Germanium ion-implanted silicon waveguide photodiode and method for manufacturing the same
Germanium ion-implanted silicon waveguides address the responsivity challenges in silicon photodetectors by introducing defect states for improved sub-bandgap linear photodetection, achieving significant responsivity enhancements in the telecom band.
Patent Information
- Authority / Receiving Office
- WO · WO
- Patent Type
- Applications
- Current Assignee / Owner
- Filing Date
- 2025-08-07
- Publication Date
- 2026-04-09
AI Technical Summary
Existing silicon photodetectors face challenges in achieving sufficient responsivity at telecom band wavelengths without added complexity or loss, particularly in sub-bandgap linear photodetection, due to issues with defect-state-absorption and excess optical loss.
Employing germanium ion-implantation in silicon waveguides to introduce defect states for enhanced sub-bandgap linear photodetection, utilizing Ge+ ions that naturally substitute silicon atoms at lattice sites, thereby improving responsivity in the 1200-1600 nm wavelength range.
The Ge+-implanted silicon waveguide photodiodes exhibit enhanced responsivity and reduced dark current, achieving up to three orders of magnitude improvement in linear photodetection compared to conventional methods, with minimal excess free carrier absorption.
Smart Images

Figure CN2025113142_09042026_PF_FP_ABST
Abstract
Description
GERMANIUM ION-IMPLANTED SILICON WAVEGUIDE PHOTODIODE AND METHOD FOR MANUFACTURING THE SAMEInventors: Yue NIU; Qianni ZHANG; and Wing On POONTechnical Field:
[0001] The present invention relates to germanium ion-implanted (Ge+-implanted) silicon waveguide photodiodes; and more particularly to Ge ion-implanted silicon waveguide photodiodes for all-silicon sub-bandgap linear photodetection at the telecommunications O-and C-bands and methods for manufacturing the same.Background:
[0002] Silicon photonics as a promising technology field that has been intensely studied due to its lower power consumption, higher aggregate bandwidth compared with electronic parts, and lower manufacturing cost for its CMOS compatibility. Silicon on insulator (SOI) platforms form the foundation of large-scale photonic integrated circuits (PICs) because of the high index contrast for a tight optical confinement. Silicon, with an indirect energy bandgap of 1.12 eV provides a transparency window at telecommunication wavelengths of 1310 nm and 1550 nm. Conversely, it lacks a band-to-band photon detection at wavelengths above ~1100 nm. All-silicon sub-bandgap photodetections have been studied for more than a decade. In intrinsic silicon, two-photon-absorption (TPA) provides an interband nonlinear photodetection at a moderate optical power (e.g., > greater than few mW) . Surface-state-absorption (SSA) leveraging the dangling Si bonds or intrinsic defects formed at the Si / SiO2 interface contributes to a sub-bandgap linear photodetection. However, SSA shows an extremely low responsivity because of a limited overlap between optical modes and surface defects. Related research on contactless integrated photonic probes (CLIPP) employing the SSA generated carriers exhibits a low responsivity and requires a demanding conductance measurement.
[0003] Various materials have been heterogeneously integrated on the SOI platform for infrared photodetection. Epitaxially-grown germanium on silicon has been widely studied for photodetection at C-and L-telecom bands, demonstrating a high responsivity exceeding 1 A·W-1 and a dark current of <10 nA upon -3 V. However, the fabrication of high-quality Ge / Si interfaces is required in order to suppress excess optical loss; fabrication of such high-quality interfaces is difficult. Two-dimensional materials such as graphene or black phosphorus and III-V compound materials have also been integrated on silicon utilizing heteroepitaxy or heterogeneous technologies, showing a high response at the telecom-band. However, the fabrication time and cost are considerable. The dark current and the excess optical absorption is relatively large.
[0004] Defect-state-absorption (DSA) for photodetection has been explored due to its CMOS compatible fabrication process. Using ion implantation, defects can be introduced into bulk silicon, maximally overlapping with the optical modes to achieve a significantly enhanced photon response.
[0005] Ion implantation with species such as B+ and P- can lead to elevated free carrier concentrations, resulting in excess free carrier absorption (FCA) loss. In contrast, self-implantation using silicon ions has been shown to mitigate these issues. Implanted silicon atoms can occupy lattice sites after thermal annealing, minimizing the generation of free carriers while enabling enhanced photon absorption through the introduction of deep-level defect states.
[0006] However, these current approaches fail to deliver sufficient responsivity at low bias without added complexity or loss. Accordingly, there is a need for a photodetector with improved defect state absorption that operates in the telecom band wavelength region over 1100 nm and demonstrates linear performance. The present invention addresses this need.Summary of Invention:
[0007] It is an objective of the present invention to provide devices and methods to address the aforementioned shortcomings and unmet needs in the state of the art.
[0008] In the present disclosure, a Ge+-implanted silicon waveguide photodetector for the sub-bandgap linear photodetection is provided at ~1310 nm and ~1550 nm wavelength bands. Germanium (Ge) , a CMOS compatible Group IV element that is heavier than Ar, is able to naturally replace a silicon atom at a lattice site and generate more defects per ion, thereby delivering a higher response of the sub-bandgap linear photodetection at O-and C-telecom bands (0.97 eV and 0.8 eV) .
[0009] The present invention is characterized by the use of Ge=-implanted silicon for infrared photodetection, which exhibits enhanced responsivity compared to conventional ion-implanted silicon photodiodes using B+, P-, or Ar+ ions.
[0010] In accordance with a first aspect of the present invention, a germanium doped silicon waveguide photodetector for linear photodetection in telecommunication wavelength bands in the 1200-1600 nm range is provided. The photodetector includes a substrate and a silicon waveguide. The substrate is selected from a silicon substrate or a substrate having at least one layer of silicon formed on the substrate. A silicon waveguide is formed on or in a silicon-including portion of the substrate. The silicon waveguide includes a coupling region for coupling to a source of optical signals having wavelengths between 1200 nm and 1600 nm. One or more regions of the silicon waveguide are implanted with germanium ions to form a germanium ion-implanted region. The germanium ions are implanted at a dose and implantation energy sufficient to introduce defect states into the silicon waveguide for sub-bandgap linear photodetection. This renders the germanium ion-implanted region of the silicon waveguide absorptive of optical signals between 1200 nm and 1600 nm. One or more electrodes communicate with the silicon waveguide to collect charge carriers generated by absorbed optical signals in the germanium ion-implanted region.
[0011] In accordance with a second aspect of the present invention, a Ge+-implanted silicon waveguide photodiode is provided. The photodiode includes a silicon-on-insulator (SOI) substrate, a rib waveguide structure, a Ge+-implanted region, a first low-temperature oxide (LTO) layer, and a second LTO layer. The SOI substrate includes a buried oxide layer and a silicon layer formed over the buried oxide layer. The rib waveguide structure is formed in the silicon layer. The Ge+-implanted region is located within the rib waveguide structure, in which the Ge+-implanted region is formed by ion implantation at a dose in a range from 1012 cm-2 to 1015 cm-2 such that the Ge+-implanted region is formed by introducing defect states into the silicon for sub-bandgap linear photodetection. The first LTO layer and the second LTO layer are formed over the silicon layer and defining an implantation window that is aligned with the Ge+-implanted region within the rib waveguide structure.
