rus
TS_pub
technospheramag
technospheramag
ТЕХНОСФЕРА_РИЦ
Issue #2/2024
M. S. Kovalev, I. M. Podlesnykh, K. E. Pevchikh, S. I. Kudryashov
Near-Infrared Planar Photonics Based on Hyperdoped Silicon: Prospects
Near-Infrared Planar Photonics Based on Hyperdoped Silicon: Prospects
DOI: 10.22184/1993-7296.FRos.2024.18.2.136.151
The development of silicon photonics over the past two decades has made silicon the preferential platform for photonic integration. In this paper, we offer our views toward the increasing field of integrated silicon-based near-infrared photonics. A comprehensive overview of the state-of-the-art key photonic devices is given, including the waveguides, light sources, and detectors.
The development of silicon photonics over the past two decades has made silicon the preferential platform for photonic integration. In this paper, we offer our views toward the increasing field of integrated silicon-based near-infrared photonics. A comprehensive overview of the state-of-the-art key photonic devices is given, including the waveguides, light sources, and detectors.
Теги: direct laser beam writing hyperdoped silicon integrated infrared photonics optoelectronics. pulsed laser annealing импульсный лазерный отжиг интегральная фотоника инфракрасного диапазона оптоэлектроника прямая лазерная запись сверхлегированный кремний
Near-Infrared Planar Photonics Based on Hyperdoped Silicon: Prospects
M. S. Kovalev 1, I. M. Podlesnykh 1, 2, K. E. Pevchikh 3, S. I. Kudryashov 1, 2
P. N. Lebedev Physical Institute of the Russian Academy of Sciences, Moscow, Russia
National Research University “Bauman Moscow State Technical University”, Moscow, Russia
Zelenograd Nanotechnology Center JSC, Zelenograd, Moscow, Russia
The development of silicon photonics over the past two decades has made silicon the preferential platform for photonic integration. In this paper, we offer our views toward the increasing field of integrated silicon-based near-infrared photonics. A comprehensive overview of the state-of-the-art key photonic devices is given, including the waveguides, light sources, and detectors.
Keywords: hyperdoped silicon, integrated infrared photonics, pulsed laser annealing, direct laser beam writing, optoelectronics.
Article received: 21.02.2024
Article accepted: 13.03.2024
Introduction
The near-infrared (NIR) range, covering the wavelengths from 0.78 μm to 3 μm [1], is an important technological area for various applications: microelectronics, telecommunications, and optoelectronics. The technical specifications of devices included in the optoelectronic systems and complexes operating in the NIR range are often subject to deterioration due to the increased sensitivity to the thermal fluctuations. Such effects require the development of systems that ensure stable cooling, especially for the radiation receivers used in such systems. The up-to-date commercial NIR-sensitive infrared photodetectors are mostly based on narrow-band-gap semiconductors such as Ge (0.7–1.8 μm), InGaAs (1.1–1.7 μm), and InAs (0.9–3.5 µm), as well as other materials. However, such photodetectors are expensive to be produced and their technology is not compatible with the silicon-based (Si) CMOS process, thus limiting their integration into the Si-optoelectronics. Moreover, silicon is the most common and cost-effective semiconductor material. However, its band gap with the width of 1.12 eV (at 300K) limits the IR photon absorption with the wavelengths of more than 1.1 μm.
The methods and approaches of Si-photonics in the NIR range using the CMOS technology make it possible to integrate a lot of optical elements into a single chip [2]. Moreover, such a technology opens up potential for developing submicron waveguides and packed placement of optical elements on the chip surface that affects the market availability of relevant products and confirms their potential suitability for various fields of technology. Despite these advantages, there are a number of technical limitations in the integration of laser sources, optical amplifiers and insulators, and receivers onto the Si platform.
In this article, we present our views toward the developing field of integrated NIR range photonics. Silicon, used as a substrate, has a unique combination of various specifications, such as large dimensions, strength, chemical and thermal stability, availability and low cost. It is also specified by a high refractive index and has a transmission window in the considered range that allows it to be used as a passive optical material. However, the integration of new materials that are hybrid-integrated with Si is a prerequisite for the integrated NIR photonic systems. This trend is similar to the application of new materials in the microelectronics industry. The purpose of this article is to provide an overview of the state-of-the-art NIR devices that are integrated using the Si-based technologies, including both monolithic and hybrid approaches, as well as a consideration of relevant material technologies beyond the area of silicon.
