Method for implanting doping atoms into a substrate

Abstract

Ion implantation in which an ion beam (10) comprising nitrogen ions is implanted into a substrate (12) is used to reduce the effects of basal plane dislocations in the substrate (12). The substrate (12) is at least partially formed from SiC in the region (14, 16, 21) to be doped or the regions to be doped. The doping concentration (c) of nitrogen doping atoms that are added to the substrate (12) as a result of irradiation with the ion beam (10) is between 1E14cm-3 and 5E17cm-3 in each region (14, 16, 21) to be doped.

Description

[0001] Method for implanting doping atoms into a substrate. The invention relates to a method for implanting doping atoms into a substrate and, in particular, to a specific use thereof. Many semiconductor devices, especially discrete high-blocking-capacity semiconductor devices, are now realized with in-situ doped SiC epitaxial layers. It is known that the 4H-SiC substrate material naturally exhibits a high density of various defects and dislocations, particularly basal plane dislocations. It is also known that these defects can expand from the base substrate to the epitaxial layer during epitaxial layer growth. Furthermore, it is known that basal plane dislocations, in particular, can expand under bipolar current loading (recombination of electrons and holes) and under UV illumination. This phenomenon is referred to as "bipolar degradation."Since bipolar processes can occur even in unipolar devices in certain switching configurations, bipolar degradation represents one of the most significant challenges for the reliable operation of SiC devices. This problem is particularly acute for high-current SiC devices, particularly those operated at 30 A / cm² or more. These high currents either cause partial dislocations (part of the basal plane dislocations) to migrate from the substrate or buffer layer into the epitaxial layer, or supply sufficient energy to the nucleation sites in the epitaxial region to form a stacking fault, which then expands and degrades the device. It is already known that the density of basal plane dislocations in an epitaxial layer can be reduced by adjusting the growth conditions and by high-temperature annealing processes.The bipolar degradation of basal plane dislocations in the substrates can also be significantly reduced by an epitaxial buffer layer with a higher doping concentration (n+, for example, by in-situ doping with nitrogen atoms). In Tawara et al., "Suppression of the forward degradation in 4H-SiC PiN diodes by employing a recombination-.

