Doping is a critical means to tune the properties of semiconductors^[1-3]. Nowadays researchers have been increasingly adopting extreme measures to gain added freedom for the tuning of the properties of semiconductors. Incorporating impurities into semiconductors with concentrations exceeding their solubility (i. e., hyperdoping) is exactly an extreme measure^[4-7]. Novel properties have indeed been obtained by using hyperdoping for all kinds of semiconductors. For silicon (Si), B hyperdoping has rendered superconductivity^[8]. When Si is hyperdoped with chalcogen^[9] and transition metal elements^[10], it effectively absorbs near-infrared light with energy below the band gap^[11-13]. In the nanoscale regime B or P hyperdoping enables localized surface plasmon resonance (LSPR) in the mid-infrared region for Si^[14,15]. Hyperdoping-induced novel properties of Si have been significantly expanding the fields of the applications of Si^[16-18].

Hyperdoped Si was initially prepared by using ion implantation followed by pulsed laser annealing^[19,20]. Doping methods such as gas immersion laser doping (GILD) subsequently appeared^[21,22]. In this review, we first introduce the hyperdoping methods. The electrical and optical properties of hyperdoped Si in different forms such as bulk Si, Si nanocrystals, Si nanowires and Si thin films are then discussed. We also demonstrate various devices like infrared photodetectors and solar cells based on hyperdoped Si. Finally, we outline the future development of the hyperdoping of Si and novel applications of hyperdoped Si.

To date, several methods including ion implantation followed by pulsed laser melting or flash lamp annealing and laser doping have been used to hyperdope Si. We now introduce typical hyperdoping methods as follows.

Until now, one of the common hyperdoping methods is ion implantation combined with subsequent heat treatment like rapid thermal annealing (RTA) or pulsed laser melting (PLA)^[23]. The process of ion implantation is to ionize and focus dopants into an ion beam in a vacuum system, then the dopant ions are injected into Si by electric field acceleration to form an implanted layer with special properties, which was proposed as early as 1969^[24]. Ion implantation with easy automation is suitable for large-area doping. The concentration of implanted ions is controllable, which is not limited by the equilibrium solid solubility.

The cooling rate of thermal annealing is pivotal for hyperdoping. Traditionally, erbium (Er)/O doped Si obtained after standard RTA cannot be used due to strong nonradiative recombination^[25]. Zhaoet al.^[26] proposed a new deep cooling (DC) method to obtain Er/O-hyperdoped Si, which had a greater cooling rate (1000 °C/s) compared to standard RTA. Compared with the samples fabricated by the traditional RTA process, photoluminescence of Er/O hyperdoped Si was two orders of magnitude higher. Secondary ion mass spectrometry (SIMS) measurements showed that the maximum concentrations of Er(O) can reach 7.5 × 10²⁰ (2.0 × 10²¹ ) cm^–3, far more exceeding its solubility of 5 × 10¹⁷ (2.75 × 10¹⁸) cm^–3 in Si. Analogously, Huet al.^[27] prepared Se-hyperdoped Si through ion implantation followed by furnace annealing in a nitrogen flow of 2.5 L/min at 700 °C for 30 min. The peak concentration of Se obtained by SIMS was 7.4 × 10²⁰ cm^–3, which was far beyond the solubility of 10¹⁷ cm^–3.

Compared to traditional RTA process, PLA possesses highly accurate doping depth and position, as well as small thermal diffusion displacement of doped atoms^[28,29]. The application of PLA enables rapid recrystallization of Si by liquid phase epitaxy after ion implantation, which results in a supersaturation of impurities and the emergence of novel optoelectronic properties. The influence of the laser shot number on the concentration of dopant atoms and carriers was studied by Bobet al.^[30]. Hyperdoping of deep-level impurities such as transition metals in Si using PLM has aroused great interest among researchers^[31,32]. Limet al.^[33] successfully fabricated Ag- and Ti-hyperdoped Si through ion implantation followed by PLA. Moreover, flash lamp annealing (FLA) has become a novel technology for thermal treatment^[34,35]. FLA is powered by an Xe lamp in the millisecond range, which is well suited for industrial production due to the “one-shot-one-wafer” processes^[36]. The dopant concentration and substitutional fraction of Se-hyperdoped Si^[37] prepared through ion implantation followed by FLA could be up to 2.3% and 70%, respectively. Wanget al.^[38] demonstrated that Te-hyperdoped Si fabricated by ion implantation and FLA was suitable for infrared photodetector due to the emergence of mid-infrared LSPR.

