Open Access
Add to BibSonomy
Abstract
A dual-epitaxy Si/Ge broadband photodetector covering the wavelength range from 400 nm to 1600 nm was developed and demonstrated. The detector was realized on silicon platform with vertically stacked epitaxial Si and Ge layers in the detector’s depletion region, which combined the response from a Si detector and a Ge detector. Under room temperature, dark current density of dual-epitaxy Si/Ge broadband photodetector is 1.3E-4 A/mm2 at 2 V reverse bias. The dual-epitaxy Si/Ge broadband has its peak spectral responsivity response of more than 0.6 A/W near 1000 nm wavelength and has spectral responsivity above 0.1A/W in the range from 450 nm to 1600 nm. Its temperature dependency was studied. The dual-epitaxy Si/Ge quadrant detector was proved working in the application of absolution optical power measurement in cryogenic radiometer.
© 2021 Optical Society of America under the terms of the OSA Open Access Publishing Agreement
1. Introduction
The primary standard for optical power measurement in a national metrology institute is a cryogenic radiometer operating at an absolute temperature of ∼13 K or even less, based on thermal equivalence between optical and electrical heating. Combined with a group of intensity stabilized lasers, it is capable of measuring laser power with very low uncertainty. The absorptance of the light receiving cavity in the cryogenic radiometer is commonly above 0.999 across the wavelengths from ultra-violet to visible and near infrared [1]. A silicon-based quadrant detector installed in the cryogenic radiometer assists the fine alignment of the laser into the cryogenic radiometer cavity. After the fine alignment, during the cryogenic radiometer operation, the silicon-based quadrant detector collects and measures the scattered light that is blocked out of the cavity. The small amount scattered light will be counted in measurement to achieve less uncertainty. However, the silicon-based quadrant detector operates at the wavelengths of 1000 nm or less which limits the operation of the cryogenic radiometer at the wavelengths above 1000 nm.
National metrology institutes usually only do the calibration of optical power measurement for silicon photodetectors with the cryogenic radiometer at several wavelengths in the silicon photodetectors’ range. Out of the silicon detector’s wavelength range, a pyroelectric photodetector with flat responses across the wavelengths will be used to transfer the spectral responsivity scale. Hence, the spectral responsivity scale out of silicon detector’s wavelength range has larger uncertainty than the scale comfortably falling in silicon detector’s wavelength range.
However, the need for lower measurement uncertainties in the wavelengths of optical communications has increased. In order to improve the responsivity scale in the wavelengths above 1000 nm, the Physikalisch-Technische Bundesanstalt (PTB) developed a spectral responsivity scale in the spectral range from 960 nm to 1550 nm by deploying external components with cryogenic radiometer to assist the laser alignment [2]. The external components are a germanium (Ge) photodiode and a spherical mirror with a central hole of the same diameter as the hole of the quadrant photodiode in front of the cavity of the cryogenic radiometer. The central hole of the spherical mirror simulates the entrance hole of the internal cavity. The spherical mirror reflects the light which does not pass through the central hole to the Ge photodiode. The Ge photodiode’s output simulates the light being blocked out of the cavity of the cryogenic radiometer. In this method, the measurement of the scattered light was performed outside the cryogenic radiometer and the optical axis of the external components must be as similar as the cryogenic radiometer. A slight difference in the axis may cause drift in the measurement. The relative standard uncertainty achieved for laser power calibration with these external components in this paper was 0.034% at the wavelengths of 1310 nm and 1550 nm. However, less than 0.01% relative standard uncertainty could be achieved at the wavelengths below 900 nm with the internal installed Si quadrant detector [3].
Another way to operate the cryogenic radiometer for visible wavelengths as well as wavelengths above 1000 nm is to install a broadband quadrant detector. There are broadband photodetectors consisting of a Ge or InGaAs photodetector and a second silicon photodetector being stacked together to cover the broad wavelength range [4,5]. An InGaAs PD with modified bandgap of the active layer would extend to 0.5 µm on the shorter wavelength side short wavelengths [6]. Integration of multiple active layers provides absorption of more than one band on one substrate. There is multiple junction integration technology for high efficiency solar cells [7]. Ge thin film layer is integrated on Si thin film layer to form two junctions to convert light from visible to near IR to electricity. An integrated design of a reversely connected dual diode - one in Si and one in Ge, together with simulation results was presented in [8]. By changing the bias voltage, one of the dual would be switched on and the other would be off. Under a certain bias voltage, the dual diode would be both partially on, resulting a coverage of both the Si band and Ge band with compromised responsivities in the wide range. None of these solutions have been used for cryogenic radiometer applications.