[0012] In accordance with a third aspect of the present invention, a method for manufacturing a Ge+-implanted silicon waveguide photodiode is provided. The method includes step as follows: providing a SOI substrate comprising a buried oxide layer and a silicon layer formed over the buried oxide layer; performing a patterning process to form a rib waveguide structure in the silicon layer; depositing a LTO layer over the silicon layer and forming an implantation window therein by removing at least one portion of the LTO layer, in which the implantation window is aligned with the Ge+-implanted region in the rib waveguide structure; depositing a layer of sacrificial oxide on the top of the Ge+-implanted region in the rib waveguide structure; and performing a Ge+ implantation into the rib waveguide structure through the implantation window at an implantation dose in a range from 1012 cm-2 to 1015 cm-2, thereby forming a Ge+-implanted region within the rib waveguide structure to introduce defect states into the silicon for sub-bandgap photodetection.Brief Description of Drawings:
[0013] Embodiments of the invention are described in more details hereinafter with reference to the drawings, in which:
[0014] FIG. 1 shows energy levels of defect states introduced into silicon by different implanted ions;
[0015] FIG. 2A is a perspective view of a Ge+-implanted silicon waveguide photodetector according to an embodiment;
[0016] FIG. 2B is a cross-sectional schematics of a Ge+-implanted silicon waveguide photodiode according to some embodiments of the present invention;
[0017] FIG. 3A and FIG. 3B show Finite-Element-Method (FEM) simulations of the electric field intensity of a fundamental TE mode in the waveguide at 1310 nm and 1550 nm wavelengths, respectively;
[0018] FIG. 3C shows Stopping Range of Ions in Matter (SRIM) modeling modeling results for Ge+ ion implantation into silicon.
[0019] FIG. 4 provides Table for SRIM modeling of various ion implantations;
[0020] FIGS. 5A-5H illustrate a fabrication process flow for a Ge+-implanted silicon waveguide photodiode according some embodiments of the present invention;
[0021] FIG. 6 shows the top-view and cross-sectional micrographs, FEM simulations of optical modes in a fabricated Ge+-implanted silicon waveguide, Ge+implantation dose layout across a 4-inch SOI wafer, scanning-electron micrograph (SEM) of the doping regions, and the Ge+-implanted region with its implantation window;
[0022] FIG. 7A-7D show the measured current–voltage (I-V) characteristics of Ge+-implanted silicon photodiodes under a bias ranging from 0 V to -5 V;
[0023] FIG. 8A shows a schematic diagram of the experimental setup used to characterize Ge+-implanted silicon photodiodes at the 1550 nm and 1310 nm wavelengths;
[0024] FIG. 8B shows the measured excess loss and photocurrent (upon a bias voltage of -3 V) spectra of a 250 μm-long Ge+-implanted silicon waveguide PD at the 1310 nm and 1550 nm wavelength bands, in which the excess loss spectra is obtained by normalizing the transmission spectrum to that of a control intrinsic silicon waveguide of the same length (with p / n doped regions in the silicon slab) ;
[0025] FIG. 9A and FIG. 9B plot the photocurrent as a function of the estimated on-chip power for a 250 μm-long Ge+-implanted silicon photodiode at a Ge+ dose of 4×1014 cm-2 and for a control intrinsic silicon waveguide photodiode at the 1310 nm and 1550 nm wavelengths, respectively, in which the bias voltage is -3 V;
[0026] FIG. 9C and FIG. 9D show the power-dependent photocurrent responses (upon a bias voltage of -3 V) of Ge+-implanted silicon photodiodes at implantation doses of 5×1012 cm-2 and 3×1015 cm-2 at the 1550 nm and 1310 nm wavelengths, respectively;
[0027] FIG. 10A and FIG. 10B compare the wavelength-dependent responsivity and internal quantum efficiency of Ge+-implanted silicon photodiodes with those of B+-, P--, and Ar+-implanted silicon photodiodes in the 1310 nm and 1550 nm wavelength bands;
[0028] FIG. 11A shows the measured illuminated current and dark current upon bias voltages ranging from –2 V to –24 V in Ge+-implanted silicon photodiodes at the 1310 nm and 1550 nm wavelengths;
[0029] FIG. 11B and FIG. 11C plot the bias-dependent responsivities of Ge+-implanted silicon photodiodes compared with those of B+-, P--, and Ar+-implanted silicon photodiodes at the 1310 nm and 1550 nm wavelengths, respectively;
[0030] FIG. 12 is a cross-sectional view of a Ge+-implanted silicon waveguide photodiode according to some embodiments of the present invention;
[0031] FIG. 13 is a cross-sectional view of a Ge+-implanted silicon waveguide photodiode according to some embodiments of the present invention;
[0032] FIG. 14 is a cross-sectional view of a Ge+-implanted silicon waveguide photodiode according to some embodiments of the present invention; and
[0033] FIG. 15 is a cross-sectional view of a Ge+-implanted silicon waveguide photodiode according to some embodiments of the present invention.Detailed Description of the Invention:
[0034] In the present invention, germanium-doped silicon waveguides are used in a variety of structures to create photodetectors demonstrating linear performance in telecom wavelength bands over approximately 1100 nm. The term “photodetector” as used herein, relates to any light-absorbing structure that generates a photocurrent, regardless of its implementation as a particular device. Examples of photodetector structures include non-junction detectors (employing a doped silicon waveguide and electrodes for photocurrent collection) , photodiodes (e.g., p-i-n or p-n junction based photodiodes) and field-effect phototransistors (e.g., JFET or MOSFET type) . In the following description, germanium ion-implanted (Ge+-implanted) silicon waveguide photodiodes are set forth as preferred examples of photodetectors due to the ease of integration with existing CMOS technology. However, it is understood that the germanium-doped silicon waveguides may be used in any of these structures. It will be apparent to those skilled in the art that modifications, including additions and / or substitutions may be made without departing from the scope and spirit of the invention. Specific details may be omitted so as not to obscure the invention; however, the disclosure is written to enable one skilled in the art to practice the teachings herein without undue experimentation.
[0035] FIG. 1 shows the energy levels of defect states introduced into silicon by different implanted ions. Ge+ implantation introduces defect states at approximately +0.51 eV and –0.27 eV, facilitating the linear sub-bandgap photodetection for photon energies of 0.8 to 0.9 eV which are located within the C-band and the O-band of telecom wavelength bands, respectively. Compared to B+ , P- and Ar+, Ge+ as a CMOS-compatible Group-IV element can naturally substitute a silicon atom at a lattice site, thus Ge ion implantation will introduce a minimal density of extra free carriers upon an exhaustive annealing process while generating more deep levels in silicon, making it well-suited for infrared photodetectors on silicon. Telecom networks use a variety of infrared wavelength bands for fiber optic communications due to optical fibers that have been optimized for transmission in these bands as well as rare-earth-based optical amplifiers that amplify optical signals for a small infrared region. Commercially important telecom bands include the O-band (1260-1360 nm) for short-reach data communication and C-band (1530-1565 nm) for long-haul, wavelength division multiplexed system. The germanium doping enables both O and C-band photodetection.
