MNE-00002; No of Pages 4

Micro and Nano Engineering xxx (xxxx) xxx–xxx

Contents lists available at ScienceDirect

Micro and Nano Engineering

journal homepage: www.elsevier.com/locate/tsf

Research paper

Cleaning surfaces from nanoparticles with polymer film: impact

of the polymer stripping
Adeline Lallart a,b,c,d,⁎, Philippe Garnier a, Elise Lorenceau b, Alain Cartellier c, Elisabeth Charlaix b
a
STMicroelectronics, 850 Rue Jean Monnet, 38926 Crolles, Cedex, France
b
Univ. Grenoble Alpes, CNRS, LIPhy, F-38000 Grenoble, France
c
Université Grenoble Alpes, CNRS, Grenoble INP, LEGI, F-38000 Grenoble, France 1
d
CEA-LETI, MINATEC Campus, 17 Rue des Martyrs, 38054 Grenoble, France

a r t i c l e i n f o a b s t r a c t

Article history: The removal of nanometric particles constitutes one of the main challenges for the Integrated Circuits
Received 31 May 2018 manufacturing. A solution based on the polymer coating and removal without substrate consumption is
Received in revised form 27 August 2018 described and its performances are evaluated. In our experiments, 60 nm SiO2 particles and 40–200 nm Si3N4 par-
Accepted 30 September 2018
ticles are used to contaminate Si wafers. Two polymer removal methods are compared, one is purely based on a
Available online xxxx
chemical action while in the other one a chemical and a physical actions are coupled. We demonstrate that a
Keywords:
physical action is required to remove particles. The process shows high Particle Removal Efficiency (PRE) up to
Polymer 87% independently of the particle size and nature. The PRE stays at a constant value, around 85%, for 3 decades
Particle removal efficiency of aging time, but the particle removal is not uniform on all the wafer.
Contamination © 2018 The Authors. Published by Elsevier B.V. This is an open access article under the CC BY license (http://
Particle size creativecommons.org/licenses/by/4.0/).
Aging

1. Introduction them present some limitations, either due to the restricted range of par-
ticle that can be cleaned off or due to pattern damages. A new promising
In the last decades, it has been observed that the number of transis- method using polymer layer, which has been barely investigated so far,
tors in integrated circuits has doubled every two years. This law, known circumvent these limitations. It consists in two steps: first a thin poly-
as Moore's Law, places enormous constraints on manufacturing mer layer, which embeds the particles within it, is spin-coated on the
processes. For example, it requires us to invent new manufacturing pro- surface. The polymer layer - and the embedded particles - is then re-
cesses but also to ensure more and more critical control over the quality moved from the surface either by siphoning it if it is liquid or by peeling
of the surrounding material. Indeed, parasitic particles, emitted from air it if it is solid [5–7]. In this work, we propose a new way to remove this
or manufacturing equipment can deteriorate the performance of inte- polymer by chemical dissolution either in a wet bench or by spray and
grated microcircuits leading to short circuit or micro masking issues. spin drying the chemistry assisted by wafer rotation. In the following,
With the drastic miniaturization of microcircuits, these critical sizes we describe our experimental protocol and discuss whether applying
get smaller and smaller. Thus, to continue to achieve high yield various a physical action during the chemical removal process via the impact
solutions are used to remove particles from microelectronic chips. of spray droplets and centrifugal forces enhance the efficiency of the
A common method to remove particles is based on the use of SC1 whole process or not.
(NH4OH:H2O2:H2O) chemistry. When put into contact with the micro-
electronic chip, this chemistry will etch the substrate and/or the parti- 2. Material and methods
cle, leading to a detachment of the particle from the surface and its
removal thanks to favorable pH conditions. However, by chemically The different steps of our experimental protocol are described on
etching the surfaces of the microelectronic circuit, this method degrades Fig. 1a). We first artificially contaminate 300 mm bare chemically oxi-
the resolution of the patterns, thus limiting the miniaturization of the dized silicon wafers by spin-drying. To do so, we used deionized water
circuits [1]. Other methods have been proposed such as acoustic solution containing either monodisperse SiO2 particles with a size of
methods [2], spray impact [3] or cryogenic method [4]. However, all of 61.7 nm + −1.6 nm or polydisperse Si3N4 particles ranging from
40 nm up to 200 nm. The initial particles count on the wafer is com-
⁎ Corresponding author.
prised between 40,000 and 60,000. After this contamination step, a
E-mail address: adeline.lallart@st.com (A. Lallart). polymer layer with a thickness of 1.8 μm is spin coated on the wafer.
1
Institute of Engineering Univ. Grenoble Alpes Then, this layer is baked during 90 s at 130 °C, which is in the order of