[0002] 14513 P 6523 WO enhanced buffer layer“, Materials Science Forum, Vol. 897, pp. 419-422, May 15, 2017, it was found in this context that such an epitaxial buffer layer between the SiC base substrate and the SiC drift zone has a positive effect on PiN diodes. Increasing the thickness of this buffer layer led to better results. The effect of this measure is only noticeable at a doping concentration of the nitrogen atoms above 1E18cm -3in a relevant way. By means of ion implantation, doping or the generation of defect profiles in substrates made of various materials, such as semiconductors (e.g., silicon, silicon carbide, gallium nitride) or optical materials (e.g., LiNbO3), with predefined depth profiles in the depth range down to several hundred micrometers can be achieved. Ion implantation is also regularly used to dope structures in semiconductor components. The present invention is based on the object of specifying a method by means of which the negative effects of basal plane dislocations in SiC semiconductor components can be significantly reduced without generating other serious disadvantages.According to the invention, ion implantation with an ion beam containing ions of nitrogen into a substrate is used to reduce the effects of basal plane dislocations in the substrate, wherein the substrate is formed at least partially from SiC, preferably 4H-SiC, in the region or regions to be doped, and wherein the doping concentration of nitrogen doping atoms added to the substrate by irradiation with the ion beam is between 1E14cm in each region to be doped. -3 and 5E17cm -3 , preferably between 5E14cm -3 and 5E17cm -3 , more preferably between 1E15cm -3 and 5E17cm -3 , more preferably between 1E15cm -3 and 3E17cm -3 , more preferably between 2E15 cm -3 and 2E17cm -3 , more preferably between 3E15cm -3 and 1E17cm -3 , particularly preferably between 4E15cm -3 and 8E16cm -3 or between 5E15cm-3 and 7E16cm -3 The applicant has surprisingly discovered that when using ion implantation for nitrogen doping, even relatively low doping levels in the substrate significantly reduce the effects of basal plane dislocations. 14513 P 6523 WO The preferred doping concentration of nitrogen doping atoms added to the substrate by irradiation with the ion beam can be adjusted in each region to be doped, even between 1E14cm -3 and 1E17cm -3 or between 1E14cm -3 and 7E16cm -3 or between 1E14cm -3 and 4E16cm -3 or between 1E14cm -3 and 2E16cm -3 The preferred doping concentration of nitrogen dopant atoms added to the substrate by irradiation with the ion beam can also be between 5E14cm -3 and 1E17cm -3 or between 5E14cm -3and 7E16cm -3 or between 5E14cm -3 and 4E16cm -3 or between 5E14cm -3 and 2E16cm -3 The preferred doping concentration of nitrogen dopant atoms added to the substrate by irradiation with the ion beam can also be between 1E15cm -3 and 1E17cm -3 or between 1E15cm -3 and 7E16cm -3 or between 1E15cm -3 and 4E16cm -3 or between 1E15cm -3 and 2E16cm -3 Likewise, the preferred doping concentration of nitrogen dopant atoms added to the substrate by irradiation with the ion beam can also be between 2E15cm -3 and 3E17cm -3 or between 2E15cm -3 and 7E16cm -3 or between 2E15cm -3 and 4E16cm -3 or between 2E15cm -3 and 2E16cm -3Likewise, the preferred doping concentration of nitrogen dopant atoms added to the substrate by irradiation with the ion beam can also be between 3E15cm -3 and 3E17cm -3 or between 3E15cm -3 and 7E16cm -3 or between 3E15cm -3 and 4E16cm -3 or between 3E15cm -3 and 2E16cm -3 Finally, the preferred doping concentration of nitrogen dopant atoms added to the substrate by irradiation with the ion beam can also be between 4E15cm -3 and 3E17cm -3 or between 4E15cm -3 and 7E16cm -3 or between 4E15cm -3 and 4E16cm -3 or between 4E15cm -3 and 2E16cm -3It is preferably provided that the doping concentration of nitrogen doping atoms, which are added to the substrate by irradiation with the ion beam, shows a continuous profile in the depth direction of the substrate from a doping start depth, particularly preferably already from the entry point of the ion beam into the substrate, up to a maximum doping depth. In this way, the negative effects of basal plane dislocations are prevented over as broad an area as possible. 14513 P 6523 WO It is further preferably provided that the doping concentration of nitrogen doping atoms, which are added to the substrate by irradiation with the ion beam, shows a profile in the depth direction up to a maximum doping depth, the lowest value of which differs from the highest value by a maximum of 50%, preferably a maximum of 20%.The relatively constant doping profile thus generated produces a uniform effect over a considerable substrate area. In a particularly preferred embodiment, the ion beam passes through an energy filter before striking the substrate. This filter is designed as a microstructured membrane for adjusting a dopant depth profile in the substrate. This ensures the generation of continuous dopant depth profiles in the substrate in a simple manner. Generally speaking, within the scope of the invention, a maximum doping depth of the ion beam into the substrate is preferably 5 to 40 µm, more preferably 10 to 30 µm. The energy of the ion beam is preferably in the range of 3 to 50 MeV, more preferably in the range of 5 to 25 MeV. It is preferably provided that an annealing step at 1,650 to 1,750 °C takes place after the implantation with nitrogen.Negative effects caused by the implantation of nitrogen atoms are thereby reduced, and nitrogen is electrically activated as a doping atom. The region(s) to be doped of the single-crystal SiC substrate are undoped or lightly n-doped with a doping concentration of no more than 6E15cm-1 before the inventive implantation of the nitrogen atoms. -3 After the implantation of the nitrogen atoms according to the invention, the region or regions of the single-crystal SiC substrate to be doped are now n-doped (in the case of a previously undoped substrate) or slightly more strongly n-doped than before (in the case of a previously weakly n-doped substrate), with a total doping concentration of between 3E14cm -3 and 5E17cm -3 , preferably between 3E14cm -3 and 2E17cm -3 , especially preferred between 4E14cm -3 and 7E16cm -3The total concentration can preferably also be between 3E15cm -3 and 5E17cm -3 , more preferably between 3E15cm -3 and 2E17cm -3 , particularly preferably between 4E15cm -3 and 7E16cm -3 According to a further aspect of the invention, a method according to the invention for implanting doping atoms into a substrate in the context of manufacturing semiconductor components or a method according to the invention for reducing the effects of basal plane dislocations in a substrate in the context of manufacturing semiconductor components comprises the following steps: - Providing a substrate with two substrate regions arranged one above the other in a depth direction of the substrate, wherein the first substrate region is a region made of monocrystalline SiC, preferably 4H-SiC, which is undoped or weakly n-doped with a doping concentration of at most 6E15cm -3 , preferably no more than 5E15cm-3 , more preferably maximum 4E15cm -3 , and wherein the second substrate region is a heavily n-doped region of single-crystalline SiC, polycrystalline SiC, amorphous SiC, silicon, graphite or layer combinations thereof with a doping concentration of at least 3E17cm -3 , preferably at least 5E17cm -3 , more preferably at least 9E17cm -3 - irradiating the substrate with an