HAL Id: hal-02275699
https://hal.archives-ouvertes.fr/hal-02275699
Submitted on 1 Sep 2019
HAL is a multi-disciplinary open access
archive for the deposit and dissemination of sci-
entic research documents, whether they are pub-
lished or not. The documents may come from
teaching and research institutions in France or
abroad, or from public or private research centers.
L’archive ouverte pluridisciplinaire HAL, est
destinée au dépôt et à la diusion de documents
scientiques de niveau recherche, publiés ou non,
émanant des établissements d’enseignement et de
recherche français ou étrangers, des laboratoires
publics ou privés.
4H-SiC P+N UV Photodiodes : A Comparison between
Beam and Plasma Doping Processes
Stéphane Biondo, Laurent Ottaviani, Mihai Lazar, Dominique Planson, Julian
Duchaine, V. Le Borgne, M A El Khakani, Frédéric Milesi, Wilfried Vervisch,
Olivier Palais, et al.
To cite this version:
Stéphane Biondo, Laurent Ottaviani, Mihai Lazar, Dominique Planson, Julian Duchaine, et
al.. 4H-SiC P+N UV Photodiodes : A Comparison between Beam and Plasma Doping Pro-
cesses. Materials Science Forum, Trans Tech Publications Inc., 2012, 717-720, pp.1203-1206.
�10.4028/www.scientic.net/MSF.717-720.1203�. �hal-02275699�
4H-SiC P
+
N UV Photodiodes : a Comparison between Beam and Plasma
Doping Processes
S. Biondo
1, a
, L. Ottaviani
1,b
, M. Lazar
2,c
, D. Planson
2,d
, J. Duchaine
3,e
, V. Le
Borgne
4,f
, M. A. El Khakani
4,g
, F. Milesi
5,h
,
W. Vervisch
1,i
, O. Palais
1,j
and F.
Torregrosa
3,k
1
IM2NP (UMR 6242) – Université Paul Cézanne, Case 231, 13397 Marseille Cédex 20, France
2
AMPERE (UMR 5005) – INSA de Lyon, 21 Av. Capelle, 69621 Villeurbanne, France
3
Ion Beam Services, Rue Gaston Imbert Prolongée, 13790 Peynier, France
4
INRS-EMT, 1650 Boulevard Lionel-Boulet, Varennes, Quebec, Canada J3X 1S2
5
CEA LETI/MINATEC, 17 rue des Martyrs, 38054 Grenoble Cedex 9, France
a
stephane.biondo@im2np.fr,
b
laurent.ottaviani@im2np.fr,
c
mihai.lazar@insa-lyon.fr,
d
dominique.planson@insa-lyon.fr,
e
julian.duchaine@ ion-beam-services.fr,
f
leborgne@emt.inrs.ca,
g
elkhakani@emt.inrs.ca,
h
frederic.milesi@cea.fr,
i
wilfried.vervisch@im2np.fr,
j
olivier.palais@im2np.fr,
k
frank.torregrosa@ion-beam-services.fr
Keywords: UV Photodiode – plasma implantation – FDTD – dark current - sensitivity
Abstract. This paper presents a study of 4H-SiC UV photodetectors based on p
+
n thin junctions.
Two kinds of p
+
layers have been implemented, aiming at studying the influence of the junction
elaborated by the ion implantation process (and the subsequent annealing) on the device
characteristics. Aluminum and Boron dopants have been introduced by beam line and by plasma ion
implantation, respectively. Dark currents are lower with Al-implanted diodes (2 pA/cm
2
@ - 5 V).
Accordingly to simulation results concerning the influence of the junction thickness and doping,
plasma B-implanted diodes give rise to the best sensitivity values (1.5x10
-1
A/W @ 330 nm).
Introduction
During the past years there has been considerable interest in systems able to record very low light
levels in the ultraviolet range in severe conditions of use. The advantage of Silicon Carbide (SiC)
with respect to nitride alloys – the major wide band-gap semiconductor used today in industry –
relies on three major points : a low residual doping for epitaxial layers (in the 10
14
cm
-3
range and
concentrations of residual defects/impurities at least one order of magnitude lower), a high thermal
conductivity allowing high temperature operations, and a very good radiation hardness. It is then
possible to use SiC for fabrication of devices capable to operate under extreme conditions.
Photodetectors based on SiC allow to obtain good wavelength selectivity in the UV range, without
any optical filters.
