Journal of ELECTRONIC MATERIALS, Vol. 38, No.

9, 2009 Regular Issue Paper

DOI: 10.1007/s11664-009-0846-8
Ó 2009 TMS

Silicon Nitride Films Deposited by RF Sputtering

for Microstructure Fabrication in MEMS

VIVEKANAND BHATT1 and SUDHIR CHANDRA1,2

1.—Centre for Applied Research in Electronics, Indian Institute of Technology Delhi, Hauz-Khas,
New Delhi 110016, India. 2.—e-mail: schandra@care.iitd.ernet.in

In the present work, we report silicon nitride films deposited by a radio-

frequency (RF) sputtering process at relatively low temperatures (<260°C) for
microelectromechanical system (MEMS) applications. The films were pre-
pared by RF diode sputtering using a 3-inch-diameter Si3N4 target in an argon
ambient at 5 mTorr to 20 mTorr pressure and an RF power of 100 W to 300 W.
The influence of the film deposition parameters, such as RF power and sput-
tering pressure, on deposition rate, Si-N bonding, surface roughness, etch
rate, and stress in the films was investigated. The films were deposited on
single/double-side polished silicon wafers and transparent fused-quartz sub-
strates. To explore the RF-sputtered silicon nitride film as a structural
material in MEMS, microcantilever beams of silicon nitride were fabricated by
bulk, surface, and surface-bulk micromachining technology. An RF-sputtered
phosphosilicate glass film was used as a sacrificial layer with RF-sputtered
silicon nitride. Other applications of sputtered silicon nitride films, such as in
the local oxidation of silicon (LOCOS) process, were also investigated.

Key words: RF sputtering, silicon nitride, structural material, surface

micromachining, MEMS

INTRODUCTION chemical vapor deposition (LPCVD),10,11 plas-

Silicon nitride films are extensively used in inte-
grated circuit fabrication processes for various
applications, such as as oxidation masks in the local
oxidation of silicon (LOCOS) process, gate dielec-
trics, diffusion barriers, isolation material, and final
passivation layers.1–3 All these applications are due
to silicon nitride’s remarkable properties such as
high thermal stability, chemical inertness, extreme
hardness, and good dielectric properties. Silicon
nitride thin films also exhibit favorable optical
properties and have been widely used in optoelec-
tronic devices and micro-optoelectromechanical
systems (MOEMS).4,5 More recently, the application
of silicon nitride films has been extended to struc-
tural materials in microelectromechanical system
(MEMS) devices.6–9 Silicon nitride films can be
prepared by a variety of methods such as low-pressure
(Received September 15, 2008; accepted May 12, 2009;
published online June 3, 2009)
ma-enhanced CVD (PECVD),12,13 pulsed laser
ablation,14 reactive evaporation,15 and sputter-
ing.16,17 Conventionally, high-quality stoichiometric
silicon nitride films are obtained by the LPCVD
process at temperatures in excess of 700°C. To
realize MEMS on the same silicon chip, it is a
requirement to integrate the control electronic cir-
cuits, usually based on complementary metal–
oxide–semiconductor (CMOS) technology, with the
sensor/actuator/mirror fabrication processing. For
post-CMOS processing, the film deposition must be
at relatively low temperatures to have minimum
effects on the circuit already fabricated. Integrating
micromechanical structures with on-chip circuitry
can also improve the performance and functionality
of devices. PECVD and the RF-sputtering technique
are two prime methods for silicon nitride deposition
at relatively low temperatures (<400°C). RF sput-
tering seems to be a promising technique because of
several intrinsic advantages over the PECVD pro-
cess, such as elimination of toxic gases, flexibility of

