Method for the preparation of a gallium arsenide substrate, gallium arsenide substrate and its use

Abstract

Method for producing a surface-treated gallium arsenide substrate, comprising the following steps: a) Providing a gallium arsenide substrate; b) Oxidation treatment of at least one surface of the gallium arsenide substrate in the dry state using UV radiation and / or ozone gas; c) Contacting at least one surface of the gallium arsenide substrate with at least one liquid medium; comprising the following sub-steps: ci) Contacting at least one surface of the gallium arsenide substrate with alkaline aqueous solution, c-ii) subsequent contact of at least one surface of the gallium arsenide substrate with water, c-iii) subsequent contact of at least one surface of the gallium arsenide substrate with acidic aqueous solution, c-iv) subsequent further contact of at least one surface of the gallium arsenide substrate with water. d) Marangoni drying of the gallium arsenide substrate.

Description

[0001] The present invention relates to a method for producing a surface-treated gallium arsenide substrate, as well as a gallium arsenide substrate as such, and the use thereof. Background of the invention

[0002] Gallium arsenide substrate wafers are used to fabricate high-frequency amplifiers and switches, as well as light-emitting devices such as semiconductor lasers and diodes. The device structures (transistors and diodes) are typically produced from epitaxially deposited stacks of mixed crystals containing various elements, which can be selected from the Ga-In-As-P-Al-N group. Different electronic properties can be achieved in individual layers through specific compositions, while maintaining the same crystal lattice parameters. This type of lattice-adapted epitaxy aims for very high layer quality and low defect densities. The layer quality depends not only on the conditions of the epitaxial process but is also influenced by the substrate properties.

[0003] The production of GaAs substrates begins with the growth of suitable bulk crystals and their subsequent singulation in a sawing / cutting process (see, for example, T. Flade et al., “State of the art 6” SI GaAs wafers made of conventionally grown LEC-crystals”, Journal of Crystal Growth, 198-199(1), 1999, pp. 336-342 and Th. Bünger et al., “Development of a vertical gradient freeze process for low EPD GaAs substrates”, Materials Science and Engineering B, 80(1), 2001, pp. 5-9). The substrates are then treated in a multi-stage process, including polishing, to achieve advantageous geometric properties (thickness, bending, wedge-shaped) and roughness (see, for example, Flade et al.). After the final polishing step, the pure, highly reactive GaAs surface is exposed, and oxide growth inevitably begins immediately. The typical subsequent cleaning in liquid media requires a sequence of oxide formation or...Oxide removal steps primarily serve to reduce the particle count and residual impurities or contamination on the substrate. During the final cleaning steps and subsequent drying of the substrate, an oxide layer forms. This layer can still change in the time leading up to the insertion of the substrates into the epitaxy apparatus (see, e.g., DA Allwood, S. Cox, NJ Mason, R. Palmer, R. Young, PJ Walker, Thin Solid Films, 412, 2002, pp. 76-83).

[0004] The composition of the oxide layer can be measured, for example, by X-ray-excited photoelectron spectroscopy (XPS), in which core electrons of the oxidized atom are spectroscopically analyzed. The oxidation states can be determined from the energy shift of the excited electrons. The measurement method is described in detail using GaAs as an example in CC Surdu-Bob, SO Saied, JL Sullivan, Applied Surface Science, Volume 183(1-2), 2001, pp. 126-136. The ratio of arsenic and gallium oxides ranges between 1 and 5, depending on the oxidation conditions. Typically, GaO, Ga2O3 and As2O, AsO, As2O3, As2O5 occur (see, e.g., F. Schröder-Oeynhausen, “Surface analytical characterization of metallic impurities and oxides on GaAs”, dissertation, University of Münster, 1996 and JS Song, YC Choi, SH Seo, DC Oh, MW Cho, T. Yao, MH Oh, Journal of Crystal Growth, 264, 2004, pp. 98-103).

[0005] Document JP H10-12 577 A is also known from the prior art. This describes a wet-chemical drying process with conventional drying.

[0006] Furthermore, document US 2008 / 0292877A1 is known. This document discloses various purification methods for GaAs substrates without using oxidation treatment with UV / ozone gas.

[0007] Furthermore, document JP H09 - 320 967 A is known. This discloses a UV / ozone treatment of GaAs with subsequent epitaxial growth, but without actual wafer cleaning.

[0008] Furthermore, document JP H11 - 204 471 A is known. This discloses a mirror-polished compound semiconductor wafer, which was obtained in a polishing and cleaning step with an organic solvent, and is oxidized in an oxidation step and subsequently cleaned in a cleaning step including an alkaline cleaning step.

[0009] Furthermore, document CN 1 787 178 A is known. This discloses a method for cleaning gallium arsenide wafers, comprising the following basic steps: (1) initial megasonic cleaning using a first solution; (2) initial rinsing; (3) initial drying; (4) cleaning with a UV-ozone cleaning machine; (5) initial megasonic cleaning using a second solution; (6) second rinsing; (7) second drying. The first solution is an alkaline aqueous solution that has a mildly corrosive effect on the gallium arsenide; the second solution is an acidic aqueous solution that has a corrosive effect on the oxide layer of the gallium arsenide.

[0010] Furthermore, the following scientific article is known from the prior art: WOLKE, Klaus [et al.]: Marangoni wafer drying avoids disadvantages. In: Solid State Technology, Vol. 39, 1996, No. 8, pp. 87-88 + p. 90. - ISSN 0038-111X. This discloses a Marangoni drying process for silicon wafers.

[0011] To influence the growth of epitaxial layers through the surface properties of the wafer, the possibility of heating surface layers (e.g., oxides) immediately before the epitaxial process in the epitaxial apparatus has frequently been utilized. The roughness of a heated surface, as well as the degree of contamination and impurity (particles), play a role in the qualitative properties of the deposited layer stacks and the resulting devices. For III / V semiconductors, the dependence of the quality of epitaxially deposited layers on parameters of the processes used for prior cleaning and the associated surface properties has been investigated. It was assumed that the wet chemical treatment is uniform across the entire wafer surface in each step.

[0012] The wet chemical cleaning of wafers is typically carried out in successive liquid baths. Process trays containing wafers are usually transferred from bath to bath using automated transport systems and are subsequently dried. The cleaning process generally comprises a sequence of acidic and alkaline wet steps with intermediate rinsing steps in deionized water (DI water). The alkaline components most commonly used are ammonium hydroxide (NH₄OH) or organic amines. Acids used include, for example, hydrogen fluoride (HF) and hydrogen chloride (HCl), as well as sulfuric acid (H₂SO₄) or organic acids. The cleaning media often contain additives such as oxidizing agents, surfactants, or complexing agents. For particle removal, ultrasound or megasound is used in individual baths.In principle, various methods are conceivable for drying wafers rinsed with DI water; in practice, conventional drying is based on the removal of the DI water by utilizing centrifugal forces during rapid rotation of the wafers or wafer carriers (spin drying) (see e.g. Song et al.).

[0013] However, conventional methods do not provide gallium arsenide substrates that meet the increasing demands of subsequent epitaxial processes in order to enable the large-area and reliable epitaxial production of components with the required layer quality and defect densities in a corresponding yield.

[0014] The object of the present invention is to provide an improved method for the production of gallium arsenide substrates which have advantageous properties for subsequent epitaxy. Summary of the invention

[0015] This problem is solved by the methods according to claims 1 and 7 and the gallium arsenide substrates according to claims 8-10, 12 and 15. Further developments are specified in the corresponding dependent claims. Useful uses are defined in claims 15 and 16.

[0016] To characterize the surface properties of differently treated gallium arsenide substrates, ellipsometric lateral substrate mappings from an optical surface analyzer were used. Details of the measurement procedure and the evaluation methods employed can be found in the description of the preferred embodiments and in the description of the examples.