ion beam with ions of nitrogen from a side of the first substrate region which is opposite the second substrate region, such that the first substrate region is doped at least in sections with nitrogen and the second substrate region is doped at least in sections with nitrogen, wherein the doping concentration of nitrogen doping atoms which are added to the substrate by the irradiation with the ion beam in each doping region is between 1E14cm -3 and 5E17cm -3 , preferably between 5E14cm -3and 5E17cm -3 , more preferably between 1E15cm -3 and 5E17cm -3 , more preferably between 1E15cm -3 and 3E17cm -3 , more preferably between 2E15 cm -3 and 2E17cm -3 , more preferably between 3E15cm -3 and 1E17cm -3 , particularly preferably between 4E15cm -3 and 8E16cm -3 or between 5E15cm -3 and 7E16cm -3The applicant has determined that, in such an application, even relatively low doping in the substrate ensures a significant reduction in the effects of basal plane dislocations. Doping into the second, previously more highly doped substrate region has no effect on the operation of the semiconductor component. However, the positive effect on suppressing the effects of basal plane dislocations is enhanced if the inventive doping with nitrogen is also carried out in the second substrate region. This is where the basal plane dislocations often originate. The preferred doping concentration of nitrogen doping atoms, which are added to the substrate by irradiation with the ion beam, can also be between 1E14cm -3 and 1E17cm -3 or between 1E14cm -3 and 7E16cm -3 or between 1E14cm -3 and 4E16cm-3 or between 1E14cm -3 and 2E16cm -3 The preferred doping concentration of nitrogen dopant atoms added to the substrate by irradiation with the ion beam can also be between 5E14cm -3 and 1E17cm -3 or between 5E14cm -3 and 7E16cm -3 or between 5E14cm -3 and 4E16cm -3 or between 5E14cm -3 and 2E16cm -3 The preferred doping concentration of nitrogen dopant atoms added to the substrate by irradiation with the ion beam can also be between 1E15cm -3 and 1E17cm -3 or between 1E15cm -3 and 7E16cm -3 or between 1E15cm -3 and 4E16cm -3 or between 1E15cm -3 and 2E16cm -3Likewise, the preferred doping concentration of nitrogen dopant atoms added to the substrate by irradiation with the ion beam can also be between 2E15cm -3 and 3E17cm -3 or between 2E15cm -3 and 7E16cm -3 or between 2E15cm -3 and 4E16cm -3 or between 2E15cm -3 and 2E16cm -3 Likewise, the preferred doping concentration of nitrogen dopant atoms added to the substrate by irradiation with the ion beam can also be between 3E15cm -3 and 3E17cm -3 or between 3E15cm -3 and 7E16cm -3 or between 3E15cm -3 and 4E16cm -3 or between 3E15cm -3 and 2E16cm -3Finally, the preferred doping concentration of nitrogen dopant atoms added to the substrate by irradiation with the ion beam can also be between 4E15cm -3 and 3E17cm -3 or between 4E15cm -3 and 7E16cm -3 or between 4E15cm -3 and 4E16cm -3 or between 4E15cm -3 and 2E16cm -3 The total n-doping in the first substrate region after the implantation according to the invention is between 3E14cm -3 and 5E17cm -3 , preferably between 3E14cm -3 and 2E17cm -3 , particularly preferred between 4E14cm -3 and 7E16cm -3 . The total concentration can preferably also be between 3E15cm -3 and 5E17cm -3 , more preferably between 3E15cm -3 and 1E17cm -3 , particularly preferably between 4E15cm -3 and 7E16cm -3 or between 3E15cm -3 and 3E16cm-3. 14513 P 6523 WO Preferably, each nitrogen-doped section in the first substrate region is aligned with a nitrogen-doped section in the second substrate region, ie a doped section in the first substrate region and an associated doped section in the second substrate region show no offset when viewed in a direction perpendicular to the depth direction. In other words, a doped section in the first substrate region and an associated doped section in the second substrate region lie in the same column in the substrate running in the depth direction, or they each extend over the entire substrate width. In the former case, the substrate must be masked during doping. According to the invention, it is preferred that the first substrate region is doped continuously with nitrogen, at least in sections, over its full extent in the depth direction.The first substrate region preferably forms the outer layer of the substrate facing the ion beam. The entry point of the ion beam into the substrate then lies on the surface of the first substrate region opposite the second substrate region. It is preferably provided that the doping concentration of nitrogen doping atoms added to the substrate by irradiation with the ion beam exhibits a continuous profile in the depth direction of the substrate from a doping start depth, particularly preferably already from the entry point of the ion beam into the substrate, up to a maximum doping depth. In this way, the negative effects of basal plane dislocations are prevented as widely as possible.It is further preferably provided that the doping concentration of nitrogen doping atoms, which are added to the substrate by irradiation with the ion beam, shows a profile in the depth direction from a doping start depth, preferably already from the entry point of the ion beam into the substrate, down to a maximum doping depth, the lowest value of which differs from the highest value by a maximum of 50%, more preferably a maximum of 20%. The relatively constant doping profile generated in this way brings about a uniform effect over a considerable substrate region. 14513 P 6523 WO In preferred embodiments, a maximum doping depth of the ion beam into the substrate is 5 to 40 µm, more preferably 10 to 30 µm. Generally within the scope of the invention, the thickness of the first substrate region in the depth direction is preferably 4 to 25 µm.Generally, within the scope of the invention, the thickness of the second substrate region in the depth direction is preferably 2 to 400 µm. Generally, within the scope of the invention, the depth of the region doped with nitrogen by implantation in the second substrate region is preferably 0.1 to 10 µm, more preferably 0.1 to 5 µm, even more preferably 0.1 to 4 µm. These ranges have proven sufficient to enhance the inventive effect. The depth of the region doped with nitrogen by implantation in the second substrate region can preferably also be between 0.15 and 5 µm or between 0.2 and 5 µm. The upper limit of these ranges can also be generally open, because the implantation in the second substrate region is not significant. However, at greater depths, increasingly higher ion energies would be required, which is why an upper limit makes sense for reasons of efficiency.It is particularly preferred that the ion beam passes through an energy filter before hitting the substrate, said filter being designed as a microstructured membrane for setting a dopant depth profile in the substrate. This ensures the generation of continuous dopant depth profiles in the substrate in a simple manner. The doping of the first and second substrate regions generally preferably takes place in the same process. Preferably, the second substrate region is formed at least partially, preferably completely, by a base substrate, preferably made of monocrystalline SiC, polycrystalline SiC, amorphous SiC, silicon, graphite or layer combinations thereof. 