1979
1980 Bhatt and Chandra

sequential deposition, low cost, and a simpler
deposition system. Furthermore, the lower deposi-
tion temperature of RF sputtering is beneficial
when fabricating devices that monolithically inte-
grate silicon nitride with temperature-sensitive
materials such as mercury cadmium telluride and
organic or biomaterials. Many researchers have
reported the deposition of silicon nitride by the
RF-sputtering process.16–20 Silicon nitride depos-
ited by LPCVD and PECVD is widely used as a
structural material in the fabrication of different
MEMS devices, such as static membranes, tunable
inductors, and tunable MEMS Fabry-Pérot optical
filters.6–9,21,22 However, sputtered silicon nitride
has not yet been explored as a structural layer in
MEMS. This is the main motivation for the present
work.
In this paper, we report low-temperature
(<260°C) RF-sputtering deposition of silicon nitride
film for structural layers in MEMS. The influence of
sputtering parameters such as RF power, sputter-
ing pressure on deposition rate, etch rate, surface
roughness, and residual stress in the film has been
investigated. Silicon nitride films were deposited on
single/double-side polished silicon and transparent
fused-quartz substrates by RF sputtering of a silicon
nitride (Si3N4) target in an argon ambient. The
optical and microstructural properties of the films
were evaluated by ultraviolet–visible (UV–Vis)
spectroscopy, Fourier-transform infrared (FTIR)
spectroscopy, scanning electronic microscopy
(SEM), and atomic force microscopy (AFM). Stress
control in the deposited films, especially when these
are employed as a structural layer in MEMS, is very
important because too much intrinsic tensile stress
causes the bridge/membrane to fracture, while
compressive stress results in upward or downward
buckling. The deposition parameters for obtaining
films with minimum residual stress, lower surface
roughness, and lower etch rate were optimized. To
demonstration RF-sputtered silicon nitride films
as a structural layer in MEMS, silicon nitride
microcantilever beams were fabricated using bulk,
surface, and surface-bulk micromachining tech-
niques. For microelectronic and MEMS application,
the use of RF-sputtered silicon nitride film as a
mask against oxidation in the LOCOS process has
been explored.
FILM DEPOSITION
AND CHARACTERIZATION
Film Deposition
The silicon nitride films were deposited in an RF
(13.56 MHz) diode sputtering unit (Alcatel, QM-311).
A diffusion-pump-based vacuum system with
liquid-nitrogen trap was used. A high-purity silicon
nitride (Si3N4) target with 75 mm diameter and
5 mm thickness was fixed to the water-cooled target
holder with conducting epoxy. The sputtering was
done in argon ambient in the ‘‘sputter-up’’ configu-
ration. The sputtering pressure was in the range of
5 mTorr to 20 mTorr and the RF power was varied
from 100 W to 300 W. The distance between the
target and the substrate was kept constant at
45 mm. Both-side optically polished transparent
fused-quartz discs of 1 inch diameter and single- or
double-side polished (100)-oriented 2-inch-diameter
silicon wafers were used as the substrates. Prior to
deposition, the substrates were cleaned in hot (90°C
to 100°C) H2SO4:H2O2 (1:1) solution for 15 min fol-
lowed by deionized (DI) water rinse. After this step,
the silicon substrates were immersed in 5% HF for
30 s to remove the native oxide formed in the hot
H2SO4:H2O2 solution. The wafers were rinsed in
running DI water and spin-dried. After loading the
samples, the chamber was evacuated to a base
pressure of 2 9 10
6 Torr and then throttled. Before
starting the deposition, the target was sputter-
cleaned for 10 min. All the depositions were carried
out without any external substrate heating. The
deposition parameters for the preparation of various
samples on both types of substrates are summarized
in Table I.

Table I. Deposition Rate, Refractive Index, and Etch Rates of Films Prepared Under Different Process
Parameters
Etch Rate
Sputtering Deposition Rate Refractive Etch Rate in 40 wt.%
Sample RF Pressure at the Center Index in BHF Aqueous KOH
Name Power (W) (mTorr) of the Wafer (Å/min) (n) (Å/min) at 70°C (Å/min)