[0017] Without limiting the invention, the following is a compilation of points that describe the subject matter, further developments and special features of the present invention: 1) Method for producing a surface-treated gallium arsenide substrate, comprising the steps of: a) Providing a gallium arsenide substrate; b) Oxidation treatment of at least one surface of the gallium arsenide substrate in the dry state using UV radiation and / or ozone gas; c) Contact of at least one surface of the gallium arsenide substrate with at least one liquid medium; and d) Marangoni drying of the gallium arsenide substrate. 2) Procedure according to point 1, wherein step c) comprises the following steps: i) Contacting at least one surface of the gallium arsenide substrate with alkaline aqueous solution, optionally using megasound; and ii) subsequent contact of at least one surface of the gallium arsenide substrate with water. 3) Method according to point 1 or 2, wherein in step c) an alkaline aqueous solution is used which is a solution of NH3 or organic amine in water, preferably of NH3, more preferably of NH3 in a concentration of 0.1-2 vol% and particularly preferably of NH3 in a concentration of 0.2-1 vol%. 4) Procedure according to point 2 or 3, wherein step c) further comprises the following steps: iii) following step ii), contacting at least one surface of the gallium arsenide substrate with acidic aqueous solution, optionally in the presence of an oxidizing agent; and iv) subsequent further contact of the at least one surface of the gallium arsenide substrate with water, wherein the water preferably contains at least an initially pH-modifying additive. 5) Method according to point 4, wherein the acidic aqueous solution is a solution of HCl or HF in water, preferably in a concentration of 0.1-0.5 vol.%, more preferably in a concentration of 0.1-0.25 vol.% and particularly preferably HCl in a concentration of 0.15-0.25 vol.%. 6) Method according to point 4 or 5, wherein the oxidizing agent in the acidic aqueous solution is ozone or H2O2, preferably ozone, more preferably ozone in a concentration of 10-50 ppm and particularly preferably ozone in a concentration of 30-50 ppm. 7) Method according to one of points 4-6, wherein the pH-modifying additive is basic or acidic, preferably basic, further preferably NH3, even more preferably NH3 in a concentration of 0.01-0.2 vol% and particularly preferably NH3 in a concentration of 0.05-0.1 vol%. 8) Procedure according to one of points 4-7, wherein in step c) following step iv) further steps are carried out according to steps i) and ii). 9) Method according to one of points 2-8, wherein water is deionized water or ultrapure water. 10) Method according to one of the preceding points, wherein the gallium arsenide substrate provided in step a) has previously been separated or isolated from a gallium arsenide bulk crystal and / or polished and preferably pre-cleaned, further preferably pre-cleaned by wet chemicals and particularly preferably pre-cleaned by wet chemicals and brush cleaning. 11) Method according to any of the preceding points, wherein the gallium arsenide substrate provided in step a) is doped or undoped. 12) Method according to one of the preceding points, wherein an aqueous isopropanol solution is used in step d). 13) Method for producing a plurality of surface-treated gallium arsenide substrates, wherein simultaneously a plurality of gallium arsenide substrates are subjected to the respective steps b)-d) in the method according to one of the preceding points. 14) Gallium arsenide substrate which has at least one surface which, when ellipsometrically lateral substrate mapping with an optical surface analyzer, exhibits a variation of the laterally resolved background-corrected measurement signal whose 1% distribution percentile normalized to the substrate mean of the phase shift signal is greater than -0.0065. 15) Gallium arsenide substrate according to point 14, which has at least one surface which, when ellipsometrically lateral substrate mapping with an optical surface analyzer, exhibits a variation of the laterally resolved background-corrected measurement signal, the distribution percentile of 1% normalized to the substrate mean of the phase shift signal being greater than -0.0060, preferably greater than -0.0055, more preferably greater than -0.0050, even more preferably greater than -0.0045, particularly preferably greater than -0.0040, particularly greater than -0.0030, particularly greater than -0.0020, even greater than -0.0010 and up to 0.0000 exclusively 0.0000. 16) Gallium arsenide substrate having at least one surface which, when ellipsometrically mapped laterally with an optical surface analyzer, is measured with a substrate diameter of 150 mm as a reference, a defect count of < 6000 and / or a total defect area of ​​less than 2 cm² 2 exhibits where a defect is defined as a contiguous area larger than 1000 µm 2 defined by a deviation from the mean measurement signal in ellipsometric lateral substrate mapping with an optical surface analyzer of at least ± 0.05%. 17) Gallium arsenide substrate according to point 16, where the number of defects is < 5000, preferably < 4000, further preferably < 3000, even more preferably < 2000, even more preferably < 1000, even more preferably < 500, even more preferably < 300, even more preferably < 250, even more preferably < 200, even more preferably < 150 and particularly preferably < 100. and / or the total defect area is less than 1 cm² 2 , preferably smaller than 0.5 cm 2 , preferably smaller than 0.1 cm 2 , even more preferably smaller than 0.05 cm 2 , even more preferably smaller than 0.01 cm2 , even more preferably smaller than 0.005 cm 2 and especially preferred to be smaller than 0.0035 cm 2 is. 18) Gallium arsenide substrate produced according to the method in accordance with one of points 1-12 (reference aspect). 19) Gallium arsenide substrate according to one of points 14-18, wherein the diameter is at least 100 mm, preferably at least 150 mm and further preferably at least 200 mm. 20) Polished and surface-treated gallium arsenide substrate with a diameter of at least 150 mm, wherein the surface treatment comprises oxidation treatment of at least one surface of the gallium arsenide substrate in the dry state by means of UV radiation and / or ozone gas, contact of the at least one surface of the gallium arsenide substrate with at least one liquid medium and Marangoni drying of the gallium arsenide substrate. 21) Gallium arsenide substrate according to point 20, which has a thickness of not greater than approximately 600 µm or not less than approximately 800 µm. 22) Gallium arsenide substrate according to point 20 or 21, wherein the thickness is in the range of approximately 100 to approximately 600 µm or the thickness is greater than approximately 800 µm and the thickness preferably is in the range of approximately 250 to approximately 500 µm or of approximately 800 to approximately 2000 µm. 23) Polished and surface-treated gallium arsenide substrate, which has a thickness of not greater than approximately 600 µm. 24) Polished and surface-treated gallium arsenide substrate, which has a thickness of not less than approximately 800 µm. 25) Polished and surface-finished gallium arsenide substrate according to one of points 20 to 24, wherein the treated surface of the substrate has the properties defined in one of points 14 to 17. 26) Gallium arsenide substrate according to one of points 14-25, wherein said gallium arsenide substrate is doped or undoped. 27) Gallium arsenide substrate, which has at least one surface which, within 9 months, preferably 12 months, after production, exhibits a substantially non-deteriorating, preferably non-deteriorating, variation of the laterally resolved background-adjusted measurement signal during ellipsometric lateral substrate mapping with an optical surface analyzer. (Reference aspect) 28) Gallium arsenide substrate according to one of points 14-26, wherein the at least one surface exhibits, within 6 months of manufacture, a substantially non-deteriorating, preferably non-deteriorating, variation of the laterally resolved background-adjusted measurement signal when ellipsometric lateral substrate mapping with an optical surface analyzer. 29) A plurality of gallium arsenide substrates which are produced according to the method according to one of points 1-13 and which exhibit a substantially similar, preferably identical, variation from substrate to substrate of the laterally resolved background-cleaned measurement signal when performing ellipsometric lateral substrate mapping of the at least one surface with an optical surface analyzer. 30) Gallium arsenide substrate according to one of points 14 to 17 and 25 to 28 or a plurality of gallium arsenide substrates according to point 29, wherein ellipsometric lateral substrate mapping is performed with an optical surface analyzer analogous to Candela CS20, preferably specifically with an optical surface analyzer Candela CS20, more preferably with an optical surface analyzer utilizing laser light of a wavelength of 405 nm and whose beam path comprises a half-wave plate, a quarter-wave plate, a polarization-sensitive beam splitter and two detectors, even more preferably with an optical surface analyzer according to the phase-shift channel of Candela CS20 and particularly with an optical surface analyzer according to Fig. 1. 31) Use of the gallium arsenide substrate according to one of points 14-28 and 30 respectively. the multitude of gallium arsenide substrates according to point 29 for epitaxial crystal growth, optionally after storage and preferably without pretreatment after providing the gallium arsenide substrate and before epitaxial crystal growth. 32) Use of the gallium arsenide substrate according to one of points 14-28 and 30 or the plurality of gallium arsenide substrates according to point 29 for the manufacture of semiconductor devices or electronic and optoelectronic devices. 33) Use of the gallium arsenide substrate according to one of points 14-28 and 30 or the plurality of gallium arsenide substrates according to point 29 for the manufacture of power devices, high-frequency devices, light-emitting diodes and lasers. 34) Use of an optical surface analyzer, in particular an optical surface analyzer analogous to the Candela CS20 or specifically the Candela CS20, for the optical non-contact quantitative characterization of the homogeneity of surface properties, in particular for the quantitative characterization of the homogeneity of the surface oxide layer, of gallium arsenide substrates by means of ellipsometric lateral substrate mapping. 35) Use according to point 34, wherein a laterally resolved measurement signal is cleared from a lower frequency background by means of discrete complex Fourier transform, preferably using the Levenberg-Marquardt algorithm. Brief description of the drawings Fig. Figure 1 schematically shows the optical setup and beam path of an optical surface analyzer according to Candela CS20 used for ellipsometric surface measurements, in particular for the so-called phase-shift channel of the measuring instrument Candela CS20. Fig. Figure 2 shows typical ellipsometric lateral substrate maps, so-called candela images, for different final purification technologies (from left to right: comparison example 1, example 1 and example 2). Fig. Figure 3 shows an example of the measured signal profile (“raw data”) of a circular track, where the line “background” describes the crystallographically induced part of the measurement signal, which can be modeled by Fourier transformation using the Levenberg-Marquardt algorithm. Fig. Figure 4 shows exemplary phase shifts of a track determined after subtraction of the crystallographically induced background signal. Fig. Figure 5 shows typical ellipsometric lateral substrate maps, so-called Candela images, after subtraction of the crystallographically induced background signal for various final purification technologies (from left to right: comparison example 1, example 1 and example 2). Fig. Figure 6 presents typical global frequency distributions of the ground-cleaned phase shifts, i.e., residues of the ground-cleaned mappings, for individual wafers, each of which was subjected to different final cleaning technologies. Fig. Figure 7 shows typical 1% percentiles of 25 wafers each, which were manufactured according to the methods from Comparative Examples 1 and 2 or the Inventive Examples 1 and 2. For an objective comparison of the percentiles, they are normalized to the respective wafer mean values ​​of the phase shift signal. Fig. Figure 8 shows typical so-called defect maps of background-cleaned candela images for different final cleaning technologies (from left to right: comparison example 1, example 1 and example 2). Fig. Figure 9 shows typical defect numbers and defect areas of 25 wafers each, which were manufactured according to the methods from comparative examples 1 and 2 or the inventive examples 1 and 2. Description of preferred embodiments

