Improved doped inp crystals

Abstract

The present invention relates to the doping of indium phosphide. It has been observed here that semiconductive S-doped InP crystals have an elevated probability of twinning, and that the lattice constant is lowered slightly as a result of the addition of the dopant S, but lowered considerably as a result of further addition of a substance which is isoelectronic to InP, for example Ga. Co-doping with a material which is isoelectronic to InP, for example Ga, thus permits the variable adjustability of the lattice constant.

Description

[0001] Improved doped InP crystals

[0002] The present invention relates to doped InP crystals and InP rods and to a method for producing them.

[0003] BACKGROUND OF THE INVENTION

[0004] Indium phosphide (InP) is playing an increasingly important role in today's semiconductor market. It is an important semiconductor material for optoelectronic and microelectronic devices. A wide variety of optical components, such as lasers for data transmission, are manufactured using indium phosphide. Other applications include solid-state light emission, microwave communications, fiber optic communications, microwave and millimeter-wave devices, and radiation-resistant solar cells.

[0005] Indium phosphide (InP) is a Group III compound semiconductor material that combines indium (In) as a Group III element and phosphorus (P) as a Group V element. Compared to germanium and silicon, InP has a direct band gap structure, high electrical-to-optical energy conversion efficiency, and high electron mobility. It can be easily processed into a semi-insulating material, making it suitable for radio-frequency, millimeter- and terahertz-wave devices and circuits. Semiconducting InP materials also have many applications; for example, such materials are suitable for optoelectronic components, such as edge-emitter lasers for data transmission or infrared photodiodes for environmental sensing.

[0006] To produce InP components, for example, an epitaxial layer is grown on a single-crystal InP wafer as a substrate. Electronic components are then fabricated from the grown material. Various dopants are possible for indium phosphide. For example, sulfur (S) or tin (Sn) can be used to produce semiconducting indium phosphide, while iron can be used for semi-insulating indium phosphide. Electrically inactive dopants are also available for InP, such as Ga, B, Al, Ti, As, Sb, and Bi.

[0007] In the prior art, the use of Sb, As and Ga as dopants for InP is known from G. JACOB, M. DUSEAUX, JP FARGES, MMB VAN DEN BOOM and PJ ROKSNOER "DISLOCATION-FREE GaAs AND InP CRYSTALS BY ISOELECTRONIC DOPING” Journal of Crystal Growth 61 (1983) 417- 424. These were added to the melt as suitable III / V compounds. Crystal growth is carried out here using a Liquid Encapsulated Czochralski (LEC) process. By doping, a reduction of the dislocation density in the upper part of the crystal rod could be achieved.

[0008] Furthermore, the article by L. Haji, R. Coquille, M. Gauneau, and Y. Toudic, "X-ray topography and tem studies of (Ga,Fe)-double-doped LEC grown InP crystals," Journal of Crystal Growth 82 (1987), 487-494, shows that doping with Fe and Ga leads to a very low dislocation density for semi-insulating InP. Here, too, the liquid-encapsulated Czochralski (LEC) process was used.

[0009] The article by Akinori Katsui and Shun-ichi Tohno, "LEC GROWTH AND CHARACTERIZATION OF Ga DOPED InP CRYSTALS," Journal of Crystal Growth 74 (1986) 221-224, also discloses the doping of InP with Ga, which is a non-electrically active dopant. Here, too, the goal of the investigations was to obtain dislocation-poor InP.

[0010] One problem that can occur with doping, especially with sulfur, is that twinning can occur during crystal growth.

[0011] Another problem can be that the incorporation of the dopant can change the lattice constant. When growing crystals, the probability of twinning is very high, especially in the cone region of the crucible. These twins can grow into the cylindrical region of the crystal and even through the entire crystal. However, it is important that all devices are produced in an epitaxial layer with the same crystal structure and orientation. This, in turn, requires that the InP substrate be single-crystalline. A substrate containing regions with undesired crystal orientations would lead to the failure of the devices in these regions and is therefore not economically viable. Cylindrical regions above the cone must therefore often be discarded due to twinning, which leads to yield losses.

[0012] A crystal twin consists of two intergrown crystals that have the same chemical composition and crystal structure. However, a twin is a crystal region with a mirror-symmetric orientation compared to the rest of the crystal.

[0013] The formation of crystal twinning can be caused either by a structural disturbance in the crystal nucleus, from which the crystal twin develops through uniform growth. This is also known as growth twinning. On the other hand, twinning can also be caused by external influences, such as mechanical stress or temperature-induced phase transitions in the material in question. These would then be, for example, pressure or transformation twinning.

[0014] To avoid unwanted twinning during crystal growth, various approaches are mentioned in the literature.

[0015] A well-known article on this topic is the article by J. Amon, F. Dumke, and G. Müller, "Influence of the crucible shape on the formation of facets and twins in the growth of GaAs by the vertical gradient freeze technique," Journal of Crystal Growth 187 (1998), pp. 1-8. This article investigated the influence of the crucible cone angle on facet growth. The optimal crucible cone angle range was specified as 95.3°–107.3°.

