Method for producing thick films of a silicon-based infrared-sensitive material

Abstract

Current infrared detectors are expensive because they are based on materials with disadvantages. Si supersaturated with deep impurities is a potential alternative, which is why attempts have been made to produce it using various production techniques, however, so far only very thin layers have been obtained, and there are issues such as diffusion of the impurities or other drawbacks which prevent its use in infrared detectors. An alternative method is proposed for producing thick films of Si supersaturated with deep impurities, combining the the low-pressure radio-frequency cathodic sputtering deposition technique using multiple magnetron cathodes and the subsequent low-temperature microwave crystallisation using solid-state sources in a single-mode cavity. The obtained layers have characteristics that allow them to be used in infrared detection devices.

Description

[0001]

[0002] Method for obtaining thick films of a silicon-based infrared-sensitive material

[0003] TECHNICAL SECTOR

[0004] The invention falls within the field of semiconductor materials. More specifically, it relates to a method for obtaining a silicon-based infrared-sensitive semiconductor material and its application in infrared detector systems for various technical fields.

[0005] BACKGROUND OF THE INVENTION

[0006] Current infrared (IR) detectors, such as night vision cameras, are expensive (up to a thousand times more expensive than a standard high-definition visible camera) and bulky because their sensing electronics rely on materials that have one or more of the following disadvantages: they are polluting or toxic, scarce (and therefore expensive), or require operation at low or cryogenic temperatures. Examples of these materials include HgCdTe, PbSe, Ge, InSb, and InGaAs. The low operating temperature requirement increases the cost of the device and its maintenance, as well as its operational efficiency, since the cooling system increases the overall size and weight of the system. Finally, these materials are more difficult to integrate with current silicon-based CMOS (Complementary Metal-Oxide-Semiconductor) microelectronics technology, which further increases the final cost of the device.

[0007] A non-Si-based device is more complex and expensive to manufacture than a Si-based ROIC (Read-Out Integrated Circuit, the part of the integrated circuit used to process the signal from a camera's pixel array, performing operations such as amplification and integration), which is necessary to "read" the image. Therefore, to manufacture an IR camera with current technology, which would be based on the aforementioned materials other than Si, it is necessary to integrate Si technology with a technology based on different semiconductors.

[0008] In some cases, a hybrid approach is used instead of a monolithic one, fabricating the sensor with materials other than silicon, and the silicon ROIC separately (often at different locations and even by different companies), and then joining them together. Integrating the detector and ROIC into a single, monolithic, all-silicon piece would reduce the number of manufacturing steps, increase performance and reliability, and significantly lower costs (A. Rogalski, “Graphene-based materials in the infrared and terahertz detector families: a tutorial”, Advances in Optics and Photonics Tutorial 2019, Vol. 2, No. 2, 314). Non-standard hybridization techniques such as indium bumps or the use of different substrates, such as GaAs, which are much more expensive, are employed. These are the main reasons for the increased cost.This additional cost also affects the resolution of the devices, and standard commercial IR devices based on materials other than silicon are manufactured with much lower resolutions than standard silicon detectors for the visible range, in order to strike a balance between performance and cost. Thus, manufacturing a standard silicon sensor with 20 megapixels costs less than €10, compared to several thousand euros for a much lower-resolution, non-silicon-based SWIR (Short-Wave Infrared) sensor. The price per pixel is in the range of €10. 6 - 10' 5 for Si devices and in the 10' range 2 - 10° for other semiconductor technologies, such as InGasAs or HgCdTe. Furthermore, the maximum resolution is typically much lower for current IR technologies (less than 3 MP) than for Si (up to 160 MP).

[0009] Although silicon (Si) is the most important semiconductor, as most current microelectronic technology is based on it, its band gap of 1.12 eV at room temperature prevents the absorption of radiation with wavelengths longer than 1100 nm. Therefore, conventional Si is not a suitable material for SWIR detection. If the Si semiconductor could be modified to absorb longer wavelengths in the SWIR range, the improved resolution and lower cost of Si-based devices could lead to the design of devices with applications in numerous fields, such as recognition or video surveillance.

