A Study of Sub-Atomospheric Chemical Vapor Deposition (SACVD) 0 3 -

TEOS UGS, BPSG, PSG, and Plasma Enhanced Chemical Vapor

Deposition (PECVD) PSG Film as Sacrificial Layers in a Micro-fluid
Dispenser Device
Ming-Tung Lee, Peter Huang, May Chang, Yi-Yueh Chen
Thin Film Process Dept., Macronix International Co., Ltd.
No.3 Creation Road 111, Science-based Industrial Park, Hsin Chu, Taiwan, R.O.C.
E-mail address: [email protected]

dispenser device are presented. Results show the film properly

Abstract - I n this paper, we have investigated SACVD 0 3 - is changed during the thermal processing steps. By using the
TEOS USG, BPSG, PSG and PECVD PSGfilm as sacrifcial FTIR spectrum analysis, mainly in comparison with peak
layers toformation afluid rank with afluid barrier liver in a height of P=O (1330cm.l) and B-0 (135Ocm.') after rapid
micro-fluid dispenser device. The wet etch rate of as deposited thermal anneal (RTA), and furnace anneal processing steps
films was measured and the SACVD PSG film performed with during MEMS fabrication. The wet etch rate variation of
the highest wet etch rate as 5913Mmin in a 6:1 buffered oxide sacrificial films is revealed. Also ffom the measurement of
etch[BOE] solution. Inversely SACVD BPSG film displayed a sacrificial layers etch out speed, and fluid tank profile
lowest wet etch rate as 165OMmin in comparison with other observation liom the pattern wafer, the optimized sacrificial
sacrifcial material above. material is explored.
By the way, in process integration consideration, EXPERIMENTAL PROCEDURE
patterned wafer with stacking sacrificial layer and fluid
barrier layers were fabricated, then sacrifcial layer etched Sacrificial film-covered (100) p-type silicon wafers
out in HF solution. Based on microscopic view the SACVD were used as substrates in this study. The SACVD 03-TEOS
PSG film again displayed a fast removal. However, SACVD USG, BPSG, PSG films and PECVD PSG film were
BPSGfilm shown a better removal than PECVD PSG film is deposited in AMAT Precision 5000 System, and the process
different with as-depositedfilm characteristic. From a series factors were shown in Table 1. The sacrificial films wet etch
of analysis by FT-IR instrument, control wafer deposited with rate investigation was performed by using Optiprobe
dflerent sacrificial films above, then processed with rapid Thermawave 2600 film thickness measurement tool for as-
thermal anneal orfirrnace anneal process shown the boron (B) deposited wafer, wet etch under 6:1 buffered oxide etch
content decrease a#er thermal anneal is the dominatorfor the [B.O.E] solution. For SACVD BPSG film with different
wet etch properly performed. boron and phosphor content, the samples were thermal anneal
by rapid thermal process (950C) or furnace process (SSOC)
Also based on wei etch rate ratio W.E.R.R test with firstly, then wet etch under 5O:l B.O.E. solution to determine
different boron (B) and phosphor (P) content under 50-1 the boron and phosphor effect on wet etch properly. FTIR
B.O.E. etched, it shown as phosphor increase W.E.R increase, spectrum analysis (Bio-Rad QS-300 Model) was employed in
but boron is inversely. According to the patterned wafer wet sacrificial films covered control wafer, to examine boron and
etch rate measurement and fluid tank profile observation by phosphor variation during thermal anneal process.
using optical microscope (OM) and scanning electron
microscope (SEW respectively, the SACVD PSG f l m display Pattern wafer were fabricated by sacrificial film
the bestperfarmonce as device requirement. PECVD PSG, SACVD PSG, BPSG deposited individually
and rapid thermal anneal (950C) in Steag SHS2000 system
firstly, photo mask and dry etch to define fluid tank region
INTRODUCTION secondly, then fluid barrier layer deposited.
The use of sacrificial materials is common in
fabrication of 3D microstructures such as integrated circuits For sacrificial layer performance evaluation, pattern
and microelectromechanical systems (MEMS) [l]. For a fluid wafers were backside fluid barrier layer and Si-substrate
dispenser device sacrificial material is general used to form a removal firstly by using dry etching and 1-st KOH solution
fluid tank structure under fluid barrier layer. Based 011process individually, then sacrificial layer etch out by using HF
integration this sacrificial material need to be easy formation, solution, and finally fluid tank formation by 2-nd KOH
and to exhibit highly selective and controlled removal of the solution etched. Optical microscope (OM) was used in
sacrificial material with minimal alternation of non-sacrificial measurement of sacrificial film wet etch rate on pattern wafer
structure during a series of processing step [2]. In this paper, processed above, and scanning electron microscope (SEM)
comprehensive studies on use of sub-atmospheric chemical also be used to observation the fluid tank profile revealed by
vapor deposition (SACVD) 03-TEOS USG, BPSG, PSG and immersion in HF solution one day.
plasma enhanced chemical vapor deposition (PECVD) PSG as
sacrificial layer to form fluid tank structure in a fluid