5SN100 100 5 30 1.84 180 10

5SN200 200 5 50 1.98 65 5
5SN300 300 5 90 2.04 28 2
10SN100 100 10 40 1.88 190 12
10SN200 200 10 90 1.99 70 6
10SN300 300 10 150 2.02 30 1
20SN100 100 20 50 1.89 340 18
20SN200 200 20 120 1.91 180 14
20SN300 300 20 190 1.96 120 10
Silicon Nitride Films Deposited by RF Sputtering for Microstructure Fabrication in MEMS
Deposition Rate
The deposition rate is a very important parameter
for controlling the required film thickness at a given
target-to-substrate spacing, sputtering pressure,
and RF power, especially for precise multilayer
coating. The deposition rate was obtained by mea-
suring the film thickness using a thin-film analyzer
(Filmetrics F20, USA) and deposition time. The
same system can also used for measuring the
refractive index n(k) and extinction coefficient k(k)
of the films.
Figure 1 shows the deposition rate of films as a
function of sputtering power at 5 mTorr, 10 mTorr,
and 20 mTorr sputtering pressure. The maximum
deposition rate is obtained at the centre of the
2-inch silicon wafer. Near the periphery of the
wafer, the deposition rate falls by around 5% for all
samples. The deposition rate exhibits strong
dependence on the RF power and increases with
increasing sputtering power, which is consistent
with published results.
Substrate Temperature Measurements
No substrate heating was provided during film
deposition. However, the substrate temperature
rises during the deposition process due to bom-
bardment of neutral species and electrons.23 The
silicon wafer surface temperature was measured by
fixing a thermocouple onto the front surface of the
wafer.
Figure 2 shows the change in substrate temper-
ature with time during sputtering when no exter-
nal heat was provided. The temperature rises
gradually and stabilizes after about 20 min from
the start of sputtering. The maximum temperature
was measured as about 60°C for 100 W and 260°C
for 300 W RF power at 10 mTorr sputtering pres-
sure. From this figure, it is clear that the wafer
surface temperature is different for different
sputtering powers.
Fig. 1. Deposition rate versus RF power at different pressures.
1981
Fig. 2. Change in substrate temperature with sputtering time for
different RF powers at 10 mTorr pressure.
Chemical Bonds in the Films
FTIR spectral data for transmission of infrared
light through the silicon nitride films were used to
investigate the chemical bonding in the deposited
films and also to obtain information about the
influence of the deposition parameters on the Si-N
bonds present in the sample. The infrared spectra
for the films deposited on both-side polished silicon
wafers were recorded using a Perkin Elmer FTIR
spectrometer with 2 cm
1 resolution in the wave-
number range 400 cm
1 to 4000 cm
1. A double-
side polished bare silicon wafer was used as a
reference. For these measurements, 5000 Å thick-
ness of silicon nitride film was used.
Figure 3 shows the IR spectra of the silicon
nitride films deposited at 10 mTorr pressure with
varying sputtering power of 100 W, 200 W, and
300 W. Samples deposited at 200 W and 300 W RF
10 mTorr, 100 W
Si-N bonds
10 mTorr, 200 W
Transmittance (arb. unit)
10 mTorr, 300 W
Si-OH bonds
Si-N bond shifted
towards Si-O bond
4000 3500 3000 2500 2000 1500 1000 500
Wavenumber (cm-1)
Fig. 3. Infrared transmission spectra of silicon nitride films deposited
at 10 mTorr pressure at 100 W, 200 W, and 300 W RF power.
1982
power exhibit a strong absorption band centered
around 840 cm
1 to 860 cm
1, which corresponds to
the stretching vibration mode of the Si-N bond.13
In this configuration, silicon atoms are nearly tetra-
hedrally surrounded by four nitrogen atoms. The
large value of the full-width at half-maximum of
this band is consistent with previously reported
work,13 and is attributed to the amorphous nature
of the silicon nitride films. The additional small
absorption band observed at 480 cm
1 corre-
sponds to Si–N breathing mode.13 We also observed
that samples deposited at 100 W RF power showed
a larger shift toward higher frequency.
Optical Transmittance
The optical transmission and reflection spectra
were obtained for the films deposited on quartz
substrates in the wavelength range of 200 nm to
2000 nm using a Perkin-Elmer Lamda 900 UV–Vis
spectrometer. For these measurements, a 500-Å-
thick silicon nitride film was deposited on 1-inch-
diameter quartz substrates.
Transmittance spectra for films prepared under
different deposition conditions (Table I) were
obtained in the 200 nm to 2000 nm wavelength
range. Figure 4 shows the transmittance spectra of
the samples deposited at 10 mTorr pressure with
varying RF power (100 W, 200 W, and 300 W). The
samples exhibit absorption edges at different
wavelengths in the ultraviolet region, which indi-
cates varying absorption characteristics with depo-
sition conditions. These spectra show that the
silicon nitride films exhibit good optical transpar-
ency in the visible and near-infrared regions.
Stress Measurement
The residual stress in the films deposited on
2-inch single-sided polished (100) silicon wafers was
Fig. 4. Transmittance spectra of silicon nitride films on quartz sub-
strates deposited at 10 mTorr pressure at 100 W, 200 W, and 300 W
RF powers.
Bhatt and Chandra
determined from the change in the radius of
curvature of the wafer, before and after film depo-
sition, using a 500TC temperature-controlled film
stress measurement system (FSM Frontier Semi-
conductor). The substrate curvature method relies
on the Stoney formula relating the average stress of
the film to the substrate curvature under the
assumption that the film is much thinner than the
underlying substrate.24 The stress in a thin film of
thickness df can be expressed as
 