[0018] Without limiting the present invention, the following detailed description of the figures, objects, further developments and special features is intended to illustrate the invention and describe special embodiments in more detail.

[0019] A first object of the present invention provides a method for producing a surface-treated gallium arsenide substrate, comprising the following steps: Providing a gallium arsenide substrate, oxidation treatment of at least one surface of the gallium arsenide substrate in the dry state using UV radiation and / or ozone gas, contacting the at least one surface of the gallium arsenide substrate with at least one liquid medium and Marangoni drying of the gallium arsenide substrate.

[0020] In the process according to the invention, it was surprisingly found that the combination of the coordinated dry oxidation treatment, contact with a liquid medium, and Marangoni drying allows the surface properties of the gallium arsenide substrate to be advantageously and very homogeneously adjusted, particularly the surface oxide. This means that, in contrast to conventionally produced substrates, the surface oxide can be thermally heated in a controlled and reproducible manner across the entire substrate surface, e.g., immediately before epitaxy in the epitaxy apparatus. Thus, the substrate according to the invention can be used in the epitaxy process without further treatment. The gallium arsenide substrate can be a wafer.

[0021] In the present invention, it was advantageously recognized that a very homogeneous surface is required, since the heating behavior and heating temperatures of different oxides differ significantly, and the heating behavior of the oxide layer on a GaAs substrate depends on its composition of gallium and arsenic oxides of different oxidation states as well as the oxide layer thickness. A further advantage is that complete heating of the homogeneous oxide layer prevents surface roughening during desorption of this oxide layer and thus avoids morphological disturbances in the epitaxially grown layer structures. Furthermore, the present invention recognized that for consistently high quality of the epitaxial layers across the entire wafer surface, and thus for high yields, the lateral surface homogeneity, or...Oxide homogeneity, i.e., the lateral uniformity of oxide composition and thickness, is of great importance. In particular, it has been recognized that on parts of the GaAs surface with an unfavorable oxide composition or thickness that deviates significantly from the rest of the wafer surface, the oxides' heating behavior can be disrupted, potentially resulting in residual oxide islands or increased roughening of the wafer surface. The risk of crystallographic defects forming in the epitaxial layers can be very high in these disturbed areas of the wafer surface, which would compromise the electrical functionality of devices fabricated from the GaAs wafer.It was further advantageously recognized that the execution of the last steps within the manufacturing process of GaAs wafers, namely the final polishing and the subsequent cleaning steps, and in particular the last cleaning steps immediately before drying and the drying of the wafers, is of particular importance for the adjustment of the surface properties.

[0022] Considering the subsequent wet treatment and the following special drying step, it is particularly important that, for the manufacturing and cleaning process of the present invention, the surface of the GaAs substrate is homogeneously oxidized in a dry state before contact with the liquid medium. According to the invention, the dry oxidation treatment is carried out in one case by irradiating the surface with UV light, preferably short-wave UV light. Firstly, the oxidation is stimulated by the energy of the UV light itself. Secondly, the light energy partially converts oxygen from the surroundings into ozone, which intensifies the oxidation process on the wafer surface. In another case, the dry oxidation takes place by gassing with ozone gas. The introduced ozone gas can be generated by an ozone generator.Dry oxidation results in the formation of a homogeneous passivating oxide layer and the oxidative destruction of organic impurities on the substrate surface. A homogeneous surface oxide layer, which serves as a sacrificial oxide in subsequent steps, is formed in a controlled manner, both laterally and in depth. Particularly advantageous are the deep oxidation of undefined surface conditions created by preceding process steps and the avoidance of selective oxide formation mechanisms that would occur during natural or wet-chemical oxidation to form a sacrificial oxide. In dry oxidation treatment, the GaAs surface is more strongly oxidized towards bulk GaAs, as the oxide on the front side partially grows towards the back of the wafer.Areas originating from the upstream processes with surface properties that differ from the rest of the wafer surface, hereinafter referred to as inhomogeneity defects, can be more easily removed from the wafer surface in their oxidized state, for example, by wet chemical processes. A high-quality oxide layer produced in the first step of the cleaning process is a prerequisite for the subsequent application of the contacting liquid medium, preferably a low-abrasive and roughness-neutral etching step. Furthermore, the present invention has been particularly recognized as being ideally suited to homogeneously hydrophilizing the entire surface. This is especially advantageous with regard to uniform wetting of the wafer surface when transferring the wafers between cleaning baths.

[0023] It was further found that wet-chemically produced oxides are difficult to completely remove from the GaAs surface. The homogeneity of the resulting oxide layer, and thus also its (extent and type of) removability, are significantly influenced by the oxidation conditions.

[0024] Following dry oxidation treatment and contact of the surface with a liquid medium, for example for suitable cleaning steps, the surface wetted with the liquid medium is then advantageously dried by Marangoni drying according to the present invention. Marangoni drying is based on the Marangoni principle. A suitable agent is used, e.g., alcohol or other organic compounds, in particular isopropanol, which, after its accumulation on the surface of the liquid medium, e.g., the water surface, reduces the surface tension of the liquid. If this liquid front, preferably the water front or aqueous front, is moved relative to the surface of the wafer, e.g., which is positioned vertically in the bath, the gradient in surface tension between the thin meniscus layer in contact with the wafer and regions of the liquid surface located further away from the wafer, preferably the water surface, causes the surface tension to decrease.aqueous surface, allowing the liquid to flow off the wafer surface without leaving any residue.