https://doi.org/10.1016/j.mne.2018.09.001
2590-0072/© 2018 The Authors. Published by Elsevier B.V. This is an open access article under the CC BY license (http://creativecommons.org/licenses/by/4.0/).

Please cite this article as: A. Lallart, et al., Cleaning surfaces from nanoparticles with polymer film: impact of the polymer stripping, (2018),
https://doi.org/10.1016/j.mne.2018.09.001
2 A. Lallart et al. / Micro and Nano Engineering xxx (xxxx) xxx–xxx

Fig. 1. a) Process and time description b) Sketch of the geometry of the spray batch equipment in top view.

the glass transition temperature (Tg) of the polymer. The time lapse be- after the initial contamination (pre particle counts) and after the com-
tween the contamination step and the spin coating of the polymer layer, plete polymer removal (post particle counts) thanks to a laser diffrac-
noted τpc, has been varied between 11 h and 19 days. To remove the tion spectrometer (Surfscan SP3 KLA Tencor) with a defect detection
polymer layer, two methods have been used. First, the wafer is im- ranging between 30 nm up to 2 μm.
mersed in different tanks containing the chemistry. A tank containing The efficiency of the method is quantified using the particle removal
Sulfuric Peroxide Mixture SPM (H2SO4:H2O2) is first used to dissolve efficiency (PRE), defined as follow:
the polymer layer associate to a second immersion in a SC1 tank. The
second method uses a spray batch equipment to dispense the chemis- initial particles amount−final particles amount
PRE ð%Þ ¼  100 ð1Þ
tries. In this case, the contaminated wafers are put on rotation at 80 initial particles amount
RPM around a central spray dispenser, which is located 25 cm away
from the center of the wafer as shown in Fig. 1b). The SPM and the
SC1 chemistries are successively dispensed by a central spray horizon- 3. Results
tally flowing droplets obtained from the co-flow of nitrogen at a flow
rate of 4 L/min and liquid chemistry at 1 L/min. Then, the wafers are fi- To quantitatively compare the methods used to remove the polymer
nally water rinsed and dried at a rotation of 300 RPM. The time between layer, we map the particle distribution on the wafer along with the par-
the polymer coating and the polymer removal (τpr) is between 6 h and ticle removal efficiency as defined in Eq. (1), in Fig. 2 right after the ini-
42 days, leading to a queue time (τaging) between the contamination and tial contamination (a), after the wet bench removal step (b) or after the
the polymer removal, corresponding to the addition of the times τpc and spray batch removal step (d). To quantify the influence of the spray im-
τpr, between 17 h and 57 days. The particles on the wafer are counted pact on the process, we also provide these two items when the spray

Fig. 2. Particles mapping and PRE after the different processes a) after the initial contamination b) after a polymer coating and the wet bench process c) after spray batch process d) after
polymer coating and the spray batch process.

Please cite this article as: A. Lallart, et al., Cleaning surfaces from nanoparticles with polymer film: impact of the polymer stripping, (2018),
https://doi.org/10.1016/j.mne.2018.09.001
 A. Lallart et al. / Micro and Nano Engineering xxx (xxxx) xxx–xxx 3

batch removal step has been carried out without any polymer coating decades of aging time. Moreover, we also varied the time between the
(c). contamination and the polymer coating (τpc) and the time between
the polymer coating and the polymer removal (τpr) and didn't observe
3.1. Comparison of the removal efficiency using the wet bench and the spray any variation.
batch process We also report in Fig. 4, PRE for different nature and size of the par-
ticles. Here also, the PRE exhibits a plateau at 85% + −5%.
When a chemical action is solely used, like in the wet bench method,
no particles are removed. Thus, both the physical action of the rotation 4. Discussion
and the spray is mandatory in order to withdraw particles from the
surface. To understand these results, we recall that when a solid particle is
deposited over a solid surface in the presence of humidity, the adhesion
3.2. Influence of the polymer presence on the particle removal efficiency force between the particle and the solid surface, which is of capillary or-
using the spray batch equipment igin, varies as a function of time τaging as written below [10]:
 