Es d2s 1 1
rf ¼
; (1)
6ð1
ms Þdf Rf R0
where Es, ms, and ds are the Young’s modulus,
Poisson ratio, and thickness of the substrate,
respectively, and R0 and Rf are the radii of substrate
curvature before and after film deposition.
Figure 5 shows the dependence of the stress on
the sputtering pressure for 1-lm-thick silicon
nitride film deposited at 100 W to 300 W RF power.
As seen from Fig. 5, the stress in the films is com-
pressive in the entire pressure range investigated,
and decreases from 1360 MPa to 390 MPa as the
sputtering pressure is changed from 5 mTorr to
20 mTorr for the film deposited at 300 W RF power.
At lower sputtering pressures, the inelastic collision
probability between plasma particles decreases,
enhancing the peening effects of the energetic
plasma particles.25,26 Therefore, the intrinsic stress
increases with decreasing pressure. In the case of
silicon nitride, deposition at higher pressure seems
to be favorable for obtaining low stress.
Figure 6 show the effect of film thickness on the
stress in the silicon nitride films. All the films were
deposited at 10 mTorr pressure and 300 W RF
power. From Fig. 6 it is clear that the stress in the
silicon nitride film is compressive in nature and
decreases as the film thickness is increased from
Fig. 5. Variation of residual stress in the silicon nitride films as a
function of sputtering pressure at different RF powers.
Silicon Nitride Films Deposited by RF Sputtering for Microstructure Fabrication in MEMS
Fig. 6. Dependence of residual stress in the silicon nitride film on
film thickness.
Fig. 7. Residual stress versus annealing time for silicon nitride films
deposited at 300 W RF power and 10 mTorr pressure for different
thicknesses.
0.5 lm to 2 lm. Deposition time is higher to obtain
thicker films, which provides more time for stress
relaxation during deposition and results in lowering
of the stress in the films. Alternatively, with
increase in the thickness of the silicon nitride film,
the electrical insulation over the substrate (silicon)
increases, which may alter/reduce the energy of the
incoming species (probably positive ions) due to the
bias over the substrate.
Figure 7 shows the effect of annealing time on the
stress in the silicon nitride film of different thick-
nesses deposited at 10 mTorr pressure and 300 W
RF power. From Fig. 7 it is observed that the stress
is released through the annealing process and
decreases initially with annealing time.
1983
Surface Roughness
The average surface roughness of the deposited
films is a critical parameter when these are
employed as an intermediate layer for wafer bond-
ing processes and for optical surface in MOEMS
devices and micromirror. An AFM was used for
measuring surface roughness. The AFM measure-
ments (surface roughness) of silicon nitride films
deposited at 100 W to 300 W RF power and at dif-
ferent sputtering pressures were carried out using a
Digital Instruments Nanoscope IIIA apparatus. The
films were deposited on high-quality n-type (100)-Si
pieces of size 10 mm 9 10 mm. The film thickness
was kept constant at 5000 Å except where surface
roughness was considered as a function of film
thickness. All the images have the same scanning
area of 4 lm 9 4 lm.
Figure 8a–c shows three-dimensional AFM ima-
ges of 5000-Å-thick silicon nitride films deposited on
silicon at 5 mTorr, 10 mTorr, and 20 mTorr pres-
sure, respectively, and at 300 W RF power. It is
clear from the images that the surface roughness
increased significantly with higher sputtering
pressure. This sputtering pressure dependence is
shown in Fig. 9, which shows that the roughness
was significantly low at lower sputtering pressure.
Atomic peening, caused by reflected high-energy
neutral species and the increased energy of sputter
gas atoms, might be the possible reason for the
smoother films at low pressure, since this phenom-
enon densifies the deposited films at low pres-
sure.25,26 The atomic peening effect decreases as the
pressure is increased. This pressure dependence can
also be explained by the gas scattering effect: the
number of scattering events in the path from the
target to the substrate increases with increasing
deposition pressure. Scattering reduces the energy
of sputtered particles and increases their angular
distribution. Less energy of sputtered particles
leads to reduced mobility on surface, while the
increased angular distribution enhances shadow-
ing. The contribution of these two factors results in
a rougher surface at a higher sputtering pressure.
Figure 9 also shows the effect of sputtering power
on the surface roughness. From Fig. 9, it is observed
that the surface roughness of the film decreases
when the sputtering power increases. The surface
mobility or diffusion of the sputtered atoms affects
the growth mechanism of the film. This mobility can
be enhanced by inputting more energy to the system
such as increasing the deposition temperature or
supplying impact energy through ion bombardment.
At high RF power (300 W), the kinetic energy of
sputtered ion/atoms increases and also the sub-
strate temperature increases due to self-heating
(Fig. 2). These two factors enhanced surface diffu-
sion and smooth the growth front, resulting in a
lower surface roughness. On the other hand, at low
RF power, ions have low energy and the substrate
temperature is also low due to lower self-heating.
1984 Bhatt and Chandra

(a) (b)

1 1
2 2
x 1.00 µm/div x 1.00 µm/div
3 3
µm z 50.00 nm/div z 50.00 nm/div
µm

(c)

1
2
3 x 1.00 µm/div
µm z 50.00 nm/div

Fig. 8. Atomic force micrographs of silicon nitride films deposited at 300 W RF power at (a) 5 mTorr, (b) 10 mTorr, and (c) 20 mTorr pressures.
The scanned area is 4 lm 9 4 lm and the film thickness is 0.5 lm for all the samples.