[0025] The drying process according to the invention is particularly advantageous for cleaning GaAs wafers of very thin or very thick wafers. In Marangoni drying, the wafers are subjected to less mechanical stress than in spin drying, which is the primary drying method used. It has been found in the present invention that the risk and tendency of wafer breakage during the cleaning of thin GaAs wafers using the method of the present invention are significantly reduced. With thick GaAs wafers, imbalance can occur during spin drying at high speeds due to the flats or notches applied to mark the crystallographic orientation, which also increases the risk of wafer breakage. Therefore, the present cleaning method also offers advantages when cleaning particularly thick GaAs wafers.The present invention offers a further advantage, particularly for drying large-area GaAs wafers, regardless of wafer thickness: In centrifugal drying, residual water is transported across the wafer surface to the edge of the GaAs wafer, which can lead to localized traces in the surface composition, so-called watermarks. In Marangoni drying, drying always takes place directly at the three-phase boundary between the wafer, the liquid level (preferably a water level or aqueous level), and the surrounding gas atmosphere. This prevents watermarks at all substrate thicknesses and ensures homogeneous drying of the GaAs substrate wafers.

[0026] Furthermore, it has been surprisingly found that, in combination with the preceding dry oxidation treatment, the Marangoni drying process in the present method produces an extremely homogeneous oxide layer on the treated GaAs substrate surface with respect to both thickness and composition. This is very advantageous for conventionally available GaAs substrate thicknesses, but especially also for relatively thin or even extremely thin GaAs substrates, as well as for relatively thick or even extremely thick GaAs substrates with a large diameter, or for large diameters and all substrate thicknesses. Large diameters are defined as at least 100 mm, preferably at least 150 mm, and more preferably at least 200 mm. Substrate thicknesses can expediently be ≥ 100 µm; the maximum thickness can be determined by the desired application of a substrate, for example, up to 5000 µm, but usually more often up to 2000 µm or even 1000 µm.In this way, the cleaning sequence according to the invention enables the production of large-area GaAs wafers of varying thicknesses with homogeneous surfaces.

[0027] The provided gallium arsenide substrate can be doped or undoped, is crystalline, and particularly preferably monocrystalline, and can be produced by singulation or separation from a GaAs bulk single-crystal body (ingot, boule). In one embodiment, said provided gallium arsenide substrate was first polished, preferably polished and then pre-cleaned, more preferably polished and then pre-cleaned by wet chemical means, and particularly preferably polished and then pre-cleaned by wet chemical means and brush cleaning. Brush cleaning can be advantageous, for example, to remove particles without the addition of organic surfactants. In the process according to the invention, impurities caused by particles and surface defects such as scratches, pits or bumps, crystal defects, and high roughness are avoided or significantly reduced.Due to the high homogeneity achieved in the top layer (the oxide layer), the GaAs wafer produced is particularly suitable as a substrate for the epitaxial growth of layers and layer systems, and apart from the usual heating process, it can be used directly for epitaxy - possibly after intermediate storage.

[0028] Step c) comprises the following steps: at least one surface of the gallium arsenide substrate is contacted with an alkaline aqueous solution, optionally using megasonography, and subsequently with water. In particular, any remaining inhomogeneities in the oxide layer, which are located very close to the surface of the wafers, can be advantageously and reliably removed.

[0029] The wet cleaning process, which includes an alkaline cleaning step, optionally using megasound, and a water rinse, preferably with deionized water (DI water) or ultrapure water, is adapted to the subsequent process steps.

[0030] When transferring wafers between different baths in an automatic or manually operated wet processing plant, a hydrophilic wafer surface is completely wetted with a film of liquid.

[0031] In principle, the avoidance of dry areas on the wafer surface during the transfer processes between the wet baths of a wet bench can also be achieved by carrying out the processes, including drying, in a single basin if possible.

[0032] The oxide layer is removed in a controlled manner by alkaline treatment in a first liquid tank. To adjust the pH, ammonia or organic amine compounds can be added to the deionized water. The concentrations of the added chemicals can be greater than 0.1% by mass (wt%). The introduction of megasound into this tank promotes the removal of particles adhering to the surface. The megasound can be introduced either by attaching suitable transducers to the outside of a quartz glass tank or by using appropriately coated vibrating elements directly within the tank.

[0033] Alternatively, the removal of the oxide layer with simultaneous particle removal can also be carried out in an acidic cleaning medium.

[0034] When the oxide layer is removed in the alkaline medium, the high surface energy of the wafer surface is retained due to the increased presence of OH groups. The resulting hydrophilic wetting behavior of the surface not only stabilizes surface homogeneity during subsequent transfer steps but also leads to greater uniformity of the GaAs surface during Marangoni drying.

[0035] Following alkaline cleaning, the wafers are rinsed with water, preferably deionized water. The rinsing process time can range from a few seconds to several minutes, depending on the chemical concentration used.

[0036] The subsequent drying of the wafers according to the Marangoni principle is capable of achieving a very uniform drying of the wafer surface. The basic principle is the reduction of the surface tension of the water or aqueous solution by the appropriate introduction of a suitable agent such as isopropanol. The speed at which the wafers or the water level or aqueous level are moved can be precisely adjusted to the surface energy of the wafers to be dried. This allows for an optimum balance between quality and wafer throughput. Suitable drying speeds for GaAs substrates in the final stage of wafer manufacturing range from a few hundredths of a millimeter per second to several centimeters per second.

[0037] GaAs wafers dried using the Marangoni process exhibit fewer traces of surface inhomogeneities compared to spin-dried wafers. This applies to both hydrophobic and hydrophilic spin-dried wafers.

[0038] The roughness of the GaAs surface is not changed by the removal of the oxide in the alkaline cleaning step and remains at approximately 0.3 nm for R. a . Metallic surface contaminations determined by total reflection X-ray fluorescence analysis (TXRF) also remain at the same level.

[0039] The cleaning sequence results in an average particle contamination of at most 10 particles with a diameter greater than 0.3 µm per substrate (KLA-Tencor Surfscan 6420). Thus, with a cleaning sequence according to the preferred embodiment of the present invention, surface homogeneity is further improved. Roughness and metal contamination remain at the same level, and particle contamination is low.

[0040] GaAs wafers with a highly sensitive surface can be processed in batches using wet chemical cleaning with various types of wafer carriers. Since the present invention preferably involves the production of large-area GaAs wafers with high surface homogeneity, the use of wafer carriers that minimize surface obstruction is preferable. An even better option is to operate entirely without carriers. For the same reasons of achieving the most homogeneous wet treatment possible, the flow conditions in the liquid baths are optimally adjusted.

[0041] A preferred embodiment is shown by way of example in the following flowchart:

[0042] Furthermore, it is provided that step c) further comprises the following steps: contacting the at least one surface of the gallium arsenide substrate with acidic aqueous solution, optionally in the presence of an oxidizing agent, following step ii), and subsequent further contacting of the at least one surface of the gallium arsenide substrate with water, wherein the water preferably contains at least an initially pH-modifying additive.

[0043] The etching solution consists of an acid and optionally an oxidizing agent. This embodiment also results in a very uniform surface oxide after the wafer is finally dried using the Marangoni process.

[0044] Hydrochloric acid or hydrofluoric acid can be used. H₂O₂ or ozone (O₃) dissolved in the liquid medium serves as the oxidizing agent. In this further preferred embodiment, all steps are coordinated such that i) inhomogeneities in the oxide layer and additionally also in the uppermost GaAs atomic layers of the surface are removed, ii) no new inhomogeneity defects are created by continuous, complete wetting of the wafer surface during the transfer processes between the wet process steps, iii) all oxide traces from the acidic cleaning step are uniformly removed by the special design of the final rinsing step and the GaAs surface is hydrophilized for drying, and iv) an extremely uniform wafer surface is achieved by Marangoni drying.