Besides, MXene, as a 2D material, has received increasing attentions in recent years. Nano-structured MXene materials have shown excellent broadband absorptions over visible and near infrared wavelength ranges [9]. However, it is still quite challenging to integrate the fabrication of MXene devices on Si processing platforms, which will somehow limit the extensive applications of this novel material for mass production.
Si and Ge are both group-IV materials. Ge can be epitaxially grown on Si. And Si can be epitaxially grown on Ge. Ge-on-SOI photodetector has high dark current at room temperature, and its dark current drops as temperature decreases [10]. We developed a monolithic broadband quadrant detector by combining Ge and Si active layers monolithically together on the same substrate, and forming p-i-n junction and depletion region with the two layers of materials. The integrated Ge/Si photodetector responses to the light in a broad wavelength range from 400 nm to 1600 nm. Spectral responsivity response of the integrated broadband photodetector was characterized. The monolithic broadband quadrant detector was fabricated and installed in the cryogenic radiometer in National Metrology Centre, A*STAR, Singapore. We setup fiber lasers of 1310 nm and 1550 nm with the cryogenic radiometer to calibrate the spectral responsivity of the InGaAs trap detector based absolute transfer standard scale at these two wavelengths.
2. Design and simulation of the dual-epitaxy Si/Ge broadband quadrant photodetector
The original quadrant detector used in the cryogenic radiometer is a silicon-based PIN photodiode. The central hole allows laser to pass through and the active area surrounding the hole is divided into 4 sections. We designed the layout of broadband Si/Ge quadrant photodetector with the diameter of 25 mm and the hole of 9 mm (Fig. 1). The active area of each quadrant is 107 mm2.
To integrate Ge and Si active layers monolithically together, the intrinsic crystalline Ge layer and the intrinsic crystalline Si layer are grown in a vertical stack on highly doped n-type (or p-type) silicon substrate as shown in Fig. 2. On top of the epitaxial silicon layer, a layer of p-type (or n-type) silicon is formed. A p-i-n junction and depletion region are formed between the top doped layer and the highly doped substrate, and a PIN photodetector comes into being with the combined intrinsic Ge and Si layers. When light is launched to the device surface, photons in the visible wavelengths can be directly absorbed in Si layer in the depletion region and converted into photon-generated carriers. Silicon is transparent to low energy photons in the infrared portion of the spectrum above 1000 nm, hence, light with longer wavelengths will pass through the Si layer without obvious attenuation and then be detected by the underneath Ge layer in the depletion region. The dual-epitaxy Ge/Si photodetector can response to the light in a broad wavelength range from 400 nm to 1600 nm. A SiO2/SiN stack layer is deposited at the incident window to minimize the light reflection at the air-silicon interface.
Fig. 2. Vertically integrated intrinsic crystalline Si and Ge on Si substrate (cross-sectional view).
Spectral responsivity of the device as in Fig. 2(a) is simulated. It has the p + Si/i-Si/i-Ge/n + Si structure. The i-Si and i-Ge can be fully depleted, and the photons being absorbed in these layers contribute to the photocurrent, whereas the photons being absorbed in the top p+ Si layer and the bottom n+ Si layer do not contribute to the photocurrent, thus, the spectral responsivity R of the dual-epitaxy Ge/Si photodetector can be estimated as Eq. (1):
In our simulation of spectral responsivity, we assume A(λ) = 1 and IQE=100% to ignore the reflection and IQE effects. An important parameter is the top Si p+ layer thickness, which depends on the fabrication process, i.e., the implantation dose/energy and the flowing thermal budget. We estimated the thickness of p+ layer in our fabrication to be about 0.2 - 0.3 µm. A thickness of 0.3 µm for p+ layer was used in the simulation. Figure 3 compares the calculated spectral responsivities of the PDs with 1.5 µm Si / 1.5 µm Ge stack, 2 µm Si / 2 µm Ge stack, 4 µm Si / 4 µm Ge stack, 4 µm Si, and 4 µm Ge.