Waveguides and passive devices
The passive components made from the optically transparent materials play a key role in the photonic integrated circuits. The material transparency in the NIR range begins at the wavelength corresponding to the occurrence of phonon absorption. The material transmission window expansion in this range provides for the replacement of lighter elements with the heavier ones to reduce the typical phonon frequency. This trend is illustrated by Fig. 1 that shows the transmission windows of various optical materials determined as the spectral regions where absorption is <1 dB/cm [3]. This figure demonstrates the ability to make a solid choice of a passive material platform in various parts of the NIR range. For the wavelengths shorter than 2 µm, the standard technologies are silicon-on-insulator (SOI) and silicon nitride (Si3N4).
There are other photonic platforms not listed above. One strategy involves the replacement of lossy silicon oxide cladding with other materials, such as silicon nitride [4] or germanium nitride. Another option is to use germanium-on-silicon (Ge-on-Si or SiGe-on-Si) that provides compatibility with the Si-processing CMOS processes, since the high-quality germanium can be epitaxially grown on Si [5]. In addition, the high refractive index of Ge implies that the Si substrate can be used as a lower cladding. On the other part, the IR-transparent chalcogenides and halides can be deposited monolithically on the Si or dielectric substrates by thermal evaporation or sputtering, while the waveguides consist of two compositions with various refractive indices as the core and cladding. When comparing to the Si or Ge layers, the disadvantage of this approach is that chalcogenides and halides are generally considered incompatible with the CMOS processes. However, the recent publications devoted to the integration of chalcogenide devices with the Si waveguides indicate that such a hybrid configuration can enable smooth integration of alternative materials with a standard Si photonics platform [6].
At present, it is well-known that doping is a simple and efficient technology that allows to change the properties of semiconductors that significantly expands the scope of their application, including in the photonic integrated circuits [7]. The silicon doping with the acceptor (B) or donor (P, As, Sb) impurities from groups III and V of the periodic table allows it to be provided with the required conductivity type and degree that is widely used in optoelectronics to produce a variety of devices. However, there is a need to expand the silicon functionality in the near and mid-IR range (0.78–10 μm), where the base materials for optoelectronics are the narrow-band-gap semiconductors of the A3B5 type (InAs, InSb) and CdHgTe. However, the processing technology for such materials is poorly compatible with the conventional Si technology. A promising approach to increase the Si absorption in the wavelength range of 0.78–10 μm is its hyperdoping, namely introduction of deep level impurities (S, Se, Te, Ti, V, Fe, Co, Ni, Ag, Au) with the ionization energy of 0.1–0.5 eV. In the case of hyperdoping, the impurity concentration exceeds 1019 cm−3 that is significantly (by several orders of magnitude) higher than the equilibrium solubility of these impurities in Si (1015–1016 cm−3) and leads to the development of impurity zones [8]. The availability of such zones promotes the absorption of low-energy photons. The recent studies devoted to Si hyperdoping have already shown promising results in terms of the high level of Si optical absorption in the wavelength range of 0.78–10 μm (up to 50%), as well as obtaining a photoresponse on the photodiode structures up to 500 mA/W at the wavelengths of 1.31 and 1.55 microns.
Table 1 demonstrates the various passive platforms available for the Si-based integrated NIR photonics, including the best results in low loss performance achieved in each category.
Silicon hyperdoping technology
One of the main conditions for ensuring intermediate-band light absorption in the hyperdoped Si samples is development of the high concentrations of deep level impurities (with an ionization energy of about 0.5 eV). Typically, the equilibrium solubility limit for such impurities does not exceed 1016 cm−3. However, this concentration is not enough to generate an intermediate-band in a semiconductor that occurs at the concentrations of about 1019 cm−3. Therefore, when silicon is hyperdoped, the nonequilibrium methods of impurity introduction into the semiconductor are applied [17].
Figure 2 provides various hyperdoping methods that have become popular among the researchers in recent decades [18]. One of the most promising methods is the ion implantation followed by the pulsed laser melting when the pulse duration ranges from 1 ns to 1 000 ns. This method makes it possible to develop a hyperdoped layer with the thickness of up to several hundred nanometers and the required impurity concentration (Fig. 2a). It is important to note that subsequent pulsed laser irradiation leads to restoration of the sample crystal structure and reduction in the radiation-induced defects resulting from the ion implantation. Another method to reduce the radiation-induced defects is annealing that is performed using a flash lamp (Fig. 2b) and is specified by the rapid semiconductor heating and cooling. The duration of lamp exposure is about 1 μs – 1 ms, and this method belongs to the solid-phase epitaxy methods. Although the lamp annealing results in negligible diffusion of impurities and incomplete restoration of the Si crystal structure, this method is an alternative for reducing the radiation-induced defects. Finally, a technologically simpler method involves the preliminary deposition of a thin film of impurity material on the Si surface and subsequent pulsed laser melting of this system (Fig. 2c). As a result of the melting process, comparable in time to the pulse duration, the liquid-phase diffusion of impurity atoms into the semiconductor is occurred that makes it possible to ensure both their high concentration and good crystallinity of the hyperdoped layer. However, in the case of high-intensity laser irradiation, a significant part of the impurity material is ablated, therefore the layer thickness often does not exceed several hundred nanometers.