[0036] FIG 2A schematically depicts a germanium doped silicon waveguide photodetector 1000 for linear photodetection in telecommunication wavelength bands in the 1200-1600 nm range according to an embodiment. The photodetector 1000 is formed on a substrate 1010 and it may be silicon or a silicon layer formed on a substrate, such as an SiO2 substrate or other insulating substrate (for example, SOI substrate) . A silicon waveguide 1020 is formed in or on the silicon layer / silicon substrate 1010. The silicon waveguide 1020 includes a coupling region 1030 for coupling to a source of optical signals 1040 having wavelengths between 1200 nm and 1600 nm. One or more regions 1050 of the silicon waveguide are implanted with germanium ions to form a germanium ion-implanted region. The germanium ions are implanted at a dose and implantation energy sufficient to introduce defect states into the silicon waveguide for sub-bandgap linear photodetection as seen in the bandgap diagram of FIG. 1. As the silicon waveguide 1020 can have a variety of shapes and thicknesses, it is understood that the dose and implantation energy will vary accordingly. Further, the predetermined size and shape of the ion-implanted region will also impact the energy dose as is discussed in more detail below. This ion-implantation renders the germanium ion-implanted region of the silicon waveguide absorptive of optical signals between 1200 nm and 1600 nm. One or more electrodes 1060 communicate with the silicon waveguide to collect charge carriers generated by absorbed optical signals in the germanium ion-implanted region. The electrodes 1060 may be positioned, for example, on either side of the Ge=-implanted region of the silicon waveguide. The charge carrier collection is routed as desired to power measurement equipment or processors depending on the photodetector application. At least one cladding layer 1070 is disposed over and / or around the silicon waveguide 1020 to provide optical confinement and electrical isolation. The cladding layer 1070 may comprise silicon dioxide or other dielectric materials compatible with standard CMOS processing.
[0037] FIG. 2B is a cross-sectional schematics of a Ge+-implanted silicon waveguide photodiode (PD) 100 used as a photodetector according to some embodiments of the present invention. The waveguide photodiode 100 includes a substrate 110, a first lightly doped region (p--doped region) 120, a second lightly doped region (n--doped region) 122, a first heavily doped region (p+-doped region) 124, a second heavily doped region (n+-doped region) 126, a first low-temperature oxide (LTO) layer 130, and a second LTO layer 132, a Ge+-implanted region 140, a first metal contact 150, and a second metal contact 152.
[0038] The substrate 110 may be a silicon-on-insulator (SOI) substrate, including a buried oxide layer (e.g., silicon oxide SiO2) 112 and a silicon layer 114 formed thereon. The buried oxide layer 112 electrically isolates the waveguide structure from the underlying bulk silicon, providing optical confinement and reduced substrate leakage. A portion of the silicon layer 114 serves as a silicon waveguide layer 116 and includes a rib-type structure which has a width W1, in which the width W1 is in a range from 400 nm to 600 nm (e.g., 500 nm) . The rib-type structure may have a thickness T1 and it rests on a silicon slab portion of the silicon layer 114 with a thickness T2, defining a rib waveguide. In some embodiments, the thickness T1 is in a range from 100 nm to 170 nm (e.g., 135 nm) , and the thickness T2 is in a range from 50 nm to 120 nm (e.g., 85 nm) . The rib waveguide provides the optical transmission and absorption region of the waveguide photodiode 100. In various embodiments, the rib waveguide may be configured, based on optical mode requirements, to support a transverse-electric (TE) polarized optical mode, a transverse-magnetic (TM) polarized optical mode, a fundamental optical mode, or a multimode configuration.
[0039] Various regions of the silicon layer 114 are heavily or lightly doped. As used herein, the term “lightly doped” refers to a doping concentration that is, relative to a “heavily doped” region, lower by approximately 1 to 3 orders of magnitude. Alternatively, “lightly doped” may refer to a level of doping which, in combination with an adjacent heavily doped region, is sufficient to extend the depletion region of a p-i-n structure under reverse bias, thereby contributing to the breakdown voltage enhancement.
[0040] The first lightly doped region (p--doped region) 120 is formed adjacent to the silicon waveguide layer 116 on one side of the intrinsic region in the silicon layer 114. The first lightly doped region 120 has a doping concentration suitable for extending the depletion region and enhancing the breakdown voltage of the photodiode. In some embodiments, the first lightly doped region 120 has a width W2 (e.g., from 0 μm to 3 μm, such as 2 μm) and a p-type doping dose of approximately 5×1013 cm-2. The first lightly doped region 120 is spaced by a distance D1 away from a waveguide core of the silicon waveguide layer 116 to reduce excess free carrier absorption (FCA) . In one embodiment, the distance D1 is in a range from 0 nm to 500 nm (e.g., 200 nm) . Herein, a distance D1 of 0 nm indicates that the boundary of the lightly doped region is laterally aligned without any spacing with the silicon waveguide sidewall.
[0041] The second lightly doped region (n--doped region) 122 is formed on the opposite side of the intrinsic region in the silicon layer 114, relative to the first lightly doped region 120. The second lightly doped region 122 may have similar dimensions and doping characteristics as those of the p--doped region. In some embodiments, the second lightly doped region 122 has a width W3 (e.g., from 0 μm to 3 μm, such as 2 μm) and an n-type doping dose of approximately 5×1013 cm-2. The second lightly doped region 122 is spaced 200 nm away from the waveguide core of the silicon waveguide layer 116 to reduce excess FCA. The second lightly doped region 122 is symmetrically positioned with respect to the first lightly doped region 120 to define a balanced p-i-n junction structure.
[0042] The first heavily doped region (p+-doped region) 124 is formed adjacent to the first lightly doped region 120 and configured to provide low-resistance electrical contact. The first heavily doped region 124 has a width W4 (e.g., from 1 μm to 8 μm, such as 6 μm) and a p-type doping dose of approximately 3×1015 cm-2. The second heavily doped region (n+-doped region) 126 is formed adjacent to the second lightly doped region 122 and is similarly configured for electrical contact. The second heavily doped region 126 also has a width W5 (e.g., from 1 μm to 8 μm, such as 6 μm) and an n-type doping dose of approximately 3×1015 cm-2.
[0043] By this configuration, the p+ / p- / intrinsic / n- / n+ structure enables a wide depletion region for efficient carrier collection and high breakdown voltage. The heavily doped p+ and n+ regions provide low-resistance contacts, while the lightly doped regions extend the depletion region for breakdown voltage enhancement.
[0044] Although the illustration and the description thereof recites a p+ / p- / intrinsic / n- / n+ structure, it is understood that other types of PN junction configurations may also be employed, such as p+ / intrinsic / n+ and p+ / n+ structures. Accordingly, references to a width of “0 μm” for the lightly doped regions indicate that these regions may be omitted in some embodiments. For example, the silicon layer 114 includes a doped region configured as one of a p+ / p- / intrinsic / n- / n+ junction, a p+ / intrinsic / n+ junction, or a p+ / n+ junction, depending on electrical or optical performance requirements.
[0045] The first LTO layer 130 is formed over the first lightly doped region 120 of the silicon layer 114 and functions as a dielectric isolation layer. The second LTO layer 132 is formed over the second lightly doped region 122 and also functions as a dielectric isolation layer. The first LTO layer 130 and the second LTO layer 132 each have a thickness T3 (e.g., the thickness T3 is in a range from 500 nm to 1000 nm, such as 650 nm) , and together define the lateral boundaries of an implantation window 134 in a width W6, which provides access for ion implantation while maintaining electrical insulation. In one embodiment, the width W6 is in a range from 0.5 μm to 0.6 μm, such as 0.55 μm)
[0046] The Ge+-implanted region 140 is located in the rib-type structure of the silicon waveguide layer 116, between the first lightly doped region 120 and the second lightly doped region 122. The Ge+-implanted region 140 is formed by performing Ge+ ion implantation through an implantation window that is defined by the first LTO layer 130 and the second LTO layer 132, introducing defect states within the silicon bandgap to facilitate sub-bandgap linear photodetection.