Fig. 9. Variation of surface roughness (Ra) as a function of Fig. 10. Surface roughness (Ra) versus thickness of silicon nitride
sputtering pressure at different RF powers. films deposited at 300 W RF power and 10 mTorr sputtering pressure.

Ions are therefore more likely to stay at the site of
arrival, thus creating a rougher surface. Figure 10
presents the effect of film thickness on the surface
roughness (Ra) deposited at 300 W RF power and
10 mTorr sputtering pressure. The average surface
roughness is observed to decrease as the film
thickness increases from 0.5 lm to 2 lm. The
deposition time for preparing thicker films is
greater, which allows the atoms to rearrange into
low-energy states by surface diffusion. This may be
Silicon Nitride Films Deposited by RF Sputtering for Microstructure Fabrication in MEMS 1985

the reason for the lower surface roughness of

thicker films.

Etching Characteristics (a) Oxidation of (100)-silicon

Etching of silicon nitride is very important in many

applications and device processing. Etching can be
accomplished by either dry or wet etching process.
The etching characteristics of RF-sputtered silicon (b) Oxide etching from top side
nitride films in buffered hydrofluoric acid (BHF) and
potassium hydroxide (KOH) were investigated. To
evaluate the etch rate of silicon nitride films in KOH,
a 40 wt.% aqueous solution was used. The etching (c) Deposition of silicon nitride
temperature was kept constant at 70°C by using a
constant-temperature bath. A reflux condenser was
used to maintain the desired concentration of the
solution throughout the etching. The etch rate was
determined by measuring film thickness before and
after etching using a thin-film analyzer.
The etch rate of silicon nitride film in BHF and
KOH, deposited at different deposition parameters,
are summarized in Table I. The value of etch rate of
RF-sputtered silicon nitride film in BHF and KOH
varied strongly with deposition parameters. The
value in BHF was observed to vary from 30 Å/min to
340 Å/min. Films deposited at 5 mTorr and
10 mTorr pressure at 300 W RF power showed (d) Lithography and etching of RF sputtered
minimum etch rate, close to that of LPCVD nitride silicon nitride
film.27 For sample 10SN300 (Table I), the etch rate
in KOH was around 1 Å/min. This film could be
advantageously used for fabricating free-standing
microstructures by selective KOH etching. Films
deposited at higher sputtering pressure and lower (e) Anisotropic etching of Si in KOH
power (100 W) showed relatively higher etch rates.
This can be explained in terms of the refractive (100)-Si Thermal SiO2
index and the position of the Si-N bond signal in the RF sputtered silicon nitride
FTIR measurements. For this film, the Si-N bond Fig. 11. Schematic process steps for fabricating a silicon nitride
shifts toward higher wavenumber and is near to the microcantilever beam using the bulk micromachining technique.
Si-O bond. This may be due to incorporation of some
oxygen in the film after deposition and is presumed
to be responsible for the higher etch rate in BHF
and KOH. From Table I it is seen that the refractive fabrication steps are shown in Fig. 11. These
index of the films decreases with decreasing depo- includes (a) thermal oxide growth (1 lm), (b) oxide
sition power. This reduction in refractive index is etching from the top side, back side oxide will act as
indicative of a reduction in the density of the film, a mask during anisotropic etching, (c) silicon nitride
which is responsible for the higher etch rates for deposition by RF sputtering (300 W, 10 mTorr,
films deposited at lower RF power. 2 lm thick), (d) lithography and etching of silicon
nitride by reactive-ion etching (RIE) in SF6 plasma,
and (e) Si anisotropic etching in KOH solution.
FREE-STANDING MICROSTRUCTURE
Figure 12 shows an SEM micrograph of the
FABRICATION
RF-sputtered silicon nitride cantilever beams fab-
To explore the RF-sputtered silicon nitride film as ricated by bulk micromachining processing.
a structural layer in microelectromechanical system,
microstructures such as microcantilever beams, Surface Micromachining
microbridges, and diaphragms of different sizes were