[0016] Furthermore, the prior art article by JP Tower, R. Tobin, PJ Pearah, RM Ware, "Interface shape and crystallinity in LEC GaAs," J. Crystal Growth 114 (1991) 665, is known. It was found that it is possible to reduce twinning by reducing the instability of the melt or the extent of facet growth.

[0017] In particular, the use of an axial magnetic field of 2000 G prevents twinning by stabilizing the melt. However, it has also been found that the magnetic field favors polycrystalline growth by enhancing the downward movement of the interface edge.

[0018] The lattice constant of the components also significantly influences their properties and function. In optical components, the lattice constant of the InP determines the band gap and thus also the wavelength of the emitted light. High doping, for example, with sulfur, may be necessary for such components, as they must be able to conduct high currents without generating a significant amount of heat.

[0019] It is possible that defined lattice constants of the substrate are required for certain epitaxial processes, because an epitaxial layer always adopts the lattice constant of the substrate.

[0020] In other words, substrates with different dopant concentrations or charge carrier concentrations have different lattice constants.

[0021] The influence of the dopant concentration on the lattice constant is particularly noticeable for semiconducting substrates with high dopant concentrations. If the dopant concentration of the substrate is significantly increased to reduce the series resistance of the component to be subsequently produced by epitaxy, the lattice constant of the substrate and thus of the epitaxial layer and the component changes.

[0022] It can therefore be considered an object of the present invention to provide doped indium phosphide crystals and doped indium phosphide rods which have the lowest possible density of twins and in which the lattice constant can be adjusted as variably as possible even at high dopant concentrations (e.g., S). SUMMARY OF THE INVENTION

[0023] This object is achieved by an indium phosphide crystal according to claim 1 and a method according to claim 9. Further advantageous embodiments of the present invention are the subject of the corresponding subclaims.

[0024] Without limiting the invention, the following points are presented to describe main aspects, preferred embodiments and special features of the present invention as follows:

[0025] 1 . Indium phosphide crystal containing: at least one first dopant (D1) which imparts semi-insulating or semiconducting properties to the crystal, at least one second dopant (D2) which is isoelectronic to indium or phosphorus, wherein the at least one first dopant (D1) and the at least one second dopant (D2) are different from each other.

[0026] Isoelectronic to indium or phosphorus means that Ga and In are located in the same main group of the periodic table of elements, thus exhibiting isoelectronic behavior with InP. Ga and In are not electrically active with each other and are incorporated isoelectronically at their respective lattice sites.

[0027] 2. Indium phosphide crystal according to item 1, wherein the indium phosphide crystal is semiconducting.

[0028] 3. Indium phosphide crystal according to item 1 or 2, wherein a first dopant (D1) is sulfur.

[0029] In this case, sulfur is the main dopant responsible for the charge carrier concentration. The indium phosphide crystal according to any one of items 1 to 3, wherein a second dopant (D2) is selected from gallium, boron, antimony, and arsenic, or a combination thereof.

[0030] Gallium, boron, antimony and / or arsenic is / are the dopant(s) which is / are isoelectronic to indium or phosphorus. Indium phosphide crystal according to item 4, wherein a second dopant (D2) comprises gallium. Indium phosphide crystal according to any one of items 1 to 5, wherein the concentration of the first dopant (D1) is 1*10 18 up to 2*10 19 Atoms / cm 3 and / or the concentration of the second dopant (D2) is at least 1 * 10 17 Atoms / cm 3 amounts.

[0031] From a concentration of the second dopant (D2) of at least 1*10 17 Atoms / cm 3there is a positive effect on twinning and / or the possibility of lowering the lattice constant.

[0032] Preferably, the concentration of the first dopant (D1 ) is 2.4*10 18 up to 1*10 19 Atoms / cm 3 , and / or the concentration of the second dopant (D2) is at least 2.8*10 17 Atoms / cm 3 This is particularly beneficial for reducing twinning.

[0033] The concentration of the first dopant in the crystal is determined, for example, using glow discharge mass spectrometry (GDMS).

[0034] The concentration of the second dopant in the crystal is determined, for example, by glow discharge mass spectrometry (GDMS). Indium phosphide crystal according to point 6, with a concentration of the second dopant (D2) of 2.8*10 17 up to 3.4*10 19 Atoms / cm 3 amounts.

[0035] The concentration of the second dopant in the crystal is determined, for example, by glow discharge mass spectrometry (GDMS). An indium phosphide crystal according to any one of items 1 to 7, wherein the crystal further contains oxygen. An indium phosphide crystal according to item 8, wherein the oxygen concentration is 5*10 16 up to 1.3*10 17 Atoms / cm 3 amounts.

[0036] The oxygen concentration was determined using SIMS (secondary ion mass spectrometry). Indium phosphide crystal according to one of points 1 to 9, which has a diameter of 45 mm to 205 mm.

