ESCAMPIG XXI, Viana do Castelo, Portugal, July 10-14 2012
Investigations in SF6 and Cl2/Ar plasmas used for titanium deep etching by
means of mass spectrometry
T. Tillocher (*)1, J. Golda 1, P. Lefaucheux 1 , B. Boutaud 2 , P. Ranson 1, R. Dussart 1
1
GREMI, Université d’Orléans/CNRS, 14 rue d’Issoudun, BP 6744, 45067 Orléans cedex 2 France
2
SORIN CRM, 4 avenue Réaumur, 92140 Clamart cedex, France
(*)
[email protected]
Titanium can be deep etched to a few hundreds of µm with SF6 and/or Cl2/Ar plasma(s). A
chlorine-based chemistry enhances the anisotropy of the profiles. However, reproducibility is low
and a better understanding of the mechanisms would help to optimize etching processes. In this
study, we present data of SF6 and Cl2/Ar plasmas chemistry under titanium etching conditions
obained by mass spectrometry.
Titanium is a biocompatible material which is of great interest in the biomedical field for
cardiology, surgery etc.... These last years, devices like bioMEMS have emerged with applications for
biosensors, drug delivery and pacemakers. Their fabrication, where vertical sidewalls and smooth
surfaces are required, is based on micromachining techniques derived from microelectronics
technologies. Most of the research work reported in the literature relies on a Cl2/Ar chemistry to deep
etch titanium. Processes are performed at room temperature of the substrate with typical etch rates
close to 1 µm/min and provide rather smooth surfaces [1, 2]. A patterned TiO2 layer is often used as a
hard mask. It has also been suggested recently that a thick (several 10s of microns) SU8 layer, a
negative photoresist, can used as a mask for deep etching of titanium [3]. However, to our knowledge,
the chemistry and the mechanisms involved in titanium deep etching are not yet completely
understood.
The study we present here was performed in an Alcatel 601E ICP etching tool.The substrates were
4’’ Ti wafers with various millimetric or submillimetric structures: circular and rectangular pillars and
rings. We have investigated different mask materials: 40-45 µm thick SU8 layers, TiO2 or Ni. This
first investigation showed that Ni was the best candidate in our experimental conditions. Actuallty, the
etched surface was rather smooth compared to what was obtained with the other mask materials. Mask
sputtering and redeposition induced roughness with SU8 and the etch rate dropped dramatically with
the TiO2 mask. Wafers were cut into pieces and glued on a 6’’ silicon carrier wafer. The carrier
wafers were electrostatically clamped and the chuck temperature was regulated at 20°C. The process
gases were SF6, Cl2, and Ar. Depending on the process run, helium flow could be used for thermal
backside contact. Etched profiles were characterized using Scanning Electron Microscopy.
Fig. 1: Examples of features etched with a SF6/Cl2/Ar chemistry. Process time is 2h and etched depth is
127 µm. ( 20 mm x 20 mm sample glued on a 6’’ Si carrier wafer)
Our results have shown that either a SF6 plasma or a Cl2/Ar plasma can be used to etch titanium. A
SF6 plasma helps to reach etch rates as high as 4 µm/min at higher pressures (a few Pa) but profiles are
isotropic. A Cl2/Ar chemistry at low pressure (almost 1 Pa) is preferred to get vertical sidewalls but, in
this latter case, the etch rate is reduced. This is why we proposed to mix these two chemistries to
ESCAMPIG XXI, Viana do Castelo, Portugal, July 10-14 2012
obtain the profiles displayed in figure 1. In this example, the mask was a 5 µm patterned Ni layer. The
samples were etched for 2h and the substrate temperature was 20°C. The sidewalls are anisotropic and
more especially negatively tapered. The depth is 127 µm, which represents an etch of almost
1 µm/min. The surface is fairly smooth. However, these performances are not reproducible since in
most cases this process leads to a high roughness. Consequently, the etch rate is reduced.