Nanosecond (ns) and femtosecond (fs) laser doping have emerged as effective methods to dope a range of impurities into crystalline semiconductors with the ability to increase the dopant concentrations to greatly exceed their solubility . Hyperdoped Si can be obtained through irradiating Si wafers with fs or ns laser pulses in the presence of dopants on surface, which are typically gases or thin films^[39,40].

The fs-pulsed laser doping has been extensively studied in the past few decades due to the simultaneous realization of hyperdoping and light-trapping structures^[41]. fs-pulsed laser can accurately process materials without generating additional thermal effects because the ablation time is much faster than the thermal diffusion time^[42-44]. However, the hyperdoping depth of Si wafer generally cannot exceed a few hundred nanometers after intensely ablation with direct solid-vapor transition. It is noted that the crystallinity of the surface of hyperdoped Si is poor due to the pressure waves generated by intense light-matter interaction^[45-47], which results in a high sheet resistance, a low charge carrier mobility and many carrier recombination centers^[48]. A ns-pulsed laser possesses smaller peakpower and longer pulse width compared to a fs-pulsed laser,leading to sufficient thermal diffusion and a deeper melting depth^[49]. Therefore, the crystallinity and impurity electroactivity of the hyperdoped region can be enhanced^[50,51]. However, the excessive thermal effect of ns-pulsed lasers results in surface damage^[52-54].

In 1986, Careyet al.^[21] first proposed gas immersion laser doping (GILD) technique to fabricate As-hyperdoped Si. Afterwards, they obtained B-hyperdoped Si using the same method^[55]. The laser doping apparatus is shown inFig. 1, which mainly consists of five components: a XeCI ns-pulsed excimer laser, an optical system, a gas cell, a XY stage and a computer. Hoummadaet al.^[56] adopted atom probe tomography to analyze the 3D compositional profile of B-hyperdoped Si obtained by GILD. The results revealed that the concentration of the incorporated B was well above the solubility (6 × 10²⁰ cm^–3). No B clusters or precipitates were formed. In 2014, Sheret al.^[57] fabricated S-hyperdoped Si using fs-pulsed laser GILD at different SF₆ pressures. After optical-pump/terahertz-probe measurements, they found that the carrier lifetime decreased with increasing dopant concentration^[58]. Then, they proposed that dopant profile and incorporation depth can be tuned by varying the precursor gas pressure and laser pulse parameters^[59,60]. Donget al.^[61] prepared N-hyperdoped Si using fs-pulsed laser GILD in NF₃ atmosphere in 2014, and the SIMS result showed that the N concentration stabilized around 3.5 × 10¹⁹ atom/cm³, which was far exceed the solubility of N atoms in Si (4.5 × 10¹⁵ cm^–3)^[62]. Then in 2018 they designed N, Se and N, S co-hyperdoped Si using first-principles calculations, with the expectation that the hyperdoped Si would be effectively applied to achieve high-efficiency photovoltaic cells^[63,64]. Concentration gradient usually exists in S-hyperdoped Si fabricated by fs-pulsed laser, leading to undesirable inhomogeneous material properties. Therefore, Linet al.^[65] designed a manufacturing method with varying laser pulses and sulfur concentrations to prepare S-hyperdoped Si with a uniform concentration depth profile. Due to the development of thin film deposition technology, substantial researches on laser doping of thin film precursors have been reported in recent years^[66-68]. The equipment for laser doping of thin film is simpler compared to GILD. Sheret al.^[17] first deposited 65 nm Se or Te film on the surface of Si wafer through vacuum thermal evaporation technique, and then obtained Se- or Te-hyperdoped Si by fs-pulsed laser doping. Analogously, Yuet al.^[69] and Qiuet al.^[70] prepared Au- and Ag-hyperdoped Si by vacuum thermal evaporation combined with fs-pulsed laser doping.

Batalovet al.^[71] used recoil-atom implantation method to fabricate Fe-hyperdoped Si, which utilized Xe⁺ ion beam to knock out Fe atoms from target to achieve incorporation of Fe into Si. The SIMS results showed that the utmost Fe concentration reached 1.7 × 10²² cm^–3 (4 nm depth), which was well beyond the equilibrium solubility of Fe in Si (3 × 10¹⁶ cm^–3 at 1300 °C). Chenet al.^[72] successfully fabricated Ni-hyperdoped Si by linear type continuous-wave laser irradiation on Ni film deposited on Si wafer. The maximum Ni concentration of 10²⁰ cm^–3 had exceeded its solubility in Si^[73].