Light sources
The light source integration on Si poses a significant challenge due to the Si’s existing indirect bandgap. However, there are currently three prospective approaches to solve this issue: heterogeneous integration of cascade and semiconductor lasers, nonlinear frequency generation or conversion, and heteroepitaxy of narrow-band-gap semiconductor compounds on silicon. All these methods have been experimentally tested.
Since the invention of quantum cascade lasers (QCLs) and intermediate-band cascade lasers in the 1990s, their production technology has been significantly advanced. Although the heterogeneous integration of QCLs with Si photonics was demonstrated by Spott [19] at 4.8 μm using a silicon-on-nitride-on-insulator wafer, the development of QCLs operating at the room temperature in the NIR region has not yet been achieved. In this case, the traditional semiconductor lasers based on type I or type II heterostructures are a good alternative. Thus, there are some papers where several examples of heterogeneous integration of IR diode lasers and III–V amplifiers with the silicon waveguides have been proposed, including InP-based Fabry-Perot and DFB lasers emitting in the range of 2.0 μm [20] and 2.3 µm [21], a GaSb-based Fabry-Perot laser operating at 2.38 µm [22], and an InP-based optical amplifier operating at 2.0 µm [23]. The structure of an InP-based silicon multiquantum well laser emitting at 2.3 μm is also well-know that is used as an example of a large group of heterogeneously integrated III–V lasers [21].
Heteroepitaxy, that is, the growth of non-silicon materials on Si, opens up potential for the integration of Si-based devices with the improved optoelectronic properties. In the telecommunications range, heteroepitaxial growth has reached its best level in the development of electrically pumped silicon laser sources based on n-doped Ge [24] and InAs / GaAs quantum dots [25]. Moreover, generation in the near-IR region has also been achieved using GaSb, GeSn and lead compounds. However, the lead compound lasers have limitations in optical pumping and use the external free-space resonators, except for the photonic crystal surface-emitting lasers. The GaSb-on-Si material is probably the most promising in terms of laser device development, while providing the electrically pumped continuous lasing at the room temperature. However, the laser structure is grown on a cut silicon wafer due to the issues with the formation of antiphase domains. It is also necessary to develop the laser integration technologies with other elements of the planar photonic circuits.
In recent years, significant efforts have been made to research and develop the Si-based lasers. Although silicon has an indirect semiconductor junction and is not typically used in the optical devices, there are several examples in the literature devoted to the silicon application to develop the Raman and quantum cascade lasers. Recently [26, 27] a Si laser demonstrating continuous emission at the room temperature at a wavelength of 1.3 μm has been developed. The typical values of threshold current density for laser oscillations were 1.1–2.0 kA / cm2, the power ratio in TE polarization and TM polarization during oscillations was 8 : 1, the optical output power was 50 μW (at a current of 60 mA) and the external differential quantum efficiency was 1%. However, some parameters of such lasers, such as operating temperature, efficiency, and wavelength, do not yet meet the requirements for practical application. Table 2 shows the main specifications of laser structures.
Waveguide-integrated photodetectors
The photodetectors play an important role in converting optical signals into the electronic form while being an integral part of photonic integrated circuits [33]. The receiver’s integration into such circuits provides a number of significant advantages. First, the integration helps improve the signal-to-noise ratio (SNR) by noise suppression. It is achieved due to the fact that various types of noise (fluctuation noise, Johnson noise, and lasing recombination noise) that often limit the detector’s SNR depend linearly on the active detector’s volume. When light is sent to the detector through a waveguide (with the core refractive index n) rather than from free space, the active detector’s volume, and hence its noise, can be reduced by about n2 times without any loss of optical absorption. It has important implications for the NIR detectors made from the narrow-band-gap semiconductors that suffer from the higher intrinsic noise levels. Second, the waveguide-integrated detectors can provide higher bandwidth than their free-space analogues. The detector’s volume decrease also reduces the RC delay and carrier propagation time in the photoelectric detectors. Finally, the waveguide-integrated detectors avoid the compromise between optical absorption and charge carrier collection that is often typical for the free-space detectors. This makes it easier to achieve high quantum efficiency without sacrificing the charge carrier collection efficiency and time constraints of the charge transfer. Thus, the waveguide detectors offer the opportunity to avoid this compromise since the optical path and carrier collection path are orthogonal, whereas these paths often coincide in the free-space detectors.
