Light emission enhancement by geometrical scaling of carrier injectors in Si-based LEDs

Abstract

In this paper we present the increased light emission for Si p-i-n light emitting diodes (LED) by geometrical scaling of the injector size for p- and n-type carriers. TCAD simulations and electrical and optical characteristics of our realized devices support our findings. Reducing the injector size decreases the diffusion current: therefore, for a particular on current, the pn-product, and hence the radiative recombination, inside the active region increases. A comparison is made among reference large-scale, micro-size and nano-size injector p-i-n diodes. We demonstrate a 4-fold increase in electroluminescence (EL) when the injectors are scaled down to micro-size and a further 10-fold increase for nano-size injectors.

Full text

Page 1

Light emission enhancement by geometrical scaling
of carrier injectors in Si-based LEDs
G. Piccolo1,*, V. Puliyankot1, A.Y. Kovalgin1, R.J.E. Hueting1, A. Heringa2, and J. Schmitz1
1MESA+Institute for Nanotechnology, University of Twente, Enschede, The Netherlands
2NXP Research, Eindhoven, The Netherlands
*g.piccolo@ewi.utwente.nl
Abstract—In this paper we present the increased light emission
for Si p-i-n light emitting diodes (LED) by geometrical scaling of
the injector size for p- and n- type carriers. TCAD simulations
and electrical and optical characteristics of our realized devices
support our findings. Reducing the injector size decreases the
diffusion current: therefore, for a particular on current, the
pn-product, and hence the radiative recombination, inside the
active region increases. A comparison is made among reference
large-scale, micro-size and nano-size injector p-i-n diodes. We
demonstrate a 4-fold increase in electroluminescence (EL) when
the injectors are scaled down to micro-size and a further 10-fold
increase for nano-size injectors.
Index Terms—LEDs, silicon photonics, p-i-n diode, carrier
injector, antifuse, electroluminescence.
I. INTRODUCTION
In recent years, there is much interest to merge electronics
and photonics in an integrated circuit (IC) in a silicon platform
[1]–[3]. Silicon wave guides, detectors and modulators have
already been successfully built [4]. Many efforts have also
been made to realize efficient light emitting sources in silicon,
despite the indirect bandgap of the material: the attempts
include chemical or structural modifications of the material
[5], [6], confinement of the carriers [7] and optimization of
the device structure using high quality silicon [8]–[10].
In this paper, on the same line of the optimization of simple
structures, we demonstrate the enhancement of light emission
from Si p-i-n LEDs by scaling down the injector size for p-
and n-type carriers. We propose a comparison among three
p-i-n diodes: a reference diode with large micro-size injectors
(device I), one with reduced micro-size injectors (device II)
and one with nano-size injectors (device III). The first two
devices are depicted in Fig. 1, in top view layout (device I
in (a) and device II in (b)) and in cross-section (Fig. 1c). A
logical step forward in scaling the dimensions of the injectors
is to go to nanometer dimensions. Successful fabrication of
such small features might require a long process engineering,
and would lead to a big complication of the flowchart. An
alternative way to maintain the process flow simple is to form
antifuse injectors in an insulator by electrical breakdown [11].
The proposed device is shown in Fig. 2.
II. THEORY AND SIMULATION
All the theoretical derivation are made for devices I and
II; in the TCAD simulations [12] we vary the injector width
Fig. 1.
Figures (a) and (b) show the top view layout of the devices I and II,
respectively. W is 100 µm, L is 5 or 10 µm and the injector size Wi is
reduced from 100 to 10 µm in design II. Figure (c) shows the cross sectional
layout along the X − X?axis of the device.
A schematic drawing of Si p-i-n diodes with micro-size injectors.
Fig. 2. A schematic drawing of a p-i-n diode with nano-size antifuse injectors
(device III), in top view (a) and cross-section (b) along the X −X?line. The
device has an intrinsic region of 6x60 µm (LxW).
Wi with respect to the active region width W (see Fig. 1).
The simulated band diagrams for device II are depicted in
Fig. 3, showing that for a forward bias of 0.3 V the full voltage
appears across the active (i.e. intrinsic) region. Consequently,
for the pn-product in the active region holds:
?EFN− EFP
where uT is the thermal voltage, VD is the applied voltage,
EFN, EFP are the electron and hole quasi-Fermi levels and
ni is the intrinsic carrier concentration. Figure 4 shows the
simulations results, applying Boltzmann approximation with
Philips’ unified mobility model [13], doping-induced bandgap
narrowing model [14], and conventional Si recombination
pn = n2
i· exp
uT
?
= n2
i· exp
?VD
uT
?
,
(1)
978-1-4577-0708-7/11/$26.00 ©2011 IEEE 175

