10 μm Pixel-to-Pixel Pitch Hybrid Backside Illuminated AlGaN-on-Si Imagers for
Solar Blind EUV Radiation Detection
Pawel E. Malinowski
*,1,2
, Jean Yves Duboz
3
, Piet De Moor
1
, Joachim John
1
, Kyriaki Minoglou
1
, Puneet
Srivastava
1,2
, Ybe Creten
1
, Tom Torfs
1
, Jan Putzeys
1
, Fabrice Semond
3
, Eric Frayssinet
3
, Boris
Giordanengo
4
, Ali BenMoussa
4
, Jean Francois Hochedez
4
, Robert Mertens
1,2
, and Chris Van Hoof
1,2
1
IMEC, Kapeldreef 75, Leuven, Belgium
2
ESAT, Katholieke Universiteit Leuven, Leuven, Belgium
3
CRHEA/CNRS, Valbonne, France
4
Royal Observatory of Belgium, Brussels, Belgium
*e-mail: Pawel.Malinowski@imec.be
Abstract
We present the first measurement results from hybrid
AlGaN-on-Si-based Extreme Ultraviolet (EUV) imagers with
10 μm pixel-to-pixel pitch. The 256x256 backside
illuminated Focal Plane Arrays (FPAs) were hybridized to
dedicated Si-based CMOS Readouts (ROICs). The AlGaN
active layer with 40% Al concentration provides an intrinsic
rejection of wavelengths larger than 280 nm (solar
blindness), together with enhanced radiation hardness (1).
Sensitivity in Deep UV (DUV), Far UV (FUV) and Extreme
UV (EUV) was verified using synchrotron radiation down to
a wavelength of 1 nm.
Introduction
Ultraviolet detection is of particular interest for solar science,
EUV microscopy and modern EUV lithography tools (2).
Having 2D imaging capability with wide bandgap
semiconductors such as Aluminum Gallium Nitride (AlGaN)
provides an interesting alternative to the currently used Si
electronics, prone to radiation damage and requiring filters
for blocking the unnecessary visible and infrared radiation
(3). Backside illumination in a hybrid design eliminates
absorption in metal contacts shadowing the active layer (4).
Using Si as a substrate for AlGaN growth enables increasing
the wafer size to 200 mm.
Fabrication
AlGaN layers were grown on Si(111) wafers using Molecular
Beam Epitaxy. Two layer types were processed, with an n-
doped layer requiring MESA etching (Layer 1, Fig. 1a) or
without MESA (Layer 2, Fig. 1b). The latter was designed
for a backside ohmic contact, producing a vertical Schottky
diode. The growth started with a 40 nm AlN nucleation layer
and was followed by the undoped active layer. Al
concentration was decreasing towards the surface, reaching
40%. For the Layer 1, the n-doping was introduced for the
top 100 nm. A total thickness of the AlGaN layer was 440
nm for Layer 1 and 340 nm for Layer 2. For the Layer 1, a
Cl
2
-based Reactive Ion Etching (RIE) was used to access the
undoped layer, with the resulting MESA depth of 200 nm.
The FPA pixels are 4 μm Schottky contacts and were
deposited by sputtering 20 nm of Au (Fig. 2a). The ohmic
grid (common contact surrounding the array) consisted of a
Ti/Al/Mo/Au stack, annealed for 1 minute at 850°C. After
that, 100 nm of SiO
2
was sputtered to act as an isolation
layer. It was opened on top of the pixel contacts with SF
6
-
based RIE. Furthermore, in the post-processing step, arrays
of 256x256 In bumps with 10 μm pixel-to-pixel pitch and
height of 2 μm were deposited by evaporation on top of the
AlGaN arrays (Fig. 2b).
Figure 1 – AlGaN-on-Si active layers in two versions: a) with n-doped layer
for the ohmic contact and requiring MESA etching; and b) simpler, with no
MESA etching required. The active layer is AlGaN with a 40% Al
concentration. Both layers were grown on Si(111).
Figure 2 – SEM micrographs of AlGaN Schottky-photodiode-based pixels
with 10 μm pitch: a) front side view before post-processing, showing the
MESA structures and the ohmic grid; and b) In bumps deposited on each
pixel, with a height of approximately 2 μm.
undoped Al
0.4
Ga
0.6
N
undoped AlN
300 nm
AlN 40 nm
Si(111) 300 µm
b) Layer 2
no MESA
a) Layer 1
Ω contact + MESA
AlN 40 nm
Si(111) 300 µm
n-Al
0.4
Ga
0.6
N MESA
200 nm
200 nm undoped AlN
b) a)
14.5.1 IEDM10-348 978-1-4244-7419-6/10/$26.00 ©2010 IEEE