[0045] Acid cleaning is used, among other things, to remove metallic contaminants. By selecting the concentrations of acid and oxidizing agent, the strength of the reaction can be controlled, taking into account the high reactivity of the GaAs surface, and thus the etching rate and surface roughening. Due to the high reactivity of the GaAs surface, the conditions are advantageously chosen so that essentially no, preferably no, locally varying oxide thicknesses, oxide compositions, or roughening—i.e., inhomogeneity defects—occur on the surface of the GaAs wafer.

[0046] In the present invention, suitable conditions and compositions for the oxidizing acidic cleaning step can be selected, whereby such a preferential removal of any inhomogeneity defects from the upstream processes on GaAs wafers is possible, wherein in this cleaning step the roughness after the last polishing step is at least preserved.

[0047] Typical oxidizing agents for cleaning semiconductor wafers in aqueous solutions are hydrogen peroxide and ozone. Hydrofluoric acid or hydrochloric acid are preferred as oxide-dissolving acids.

[0048] Extensive investigations according to this preferred embodiment of the present invention have revealed conditions under which the elimination of any inhomogeneity defects is possible, recontamination by particles remains low, and the roughness achieved with the final polishing step remains constant. For this purpose, the acid concentration is preferably less than 0.25% and greater than 0.1%, determined as volume percent in the total liquid. The rate of material removal can be essentially determined by the concentration of the oxidizing agent. Thus, a typical R a -Roughness improved from 0.30±0.03 nm to 0.20±0.03 nm after the last polishing step across the entire ozone concentration range.

[0049] The etching process, preferably carried out according to this embodiment, is based on simultaneous oxide formation and dissolution processes. When the wafers are removed from the acidic etching solution for transfer to the subsequent liquid bath, their surface is covered with an oxide layer. Due to the simultaneous hydrophobic effect of the acid, the wafers are not automatically completely wetted during transfer, despite the oxide layer. Rather, the wetting behavior after the acidic cleaning step is advantageously determined by the balance between acid and oxidizing agent. Good wetting behavior of the GaAs wafers is also necessary during this transfer process to avoid local disturbances of oxide homogeneity in the dry areas for the reasons mentioned above. The investigations carried out showed that the acid concentration is preferably kept below 0.25%, also for the sake of good wettability.

[0050] When using ozone concentrations above 20 ppm and low concentrations of HCl below 0.5%, and applying appropriate conversion times, GaAs wafers remain completely wetted during conversion in air. This results in the particularly preferred configuration for the acidic cleaning step in this embodiment of the invention. The use of HCl at a concentration between 0.15% and 0.25% together with ozone at a concentration of 10–100 ppm and a residence time of the wafers in this solution between 2 and 5 minutes is particularly advantageous.

[0051] Preferably, after acid cleaning in the presence of an oxidizing agent, a subsequent rinsing step is carried out with the addition of a pH-modifying agent ("spiking"). Such an additional step can be appropriately tailored to both the preceding acid cleaning and the subsequent Marangoni drying and helps to prevent the potential reformation of inhomogeneities. By adding an acidic or basic substance to the rinsing bath circulation circuit, the oxide layer on the GaAs wafer is immediately and uniformly removed upon immersion. Any inhomogeneity defects are avoided with both ammonia and HCl. After the pH-modified rinsing water has circulated around the wafers for a short time, fresh water is preferably added, and the actual rinsing of the wafers with fresh DI water begins.Both acidic and basic chemicals are suitable for removing the oxide layer in the rinsing step. The roughness of the GaAs surface during the gentle etching process is not increased within the preferred concentration range of the pH-modifying agent compared to the roughness before the cleaning sequence and remains at least at typically 0.3 ± 0.03 nm for R. a .

[0052] According to the invention, it can be achieved that the metallic surface contamination determined by TXRF remains at the same low level. The average particle contamination in this cleaning sequence is approximately 10 particles with a diameter greater than 0.3 µm per wafer (KLA-Tencor Surfscan 6420). Thus, with a cleaning sequence according to this preferred embodiment of the present invention, surface homogeneity is improved. Roughness and metal contamination remain at least at the same level, and particle contamination is low.

[0053] This particularly preferred embodiment is shown by way of example in the following flowchart:

[0054] In a further embodiment of the method according to the invention, in step c) following step iv), further steps according to steps i) and ii) are carried out. In this embodiment, after acid cleaning and subsequent rinsing, the wafers are subjected to a further alkaline cleaning, optionally and preferably under the influence of megasound, before drying. This allows for even more effective particle removal than when using megasound in the DI water rinse bath. Furthermore, the hydrophilic character of the GaAs surface is enhanced by the alkaline treatment, which can have a beneficial effect on the behavior of the wafers during drying.

[0055] This embodiment is shown by way of example in the following flowchart:

[0056] Another object of the present invention is a method for producing a plurality of surface-treated gallium arsenide substrates, wherein simultaneously a plurality of gallium arsenide substrates are subjected to the respective steps b)-d) in the method according to one of the preceding points.

[0057] Batch production of a large number of wafers enables, on the one hand, economical production, and on the other hand, a high homogeneity of the product properties, especially the surface properties, of the substrates among themselves.

[0058] A further object of the present invention is a gallium arsenide substrate which has at least one surface which, when ellipsometrically lateral substrate mapping with an optical surface analyzer, exhibits a variation of the laterally resolved background-corrected measurement signal, the distribution percentile of which is normalized to the substrate mean of the phase shift signal, i.e., for the distribution, the 1% percentile, is greater than -0.0065, preferably greater than -0.0060, more preferably greater than -0.0055, even more preferably greater than -0.0050, even more preferably greater than -0.0045, particularly preferably greater than -0.0040, particularly greater than -0.0030, particularly greater than -0.0020, preferably greater than -0.0010 and even up to 0.0000, wherein the absolute value 0.0000 can be excluded.

[0059] A phase shift signal is a measured or detected signal based on the phase-dependent properties of differently polarized light components, e.g., perpendicularly and parallel polarized light. Preferably, the phase shift signal is measured as the intensity difference between detector signals. More preferably, two detectors detect differently polarized light components from the substrate surface of laser light reflected from the substrate surface, with the differently polarized light components being additionally spatially separated after the interaction of the laser light and the substrate.

[0060] The substrate mean of the phase shift signal is the arithmetic mean of all phase shift signals measured on a substrate.

[0061] All deviations from the ideally homogeneous surface (residues of the background-corrected mapping of the phase-shift signals) can be represented in a histogram. Since this distribution typically does not correspond to a Gaussian normal distribution, the commonly used standard deviation cannot be used as a measure of homogeneity. Rather, it is appropriate to define and calculate suitable percentiles. To evaluate the homogeneity of the GaAs surfaces according to the present invention, the percentiles P1 and Q are used. 0,01 used for comparisons. A distribution percentile of 1% or 1% percentile for the distribution means the percentile rank P1 below which 1% of the total number of values ​​in the distribution lie, and corresponds to the quantile Q. 0,01(See, for example, F. Schoonjans, D. De Bacquer, P. Schmid, “Estimation of population percentiles”. Epidemiology, 22, 2011, pp. 750-751). To compensate for fluctuations in the incident light intensity, the determined percentiles are normalized to the respective wafer means of the phase shift signal, i.e., they are divided by the arithmetic mean of all phase shift signals measured on a substrate.

