Silicon has been extensively utilized in fiber-optic communications, photodetection, and solar cells owing to its low purification cost, high purity, and compatibility with complementary metal-oxide-semiconductor (CMOS) technology. However, silicon-based photodetectors are restricted in near-infrared (NIR) photodetection due to their indirect bandgap of 1.12 eV at room temperature. To address this limitation, significant research efforts have focused on NIR photodetection using bulk materials such as germanium (Ge), cadmium sulfide (CdS), and gallium arsenide (GaAs), as well as two-dimensional materials like graphene. Among these, femtosecond-laser hyperdoped silicon photodetectors have emerged as promising candidates for silicon-based NIR photodetection. These devices combine extended spectral responsivity beyond 1100 nm where conventional silicon photodetectors are unable to respond with cost-effective CMOS-compatible fabrication, simultaneously achieving broadband high responsivity and industrial-scale manufacturability. Despite these advantages, the underlying mechanisms of femtosecond-laser hyperdoped silicon photodetectors remain poorly understood. Further research on the working mechanism of femtosecond-laser hyperdoped silicon photodetectors will not only facilitate in-depth tuning of the photodetectors, but also provide important guidance for further functional design of femtosecond-laser hyperdoped devices. In this study, we report a low-concentration femtosecond-laser hyperdoped silicon photodetector demonstrating light-intensity inverse-saturation response characteristic and bias inverse-saturation response characteristics. This unique inverse-saturation response characteristic offers a robust solution for photodetection in environments with intense background illumination. Furthermore, by systematically analyzing current-output properties of hyperdoped silicon devices under different light intensities and reverse biases, our findings provide critical insights into working mechanisms of femtosecond-laser hyperdoped silicon photodetectors, facilitating their future performance enhancement and application-specific customization.

In this study, sulfur-hyperdoped silicon is fabricated by irradiating N-type single-crystalline (100) silicon wafers with a femtosecond laser in an SF₆ atmosphere. A Pockels cell electro-optic switch is utilized to precisely control the number of scanning laser pulses, enabling the preparation of two distinct doping concentration levels. Hyperdoped silicon samples subsequently undergo rapid thermal annealing, followed by thermal evaporation of aluminum electrodes to prepare photodetectors. The surface morphology of the hyperdoped silicon is characterized using field-emission scanning electron microscopy (SEM), while depth-resolved dopant concentration profiles are quantitatively analyzed via secondary ion mass spectrometry (SIMS). For the low-concentration sulfur-hyperdoped silicon photodetector, spectral responsivity measurements are performed under varying reverse biases (DC power supply) and incident light intensities (modulated by a calibrated neutral density filter). Current-output properties are systematically investigated at different reverse biases and light intensities.

The high-concentration sulfur-hyperdoped silicon exhibits quasi-periodic peak structures (about 1 μm average height) distributed across the surface. In contrast, the low-concentration sulfur-hyperdoped silicon displays a flat morphology without microscale surface features. SIMS measurements reveal a sulfur peak concentration exceeding 10²⁰ cm^-3 within a 600 nm depth for the high-concentration sample, whereas the low-concentration sample achieves a peak doping concentration above 10¹⁸ cm^-3 confined to a 50-nm subsurface region [Fig. 2(c)]. Photocurrent measurements reveal that the low-concentration hyperdoped silicon photodetector exhibits a unique inverse-saturation response characteristic under illumination intensities exceeding 166.2 μW, where the photocurrent decreases with increasing light intensity [Fig. 3(a)]. This inverse-saturation response characteristic is modulated by the irradiation wavelength, with the most pronounced effect observed under 1000 nm illumination [Fig. 4(b)]. By regulating the photodetector’s reverse bias between 2?8 V, we achieve controllable modulation of the inverse-saturation intensity (Fig. 5). At an illumination intensity of 5 μW, the photodetector demonstrates a peak spectral responsivity of 252.71 A/W at 1000 nm (Fig. 8), surpassing that of commercial silicon-based photodetectors by over two orders of magnitude, while possessing a unique light inverse-saturation response characteristic, indicating superior weak-signal photodetection capability in high-intensity optical interference environments.

In this study, two sulfur-hyperdoped silicon photodetectors with distinct doping concentrations are successfully fabricated by adjusting the pulse number of femtosecond laser irradiation on silicon. We observe both light-intensity inverse-saturation response characteristic and bias inverse-saturation response characteristic in the low-concentration hyperdoped silicon photodetector. To explain the light-intensity inverse-saturation phenomenon, we extend the existing defect-assisted photoconductive gain model by incorporating discussions on the hole quasi-Fermi level variation in hyperdoped silicon under varying light intensities. Furthermore, the bias inverse-saturation mechanism is elucidated through the introduction of doping-generated junction effects on defect-assisted photoconductive gain. The refined model effectively describes the current-output characteristics of femtosecond-laser hyperdoped silicon under varying reverse biases and light intensities, providing critical insights for optimizing femtosecond-laser hyperdoped devices and advancing their exploration for environment-specific applications. Additionally, we develop a novel photodetector exhibiting a unique light-intensity inverse-saturation response characteristic. This device demonstrates straightforward fabrication, exceptional performance (peak spectral responsivity: 252.71 A/W), and dual functionality—sensitive weak-light photodetection coupled with distinct strong-light inverse-saturation response characteristic. The easily tunable inverse-saturation threshold and CMOS-compatibility position this technology as a promising solution for weak-signal photodetection in complex optical interference environments, showcasing significant application potential.