[0010] To overcome this limitation of Si, this semiconductor can be supersaturated with impurities that generate deep levels, that is, impurities that introduce allowed energy states for electrons near half of the semiconductor's band gap (supersaturation consists of introducing a very high concentration of said impurity, typically above 10). 20 cm -3Due to the lower energy transitions involved in these impurities, the absorption range can be extended into the IR region (Fu et al., “Hyperdoped Crystalline Silicon for Infrared Photodetectors by Pulsed Laser Melting: A Review”, Physica Status Solidi A 2022219, 2100772). Furthermore, when the concentration of these impurities is sufficiently high, an impurity band can form, improving the material's properties, such as the absorption coefficient and minority carrier lifetime (A. Luque et al., “Understanding intermediate-band solar cells”, Nature Photonics 2012, 6, 146-152).

[0011] Supersaturated (or hyperdoped) silicon-based materials have been studied for about 20 years. Until now, laser or flash annealing processes have been used to fabricate them, with very interesting results. However, although there is an improvement in the IR sensitivity of supersaturated silicon fabricated with pulsed-laser melting (PLM), the reported quantum efficiency (ratio between electrons generated in the device and photons incident on it) at IR wavelengths (for example, at 1350 nm) is very low for the fabricated prototypes, around 0.1% (Daniel Montero Álvarez, “Near Infrared Detectors Based on Silicon Supersaturated with Transition Metals”. Springer, 2021).These low quantum efficiency values ​​are related to several causes, but the main one is that the laser annealing technique can only create very thin sheets (on the order of 250 nm at most), because laser annealing can only modify the surface of the semiconductor. A much thicker layer (on the order of 2 pm) would have to be obtained to absorb at least 95% of the incident light.

[0012] In addition to laser annealing, other manufacturing techniques have been used to obtain supersaturated silicon-based materials, such as flash annealing (FLA) and rapid thermal annealing (RTA). Because deep impurities typically have very low solid solubility in silicon, it is known that obtaining a supersaturated semiconductor requires a manufacturing technique that is not in thermodynamic equilibrium. This means that the material's temperature must be raised and lowered in very short periods, such as laser annealing, which melts and recrystallizes the material in tens of nanoseconds. This rapid process is essential for trapping impurities and achieving a concentration above the supersaturation limit.FLA and RTA techniques involve much longer times than laser annealing, since the FLA process lasts milliseconds and the RTA process several seconds, so the resulting materials often have worse properties than materials manufactured by laser annealing or turn out not to be supersaturated materials.

[0013] Another problem inherent in PLM, FLA, or RTA processes is the diffusion of introduced impurities due to the high temperatures used (typically above 900 °C) and the process dynamics. This diffusion creates zones where the impurity concentration is not high enough to supersaturate (above approximately 10). 20 cm -3 (depending on the impurity), losing the desired properties. Moreover, the presence of deep centers at lower concentrations (between 10 11 cm -3 and 10 2 ° cm -3This produces undesirable effects, such as the reduction in the lifetimes of photogenerated carriers due to recombination processes (A. Rohatgi et al., “Effect of titanium, copper and iron on silicon solar cells”, Solid-State Electronics 1980, vol. 23, 415), and ultimately the reduction in quantum efficiency and the ability to detect IR. To mitigate this problem, low-temperature manufacturing techniques should be used, reducing diffusion.

[0014] On the other hand, ion implantation is also typically used to introduce deep impurities into the semiconductor, and the material is subsequently recrystallized using PLM, FLA, or RTA (see, for example, the literature review on this topic: MJ Sher and E. García, “Hyperdoped silicon materials: from basic materials properties to sub-bandgap infrared photodetectors”, Semiconductor Science and Technology 2023, vol. 38, 033001). One disadvantage of this method is that ion implantation inherently produces what are called “implantation tails,” which are simply the aforementioned areas with low concentrations of deep impurities. Therefore, ion implantation should also be avoided for obtaining supersaturated semiconductor sheets.Combining ion implantation techniques with PLM, FLA, or RTA exacerbates the problem of implantation tails due to the diffusion of impurities at high temperatures into deeper areas of the material.