Page 2

Fig. 3.
Fig. 1) under a forward bias of 0.3V. The axis origin is at the left-end side
of the device. ECand EV are the conduction and valence band energies.
W=100µm, Wi=10µm, L=5µm, injector length of L∗
used for simulation. The full applied voltage appears across the active region.
The minority quasi-Fermi levels drop in the injector regions indicates that the
current there is limited by diffusion. In the inset, the band diagram along the
Y −Y?axis is shown. The Y axis origin is the midpoint of the active region.
The full applied voltage is available along the whole Y axis.
Simulated band diagram along the X − X?axis of device II (see
p= L∗
n= 3µm was
models. The following Si recombination parameters were
used: BRad = 10−14cm3s−1[15], τSRH = 2.5 · 10−5s,
Cn = 1.83 · 10−31cm6s−1and Cp = 2.81 · 10−31cm6s−1
[16], where BRad is the radiative recombination coefficient.
τSRH is the Shockley-Read-Hall coefficient and Cn and Cp
are the Auger parameters. The uniformly doped p+and n+
injector regions were doped with 1019cm−3. Figure 4a shows
the IV curves of the devices I and II. As we reduce Withe
injector current, which is limited by the diffusion (see Fig.3),
reduces at a given voltage. This decreases the total current
flowing through the diode. Hence, for a particular on current,
the voltage drop across the active region increases for the
diodes with lower Wi. As the applied voltage determines the
pn-product (Eq.(1)), the higher voltage drop will enhance the
radiative recombination rate RRad= BRad(pn − n2
the light emission. The increase in pn-product is shown by
Fig. 4b. An increase of 7.8 times is noticed as Wiis reduced
from 100µm (device I) to 10µm (device II) at 1mA current.
An even larger increase can be expected with further reducing
the injector size to the sub-micron range.
i), hence
III. EXPERIMENTAL
A. Device fabrication
All proposed devices have been fabricated on high qual-
ity Soitec 340-nm SOI substrates (400 nm BOX, p-type,
ρ = 14 − 22 Ω·cm). After standard cleaning, the central
regions were patterned by wet etching in TMAH solution,
and subsequently encapsulated in a 10-nm thick thermal
oxide (900◦C in dry oxygen atmosphere). The devices with
micro-size injectors have been subsequently implanted with
6 · 1015cm−2B+and 5 · 1015cm−2P+, at 30 and 50 kV
respectively; dopants have been activated with a 5 s peak
rapid thermal anneal (RTA) at 900◦C, after capping with
30 nm LPCVD oxide. For the devices with antifuse injectors,
after the oxide encapsulation, polysilicon electrodes have been
deposited and defined via wet etch so to partially overlap the
Fig. 4.
W=100µm, L=5µm, injector length of L∗
simulation. Figure (a) shows the I-V curves. As Wi reduces from 100µm
(Wi = W) to 10µm the total current also reduces. Figure (b) shows the
increase in pn-product at an injected current of 1mA for two extreme cases
of Wi. A 7.8 fold increase is noted.
Simulation results showing the effect of reduction of Wifrom W.
p= L∗
n= 3µm was used for
intrinsic island (see Fig. 2). P+and n+doping is achieved
by implantation of 5 · 1015cm−2of B+and P+, at 30 and
60 kV respectively. The dopants have been activated by 30
min furnace anneal in nitrogen atmosphere after capping. The
contact pads are Al/TiW sintered for 5 min at 400
forming gas.
◦C in
B. Measurement setup
All devices have been characterized on wafer level on a
probe station Karl S¨ uss with a Keithley parameter analyzer
4200. Light emission is measured equipping the probe station
with a XenICs infrared (IR) camera featuring a cooled InGaAs
sensor (an array of 256 × 320 photodiodes) with a detection
range from 900 to 1700 nm (near infrared NIR). Spectra are
captured with a Specim ImSpector spectrometer; to be noted
that this element introduces further losses of signal, so spectra
cannot be acquired below a defined signal threshold.
C. Characterization of micro-size injectors
In Fig.5 we show the I-V characteristics of the diodes for
two active region lengths. As we reduce the injector width Wi
the total current flowing through the diode also diminishes.
As a result, if the device is driven at a constant current,
higher applied bias is available across the active region for
reduced Wi. The internal voltage drop can be determined by
extrapolating the exponential curves yielding a 40mV shift,
and hence a pn-product increase of about 4.7 for both values
of L. This increase enhances the light emission. Figure 6
176