14513 P 6523 WO To further enhance the effect according to the invention, it is preferably provided that the second substrate region is formed at least partially by a buffer layer, which preferably consists of monocrystalline SiC.The thickness of the buffer layer is preferably a maximum of 30% of the first substrate region. The thickness of the buffer layer is also preferably a maximum of 2 µm, more preferably a maximum of 1 µm. Within the scope of the invention, the first substrate region is preferably formed entirely from SiC. The second substrate region can be formed from a base substrate on which the first substrate region was formed by epitaxial deposition. Depending on the design of the base substrate, the second substrate region can thus also be formed from monocrystalline SiC, but also from polycrystalline SiC or amorphous SiC. In the case of epitaxially produced substrates, however, a buffer layer can also be provided between the first substrate region and the actual base substrate. This buffer layer also has an n+ doping. In such cases, the second substrate region can also be this buffer layer or comprise it, as defined in this application.The second substrate region can either extend only to this buffer layer or it can extend beyond the buffer layer into the base substrate. In the latter case, the "second substrate region" would thus possibly be composed of different materials. Nevertheless, the term "second substrate region" is intended, within the scope of the invention, to also encompass such a structure as a zone of higher n-doping. A transition region in which the doping concentration increases steeply can also be located between the first and second substrate regions. The first and second substrate regions, with their respective defined doping concentrations, can therefore, but in practice will not, be directly adjacent to one another. In such a case, it is preferred if the transition region is also doped by means of implantation. If a transition region is present, it is generally no more than 0.5 µm thick.14513 P 6523 WO In the case of an alternative production of the first substrate region, for example by cleaving the first substrate region from a donor substrate and applying this cleaved part to the second substrate region, this second substrate region can also be formed from other materials such as silicon or graphite. The second substrate region is then formed by the entire carrier substrate or a part of the carrier substrate. Generally speaking, within the scope of the invention, the single-crystalline SiC used is preferably SiC in which the wafer surface is arranged at an angle of 4° to a direction perpendicular to the c-direction of the crystal structure. There are several known methods for determining the doping concentration which deliver identical results, for example SIMS. A measurement at 50 nm intervals provides sufficient resolution in the depth direction.In order to determine the concentration of nitrogen ions additionally introduced by implantation into a pre-doped substrate, an otherwise identical reference substrate that is not subjected to implantation and on which the same measurements are carried out as on the target substrate can be used by subtracting the measurement results. Brief description of the drawings Fig. 1 is a schematic view of an irradiation arrangement with an energy filter for ion implantation into a substrate. Fig. 2 is a schematic representation of the mode of operation of an energy filter that can be used in the method according to the invention. Fig. 3 is a schematic representation of various doping profiles that can be generated using differently structured energy filters. 14513 P 6523 WO Fig.4 schematically shows the process of an ion implantation into a layer beginning at a surface of the substrate, as well as a resulting doping profile of the substrate, as can be produced within the scope of the invention. Figs. 5a-c show various further possibilities for doping profiles that can be produced using the method according to the invention. Figs. 6a-c are schematic examples of implantations according to the invention based on cross-sectional views of the respective substrate; Fig. 7 is a representation of the doping concentration of a substrate against the doping depth before and after the implantation; Fig. 8 is a representation of the doping concentration of another substrate against the doping depth before and after the implantation; Figs. 9 to 13 document test results to demonstrate the effectiveness of the implantation according to the invention. Fig. 1 shows an irradiation chamber 8, in which a high vacuum is usually present.In the irradiation chamber 8, the substrate 12 to be doped is held in a substrate holder 30. An ion beam 10 is generated by an ion beam generation device 11, which is usually designed as a particle accelerator, and guided into the irradiation chamber 8. There, the energy of the ion beam 10 is spread by an energy filter 20 and it impinges on the substrate 12 to be irradiated. Alternatively, the energy filter 20 can be arranged in a separate, valve-closable vacuum chamber within the irradiation chamber 8 or directly adjacent to the irradiation chamber 8. The substrate holder 30 does not have to be stationary, but can optionally be provided with a device for displacing the substrate 12 in xy (in the plane perpendicular to the plane of the page).A wafer wheel, on which the substrates 12 to be implanted are fixed and which rotates during implantation, can also be considered as the substrate holder 30. A displacement of the substrate holder 30 in the beam direction (z-direction) is also possible. Furthermore, the substrate holder 30 can optionally be provided with a heater or cooler. The basic principle of the energy filter 20 is shown in Fig. 2. The monoenergetic ion beam 10 is modified in its energy as it passes through the energy filter 20, which is designed as a microstructured membrane, depending on the entry point. The resulting energy distribution of the ions of the ion beam 10 leads to a modification of the depth profile of the implanted substance in the matrix of the substrate 12. E1 denotes the energy of a first ion, E2 denotes the energy of a second ion, c denotes the doping concentration, and d denotes the depth in the substrate 12.In the diagram on the right, the usual Gaussian distribution is identified by reference symbol A, which arises without the use of an energy filter 20. In contrast, a rectangular distribution, which can be achieved when using an energy filter 20, is sketched as an example with reference symbol B. The layouts or three-dimensional structures of energy filters 20 shown in Fig. 3 illustrate the basic possibilities of generating a multitude of dopant depth profiles or defect depth profiles using energy filters 20. c again denotes the doping concentration and d again denotes the depth in the substrate 12. The filter structure profiles can in principle be combined with one another to obtain new filter structure profiles and thus new dopant depth profiles or defect depth profiles. These structures of the energy filter 20 shown in Fig. 3 are also applicable within the scope of the invention, as are many other structure profiles.However, relatively uniform profiles are preferred, such as those shown in Fig. 3 (a) or 3 (b). The microstructures 50 in Fig. 3 (a) have the shape of strip-shaped ribs with a triangular cross-section. The microstructures 50 in Fig. 3 (b) also have the shape of strip-shaped ribs with a triangular cross-section, but with smaller projection dimensions. The microstructures 50 in Fig. 3 (c) have the shape of strip-shaped ribs with an essentially triangular cross-section, but with flattened tips. The