[0015] Therefore, with the current state of thin-film technology, it is not possible to obtain a viable Si-based SWIR photodetector that operates at room temperature. It is therefore necessary to explore other techniques that can reproduce or improve the properties of supersaturated Si, but with thicker layers and low-temperature fabrication, avoiding the diffusion of deep impurities.

[0016] The present invention proposes the use of microwave annealing (MWA) processes to obtain supersaturated silicon sheets with deep centers of the required thicknesses while minimizing diffusion. The MWA process has been known for decades and is applied in many areas of materials science. However, it has been little used in the field of microelectronics. The inventors' previous paper (J. Olea et al., “Ti supersaturated Si by microwave annealing processes”, Semicond. Sc / . Technol. 2023, 38, 024004) shows the results of initial experiments, which were unsuccessful, and seem to conclude that the microwave annealing process is not suitable for obtaining supersaturated silicon sheets, as it does not lead to the desired results.Specifically, in this research sheets are manufactured using the ion implantation technique, producing depth profiles of Ti with the aforementioned implantation tails, that is, with areas with a Ti concentration well below 1O. 20 cm -3(See Figure 2 of the reference). Furthermore, the crystalline quality of the resulting sheets is low, containing a high density of defects (Figure 3 of the reference). Finally, the thickness of the sheets is low, on the order of 50 nm, and this thickness is well below the thickness required to absorb all photons with energy below the semiconductor's bandgap, making them unsuitable for fabricating IR photodetector devices. The present invention discloses an optimized process for obtaining high-quality, deep-center, supersaturated Si material with impurities that enable the fabrication of materials and structures with thicknesses suitable for integration into IR radiation detectors.

[0017] EXPLANATION OF THE INVENTION

[0018] This invention proposes the use of a deposition technique to deposit a semiconductor layer (initially in an amorphous phase) of silicon supersaturated with impurities that generate deep centers, onto a crystalline silicon substrate, followed by a low-temperature microarray process to crystallize the layer to the appropriate thickness and obtain a high-quality material with the desired properties. The success of the process lies in the combination of both techniques, since neither alone or in combination with others (PLM, FLA, or RTA) would provide the desired results.

[0019] On the one hand, the deposition technique avoids the use of ion implantation (which is not a deposition technique, but a sheet modification technique), typically used in this field (for example, in J. Olea et al., “Ti supersaturated Si by microwave annealing processes”, Semicond. Sc / . Technol. 2023, 38, 024004), thus preventing the appearance of the harmful implantation tails (1) in Figure 1, and also enables the fabrication of thick sheets with a constant impurity concentration profile, (2) in Figure 1. On the other hand, the MWA technique, unlike laser annealing, allows the crystallization of thick sheets, even several microns thick, with low-temperature processes (below 600 °C), minimizing diffusion and therefore drastically reducing or even avoiding areas with a low concentration of deep impurities.Due to its duration (tens of minutes or even hours, depending on the thickness of the laminate being manufactured), the MWA process could be considered a process within thermodynamic equilibrium, and therefore might initially seem unsuitable for supersaturating semiconductors. However, surprisingly, and as shown below in the experimental results, thanks to the fact that microwaves can be absorbed locally by defects at the nanoscale, crystalline defects in laminates can be eliminated using low-temperature processes (< 600 °C) and times exceeding 30 minutes (preferably between 50 and 600 minutes, depending on the thickness of the laminate to be crystallized), with satisfactory results.The idea is that while the defect (of nanometric size) absorbs microwaves, raising its temperature and repairing itself, the surrounding volume remains at a low temperature. Therefore, the overall temperature required to crystallize a sheet is much lower than in a standard PLM, FLA, or RTA process. If the same temperature and / or time parameters were used in FLA or RTA processes, supersaturated semiconductor sheets with high crystalline quality would not be obtained.