0-7803-7604-8/02/$17.00
02002 IEEE

208
 RESULT AND DISCUSSION barrier layer formation process (SSOC), so SACVD BPSG
film's wet etch rate became fast than as-deposited film. On the
(I). As-deposited sacrificialfilms wet etch rate other hand, no obviously phosphor content changed during
thermal process, so SACVD PSG film performed a best wet
Consideration the meaning of sacrifice, the wet etch etch rate as device requirement. However, the PECVD PSG
rate of asdeposited sacrificial films is measured under 6:l film show a slowest wet etch rate performance, may be due to
B.O.E. solution fmtly, and the result is shown in Fig.1. From less phosphor content exist in film as FTIR spectrum show
the data, it can be seen that presence of impurities in the oxide P=O (1330cm") peak height in Fig2 (c).
strongly affect the etch rate. Boron results in a reduced etch
rate as comparison of SACVD PSG and BPSG films wet etch (5). Fluid tank profire observation from pattern wafer afrr
rate. And phosphor increases it as comparison of SACVD sacrificialfilm andfluid barrier layer removal
PSG and USG films.
Designing an optimum fluid dispenser device. The
(2). Boron andphosphor content effect on SACVD BPSG films fluid tank profile is an import factor, not only it affected on
wet etch rate fluid dispenser efficiency and fluid droplet formation but also
the device dispense reliability. [4] Scanning electron
Based on the result (I), the SACVD BPSG films microscope (SEM) was used to observation the fluid tank
with different boron and phosphor content are thermal anneal profile in pattem wafer. From the Fig.3 (a), (b) it shown use
by rapid thermal tool (950C) or horizontal furnace (SSOC) to the SACVD PSG film as sacrificial material, the profile
investigate thermal budget effect, then wet etch tested under revealed is better than that use SACVD BPSG as sacrificial
5 0 1 B.O.E. solution. The result is listed as Table 2, again the material. Since latter one got a much narrow fluid tank wall, it
boron cause a reduced wet etch rate ratio (W.E.R.R.) compare is easy to crack during the fluid dispenser operation, and
with thermal oxide film, and phosphor display an inversely induce an unexpected device failure. Based on the experiment
behavior, whatever the sample is rapid thermal anneal or data, during the sacrificial material etching the fluid barrier
furnace anneal. Furthermore, the furnace annealed exist a layer is also etched and the wet etch rate measured under HF
lower W.E.R.R., it could be explained due to the boron and solution is about 1 IENmin. By the way, for pattern wafer the
phosphor out-diffusion during thermal process. [3] Since more time needed for sacrificial material etching out is more longer
elapsed time in furnace than in rapid thermal anneal cause for SACVD BPSG film about 18 minutes with respective to
more doping atom outdiffusion and also more pure and dense SACVD PSG film about 4 minutes. So the Si-substrate
silicate film formed, so W.E.R.R. reduced more. exposed in 2-nd KOH solution etching process is larger in
pattern wafer used SACVD BPSG as sacrificial material, and
(3). Investigation of boron and phosphor variation in finally cause the unwanted fluid tank profile as shown in Fig.
sacrificial films from FTIR spectrum during thermal anneal 3 (a). Although the KOH solution also have the effect on Si-
process substrate etching as literature reported[Sl, and in Fig.3 (a), it
seems the KOH induced roughness surface appeared in the
Control wafers with SACVD PSG, BPSG, and bottom of fluid tank profile. The repeated test result as in
PECVD PSG films covered were individually subjected into Fig.3 (c) show in spite of smooth fluid tank profile in the
rapid thermal process (9SOC), then furnace anneal process pattem wafer with SACVD BPSG film as sacrificial material,
(SSOC) to simulate the integration process as device it did cause unwanted fluid tank profile, regardless of KOH
fabrication and the boron and phosphor content variation were solution used in this study.
analyzed by accompanying FTIR spectrum measurement.
From Fig2 (a) it shown boron and phosphor both decrease
CONCLUSION
during thermal anneal process as P=O (133Ocm") and B-0
(1350cm.l) peak height reduced in gradually. But in contrast, It has been shown how the boron and phosphor
the phosphor content in SACVD PSG and PECVD PSG is content affects the wet etch rate during the thermal process in
almost the same between rapid thermal anneal and furnace device fabrication. To choose a optimized material as
anneal as shown in Fig2 (b), (c), P O (133Ocm'l) peak height. sacrificial layer the key contains not only a fast wet etching
It probably due to only when boron and phosphor exist performance but also material needs to be easy formation, and
together would enhance the out-diffusion as literature [3] to exhibit highly selective and controlled removal of non-
illustrated
sacrificial structure during a series of processing step. Based
on this study, SACVD- PSG film is selected, and well
(4). Measurement of sacri$cialfilms wet etch rate performed performance obtained.
on pattern wafer
For sacrificial layer performance evaluation in ACKNOWLEDGEMENT
pattem wafers, the result as in Table 3. The SACVD PSG film
display a most fast wet etch rate about 67.5udmin than The authors would like to thank their colleagues
SACVD BPSG ls,sum/min and PECVD PSG 1 0 , 9 d m i n S.L.Tu and Ben Lee for their collaboration, advice and fruitful
films, regardless of film thickness is 6.5KA or 9.0KA. It can discussion in this research
be explain due to boron and phosphor content out-diffusion
during rapid thermal anneal (950C) firstly, and then fluid