Page 3

Fig. 5.
of L=5µm and L=10µm. As Wi decreases the total current in the diode
also decreases. The device dimensions are: W=100µm, injector length of
L∗
Measured I-V curves for two values of Wifor active region length
p= L∗
n= 10µm.
Fig. 6.
show the device structure under weak external light. Figures (b) and (d) show
the same devices emitting NIR light at 1mA forward current. The device
dimensions are W=100µm, L=5µm, injector length of L∗
When Wiis reduced from 100µm (figures (a) and (b), device I) to 10µm
(figures (c) and (d), device II) the light emission increases. The light intensity
is measured along X−X?axis. Figure (e) shows the increase in EL intensity
for device II. A four fold increase in intensity is observed.
IR images of the LEDs and their EL intensity. Figures (a) and (c)
p= L∗
n= 10µm.
shows the IR images from LEDs with a 5µm long active
region. Left-hand side shows the device structures under the
microscope. On the right-hand side we see the NIR light
emission from the same devices at an injected current of
1mA. Figures 6c and 6d show the diodes with reduced Wi.
The reduction in Wiincreases the light emission at the same
current as predicted by the electrical measurements. Figure
6e shows the EL intensity along the active region at 1mA
current. Approximately 4.2 times higher intensity is seen for
reduced injector width (Wi=10µm). Similarly, a 3.7 times
higher intensity has been seen for devices with 10µm active
region length when injector width Wiwas reduced to 10µm.
The increase in light intensity almost corresponds with the
increase in pn-product.
D. Characterization of nano-size injectors
The antifuses are created and conditioned electrically as de-
scribed in [17]. The properties of the devices vary accordingly
to the programming current Ip: the higher the current, the
higher the conductivity of the nano-links and therefore the
lower the total series resistance of the device. The dimensions
of the nano-injectors also vary with the programming current
Fig. 7.
i-n (black), the one with micro-size injectors (red) and that with nano-size
injectors (blue). For the latter, diffusion is not the limiting current.
Semi-logarithmic plot of the I-V characteristic of the reference p-
Fig. 8. IR image of a diode with antifuse injectors under weak external light
(a) and when emitting NIR light in dark (b).
[11]. To be able to drive the device at a comparable current
as that of the standard p-i-n devices, the antifuses have been
programmed at 12 mA. When the operating current is kept
below the programming current, the devices are stable and the
performance is hardly affected by frequent use of the device.
A comparison among I-V curves for all devices is shown in
Fig. 7: to be noted the reduction in current for the p-i-n with
nano-size injectors (device III). Moreover, the slope of the
curves at low voltage indicates that for the devices with micro-
size injectors (devices I and II) current is limited by diffusion,
whereas for device III the diffusion has reduced to the extent
of not being observable. Figure 8 shows an IR image of the
device. It can be observed that, due to the high reflectance of
the interface Si-SiO2, most of the light is able to escape from
the silicon volume only in presence of structural irregularities,
i.e. at the edges of the Si island or where the antifuse injectors
are formed.
E. Spectral measurements
Figure 9 shows the EL spectra measured for the three
devices, normalized with respect to the integration time. For
the reference p-i-n diode no signal could be acquired for
currents lower than 20 mA, whereas for the other two devices
the current has been fixed at 10 mA. All the spectra show two
emission peaks, located at approximately 1050 nm and 1150
nm. The first peak corresponds to an energy of approximately
1.2 eV and its attribution is still under investigation. The
second peak corresponds to the band to band phonon assisted
177