Four grades of waveguide-integrated detectors have been shown in the NIR range, when the detector active material includes the following: (1) hybrid narrow-band-gap semiconductors, (2) monolithically deposited or grown narrow-band-gap semiconductors, (3) narrow-band-gap semiconductors or van der Waals semimetals, or (4) ion-implanted Si with the intentionally introduced impurities. Table 3 compares the waveguide-integrated NIR detectors based on the latest technology.
Since the 1950s, active researches have been performed aimed at increasing the silicon sensitivity in the IR region. One of the methods is the development of deep level impurities, the acceptor or donor levels of which are close to the middle of the Si band gap (with an ionization energy of about 0.5 eV), when their concentration exceeds the equilibrium solubility values in Si (more than 1019 cm−3; for equilibrium solubility it is about 1015–1016 cm−3). It results in the hyperdoped silicon, namely a material with remarkable photovoltaic properties at the wavelengths of more than 1.1 µm and an absorption of about 103 cm−1 at a wavelength of 1.55 µm. This approach is low cost and simple to be performed without the need for unconventional materials. The nonequilibrium technologies, ion implantation and pulsed laser doping, have also played a significant role in the development of silicon microelectronics and made it possible to achieve the required high concentrations of low-solubility impurities that was impossible when using standard methods, for example, thermal diffusion. However, silicon processing using the ion implantation and pulsed laser doping technologies can cause generation of the radiation-induced defects that increase the likelihood of non-radiative recombination of charge carriers. This effect is undesirable for photodiodes. Therefore, after processing, the post-processing procedure is required to restore the material crystalline structure and activate the dopant impurities. For this purpose, for example, the pulsed laser annealing of silicon is used that makes it possible to partially or completely restore the damaged layer, while transforming the amorphous phase into a quasi-monocrystalline or polycrystalline structure with a high impurity concentration.
The cryogenically cooled external Si and Ge detectors that use the electronic transitions from the impurity states to the conduction band or valence band, including the blocked impurity band detectors, are widely applied for the infrared signal measurement. The studies have demonstrated the performance of SOI detectors (Fig. 4) with implantation of Zn and S ions at the room temperature by taking advantage of the relatively deep levels related to the Zn and S impurities [34]. The waveform SOI detectors with Si+ and Ar+ implants have also been considered. These detectors are fully compatible with the standard CMOS production and avoid application of foreign materials in the Si platform. However, the main disadvantage of these detectors is their low sensitivity occurred due to the weak impurity concentration or absorption mediated by the defect levels. The device operation at the high bias voltage in the avalanche mode can significantly improve its photosensitivity, but this is accompanied by deterioration in the noise performance. Table 3 compares the specifications of waveguide-integrated NIR detectors based on the hyperdoped silicon technology.
Conclusion
It should be noted that most of the described examples of key optoelectronic devices for optoelectronic systems, namely the waveguides, emitters and radiation detectors promising for use in the near-infrared range, are selected from the vast body of references published over the last decade. An analysis of the modern state of production technology for such devices based on the Si platform demonstrates the prospects of their photonic integration in the NIR range and occurrence of a new area in silicon technology, namely the production technology of silicon-based photonic elements. The integration of multiple materials beyond the traditional silicon-based materials is a key to progress in the NIR element design. The integration advances play an important role in overcoming technological limitations, and the transition from optimization of individual photonic devices to the system-level integration opens up great prospects for NIR photonics.
Funding
The results have been obtained as a part of the implementation of state assignment No. FSFN‑2024–0019.
AUTHORS
M. S. Kovalev, Ph.D. in technical sciences, senior researcher, Lebedev Physical Institute of the Academy of Sciences, Moscow, Russia.
ORCID: 0000-0001-5074-0718
I. M. Podlesnykh, research assistant, Lebedev Physical Institute of the Academy of Sciences, engineer of the Bauman Moscow State Technical University, Moscow, Russia.
ORCID: 0000-0003-3381-2972
K. E. Pevchikh, Ph.D. in technical sciences, head of the technological area of integrated photonics, Zelenograd Nanotechnology Center JSC, Zelenograd, Moscow, Russia.