209
 Table 1. A summary of the processing conditions for sacrificial
BPSG23-1: Asdeposited
BPSG23-2:Rapid thermal anneal

Asdeposited sacriilcialf i s wet etch rate Figure2 (a). FTIR spectrum of SACVD BPSG film undergoes
with different thermal annealing.
12000
1
I loo00 '
I A SAPSGOB-I: As-deuosited
SAPSGOB-2:Rapid &mal anneal
Si-0
(l0804-')

SAPSGO8J:RTA + Furnace
D

I
Figure 1. Wet etch rate ofseveral as-deposited sacrificial films i n 2 (b). ~
~i~~~~ I spectrum
R of SAC-D PSG filmundergoes
6 1 buffered oxide etch solution. with different thermal annealing.
. .. .
Table 2. Wet etch rate ratio of SACVD BPSG film compared
with thermal oxide with different boron and phosphorus content 1.
by adjusting process factor TEPO, TEB flow undix rapid PEPSG23-1: As-deposited
thermal anneal o r furnace anneal process. Wet etched in S0:l PEPSG23-2Rapidthermal anneal
buffered oxide etch solution. PEPSG23-3:RTA + Furnace

rfFSe-2
IF90.

Figure 2 (c). FTIR spectrum of PECVD PSG film undergoes with

different thermal annealing.

210
 ~

Table 3. The opriral microscope measurement result of we1

ctrhine rate bv.mineISACVD PSG BPSG and PECVD PSG films
as sacrificial layer in pattern wafer.
SACVD BPSG 6500A film as sacrificiallayer
HF solution 16 minutes 110 minutes

HF solution Film thickness

wet etch time Film thickness 6500A 90WA

4 minutes Figure 3 . Scanning electron microscope picture shown fluid
4 minutes tank profile of fluid dispenser device with SACVD BPSG (a), (e)
Removal 41um 46.5um and SACVD PSG (b) as sacrificial layers in respectively.
Wet etch rate 10.25udmin. 11.63udmin.
REFERENCES
[I] Steven D. Leith et al., “In-situ fabrication of sacrificial
layers in electrodeposited NiFe microstructures”, I.
Micromech. Microeng., vol. 9, pp97-104, 1999.

[Z] H. Bemey et al., “Investigation of the effect of processing

steps on stress in a polysilicon structural membrane”, I.
Micromech. Microeng., vol. IO, pp223-234,2000,

[3] Masaki Yoshimaru et al., “Microcrystal Growth on

Borophosphosilicate Glass Film during High-Temperature
Annealing”, J. Electrochem. Soc., vol 143, pp666-
671,1996.

[4] T. Laurel1 et al., “Design and development of a silicon

microfabricated flow-through dispenser for on-line
picolitre sample handling”, J. Micromech. Microeng., vol.
9, pp369-376, 1999.

[SI 0. Powell et al., “Anisotropic etching of {IOO} and { 110)

planes in (100) silicon”, J. Micromech. Microeng., vol. 11,
pp217-220,2001.

AUTHOR BIOGRAPHY

Ming-Tung Lee received B. S . and M. S . degree in Materials

Science and Engineering ffom Fung-Chia University and
Cheng-Kung University in respectively. He joined Macronix
International Company (MXIC) Taiwan as an Engineer in
1998 and was engaged in Thin Film Module. He is now a thin
film process leader of MXIC FABI in the field of
metallization.

Peter Huang received a B. S . degee in the Electric

Engineering 6om Chung-Hua University. He joined Macronix
lntemational Company (MXIC) Taiwan as an Engineer in
1996 and was engaged in Metal Sputter Hardware. He is now
a thin film process engineer of MXIC FABI in the field of
metallization.

21 1