[0020] More specifically, the proposed deposition technique is radiofrequency sputtering with magnetron cathodes in low-pressure processes using a multi-cathode configuration. Using a magnetron cathode makes the deposition process more efficient and faster than other techniques because the magnetic field concentrates the ion trajectories, resulting in faster material removal. This speed also produces sheets with lower contamination. For example, deposition rates of around 10 nm / min can be achieved, so manufacturing a 1 pm supersaturated sheet would take approximately 100 minutes. Furthermore, by performing low-pressure deposition processes (below 6 x 10⁻⁶ N / min), the process is significantly faster. -3mbar) we also minimize contamination of the sheets, which is fundamental to obtaining sheets with good characteristics, mainly high carrier mobility and long lifetimes of minority carriers.

[0021] Finally, the multi-cathode configuration gives the system great versatility in manufacturing sheets with different characteristics. This allows varying the impurity concentration depending on the depth within the sheet, enabling the design of materials with different properties (for example, sheets with different absorption coefficients, varying mobility and conductivity, or minority carrier lifetimes). This is very useful for optimizing the design of a device where different impurity concentrations may be required depending on the depth within the sheet. For example, a system could be configured with one Si cathode and one Ti cathode to deposit Si sheets supersaturated with Ti at a chosen Ti concentration, which could be 1%, 2%, etc., even varying according to the power applied to each cathode during the deposition process.With a single-cathode configuration, this is not possible, as the impurity concentration is predefined. Furthermore, this configuration would allow the controlled deposition of various deep impurities on the same supersaturated sheet. The deep-level impurity or impurities in Si can be Ti, V, Cr, Cu, Ni, Zn, Co, Au, Se, S, Te, Ta, Sr, Cs, Ba, Mo, Na, Pd, Pt, Ag, or a combination thereof. For example, the system could be configured with one cathode of Si, one of Ti, one of Se, and one of P to fabricate the different configurations shown in Figure 2. The P cathode would be necessary to fabricate Si with type ny doping, thus completing the device.

[0022] Following the deposition process using the technique described in the preceding paragraphs, which yields a supersaturated sheet in the amorphous phase, the MWA process would need to be performed to crystallize this sheet and obtain the optimized properties. Since the results in the reference J. Olea et al., “Ti supersaturated Si by microwave annealing processes”, Semicond. Sci. Technol. 2023, 38, 024004 (2023 Olea), are not satisfactory, a modified method is proposed to achieve this objective. In this procedure, the MWA process is modified in several aspects, which are described below.

[0023] Regarding microwave sources, the 2023 Olea reference uses magnetron-type sources. The modified procedure uses solid-state sources with a power output between 2 kW and 5 kW, depending on the thickness of the sheet to be crystallized. The fundamental difference is that solid-state sources make the process more spatially homogeneous because they electronically vary the process frequency, eliminating localized hot spots. This improves the crystallization process, producing a material with better crystalline quality and greater spatial homogeneity, and in a shorter manufacturing time. In the field of microelectronics, where devices are very small (on the order of 1 micrometer or less), spatial homogeneity is essential.When inhomogeneities occur, undesirable processes such as “cellular breakdown” appear, leading to sheets with a very inhomogeneous concentration of the deep impurity, and frequently with worse crystalline quality (F. Liu et al., “Suppressing the cellular breakdown in silicon supersaturated with titanium”, J. Phys. D: Appl. Phys. 2016, vol. 49, 245104).

[0024] Regarding the process frequency, in the 2023 Olea reference, the magnetron sources are 5.8 GHz, while in the modified procedure we propose the use of solid-state sources at 2.4–2.5 GHz. The 2.45 GHz frequency is more standard than 5.8 GHz, with more equipment available on the market at a much lower cost. At an industrial level, this is a very important decision because at 5.8 GHz, all equipment is much more expensive, spare parts are less readily available, and it is less reliable. Furthermore, the microwave cavity design is much more complex, and achieving a homogeneous process is more difficult.