microstructures 50 in Fig. 3 (d) have a pyramid shape. The pyramid shape can also have flattened tips. In addition to the shapes described, many other suitable shapes exist. The energy filter 20 is generally made of silicon, preferably of single-crystal, polycrystalline or amorphous silicon.Alternatively, it can be made of SiC, preferably single-crystal, polycrystalline, or amorphous SiC. Other possible materials include ceramic materials. Layer systems with different layers made of the aforementioned materials are also possible. The energy filter 20 has a thickness of between 3 µm and 60 µm, preferably between 5 µm and 50 µm, and particularly preferably between 7 µm and 20 µm. It can be held in a filter frame (not shown). The filter frame can be exchangeably accommodated in a filter holder (not shown). During the doping of a layer or the region 21 to be doped, which begins at the surface of the substrate, as shown in Fig. 4, the ion implantation into the substrate 12 takes place from a front side of the substrate 12.The short, black-filled arrow indicates the ions of minimum energy transmitted through the energy filter 20, and the long, black-filled arrow indicates the ions of maximum energy transmitted through the energy filter 20. The resulting doping profile along section AA' is shown on the right in the coordinate system. c again represents the doping concentration. The doping profile is approximately uniform across the entire region 21. Region 21 extends from the surface of the substrate 12 facing the ion beam 10 to a predetermined doping depth T, where a remaining portion 22 of the substrate 12 can adjoin, which is not affected by the ion implantation by means of the energy filter 20. However, it is also conceivable for the entire substrate 12 to be doped by means of ion implantation. By using masks, it is also possible to dope only partial regions of the substrate 12.It is also conceivable to start the doping at a certain depth in the substrate 12, i.e. not at the surface. All of the embodiments described here for Fig. 4 can be used within the scope of the invention. In general, implantation with ions of nitrogen or phosphorus is particularly suitable for forming an n-doped region 21, while implantation with ions of boron or aluminum is particularly suitable for a p-doped layer. According to the invention, nitrogen ions are implanted at least in partial regions of the substrate 12, because this brings about the effect according to the invention. The thickness of the doped layer or region 21 is preferably between 4 and 30 µm, more preferably between 5 and 25 µm. Figs. 5a-c show possible further preferred doping profiles in the substrate 12. According to the invention, many profile shapes of the doping profile can be produced. However, relatively uniform profiles are preferred, such as, for example,B. shown in Fig. 5a. Figs. 6a-c are schematic examples of implantations according to the invention based on cross-sectional views of the respective substrate, which is divided into two partial regions. In all exemplary embodiments, a first substrate region 14 and a second substrate region 16 are shown. The area outlined in black is the area of ​​the inventive implantation with nitrogen. In the example in Fig. 6a, the doping with nitrogen takes place only in the first substrate region 14 and not continuously down to its maximum depth. The first substrate region 14 could, however, also be doped continuously down to its maximum depth. In the example in Fig. 6b, the doping with nitrogen takes place continuously in the first substrate region 14 from the entry point 26 of the ion beam 10 into the substrate 12 and into the second substrate region 16. In the example in Fig.6c, the doping with nitrogen takes place exclusively in the transition region between the first substrate region 14 and the second substrate region 16, starting at a doping start depth 24 and up to the maximum doping depth T. Fig. 7 shows the doping concentration c of a substrate 12 versus the doping depth d (in µm) before and after the implantation. The implantation region is shown as a dashed rectangle. The doping concentration of the first substrate region 14 before the implantation according to the invention is shown as a solid graph, and the combined doping concentration after the implantation according to the invention is shown as a dashed graph. It can be seen that the implantation according to the invention in the second substrate region 16 has no influence on the overall doping due to the high pre-doping. Fig.Figure 8 is a representation of the doping concentration c of a further substrate 12 against the doping depth d (in µm) before and after the implantation. The same statements apply as for Figure 7. In contrast to Figure 7, here in the second substrate region 16 there is a buffer layer 32 which lies between the base substrate and the first substrate region 14 and has an n-type pre-doping which lies between the doping of the first substrate region 14 and the base substrate. In both cases there is a transition region 34 with a steeply rising flank of the doping profile between the first substrate region 14 and the buffer layer 32 or the second substrate region 16. Figures 9 to 13 document test results to demonstrate the effectiveness of the implantation according to the invention. Basal plane dislocations can act as nuclei for Shockley stacking faults (e.g.Simple Shockley stacking faults (SBS) act when minority charge carriers are injected in the forward direction in a pn diode and energy is deposited at the basal plane dislocations through charge carrier recombination. It is generally accepted that comparable changes occur when the samples are exposed to UV irradiation, simulating the bipolar current load during operation. Example 1 (Fig. 9) The starting point is a three-layer substrate with a first substrate region 8.18 µm thick, an adjacent buffer layer 0.5 µm thick, and a base substrate adjacent to the buffer layer 350 µm thick. All three layers consist of single-crystal 4H-SiC. The buffer layer and the first substrate region are each formed on the base substrate using epitaxial processes. 14513 P 6523 WO The initial n-doping concentration in the first substrate region is 4.75E15cm. -3The initial n-doping concentration in the buffer layer is 1E18cm -3 . The initial n- doping concentration in the base substrate is in the range of at least 5E19cm -3First, the existing basal plane dislocations and simple Shockley stacking faults in the starting substrate were recorded and evaluated using photoluminescence spectroscopy, with the peak of the excitation wavelength at 313 nm. The detection area per image was 1.33 x 1.33 mm. The radiation emitted by the substrate in response to the excitation was measured at a wavelength of 420 nm and, via a high-pass filter, at more than 700 nm. In Fig. 9, the top left row shows the results for a sample location R-1 in the starting state, measured at 420 nm and more than 700 nm. Some mixed dislocations (including basal plane dislocations and simple Shockley stacking faults) can be seen, particularly at > 700 nm (in yellow). The substrate was then irradiated with a Nd:YAG laser in the UV range with a wavelength of 355 nm and an irradiance of 72 W / cm 2irradiated for over 2,000 seconds and the photoluminescence measurements were repeated. In Fig. 9, the results for sample location R-1 in the UV-excited state are shown in the bottom row on the left, again measured at 420 nm and more than 700 nm. As can be clearly seen, the simple Shockley stacking faults (at the locations marked in blue) have spread significantly, i.e. expanded, or more simple Shockley stacking faults were identified. The number of mixed dislocations (yellow) remained unchanged. The expanded simple Shockley stacking faults were then contracted again using a known process (800°C annealing in an argon atmosphere for 60 minutes). 