Extended Abstracts of the 2013 International Conference on Solid State Devices and Materials, Fukuoka, 2013, pp1006-1007
K-3-3
GeSn Metal-Semiconductor-Metal Photodetectors with Suppressed Dark
Current by Ammonium Sulfide Surface Passivation
Yuan Dong,1 Wei Wang,1 Xin Xu,1 Xiao Gong,1 Pengfei Guo,1 Qian Zhou,1 Lanxiang Wang,1 Genquan Han,1 Zhe Xu,2
Soon-Fatt Yoon,2 and Gengchiau Liang1, Yee-Chia Yeo.1, *
1
I.
Department of Electrical and Computer Engineering, National University of Singapore, Singapore 117576.
2
School of Electrical and Electronic Engineering, Nanyang Technological University, Singapore 639798.
*Phone: +65 6516-2298, Fax: +65 6779-1103, E-mail: [email protected]
INTRODUCTION
Optical communication based on silicon photonics requires the
development of high performance near-infrared (NIR) photodetectors.
Germanium-tin (GeSn), as one of the group IV semiconductor alloys,
is attracting great interest for applications in NIR photodetectors
which can cover all the optical communication bands [1]-[4].
Among all the photodetector structures, metal-semiconductor-metal
(MSM) photodetectors exhibit the advantages of the highest response
speed and ease of integration [5]. However, MSM photodetectors
generally suffer from large dark current, which greatly limits its applications. Therefore, it is important to explore techniques to suppress the dark current of MSM photodetectors.
In this paper, we report, for the first time, the dark current suppression of GeSn MSM photodetectors by a (NH4)2S surface passivation. Reduction of dark current as large as 67.2% was achieved
with (NH4)2S passivation as compared with that of the unpassivated
ones, which can be attributed to the -reduction of surface states by
(NH4)2S surface passivation. Responsivity of 181 mA/W at bias
voltage of 3 V was also achieved at the wavelength of 1550 nm.
II.
MATERIAL CHARACTERIZATION
Epitaxial GeSn films were grown on Ge(100) substrate using
molecular beam epitaxy (MBE) at 170 °C. Cross-sectional TEM
image of a 780 nm-thick GeSn film on Ge(100) substrate is shown in
Fig. 1(a). High resolution TEM (HRTEM) image in Fig. 1(b) shows
the single-crystalline GeSn film with high crystalline quality. The
surface region and interface region of the GeSn film are shown in Fig.
1(c) and (d), respectively. High resolution XRD (HRXRD) (004) and
(224) curves in Fig. 2(a) and (c) show that the as-grown GeSn film
has a substitutional Sn composition of 2.6%. Fig. 2(b) and (d) are
(004) and (224) curves of the GeSn film after 400 °C annealing for 30
minutes in N2 ambient. Negligible change of substitutional Sn
composition was observed after annealing, indicating good thermal
stability of the as-grown GeSn film. AFM images in Fig. 3(a) and
(b) show the RMS surface roughness of the as-grown GeSn film and
the one after annealing, respectively. Surface smoothness of the
GeSn film was improved by annealing.
III.
DEVICE FABRICATION AND CHARACTERIZATION
GeSn MSM photodetector along the dash line A-A’ shown in Fig. 4(b).
Electrode width (w), as defined in Fig. 4(c), was designed to be equal
to electrode spacing (s) with the values ranging from 6 to 10 μm.
B. Electrical and Optical Characteristics
I-V characteristics of GeSn MSM photodetectors with and
without (NH4)2S surface passivation are shown in Fig. 5. Electrode
spacings of the GeSn MSM photodetectors in Fig. 5(a), (b), and (c)
are 10, 8, and 6 μm, respectively. At a bias voltage of 3 V, the dark
currents of (NH4)2S-passivated photodetectors are 67.2%, 50%, and
39% smaller than those of the unpassivated ones, as shown in Fig.
5(a), (b), and (c), respectively. This can be attributed to the reduction
of surface states by (NH4)2S passivation of the GeSn surface.
The photocurrent of a photodetector is defined as the difference
between the total current under illumination and its dark current. Fig.