[0047] Typically, the implantation dose for Ge+ ions is 1012 cm-2 to 1015 cm-2. In Examples of doses include implantation doses of Ge+ ions for the Ge+-implanted region 140 is approximately 5×1012cm-2, 4×1014cm-2, or 3×1015 cm-2, and the ion projective range is centered around a range from 65 nm to 70 nm (e.g., 68 nm) beneath the silicon surface (the top-most surface of the silicon waveguide layer 116) . Herein, the “projective range” refers to the depth at which the implanted Ge+ with a peak ion concentration come to rest within the silicon after implantation. A projective range centered around a range from 65 nm to 70 nm (e.g., 68 nm) beneath the silicon surface indicates that the peak concentration of implanted ions is positioned to overlap with the optical mode in the waveguide core, thereby enhancing photon absorption. However, it is understood that other implantation depths may be selected depending on the selected dimensions of the waveguide and other device parameters. Implantation energy is proportional to the depth of implantation and therefore is determined based on the desired implantation depth. For example, in some embodiments, to spatially align the defect states with the optical field maxima of the fundamental transverse-electric mode (i.e., TE00 mode) and thereby maximize absorption efficiency, the projected range of the Ge+ ion implantation is configured to be centered between 55 nm and 75 nm beneath the silicon surface. In some embodiments, around the projected range center within the silicon waveguide layer 116, the Ge+ ion concentration reaches its maximum and decays (or symmetrically decays) toward both shallower and deeper regions.
[0048] The first metal contact 150 and the second metal contact 152 are formed over the first heavily doped region 124 and the second heavily doped region 126, respectively, to provide electrical connection to external circuitry. In some embodiments, the first metal contact 150 and the second metal contact 152 include an aluminum (Al) layer deposited and patterned to overlap the first heavily doped region 124 and the second heavily doped region 126, thereby forming ohmic contacts. The first metal contact 150 and the second metal contact 152 may be used to apply a reverse bias across the photodiode and to collect photocurrent generated within the Ge+-implanted region 140.
[0049] In some embodiments, the waveguide structure of the Ge+-implanted silicon waveguide photodiode 100 may be implemented as a straight waveguide or a ring waveguide, and may be cladded with materials such as air, silica, or silicon nitride. When air is used as the cladding medium, the resulting structure may correspond to that shown in FIG. 2B. Alternatively, when a solid dielectric material is employed as the cladding, a corresponding cladding layer may be formed over the structure illustrated in FIG. 2B, such that the layer at least covers the silicon waveguide layer 116.
[0050] FIG. 3A and FIG. 3B show Finite-Element-Method (FEM) simulations of the electric field intensity of the fundamental TE00 mode in a silicon waveguide core at wavelengths of 1310 nm and 1550 nm, respectively. FIG. 3C illustrates the Stopping Range of Ions in Matter (SRIM) modeling of Ge+ implantation into silicon upon an implantation energy of 130 keV.
[0051] EXAMPLE
[0052] In one embodiment, a sacrificial oxide layer (e.g., a sacrificial oxide layer with a thickness ranging from 20 nm to 30 nm, such as a 25 nm-thick sacrificial oxide layer) is provided on top of the silicon waveguide (e.g., the rib-type structure of the silicon waveguide layer 116) to protect the surface from implantation-induced damage that may lead to excessive scattering loss. The projective range of Ge+ ions is targeted at approximately 68 nm within the 220-nm-thick silicon layer (i.e., a sum of 85 nm and 135 nm) . In one embodiment, the implantation energy of the Ge+ ion implantation is selected to be 130 keV with a tilt angle of 7°. Herein, the tilt angle of 7° refers to the angle between the incident Ge+ ion beam and the normal direction of the silicon surface during implantation. The peak ion concentration is located approximately at the core of the ridge silicon waveguide, while being slightly offset from the center of the optical mode, balancing between achieving sufficient overlap with the optical mode and introducing minimal excess FCA loss. For comparison, control devices are designed and fabricated, in which silicon photodiodes are implanted with various ions including B+, P-, and Ar+. Additional details of the SRIM modeling for these ion implantations are provided in Table of FIG. 4.
[0053] FIG. 5A, FIG. 5B, FIG. 5C, FIG. 5D, FIG. 5E, FIG. 5F, FIG. 5G, and FIG. 5H illustrate a process flow for manufacturing a Ge+-implanted silicon waveguide photodiode according to some embodiments of the present invention.
[0054] In FIG 5A, a standard 4-inch silicon-on-insulator (SOI) wafer 500 is provided, including a 220-nm-thick silicon layer 502 over a 3-μm-thick buried oxide (BOX) layer 504 that is formed on a silicon substrate 506. The silicon layer 502 serves as the active region for the formation of the waveguide and photodiode structure. A rib-type silicon waveguide structure 508 is formed by patterning and etching the silicon layer 502 using photolithography and dry etching processes. The waveguide layout is defined to support a fundamental TE optical mode. In one embodiment, an i-line photolithography stepper with a resolution of less than 0.5 μm and an overlay alignment accuracy of approximately 0.1 μm is employed during the patterning process.
[0055] In FIG. 5B, a sacrificial oxide layer 510 is formed on the silicon layer 502 using plasma-enhanced chemical vapor deposition (PECVD) . The sacrificial oxide layer 510 functions to protect the silicon surface from ion implantation-induced damage in subsequent steps, thereby preventing excessive surface roughness or optical scattering loss in the waveguide region.
[0056] In FIG. 5C, p-type and n-type doped regions are formed on opposing sides of the waveguide core of the rib-type silicon waveguide structure 508. Ion implantation is performed using suitable dopants to define the p+, p-, n-, and n+ regions, thereby establishing a lateral p-i-n junction configuration. In one embodiment, after the ion implantation, the sacrificial oxide layer 510 is removed.
[0057] In FIG. 5D, a LTO layer 512 having a thickness of approximately 650 nm is deposited over the surface of the silicon layer 502. Subsequently, metal contacts (e.g., aluminum contacts) 513 are formed by evaporation and etching, using an evaporator and an aluminum etcher, to define the metal pads for electrical interfacing. In one embodiment, a forming gas annealing (FGA) process is performed at 400℃ for 30 minutes to improve the contact quality between the metal and the underlying silicon.
[0058] In FIG. 5E, an implantation window 514 is formed by removing a portion of the LTO layer 512, thereby exposing the rib-type silicon waveguide structure 508 in the target implantation region.
[0059] In FIG. 5F, a 25-nm-thick sacrificial oxide layer 516 is deposited using tetraethyl orthosilicate (TEOS) as the precursor to protect the surface of the rib-type silicon waveguide structure 508 during subsequent ion implantation. In one embodiment, the sacrificial oxide layer 516 is formed by a PECVD process.
[0060] In FIG. 5G, a high-temperature-resistant photoresist 518 is coated onto the resulting structure of FIG. 5F (e.g., disposed on the sacrificial oxide layer 516) . A what we call “branch exposure method” is employed using an i-line stepper to pattern the photoresist and define the implantation window 514 in the LTO layer 512, exposing the target implantation region that covers the rib-type silicon waveguide structure 508 where Ge+ ions 520 are to be implanted upon various doses in sequence before wafer dicing. Thereafter, Ge+ ion implantation is performed through the defined implantation window 514 using an implanter at an implantation energy of 130 keV with a tilt angle of 7°. In one embodiment, the tilt angle of 7° is achieved by adjusting the incident angle of the ion beam relative to the normal direction of the wafer (e.g., the wafer 500) surface during the implantation process.