In addition to bulk Si, the hyperdoping of Si nanocrystals and nanowires has also been investigated. For example, Luanet al.^[74] synthesized B- and P-hyperdoped Si nanocrystals through nonthermal plasma. They then fabricated bulk Si with the resistivity of 0.8 mΩ·cm through hot pressing hyperdoped Si nanocrystals. Chenet al.^[75] fabricated B-hyperdoped Si nanoparticles with B concentrations of 15, 25 and 42 at% through laser pyrolysis of SiH₄ and B₂H₆. Moutanabbiret al.^[76] synthesized Al-hyperdoped Si nanowires in an ultrahigh-vacuum chemical deposition system. The Al particles served as energetically favored sites for SiH₄ adsorption and nucleation sites for one-dimensional growth.

The radiative emission of Er³⁺ is around the wavelength of 1.54µm, which can be used for long-distance optical fiber telecommunication^[77]. Therefore, doping Si with Er³⁺ is considered as a promising approach to produce Si-based light-emitting devices^[78]. Wenet al.^[79] fabricated Er/O hyperdoped Si by DC process, which exhibitd a strong photoluminescence at 1536 nm (Fig. 2(b)). Compared to RTA sample, the decay time of Er/O hyperdoped Si remained nearly constant (Fig. 2(c)).

Bustarretet al.^[80] reported that B-hyperdoped Si achieved by GILD was a superconductor when the temperature was below 0.35 K and the critical field was about 0.4 T. According to Bardeen-Cooper-Schrieffer theory, electrons in the lattice attract positive charges at neighboring lattice points, forming a localized region of high positive charge to attract electrons with opposite spin. Electrons with opposite spin and momentum can form Cooper pairs, which can move in the lattice to generate superconducting currents^[81]. Furthermore, ab initio calculations and Raman results indicated that B atoms were in substitutional sites. As shown inFig. 3, deep-level impurity band and sub-bandgaps were generated in Si hyperdoped with deep-level impurities^[82-84]. Compared with chalcogen elements, transition-metal elements are more electroactive. Their impurity bands are closer to the center of intrinsic Si bandgap, which contributes to the decrease of the dark current and improves the sub-bandgap photoresponse^[85-88]. The delocalization of impurity-band electrons leads to high conductivity and significant broadband infrared light absorption^[89]. Zhouet al.^[90] prepared Co-hyperdoped Si by ion implantation followed with PLA, and deduced from resistance-temperature curves that the fitting location of intermediate band was at 0.49–0.51 eV below the conduction band edge of Si. The maximum Co concentration could almost reach 10²² cm^–3, far exceeding the solubility of Co in Si (below 10¹⁷ cm^–3)^[91]. The Co-hyperdoped Si with intermediate band could be a potential candidate in the field of photovoltaic devices, which was confirmed by simulations and calculations of Donget al.^[92,93]. The high absorption coefficient of Ti-hyperdoped Si fabricated by Oleaet al.^[94] could reach between 4 × 10³ and 10⁴ cm^–1. Tullet al.^[95] found that the infrared absorptance of chalcogens (S, Se, and Te)-hyperdoped Si decreased with increasing annealing time (Fig. 4(a)) and the reduction was associated with a characteristic dopant diffusion length (Fig. 4(b)). Therefore, they proposed a simple diffusion model to explain that the diffusion of dopants and the precipitation at the grain boundarieswere the reasons of the decrease in infrared absorptance. After that, Newmanet al.^[96] proposed that annealing at 1350–1550 K followed by rapid cooling (>50 K/s) or fs-pulsed laser treatment contributed to reactivation of sub-bandgap absorptance, as can be seen inFig. 4(c).

Wenet al.^[97] evaluated the effects of annealing on morphology and optoelectronic properties of S-hyperdoped Si. As can be seen inFig. 5(a), conical structure arrays with small protrusions on the surface were obtained after ns-pulsed laser processing, resulting in additional reflection of visible light. The reflectance (400–1000 nm) decreased after annealing owing to the decreasing number of protrusions (Fig. 5(b)). Moreover, the infrared absorptance (1100–2400 nm) increased after annealing at 575 K and decreased as the temperature further increased (Fig. 5(c)) due to the precipitation of S dopants. Wanget al.^[98] reported a novel S-hyperdoped Si prepared by surface texturing combined with ion implantation and PLA.Fig. 5(d) displayed the morphology of Si wafer surface after surface chemical texturing using alkaline solutions. After PLA, the pyramid structures changed into dome structures(Fig. 5(e)), which prolonged the effective path of incident light to increase the infrared absorptance from 30% to 70% (Fig. 5(f)).