6 shows the photocurrent of the (NH4)2S-passivated GeSn MSM photodetectors as a function of bias voltages at different laser powers.
Fig. 7 shows the photocurrent of the GeSn MSM photodetectors at
different laser powers with and without (NH4)2S surface passivation.
Difference in photocurrent with and without passivation is very small,
which indicates that (NH4)2S surface passivation has negligible influence on photocurrent of the GeSn MSM photodetectors.
It is noted that the dark current is still relatively high after surface passivation, which degrades the signal-to-noise ratio (SNR).
This is due to Fermi level pinning near the valence band of GeSn,
resulting in a small hole Schottkey barrier height (SBH) at the NiGeSn/GeSn interface and leading to large hole leakage current. This
dark current cannot be suppressed by surface passivation. Therefore,
our further investigations will be focused on combination of surface
passivation, which reported in this work, with techniques that can
increase the hole SBH to improve the SNR.
Fig. 8(a) shows photocurrent of the (NH4)2S-passivated GeSn
MSM photodetectors as a function of bias voltage at different light
wavelengths. The wavelengths are 1550, 1580, 1610, and 1630 nm,
and the laser power is 1 mW. Fig. 8(b) shows the responsivity-wavelength characteristics of the (NH4)2S-passivated GeSn MSM
photodetectors at a bias voltage of 3 V. A responsivity of 181 mA/W
was achieved at the wavelength of 1550 nm.
A. Device Fabrication
IV.
The process flow for fabricating GeSn MSM photodetectors is
shown in Fig. 4(a). Mesa structure was first formed by wet etch after
the epitaxial growth of GeSn film. One sample was immersed in a
(NH4)2S (24%) solution for 1 hour at room temperature for surface
passivation. For the control sample, this passivation step was
skipped. A 100-nm thick SiO2 layer was then deposited for both of
the two samples by sputtering. The inner and outer electrode contact
regions were then patterned and opened by wet etch. Right after this,
15 nm-thick Ni was deposited and followed by rapid thermal annealing (RTA) at 350 °C for 30 s to form the NiGeSn metallic contacts.
The unreacted Ni was then removed by concentrated sulfuric acid.
Finally, Al electrodes were formed by electron-beam evaporator
(E-beam) deposition and lift-off.
We demonstrated, for the first time, the dark current suppression of GeSn MSM photodetectors by (NH4)2S surface passivation.
Maximum suppression of 67.2% was achieved with passivation. The
large dark current suppression can be attributed to the reduction of
surface states by (NH4)2S surface passivation. Responsivity of 181
mA/W was achieved at the wavelength of 1550 nm.
Top-view SEM image of the GeSn MSM photodetector is depicted in Fig. 4(b). Fig. 4(c) shows the cross-sectional schematic of
CONCLUSION
ACKNOWLEDGEMENT. The authors acknowledge support from
the National Research Foundation under Grant NRF-CRP6-2010-4.
REFERENCES
[1] S. J. Su et al., Optics Express 19, 6400, 2011.
[2] J. Werner et al., APL 98, 061108, 2011.
[3] J. Mathews et al., APL 95, 133506, 2009.
[4] M. Oehme et al., APL 101, 141110, 2012.
[5] Y. H. Chen et al., IEEE EDL 28, 1111, 2007.
- 1006 -
-0.4
(b)
(d)
GeSn
(a) Key Process Steps
A
A’
(NH4)2S Surface Passivation
SiO2 Deposition and
Contact Region Opening
Ni deposition (~ 15 nm)
Ni(GeSn) Metallic
Contact Formation
RTA: 350 ºC, 30 s
(c)
Light
(NH4)2S
Passivation
A
A’
10-4
10-5
10-6
Photocurrent Iphoto (mA)
Photocurrent Iphoto (A)
10-4
λ = 1550 nm
1
2
With
0.4
3
Fig. 6. Photocurrent Iphoto of the
(NH4)2S-passivated GeSn MSM
photodetectors as a function of the
bias voltage at different laser power
values.