Page 4

Fig. 9.
p-i-n (black circles), the micro-size injectors (blue triangles) and nano-size
injectors (red squares).
Comparison of the (normalized) emission spectra for the reference
recombination of electrons and holes in Si. The spectral mea-
surements confirm the improvement in light emission already
for the micro-size injectors; the further decrease of injector
dimensions to the nano-size achieves an additional 10-fold
increase in the peaks height. This translates into a similar
increase of the integrated EL, as depicted in Fig.10.
IV. CONCLUSION
In summary, we have shown that light intensity can be
increased by scaling down of the carrier injector size, without
using technologically demanding processes. The simulations
demonstrate that decreasing the injector width Wi decreases
the diffusion current, which reduces the total current flowing
through the diode, availing more applied voltage across the
active region at a particular on current. The higher voltage
increases the pn-product inside active region, resulting in a
higher EL intensity. We confirmed the theoretical expecta-
tion by electrical and optical measurements, comparing the
reference device with diodes having micro- and nano-size
injectors. Shrinking the injectors to micro-size resulted in a
4-fold increased light emission, with a further 10-fold increase
for the nano-size injectors.
ACKNOWLEDGMENT
The authors gratefully acknowledge the support of the Smart
Mix Programme of the Netherlands Ministry of Economic
Affairs and the Netherlands Ministry of Education, Culture
and Science.
REFERENCES
[1] R. Soref, “The past, present, and future of silicon photonics,” IEEE
Journal of Selected Topics in Quantum Electronics, vol. 12, pp. 1678–
1687, 2006.
[2] B. Jalali and S. Fathpour, “Silicon photonics,” J. Lightwave Tech.,
vol. 24, no. 12, pp. 4600–4615, December 2006.
[3] L. Pavesi, “Will silicon be the photonic material of the third millen-
nium?” J. Phys. Cond. Matter, vol. 15, no. 26, pp. 1169–1196, 2003.
[4] N. Daldosso and L. Pavesi, “Nanosilicon photonics,” Laser &
Photonics Reviews, vol. 3, no. 6, pp. 508–534, 2009. [Online].
Available: http://dx.doi.org/10.1002/lpor.200810045
Fig. 10.
and the p-i-n with micro-size (device II) and nano-size (device III) injectors
respectively.
Integrated EL: comparison among the reference device (device I)
[5] A. Kenyon, “Erbium in silicon,” Semiconductor Science and Technology,
vol. 20, no. 12, pp. 65–84, 2005.
[6] M. Reiche, M. Kittler, D. Buca, A. H¨ ahnel, Q.-T. Zhao, S. Mantl, and
U. G¨ osele, “Dislocation-based si-nanodevices,” Japanese Journal of
Applied Physics, vol. 49, no. 4, p. 04DJ02, 2010. [Online]. Available:
http://jjap.jsap.jp/link?JJAP/49/04DJ02/
[7] W. L. Ng, M. A. Lourenc ¸o, R. M. Gwilliam, S. Ledain, G. Shao, and
K. P. Homewood, “An efficient room-temperature silicon-based light-
emitting diode,” Nature, vol. 410, pp. 192–194, 2001.
[8] M. Green, J. Zhao, A. Wang, P. Reece, and M. Gal, “Efficient silicon
light-emitting diodes,” Nature, vol. 412, pp. 805–808, 2001.
[9] T. Hoang, P. LeMinh, J. Holleman, and J. Schmitz, “Strong efficiency
improvement of SOI-LEDs through carrier confinement,” IEEE Elect.
Dev. Lett., vol. 28, pp. 383–385, 2007.
[10] J. Potfajova, B. Schmidt, M. Helm, T. Gemming, M. Benyoucef,
A. Rastelli, and O. G. Schmidt, “Microcavity enhanced silicon light
emitting pn-diode,” Appl. Phys. Lett., vol. 96, pp. 151113–151115,
2010.
[11] N. Akil, “Modeling of light-emission spectra measured on silicon
nanometer-scale diode antifuses,” J. Appl. Phys., vol. 88, no. 4, pp.
1916–1922, August 2000.
[12] Synopsys Sentaurus Device User Guide, Version a-2008.09 ed., 2009.
[13] D. B. M. Klaassen, “A unified mobility model for device simulation-I.
Model equations and concentration dependence,” Solid-State Electronics,
vol. 35, pp. 953–959, 1992.
[14] D. B. M. Klaassen, J. W. Slotboom, and H. C. de Graaff, “Unified
apparent bandgap narrowing in n- and p-type silicon,” Solid-State
Electronics, vol. 35, pp. 125–129, 1992.
[15] M. S. Y. P. M. Levinshtein, S. Rmyantsev, Ed., Handbook Series on
Semiconductor Parameters. World Scientific, 1996, vol. 1.
[16] D. B. M. Klaassen, “A unified mobility model for device simulation-II.
Temperature dependence of carrier mobility and life time,” Solid-State
Electronics, vol. 35, pp. 961–967, 1992.
[17] G. Piccolo, T. Hoang, J. Holleman, A. Kovalgin, and J. Schmitz, “Silicon
LEDs with antifuse injection,” Group IV Photonics, 5th IEEE Int. Conf.
Proc., pp. 49–51, 2008.
178