14513 P 6523 WO The substrate was then doped with nitrogen doping atoms by ion implantation using an energy filter.The beam energy was 19.0 MeV and the doping concentration of nitrogen dopant atoms added to the substrate by irradiation with the ion beam was 5E15cm. -3. The implantation was carried out with a uniform doping profile starting from the entry point of the ion beam in the first substrate region down to the buffer layer, specifically a total of 8.35 µm deep, i.e. across the entire first substrate region and 0.17 µm deep into the buffer layer. This was followed by activation annealing to electrically activate the dopants in an argon atmosphere at 1,700°C for 30 minutes. Photoluminescence measurements were then carried out again at the same sample location R-1 as described above. In Fig. 9, the top right row shows the results for sample location R-1 doped according to the invention in the initial state, measured at 420 nm and more than 700 nm. In particular, at > 700 nm the same mixed dislocations (yellow) can be seen as before doping. Sample location R-1 is therefore clearly identifiable.The substrate was then irradiated again with the Nd:YAG laser in the UV range at a wavelength of 355 nm as described above (stress), and the photoluminescence measurements were repeated. Fig. 9 shows the results for sample location R-1 (same sample location as on the left in Fig. 9) in the UV-excited state in the bottom right row, again measured at 420 nm and more than 700 nm. As can be clearly seen, the simple Shockley stacking faults have hardly propagated, or hardly any more simple Shockley stacking faults were identified than in the initial state. After UV irradiation, the defect density without ion implantation at sample location R-1 was 108 / cm. 2 , while the defect density with ion implantation after UV irradiation is only 2 / cm 2was (counting of defects in the 420 nm image, comparison with Fig. 9, bottom row of images 1st and 3rd). The effects of the expansion of basal plane dislocations in the substrate were thus significantly reduced by the implantation of the nitrogen atoms. 14513 P 6523 WO Example 2 (Fig. 10) The starting point is the same three-layer substrate as described in Example 1, but sample location R-2 was investigated. First, the existing basal plane dislocations and simple Shockley stacking faults in the starting substrate were recorded and evaluated using photoluminescence spectroscopy. The procedure was as described in Example 1. The measurement of the radiation emitted by the substrate in response to the excitation was carried out at wavelengths of 420 nm and more than 700 nm, as described in Example 1. In Fig.10, the results of sample location R-2 in the initial state, measured at 420 nm and more than 700 nm, are shown on the left in the upper row.Many mixed dislocations (including basal plane dislocations and Shockley stacking faults), in particular triangular defects (in yellow) and simple Shockley stacking faults (in purple), are already visible. A defect cluster is present here. The substrate was then irradiated with the Nd:YAG laser in the UV range as described in Example 1, and the photoluminescence measurements were repeated. In Fig. 10, the bottom left row shows the results for sample location R-2 in the UV-excited state, again measured at 420 nm and more than 700 nm. As can be clearly seen, the simple Shockley stacking faults (at the locations marked in blue) have spread significantly, i.e., expanded, or more simple Shockley stacking faults were identified. The number of defects marked in purple remained unchanged.The expanded simple Shockley stacking faults were subsequently recontracted using a known method (800°C annealing in an argon atmosphere for 60 minutes). 14513 P 6523 WO The substrate was then doped with nitrogen dopant atoms by ion implantation using an energy filter. The beam energy was 19.0 MeV, and the doping concentration of nitrogen dopant atoms added to the substrate by irradiation with the ion beam was 5E15cm-3. -3. The implantation was carried out with a uniform doping profile starting from the entry point of the ion beam in the first substrate region down to the buffer layer, specifically a total of 8.35 µm deep, i.e. across the entire first substrate region and 0.17 µm deep into the buffer layer. This was followed by activation annealing in an argon atmosphere (at 1,700 °C for 30 minutes) to electrically activate the dopants. Photoluminescence measurements were then carried out again at the same sample location R-2 as previously described. In Fig. 10, the top right row shows the results for sample location R-2 doped according to the invention in the initial state, measured at 420 nm and more than 700 nm. In particular, at > 700 nm the same mixed dislocations can be seen as before doping. This indicates a defect cluster.The substrate was then irradiated again with the Nd:YAG laser in the UV range at a wavelength of 355 nm as described above, and the photoluminescence measurements were repeated. In Fig. 10, the bottom right row shows the results for sample location R-2 in the UV-excited state, again measured at 420 nm and above 700 nm. As can be clearly seen, the simple Shockley stacking faults have hardly propagated, or hardly any more Shockley stacking faults were identified than in the initial state. After UV irradiation, the defect density without ion implantation was 234 / cm. 2 , while the defect density with ion implantation after UV irradiation was only 31 / cm 2The effects of the expansion of basal plane dislocations in the substrate were thus significantly reduced by the implantation of nitrogen atoms. 14513 P 6523 WO Example 3 (Fig. 11) The starting point is a three-layer substrate with a first substrate region of 5.16 µm thickness, an adjacent buffer layer of 0.5 µm thickness, and a base substrate of 350 µm thickness adjacent to the buffer layer. All three layers consist of single-crystal 4H-SiC. The buffer layer and the first substrate region are each formed on the base substrate by epitaxial growth. The initial n-doping concentration of nitrogen atoms in the first substrate region is < 5E14cm -3 The initial n-doping concentration in the buffer layer is 1E18cm -3 . The initial n-doping concentration in the base substrate is in the range of at least 1E19cm -3The substrate was doped with nitrogen dopant atoms by ion implantation using an energy filter. The beam energy was 12.5 MeV, and the doping concentration of nitrogen dopant atoms added to the substrate by ion beam irradiation was 4.85E15cm -3. The implantation was carried out with a uniform doping profile starting from the entry point of the ion beam into the first substrate region down to the buffer layer, specifically a total of 5.45 µm deep, i.e. across the entire first substrate region and 0.29 µm deep into the buffer layer. This was followed by activation annealing in an argon atmosphere (at 1,700 °C for 30 minutes) to electrically activate the dopants. Photoluminescence measurements were then carried out at two sample locations T-1 and T-2 as described in Example 1. The top row of Fig. 11 shows the results for the sample locations T-1 and T-2 doped according to the invention in the initial state, each measured at 420 nm and more than 700 nm. A few simple Shockley stacking faults can be seen.The substrate was then irradiated again with the Nd:YAG laser in the UV range at a wavelength of 355 nm as described in Example 1, and the photoluminescence measurements were repeated. 