We can see that the responsivity of Si/Ge stack is similar to that of Si at short wavelengths from 400 nm to 500 nm and is similar to that of Ge at the long wavelengths of above 1100 nm. At the middle wavelengths ranging from 500-1100 nm, the responsivity is significantly enhanced as compared to the single layer PDs. Moreover, we found that the thickness of i-Si doesn’t affect the overall responsivity too much as the 1.5 µm Si /1.5 µm Ge stack has almost similar responsivity as the 4 µm Si / 4 µm Ge stack in the range from 400 nm to 900 nm. This is because silicon has large absorption in the short wavelengths and the photons transmitted through the i-Si layer in the middle wavelengths can be absorbed by the i-Ge. On the other hand, the thicker the i-Ge thickness, the better for the long wavelength photon absorption. However, thicker Ge epitaxial layer would induce larger stress across the wafer, wafer warpage and higher defect density. Hence, considering the fabrication feasibility, the optimal design was 1.5 µm Si / 2 µm Ge stack.
We designed PECVD SiO2/SiN stack on the PD surface to protect the PD and also acted as the anti-reflection (AR) layer for the air-Si interface. The thicknesses of SiN and SiO2 layers were optimized to be 80 nm and 120 nm, respectively, to achieve <10% reflection in the wavelength range from 400 nm to 1600 nm.
3. Fabrication and characterization
The dual-epitaxy Si/Ge broadband quadrant detector was realized on silicon platform with vertical integrated Si and Ge layers in a detector’s depletion region. Single broadband Si/Ge detectors with the active areas size of 1 mm by 1 mm and 6 mm by 6 mm were put in the layout besides the quadrant detectors for testing purpose. Non-selective Epitaxial growth of intrinsic Ge was done on doped Si substrate and followed by non-selective epitaxial growth of intrinsic Si on top of the intrinsic Ge. The thicknesses of Ge Epi and Si Epi in the fabrication are in the range from 1.5 µm to 2 µm. Due to the lattice constant mismatch between Si and Ge, there is a high-defect buffer layer between the epi-Ge and epi-Si layers which will recombine the photo generated carriers, thus reduce the internal quantum efficient. We optimized the epitaxy process to reduce the buffer layer thickness and improve the epitaxy layer quality.
Figure 4 schematically shows the fabrication process flow of our dual-epitaxy Si/Ge broadband PD. It started with a highly doped n+ silicon wafer. Successive intrinsic Ge and Si layers were epitaxially grown non-selectively by PECVD, where the target thickness of each layer is under well control. Mesa was etched to isolate the active regions from adjacent devices. Boron was implanted from the top of Si epi layer to form p+ layer. Metallization and AR coating were fabricated afterwards. Figure 5 shows the cross-sectional view of the fabricated device with the labels for each layer.
3.1 Dark Current
We studied the I-V curves of the dual-epitaxy Si/Ge broadband photodetector under difference temperature. The Si/Ge photodetector was placed on a TEC and measured at dark condition. Due to the size limitation of the TEC used in the experiment, we measured the small single detectors in this setup.
The high dark current density is expected because a high number of defects are formed when Ge is grown directly on Si and Si is grown on top of Ge. Figure 6 shows the dark current density in a 1 mm by 1 mm Si/Ge photodetector measured under room temperature 25°C and reverse bias. The dark current density is 1.3E-4 A/mm2 at 2 V reverse bias. It is comparable to the dark current of 2E-8 A at 2 V reverse bias in a 10 µm by 10 µm Ge-on-SOI photodetector reported in Ref. [10], which means the dark current density of 2E-4 A/mm2.
The dark current decreases tremendously as the temperature decreases. The dark current is dominated by trap-assisted generation and recombination mechanism [10] and it is proportional to the detector’s active area. Figures 7 and 8 are the plots of dark current density measured in the 1 mm by 1 mm and 6 mm by 6 mm Si/Ge photodetector, respectively, across the temperature at reverse bias near 0 V. Both detectors show the exponential decrease of the dark current as temperature decreases. The dark current densities in Si/Ge photodetectors with different size are about the same under the same measurement conditions.
Fig. 7. Dark current density in the 1 mm by 1 mm dual-epitaxy Si/Ge PD under reverse bias across temperature.
Fig. 8. Dark current density in the 6 mm by 6 mm dual-epitaxy Si/Ge PD under reverse bias across temperature.
The measurement cavity of the cryogenic radiometer operates at the temperature of ∼13 K or less, and the quadrant detector is at the entrance of the cavity. Hence, the temperature of the quadrant detector is expected to be close to that of the cavity during operation. Besides, the quadrant detector in the cryogenic radiometer will be operated at zero bias condition, we expect the dark current of the quadrant detector in cryogenic radiometer operation is not significant.