Cantilever beams using an RF-sputtered silicon
fabricated on (100)-silicon wafer using bulk, surface,
nitride as a structural material and RF-sputtered
and surface-bulk micromachining technology.
phosphosilicate glass (PSG) as a sacrificial layer
were fabricated. The sequential fabrication process
Bulk Micromachining
using a two-mask process is shown in Fig. 13.
RF-sputtered silicon nitride cantilever beams A PSG film of 3 lm thickness was deposited from
were fabricated using a single-mask process. The the PSG target on the (100)-silicon wafer, as shown
1986
Fig. 12. SEM photograph of an RF-sputtered silicon nitride micro-
cantilever beam fabricated by bulk micromachining processing.
(100)-Si
(a) Deposition of RF sputtered PSG film
(sacrificial layer)
(100)-Si
(b) Deposition and patterning of PSG
(100)-Si
(c) Deposition and patterning of silicon nitride
(100)-Si
(d) Removal of PSG
(100)-Si RF sputtered silicon nitride
RF sputtered PSG
Fig. 13. Schematic process sequence for fabricating a silicon nitride
microcantilever beam by surface micromachining technique.
in Fig. 13a and patterned using a photolithography
process. It is required that the etching of the PSG
film during platform formation results in a smooth,
gentle slope. This contributes to the conformal step
coverage of the structural material layer deposition.
Etching in BHF gives a gentle slope in the patterned
PSG film, as shown in Fig. 13b. In the next step,
1.5-lm-thick silicon nitride film was deposited at
Bhatt and Chandra
Fig. 14. SEM photograph of an RF-sputtered silicon nitride micro-
cantilever beam fabricated by surface micromachining processing
using RF-sputtered PSG as a sacrificial layer.
300 W RF power and 10 mTorr pressure. The silicon
nitride film was then patterned by photolithography
employing the second mask, followed by RIE in SF6.
Due to the topography of the wafer surface, normal
photoresist could not protect the sidewalls. A high-
viscosity photoresist (Shipley STR 1045) was coated
to overcome this problem. After patterning, the
sacrificial layer (RF-sputtered PSG) was removed in
a BHF solution and the cantilever beams were
completely released, as shown in Fig. 13d.
Figure 14 shows an SEM micrograph of an
RF-sputtered silicon nitride microcantilever beam
fabricated by the surface micromachining tech-
nique. Cantilever beams can be seen to be straight,
without any upward or downward bending, indi-
cating minimal stress gradient across the silicon
nitride film. This experiment also illustrates that an
RF-sputtered PSG can be used as a sacrificial layer
with RF-sputtered silicon nitride as a structural
material in surface micromachining processing.
Surface-Bulk Micromachining
The microstructures fabricated by a conventional
surface micromachined process have two ‘‘bends’’
around the anchor point. Due to this, local stress
may develop at the bends, which degrades the
mechanical performance of the beam. Furthermore,
at these bends, the structural layer thickness is
reduced due to the well-known step-coverage effect
during film deposition, and these become weak
points in the microstructures. Using a modified
surface micromachining technique, these short-
comings can be minimized.28 Completely planar
microstructures of RF-sputtered silicon nitride
were fabricated using RF-sputtered PSG film as a
sacrificial layer by this modified surface microma-
chining technology following a three-mask process.
The complete process flow is shown in Fig. 15,
which includes (a) oxidation of (100)-Si wafer,
Silicon Nitride Films Deposited by RF Sputtering for Microstructure Fabrication in MEMS
(100)-Si
(a) Oxidation of (100)-Si wafer
(b) Lithography, etching of the oxide
and anisotropic etching
(c) Global etching of thermal oxide
(d) Deposition of RF sputtered PSG
(Sacrificial layer)
(e) Patterning of PSG film
Legend for cross-sectional schematic
(100)-Si Thermal SiO2
Fig. 15. Schematic process sequence for fabricating a silicon nitride microcantilever beam by a modified surface micromachining technique.
(b) lithography (using mask 1) and etching of the
oxide in BHF followed by anisotropic etching of Si in
KOH to create 3-lm-deep cavities of different sizes,
(c) global etching of oxide, (d) deposition of PSG by