14513 P 6523 WO In Fig. 11, the bottom row shows the results for sample locations T-1 and T-2 in the UV-excited state, again measured at 420 nm and more than 700 nm. As can be clearly seen, the simple Shockley stacking faults have hardly propagated, or hardly any more Shockley stacking faults were identified than in the initial state. No new faults were created by ion implantation. The effects of basal plane dislocations in the substrate were almost eliminated here by the implantation of the nitrogen atoms. Example 4 (Fig.12a and 12b) The starting point is a three-layer substrate with a first substrate region 5.01 µm thick, an adjacent buffer layer 0.5 µm thick, and a base substrate 350 µm thick adjacent to the buffer layer. All three layers consist of single-crystal 4H-SiC. The buffer layer and the first substrate region are each formed on the base substrate by epitaxial growth. The initial n-type doping concentration of nitrogen atoms in the first substrate region is < 5E14cm³. -3 The initial n-doping concentration in the buffer layer is 1E18cm -3 . The initial n-doping concentration in the base substrate is in the range of at least 1E19cm -3The substrate was partially doped with nitrogen doping atoms by ion implantation using an energy filter. The beam energy was 19.0 MeV. As shown in Fig. 12a, the substrate was divided into four quadrants Q1 to Q4, one of which was not doped at all and the others were doped to varying degrees. While a first quadrant Q1 (bottom left) was not doped at all by ion implantation, in a second quadrant Q2 (top right), the doping concentration of nitrogen doping atoms added to the substrate by irradiation with the ion beam was 1E15cm. -3 , in a third quadrant Q3 (top left) 5E15cm -3 and in a fourth quadrant Q4 (bottom right) 1E16cm -3The same mask was used for each implantation, leaving only one quadrant free at a time. The substrate, on the other hand, was mounted so that it could rotate and was moved into the desired position for the 14513 P 6523 WO implantation in each quadrant. The three implantations were carried out one after the other. The implantation took place in the second, third and fourth quadrants with a uniform doping profile starting from the entry point of the ion beam, specifically a total of 8.35 µm deep, i.e. over the entire first substrate area and the entire buffer layer up to 2.84 µm deep into the base substrate. This was followed by activation annealing in an argon atmosphere (at 1,700 °C for 30 minutes) to electrically activate the dopants. The substrate was then irradiated with a Nd:YAG laser in the UV range with a wavelength of 355 nm as described in Example 1, but with an irradiance of 142 W / cm 2over 4,000 seconds, and photoluminescence measurements were subsequently carried out as described in Example 1. The radiation emitted by the substrate in response to the excitation was measured at a wavelength of 420 nm. In Fig. 12b, the density D of the expanded simple Shockley stacking faults is plotted graphically for the individual quadrants Q1 to Q4. The density D in the first quadrant was Q191 / cm 2 , in the second quadrant Q28 / cm 2 , in the third quadrant Q32 / cm 2 and in the fourth quadrant Q46 / cm 2This clearly demonstrates the effect of ion implantation in the second, third, and fourth quadrants. Example 5 (Figs. 13a and 13b) The starting point is a three-layer substrate with a first substrate region of 5.0 µm thickness, an adjacent buffer layer of 0.5 µm thickness, and a base substrate of 350 µm thickness adjacent to the buffer layer. All three layers consist of single-crystal 4H-SiC. The buffer layer and the first substrate region are each formed on the base substrate by epitaxial growth. The initial n-type doping concentration of nitrogen atoms in the first substrate region is < 5E14cm -3 The initial n-doping concentration in the buffer layer is 1E18cm -3 . The initial n-doping concentration in the base substrate is in the range of at least 1E19cm -3. 14513 P 6523 WO The substrate was partially doped with nitrogen doping atoms by ion implantation using an energy filter. The beam energy was 19.0 MeV. As shown in Fig. 13a, the substrate was divided into four quadrants Q1 to Q4, one of which was not doped at all and the others were doped to varying degrees. While a first quadrant Q1 (bottom left) was not doped at all by ion implantation, in a second quadrant Q2 (top right), the doping concentration of nitrogen doping atoms added to the substrate by irradiation with the ion beam was 5E14cm -3 , in a third quadrant Q3 (top left) 1E14cm -3 and in a fourth quadrant Q4 (bottom right) 1E15cm -3The same mask was used for each implantation, leaving only one quadrant exposed. The substrate, however, was mounted so that it could rotate and then moved into the desired position for implantation in each quadrant. Thus, the three implantations were performed one after the other. The implantation took place in the second, third, and fourth quadrants with a uniform doping profile starting from the entry point of the ion beam, specifically a total of 8.35 µm deep, i.e., across the entire first substrate area and the entire buffer layer, extending to a depth of 2.85 µm into the base substrate. This was followed by activation annealing in an argon atmosphere (at 1,700 °C for 30 minutes) to electrically activate the dopants. The substrate was then irradiated with an Nd:YAG laser in the UV range at a wavelength of 355 nm, as described in Example 1, but with an irradiance of 142 W / cm 2over 4,000 seconds, and photoluminescence measurements were subsequently carried out as described in Example 1. The radiation emitted by the substrate in response to the excitation was measured at a wavelength of 420 nm. In Fig. 13b, the density D of the expanded simple Shockley stacking faults is plotted graphically for the individual quadrants Q1 to Q4. The density D in the first quadrant was Q110 / cm 2 , in the second quadrant Q22 / cm 2 , in the third quadrant Q33 / cm 2 and in the fourth quadrant Q44 / cm 2This clearly demonstrates the effect of ion implantation in the second, third, and fourth quadrants compared to the untreated substrate in the first quadrant, which already had only a few expanded simple Shockley stacking faults. 14513 P 6523 WO In the previous description, a preferred implantation using an energy filter was described. However, it is also possible to use monoenergetic ion beams without an energy filter. In this case, many individual pulses with different beam energies are required. The processes described generally serve to contribute to the production of semiconductor components. For example, the implantation according to the invention can not only reduce the effect of basal plane dislocations, but the drift zone of the semiconductor component can also be partially or completely doped in the same process.To manufacture a semiconductor component, further processes are necessary after the implantation according to the invention. The substrate 12 can be transformed into a finished semiconductor component through further steps, for example, by implanting additional active regions, producing oxides, depositing gate electrodes, contacts, lines, or vias, etc. 14513 P 6523 WO.