[0037] Standardized nominal dimensions are, for example, 50.8 mm, 76.2 mm, 100 mm, 150 mm, and 200 mm. Indium phosphide crystal according to item 9, which has a diameter of 45 mm to 155 mm. Indium phosphide crystal according to item 10, which has a diameter of 95 mm to 155 mm. Indium phosphide crystal according to item 6, wherein the concentration of the first dopant (D1) is 1.3*10 18 up to 8.5*10 18 Atoms / cm 3 and / or the concentration of the second dopant (D2) is at least 2.9*10 18 Atoms / cm 3 Indium phosphide crystal according to item 12, wherein the concentration of the second dopant (D2) is 2.9*10 18 up to 2*10 19 Atoms / cm 3 Indium phosphide crystal containing S as the first dopant (D1 ) with a concentration of 1 .9*10 18 up to 4.4*10 18 Atoms / cm 3wherein the lattice constant of the doped indium phosphide is 5.868 to 5.87 Angstroms (586.8 to 587.00 pm), preferably 5.8685 to 5.87 Angstroms (586.85 to 586.70 pm).

[0038] 16. Indium phosphide crystal according to item 15, wherein the first dopant (D1) is sulfur and the second dopant (D2) is gallium.

[0039] 17. Indium phosphide crystal according to item 13 or 14, wherein the lattice constant of the doped indium phosphide is 5.868 to 5.87 angstroms (586.8 to 587.00 pm), preferably 5.8685 to 5.869 angstroms (586.85 to 586.90 pm).

[0040] 18. Indium phosphide crystal according to any one of items 1 to 17, wherein the Vickers hardness of the crystal is > 380 HV0.1. For S-Ga co-doped InP, the Vickers hardness can be > 400 HV0.1.

[0041] 19. Indium phosphide rod, where the resistivity is higher at the tip than at the end.

[0042] 20. Indium phosphide wafer consisting of indium phosphide crystal according to any one of items 1 to 18, wherein the wafer has a diameter of at least 50.8 mm, preferably 76.2 mm, more preferably at least 100 mm, even more preferably at least 150 mm, and even more preferably 200 mm, and / or a thickness between 450 pm and 675 pm.

[0043] Wafers with diameters of 150 mm and above typically have a thickness of 675±25 pm, while wafers with diameters of 100 mm and less typically have a thickness of 625±25 pm.

[0044] 21. A method for producing an indium phosphide rod, comprising the following steps: a) providing a crucible with a crystal nucleus; b) filling the crucible with polycrystalline indium phosphide c) adding a first dopant or a first dopant compound (DV1) and adding a second dopant or dopant compound (DV2) different from the first, d) optionally adding boron oxide (B2O3); e) creating an inert atmosphere in the crucible; f) melting the contents of the crucible by means of at least one heater, whereby the crystal nucleus is at least partially melted; g) forming a temperature gradient in the longitudinal direction; h) shifting the temperature field in order to achieve solidification of the melt.

[0045] 22. The method according to item 21, wherein the first dopant compound (DV1) is In2Ss and the second dopant compound (DV2) is GaP.

[0046] Adding pure Ga wouldn't work, as it would flow to the nucleus in liquid form and not form a single crystal. Furthermore, P has the advantage that it is not a contaminant and does not change the stoichiometry.

[0047] 23. The process according to item 22, wherein In2Ss is added in powder form and / or GaP is added in the form of crystalline GaP.

[0048] 24. The method according to item 21, wherein GaP is added in the form of GaP wafer pieces.

[0049] 25. The method according to item 21, wherein the first dopant compound (DV1) is In2Ss and the second dopant compound (DV2) is GaAs.

[0050] 26. A process according to any one of items 21 to 25, wherein in step e) the inert gas is selected from a noble gas and nitrogen or a combination thereof.

[0051] 27. A process according to item 26, wherein the inert gas is argon. 28. A process according to any one of items 21 to 27, wherein the process is a VGF (vertical gradient freeze) process.

[0052] 29. A process according to any one of items 21 to 28, wherein after step c) or d) the crucible is sealed into a quartz glass ampoule.

[0053] 30. Process according to one of items 21 to 29, wherein during at least one of steps f), g) and h) the melt is mixed by at least one electromagnetic field.

[0054] This generates convection, which is caused by the action of at least one electromagnetic field on the electrically charged melt. Due to the high effective distribution coefficient of gallium (approximately 4), gallium is incorporated at the crystal end at a significantly lower concentration than at the crystal head. Convection enhances mixing and influences the shape of the phase boundary, which affects the axial homogeneity of the through-material incorporation. Furthermore, constitutional undercooling can be avoided with high dopant additions.

[0055] 31 .A method according to any one of items 21 to 30, further comprising singulating an InP rod into a plurality of wafers.

[0056] DESCRIPTION OF THE CHARACTERS

[0057] In the following, preferred embodiments of the present invention are described in more detail with reference to the accompanying figures.

[0058] Fig. 1 shows a diagram comparing the lattice constants for an undoped indium phosphide crystal and several doped indium phosphide crystals. Fig. 2 shows an indium phosphide crystal (one photographic view and one schematic view), each as a sulfur-doped rod and as a rod with sulfur-gallium codoping.

[0059] Fig. 3 again shows an indium phosphide crystal (one view as a photograph and one view as a schematic representation), each as a sulfur-doped rod and as a rod with a sulfur-gallium co-doping, but here on an etched surface.

[0060] Fig. 4 shows <110> -oriented structure-etched longitudinal sections of an indium phosphide rod (one view as a photograph and one view as a schematic representation), respectively as a sulfur-doped rod and as a rod with a sulfur-gallium codoping.

[0061] Fig. 5 shows two diagrams showing EPD mappings of the head of a rod with a sulfur-gallium codoping and the head of a rod without gallium codoping.