RF sputtering (thickness is more than cavity depth),
(e) pattering of PSG film using mask 2, (f) planari-
zation by CMP using Mastermet colloidal silica
polishing suspension (Buehler Co.), (g) deposition of
1.5-lm silicon nitride (as a structural material) by
RF sputtering at 300 W and 10 mTorr pressure, (h)
photolithography and etching of silicon nitride film
by RIE using SF6 gas (mask 3), and (i) removal of
RF sputtered PSG in BHF. Following the PSG etch,
the samples were rinsed in two successive DI water
baths, each lasting for 5 min, followed by a 10-min
dip in acetone. The final rinsing was done for 10 min
in a surface-tension-weakening solution (methanol)
and the wafers were dried under an IR lamp.
1987
(f) Planarization by CMP process
(g) Deposition of silicon nitride by
RF sputtering
Top view of mask # 3
(h) Photolithography and etching of
RF sputtered silicon nitride film
(i) Removal of RF sputtered PSG
(sacrificial layer)
RF sputtered PSG RF sputtered silicon nitride
Figure 16 shows SEM photographs of completely
planar microcantilever beams.
RF-SPUTTERED SILICON NITRIDE
AS A MASK IN OXIDATION
An RF-sputtered silicon nitride film was explored
as a mask in thermal oxidation processing.
Figure 17 shows a schematic view of the process
sequence used in LOCOS. The process was per-
formed on a silicon substrate having 2000 Å silicon
nitride deposited over 500 Å SiO2 (pad oxide, ther-
mally grown). Silicon nitride film was deposited by
RF sputtering at 300 W power and 10 mTorr pres-
sure. After silicon nitride deposition, the lithogra-
phy was performed to open windows. Silicon nitride
was etched in SF6 plasma (RIE mode) and the
pad oxide in BHF. Figure 18 shows an optical
1988
Fig. 16. SEM photograph of an RF-sputtered silicon nitride micro-
cantilever beam fabricated by the modified surface micromachining
process using RF-sputtered PSG as a sacrificial layer.
(a) Silicon nitride deposition
(b) Lithography and etching of Si3N4 and pad oxide
(c) Thermal wet-oxidation
(d) Removal of nitride
(100)-Si SiO2
Deposited silicon nitride
Fig. 17. Schematic process sequence of the LOCOS process.
photograph of silicon nitride lines after patterning
and etching process. Total thickness of silicon
nitride/pad oxide was measured by alpha step before
thermal oxidation. Wet oxidation at 1000°C was
then carried out for 1 h. After oxidation, silicon
nitride was globally etched out in orthophosphoric
acid. The oxide thickness was measured on areas
that were protected by silicon nitride during the
oxidation process. SEM was used to observe the
bird’s beak, which is typical in the LOCOS
process.29
Bhatt and Chandra
Fig. 18. Optical photograph of the silicon nitride lines after
patterning.
Fig. 19. SEM micrograph showing typical bird’s beak in the LOCOS
process.
It was observed that the combined thickness of
silicon nitride-pad-SiO2 remained unchanged after
the thermal oxidation. It was further confirmed that
the pad oxide thickness was also not affected after
removing the silicon nitride globally. Figure 19
shows an SEM photograph of the bird’s beak after
the LOCOS process. These observations demon-
strate that an RF-sputtered silicon nitride film
can be advantageously used for LOCOS isolation
technology.
CONCLUSIONS
Silicon nitride films deposited at low tempera-
tures (<260°C) by RF sputtering have been inves-
tigated for application in MEMS. The effect of
deposition parameters such as RF power and
Silicon Nitride Films Deposited by RF Sputtering for Microstructure Fabrication in MEMS
sputtering pressure on the chemical bonds, optical
transmittance, stress, surface roughness, and etch
rates were studied. It was observed that these
properties are strongly dependent on the sputtering
parameters. The films have good transparency in
the visible and near-infrared range, which is a
requirement for many optical applications. AFM
images show smooth-textured films, and the surface
roughness becomes smaller with increasing RF
power and film thickness. The etch rate of the films
was measured in BHF and KOH, and results are
comparable to those of LPCVD films. The wafer
curvature technique was used to measure the stress
in the film. For the investigated deposition param-