We also observe that the presence of the polymer layer is mandatory
1 τ
F adhesion τaging ¼ γd   ln aging ð2Þ
to obtain high value of PRE. Indeed, when using the spray batch equip- P sat τ0
ln
ment on contaminated wafers without the polymer layer coating Pv
(Fig. 2c)), no particles are removed whereas with the polymer layer
(Fig. 2d)) a removal efficiency up to 87% is achieved. This ensures that where γ is water/air surface tension, d is a distance taking into account
physical action applied by the impact of the spray droplet associated the geometrical characteristics of the contact and is proportional to the
with the fast rotation of the wafer does not participate by itself in the particle radius, Psat is the saturated water pressure, Pv the water vapor
cleaning process. pressure and τ0 order of time needed to condense one liquid layer.
This time dependence of the adhesion force can be explained consider-
3.3. Impact of the polymer removal method on the particle removal ing either that the contact area of the particles with the surface increases
efficiency over time due to the plastic deformation of the beads or the formation of
silica bridges between the surface and the particle due to water action.
When comparing Fig. 2b) and d), we observe the following features: Eq. (2), which has been used successfully in several different experi-
first, when the polymer is chemically removed using the wet bench mental configurations, suggests that the longer the time between con-
equipment, so without any physical action, no particles are removed tamination and cleaning, the more firmly attached the particles and
from the surface. Then, when the polymer is removed thanks to a the less efficient the cleaning process. In addition, the adhesion force
spray batch method which combines chemical and physical actions of eq. (2) depends on the particle: the higher the adhesion energy (via
(via the spray dispense and the wafer rotation) we obtain that 87% of γ), the greater the adhesion force.
the particles initially present on the surface are withdrawn. Another ob- Fig. 3 suggests, however, that PRE is neither dependent of aging time
servation is that the particle removal is not uniform. Denser zones are nor of the particle nature despite the fact that the adhesion energy on
observed near the central spray dispenser and along straight lines. To SiO2 of Si3N4 particles is higher than the one of SiO2 particles.
understand the origin of these straight lines, we determine their inter- To understand these two surprising features, we assume that the
section point from images such as the one of Fig. 2d) and find that particles are bounded to the polymer film due to the large surface
they intersect at the position of the central spray dispenser which is they exhibit at their contact. Thus, when the film detaches from the sur-
also the rotation center of the wafer. This suggests that the rotation of face, the particles follow it and also detaches from the surface. This
the wafer contributes to the particle removal through centrifugal forces. mechanism explains why the PRE we measure is neither influenced by
the aging time nor by the particle size or nature. To understand under
3.4. Performances of the process, aging time and contamination influence which conditions the film can detach from the surface, we use the the-
on the particle removal efficiency oretical framework used to describe rupture and delamination of thin
films deposited over a solid surface. When a thin film (such as our poly-
In the classical cleaning methods, the process performances can vary mer film) deposited over a solid substrate is put into tension, the gen-
with different parameters such as the aging time between the contam- eral principle is that the crack propagates if the elastic energy released
ination and the cleaning process [8,9] or like the contamination nature during the propagation exceeds the crack energy. However, in that
[8]. Thus, to gain quantitative comparison between the different pro- case, the film fracture and delaminates in a limited volume defined by
cesses, we measure the PRE for various aging time, in Fig. 3. As can be Lh2 where L is the length of the fracture and h the polymer film (in
seen, the particle removal efficiency is constant around 85%, for 3 the order of 2 μm) [11]. This would only induce delamination in a very

Fig. 3. Influence of the aging time on the particle removal efficiency.

Please cite this article as: A. Lallart, et al., Cleaning surfaces from nanoparticles with polymer film: impact of the polymer stripping, (2018),
https://doi.org/10.1016/j.mne.2018.09.001
4 A. Lallart et al. / Micro and Nano Engineering xxx (xxxx) xxx–xxx

This method is still under study and need improvement. Indeed, new
chemical solutions for polymer removal are required for its compatibil-
ity with all kind of substrates. The polymer and the chemical solution
used for its removal need to be environmental friendly and finally the
efficiency of this method should be demonstrated at an industrial scale.