[0062] Fig. 1 shows the lattice constants of indium phosphide for different crystals: for an undoped InP material, two examples for a material doped only with sulfur, and six examples for a material co-doped with sulfur and gallium.

[0063] When doping only with sulfur, the lattice constant drops slightly - and when co-doping with Ga is also carried out, it can be seen that at a minimum concentration of gallium the lattice constants decrease noticeably.

[0064] Fig. 2 compares an indium phosphide crystal doped only with sulfur (a) and one co-doped with gallium and sulfur (b). Here, the crystals are viewed axially from the direction of the nucleus. A photograph of each is shown at the top, and a schematic sketch with outlines in solid lines and dashed lines for the twins is shown below. This shows that twinning occurs particularly at the tip of the rod when doping only with sulfur and an isoelectronic dopant such as gallium is not added as a co-dopant.

[0065] However, if codoping with sulfur and gallium is present, no twinning can be detected.

[0066] Fig. 2 c) shows an enlarged view of Fig. 2 b).

[0067] In Fig. 3, the crystals are photographed from a different direction, namely from above, after the crystals had been etched. Fig. 3(a) shows that dislocations occur across the surface, primarily with sulfur-doped indium phosphide, thus forming twins. However, when doped with both sulfur and gallium, as shown in Fig. 3(b), twinning does not occur.

[0068] In Fig. 4, longitudinal sections <110> oriented through the crystals shown above after structural etching, shown both as a photo and as a sketch. The solid line is the outline, the dashed lines are twins, and the dash-dotted lines are the envelopes of the individual facets. Fig. 4 (a) shows that clear twins have formed in the region of the arrow when doped only with sulfur. Without codoping - Fig. 4 (b) - the facet rebuilds very slowly after its collapse, which is seen from the nucleus, and accordingly no twinning occurs. In the codoped case, the facet rebuilds very quickly after collapse; in this case, no twinning occurs. The edge of the facet towards the interior of the crystal is less sharply defined than in the case without codoping.

[0069] Fig. 5 shows two EPD maps. Here, a test slice was cut from the sample toward the nucleus. The EPD map is created from a test slice that was structurally etched. The structural etch is adjusted to the crystal orientation of the wafer surface {100}. Only the {100} wafer surface is etched in this case, and thus etch pits are only visible on {100}-oriented planes. The etch pit density is shown on fields of 0.5 mm x 0.5 mm. There are therefore 200 fields per 100 mm. Fig. 5 a) shows an EPD map of a test slice from a sample in which no twins occur. The average etch pit density is 669 cm -2 . The etch pits can be seen in all areas of the map.

[0070] Fig. 5 b) shows an EPD map of a test slice of a sample in which twinning occurs. The surface of a twin on such a test slice exhibits a different crystallographic orientation, which is why dislocations are not etched. The areas at the top and right of the image show correct crystal orientation. However, there is a sharp demarcation from the lower left quarter of the image, where incorrect crystal orientation is present due to twinning. No etch pits can be detected here using structural etching for {100} surfaces.

[0071] Detailed description of the invention

[0072] An indium phosphide crystal according to the invention contains at least one first dopant which imparts semi-insulating or semiconducting properties to the crystal and at least one second dopant which is isoelectronic to indium or phosphorus.

[0073] The first dopant and the second dopant are different from each other.

[0074] The indium phosphide which is doped only with a first dopant, for example sulfur or iron, has the risk of twinning occurring during the growth of a rod.

[0075] It has been shown within the present invention that twinning can be significantly reduced by adding isoelectronic elements as a second dopant.

[0076] Due to a high concentration of the dopant element for semiconducting InP (in this case S), the lattice constant of the crystal decreases slightly.

[0077] It was also shown in the present application that by adding a dopant which is isoelectronic to indium or phosphorus, the lattice constant could be significantly reduced again, so that corresponding codoped indium phosphide crystals have a significantly lower lattice constant than completely undoped indium phosphide.

[0078] Preferably, the first dopant is sulfur, and a second dopant is selected from gallium, boron, antimony, and arsenic, or a combination thereof. Preferably, the second dopant contains gallium.

[0079] Co-doping with gallium leads to the fact that, due to its electronic structure, gallium basically occupies In sites and is slightly smaller in ionic radius than indium, which leads to a relaxation of the lattice.

[0080] According to Vegard's law, there are several influences on the lattice constant – a size effect and an electronic effect. Sulfur and phosphorus have approximately the same atomic size – the size effect is considered negligible here. However, the electronic effect is present, which is why the lattice constant of S-doped InP is slightly smaller than that of undoped InP. Gallium, however, has a different atomic size than indium, which is why the size effect becomes dominant when co-doped with Ga, and the lattice constant is significantly reduced.

[0081] Preferably, the concentration of the first dopant is 1*10 18 up to 2*10 19 Atoms / cm 3 and the concentration of the second dopant is at least 1*10 17 Atoms / cm 3 .

[0082] It was observed in the context of the present invention that, especially with higher doping with the first dopant, a concentration of at least 1*1017 Atoms / cm 3 , preferably at least 2.8*10 17 Atoms / cm 3 sufficient to observe at least one of the corresponding effects.