3.2 Spectral Responsivity
The spectral responsivity of the broadband Si/Ge photodetector was measured by using a facility consisting of a broadband source, a monochromator and reference standard photodetectors. Tungsten lamp was used as the broadband source in the measurement. A monochromator was used to select the wavelength from the source in the range from 400 nm to 1600 nm. After the monochromator, the light of the selected wavelength falling on the photodetector had a power of 1 to 10 µW. The measurement was carried out with 0 V bias voltage in the wavelength range from 400 nm to 1600 nm for the quadrant detector as well as the single detectors with difference sizes.
Figure 9 shows the spectral responsivity of the Si/Ge PDs with 1 mm by 1 mm size, as well as the responsivity of reference Si, InGaAs and Ge photodetectors. The measurements were done at room temperature. The silicon reference PD responsivity cuts off at around 1000 nm. Typical Ge PD and InGaAs PD responsivities start from around 800 nm and cut off at round 1600 nm. The broadband Si/Ge PD responsivity covers the both bands of Si PD and Ge PD. The Si/Ge PD has responsivity peak at near 1000 nm and decrease at shorter and longer wavelengths. The decrease at shorter wavelength is similar to the normal silicon photodetector. The Si/Ge PD has lower responsivity at short wavelengths near 400 nm than pure Si PD. This is due to recombination process of the photon-generated carriers while the carriers move across the depletion region. The Si/Ge PD has lower responsivity than normal surface illuminated Ge PD at longer wavelength may be due to not enough Ge epi thickness so that the light with longer wavelengths is not fully absorbed by the material. Taking the PD surface reflection and internal quantum efficiency into consideration, the measured spectral responsivity of the Si/Ge PD is very close to the simulation results in Fig. 3.
By improving the Si and Ge epi quality, the recombination process can be decreased and the dark current will be decreased. Hence, having good epi quality is very important for the Si/Ge PD fabrication. By improving the Ge epi thickness, the responsivity of the Si/Ge PD at longer wavelengths can be improved. However, the thicker the epitaxial layer, the more challenging is the epitaxial process.
Figure 10 shows the spectral responsivity of 3 pieces of Si/Ge PDs with 1 mm by 1 mm size selected at different locations across the wafer. The standard deviation of the spectral responsivities of the 3 PDs were calculated across the wavelength. The average standard deviation is 1.1% of the responsivity.
Fig. 10. Spectral responsivity of the 1 mm by 1 mm dual-epitaxy Si/Ge broadband PDs on wafer 1 under 0 V bias.
The spectral responsivity difference among 3 different wafers were evaluated and shown in Fig. 11. The average standard deviation of the responsivity of Si/Ge PDs with 1 mm by 1 mm size from the 3 difference wafers is 10%. The variation is mainly due to the difference in defect density in the depletion regions and the anti-reflection coatings across the wafers.
Fig. 11. Spectral responsivity of the 1 mm by 1 mm dual-epitaxy Si/Ge broadband PD under 0 V bias from 3 different wafers.
Decreased spectral responsivity is observed in larger size Si/Ge PDs at room temperature. The loss of photocurrent may be due to the more recombination process of the carriers in the depletion region in larger area Si/Ge PDs, hence the responsivity is affected. The location where the light spot falls on the large area Si/Ge PD also affect the loss of photocurrent. The more the distance from the light spot to the metal contact for the doped silicon layer, the more the loss of photocurrent due to carrier recombination.
Figure 12 shows the spectral responsivity of the four quadrants, respectively, of one quadrant Si/Ge PD. The measurement was done with the light spot falling on the area near the metal contact for top doped layer under room temperature. The responsivity of the quadrant falls to about 60% of the responsivity of the Si/Ge PD with 1 mm by 1 mm size.
A further study was on the temperature dependency of the responsivity. The spectral responsivity of Si/Ge PDs with 1 mm by 1 mm size didn’t have obvious change when temperature changed from 24°C to 8°C, except that the responsivity at the edges of the spectral responsivity curve was affected due to the material absorption spectrum shifting towards shorter wavelengths at lower temperature. Both Si and Ge have broad absorption spectrum. Consequently, the temperature-induced shifts only affect the edges of the spectral responsivity curve significantly.