S. I. Kudryashov, doctor of physical and mathematical sciences, senior researcher, Lebedev Physical Institute of the Academy of Sciences, lead researcher, Bauman Moscow State Technical University, Moscow, Russia.
ORCID: 0000-0001-6657-2739
Distribution of functions
between the authors
M. S. Kovalev: idea of the study, discussion of the results, preparation and writing of the manuscript; I. M. Podlesnykh: material collection and processing of results; K. E. Pevchikh: suggestions for area selection, writing and editing of the manuscript; S. I. Kudryashov: work arrangement, discussion of results and editing of the manuscript.
Conflict of interest
The authors declare no conflict of interest. Each member of the team of authors has supplemented the manuscript in his own part of the work, and all team members have taken part in the discussion of results.
M. S. Kovalev 1, I. M. Podlesnykh 1, 2, K. E. Pevchikh 3, S. I. Kudryashov 1, 2
P. N. Lebedev Physical Institute of the Russian Academy of Sciences, Moscow, Russia
National Research University “Bauman Moscow State Technical University”, Moscow, Russia
Zelenograd Nanotechnology Center JSC, Zelenograd, Moscow, Russia
The development of silicon photonics over the past two decades has made silicon the preferential platform for photonic integration. In this paper, we offer our views toward the increasing field of integrated silicon-based near-infrared photonics. A comprehensive overview of the state-of-the-art key photonic devices is given, including the waveguides, light sources, and detectors.
Keywords: hyperdoped silicon, integrated infrared photonics, pulsed laser annealing, direct laser beam writing, optoelectronics.
Article received: 21.02.2024
Article accepted: 13.03.2024
Introduction
The near-infrared (NIR) range, covering the wavelengths from 0.78 μm to 3 μm [1], is an important technological area for various applications: microelectronics, telecommunications, and optoelectronics. The technical specifications of devices included in the optoelectronic systems and complexes operating in the NIR range are often subject to deterioration due to the increased sensitivity to the thermal fluctuations. Such effects require the development of systems that ensure stable cooling, especially for the radiation receivers used in such systems. The up-to-date commercial NIR-sensitive infrared photodetectors are mostly based on narrow-band-gap semiconductors such as Ge (0.7–1.8 μm), InGaAs (1.1–1.7 μm), and InAs (0.9–3.5 µm), as well as other materials. However, such photodetectors are expensive to be produced and their technology is not compatible with the silicon-based (Si) CMOS process, thus limiting their integration into the Si-optoelectronics. Moreover, silicon is the most common and cost-effective semiconductor material. However, its band gap with the width of 1.12 eV (at 300K) limits the IR photon absorption with the wavelengths of more than 1.1 μm.
The methods and approaches of Si-photonics in the NIR range using the CMOS technology make it possible to integrate a lot of optical elements into a single chip [2]. Moreover, such a technology opens up potential for developing submicron waveguides and packed placement of optical elements on the chip surface that affects the market availability of relevant products and confirms their potential suitability for various fields of technology. Despite these advantages, there are a number of technical limitations in the integration of laser sources, optical amplifiers and insulators, and receivers onto the Si platform.
In this article, we present our views toward the developing field of integrated NIR range photonics. Silicon, used as a substrate, has a unique combination of various specifications, such as large dimensions, strength, chemical and thermal stability, availability and low cost. It is also specified by a high refractive index and has a transmission window in the considered range that allows it to be used as a passive optical material. However, the integration of new materials that are hybrid-integrated with Si is a prerequisite for the integrated NIR photonic systems. This trend is similar to the application of new materials in the microelectronics industry. The purpose of this article is to provide an overview of the state-of-the-art NIR devices that are integrated using the Si-based technologies, including both monolithic and hybrid approaches, as well as a consideration of relevant material technologies beyond the area of silicon.
Waveguides and passive devices
The passive components made from the optically transparent materials play a key role in the photonic integrated circuits. The material transparency in the NIR range begins at the wavelength corresponding to the occurrence of phonon absorption. The material transmission window expansion in this range provides for the replacement of lighter elements with the heavier ones to reduce the typical phonon frequency. This trend is illustrated by Fig. 1 that shows the transmission windows of various optical materials determined as the spectral regions where absorption is <1 dB/cm [3]. This figure demonstrates the ability to make a solid choice of a passive material platform in various parts of the NIR range. For the wavelengths shorter than 2 µm, the standard technologies are silicon-on-insulator (SOI) and silicon nitride (Si3N4).