[0083] Preferably, the indium phosphide crystal has a diameter of 45 mm to 205 mm. These are standard sizes, and twinning can be avoided at all diameters. Likewise, the lattice constant can be reduced for all diameters. Further preferably, the concentration of the first dopant is 1.3*10 18 up to 8.5*10 18 Atoms / cm 3 , and the concentration of the second dopant is at least 2.9*10 18 Atoms / cm 3 .

[0084] Such concentrations lead in particular to the possibility of a reduction in the lattice constant - despite the addition of sulfur as a dopant.

[0085] More preferably, the lattice constant of the doped indium phosphide is 5.8680 to 5.870 angstroms (586.8 to 587.00 pm).

[0086] Preferably, the first dopant is sulfur and the second dopant is gallium.

[0087] Such values ​​are lower than the lattice constant of undoped InP.

[0088] An indium phosphide wafer according to the invention consists of an indium phosphide crystal and has a diameter of at least 50.8 mm, preferably 76.2 mm, more preferably at least 100 mm, even more preferably at least 150 mm, and even more preferably 200 mm, and / or has a thickness between 450 pm and 675 pm.

[0089] A method according to the invention for producing an indium phosphide rod, for example by a vertical gradient freeze (VGF) method, comprises the following steps: a) providing a crucible with a crystal nucleus; b) filling this crucible with polycrystalline indium phosphide and adding a first dopant compound and a second dopant compound and optionally boron oxide; c) creating an inert atmosphere in the crucible; d) melting the contents of this crucible by at least one heater, whereby the crystal nucleus is at least partially melted; e) forming a temperature gradient in the longitudinal direction; f) shifting the temperature field to achieve solidification of the crystal.

[0090] In this way, easily co-doped indium phosphide rods can be produced, which can later be singulated into wafers.

[0091] Preferably, the first dopant compound is In2Ss, and the second dopant compound is GaP, wherein In2Ss is added in powder form, and GaP is added in the form of crystalline GaP.

[0092] Adding elemental Ga would not work, as it would flow to the nucleus in liquid form, preventing the formation of a single crystal. GaP has the advantage that P is not a contaminant and the stoichiometry does not change.

[0093] Experimental

[0094] Production of crystals and wafers

[0095] Processes P1 to P6

[0096] The crystals were grown using a VGF (vertical gradient freeze) process. Instead of moving the seed, the crystal grew upwards from the top of the seed in a crucible (see, for example, M. Jurisch et al. in "Handbook of Crystal Growth Bulk Crystal Growth: Basic Techniques," Volume II, Part A, Second Edition, Chapter 9, "Vertical Bridgman Growth of Binary Compound Semiconductors," 2015). The crucible was not moved; instead, the temperature field above the seed exhibited a positive temperature gradient and was shifted upwards by appropriate heater control.

[0097] A crystal seed with a length of 37 mm and a diameter between 7 and 9 mm was drilled axially from an oriented InP single crystal in the

[0100] direction. The seed crystal had a relatively small diameter compared to the target diameter of the single crystal. Therefore, the crucible had a conical region between the seed channel and the cylindrical region to accommodate the crystals to be produced, in which the crucible diameter was continuously increased. Crystal recycle or ingots of polycrystalline InP (GDMS analysis: total metallic impurities < 1E) were used as the base material for the InP. 16 cm -3) or combinations thereof. For 3" diameter crystal rods, 3.5 kg to 5 kg of InP were used, and for 4" diameter crystal rods, 4 kg to 6 kg of InP were used. For sulfur doping, In2Ss powder was added. GaP was added as a co-dopant. Furthermore, between 0.5 and 1.3 kg of boron oxide (B2O3) was added.

[0098] All materials were placed in a crucible. The furnace was filled with an argon atmosphere, and the pressure was raised to 22 bar at room temperature, reaching 42 bar at maximum temperature.

[0099] The InP melting took place at a power of 12 to 14 kW for 20 to 30 hours. The nucleus was also slightly melted (a few mm), for which purpose the power was increased. A vertical temperature gradient was created by appropriately controlling the furnace heating elements. By shifting the temperature field parallel to the crucible's longitudinal axis, the melt solidified completely, starting at the nucleus crystal.

[0100] Growth in the cone occurred at a rate of 2-3 mm / h, while growth in the cylinder occurred at a rate of 1-3 mm / h. Cooling took place over 48 hours, from 1000°C to 70°C.

[0101] The crucibles have a 5-10 cm long germination channel with a diameter of 5 to 10 mm. The germination channel is connected to the cylindrical portion of the crucible by a conical portion. The cylindrical portion has an inner diameter 0.1 to 2 cm larger than the nominal diameter of the wafers, i.e., 76.3 to 78.3 mm or 100.1 to 102 mm.

[0102] After the raw material has melted in the crucible, the nucleus is melted at temperatures for process P1 of 1035°C for the lower heater, 1084°C for the middle heater, and 1063°C for the upper heater. In process P2, the temperature of the middle heater is slightly higher in all phases than in process P1. In process P3, the temperature of the middle heater is only slightly higher during cone growth and cylinder growth than in process P1. In processes P4 and P5, similar conditions prevail to process P1. In process P6, the power of the upper heater is slightly higher during cone growth than in process P1, and during cylinder growth the power of the upper heater and the temperature of the middle heater are slightly higher than in process P1.