[0061] In various embodiments, the implantation dose ranges from approximately 5×1012 cm-2 to 3×1015 cm-2 to introduce defect states into the silicon for sub-bandgap photodetection. In some embodiments, the implantation dose is selected to be exactly 5×1012 cm-2, 4×1014 cm-2, 3×1015 cm-2, or another value in a range from 1012 cm-2 to 1015 cm-2. These different doses are employed for comparative analysis of device performance under varying implantation conditions. The peak ion concentration is located near the waveguide core of the rib-type silicon waveguide structure 508.
[0062] In FIG. 5H, the photoresist 518 and the sacrificial oxide layer 516 are removed, and a post-implantation annealing process is conducted in a in inert atmosphere such as nitrogen (N2) atmosphere at an elevated temperature. In this example, the annealing is performed in nitrogen at 350℃ for 10 minutes. It is understood that annealing temperatures and times may vary based on the device configuration, layer thickness, and implantation parameters. In various embodiments, the annealing is performed for optimizing the manufactured device’s performance, including the dark currents and responsivities.
[0063] FIG. 6, including parts (a) , (b) , (c) , and (d) , illustrates the optical image, layout, cross-sectional-view and top-view SEMs of a Ge+-implanted silicon waveguide photodiode, showing the photodiode structure, implantation window, and doping region dimensions. Specifically, part (a) of FIG. 6 presents a top-view micrograph of a 250 μm-long Ge+-implanted silicon waveguide photodiode, and the FEM simulated electric field intensity distributions of the fundamental TE00 mode in our fabricated silicon waveguides (accounting for the misalignment of the implantation window, at 1310 nm and 1550 nm wavelengths) , indicating a good optical confinement in our fabricated silicon waveguides; part (b) of FIG. 6 shows the Ge+ implantation dose layout on a 4-inch wafer, as referenced in the process flow of FIGS. 5A-5H; part (c) of FIG. 6 illustrates a scanning electron micrograph (SEM) of the cross section of a Ge+-implanted photodiode. The center shift of the Ge+ implantation window is attributed to limited overlay alignment accuracy and non-uniformity in the wet etching process. Part (d) of FIG. 6 provides a top-view SEM image of the Ge+-implanted region in the silicon waveguide photodiode. An approximately 550 nm-wide Ge+ implantation window is formed on top of the silicon waveguide.
[0064] FIGS. 7A-7D show the measured current–voltage (I-V) characteristics of Ge+-implanted silicon photodiodes under a bias voltage ranging from 0 V to -5 V. Device characterization is performed using a Perfect Lab Probe Station equipped with a Keysight B1500A Semiconductor Analyzer.
[0065] FIG. 7A shows a graph for comparing the performance of Ge+-implanted silicon photodiodes at a dose of 4×1014 cm-2 before and after annealing, as well as that of an intrinsic silicon photodiode. A two-order-of-magnitude increase in dark current is observed after Ge+ implantation, rising from approximately 100 pA to ~10 nA at -3 V. After annealing in nitrogen at 350℃ for 10 minutes, the dark current decreases to 0.8 nA at -3 V. FIG. 7B presents the I-V curves of Ge+-implanted silicon photodiodes at implantation doses of 5×1012 cm-2 and 3×1015 cm-2, both showing a nearly consistent dark current in the range of 0.8 to 0.9 nA upon a bias voltage of -3 V.
[0066] FIG. 7C shows a graph for comparing the dark current measured upon a bias voltage of -3 V for intrinsic and Ge+-implanted silicon waveguide photodiodes in various photodiode lengths, following annealing at 350℃ for 10 minutes. FIG. 7D shows I-V curves of 250 μm-long Ge+-implanted silicon photodiodes alongside control devices implanted with B+, P-, and Ar+ ions, all at a dose of 4×1014 cm-2 and annealed under the same conditions of 350℃ for 10 minutes. B+-, P--, and Ar+-implanted silicon photodiodes exhibit dark currents ranging from approximately 20 pA to 100 pA at a bias voltage of -1 V, comparable to those of intrinsic silicon photodiodes. In contrast, a Ge+-implanted silicon photodiodes exhibits a dark current approximately one-order-of-magnitude higher than those of the control devices after annealing, indicating an increased number of thermally excited electron-hole pairs in the depletion region due to more defect states introduced by the Ge+ implantation than the lighter ion-types B+, P-and Ar+ implantations.
[0067] FIG. 8A shows a schematic diagram of the experimental setup used to characterize Ge+-implanted silicon photodiodes at the 1550 nm and 1310 nm wavelengths. Two Santec tunable lasers are employed to provide optical signals in the 1550 nm and 1310 nm bands, respectively. Erbium-doped fiber amplifiers (EDFAs) are used to amplify the optical input at 1550 nm. A variable optical attenuator (VOA) enables continuous adjustment of the input power, while a 99: 1 beam splitter (BS) taps out 1%of the optical power for monitoring purposes.
[0068] A polarization controller (PC) is utilized to ensure excitation of a fundamental TE-polarized mode. A symmetric edge-coupling configuration is adopted using lensed single-mode fibers (SMFs) to couple light in and out of the photonic chip.
[0069] A source meter is used to apply a reverse voltage bias to the photodiode through a ground-signal-ground (G-S-G) RF probe and to acquire the real-time photocurrent signal. Transmitted light from the chip, as well as the reference signal from the 1%port of the beam splitter, is detected by commercial InGaAs photodetectors (PD1 and PD2) , with outputs monitored on an oscilloscope.
[0070] FIG. 8B shows the measured spectra of the excess loss and of the photocurrent (upon a bias voltage of –3 V) of a 4×1014cm-2 Ge+-implanted silicon waveguide PD (annealed at 350 ℃) at 1310 nm and 1550 nm wavelength bands. The excess loss spectra of the Ge+-implanted silicon waveguide are obtained by normalizing the transmission spectrum to that of the control intrinsic silicon waveguide of the same length (with p / n doped regions in the silicon slab) . And the transmission spectra are obtained by normalizing the transmitted power to the input power. We measure an excess loss of ~3 dB (~1 dB) at the 1550 nm (1310 nm) wavelength, corresponding to an optical absorption loss of 0.012 dB·μm-1 (0.004 dB·μm-1) . To achieve this level of excess loss, the factors include the lateral width of the Ge+-implanted region, the projective range of the implanted ions within the silicon, the width of the implantation window, and the tilt angle during Ge+ ion implantation. Proper alignment among these factors is essential to achieve a balance between enhanced photoresponse and acceptable optical absorption loss within the waveguide photodiode.
[0071] FIG. 9A and FIG. 9B plot the photocurrent as a function of the estimated on-chip power for a 250 μm-long Ge+-implanted silicon photodiode at a Ge+ dose of 4×1014 cm-2 and for a control intrinsic silicon waveguide photodiode at the 1310 nm and 1550 nm wavelengths, respectively. The bias voltage is –3 V. The on-chip power is estimated by subtracting the measured facet loss from the input power monitored by photodetector PD1. A variable optical attenuator (VOA) is used to control the on-chip power within approximately 1 mW. The error bar of the estimated on-chip power reflects the zero shift obtained from a least squares linear fit. A linear fit with a fixed zero intercept is applied to extract the slope of the photocurrent versus estimated on-chip power.
[0072] Based on this analysis, a linear responsivity of 8.64 ± 0.11 mA·W-1 at a bias voltage -3 V is obtained at 1550 nm. At 1310 nm wavelength, the responsivity increases to 31.2 ± 0.5 mA·W-1 at -3 V, representing an enhancement of more than 5 dB relative to the 1550 nm result.