There are other photonic platforms not listed above. One strategy involves the replacement of lossy silicon oxide cladding with other materials, such as silicon nitride [4] or germanium nitride. Another option is to use germanium-on-silicon (Ge-on-Si or SiGe-on-Si) that provides compatibility with the Si-processing CMOS processes, since the high-quality germanium can be epitaxially grown on Si [5]. In addition, the high refractive index of Ge implies that the Si substrate can be used as a lower cladding. On the other part, the IR-transparent chalcogenides and halides can be deposited monolithically on the Si or dielectric substrates by thermal evaporation or sputtering, while the waveguides consist of two compositions with various refractive indices as the core and cladding. When comparing to the Si or Ge layers, the disadvantage of this approach is that chalcogenides and halides are generally considered incompatible with the CMOS processes. However, the recent publications devoted to the integration of chalcogenide devices with the Si waveguides indicate that such a hybrid configuration can enable smooth integration of alternative materials with a standard Si photonics platform [6].
At present, it is well-known that doping is a simple and efficient technology that allows to change the properties of semiconductors that significantly expands the scope of their application, including in the photonic integrated circuits [7]. The silicon doping with the acceptor (B) or donor (P, As, Sb) impurities from groups III and V of the periodic table allows it to be provided with the required conductivity type and degree that is widely used in optoelectronics to produce a variety of devices. However, there is a need to expand the silicon functionality in the near and mid-IR range (0.78–10 μm), where the base materials for optoelectronics are the narrow-band-gap semiconductors of the A3B5 type (InAs, InSb) and CdHgTe. However, the processing technology for such materials is poorly compatible with the conventional Si technology. A promising approach to increase the Si absorption in the wavelength range of 0.78–10 μm is its hyperdoping, namely introduction of deep level impurities (S, Se, Te, Ti, V, Fe, Co, Ni, Ag, Au) with the ionization energy of 0.1–0.5 eV. In the case of hyperdoping, the impurity concentration exceeds 1019 cm−3 that is significantly (by several orders of magnitude) higher than the equilibrium solubility of these impurities in Si (1015–1016 cm−3) and leads to the development of impurity zones [8]. The availability of such zones promotes the absorption of low-energy photons. The recent studies devoted to Si hyperdoping have already shown promising results in terms of the high level of Si optical absorption in the wavelength range of 0.78–10 μm (up to 50%), as well as obtaining a photoresponse on the photodiode structures up to 500 mA/W at the wavelengths of 1.31 and 1.55 microns.
Table 1 demonstrates the various passive platforms available for the Si-based integrated NIR photonics, including the best results in low loss performance achieved in each category.
Silicon hyperdoping technology
One of the main conditions for ensuring intermediate-band light absorption in the hyperdoped Si samples is development of the high concentrations of deep level impurities (with an ionization energy of about 0.5 eV). Typically, the equilibrium solubility limit for such impurities does not exceed 1016 cm−3. However, this concentration is not enough to generate an intermediate-band in a semiconductor that occurs at the concentrations of about 1019 cm−3. Therefore, when silicon is hyperdoped, the nonequilibrium methods of impurity introduction into the semiconductor are applied [17].
Figure 2 provides various hyperdoping methods that have become popular among the researchers in recent decades [18]. One of the most promising methods is the ion implantation followed by the pulsed laser melting when the pulse duration ranges from 1 ns to 1 000 ns. This method makes it possible to develop a hyperdoped layer with the thickness of up to several hundred nanometers and the required impurity concentration (Fig. 2a). It is important to note that subsequent pulsed laser irradiation leads to restoration of the sample crystal structure and reduction in the radiation-induced defects resulting from the ion implantation. Another method to reduce the radiation-induced defects is annealing that is performed using a flash lamp (Fig. 2b) and is specified by the rapid semiconductor heating and cooling. The duration of lamp exposure is about 1 μs – 1 ms, and this method belongs to the solid-phase epitaxy methods. Although the lamp annealing results in negligible diffusion of impurities and incomplete restoration of the Si crystal structure, this method is an alternative for reducing the radiation-induced defects. Finally, a technologically simpler method involves the preliminary deposition of a thin film of impurity material on the Si surface and subsequent pulsed laser melting of this system (Fig. 2c). As a result of the melting process, comparable in time to the pulse duration, the liquid-phase diffusion of impurity atoms into the semiconductor is occurred that makes it possible to ensure both their high concentration and good crystallinity of the hyperdoped layer. However, in the case of high-intensity laser irradiation, a significant part of the impurity material is ablated, therefore the layer thickness often does not exceed several hundred nanometers.