[0103] Process P7

[0104] The crystals were prepared using a VGF (vertical gradient freeze) process. A crystal seed with a length of 37 mm and a diameter between 7 and 9 mm was drilled axially from an oriented InP single crystal in the

[0100] direction and then etched.

[0105] The base material for the InP was crystal recycle or ingot of polycrystalline InP (GDMS analysis: total metallic impurities < 1E 16 cm -3 ) or combinations thereof. For 4" diameter crystal rods, 3 kg to 3.5 kg of InP were used. For sulfur doping, In2Ss powder was added. GaP was added as a co-dopant. Furthermore, between 0.5 and 1.3 kg of boron oxide (B2O3) were added.

[0106] All materials were placed in a crucible. The furnace was then evacuated, and the furnace was baked under vacuum to 100°C. The atmosphere in the furnace was argon. At room temperature, the pressure was increased to 22 to 27 bar, reaching a maximum pressure of 37 bar.

[0107] The InP melting took place at a power of 14 to 16 kW for 35 to 50 hours. The nucleus was also slightly melted (a few mm), for which purpose the power was increased. By appropriately controlling the furnace heating elements, a vertical temperature gradient was created. By shifting the temperature field parallel to the crucible's longitudinal axis, the melt solidified completely, starting at the nucleus crystal.

[0108] Growth in the cone region occurred at a rate of 2-3 mm / h, while growth in the cylindrical region occurred at a rate of 1-3 mm / h. Cooling took place over 72 hours, from 1000°C to room temperature.

[0109] Detection of twins or freedom from twins at the crystal

[0110] The resulting crystal was removed from the crucible and cleaned. The exterior was then inspected with the naked eye. Twinning is easily recognizable by the different reflectivity of the twinned single-crystal regions. The grain boundary between the twins is visible to the naked eye.

[0111] Determination of facet properties

[0112] The resulting crystal was removed from the crucible and cleaned. The exterior was then inspected with the naked eye. Facets were already visible here. Before the longitudinal sections were taken, the entire crystal was etched in aqua regia and inspected, followed by longitudinal sections as described below. To determine the exact shape of the facets, a longitudinal section was made through the conical region. The crystal was then etched using a structural etch. The so-called AB etch was used, as described in the following literature: M.S. Abrahams, C.J. Buiocchi, J. Appl. Phys., 1965, 36(9), 2855-63; G. Brown, B. Cockayne, W. MacEwan, J. Mater. Sci. 15 (1980) 2539. The crystal was then inspected again using a light microscope using phase contrast.

[0113] Wafer production

[0114] The resulting single crystal was removed from the crucible. By separating the conical and end regions, a cylinder with an irregular outer surface was created. This cylinder was ground on the outer edges to ensure a constant cylinder diameter along its entire length. Wire-cutting was used to separate the crystals into wafers (see, for example, H.-J. Möller in "Handbook of Crystal Growth Bulk Crystal Growth: Basic Techniques," Volume II, Part A, Second Edition, Chapter 18, "Wafer Processing," 2015).

[0115] Measurement of the lattice constant

[0116] InP samples were mechanically broken off from grown VGF crystals and crushed using a mortar and pestle. X-ray diffraction measurements were performed only on small particles (~10 pm). The powder X-ray diffraction measurements were performed using monochromatic Mo-Ka radiation (A=0.71073 Å) in transmission geometry. A Huber G670-360 X-ray camera served as the detector. Both Ai and A2 were used to maximize the number of measured reflections. A LaBe-SRM 660 powder sample from the National Institute of Standards and Technology was used to calibrate the instrument. The use of the calibration standard made it possible to optimize the experimental geometry and match the sample to the source distance. All samples were measured sequentially using the same sample holder at a fixed temperature of 22 °C to minimize experimental errors.

[0117] Although the measurement of lattice parameters is relatively simple, several factors complicate such experiments, in particular a very strong background at small diffraction angles (20) and the overlap of peaks at high diffraction angles (20) when both Ai and A2 are used. Furthermore, a preferential alignment of small crystallite particles distorts the amplitudes of the reflections, especially the (111) reflections, the strongest reflection occurring at the lowest scattering angle. InP crystals cleave in the {111} plane, which allows for the preferential alignment. To make matters worse, the polar character of the

[0111] direction in the zincblende structure obscures the amplitude of the {111} reflections. For these reasons, the analysis was limited to the area observed at 19 < (20) / 0 Limited to < 100 measured reflexes.

[0118] The measured diffraction patterns were fitted with pseudo-Voigt peak functions, and the background was fitted with the 5th-order Chebyshev polynomial. The extracted positions, widths, and amplitudes of the reflections were then used in the Le Bail analysis to refine the unit cell of InP using the WinXPOW software package from STOE & Cie. GmbH.

[0119] The procedure is described in more detail in the following article: D. Coster, KS Knol & JA Prins, ZS. F. Physik. 63, 345 (1930)

[0120] Determination of concentrations of foreign atoms in the crystal

[0121] Method A1: Glow Discharge Mass Spectrometry (GDMS)

[0122] For the measurement with GDMS, a solid sample with the dimensions 2x2x20 mm 3 used. The pretreatment of the sample included cutting the sample into a pin shape with the shape of 2 x 2 x 20 mm 3using a scalpel and subsequent chemical etching in a bromine-methanol solution, the concentration of bromine being not more than 10 vol%.