[0025] In the experiments referenced in 2023 Olea, the microwave cavity used is a multimode cavity. In the modified procedure, the cavity must be single-mode, which implies a completely different design for the MWA equipment. The fundamental difference is that in single-mode equipment, the power density achieved is much higher, and therefore the manufacturing process is more energy-efficient (and thus more economical) and faster. A shorter processing time helps minimize the diffusion of impurities and therefore the formation of tailings in the impurity depth profile.

[0026] In the experiments referenced in Olea 2023, MWA processing times of up to 10 minutes were used to recrystallize a 60 nm thick sheet with a Ti concentration of approximately 1%. As shown in Figure 3, a 10-minute process is clearly insufficient, consistently producing sheets with a high number of defects. In the modified procedure, the times depend on the thickness of the sheets to be recrystallized and the concentration of the deep impurity, but are always above 30 minutes to ensure the material crystallizes properly and achieves the desired properties. For sheets with thicknesses between 500 nm and 2 pm, between 2.5 and 10 hours are required. For example, for a thickness of 1 pm, approximately 5 hours will be necessary.

[0027] Using these modifications in the manufacturing process, the results shown in figures 3 to 6 are achieved. Si sheets supersaturated with 1% Ti are manufactured, processed by MWA for 10 minutes (as a reference) and for 40 and 50 minutes (for optimized crystallization), and the results are analyzed and it is verified that a monocrystalline sheet with optimized quality can be obtained.

[0028] Thus, the thick sheet material (more than 500 nm thick) of Si supersaturated with deep-level impurities, obtained by the described procedure, has an absorption coefficient greater than 10 4 cm -1 in the 1000-3000 nm wavelength range of light, carrier mobility of at least 10 cm 2 / Vs, and lifetimes of minority carriers of at least 1 ps.

[0029] All these results verify that the proposed method is viable for manufacturing supersaturated silicon sheets oriented towards the field of infrared detectors. This involves integrating the supersaturated material sheet, along with electrical contacts and other components (such as an antireflective coating or insulating grooves), into an IR photodetector device, depending on the specific design.

[0030] BRIEF DESCRIPTION OF THE DRAWINGS Figure 1. Ti concentration profiles in Si as a function of depth for two supersaturated samples, one of them manufactured by ion implantation (solid line), with implantation tails (1), and the other of them manufactured by radiofrequency sputtering deposition with magnetron cathode (dashed line), without implantation tails (2).

[0031] Figure 2. Examples of structures formed by supersaturated Si sheets with one (Ti or Se) or two deep centers (Ti and Se) on a crystalline Si substrate. One of the structures also has a P-doped Si sheet to form a double bond between the supersaturated sheet and two Si layers, one ny-type and the other p-type.

[0032] Figure 3. Transmission and inset electron microscopy image with the electron diffraction pattern of a Ti-supersaturated Si sheet processed by MWA for 10 minutes, and fabricated on a crystalline Si substrate. The sample is observed to have a high density of structural defects, which is confirmed by the electron diffraction pattern, showing bright spots between the main points.

[0033] Figure 4. Transmission and inset electron microscopy image with the electron diffraction pattern of a Ti-supersaturated Si sheet processed by MWA for 40 minutes, and fabricated on a crystalline Si substrate. No structural defects are observed and the electron diffraction pattern does not show intermediate bright spots, from which it is concluded that the sheet has optimal crystalline quality.

[0034] Figure 5. Raman spectroscopy results of 3 samples: a crystalline Si reference substrate without Ti, and two Si samples supersaturated with Ti and processed by MWA for 10 and 40 minutes, respectively. The sample processed for 10 minutes does not have good crystalline quality, while the sample processed for 40 minutes has practically perfect crystalline quality, similar to that of the reference sample.