Light sources
The light source integration on Si poses a significant challenge due to the Si’s existing indirect bandgap. However, there are currently three prospective approaches to solve this issue: heterogeneous integration of cascade and semiconductor lasers, nonlinear frequency generation or conversion, and heteroepitaxy of narrow-band-gap semiconductor compounds on silicon. All these methods have been experimentally tested.
Since the invention of quantum cascade lasers (QCLs) and intermediate-band cascade lasers in the 1990s, their production technology has been significantly advanced. Although the heterogeneous integration of QCLs with Si photonics was demonstrated by Spott [19] at 4.8 μm using a silicon-on-nitride-on-insulator wafer, the development of QCLs operating at the room temperature in the NIR region has not yet been achieved. In this case, the traditional semiconductor lasers based on type I or type II heterostructures are a good alternative. Thus, there are some papers where several examples of heterogeneous integration of IR diode lasers and III–V amplifiers with the silicon waveguides have been proposed, including InP-based Fabry-Perot and DFB lasers emitting in the range of 2.0 μm [20] and 2.3 µm [21], a GaSb-based Fabry-Perot laser operating at 2.38 µm [22], and an InP-based optical amplifier operating at 2.0 µm [23]. The structure of an InP-based silicon multiquantum well laser emitting at 2.3 μm is also well-know that is used as an example of a large group of heterogeneously integrated III–V lasers [21].
Heteroepitaxy, that is, the growth of non-silicon materials on Si, opens up potential for the integration of Si-based devices with the improved optoelectronic properties. In the telecommunications range, heteroepitaxial growth has reached its best level in the development of electrically pumped silicon laser sources based on n-doped Ge [24] and InAs / GaAs quantum dots [25]. Moreover, generation in the near-IR region has also been achieved using GaSb, GeSn and lead compounds. However, the lead compound lasers have limitations in optical pumping and use the external free-space resonators, except for the photonic crystal surface-emitting lasers. The GaSb-on-Si material is probably the most promising in terms of laser device development, while providing the electrically pumped continuous lasing at the room temperature. However, the laser structure is grown on a cut silicon wafer due to the issues with the formation of antiphase domains. It is also necessary to develop the laser integration technologies with other elements of the planar photonic circuits.
In recent years, significant efforts have been made to research and develop the Si-based lasers. Although silicon has an indirect semiconductor junction and is not typically used in the optical devices, there are several examples in the literature devoted to the silicon application to develop the Raman and quantum cascade lasers. Recently [26, 27] a Si laser demonstrating continuous emission at the room temperature at a wavelength of 1.3 μm has been developed. The typical values of threshold current density for laser oscillations were 1.1–2.0 kA / cm2, the power ratio in TE polarization and TM polarization during oscillations was 8 : 1, the optical output power was 50 μW (at a current of 60 mA) and the external differential quantum efficiency was 1%. However, some parameters of such lasers, such as operating temperature, efficiency, and wavelength, do not yet meet the requirements for practical application. Table 2 shows the main specifications of laser structures.
Waveguide-integrated photodetectors
The photodetectors play an important role in converting optical signals into the electronic form while being an integral part of photonic integrated circuits [33]. The receiver’s integration into such circuits provides a number of significant advantages. First, the integration helps improve the signal-to-noise ratio (SNR) by noise suppression. It is achieved due to the fact that various types of noise (fluctuation noise, Johnson noise, and lasing recombination noise) that often limit the detector’s SNR depend linearly on the active detector’s volume. When light is sent to the detector through a waveguide (with the core refractive index n) rather than from free space, the active detector’s volume, and hence its noise, can be reduced by about n2 times without any loss of optical absorption. It has important implications for the NIR detectors made from the narrow-band-gap semiconductors that suffer from the higher intrinsic noise levels. Second, the waveguide-integrated detectors can provide higher bandwidth than their free-space analogues. The detector’s volume decrease also reduces the RC delay and carrier propagation time in the photoelectric detectors. Finally, the waveguide-integrated detectors avoid the compromise between optical absorption and charge carrier collection that is often typical for the free-space detectors. This makes it easier to achieve high quantum efficiency without sacrificing the charge carrier collection efficiency and time constraints of the charge transfer. Thus, the waveguide detectors offer the opportunity to avoid this compromise since the optical path and carrier collection path are orthogonal, whereas these paths often coincide in the free-space detectors.
Four grades of waveguide-integrated detectors have been shown in the NIR range, when the detector active material includes the following: (1) hybrid narrow-band-gap semiconductors, (2) monolithically deposited or grown narrow-band-gap semiconductors, (3) narrow-band-gap semiconductors or van der Waals semimetals, or (4) ion-implanted Si with the intentionally introduced impurities. Table 3 compares the waveguide-integrated NIR detectors based on the latest technology.