However, the spectral responsivity of Si/Ge PDs with larger size significantly improved when temperature decreased. The carrier recombination decreases as temperature decreases, hence, the photocurrent loss in large size Si/Ge PDs decreases at lower temperature. Figure 13 shows the spectral responsivity of Si/Ge PDs with 6 mm by 6 mm size from 8°C to 24°C measured at the center location of the active area, comparing with the 1 mm by 1 mm PD measured at room temperature. As water condensation happened when temperature was low, and liquid water had significant absorption at wavelengths above 1000 nm, the spectral responsivity measurement at wavelengths above 1000 nm at low temperatures were affected by the water droplets condensed on the PD surface. The trend of responsivity across the temperature at the wavelengths in the range from 400 nm to 1600 nm were shown in Fig. 14. As the spectral responsivity measurement at wavelengths above 1000 nm at 8°C were affected by the water droplets condensed on the PD surface, the data at 8°C for 1310 nm and 1550 nm were not included in Fig. 14(b). Due to the limitation of our experiment setup, we didn’t obtain the data at temperature below 8°C. We expect the spectral responsivity of large area Si/Ge PD to be closer and closer to that of small area PD with shift towards shorter wavelength if the temperature keeps decreasing. The same phenomenon was observed in large area quadrant Si/Ge PD. However, due to the TEC size limit in our experiment, the temperature of the quadrant Si/Ge PD couldn’t be controlled well.
Fig. 13. Spectral responsivity of dual-epitaxy Si/Ge broadband PD with 6 mm by 6 mm size across temperature.
Fig. 14. Responsivity of dual-epitaxy Si/Ge broadband PD with 6 mm by 6 mm size across temperature at the wavelengths in the range from 400 nm to 1600 nm.
4. Testing of dual-epitaxy Si/Ge broadband quadrant PD in cryogenic radiometer
A dual-epitaxy Si/Ge quadrant detector was picked and installed in the cryogenic radiometer in National Metrology Centre, A*STAR, Singapore. The quadrant detector was electrically connected out to a 4-channel trans-impedance amplifier and a reader. Without applying bias voltage, we observed significant readings in the 4 quadrants under dark condition at room temperature. The reading decreased rapidly as the cryogenic radiometer was cooled down from the room temperature. Comparing to the pure silicon-based quadrant detector, the Si/Ge quadrant detector had much larger dark reading under the cryogenic radiometer operating condition. We adjusted the reader controller to cancel the dark reading to zero under the low temperature condition.
The system was tested with two lasers at 488 nm and 1310 nm. As we expected that Si/Ge quadrant detector had the spectral responsivity very close to the Si/Ge detector with size of 1 mm by 1 mm, we applied the responsivity at 488 nm and 1310 nm measured in the detector with 1 mm by 1 mm size, which was the nearest to the picked quadrant detector on the same wafer, for the quadrant detector to estimate the scattered light being blocked out of the absorbing cavity of the cryogenic radiometer.
The laser was collimated and the beam had a diameter of 3 mm to 4 mm. The laser beam was aligned with the absorbing cavity in cryogenic radiometer with the assistance of the Si/Ge quadrant detector installed in the cryogenic radiometer. The collimated laser went through the 9 mm central hole of the Si/Ge quadrant detector. The 4 quadrants captured the light being blocked by them. The photocurrents were amplified and the reader showed the readings. Then we adjusted the laser alignment so that the 4 quadrants’ readings were minimal and balanced. Although the measurement for the scattered laser had large uncertainty due to the uncertainty of the spectral responsivity of the quadrant detector in the operating condition, the overall influence on the uncertainty of the total laser power measurement was limited because the scattered laser power could be less than 0.01% of the total laser power after good alignment. We observed that the Si/Ge quadrant detector’s readings drifted slightly during the measurement which may be due to the drift of the dark current with the temperature.
Laser powers were calibrated at 488 nm and 1310 nm, respectively, with the cryogenic radiometer installed with the Si/Ge quadrant detector. The relative combined uncertainty of laser power calibration at 488 nm by using the cryogenic radiometer with internal Si/Ge quadrant detector was 0.03%, which was about 3 times of that achieved by using internal pure silicon quadrant detector. The relative combined uncertainty of laser power calibration at 1310 nm was 0.02%. As a result, the expanded uncertainties (estimated a level of confidence of approximately 95% with coverage factor k=2) of the laser power calibrations at 488 nm and 1310 nm were 0.06% and 0.04%, respectively. The uncertainty at 1310 nm was better than that at 488 nm due to better power stability and beam quality from the 1310 nm single mode fiber laser, and higher Si/Ge quadrant detector spectral responsivity at 1310 nm than 488 nm.