Hyperdoped Si nanocrystals give rise to LSPR in infrared spectral range due to large free carrier concentrations. As can be seen inFig. 6(a), with the increase of the doping level of B, an impurity band forms close to the valence band and expands. Moreover, the conduction band expands and the valence band edge moves downward^[99]. After the ionized B concentration exceeded 1.8 × 10²⁰ cm^–3, band gap started narrowing (Fig. 6(b))^[100]. The free holes from the impurity band will oscillate collectively under external excitation, which leads to remarkable light absorption^[14,15,101]. Niet al.^[102] found that the LSPR of B-hyperdoped Si nanocrystals could be tuned by not only the nanocrystals size but also the doping level. When the size decreased or doping level increased, the LSPR induced absorption blueshifted(Figs. 7(a) and7(b)). Rohaniet al.^[103] synthesized B-hyperdoped Si nanocrystals with an average size of 25 nm through laser pyrolysis. The LSPR peak shifted toward higher frequencies as more B was incorporated into Si nanocrystals, which was consistent with the results of Niet al. Moreover, they found the LSPR peak disappeared after annealing at 800 °C due to B segregation (Fig. 7(c)). The outstanding air stability of B-hyperdoped Si nanocrystals shown inFig. 7(d) is advantageous for plasmonic devices.

Compared with bulk Si, nanostructured Si like Si nanowires with a larger specific surface area can be used in nanoelectronics as building blocks to overcome the Schottky limit^[104-108]. Berencenet al.^[109] reported Se-hyperdoped Si nanowires with specific resistivity of 1.27 Ω·cm and sub-bandgap optoelectronic photoresponse. The Si nanowires were fabricated by vapor-liquid-solid method. The hyperdoping of Se was achieved through ion implantation and FLA. Moutanabbiret al.^[76] successfully synthesized Al-hyperdoped Si nanowires using Al-Si nanoparticles as the energetically favored sites for SiH₄ and the nucleation sites for one-dimensional growth. From the concentration profiles of Al, they found that Al was homogeneously distributed in the Al-hyperdoped Si nanowire. The average Al concentration of ~(2.0 ± 0.5) × 10²⁰ cm^–3 was about four orders of magnitude greater than the equilibrium solubility of Al in Si^[110]. Changet al.^[111] adopted supercritical fluid-liquid-solid method to obtain P-hyperdoped Si nanowires. It is worth mentioning that they used red P nanoparticles rather than toxic PH₃^[112]. It can be seen inFig. 8 that the P concentration was as high as 2 at% in Si nanowires. The resistivity of P-hyperdoped Si nanowires was 4.3 × 10^–3 Ω·m.

Dimensionally constrained Si films can control phonon transmittance and reduce Si thermal conductivity, which allows them to find application niches in the field of microelectronics like flexible solar cells^[113-116]. Wanget al.^[117] prepared Ti-hyperdoped Si nanofilms through the method of magnetron sputtering combined with PLA. Three layers were deposited alternately on a Si substrate to form a sandwich structure, as shown inFig. 9(a). Unlike SIMS results of other hyperdoped Si, the Ti atom distribution was homogeneous (Fig. 9(b)) and the average Ti concentration could reach 5 × 10²⁰ cm^–3 (far exceeding the solubility of Ti in Si)^[118]. It can be seen inFig. 9(c) that the maximum light absorption coefficient of Ti-hyperdoped Si sample was about 1.2 × 10⁴ cm^–1, which was much higher than that of Si and the sample without PLA. Wenet al.^[119] prepared Si–S–Si multilayered films (Fig. 10(a)) through film deposition and ns-pulsed laser doping. The multilayered structure was confirmed by TEM (Fig. 10(b)). As can be seen inFigs. 10(c)–10(f), the maximum absorptance, sheet carrier concentration and Hall mobility of S-hyperdoped Si samples were 90%, 1.139 × 10¹⁵ electrons/cm² and 72.29 cm²/(V·s), respectively. Furthermore, the samples exhibited high near-infrared light absorptance (75%–90%) and low sheet resistance (75.8 Ω/□).

