4 Silicon Plasmonics
163
to the combined effect of SB lowering due to the image force and the space-chargelimited nature of the photocurrent [52]. Responsivity was extracted by measuring the
current across the Schottky contact under weak reverse bias of 0.1 V as a function of
the incident optical power. A representative measurement result at the wavelength
of 1.55 μm is shown in Fig. 4.9b. Indeed, the measured photocurrent showed linear
dependence in the incident optical power. The obtained internal responsivity of the
device was found to be 0.25, 1.4 and 13.3 mA/W for optical wavelengths of 1.55, 1.47
and 1.31 μm respectively. External responsivity was actually lower due to fiber to
photonic waveguide and photonic waveguide to plasmonic waveguide coupling loss.
Very recently, quantum efficiency was improved by nearly two orders of magnitude
using the concept of roughness engineering, suppressing the reflection of electrons
from the metal-silicon boundary [63].
4.3 Concluding Remarks
Being compatible with CMOS technology on one hand and allowing true subwavelength confinement on the other, the field of silicon plasmonics holds a great promise.
Currently, the integration between plasmonics and silicon technology is still at its
infancy. Keeping in mind that silicon plasmonic based devices have been emerged
only over the last five years it would be reasonable to believe that plenty of opportunities for exploring advanced components, functionalities and configurations are
available. However, in order to make a significant breakthrough in the field one should
first be aware of the opportunities and not of lesser importance—the limitations of
this platform.
To start, it is unlikely that silicon plasmonics will be used for guiding of signals
over long distances, owing to the large propagation loss involved. The attempts to
compete with dielectric guiding for centimeters scale range do not show a great
promise. If silicon plasmonics is to be used for guiding, it is most likely be in
the local interconnect level (micrometers length scale), interfacing between single
elements and units on a chip. On the other hand, this very same nature of high
propagation loss makes the silicon plasmonic platform an attractive choice for the
construction of advanced photodetectors with enhanced wavelength photo response,
as was shown for example by the several demonstrations of plasmonic enhanced
Schottky detectors. The strong absorption and the high confinement may also play
a positive role in the construction of silicon plasmonic modulators based on the
effect of electro absorption, owing to the small device footprint and the relatively
low number of carriers required to achieve a significant modulation. This may lead
to ultra low energy per bit modulations at high rates.
Another promising direction is related to light matter interactions. The capability of plasmonics to concentrate electromagnetic energy into deep subwavelength
scale volumes makes it attractive for enhancing spontaneous emission from silicon.
While silicon based SPASERS are not expected to emerge in the near future, it may
be sufficient to replace some of their functionalities using nanoscale silicon based
163
to the combined effect of SB lowering due to the image force and the space-chargelimited nature of the photocurrent [52]. Responsivity was extracted by measuring the
current across the Schottky contact under weak reverse bias of 0.1 V as a function of
the incident optical power. A representative measurement result at the wavelength
of 1.55 μm is shown in Fig. 4.9b. Indeed, the measured photocurrent showed linear
dependence in the incident optical power. The obtained internal responsivity of the
device was found to be 0.25, 1.4 and 13.3 mA/W for optical wavelengths of 1.55, 1.47
and 1.31 μm respectively. External responsivity was actually lower due to fiber to
photonic waveguide and photonic waveguide to plasmonic waveguide coupling loss.
Very recently, quantum efficiency was improved by nearly two orders of magnitude
using the concept of roughness engineering, suppressing the reflection of electrons
from the metal-silicon boundary [63].
4.3 Concluding Remarks
Being compatible with CMOS technology on one hand and allowing true subwavelength confinement on the other, the field of silicon plasmonics holds a great promise.
Currently, the integration between plasmonics and silicon technology is still at its
infancy. Keeping in mind that silicon plasmonic based devices have been emerged
only over the last five years it would be reasonable to believe that plenty of opportunities for exploring advanced components, functionalities and configurations are
available. However, in order to make a significant breakthrough in the field one should
first be aware of the opportunities and not of lesser importance—the limitations of
this platform.
To start, it is unlikely that silicon plasmonics will be used for guiding of signals
over long distances, owing to the large propagation loss involved. The attempts to
compete with dielectric guiding for centimeters scale range do not show a great
promise. If silicon plasmonics is to be used for guiding, it is most likely be in
the local interconnect level (micrometers length scale), interfacing between single
elements and units on a chip. On the other hand, this very same nature of high
propagation loss makes the silicon plasmonic platform an attractive choice for the
construction of advanced photodetectors with enhanced wavelength photo response,
as was shown for example by the several demonstrations of plasmonic enhanced
Schottky detectors. The strong absorption and the high confinement may also play
a positive role in the construction of silicon plasmonic modulators based on the
effect of electro absorption, owing to the small device footprint and the relatively
low number of carriers required to achieve a significant modulation. This may lead
to ultra low energy per bit modulations at high rates.
Another promising direction is related to light matter interactions. The capability of plasmonics to concentrate electromagnetic energy into deep subwavelength
scale volumes makes it attractive for enhancing spontaneous emission from silicon.
While silicon based SPASERS are not expected to emerge in the near future, it may
be sufficient to replace some of their functionalities using nanoscale silicon based