Claims

Claims 1. Use of an ion implantation with an ion beam (10) with ions of nitrogen into a substrate (12) for reducing the effects of basal plane dislocations in the substrate (12), wherein the substrate (12) in the region (14, 16, 21) to be doped or the regions to be doped is at least partially formed from SiC, preferably 4H-SiC, and wherein the doping concentration (c) of nitrogen doping atoms, which are added to the substrate (12) by the irradiation with the ion beam (10), in each region (14, 16, 21) to be doped is between 1E14cm -3 and 5E17cm -3 , preferably between 5E14cm -3 and 5E17cm -3 , more preferably between 1E15cm -3 and 5E17cm -3 , more preferably between 1E15cm -3 and 3E17cm -3 , more preferably between 2E15 cm -3 and 2E17cm -3 , more preferably between 3E15cm -3 and 1E17cm -3 , particularly preferably between 4E15cm-3 and 8E16cm -32. Use of an ion implantation according to claim 1, characterized in that the doping concentration (c) of nitrogen doping atoms, which are added to the substrate (12) by irradiation with the ion beam (10), exhibits a continuous profile in the depth direction of the substrate (12) from a doping start depth (24), preferably already from the entry point (26) of the ion beam (10) into the substrate (12), up to a maximum doping depth (T). 3.Use of an ion implantation according to claim 1 or 2, characterized in that the doping concentration (c) of nitrogen doping atoms, which are added to the substrate (12) by irradiation with the ion beam (10), shows a profile in the depth direction from a doping start depth (24), preferably already from the entry point (26) of the ion beam (10) into the substrate (12), up to a maximum doping depth (T), the lowest value of which differs by a maximum of 50%, preferably a maximum of 20%, from the highest value.

4. Use of an ion implantation according to one of the preceding claims, characterized in that the ion beam (10) passes through an energy filter (20) before striking the substrate (12), which energy filter is designed as a microstructured membrane for setting a dopant depth profile in the substrate (12). 14513 P 6523 WO.

5. Use of an ion implantation according to one of the preceding claims, characterized in that a maximum doping depth (T) of the ion beam (10) into the substrate (12) is 5 to 40 µm, preferably 10 to 30 µm.

6. Use of an ion implantation according to one of the preceding claims, characterized in that the energy of the ion beam (10) is in the range of 3 to 50 MeV, preferably in the range of 5 to 25 MeV.

7. Use of an ion implantation according to one of the preceding claims, characterized in that an annealing step at 1,650 to 1,750 °C takes place after the implantation with nitrogen. 8.Method for implanting doping atoms into a substrate (12) or for reducing the effects of basal plane dislocations in a substrate during the production of semiconductor components, comprising the following steps: - providing a substrate (12) with two substrate regions (14, 16) arranged one above the other in a depth direction of the substrate (12), wherein the first substrate region (14) is a region made of monocrystalline SiC, preferably 4H-SiC, which is undoped or weakly n-doped with a doping concentration of at most 6E15cm. -3 , and wherein the second substrate region (16) is a heavily n-doped region of single-crystalline SiC, polycrystalline SiC, amorphous SiC, silicon, graphite or layer combinations thereof with a doping concentration of at least 3E17cm -3 , preferably at least 5E17cm -3 , more preferably at least 9E17cm -3- irradiating the substrate (12) with an ion beam (10) with ions of nitrogen from a side of the first substrate region (14) which is opposite the second substrate region (16), such that the first substrate region (14) is doped with nitrogen at least in sections and the second substrate region (16) is doped with nitrogen at least in sections, wherein the doping concentration (c) of nitrogen doping atoms which are added to the substrate (12) by the irradiation with the ion beam (10) is between 1E14cm -3 and 14513 P 6523 WO 5E17cm -3 , preferably between 5E14cm -3 and 5E17cm -3 , more preferably between 1E15cm -3 and 5E17cm -3 , more preferably between 1E15cm -3 and 3E17cm -3 , more preferably between 2E15 cm -3 and 2E17cm -3 , more preferably between 3E15cm -3 and 1E17cm -3, particularly preferably between 4E15cm -3 and 8E16cm -39. The method according to claim 8, characterized in that each nitrogen-doped section in the first substrate region (14) is aligned with a nitrogen-doped section in the second substrate region (16).

10. The method according to claim 8 or 9, characterized in that the first substrate region (14) is doped with nitrogen at least in sections over its full extent in the depth direction, wherein the first substrate region (14) preferably forms the outer layer of the substrate (12) facing the ion beam (10). 11.Method according to one of claims 8 to 10, characterized in that the doping concentration (c) of nitrogen doping atoms, which are added to the substrate (12) by the irradiation with the ion beam (10), shows a continuous profile in the depth direction from a doping start depth (24), preferably already from the entry point (26) of the ion beam (10) into the substrate (12), up to a maximum doping depth (T).

12. The method according to any one of claims 8 to 11, characterized in that the doping concentration (c) of nitrogen doping atoms added to the substrate (12) by irradiation with the ion beam (10) exhibits a profile in the depth direction from a doping start depth (24), preferably already from the entry point (26) of the ion beam (10) into the substrate (12), up to a maximum doping depth (T), the lowest value of which differs from the highest value by a maximum of 50%, preferably a maximum of 20%.Method according to one of claims 8 to 12, characterized in that a maximum doping depth (T) of the ion beam (10) into the substrate (12) is 5 to 40 µm, preferably 10 to 30 µm. 14513 P 6523 WO.

14. The method according to any one of claims 8 to 13, characterized in that the thickness of the first substrate region (14) in the depth direction is 4 to 25 µm.

15. The method according to any one of claims 8 to 14, characterized in that the thickness of the second substrate region (16) in the depth direction is 2 to 400 µm.

16. The method according to any one of claims 8 to 15, characterized in that the depth of the region doped with nitrogen by means of the implantation in the second substrate region (16) is at least 0.1 µm, preferably at least 0.15 µm, more preferably at least 0.2 µm.

17. The method according to any one of claims 8 to 16, characterized in that the ion beam (10) passes through an energy filter (20) before impinging on the substrate (12), which energy filter is designed as a microstructured membrane for adjusting a dopant depth profile in the substrate (12).Method according to one of claims 8 to 17, characterized in that the second substrate region (16) is formed at least partially or completely by a base substrate, preferably made of monocrystalline SiC, polycrystalline SiC, amorphous SiC, silicon, graphite or layer combinations thereof.

19. Method according to one of claims 8 to 18, characterized in that the second substrate region (16) is formed at least partially by a buffer layer (32), which preferably consists of monocrystalline SiC.

20. Method according to claim 19, characterized in that a thickness of the buffer layer (32) is at most 2 µm, preferably at most 1 µm. 14513 P 6523 WO.

Images