Hyperdoped Si obtained by non-equilibrium doping methods such as ion implantation and laser doping not only can be used in novel infrared photodetectors and intermediate band solar cells^[120-122], but also holds promise for the development of superconductors^[123], cryogenic circuit^[124], gas sensor^[125] and light-emitting diodes^[79]. At present, most of the hyperdoped Si devices are based on bulk Si. Extensive theoretical calculations about band structure engineering and optoelectronic properties have also been carried out, which provide theoretical guidance for future applications of hyperdoped Si^[126-128].

Qiuet al.^[129] reported a Ag-hyperdoped Si-based photodetector with responsivities of 504 mA/W at 1310 nm and 65 mA/W at 1550 nm under –3 V bias. The enhanced photoresponse was induced by the deep-level electron traps and two-stage carrier excitation. The Er/O hyperdoped Si wafers prepared by Zhaoet al.^[26] were further doped with B to fabricate photodiodes, which exhibited significant photoresponsivity (up to 100 mA/W) at the communication wavelength of 1510 nm. In additon, the 3 dB bandwidth of the Er/O hyperdoped Si waveguide photodiode reached 30 kHz. Batalovet al.^[71] prepared Fe-hyperdoped Si layers on p-type Si wafer by recoil-atom implantation. The schematic representation of the doping process and the HRTEM image of Si:Fe/p-Si structure can be observed inFigs. 11(a) and11(b). The darkI–V characteristics inFig. 11(c) displayed diode behavior of the hyperdoped sample compared with the ohmic behavior of p-Si substrate.Fig. 11(d) shows the photoresponse of the Si:Fe/p-Si sample under different reverse bias (0–1.6 V). It can be observed that the photoresponse of the sample under –1.0 V bias is close to that of a commercial Si photodiode (FD-27K).

It is reported that a N-hyperdoped Si photodiode fabricated exhibited a photoresponsivity of 5.3 mA/W at 1.31µm and good thermal stability^[130]. The double-absorbing-layer photodiode displayed a higher responsivity compared with single-absorbing-layer photodiode. Wanget al.^[131] proposed an etching treatment (reactive ion etching based on SF₆ plasma) to improve the photoresponse of a S-hyperdoped Si-based photodiode (Fig. 12(a)) working at 1064 nm. They investigated the effects of laser fluence and etching time on the photoresponse(Figs. 12(b) and12(c)). The photoresponse degradation after laser fluence above 0.2 J/cm² was ascribed to the reduction in carrier lifetime and carrier mobility induced by substantial structure defects and recombination sites^[132]. It is worth mentioning that the photoresponsivity could be improved to 0.45 A/W by controlling the etching time.

García-Hemmeet al.^[133] fabricated infrared photodetectors based on Ti-hyperdoped Si. The high sub-bandgap responsivity of 34 mV/W and specific detectivity of 1.7 × 10⁴ cm·Hz^1/2/W (660 Hz) were obtained at the wavelength of 1.55µm. A Te-hyperdoped Si photodiode prepared by Wanget al. exhibited the maximum room-temperature specific detectivity of 3.2 × 10¹² (9.2 × 10⁸) cm· Hz^1/2/W at 1.0 (1.55)µm, which was comparable to that of the commercial devices^[134,135]. As shown inFig. 13(a), the room-temperature responsivity of 79 (0.3) mA/W was obtained at 1.12 (1.55)μm. As the temperature decreased, the photoresponse range expanded to the mid-wavelength infrared range (Figs. 13(a)–13(g)). In the temperature range of 20–26 K, the wavelength with photoresponse was extended to 5μm. The sub-bandgap photoresponse mechanism of the Te-hyperdoped Si photodetector was depicted inFig. 13(h).