Since the 1950s, active researches have been performed aimed at increasing the silicon sensitivity in the IR region. One of the methods is the development of deep level impurities, the acceptor or donor levels of which are close to the middle of the Si band gap (with an ionization energy of about 0.5 eV), when their concentration exceeds the equilibrium solubility values in Si (more than 1019 cm−3; for equilibrium solubility it is about 1015–1016 cm−3). It results in the hyperdoped silicon, namely a material with remarkable photovoltaic properties at the wavelengths of more than 1.1 µm and an absorption of about 103 cm−1 at a wavelength of 1.55 µm. This approach is low cost and simple to be performed without the need for unconventional materials. The nonequilibrium technologies, ion implantation and pulsed laser doping, have also played a significant role in the development of silicon microelectronics and made it possible to achieve the required high concentrations of low-solubility impurities that was impossible when using standard methods, for example, thermal diffusion. However, silicon processing using the ion implantation and pulsed laser doping technologies can cause generation of the radiation-induced defects that increase the likelihood of non-radiative recombination of charge carriers. This effect is undesirable for photodiodes. Therefore, after processing, the post-processing procedure is required to restore the material crystalline structure and activate the dopant impurities. For this purpose, for example, the pulsed laser annealing of silicon is used that makes it possible to partially or completely restore the damaged layer, while transforming the amorphous phase into a quasi-monocrystalline or polycrystalline structure with a high impurity concentration.
The cryogenically cooled external Si and Ge detectors that use the electronic transitions from the impurity states to the conduction band or valence band, including the blocked impurity band detectors, are widely applied for the infrared signal measurement. The studies have demonstrated the performance of SOI detectors (Fig. 4) with implantation of Zn and S ions at the room temperature by taking advantage of the relatively deep levels related to the Zn and S impurities [34]. The waveform SOI detectors with Si+ and Ar+ implants have also been considered. These detectors are fully compatible with the standard CMOS production and avoid application of foreign materials in the Si platform. However, the main disadvantage of these detectors is their low sensitivity occurred due to the weak impurity concentration or absorption mediated by the defect levels. The device operation at the high bias voltage in the avalanche mode can significantly improve its photosensitivity, but this is accompanied by deterioration in the noise performance. Table 3 compares the specifications of waveguide-integrated NIR detectors based on the hyperdoped silicon technology.
Conclusion
It should be noted that most of the described examples of key optoelectronic devices for optoelectronic systems, namely the waveguides, emitters and radiation detectors promising for use in the near-infrared range, are selected from the vast body of references published over the last decade. An analysis of the modern state of production technology for such devices based on the Si platform demonstrates the prospects of their photonic integration in the NIR range and occurrence of a new area in silicon technology, namely the production technology of silicon-based photonic elements. The integration of multiple materials beyond the traditional silicon-based materials is a key to progress in the NIR element design. The integration advances play an important role in overcoming technological limitations, and the transition from optimization of individual photonic devices to the system-level integration opens up great prospects for NIR photonics.
Funding
The results have been obtained as a part of the implementation of state assignment No. FSFN‑2024–0019.
AUTHORS
M. S. Kovalev, Ph.D. in technical sciences, senior researcher, Lebedev Physical Institute of the Academy of Sciences, Moscow, Russia.
ORCID: 0000-0001-5074-0718
I. M. Podlesnykh, research assistant, Lebedev Physical Institute of the Academy of Sciences, engineer of the Bauman Moscow State Technical University, Moscow, Russia.
ORCID: 0000-0003-3381-2972
K. E. Pevchikh, Ph.D. in technical sciences, head of the technological area of integrated photonics, Zelenograd Nanotechnology Center JSC, Zelenograd, Moscow, Russia.
S. I. Kudryashov, doctor of physical and mathematical sciences, senior researcher, Lebedev Physical Institute of the Academy of Sciences, lead researcher, Bauman Moscow State Technical University, Moscow, Russia.
ORCID: 0000-0001-6657-2739
Distribution of functions
between the authors
M. S. Kovalev: idea of the study, discussion of the results, preparation and writing of the manuscript; I. M. Podlesnykh: material collection and processing of results; K. E. Pevchikh: suggestions for area selection, writing and editing of the manuscript; S. I. Kudryashov: work arrangement, discussion of results and editing of the manuscript.
Conflict of interest
The authors declare no conflict of interest. Each member of the team of authors has supplemented the manuscript in his own part of the work, and all team members have taken part in the discussion of results.
Readers feedback