[0073] A second-order polynomial function is used to fit the photocurrent response, from which the first-order coefficient is extracted to yield linear responsivities of 0.0068 ± 0.0002 mA·W-1 and 0.0370 ± 0.0009 mA·W-1 at a bias voltage of -3 V at the 1550 nm and 1310 nm wavelengths, respectively. These results indicate extremely weak photodetection based on surface-state absorption (SSA) in intrinsic silicon. In contrast, Ge+-implanted silicon photodiodes demonstrate defect-state absorption (DSA) -enhanced linear photodetection with up to three orders of magnitude improvement at both 1550 nm and 1310 nm wavelengths.
[0074] A second-order coefficient from the fitted response suggests a measurable two-photon absorption (TPA) effect in intrinsic silicon, with a TPA-induced photocurrent of approximately 18.4 μA·mW-2 at a bias voltage -3 V under ~1 mW optical input. The TPA contribution in intrinsic silicon is power-dependent and has been previously characterized to yield 1.9 μA·mW-2 at 100 μW and 25 μA·mW-2 at 25 mW input power.
[0075] As used herein, a “bias voltage of -3 V” refers to a voltage applied across the anode and cathode terminals of the photodiode in a direction that reverse biases the p-n junction, thereby enabling the sweep-out of photogenerated carriers and facilitating the measurement of photocurrent under sub-bandgap optical illumination. For example, a Ge=-implanted silicon waveguide photodiode (e.g., in the waveguide photodiode 100) includes a lateral p–i–n junction, a Ge=-implanted region that spatially overlaps with the optical mode of the waveguide, and metal contacts electrically connected to the p+ and n+ regions. The metal contacts are configured to apply a bias voltage of -3 V across the junction during operation, under which the device exhibits a measurable sub-bandgap photocurrent response.
[0076] FIG. 9C and FIG. 9D show the power-dependent photocurrent responses of Ge+-implanted silicon photodiodes at implantation doses of 5×1012 cm-2 and 3×1015 cm-2 under 1550 nm and 1310 nm wavelengths, respectively. The results indicate that a Ge+ dose of 4×1014 cm-2 provides superior responsivity compared to the other two conditions, which exceeds that of the 3×1015 cm-2 device (achieving 4.57 ± 0.01 mA·W-1) , and is notably higher than the 2.66 ± 0.02 mA·W-1 value measured for the 5×1012 cm-2 device. A similar trend is observed at the 1310 nm wavelength, where responsivity enhancements of approximately 6.3 dB and 4.4 dB are achieved over the corresponding values at the 1550 nm wavelength, for the 5×1012 cm-2 and 3×1015 cm-2 doses, respectively.
[0077] FIG. 10A compares the wavelength-dependent responsivities of Ge+-implanted silicon photodiodes with those of B+-, P--, and Ar+-implanted silicon photodiodes in the 1310 nm and 1550 nm wavelength bands. Upon an identical implantation dose of 4×1014 cm-2 and upon a bias voltage of -3 V, the B+-, P--, and Ar+-implanted silicon photodiodes exhibit responsivities of 1.35 ± 0.01 mA·W-1, 1.60 ± 0.01 mA·W-1, and 1.93 ± 0.01 mA·W-1, respectively, at the 1550 nm wavelength.
[0078] FIG. 10B compares the wavelength-dependent internal quantum efficiency of Ge+-implanted silicon photodiodes with those of B+-, P--, and Ar+-implanted silicon photodiodes in the 1310 nm and 1550 nm wavelength bands. Upon a bias voltage of -3 V, an enhanced internal quantum efficiency is obtained in a Ge+- implanted silicon waveguide PD compared to those of B+-, P--, and Ar+-implanted silicon waveguide PDs by factors of 4.9 to 16.8.
[0079] As shown in FIG. 10A, Ge+-implanted silicon photodiodes exhibit enhanced photoresponse compared with B+-, P--, and Ar+-implanted silicon photodiodes across both the O-and C-band telecommunication wavelengths. A consistent decreasing trend in responsivity with increasing wavelength is observed among Ge+-, B+-, P--, Ar+-implanted, and intrinsic silicon photodiodes. In the case of the Ge+-implanted silicon photodiode, the responsivity decreases by approximately 6 dB from 1260 nm to 1620 nm. Downward trends are observed, where a 5 dB drop is observed at the 1.9 μm wavelength in a Si+-implanted device and a 3 dB drop is observed at the 2 μm wavelength in a B+-implanted device, both relative to the 1.55 μm wavelength reference.
[0080] As shown in FIG. 10B, the optical confinement factor in the fabricated silicon waveguide is estimated to decrease by approximately 0.9 dB over the wavelength range of 1260 nm to 1620 nm. The majority of the responsivity difference arises from the wavelength-dependent internal quantum efficiency.
[0081] FIG. 11A shows the measured illuminated current and dark current upon bias voltages ranging from -2 V to -24 V in a Ge+-implanted silicon photodiode. The illuminated currents are measured upon an estimated on-chip power of ~0.27 mW at the 1310 nm wavelength, and of ~0.25 mW at the 1550 nm wavelength. The ratio of the illuminated current at the 1310 nm wavelength to the dark current is nearly four orders of magnitude and drops to ~156 upon a bias of -6 V, indicating the onset of the avalanche effect..
[0082] FIG. 11B plots the bias-dependent responsivity of Ge+-implanted silicon photodiodes, compared with those of B+-, P--, and Ar+-implanted silicon photodiodes at the 1310 nm wavelength. An avalanche responsivity of 0.408 ± 0.006 A·W-1 is obtained at a bias voltage of –24 V in a Ge+-implanted silicon photodiode (implantation dose: 4×1014 cm-2) . Upon an identical implantation dose of 4×1014 cm-2, the B+-, P--, and Ar+-implanted silicon photodiodes exhibit responsivities of 0.015 ± 0.001 mA·W-1 (-22 V) , 0.086 ± 0.001 mA·W-1 (-22 V) , and 0.175 ± 0.001 mA·W-1 (-28 V) , respectively.
[0083] FIG. 11C plots the bias-dependent responsivity of Ge+-implanted silicon photodiodes, compared with those of B+-, P--, and Ar+-implanted silicon photodiodes at the 1550 nm wavelength. An avalanche responsivity of 0.041 ± 0.001 A·W-1 is obtained at a bias voltage of –28 V in a Ge+-implanted silicon photodiode (implantation dose: 4×1014 cm-2) . Upon an identical implantation dose of 4×1014 cm-2, the B+-, P--, and Ar+-implanted silicon photodiodes exhibit responsivities of 0.004 ± 0.0001 mA·W-1 (-22 V) , 0.031 ± 0.0001 mA·W-1 (-22 V) , and 0.034 ± 0.0001 mA·W-1 (-28 V) , respectively.
[0084] In summary, Ge+-implanted silicon waveguide photodiodes have been designed, fabricated, and experimentally characterized at O-and C-band telecommunication wavelengths. A responsivity of 8.64 ± 0.11 mA·W-1 at 1550 nm and 31.2 ± 0.5 mA·W-1 at 1310 nm is measured in a 250 μm-long Ge+-implanted silicon waveguide photodiode with an implantation dose of 4×1014 cm-2, upon a bias voltage of -3 V and a dark current of approximately 0.8 nA.