Conflict of Interest

I confirm that there are no known conflicts of interest associated

with this publication

Acknowledgement

The experiments were performed in the frame of the joint develop-

ment program with STMicroelectronics, “Laboratoire Interdisciplinaire
de Physique (LIPhy)”, “Laboratoire des Ecoulements Géophysique et
Industriels (LEGI)” and the “CEA LETI”.
Fig. 4. Influence of the particle nature and size on the removal efficiency.
References
limited volume. However, Marthelot et al. [12] recently highlighted a
[1] F. Tardif, A. Danel, O. Raccurt, Understanding of wet and alternative particle removal
novel fracture mechanism of thin films where delamination and propa- processes in microelectronics: theoretical capabilities and limitations, J.
gation occur simultaneously. In that case, the mechanism is active Telecommun. Inf. Technol. 1 (2005) 8.
below the standard critical tensile load for channel cracks and selects [2] W. Kim, T.H. Kim, J. Choi, H.-Y. Kim, Mechanism of particle removal by megasonic
waves, Appl. Phys. Lett. 94 (2009) 081908.
a robust interaction length scale on the order of 30 times the film, [3] J. T. Snow, K. Miya, M. Sato, M. Kato, and T. Tanaka, “Advances in Particle Removal
which spontaneously replicates and propagates. without Damage,” 2015.
In our case, the centrifugal force and the additional stress to which [4] N. Narayanswami, R. Analytics, K.C. Optomec, S. Carolina, Evaluation of Particle Re-
moval Efficiency in Wafer cleaning Processes, Semicond. Int. Mag. (January) (2000).
the polymer layer is subjected due to solvent addition, would be suffi- [5] P. Lin, S. Pi, H. Jiang, Q. Xia, Mold cleaning with polydimethylsiloxane ( PDMS ) for
cient to induce this spontaneous delamination. The inhomogeneity in nanoimprint lithography, Nanotechnology 24 (32) (2013).
PRE observed in 2d) also lies in this picture. The polymer located far [6] H.F. Okorn-Schmidt, et al., Particle cleaning Technologies to Meet Advanced Semi-
conductor device Process Requirements, ECS J. Solid State Sci. Technol. 3 (1)
from the point around which the polymer rotates will be subjected to (2013) N3069–N3080.
a higher tension due to centrifugal forces and thus more prone to [7] T. W. Walker and C. W. Frank, “Enhanced particle removal using viscoelastic fluids,”
crack and delaminate. no. May 2015, 2013.
[8] C.-C. Chiang, B. Wu, S. Raghavan, Particle Deposition and Removal of Relevance to
Wet Processing in Semiconductor Manufacturing, Part. Sci. Technol. 33 (5) (2015)
5. Conclusion 150706133355006.
[9] J. Tang, A.A. Busnaina, The effect of Time and Humidity on Particle Adhesion and Re-
moval, J. Adhes. 74 (2000) 411–419.
Miniaturization in the microelectronic field lead to new challenges
[10] J. Bocquet, L. Charlaix, E. Ciliberto, C. Crassous, Moisture-induced ageing in granular
concerning particles removal. The cleaning method based on the use media and the kinetics of capillary condensation, Nature 396 (1998) 735–737.
of a polymer coating and removal open new trails with high removal ef- [11] J. Marthelot, Rupture et délamination de films minces, Université Pierre et Marie
ficiency (87%) whatever the aging time, particle nature or size. This Curie - Paris VI, 2014.
[12] J. Marthelot, B. Roman, J. Bico, J. Teisseire, D. Dalmas, F. Melo, Self-replicating cracks:
method, which combines a physical and a chemical action to remove a collaborative fracture mode in thin films, Phys. Rev. Lett. 113 (8) (2014) 22–26.
nanoparticles is thus very efficient, yet the removal is still non-uniform.

Please cite this article as: A. Lallart, et al., Cleaning surfaces from nanoparticles with polymer film: impact of the polymer stripping, (2018),
https://doi.org/10.1016/j.mne.2018.09.001