[0123] If necessary, the sample was stored under pure ethanol. The measurement was performed using a VG9000 from FI Elemental-FISON Instruments, UK. A matrix element was used as an internal standard, against which minor trace and ultratrace elements were measured accordingly. Quantification was achieved using relative sensitivity factors (RSF factors). Relative sensitivity factors in GDMS have been published, among others, in the following journal: Vieth, Wojciech. and John C. Huneke. “Analysis of high-purity gallium by high-resolution glow discharge mass spectrometry.” Analytical Chemistry 64 (1992): 2958-2964.

[0124] Method A2: Glow Discharge Mass Spectroscopy (GDMS) - modified method

[0125] Glow discharge mass spectroscopy (GDMS) was performed in combination with a glow discharge ion source and a high-resolution mass spectrometer. The HR-GDMS from NU Astrum was used with 6N + Argon was used as the carrier gas at a voltage of 1 kV at 3 mA with a resolution of approximately 3500. An InP sample was clamped into a pre-etched and pre-analyzed high-purity tantalum fork. The InP sample then imaged the cathode of the discharge. Plasma was maintained for 10 minutes before data acquisition to clean the sample surface.

[0126] Method A3: Inductively coupled plasma optical emission spectrometry (ICP-OES)

[0127] The measurement of the impurity atom concentrations by emission spectroscopy was carried out after digestion of the InP samples, using an iCAP 6500 DUO from Thermo Scientific.

[0128] The measurement conditions were as follows: https: / / assets.thermofisher.com / TFS-Assets / CMD / Application- Notes / AN40984%20Silver%20SSEA,%20Solid%20Sampling.pdf downloaded 2023- 09-08

[0129] In contrast to GDMS, in ICP-OES the plasma is generated inductively, and the impurity concentrations are determined not by mass spectroscopy but by optical detection of the emission lines characteristic of each chemical element from the excited electrons in the plasma. GDMS can be performed on the solid sample, while the sample is dissolved for ICP-OES.

[0130] Method A4: Secondary ion mass spectroscopy (SIMS)

[0131] The following device was used here: Cameca IMS 6f.

[0132] For the measured values ​​presented here, the values ​​from the measurement data at a depth of 10 pm were taken to exclude surface effects.

[0133] The measurements were carried out according to the procedure described in: Richard L. Hervig, Frank K. Mazdab, Peter Williams, Yunbin Guan, Gary R. Huss, Laurie A. Leshin “Useful ion yields for Cameca IMS 3f and 6f SIMS: Limits on quantitative analysis” Chemical Geology, Volume 227, Issues 1-2, 2006, Pages 83-99.

[0134] Procedure A5:

[0135] The charge carrier concentration is measured using the Hall effect. Under the reasonable assumption that all sulfur atoms (dopant D1) are electrically active, the concentration of sulfur atoms is equal to the charge carrier concentration. The determination of the charge carrier concentration is described in the "SEMI M39" standard.

[0136] Examples

[0137] Measurement series 1 : Dependence of twinning on material parameters

[0138] Table 1 shows measurement results from examples 1 to 37, which include various crystals with S doping, but also co-doping with Ga, As, B, and Sb. In general, it is noticeable that in the examples in which only sulfur was present as a dopant, but no co-dopant, twinning occurred in 16 cases (examples 1, 2, 5, 7, 10, 11, 13, 14, 15, 17, 19, 20, 21, 22, 27, and 30). Only in four cases in which no co-dopant was added did no twinning occur (examples 4, 6, 9, and 23). The probability of twinning with sulfur doping is therefore very high – however, with a co-dopant such as gallium, arsenic, boron, or antimony, this probability can be significantly reduced.

[0139] In cases where gallium was used as a co-dopant in addition to sulfur, no twinning occurred - this applies to examples 3, 8, 12, 16, 18, 24, 26, 28 and 29.

[0140] When other dopants were used, such as arsenic, it can be seen from Table 1 that no twinning occurred in Example 31, nor in Example 37 (here only defect formation occurred).

[0141] In the case of antimony (Example 32), twinning could not be avoided. In the case of boron (Examples 33 to 36, Example 25), twinning occurred in only three cases and not in two.

[0142] This shows that the addition of co-dopants significantly reduces twinning.

[0143] Example 16 shows that even when using very high S concentrations, twinning can be avoided if a correspondingly high amount of gallium is also used.

[0144] Furthermore, it was shown (see Examples 16 and 26) that the amount of gallium present in the system is not always completely distributed throughout the rod. In general, the gallium concentration is always higher at the beginning of the rod than at the end. However, no significant differences in hardness were observed at the beginning and end of the rod. For Ga contents greater than approximately 3E18 (measured via GDMS), HV0.1 is > 400, regardless of whether the measurement was taken at the beginning or end of the rod.

[0145] It is assumed that the stacking fault energy of the material is influenced by the addition of the co-dopant gallium, thereby reducing the tendency to twinning. The energy of the twin boundary corresponds to half the stacking fault energy.