[0062] A further object of the present invention is a gallium arsenide substrate which has at least one surface which, when substrate-cleaned ellipsometric lateral substrate mapping with an optical surface analyzer, based on a substrate diameter of 150 mm as a reference, exhibits a defect count of < 6000, preferably < 5000, more preferably < 4000, even more preferably < 3000, even more preferably < 2000, even more preferably < 1000, even more preferably < 500, even more preferably < 300, even more preferably < 250, even more preferably < 200, even more preferably < 150 and particularly preferably < 100 and / or a total defect area of ​​less than 2 cm² 2 , preferably smaller than 1 cm 2 , preferably smaller than 0.5 cm 2 , even more preferably smaller than 0.1 cm 2 , even more preferably smaller than 0.05 cm 2 , even more preferably smaller than 0.01 cm 2, even more preferably smaller than 0.005 cm 2 and especially preferred to be smaller than 0.0035 cm 2 exhibits where a defect is defined as a contiguous area larger than 1000 µm 2 with a deviation from the mean measurement signal of at least ± 0.05% when using ellipsometric lateral substrate mapping with an optical surface analyzer. With these measured values, sufficient differentiation of the gallium arsenide substrates according to the invention from conventional gallium arsenide substrates is achieved, which can be analytically determined using the described ellipsometric lateral substrate mapping (see also the examples described below).

[0063] Ellipsometric lateral substrate mapping is preferably performed with an optical surface analyzer analogous to the Candela CS20, and more preferably specifically with the Candela CS20 optical surface analyzer. In particular, suitable optical surface analyzers are those in which the analysis laser light used has a wavelength of 405 nm and whose beam path includes a half-wave plate, a quarter-wave plate, a polarization-sensitive beam splitter, and two detectors. Specifically, for example, an optical surface analyzer can operate according to the phase-shift channel of the Candela CS20; a typically usable arrangement for a preferred optical surface analyzer is described in Fig. 1 shown.

[0064] Ellipsometry is based on the effects of polarized light as it propagates through optically active media. Selecting a suitable wavelength and using a reflected measurement setup allows for high surface sensitivity. Fig. Figure 1 schematically shows the optical setup and beam path of an optical surface analyzer according to Candela CS20 used for the ellipsometric surface measurements, i.e. for ellipsometric lateral substrate mappings, of the present invention, in particular for the so-called “phase-shift” channel of the Candela CS20 measuring device from KLA-Tencor.

[0065] For the characterization of GaAs substrate surfaces, polarized laser light with a wavelength of 405 nm is used. After passing through a half-wave plate, the laser is directed onto the substrate surface at an angle (θ) of 60° to the normal via a mirror and a focusing lens. The perpendicularly and parallel polarized components are reflected at the substrate surface according to the optical properties of the oxide layer and directed by a converging lens, a mirror, and a quarter-wave plate onto a polarization-sensitive beam splitter. Here, the differently polarized light components resulting from the interaction with the substrate surface are separated and analyzed in detectors D1 and D2.According to a preferred embodiment, the intensity difference between the detector signals is referred to as the phase shift signal, and the phase shift signal characterizes the optical properties of the reflecting substrate surface. By rotating the wafer and simultaneously moving the optical measuring system radially, the entire surface of the wafer can be scanned in a spiral pattern. By means of rapid ellipsometric mapping or scanning, complete high-resolution mappings or images of the optical properties of even large-area GaAs substrates (substrate mappings) can be generated for assessing surface homogeneity. According to the invention, an optical surface analyzer analogous to the Candela CS20 from KLA-Tencor, i.e., a measuring device corresponding to the Candela CS20, can be used; however, the Candela CS20 device is particularly preferred (see also L. Bechtler, V. Velidandla, Proc.SPIE 4944, Integrated Optical Devices: Fabrication and Testing, 109, 2003; doi:10.1117 / 12.468295 and F. Burkeen, Compound Semiconductor, 14 (10), 2008), in particular the phase-shift channel of Candela CS20. In principle, however, similar or different ellipsometric measuring devices and mapping ellipsometers can also be used, whereby appropriate adaptation can be made depending on the respective optical setup, beam path and measuring principle.

[0066] The Candela CS20 uses the interaction of a 405 nm laser beam for measurement, which strikes the wafer surface at an angle of 60° to the normal. By rotating the wafer and simultaneously moving the optical measurement system radially, the entire surface of the wafer can be scanned in a spiral pattern. The signal intensity measured at each point is encoded in color, grayscale, or false colors and displayed in a high-resolution image (substrate mapping). Fig. Figure 2 shows typical ellipsometric lateral substrate maps, so-called candela images, for various final cleaning technologies. Both the anisotropic reflection properties of GaAs and changes in the optical properties of the transparent surface oxide layer, which can be caused, for example, by impurities or uneven wet cleaning of the wafers, lead to locally varying reflection behavior due to changes in layer thickness and / or refractive indices. This results in local fluctuations in the phase shift, which can be quantitatively investigated as a measure of surface homogeneity or the optical homogeneity of the oxide layer.

[0067] The radial and azimuthal resolutions are adjustable over a very wide range. For characterizing large-area GaAs wafers, a radial resolution of 50 µm (or 45 µm with a 5 µm beam width) and an azimuthal resolution of 16384 measurement points per track (corresponding to a resolution of 29 µm or 25 µm after subtracting the beam diameter at the outer circumference of a 150 mm wafer, or 0.01 µm in the innermost measurement circle) has proven suitable.

[0068] The Candela measuring system combines four different detectors for the simultaneous measurement of scattering intensity, topography, reflectivity, and phase shift of the gallium arsenide substrate. This combination enables comprehensive characterization and defect detection with regard to contamination from process residues, point defects, topographic anomalies, and surface or (oxide) layer homogeneity of the gallium arsenide substrate. The phase shift (phase shift channel) is used to characterize the surface homogeneity according to the present invention. Referring again to the schematic representation of Fig. 1. Referring to this, the measurement principle is explained in more detail. For this particular form of ellipsometric measurement, the laser beam striking the wafer surface is polarized in a specific way. Referred to as Q-polarization, this is a hybrid of perpendicularly and parallel incident, each linearly polarized, beam components. Without being bound to this theory, it is assumed that differing reflection behaviors of the two components at the surface of the oxide layer and at the interface between the oxide layer and the substrate, as well as differing refractive properties within the oxide layer, cause an optical path length difference between the s- and p-components in the reflected laser beam, resulting in a phase shift between the two components.This phase shift can be determined as the difference between the signals of the two detectors after optical separation of the two components and their spatially separated detection. Without being bound to this theory, it is further hypothesized that changes in the optical properties of the transparent surface oxide layer, which may be caused, for example, by impurities or uneven wet cleaning of the wafers, lead to locally differing refraction and reflection behavior due to altered layer thicknesses and / or refractive indices. This results in local fluctuations in the phase shift, which can be quantitatively investigated as a measure of surface homogeneity or the optical homogeneity of the oxide layer. In the mapping generated during the measurement, local differences in these layer properties are represented as brightness differences.

[0069] The high resolution of the measurement method and the high sensitivity lead to a very accurate representation of the optical surface properties, which is not achieved, or at least not usually achieved, with ellipsometers other than those described and defined here.

[0070] When measuring GaAs surfaces, a crystallographically induced, twofold anisotropic reflection of light occurs. The associated signal fluctuation is superimposed on the actual measurement signal. To improve the sensitivity of the method, the twofold intensity fluctuations of the background caused by anisotropic reflection are corrected. For data traces recorded, for example, in a spiral or circular pattern, the twofold behavior is modeled according to the following equation using a discrete complex Fourier transform with corresponding frequencies and the Levenberg-Marquard algorithm (see, for example, JJ Moré, in G.A. Watson (ed.): Numerical Analysis. Dundee 1977, Lecture Notes Math. 630, 1978, pp. 105-116): f(φ)=a0+a2 cos(4πφ+α), Here, φ is the function variable, α is an angular offset, a0 is the absolute offset, and a2 is the amplitude of the model function. Fig. Figure 3 illustrates a typical, measured signal waveform (raw data) of a circular track, i.e., along a measurement circle as a function of the spatial coordinate. The line "background" describes the crystallographically induced component of the measurement signal, which can be modeled by Fourier transformation using the Levenberg-Marquard algorithm. The long-wavelength, crystallographically induced background vibration can be separated from the surface-related component of the measurement signal by modeling the twofold waveform with the cosine function shown above for each circularly recorded data track using a discrete complex Fourier transform with corresponding frequencies, employing the Levenberg-Marquard algorithm.