A more detailed study on the Si/Ge quadrant detector’s characteristics under lower temperature will benefit its application in cryogenic radiometer and may further reduce the uncertainty in the laser power calibration.
5. Conclusions
Dual-epitaxy Si/Ge broadband detector was realized on silicon platform with vertically stacked Si and Ge layers in the detector’s depletion region. The detector can response to 400–1000 nm as Si detectors do as well as to 900-1600 nm as Ge detectors do. Compared with pure Si PD, the dual-epitaxy Si/Ge PD showed lower responsivity at short wavelengths near 400 nm, mainly due to recombination process of the photon-generated carriers at the defects while the carriers move across the depletion region. At longer wavelengths, the responsivity of the dual-epitaxy Si/Ge PD was lower than normal surface illuminated Ge PD, which may be attributed to insufficient light absorption due to limited thickness of epitaxial Ge layer. It would be better if thicker Ge layer can be adopted to have sufficient light absorption. However, the higher amount of defects in thicker epitaxial layers will lead to higher dark current and higher loss of photocurrent, especially when the detector has a large active area. A tradeoff has to be taken into consideration in device design and fabrication.
The dark current and recombination process decrease at low temperature. A dual-epitaxy Si/Ge quadrant detector installed in the cryogenic radiometer in National Metrology Centre, A*STAR, Singapore. It was tested at 488 nm and 1310 nm to prove it working for the wide wavelength range in this application.
Funding
A*STAR Singapore Aerospace Programme Cycle 14 (A2015a0119).
Acknowledgments
The authors thank Mr Huang Xuebo from National Metrology Centre, A*STAR for his guidance and advice on the detector’s application in cryogenic radiometer. They thank Mr Edward Koh Sing Chee from Institute of Microelectronics, A*STAR, and Mr Yong Chee Meng from National Metrology Centre, A*STAR, for their support in the characterization of the detectors. They also thank Mr Ian Foo Mingze for his support in testing the quadrant detector in the cryogenic radiometer during his attachment in National Metrology Centre, A*STAR.
Disclosures
The authors declare no conflicts of interest.
Data availability
Data underlying the results presented in this paper are not publicly available at this time but may be obtained from the authors upon reasonable request.
References
1. M. López, H. Hofer, and S. Kück, “Measurement of the absorptance of a cryogenic radiometer cavity in the visible and near infrared,” Metrologia 42(5), 400–405 (2005). [CrossRef]
2. L. Werner, R. Friedrich, U. Johannsen, and A. Steiger, “Precise scale of spectral responsivity for InGaAs detectors based on a cryogenic radiometer and several laser sources,” Metrologia 37(5), 523–526 (2000). [CrossRef]
3. R. Goebel, R. Pello, R. Kohler, P. Haycocks, and N. Fox, “Comparison of the BIPM cryogenic radiometer with a mechanically cooled cryogenic radiometer from the NPL,” Metrologia 33(2), 177–179 (1996). [CrossRef]
4. Optical measurement method using stacked germanium and silicone detectors, US patent 5124553 (1990).
5. Thorlabs product details on dual band Si/InGaAs photodiode, DSD2, https://www.thorlabs.com/thorproduct.cfm?partnumber=DSD2.
6. Hamamatsu product details on wide spectral response range InGaAs PIN photodiode, G10899-01 K, https://www.hamamatsu.com/eu/en/product/type/G10899-01K/index.html.
7. G. Sun and F. Chang, “1 and R. A. Soref, “High efficiency thin-film crystalline Si/Ge tandem solar cell”,” Opt. Express 18(4), 3746–3753 (2010). [CrossRef]
8. A. D. Iacovo, A. Ballabio, J. Frigerio, L. Colace, and G. Isella, “Design and simulation of Ge-on-Si photodetectors with electrically tunable spectral response,” J. Lightwave Technol. 37(14), 3517–3525 (2019). [CrossRef]
9. K. Chaudhuri, M. Alhabeb, Z. Wang, V. M. Shalaev, Y. Gogotsi, and A. Boltasseva, “Highly broadband absorber using plasmonic titanium carbide (MXene),” ACS Photonics 5(3), 1115–1122 (2018). [CrossRef]
10. S. J. Koester, L. Schares, C. L. Schow, G. Dehlinger, and R. A. John, “Temperature-dependent analysis of Ge-on-SOI photodetectors and receivers,” Proc. 3rd IEEE Int. Conf. Group IV Photonics, 179–181 (2006).