This non-reproducibility may be due to the silicon carrier: SiClx species, coming from the etch
by-products of the silicon wafer may redeposit on the surface [5] and induce a micro-masking effect.
The plasma was characterized using a Hiden Analytical EQP mass spectrometer. This device can
be used to analyse both neutral and charged species (positive and negative ions). In the first case, the
species are ionized at their entrance in the system and discriminized as a function of the m/z number
by the quadrupole mass filter. Ions can be analysed depending on their energy through an electrostatic
sector and depending on their mass through the mass filter. The mass spectrometer was moved into the
plasma bulk with a linear shift mechanism to analyse the plasma close to the samples.
The two chemistries (SF6 and Cl2/Ar) were studied separately as a function of time in different
etching conditions: without any wafer, with a piece of titanium glued on the silicon carrier and a bare
titanium carrier. Masses were scanned between 1 and 180 amu since no significant signal was detected
for higher values. For neutral analysis, electron energy was set to 70 eV. Lower electron energy did
not help to detect larger molecules or significantly increased intensities at larger m/z ratio. Typical
spectra obtained for both chemistries are represented in figure 2. This helps to identify the species
present in the plasma. We will present the study of their kinetics and propose preferential reaction
pathways and physical mechanisms. Finally, we will correlate these results with the etching results.
+
+
NO
HCl
+
+
+
F2
+
NO2
+
N
+
HO
+
+
CO2
+
N2O
+
O
O2
3
10
H
10
5
Ti
+
+
10
+
4
Cl2iso
+
TiF2
+
TiF
TiCl
0
10
10
3
20
30
40
10
2
60
50
70
80
+
+
TiF3
+
TiCl3iso
+
TiCl2iso
10
3
10
2
+
TiCl3iso
+
TiCl2iso
(b)
NO
HCl
Cl
5
10
+
+
+
H2O
HF
S
+
O2
+
HO
4
N
+
+
Cl iso
HCl
iso
+
F2
SO
+
Ti
+
CO2
+
N2O
+
O
S2
+
SO2
NO2
+
+
+
N2
+
100
110
5
10
+
SOF2
+
S2F
140
150
160
170
180
6
10
SO2F
+
TiF
+
SOF
+
SF3
SF
+
TiF3
+
SOF3
+
SF2iso
+
Cl2iso
4
10
+
+
SF3iso
10
+
SF4
3
2
5
+
S2F2
+
SO2F2
10
10
30
m/z (amu)
40
50
60
4
10
+
SF5
3
10
2
2
10
20
130
+
SF iso
+
10
120
m/z (amu)
+
+
SF2
+
Cl2
+
+
+
10
90
TiF2
6
10
Counts (c/s)
10
Counts (c/s)
+
+
TiCl3
4
10+
TiCl2
+
F
+
H3O
6
0
5
m/z (amu)
m/z (amu)
H
10
+
2
3
6
+
Cl2
Cl2iso
10
10
10
+
Ar
Cl
H2O
4
10
6
Counts (c/s)
Counts (c/s)
Al
+
N2
+
HF
++
Ar
Counts (c/s)
+
5
10
10
+
Cl iso
Counts (c/s)
F
+
H3O
6
10
70
80
90
m/z (amu)
100
110
10
120
130
140
150
160
170
180
m/z (amu)
Fig. 2: Mass spectra of (a) a Cl2/Ar plasma (1 Pa – 2000 W source power) and (b) a SF6 plasma
(1Pa – 1000 W source power) in presence of a titanium wafer.
References
[1] M.F. Aimi et al., Nature Materials, 3 (2004)103.
[2] E. R. Parker et al., Journal of The Electrochemical Society, 152 (2005) C675-C683.
[3] Gang Zhao et al., 4th IEE Int. Conf. on Nano/micro Engineering and molecular systems, 514
(2009)
[4] R. d’Agostino et al., Journal of Applied Physics, 71 (1991) 462.
[5] C. Y. Dulard et al., Journal of Physics D, 42 (2009) 115206