Experimental
The role of the p
+
emitter layer properties has been particularly studied in this paper. Among these
properties, the doping and the thickness are thoroughly key parameters for controlling the device
reliability. Photodetector simulations based on finite element method were performed, optimizing
the design of the thin junctions for improvement of the light absorption and the carrier harvest. We
also investigated the technological process giving rise to the dopant introduction into the SiC
matrix. The comparison between standard ion implantation and pulsed-Plasma Immersion Ion
Implantation (PIII) processes is expected to be fruitful, since PIII technology produced impressive
results for Si solar cells in the UV range [1]. To our knowledge, PIII doping has never been carried
out in SiC material. 4H-SiC n-type epilayers were either implanted with Aluminum by standard ion
implantation at 27 keV, or with Boron by PULSION
TM
system (pulsed-plasma ion immersion) –
B
2
H
6
at 8 kV, in order to produce p
+
-type layer thicknesses of 30 and 10 nm, respectively. The
doses were adjusted for obtaining peak concentrations of few 10
19
cm
-3
for Al (samples A) and few
10
20
cm
-3
for B (samples B). This concentration discrepancy takes into account the difference of
ionisation energies between Al and B dopants, and should give rise to similar values of the final
hole concentrations in p
+
layers. Each sample was then annealed at 1700°C (samples A1, B1) or at
1650°C (samples A2, B2), aiming at analysing the influence of the annealing temperature on the
device characteristics.
A prototype of furnace was used during this work (purchased from VEGATEC
TM
), consisting in
a vertical resistive reactor allied with a lift system. This allows to perfectly control the heating-up
and the cooling-down rates, up to ~ 20°C/s. After Al implantations, we observed that a high
heating-rate improved the sheet resistance whatever the annealing temperature, and preserved the
surface roughness for annealing temperatures lower than 1700°C, which is crucial for thin
implanted layers. The heating rate has indeed proven to be an important parameter for controlling
the reverse current of the related diodes [2].
After thermal annealing, ohmic contacts were realized by sputtering with Ti/Al/Ni on p-type
implanted layer (top contact) and Ni on n
+
-type substrate layer (bottom contact). The back contact
on the substrate was annealed at 900°C and the contact on implanted layer was annealed at 800°C.
Both contacts have been annealed during 2 min under Argon atmosphere. Finally, the UV-
photodetector surface shape has a circle geometry with 250 µm-diameter. A window area allows to
detect the UV photons. Fig.1 shows the photodetector structure.
The optical simulations of photodetectors under the UV light have been realised by FDTD
method (Finite Difference Time Domain), using the commercial software Sentaurus edited by
Synopsys society [3]. Electromagnetic solver based on the FDTD method is used to calculate the
electromagnetic field propagation inside UV-photodetector device.
Plasma Implantation in SiC
We propose to study the combination of PIII with a proper annealing, which should results in thin
p
+
implanted layers (lower than 30 nm) particularly suitable for UV photon detection. PIII were
performed on PULSION
TM
(Plasma ion implantation tool from the french company I.B.S.) using
B
2
H
6
gas (see Fig.2). Specificity of PULSION
TM
consists in using a pulsed DC polarization and a
remote ICP plasma source allowing to work at low pressure (< 1x10
-3
mbar) with the use of low gas
flow rate (< 10 sccm). This helps to minimize parasitic etching or deposition usually encountered
on Plasma doping tools.
Ohmic
contact
SiC epi
Substrate n
+
- Thickness @ 350 µm
- Concentration @ 10
18
cm
-3
Epitaxial layer n
-
- thickness @ 10 µm
- Concentration @ 5x10
15
cm
-3
Implantated layer
Fig.1 Photodetector pn structure
Fig.2 PULSION
TM
set-up
A former study proved that, at a given energy, the plasma-process leads to a better surface
morphology, a lower defect concentration and a thinner junction than a standard beam implantation
process. This is accompanied with some dopant outdiffusion during the annealing, and a higher
sheet resistance of the implanted layer [4].
Results and Discussion
Simulation
Figures 3 display the variation of the current density with the reverse bias (for an incident light
wavelength at 200 nm), varying the p
+
-layer thickness (Fig. 3a) and the p
+
-layer concentration (Fig.
3b). In a general way, the current density increases with a thinner junction and a lower hole
concentration. When the space charge region is closer to the surface, much more carriers
undergoing the electric field are then harvested, leading to a better UV photodetector response.
Fig.3a Simulated current density vs reverse
bias @ 200 nm, with a p
+
doping fixed at
5x10
19
cm
-3
Fig.3b Simulated current density vs reverse
bias @ 200 nm, with a p
+
thickness fixed at
30 nm
Device Characteristics
The evolution of dark currents with reverse bias of the realised devices is shown in Fig.4. Dark
currents reveal to be lower with Al-implanted diodes (2 pA/cm
2
@ - 5 V), whatever the annealing
temperature. On the contrary, B-implanted diodes show higher forward currents than Al-implanted
diodes (not shown here), revealing a “JBS-behaviour” due to in-diffusion of B atoms in the ternary
compound formed within the top metal during the annealing at 800°C (see Ref. 5 for details). The
SIMS profile of B atoms shows no diffusion during the annealing [5].