[0085] Ge+-implanted silicon photodiodes exhibit enhanced responsivity (upon a bias voltage of -3 V) and internal quantum efficiency compared to intrinsic silicon and B+-, P--, and Ar+-implanted silicon photodiodes across both O-and C-band wavelengths. In avalanche mode, a responsivity of 0.408 ± 0.006 A·W-1, and of 41.6 ± 1.7 mA·W-1 is achieved at 1310 nm wavelength upon a bias voltage of -24 V, and at 1550 nm wavelength upon a bias voltage of -28 V, respectively.
[0086] Moreover, as expressed per unit length, the Ge+-implanted silicon waveguide photodetector developed in the present invention achieves responsivities of 124.8 mA·W-1·mm-1 at 1310 nm and 31.2 mA·W-1·mm-1 at 1550 nm, both under a low CMOS-compatible bias voltage of -3 V. These performance levels exceed those of other ion-implanted all-silicon photodetectors, for example, including B+ (16.4 mA·W-1·mm-1) , P- (18.8 mA·W-1·mm-1) , and Ar+ (24.4 mA·W-1·mm-1) , by factors ranging from approximately 1.5 to 4.9. In addition, the Ge+-implanted device features ultralow absorption loss (0.004–0.012 dB·μm-1) and a low dark current of ~0.8 nA, outperforming Si+-implanted photodetectors that typically suffer from higher dark currents (ranging from tens of nA to μA levels) . Accordingly, compared to related technologies (e.g., which includes conventional epitaxial Ge / Si photodetectors, III-V-on-silicon hybrid devices, and avalanche-enhanced resonant detectors) , the Ge+-implanted photodetector of the present invention offers a favorable trade-off between responsivity, dark current, and power consumption. Taken together, these results demonstrate that the proposed Ge+-implanted photodetector not only matches but surpasses the performance of comparable photodetectors, achieving a commercially acceptable level in terms of responsivity, linearity, power efficiency, and integration scalability.
[0087] Furthermore, to meet varying requirements while achieving the desired device performance, the profile of the LTO layer can be adjusted to accommodate different process conditions, thereby enabling high reliability or yield in the Ge+ ion implantation.
[0088] FIG. 12 is a cross-sectional view of a Ge+-implanted silicon waveguide photodiode 600 according to some embodiments of the present invention. The Ge+- implanted silicon waveguide photodiode 600 has a configuration similar to that of the Ge+-implanted silicon waveguide photodiode 100, except that the first and second LTO layers 630 and 632 have slanted sidewalls.
[0089] The slanted sidewalls of the first and second LTO layers 630 and 632 define an inverted trapezoidal implantation window 634. The sidewalls of the first and second LTO layers 630 and 632 face the Ge+-implanted region 640 formed in the rib-type structure of the silicon waveguide layer 616. Each sidewall of the first and second LTO layers 630 and 632 includes an upper slanted portion and a lower vertical portion. The Ge+-implanted region 640 is located between vertically oriented sidewalls, such that the bottom ends of the slanted sidewalls remain positioned above the top surface of the Ge+-implanted region 640 (i.e., a boundary interface between the upper slanted portion and the lower vertical portion is positioned above the top surface of the Ge+-implanted region 640) . The width of the implantation window 634 defined by the bottom ends of the slanted sidewalls remains the width W6.
[0090] FIG. 13 is a cross-sectional view of a Ge+-implanted silicon waveguide photodiode 700 according to some embodiments of the present invention. The Ge+- implanted silicon waveguide photodiode 700 has a configuration similar to that of the Ge+-implanted silicon waveguide photodiode 100, except that the first and second LTO layers 730 and 732 have slanted sidewalls.
[0091] The slanted sidewalls of the first and second LTO layers 730 and 732 define an inverted trapezoidal implantation window 734. The sidewalls of the first and second LTO layers 730 and 732 face the Ge+-implanted region 740 formed in the rib-type structure of the silicon waveguide layer 716. In the illustrated embodiment, each sidewall of the first and second LTO layers 730 and 732 is formed as a continuous slanted profile extending from the top surface to the bottom surface of the respective LTO layer. The Ge+-implanted region 740 is located between vertically oriented sidewalls, and the implantation window 734 defined by the bottom ends of the slanted sidewalls maintains the width W6.
[0092] FIG. 14 is a cross-sectional view of a Ge+-implanted silicon waveguide photodiode 800 according to some embodiments of the present invention. The Ge+-implanted silicon waveguide photodiode 800 has a configuration similar to that of the Ge+-implanted silicon waveguide photodiode 100, except that the first and second LTO layers 830 and 832 are formed to be closer to each other.
[0093] The first and second LTO layers 830 and 832 are formed with vertical sidewalls and are laterally positioned closer to the Ge+-implanted region 840 in the rib-type structure of the silicon waveguide layer 816. The first and second LTO layers 830 and 832 define an implantation window 834. In some embodiment, the first and second LTO layers 830 and 832 cover the silicon waveguide layer 816 and make contact with the sidewalls of the Ge+-implanted region 840 (i.e., making contact with the opposite sidewalls of the rib waveguide structure) . Under conditions enabling ultra-high-precision implantation, this configuration facilitates improved spatial confinement of the implanted ions and reduces the likelihood of misalignment or lateral spread during the Ge+ ion implantation process, thereby enhancing the overall precision and uniformity of the implantation profile.
[0094] FIG. 15 is a cross-sectional view of a Ge+-implanted silicon waveguide photodiode 900 according to some embodiments of the present invention. The Ge+- implanted silicon waveguide photodiode 900 has a configuration similar to that of the Ge+-implanted silicon waveguide photodiode 100, except that the first and second LTO layers 930 and 932 include upper slanted portions and lower vertical portions.
[0095] The upper slanted portions of the first and second LTO layers 930 and 932 define an inverted trapezoidal implantation window 934. The lower vertical portions of the first and second LTO layers 930 and 932 cover the silicon waveguide layer 916 and make contact with the sidewalls of the Ge+-implanted region 940 (i.e., making contact with the opposite sidewalls of the rib waveguide structure) . This configuration provides both mechanical stability and enhanced angular access for ion implantation. The slanted upper sidewalls of the first and second LTO layers 930 and 932 may improve ion entry at a tilt angle, while the narrower vertical bottom region maintains spatial alignment with the Ge+-implanted region 940.
[0096] It should be understood that the physical values, parameters, and dimensions described herein (such as implantation dose, energy, length, width, and depth) may vary within a reasonable tolerance range due to fabrication and process variations. For example, an implantation dose of 4×1014 cm-2 may encompass values within ±10%of the nominal dose, including 3.6×1014 cm-2 to 4.4×1014 cm-2. In some embodiments, tolerances within ±1%, ±5%, or ±10%may be acceptable depending on the specific process requirements and design margins. Unless explicitly stated otherwise, such variations are considered to fall within the scope of the present disclosure.
[0097] Spatial references such as “on, ” “above, ” “below, ” and similar terms are defined relative to a component or plane as shown in the figure. These terms are for illustration only and do not limit the actual arrangement, provided the described embodiments retain their intended benefits.
[0098] It should be noted that while various structures are depicted as approximately rectangular in the illustrations, their actual shapes may differ in practice due to fabrication conditions. These shapes may include curves, rounded edges, or variations in thickness. The use of straight lines and right angles in the figures is merely a representational convenience for depicting layers and features.
[0099] In this disclosure, the terms “a, ” “an, ” and “the” should be interpreted to include both singular and plural forms unless explicitly specified otherwise by the context. Additionally, when describing embodiments, a component positioned “on” or “over” another component can refer to cases where the two components are directly in contact or where one or more intermediate components are situated between them.
