Connecting TCAD To Tapeout A Journal for Process and Device Engineers
Enhanced Silicon Light Emission Intensity with
Multiple SiGe Quantum Well Structure
The measured I-V curve from a ten period Si/SiGe
MQW pin LED fabricated using a UHVCVD system
is compared with ATLAS simulation results. A sizable
silicon emission peak is observed at high current injec-
tion mode at room temperature. This phenomena can
be explained as follows: because of the hetero-junction
between the top silicon buffer layer and MQW, when
bias increases there is a potential barrier formed due to
band bending. Thus there will be a large accumulation
of holes in the buffer layer. The recombination rate in
this layer increases which results in increased silicon
light intensity. Figure 1. Sample cross section.
for all of the epitaxial layers. After depositing a 25 nm
The use of silicon germanium (SiGe) for optoelectronic undoped Si layer on the n+ substrate, the 10 periods
components is highly advantageous since SiGe is com- consisting of Si/Si0.5Ge0.5 making up the MQW structure
patible with Si based technologies. Improved growing were grown. Each period of the MQW consists of a 3.9
techniques for heterostructures have also made the nm Si0.5Ge0.5 well and a 3nm Si barrier. However, because
manufacturing of SiGe based devices much easier. of the background doping of the UHV-CVD, this region
Advantages of using SiGe for optoelectronic structures was actually lightly p-type doped (NA~1016 cm-3) and de-
include the low defect density of the material, which noted as P- region. After the growth of the MQW, a 24nm
enhances operation at room temperatures. Also a SiGe undoped Si layer was deposited. Finally, a silicon layer
based device’s operating wavelength can be tuned over was deposited on top acting as the buffer layer. The top
the range of 1.3um to 1.55um making them ideal choices layer is heavily doped p-type (NA=1019cm-3) in order to
for optical fiber communications. Therefore there is wide form an ohmic contact.
spread interest in SiGe and SiGe based devices. The de-
vice studied in this article is a ten period Si/SiGe multi
Continued on page 2 ...
quantum well (MQW) structure. ATLAS is then used to
simulate the device and the simulated data is compared
to the measured I-V data. In this way a more physical
insight into the device operation can be obtained. Comparison of 3 Dimensional Quantum
Effect of Nano Device Using BQP
Preparation Model on ATLAS3D................................................ 4
Evaluating of the Avalanche Failure of Power
The device studied in the article utilizes a p-i-n structure
MOSFETs Using ATLAS .................................... 8
with a silicon buffer layer. The sample was grown on n-
Si(001) substrates by a UHV chemical vapor deposition Interconnect Parasitic Extraction of BiCMOS
Cell Using Simucad CLEVER .............................. 11
(UHV-CVD) system at a pressure of 5×10 -9 Torr at 600oC
Volume 16, Number 5, May 2006
May 2006 Page 1 The Simulation Standard
Figure 2. Simulated (red) and measure (green) IV curve. Figure 3. Band diagram for the device at zero bias.
The material parameters and the MQW module param- nc300=2.8e19 nv300=1.8e19 vsatn=2.4e7
eters used in ATLAS are as following: vsatp=1.775e7 permi=13.95 mun=101.5
mqw ww=0.0039 wb=0.003 nwell=10 nx=5
ny=240 acceptors=1e16 \ Discussion
xmin=0 xmax=0.05 ymin=-0.094 ymax=- Figure 1 shows a cross section of the device. Figure 2
0.022 material=SiGe xcomp=0.5 shows the measured and simulated IV curves. Figure 3
material material=silicon EG300=1.12 shows the simulated device band diagram at zero bias.
affinity=4.05 taun0=1e-7 taup0=1e-7 The band diagram and corresponding hole distribution
ni=1e10 \ for three different injection currents are shown in Figure
nc300=2.8e19 nv300=1.8e19 mun=1450 4. Figure 5 shows the intensity versus energy for two dif-
mup=450 vsatn=2.4e7 vsatp=1.65e7 bn=1 ferent injection currents. From Figure 5 we see that for low
bp=0 injection levels the light intensity has two peaks, one for
Si and one for SiGe. Both the peaks are somewhat compa-
material material=sige EG300=0.917
rable. As the injection level is increased the peak corre-
affinity=4.033 taun0=4e-11 taup0=4e-11
sponding to SiGe is considerably reduced whilst the peak
mso=0.1625 ni=1e12 \
Figure 4 The simulation results of band diagram and hole distribution with different currents.
The Simulation Standard Page 2 May 2006
Figure 5. Compared EL intensity (a) injection current=50mA (b) injection current=250mA.
corresponding to the Si material is increased. Even though light emission is dependent on the radiative recombina-
excellent confinement is achieved by the quantum wells, tion and the radiative recombination in turn is propor-
the main component in the optical spectrum at high injec- tional to the electron - hole product, the Si light emission
tion levels is not from SiGe, as expected, but from Si. increases considerably with higher injection currents
whilst the SiGe light emission receives little increase.
To try to understand the phenomena simulations were
performed to analyse the electron and hole distributions
in the structure (shown in Figure 6). For initial injec- Conclusion
tion currents, holes flow into the quantum wells and We successfully simulated a MQW SiGe LED which can
are stored in them. This in turn promotes a build up enhance Si light emission using ATLAS. This has en-
of the internal electric field at the interface of the QW abled the underlying physical behaviour of the device to
/ Si buffer regions. This electric field - band bending be more thoroughly understood.
forms a barrier to hole flow. Figure 6 shows that the hole
distribution in the quantum well for injection currents Reference
of 50mA and 250mA is almost the same, however in the  Y. H. Peng, C.H. Hsu, P. S. Chen, M. -J. Tsai, C. H. Kuan, and C. W.
Si buffer region the hole distribution is increased by a Liu, Y. W. Suen, “The Electroluminescence Evolution of Ge Quan-
tum-Dot Diodes with the Fold Number,” Applied Physics Letters,
factor of almost five for the two injection currents. As the Dec. 2004.
Figure 6. Electron and hole distribution.
May 2006 Page 3 The Simulation Standard