[0071] Fig. Figure 4 shows exemplary phase shifts of a track determined after subtracting the crystallographically induced sub- or background signal; that is, a typical course of the measurement signal along a measurement circuit as a function of the spatial coordinate after subtracting the crystallographically induced background vibration is shown. This corresponds to the laterally resolved background-corrected measurement signal.

[0072] The background-cleaned mappings no longer show twofold symmetry, which could interfere with the characterization of deviations in the homogeneity of surface properties. This is true, firstly, for typical ellipsometric lateral substrate maps, so-called Candela images, after subtraction or correction of the crystallographically induced background signal for various final cleaning technologies in Fig. 5 illustrates. Inhomogeneities in surface properties now become more apparent (cf. Fig. 2).

[0073] Furthermore, all deviations from the ideal homogeneous surface (residues of the background-cleaned mapping) can be represented in a histogram. Fig. Figure 6 presents typical global frequency distributions of the background-cleaned phase shifts, i.e., residues of the background-cleaned mappings, for various final cleaning technologies. Since this distribution typically does not correspond to a Gaussian normal distribution, the commonly used standard deviation cannot be used as a measure of homogeneity. Rather, it is appropriate to define and calculate suitable percentiles. To evaluate the homogeneity of the GaAs surfaces according to the present invention, the percentiles P1 and Q were used. 0,01 (see, for example, F. Schoonjans, D. De Bacquer, P. Schmid P, “Estimation of population percentiles”. Epidemiology, 22, 2011, pp. 750-751) for comparisons.

[0074] Fig. Figure 7 presents comparative typical 1% percentiles of a suitable number of substrates / wafers—specifically, 25 wafers each—which were fabricated according to the methods described below for Comparison Examples 1 and 2, as well as for Examples 1 and 2. The percentiles are normalized to the respective wafer mean values ​​of the phase shift signal to compensate for variations in the incident light intensity. Box-and-whisker plots are used for the statistical representation (see, for example, PJ Govaerts, T. Somers, FE Officers, Otolaryngology - Head and Neck Surgery, 118(6), June 1998, pp. 892–895 and JW Tukey: Exploratory data analysis. Addison-Wesley 1977, ISBN 0-201-07616-0). Fig. 7 (see also Examples 1-2 and Comparative Examples 1 and 2) shows, based on the respective 1% percentiles of the signal deviations of the phase shift of the light reflection normalized to the wafer mean of the phase shift signal, which can be used to evaluate the homogeneity of the GaAs surfaces, that the previously used technology produces wafers with surface homogeneities that exhibit the 1% percentiles defined above of less than -0.0065. The differences between the wafers produced according to the invention and the reference wafers are significant and reproducible. In general, and as demonstrated here in detail, wafers produced according to the invention, in contrast to reference wafers, show values ​​greater than -0.0065, preferably greater than -0.0060, more preferably greater than -0.0055, even more preferably greater than -0.0050, even more preferably greater than -0.0045, and most preferably greater than -0.0040.Even values ​​greater than -0.0030, preferably greater than -0.0020, further preferably greater than 0.0010 and up to 0.0000 exclusively 0.0000 are considered.

[0075] The background-cleaned mappings can alternatively be characterized by defect classification, for example, using the evaluation software available for the Candela CS20 and described accordingly. The mappings show light and dark areas as well as striped structures that can be assigned to inhomogeneities or, more generally, "defects." The intensity and number of deviations from the background characterize the homogeneity of a surface. The intensity differences are classified and counted according to their magnitude and area. Specific parameters are selected to define the signals and defects to be evaluated. In this case, the following definitions were made, particularly using the Candela CS20 evaluation software (see Table 1 for the definition of the measurement signals to be evaluated). Table 1 Signal Typ Neg. Abweichung [%] Pos. Abweichung [%] Kernel Länge[µm] Kernel Typ Radial Stitching[pixel] Circular Stitching[pixel] QAbsPhase 0,05 0,05 1000 Median 10 10

[0076] Negative and positive intensity fluctuations (deviations) are taken into account starting at a threshold of 0.05%. The kernel length describes the averaging area, and the kernel type specifies the averaging method. If the positive and negative exceedances of the threshold values ​​meet the spacing criteria defined under "Radial stitching" and "Circular stitching," they are counted as individual defects and summed across the wafer. The stitching parameters specify the minimum distance between measurement points that positive or negative exceedances must have to be counted as separate defects. Defects with a detected area greater than 1000 µm have proven to be relevant. 2 The sum of such counted defects is a measure of the homogeneity of a wafer. Fig. Figure 8 shows typical defect maps of substrate-cleaned Candela images for various final cleaning technologies. Such defect maps can be used for defect classification. As described above, the evaluation software of the Candela CS20 measuring device is used for this purpose, classifying and counting the intensity differences of the phase shift with respect to their magnitude and area. The sum of such defects is a measure of the homogeneity of a substrate or wafer. For substrates or wafers produced using conventional technology, this results in defect counts of > 6000 and / or defect areas of > 2 cm², based on a substrate or wafer diameter of 150 mm as a reference. 2Substrates or wafers according to the invention, which can be produced by the inventive method, exhibit, based on a substrate or wafer diameter of 150 mm as a reference, defect counts of < 6000, preferably < 5000, more preferably < 4000, even more preferably < 3000, even more preferably < 2000, even more preferably < 1000, even more preferably < 500, even more preferably < 300, even more preferably < 250, even more preferably < 200, even more preferably < 150 and particularly preferably < 100, and / or they exhibit defect areas of < 2 cm² 2 , preferably < 1 cm 2 , preferably < 0.5 cm 2 , even more preferably < 0.1 cm 2 , even more preferably < 0.05 cm 2 , even more preferably < 0.01 cm 2 , even more preferably < 0.005 cm 2 and especially preferred < 0.0035 cm 2 .

[0077] For a comparison of conventional and inventive values ​​for defect number and defect area, see also Fig. 9 as well as examples 1 and 2 and comparative examples 1 and 2. Fig. Figure 9 shows typical defect numbers and defect areas of 25 wafers each, which were manufactured according to conventional methods (see comparative examples 1 and 2) and methods according to the invention (see examples 1-2). The differences between the wafers manufactured according to the invention and the comparative wafers are significant and reproducible.

[0078] In another aspect, said gallium arsenide substrate can have a diameter of at least 100 mm, preferably at least 150 mm and more preferably at least 200 mm.

[0079] The advantageous features and properties achievable with the GaAs wafers obtained according to the invention are not limited to the conventionally used thickness of GaAs wafers, i.e., in the range of approximately 600 µm to approximately 800 µm, and especially with the usual standard thickness of approximately 675 µm (± 25 µm). Rather, significantly thinner and thicker GaAs wafers are now also accessible. This is due in particular to the overall gentler processing and the very homogeneous and considerably defect-reduced surface properties according to the present invention.

[0080] The present invention thus provides, in an independent aspect, for the first time finished gallium arsenide substrates with thickness ranges that were previously unattainable due to the product properties not achievable, namely a thickness in the range of ≤ approx. 600 µm and alternatively a thickness in the range of ≥ approx. 800 µm. Further preferred thickness ranges for the thinner substrates are in the range of approx. 100 to approx. 600 µm, more preferably in the range of approx. 250 to approx. 500 µm, and for the thicker substrates in the range of approx. 800 to approx. 2000 µm. The diameter of the gallium arsenide substrates according to the invention is preferably at least 150 mm. The gallium arsenide substrate products provided according to the invention are finished, i.e., at least surface-treated and preferably polished and surface-treated.The surface finishing treatment includes, in particular, oxidation treatment of at least one surface of the gallium arsenide substrate in the dry state using UV radiation and / or ozone gas, contacting the at least one surface of the gallium arsenide substrate with at least one liquid medium, and Marangoni drying of the gallium arsenide substrate. Known methods can be used for polishing the GaAs surface. For further details regarding the surface finishing treatment, reference is made to the more detailed description of the process according to the invention.