eters, all the films showed compressive stress.
Based on these studies, an optimized set of deposi-
tion parameters has been obtained, particularly for
fabricating microstructures for MEMS. The
optimized deposition parameters are 10 mTorr
sputtering pressure and 300 W RF power, which
results in lower surface roughness and lower etch
rate. The use of RF-sputtered silicon nitride films as
a structural layer in bulk micromachining process is
demonstrated by fabricating microcantilevers. It is
further demonstrated that an RF-sputtered silicon
nitride film can be used as a structural layer with
RF-sputtered PSG as a sacrificial layer in conven-
tional and modified surface micromachining tech-
nologies. The LOCOS process further confirms that
RF sputtering is a useful, safe, and low-cost method
for preparing silicon nitride films for rapid proto-
typing of MEMS and other devices.
REFERENCES
1. S. Isomae, S. Yamamoto, S. Aoki, and A. Yajima, IEEE
Electron. Dev. Lett. Ed. L-7, 368 (1986). doi:10.1109/
EDL.1986.26403.
2. H. Hwang, W. Ting, B. Maiti, D. Kwong, and J. Lee, Appl.
Phys. Lett. 57, 1010 (1990). doi:10.1063/1.103550.
3. M.D. Dange, J.Y. Lee, and K. Sooriakumar, Microelectron.
J. 22, 19 (1991). doi:10.1016/0026-2692(91)90010-K.
4. L. Shi, C.A.M. Steenbergen, A.H. de Vreede, M.K. Smit,
T.L.M. Scholters, F.H. Groen, and J.W. Pedersen, J. Vac.
Sci. Technol. A 14, 471 (1996).
5. M. Modreanu and M. Gartner, J. Mol. Struct. 565–566, 519
(2001). doi:10.1016/S0022-2860(00)00907-8.
1989
6. K.J. Winchester and J.M. Dell, J. Micromech. Microeng. 11,
589 (2001). doi:10.1088/0960-1317/11/5/323.
7. L. Wang, Z. Cui, J.S. Hong, E.P. McErlean, R.B. Greed, and
D.C. Voyce, Microelectron. Eng. 83, 1418 (2006). doi:10.1016/
j.mee.2006.01.067.
8. A.T. Lopes and M.N.P. Carreno, J. Non-Cryst. Solids 338–
340, 788 (2004). doi:10.1016/j.jnoncrysol.2004.03.092.
9. H.-J. Nam, Y.-S. Kim, C.S. Lee, W.-H. Jin, S.-S. Jang, I.-J.
Cho, J.-U. Bu, W.B. Choi, and S.W. Choi, Sens. Actuators A
Phys. 134, 329 (2007). doi:10.1016/j.sna.2006.05.030.
10. K. Watanabe and S. Wakayawa, J. Appl. Phys. 53, 568
(1982). doi:10.1063/1.329920.
11. P. Pan and W. Berry, J. Electrochem. Soc. 132, 3001 (1985).
doi:10.1149/1.2113711.
12. T.J. Cotler and J. Jonathan, Electrochem. Soc. 140, 2071
(1993). doi:10.1149/1.2220766.
13. G.N. Parsons, J.H. Souk, and J. Batey, J. Appl. Phys. 70,
553 (1991). doi:10.1063/1.349544.
14. I. Umezn, K. Kohna, T. Yamaguchi, A. Sugimura, and M.
Inada, J. Vac. Sci. Technol. A 20, 30 (2002). doi:10.1116/
1.1419083.
15. J.S. Yoon, C.V. Deshpandey, and R.F. Bunshah, Thin Solid
Films 220, 80 (1992). doi:10.1016/0040-6090(92)90552-M.
16. H.P. Lobi and M. Huppertz, Thin Solid Films 317, 153
(1998). doi:10.1016/S0040-6090(97)00512-9.
17. R.D. Gould and S.A. Awan, Thin Solid Films 515, 509
(2006). doi:10.1016/j.tsf.2005.12.285.
18. M. Vila, C. Prieto, P. Miranzo, M.I. Osendi, and R. Ramirez,
Surf. Coat. Technol. 151–152, 67 (2002). doi:10.1016/S0257-
8972(01)01600-0.
19. B.S. Yau and J.L. Huang, Surf. Coat. Technol. 176, 290
(2004). doi:10.1016/S0257-8972(03)00768-0.
20. L. Yu, J. Xu, S. Dong, and I. Kojima, Thin Solid Films 516,
1781 (2008). doi:10.1016/j.tsf.2007.07.158.
21. C. Iliescu, F.E.H. Tay, and J. Wei, J. Micromech. Microeng.
16, 869 (2006). doi:10.1088/0960-1317/16/4/025.
22. Q. Song, Z. Cui, S. Xia, Z. Chen, and J. Zhang, Sens. Actu-
ators A. Phys. 112, 122 (2004). doi:10.1016/j.sna.2003.
12.017.
23. L. Maissel and R. Glang, Handbook of Thin Film Technology
(New York: McGraw-Hill, 1970), pp. 4–15.
24. M. Ohring, The Material Science of Thin Film, Deposition
and Structures. Mechanical properties of Thin Films, 2nd
ed. (New York: Academic, 2002), chapter 12.
25. J.A. Thornton and D.W. Hoffman, Thin Solid Films 171, 5
(1989). doi:10.1016/0040-6090(89)90030-8.
26. I. Sugimoto, S. Nakano, and H. Kuwano, Thin Solid Films
268, 152 (1995). doi:10.1016/0040-6090(95)06903-8.
27. R. Wolf, K. Wandel, and B. Gruska, Surf. Coat. Technol.
142–144, 786 (2001). doi:10.1016/S0257-8972(01)01184-7.
28. V. Bhatt and S. Chandra, Sens. Lett. 5, 387 (2007).
doi:10.1166/sl.2007.207.
29. S.K. Gandhi, VLSI Fabrication Principles (New York:
Wiley, 1982), p. 580.