0 2 4 6 8 10 12 1 4 16 18 2 0
1E -12
1E -11
1E -10
Current Density (A.cm
-2
)
R ev erse b ia s (V)
B1
B2
Al1
Al2
Fig. 4 Dark currents of Al- and B- implanted
photodiodes
0 2 4 6 8 10 12 14 16 18 20
10
4
10
5
Ratio
Reverse bias (V)
Al
B
F
Fig.5 Ratio between UV(365 nm) and dark
current for Al- and B- implanted photodiodes
(annealed at 1700°C)
0 2 4 6 8 10 12 14 16 18 20
10
-10
10
-9
10
-8
Current density (A.cm
-2
)
Reverse bias (V)
thickness @ 0.01 µm
thickness @ 0.03 µm
thickness @ 0.05 µm
thickness @ 0.10 µm
thickness @ 0.30 µm
thickness @ 0.40 µm
thickness @ 0.50 µm
0 2 4 6 8 10 12 14 16 18 20
6x10
-9
7x10
-9
8x10
-9
9x10
-9
10
-8
1.1x10
-8
1.2x10
-8
Current density (A.cm
-2
)
Reverse bias (V)
p
+
concentration @ 5x10
17
cm
-3
p
+
concentration @ 5x10
18
cm
-3
p
+
concentration @ 5x10
19
cm
-3
p
+
concentration @ 5x10
20
cm
-3
Characteristics of the diodes have been then measured under light, with an incident wavelength
of 365 nm. Fig.5 displays the evolution of the “signal-to-noise” ratio with reverse bias. Al-
implanted photodiodes reveal a ratio six times higher than B-implanted diodes. Fig.6 gives the
spectral response of these A1 photodiodes. As seen in Fig.5, there is no influence of the reverse
bias, which is a clear advantage if a fully autonomous system is required (for space applications).
The spectral responsivities of the four kinds of diodes are compared in Fig.7. For a given dopant,
the sensivity increases with the annealing temperature, which is surely related to a better
recombination of the defects produced by the implantation process [6]. For a given annealing
temperature, B-implanted diodes give rise to a higher signal than Al-implanted diodes. This can be
due to a thinner junction and/or a lower hole concentration in the p
+
-layer, which lead to increase
the current density of the device under light.
200 220 240 260 280 300 320 340 360 380 400
10
-6
10
-5
10
-4
10
-3
10
-2
10
-1
-10V
0V
Responsivity (A/W)
Wavelength (nm)
Fig.6 Spectral responsivities of Al-
implanted photodetector (the power source
light is shown in the inset)
Fig.7 Spectral responsivities of Al- and B-
implanted photodetectors
Summary
4H-SiC UV photodetectors were realised, based on implanted p
+
n junctions either by Al standard
beam or by B plasma. Thanks to the optimised furnace for post-implantation annealings, the leakage
current of the diodes remain as low as 2 pA/cm
2
. Boron plasma-implanted devices give rise to the
best spectral responsivities. The behaviour of the diodes after irradiations is currently under study.
Acknowledgement
The authors wish to thank Daniel Ehret (VEGATEC) for technical support, DGE (Direction
Générale des Entreprises) for SiC-HT
2
project and ARCSIS for financial support.
References
[1] V. Vervisch, H. Etienne, F. Torregrosa, L. Ottaviani, M. Pasquinelli, T. Sarnet, P. Delaporte, J.
Vac. Sc. Tech. B26 (2008) 286.
[2] R. Nipoti, A. Carnera, F. Bergamini, M. Canino, A. Poggi, S. Solmi, M. Passini, Mater. Res.
Soc. Symp. Proc. B11-01 (2006) 911.
[3] Information on http://www.synopsys.com
[4] L. Ottaviani, S. Biondo, R. Daineche, O. Palais, F. Milesi, J. Duchaine, F. Torregrosa, AIP
Conf. Proc. 1321 (2010) 245.
[5] S. Biondo, M. Lazar, L. Ottaviani, W. Vervisch, O. Palais, R. Daineche, D. Planson, F. Milesi, J.
Duchaine, F. Torregrosa, Mater. Sci. For. 711 (2012) 114.
[6] M. Rambach, A. Bauer, H. Ryssel, phys. stat. sol. (B) 245 (2008) 1315.
200 250 300 350 400
0.1
1
10
Power (µW)
Wavelength (nm)
250 300 350 400
10
-4
10
-3
10
-2
10
-1
Responsivity (A/W)
Wavelength (nm)
QA1
QA2
QB1
QB2