0.0
1
-2
Without
passivation 39%
-3
-3
10
10-4
With
passivation
10-5
Without
passivation
1
2
10
10-4
10-5
With
passivation
10-6 0
3
1
2
3
Bias Voltage (V)
Bias Voltage (V)
(a)
0.6 passivation
0.2
2
(c) 10
50%
10-6 0
3
3
2
Fig. 5. I-V characteristics of the GeSn MSM photodetectors with (a) w = s = 10 μm, (b) w
= s = 8 μm, and (c) w = s = 6 μm with and without (NH4)2S surface passivation. At a bias
voltage of 3 V, the dark currents of (NH4)2S-passivated photodetectors are 67.2%, 50%,
and 39% smaller than those of the unpassivated ones, as shown in (a), (b), and (c),
respectively.
Without
passivation
10-4
10
(b) 200
150
100
50
P = 1 mW
0
1 2 3 4 5
Laser Power (mW)
Fig. 7. Photocurrent Iphoto of the
GeSn MSM photodetectors with
and without (NH4)2S surface
passivation at different laser
powers. Negligible change of
photocurrent was observed.
1550 nm
1580 nm
1610 nm
1630 nm
-5
λ = 1550 nm
-5
1
2
Bias Voltage (V)
0
0.8
1 mW
2 mW
3 mW
4 mW
5 mW
0
Without
passivation
Bias Voltage (V)
Fig. 4. (a) Key process steps for fabricating the
GeSn MSM photodetector with (NH4)2S surface
passivation. (b) Top-view SEM image of the GeSn
MSM photodetector. (c) Cross-sectional schematic
of the GeSn MSM photodetector along line A-A’
shown in (b). Electrode width (w) and spacing (s)
between electrodes are defined in this figure.
10
With
passivation
5
4 μm
1
0
0
Fig. 3. RMS surface roughness of the
(a) as-grown GeSn film and (b) the one
after 400 ºC RTA for 30 minutes. RTA
improves the surface smoothness of the
GeSn film.
-2
10
Al Electrode Deposition
and Lift Off
0 0
RMS : 2.01 nm
3
0.2
(b) 10
67.2%
-3
Selective Removal of Ni
using H2SO4
10-3
0.0
∆θ (degree)
-2
(a) 10
Dark Current (A)
Mesa Formation by
Patterning and Etching
-0.2
20 nm
10
0
5
μm 4
Fig. 2. HRXRD (a) (004) and (c) (224) curves
show that the as-grown GeSn film has a
substitutional Sn composition of 2.6%.
Negligible change of the substitutional Sn
composition was observed after 400 ºC RTA
for 30 minutes, as shown in HRXRD (b) (004)
and (d) (224) curves.
Epitaxial Growth of
Ge0.974Sn0.026 Layer by MBE
1
(b)
(224)
400ºC
-0.4
15 μm
(b)
1
2
400 ºC Annealed
Ge (100)
Fig. 1. (a) Cross-sectional TEM image of an
epitaxial GeSn film on Ge(100) substrate. HRTEM
image of the (b) bulk region, (c) top surface region,
and (d) GeSn/Ge interface region of the GeSn film.
3
2
3
(224)
5
4 μm
RMS : 2.51 nm
As-grown
(c)
GeSn
2 nm
20 nm
10
0
5
μm 4
(004)
(b)
(d)
3 nm
As-grown
Dark Current (A)
5 nm
Ge (100)
Epoxy
(a)
400ºC
Photocurrent Iphoto (A)
(c)
0.2
(004)
(a)
Intensity (a. u.)
GeSn
GeSn
250 nm
0.0
As-grown
Responsivity (mA/W)
~780 nm
-0.2
Dark Current (A)
(a)
1
2
3
Bias Voltage (V)
P = 1 mW
0
1550 1575 1600 1625
Wavelength (nm)
Fig. 8. (a) Photocurrent Iphoto of the (NH4)2S-passivated GeSn MSM
photodetectors as a function of the bias voltage at different light
wavelengths, measured at a laser power of 1 mW. (b) Responsivitywavelength characteristics of the photodetectors at a bias voltage of 3
V.
- 1007 -