As another example, a photodetector based on Te-hyperdoped Si showed superior performance^[67]. The maximum Te concentration in the hyperdoped Si was 1.2 × 10²⁰ cm^–3, which was far beyond the solubility of Te in Si (3.5 × 10¹⁶ cm^–3)^[136]. The maximum responsivity of 120.6 A/W was obtained at 1120 nm. Moreover, the responsivity at 1550 nm was 56.8 mA/W with low noise. Mailoaet al.^[27] fabricated a Au-hyperdoped Si photodiode (Fig. 14(a)) with the largest Au concentration up to 5 × 10²⁰ cm^–3, which was above the solubility of Au in Si (<10¹⁵ cm^–3)^[137]. As shown inFig. 14(c), carrier drift velocity in the Si:Au layer increased as the reverse bias on the junction increased, promoting the gathering of carriers excited by the sub-bandgap light. Mapped external quantum efficiency (EQE) confirmed that sub-bandgap response was attributed to Si:Au layer, while no sub-bandgap response was observed in the reference photodiode. The EQE for three different sub-bandgap wavelengths (1310, 1550 and 1650 nm) increased with the increase of the Au concentration (Fig. 14(f)). The Ni-hyperdoped Si photodiode first reported by Chenet al.^[72] exhibited a remarkable photoresponse of 0.15–0.18 V/W in the wavelength range of 1200–1750 nm. The performance of photodetectors based on hyperdoped Si is summarized inTable 1.

Hyperdoped Si may be also used for intermediate band solar cells^[138]. Fs laser doping can simultaneously obtain hyperdoping and surface texturing, which has been employed for solar cell fabrication^[139]. Wenet al.^[119] designed a novel structure in which S-hyperdoped Si layer was grown on the front surface of a commercial Si solar cell substrate. Compared to the back-surface structure, the front-surface structure (Fig. 15) increased photocurrent for near-infrared and visible light. Moreover, the carrier recombination loss was reduced due to the narrow transport distance of photogenerated carriers to the PN junction. Gimpelet al.^[140] compared the effect of different metal layer systems and deposition techniques on the contact behavior of devices based on S-hyperdoped Si.I–V curves and reactance spectra results showed that pulsed laser deposition Ti/Pd/Ag was the most promising front contact for solar cells. After exploring the temperature dependence of absorption and quantum efficiency for S-hyperdoped Si solar cells, they found that annealing enhanced the photoelectrical conversion below Si bandgap, which was important for the development of multijunction solar cells^[141].

As mentioned above, GILD can help realize B-hyperdoped Si with superconductivity^[142]. Grockowiaket al.^[143,144] fabricated a series of B-hyperdoped Si through GILD with B concentrations varying between ~3 × 10²⁰ to ~6 × 10²¹ cm^–3. They concluded that critical temperature was entirely determined by the B concentration. Chiodiet al.^[123] observed a reduction of contact resistance to 0 Ω as the B concentration exceeded 10²⁰ cm^–3 for a bilayer-based superconductor junction, as shown inFig. 16(a). Furthermore, they measured resistance of the superconductor junction at low temperature. It was found that the junction was superconducting at 215 mK (Fig. 16(b)), which was higher than the critical temperature (~160 mK) for normal metals. The voltage–current characteristic of the junction at 80 mK (the inset of Fig. 16(b)) indicated that the critical current for the transition from superconducting to ohmic state was 2.6µA, while the retrapping current at which the junction switched back to superconducting state was 1.0µA. Fs laser hyperdoped Si can be used for gas sensors due to its advantages of abundant surface states and unique surface morphology^[145,146]. A NO₂ gas sensor^[147] was fabricated by using N-hyperdoped Si (Fig. 17(a)), exhibiting outstanding performance including high response and good selectivity. The average response, response and recovery time for 20 ppm NO₂ were about 79.34%, 12 s and 36 s, respectively (Figs. 17(b) and17(c)).

Hyperdoping methods such as ion-implantation with subsequent heat treatment and laser doping have been employed to tune the properties of Si, which are compatible with the processing of Si materials and devices. Novel properties like superconductivity and near-infrared photoresponse have been enabled by using hyperdoping for Si, which may be in the form of bulk Si, Si films, Si nanowires or Si nanocrystals. Hyperdoped Si has been used to fabricate devices such as infrared photodetectors and solar cells.

Surface texturing induced by laser irradiation is advantageous for the enhancement of infrared absorption. Further investigation on this aspect may lead to infrared photodetectors with improved performance. Hyperdoping has made Si be a plasmonic material. The plasmonic properties of all kinds of Si deserve careful investigation. The effect of the hyperdoping processes on the performance of devices based on hyperdoped Si needs to be studied to facilitate novel device design. It is clear that the research on the hyperdoping of Si should inspire work on hyperdoping other semiconductors and exploring their use in all types of devices.

This work was supported by the National Key Research and Development Program of China (Grant Nos. 2017YFA0205704 and 2018YFB2200101) and the Natural Science Foundation of China (Grant Nos. 91964107 and U20A20209). Partial support was provided by the Natural Science Foundation of China for Innovative Research Groups (Grant No. 61721005).