[0035] Figure 6. Absorption spectrum (absorptance) of 3 samples: a reference substrate of crystalline Si without Ti, and two samples of Si supersaturated with Ti, processed by MWA, for 40 minutes and 50 minutes, respectively. The Si supersaturated with Ti samples show absorption in the IR range.

[0036] PREFERRED EMBODIMENT OF THE INVENTION

[0037] The present invention is illustrated by the following examples, which are not intended to be limiting of its scope.

[0038] Example 1.

[0039] This example describes the processes necessary to manufacture a sheet of Si-based, high-quality crystalline, IR-sensitive supersaturated material that would be integrated into an IR photodetector device.

[0040] The process begins with a crystalline silicon wafer with a resistivity in the range of 1–10 Ωcm, p-type, grown in the (111) direction using the Czochralski technique. The native silicon oxide is removed by chemical etching with buffered hydrofluoric acid (BHF), preferably in an inert atmosphere transfer chamber, so that the supersaturated wafer is deposited on a pristine surface. This will aid crystallization during the MWA process.

[0041] Secondly, the wafer is passed directly from the transfer chamber to the deposit chamber, and the chamber is immediately emptied to a pressure on the order of 10' 7mbar or less to prevent native oxide regrowth. A 1 pm thick Si sheet, doped with 2% Ti, is co-deposited on the polished side of the wafer using radio frequency sputtering. For this, the sputtering equipment must have at least one Si magnetron cathode (minimum purity 99.999%) and one Ti magnetron cathode (minimum purity 99.999%), both at least 2 inches in diameter. Additionally, a third P-doped Si magnetron cathode (minimum purity 99.999%) could be used to fabricate the n-type contact on the supersaturated sheet, thus obtaining a complete device. Furthermore, another Se cathode (minimum purity 99.999%) could be used to fabricate a Si sheet supersaturated with Ti and Se. During the process, the sample holder with the Si wafer should rotate at about 10 rpm to provide greater homogeneity to the deposit.The deposition process is carried out at a pressure of 5x10. -3 The silicon cathode is heated in an argon atmosphere, and may incorporate 2% hydrogen for passivation purposes. The silicon cathode is configured with a power of 80 W and the titanium cathode with a power of 40 W, to achieve a deposition rate of approximately 10 nm / min, with the entire deposition process lasting approximately 1 hour and 40 minutes.

[0042] Following the deposition process, the wafer with the deposited supersaturated foil, which is currently amorphous, is processed using MWA in a single-mode chamber to crystallize it and activate the carriers. This enables the detection of energies in the IR range and optimizes carrier mobility and minority carrier lifetimes. Solid-state power sources are used, automatically vaping the frequency between 2.4 GHz and 2.5 GHz throughout the process to achieve greater homogeneity. The power of the sources is 3 kW. The process can be performed in either an atmosphere with a continuous flow of Ar or N2. To achieve optimal crystallization of the foil with a thickness of 1 pm, the MWA process takes approximately 5 hours.

[0043] Example 2.

[0044] This example shows the results of characterizing a Si sheet supersaturated with Ti using the proposed method.

[0045] Figures 3 and 4 show electron microscopy results, including images and electron diffraction patterns. As can be seen, for an MWA process lasting 10 minutes, the resulting sheet contains many defects visible in the image (including stacking faults), which are also evident in the electron diffraction pattern (inset of the image). This diffraction pattern shows bright spots between the main points, indicating that the crystal lattice has a large number of defects. In contrast, when the MWA process lasts 40 minutes (Figure 4), the resulting sheet is monocrystalline with an orientation consistent with the substrate and no visible defects. The electron diffraction pattern shows only the points corresponding to the main crystal lattice, without any additional points that would correspond to defects in the crystal lattice.Raman spectroscopy can also be used to analyze crystalline quality and corroborate the results. Figure 5 shows three spectra. The spectrum corresponding to the Si sample without Ti provides a reference for what should result from the analysis of a perfect crystal lattice. A Si sample supersaturated with Ti and processed by MWA for only 10 minutes produces a spectrum with a principal vibrational mode associated with Si in the crystalline phase (at 520.5 cm⁻¹). -1) with a very low intensity, and without secondary vibration modes, which corresponds to a defective sheet, in accordance with what was seen in the electron microscopy images of figures 3 and 4. Finally, for an MWA process of 40 minutes, the Si sheet supersaturated with Ti produces a spectrum practically the same as the reference, with a very intense main vibration mode, and clearly resolved secondary modes, from which it is concluded that the sheet has an optimized crystalline quality.