[0100] The foregoing description of the present invention has been provided for the purposes of illustration and description. It is not intended to be exhaustive or to limit the invention to the precise forms disclosed. Many modifications and variations will be apparent to the practitioner skilled in the art.
[0101] The embodiments were chosen and described in order to best explain the principles of the invention and its practical application, thereby enabling others skilled in the art to understand the invention for various embodiments and with various modifications that are suited to the particular use contemplated.
[0102] PRACTICAL PARAMETERS AND CORRESPONDING DEVICE PERFORMANCE
[0103] In the examples of present invention, all-silicon sub-bandgap photodetection at O-and C-telecom bands is provided using germanium-ion-implanted silicon waveguide photodiodes. Within an estimated on-chip optical power of approximately 1 mW, a responsivity of 8.64 ± 0.11 mA·W-1 at 1550 nm and 31.2 ± 0.5 mA·W-1 at 1310 nm is measured, with a dark current of 0.8 nA upon a bias voltage of -3 V, in a 250 μm-long device implanted with Ge+ at a dose of 4×1014 cm-2. Enhanced responsivity is observed compared to intrinsic silicon, as well as B+-, P--, and Ar+-implanted silicon photodiodes. In avalanche mode, a responsivity of 0.408 ± 0.006 A·W-1, and of 41.6 ± 1.7 mA·W-1 is achieved at 1310 nm wavelength upon a bias voltage of -24 V, and at 1550 nm wavelength upon a bias voltage of -28 V, respectively. The Ge+-implanted silicon photodiodes are suitable for use in photonic networks as photomonitors, offering improved photoresponse and a simplified fabrication process.
Claims
1.A germanium doped silicon waveguide photodetector for linear photodetection in telecommunication wavelength bands in a 1200-1600 nm range, comprising:a substrate selected from a silicon substrate or a substrate having at least one layer of silicon formed thereon;a silicon waveguide formed on or in a silicon-including portion of the substrate, wherein the silicon waveguide includes a coupling region for coupling to a source of optical signals having wavelengths between 1200 nm and 1600 nm;one or more region of the silicon waveguide being implanted with germanium ions to form a germanium ion-implanted region, the germanium ions implanted at a dose and implantation energy sufficient to introduce defect states into the silicon waveguide for sub-bandgap linear photodetection to render the germanium ion-implanted region of the silicon waveguide absorptive of optical signals between 1200 nm and 1600 nm; andone or more electrodes communicating with the silicon waveguide to collect charge carriers generated by absorbed optical signals in the germanium ion-implanted region.2.A Ge+-implanted silicon waveguide photodiode detecting wavelengths in telecommunications wavelength bands of approximately 1200 nm to 1600 nm, comprising:a silicon-on-insulator (SOI) substrate comprising a buried oxide layer and a silicon layer formed over the buried oxide layer;a waveguide structure formed in the silicon layer;a Ge+-implanted region located within the waveguide structure, wherein the Ge+-implanted region is formed by ion implantation at a dose in a range from 5×1012 cm-2 to 3×1015 cm-2 such that the Ge+-implanted region is formed by introducing defect states into the silicon for sub-bandgap linear photodetection; andfirst and second electrodes positioned adjacent to the Ge+-implanted region to collect charge carriers generated by absorbed optical signals in the germanium ion-implanted region.3.The photodiode of claim 2, wherein the waveguide structure is configured to support a transverse-electric (TE) polarized optical mode, a transverse-magnetic (TM) polarized optical mode, a fundamental optical mode, or a multimode configuration.4.The photodiode of claim 2, wherein the silicon layer comprises a doping profile selected from a group consisting of a p+ / p- / intrinsic / n- / n+ junction, a p+ / intrinsic / n+ junction, and a p+ / n+ junction.5.The photodiode of claim 2, wherein rhe silicon layer comprises a doping profile configured as a p+ / p- / intrinsic / n- / n+ junction, and the photodiode further comprises:a first lightly doped region and a second lightly doped region laterally adjacent to opposite sides of the rib waveguide structure;a first heavily doped region adjacent to the first lightly doped region and a second heavily doped region adjacent to the second lightly doped region; anda pair of metal contacts formed over the first and second heavily doped regions, respectively, for electrical connection.6.The photodiode of claim 2, wherein the first and second LTO layers comprise slanted sidewalls that define an inverted trapezoidal cross-sectional profile for the implantation window.7.The photodiode of claim 2, wherein each of the first and second LTO layers comprises a sidewall having an upper slanted portion and a lower vertical portion, wherein a boundary interface between the upper slanted portion and the lower vertical portion is positioned above a top surface of the Ge+-implanted region.8.The photodiode of claim 2, wherein each of the first and second LTO layers comprises a sidewall facing the Ge+-implanted region, and the sidewalls of the first and second LTO layers extend continuously from top surfaces to bottom surfaces of the first and second LTO layers with a slanted profile.9.The photodiode of claim 2, wherein each of the first and second LTO layers comprises a sidewall facing the Ge+-implanted region and making contact with opposite sidewalls of the rib waveguide structure.10.The photodiode of claim 2, wherein the photodiode has a responsivity of at least 8.64 ± 0.11 mA·W-1 at a wavelength of 1550 nm upon a bias voltage of -3 V and at least 31.2 ± 0.5 mA·W-1 at a wavelength of 1310 nm upon a bias voltage of -3 V.11.A method for manufacturing a Ge+-implanted silicon waveguide photodiode, comprising:providing a silicon-on-insulator (SOI) substrate comprising a buried oxide layer and a silicon layer formed over the buried oxide layer;performing a patterning process to form a rib waveguide structure in the silicon layer;depositing a low-temperature oxide (LTO) layer over the silicon layer and forming an implantation window therein by removing at least one portion of the LTO layer, wherein the implantation window is aligned with the rib waveguide structure; andperforming a Ge+ ion implantation into the rib waveguide structure through the implantation window at an implantation dose in a range from 5×1012 cm-2 to 3×1015 cm-2, thereby forming a Ge+-implanted region within the rib waveguide structure to introduce defect states into the silicon for sub-bandgap linear photodetection.12.The method of claim 11, wherein the implantation window has a width greater than a width of the Ge+-implanted region.13.The method of claim 11, wherein the patterning process is performed such that the rib waveguide structure is configured to support a transverse-electric (TE) polarized optical mode, a transverse-magnetic (TM) polarized optical mode, a fundamental optical mode, or a multimode configuration.14.The method of claim 11, further comprising:depositing a sacrificial oxide layer with a thickness over the rib waveguide structure prior to the performing the Ge+ ion implantation.15.The method of claim 11, wherein the Ge+ ion implantation is performed at an implantation dose in a range from 1012 cm-2 to 1015 cm-2.16.The method of claim 11, wherein the Ge+-implanted region is formed to have a responsivity of 8.64 ± 0.11 mA·W-1 at 1550 nm and 31.2 ± 0.5 mA·W-1 at the 1310 nm wavelength upon a bias voltage of -3 V.17.The method of claim 11, wherein the implantation window is defined by slanted sidewalls of the LTO layer to form an inverted trapezoidal opening for the implantation window.
Citation Information
Patent Citations
Optical waveguide detector and optical module
CN105655417A
Waveguide type photoelectric detector and preparation method thereof
CN112201723A
Photodetector and method of manufacture thereof
CN112534590A
Vertical germanium-silicon photoelectric detector and preparation method thereof
CN117239002A
Photodetector apparatus and method of detecting light
US20240222542A1