[0146] Measurement series 2: Dependence of the lattice constant on the dopant concentration

[0147] Table 2 shows the dependence of the lattice constant on the concentration of Ga and S as dopants.

[0148] Table 2:

[0149] Sample 1 is an undoped InP sample, samples 2 and 3 are S-doped samples, and samples 4 to 6 are S and Ga-doped samples. For samples 2 to 9, the S content varied between 1.9x10 18 cm -3 and 6.5x10 18 cm -3 , in samples 4 to 9 the Ga content varied between 1 .94x10 18 cm -3 and 2x10 19 cm -3The S concentration was derived from the Hall effect measurements in semiconducting InP, the Ga concentration was determined by ICP-OES measurements or estimated from the sample weight during crystal growth.

[0150] The lattice parameter of undoped InP is in good agreement with the values ​​reported in the literature (for example, 5.87 Ä from C. Radautsan, Czech. J. Phys., 12, 382, ​​1962 and V. Geist, C. Äscheren, Cryst. Res. Technol. 19, 1231, 1984).

[0151] When doped with S at a concentration of 2x10 18 cm -3 or 8.5x10 18 cm -3 the lattice parameter becomes slightly smaller (Problems 2 and 3). Codoping S-doped InP with Ga significantly lowers the lattice parameter. Due to its electronic structure, Ga is assumed to primarily occupy In sites, and since it is slightly smaller in ionic radius than In, it can relax the lattice.

[0152] It was thus shown that with a co-doping of gallium, especially from a gallium concentration of 1x10 17 cm -3 , a significantly reduced lattice constant could be achieved compared to InP doped only with sulfur.

[0153] If this gallium concentration of 1x10 17 cm -3 below this value, the described effect of lowering the lattice constant does not occur. An even more pronounced effect is observed at gallium concentrations above 2.9x10 18 cm -3 to see (samples 4, 5 and 6).

[0154] Other possible embodiments

[0155] The production of InP rods is possible using the LEC (Liquid Encapsulated Czochralski) process, in which a single-crystal seed crystal is immersed in a melt of the same material. As the seed crystal is pulled upward, the single crystal grows around it, since the temperature above the melt is below the solidification temperature of the material.

[0156] A VB (Vertical Bridgman) method is also possible. Instead of shifting the temperature field, the crucible is moved downwards out of the temperature field. Vertical Gradient Freeze and Vertical Bridgman methods can be combined.

[0157]

Claims

1 . Indium phosphide crystal which contains: at least one first dopant (D1 ) which imparts semi-insulating or semiconducting properties to the crystal, at least one second dopant (D2) which is isoelectronic to indium or phosphorus, wherein the at least one first dopant (D1 ) and the at least one second dopant (D2) are different from one another.

2. Indium phosphide crystal according to claim 1, wherein the indium phosphide crystal is semiconducting, wherein a first dopant (D1) is sulfur.

3. Indium phosphide crystal according to claim 1 or 2, wherein a second dopant (D2) is selected from gallium, boron, antimony and arsenic or a combination thereof, wherein the second dopant (D2) preferably comprises gallium.

4. Indium phosphide crystal according to one of claims 1 to 3, wherein the concentration of the first dopant (D1) is 1*10 18 up to 2*10 19 Atoms / cm3 and / or the concentration of the second dopant (D2) is at least 1 * 10 17 Atoms / cm 3 amounts.

5. Indium phosphide crystal according to one of claims 1 to 4, which has a diameter of 45 mm to 205 mm.

6. Indium phosphide crystal according to one of claims 4 or 5, wherein the concentration of the first dopant (D1) is 1.3*10 18 up to 8.5*10 18 Atoms / cm 3 and / or the concentration of the second dopant (D2) is at least 2.9*10 18 Atoms / cm 3 amounts.

7. Indium phosphide crystal according to one of claims 1 to 6, wherein the lattice constant of the doped indium phosphide is 5.8680 to 5.8700 Angstroms (586.8 to 587.00 pm), preferably 5.8685 to 5.8690 Angstroms (586.85 to 586.90 pm).

8. Indium phosphide wafer consisting of indium phosphide crystal according to one of claims 1 to 7, wherein the wafer has a diameter of at least 50.8 mm, preferably 76.2 mm, more preferably at least 100 mm, even more preferably at least 150 mm, and even more preferably 200 mm, and / or a thickness between 450 pm and 675 pm.

9. A method for producing an indium phosphide rod, comprising the following steps: a) providing a crucible with a crystal nucleus; b) filling this crucible with polycrystalline indium phosphide c) adding a first dopant or a first dopant compound (DV1) and adding a second dopant or dopant compound (DV2) different from the first, d) optionally adding boron oxide (B2O3); e) creating an inert atmosphere in the crucible; f) melting the contents of this crucible by means of at least one heater, whereby the crystal nucleus is at least partially melted; g) forming a temperature gradient in the longitudinal direction; h) shifting the temperature field in order to achieve solidification of the crystal.

10. The method according to claim 9, wherein the first dopant compound (DV1) is In2Ss, and the second dopant compound (DV2) is GaP, wherein In2Ss is preferably added in powder form, and GaP is added in the form of crystalline GaP.

Images