1 with increasing nitrogen flow rate. The N/Ti
ratios corresponding to 0.25 sccm and 0.50 sccm nitrogen
flow rates are only 0.4 and 0.5, respectively. According to Acknowledgments
the Ti–N binary phase diagram, the N/Ti composition range
of TiN is 0.6–1.2 [23]. It is obvious that these films contain This research was funded by the National Science
both TiN and Ti. As shown in the Ti 2p XPS spectra (Fig. Council of the Republic of China under the contract NSC
8), the binding energies for the samples prepared at 0.25 and 91-2216-E-007-039. Residual stress measurements were
0.50 sccm nitrogen flow rates are 454.2 and 454.4 eV, carried out in the Department of Physics and Materials
respectively. The spectra shift from TiN at Ti 2p3/2=454.6 Science, City University of Hong Kong. TEM analysis was
eV to pure Ti metal at Ti 2p3/2=454.0 eV [24], which performed in the Center for Microanalysis of Materials,
supports that Ti exists in both samples. Frederick Seitz Materials Research Laboratory, University
The deconvoluted XRD patterns for the specimens of Illinois at Urbana-Champaign. Prof. J.M. Zuo of the
prepared at 0.25 and 0.50 sccm nitrogen flow rates also Department of Materials Science and Engineering, Uni-
reveal that the films contain both Ti and TiN. Estimating the versity of Illinois was appreciated for the assistance of
ratio of TiN phase from the XRD patterns, it is found that TEM analysis. The authors also acknowledge Dr. Wen-Jun
the fractions of TiN are 0.35 and 0.54 for 0.25 and 0.50 Chou for assisting in the operation of UBM sputtering
sccm nitrogen flow rates, respectively. After subtracting the machine.
fraction of Ti, the N/Ti ratios of the TiN in the films are 1.1
and 0.9 for 0.25 and 0.50 sccm nitrogen flow rates,
respectively. Since the grain size measured from XRD is References
less than 10 nm, these two films can be categorized as nano-
composite films of Ti and TiN. Fig. 9 shows the TEM [1] P. Patsalas, C. Charitidis, S. Logothetidis, Surf. Coat. Technol. 125
microstructure for the specimen deposited at 0.25 sccm (2000) 335.
24 J.-H. Huang et al. / Surface & Coatings Technology 191 (2005) 17–24

[2] J.E. Greene, J.-E. Sundgren, L. Hultman, I. Petrov, D.B. Bergstrom, [15] JCPDS, File No. 38.1420.
Appl. Phys. Lett. 67 (1995) 2928. [16] J. Pelleg, L.Z. Zevin, S. Lungo, Thin Solid Films 197 (1991) 117.
[3] M.I. Jones, J.R. McColl, M.D. Grant, Surf. Coat. Technol. 132 (2000) [17] A. Nelkel, in: R. Freer (Ed.), The Physics and Chemistry of Carbides;
143. Nitrides and Borides, Kluwer Academic Publishers, Netherlands,
[4] K. Laing, J. Hampshire, D. Teer, G. Chester, Surf. Coat. Technol. 112 1990, p. 485.
(1999) 177. [18] C.T. Chen, Y.C. Song, G.-P. Yu, J.-H. Huang, J. Mater. Eng. Perform.
[5] J.H. Huang, Y.P. Tsai, G.P. Yu, Thin Solid Films 355 (1999) 440. 7 (3) (1998) 324.
[6] W.J. Chou, G.P. Yu, J.H. Huang, Surf. Coat. Technol. 149 (2002) 7. [19] J.-E. Sundgren, Thin Solid Films 128 (1985) 21.
[7] W.J. Chou, G.P. Yu, J.H. Huang, Surf. Coat. Technol. 140 (2001) 206. [20] I.A. Ovid’ko, Science 295 (2002) 2386.
[8] P.J. Kelly, R.D. Arnell, Vacuum 56 (2000) 159. [21] J.E. Sundgren, B.O. Johansson, A. Rockett, S.A. Barnett, J.E. Greene,
[9] W.J. Chou, C.H. Sun, G.P. Yu, J.H. Huang, Mater. Chem. Phys. 82 Physics and Chemistry of Protective Coatings, in: W.D. Sproul, J.E.
(2003) 228. Greene, J.A. Thornton (Eds.), American Institute of Physics,
[10] P. Scherrer, Gftt. Nachr. 2 (1918) 98. Massachusetts, 1986, p. 95.
[11] C.-H. Ma, J.-H. Huang, Haydn Chen, Thin Solid Films 418 (2002) 73. [22] J.-H. Huang, C.-H. Lin, Haydn Chen, Mater. Chem. Phys. 59 (1999)
[12] K.W. Whang, Y.W. Seo, J. Vac. Sci. Technol., A 11 (1993) 1496. 49.
[13] C. Kittel, H. Kroemer, Thermal Physics, 2nd ed., W.H. Freeman and [23] Joanne L. Murray, Phase Diagram of Binary Titanium Alloys, ASM
Company, 1980, p. 395. International, Ohio, 1987, p. 176.
[14] J. Musil, V. Poulet, V. Valvoda, R. Kuzel, H.A. Jehn, M.E. [24] Kozo Tanaka, Hiroyuki Yanashima, Tadaaki Yako, Kunimasa Kamio,
Baumgartner, Surf. Coat. Technol. 60 (1993) 484. Kazumi Sugai, Shunji Kishida, Appl. Surf. Sci. 171 (2001) 71.