[0046] Example 3.

[0047] This example refers to the use of the obtained material in a device for IR detection.

[0048] Once the proposed method is proven suitable for fabricating high-quality crystalline Si sheets supersaturated with Ti, it is verified that these sheets possess interesting properties for integration into devices for IR detection. Figure 6 shows IR absorption spectra in the 1–2.5 pm range. As can be seen, a Si sample without Ti exhibits no absorption in the IR range, while the properly crystallized Si sheets supersaturated with Ti (MWA times greater than 30 minutes) show clearly visible absorption, exceeding 1%. Since the crystalline quality has been shown to be optimal using the proposed fabrication method, this absorption cannot originate from structural defects, but rather from the incorporation of Ti at high concentrations. This confirms that these supersaturated materials could indeed be applied to IR detection.There is no difference in absorption between the two sheets shown in the figure (40 minutes and 50 minutes), demonstrating that longer MWA processes do not worsen the material's properties, indicating that the MWA process is a reliable technique that produces stable materials. This is because the MWA process is a low-temperature process, unlike standard industry and scientific processes (PLM, FLA, or RTA).

Claims

CLAIMS 1. A method for obtaining thick films of a deep-level impurity-supersaturated Si-based infrared-sensitive material, characterized in that it combines a deep-level impurity deposition process using radiofrequency sputtering with magnetron cathodes, in low-pressure processes, using a multi-cathode configuration, and a subsequent low-temperature microwave annealing (MWA) process to crystallize the deposited layer, comprising: Depositing a deep-level impurity supersaturated semiconductor layer onto a crystalline Si substrate by sputtering with magnetron cathodes in a process at pressures below 6x 10 -3 mbar in an Ar atmosphere for a time between 50 and 200 minutes to obtain a supersaturated sheet in amorphous phase with deep level impurities between 0.5 pm and 2 pm thick. Crystallize the supersaturated sheet obtained by MWA using solid state sources at a frequency between 2.4 and 2.5 GHz, with a power between 2 kW and 5 kW in a single-mode chamber and at a temperature below 600°C for a time between 2.5 and 10 hours to obtain a crystalline sheet with a thickness between 0.5 pm and 2 pm.

2. Method, according to claim 1, wherein the deep-level impurity or impurities are Ti, V, Cr, Cu, Ni, Zn, Co, Au, Se, S, Te, Ta, Sr, Cs, Ba, Mo, Na, Pd, Pt, Ag or a combination thereof.

3. Method, according to claims 1 and 2, wherein the impurity is Ti in a concentration between 1% and 2%.

4. Method, according to the preceding claims, wherein the deposit is made at a pressure of 5x10 -3 mbar, in an Ar atmosphere, for 100 minutes, at a deposition rate of 10 nm / min to obtain a supersaturated sheet 1 pm thick.

5. Method, according to claim 4, wherein the supersaturated sheet is crystallized by MWA at a frequency of 2.45 GHz, a power of 3 kW for 5 hours in an Ar atmosphere to obtain a fully crystalline sheet.

6. Crystalline material in the form of a thick Si sheet supersaturated with deep-level impurities according to claim 2 characterized in that it has a thickness greater than 500 nm, an absorption coefficient greater than 10 4 cm -1 in the 1000-3000 nm wavelength range of light, carrier mobility of at least 10 cm 2 / Vs, and lifetimes of minority carriers of at least 1 ps.

7. Infrared detector device that integrates the material described in claim 6.

Images