[0081] The treated surface of the polished and surface-finished gallium arsenide substrate preferably exhibits the surface features already described above in connection with ellipsometric lateral substrate mapping using an optical surface analyzer, to which reference is made here. Therefore, finished gallium arsenide substrates according to the invention can alternatively be characterized as follows: (i) by the layer thickness being relatively thinner or thicker compared to the standard thickness; (ii) by the properties that, unlike those of conventional finishes, are only obtainable through the surface finish treatment according to the invention; and (iii) by the differences detectable by ellipsometric lateral substrate mapping using an optical surface analyzer. Reference is made to further explanations and definitions provided in this application.

[0082] The term “circa” or “ca.” used here means that in practice it is not important to specify the exact value; rather, tolerances of, for example, ± 25 µm are possible, with preferred tolerance ranges being ± 20 µm, more preferably ± 15 µm, even more preferably ± 10 µm, and most preferably ± 5 µm.

[0083] The gallium arsenide substrate according to the invention can be doped or undoped.

[0084] Another aspect of the present invention provides for a gallium arsenide substrate which has at least one surface which, within 9 months, preferably 12 months, after production, exhibits a substantially non-deteriorating, preferably non-deteriorating, variation of the laterally resolved background-corrected measurement signal during ellipsometric lateral substrate mapping with an optical surface analyzer.

[0085] "Essentially" here means a change of ≤ 10%, preferably ≤ 5%, of the 1% distribution percentile normalized to the substrate mean of the phase shift signal and / or, with reference to a substrate diameter of 150 mm, the number of defects and / or the total defect area, with reference to the preceding explanations and definitions of 1% distribution percentile, number of defects and total defect area.

[0086] Another aspect of the present invention provides for a gallium arsenide substrate, wherein the at least one surface exhibits, within 6 months of production, a substantially non-deteriorating, preferably non-deteriorating, variation of the laterally resolved background-corrected measurement signal during ellipsometric lateral substrate mapping with an optical surface analyzer.

[0087] "Essentially" here means a change of ≤ 10%, preferably ≤ 5%, of the 1% distribution percentile normalized to the substrate mean of the phase shift signal and / or, with reference to a substrate diameter of 150 mm, the number of defects and / or the total defect area, with reference to the preceding explanations and definitions of 1% distribution percentile, number of defects and total defect area.

[0088] The inventive method not only produces a very homogeneous surface over a large area, but also advantageously maintains it very stably for a period of at least 6 months, preferably 9 months, and more preferably 12 months. This is verified and confirmed by candela measurements over time after production. Proper storage, in particular storage of the substrate in darkness under a particle-free inert gas atmosphere (e.g., N2), can contribute to the longer-term stability of the substrate surface.

[0089] Another reference aspect concerns a plurality of gallium arsenide substrates which are produced according to the inventive method and which exhibit a substantially similar, preferably identical, variation from substrate to substrate of the laterally resolved background-cleaned measurement signal when performing ellipsometric lateral substrate mapping of the at least one surface with an optical surface analyzer.

[0090] "Essentially" here means a change of ≤ 10%, preferably ≤ 5%, of the 1% distribution percentile normalized to the substrate mean of the phase shift signal and / or, with reference to a substrate diameter of 150 mm, the number of defects and / or the total defect area, with reference to the preceding explanations and definitions of 1% distribution percentile, number of defects and total defect area.

[0091] The highly reproducible process and the possibility of batch production allow for the creation of a wide variety of GaAs substrates that exhibit very low variability in their surface properties. This is verified and confirmed using ellipsometric candela mapping.

[0092] Another aspect of the invention is the use of the gallium arsenide substrate according to the present invention for epitaxial crystal growth, optionally after storage and preferably without pretreatment after providing the gallium arsenide substrate and before epitaxial crystal growth.

[0093] The gallium arsenide substrate can be used, among other things, for the fabrication of semiconductor devices or electronic and optoelectronic components, power devices, high-frequency devices, light-emitting diodes, and lasers. The excellent surface properties of the substrate according to the invention enable the reproducible fabrication of epitaxial layers with high yield.

[0094] Another object of the present invention relates to the use of an optical surface analyzer, preferably a Candela CS20 optical surface analyzer, for the optical contactless quantitative characterization of the homogeneity of surface properties of gallium arsenide substrates by means of ellipsometric lateral substrate mapping, wherein a laterally resolved measurement signal is more preferably cleaned of a lower frequency background by means of discrete complex Fourier transform, preferably using the Levenberg-Marquardt algorithm. Examples, materials and methods

[0095] Candela ellipsometry: The surface properties of the substrate or wafer (the properties of the oxide surface on the wafer) are characterized after final cleaning using an optical surface analyzer (OSA). The phase shift measurement ("phase shift" channel) of the Candela CS20 from KLA-Tencor is used to characterize the homogeneity of the surface properties. The measurement principle and setup have already been described above in connection with ellipsometric lateral substrate mapping with an optical surface analyzer, to which reference is made here (see also...). Fig. 1) Due to the oblique incidence of the light, the laser irradiates an elliptical area on the wafer with a radial extent of approximately 5 µm and a perpendicular extent of approximately 4 µm. The fluctuations in the phase shift between the s- and p-components of the reflected laser beam, measured with the CS20, are used as a measure of the surface homogeneity of a GaAs wafer. In the mapping generated during the measurement, local differences in these layer properties are represented as brightness variations.

[0096] When measuring GaAs surfaces, a crystallographically induced, bifold anisotropic reflection of light also occurs. The associated signal fluctuation is superimposed on the actual measurement signal. Typical results ("Candela maps") of ellipsometric lateral mappings of surfaces of differently produced and purified GaAs substrates can be found in Fig. 2. The different phase shift values ​​are represented as light / dark contrast. The typical twofold light / dark variation is visible in the background, resulting from the anisotropy of light reflection at the GaAs itself. This variation is relatively slow and exhibits a relatively low frequency. Furthermore, bright and dark areas, as well as fringe structures, are visible, standing out from the background signal and indicating inhomogeneities or defects. The intensity and number of deviations from the background characterize the homogeneity of a surface. 1. Candela measurement

[0097] The parameters used for the candela measurement are shown in Table 2 (showing the so-called "recipe parameters" of the measurement, "scan recipe") and Table 3 (showing the "wafer setup"). The parameters in the "Scan Area" column describe the wafer area swept by the measurement and are set to the dimensions of a 150 mm wafer as an example in Table 3. The rotational speed of the measuring cylinder is set with the "Speed" parameter. The "Sampling Average" parameter defines the number of repeated measurements from which the measurement result is determined. The "Step Size" parameter sets the radial measurement resolution, from which the number of tracks in the "Total Tracks" parameter is derived. The "Encoder Multiplier" parameter sets the azimuthal resolution. The setting "16x" corresponds to 16,384 measurement points per track.The radial and azimuthal resolution shown in the lower rows of the "Scan resolution" column results from the settings for "Step size" and "Encoder Multiplier".

[0098] The "Laser" column specifies the use of the azimuthal laser, which is necessary for measuring the phase shift, as is the Q-polarization of the laser beam in the following column. Finally, the last column of Table 2 specifies certain voltages and offsets, which, like the other parameters, have proven advantageous for measurements within the scope of the present invention. Testing of various radial resolution settings has shown that a track spacing between 10 and 75 µm has no effect on the quantified surface homogeneity measurements. The azimuthal resolution of 29 µm used for the measurements is higher than the radial resolution of 50 µm, even in the outermost measurement circle. Table 2 Scan Area Spindlecontrol Scan resolution Laser Q-Polarization Gain and offsets Q-Polarization Start:r=75000µm Angle=0° Speed =3000 rpm Step size = 50 µm Circumferential Phase Sp1=0,5V; Offset = 47 Stop:r=0µm,Angle =360° Samplingaverage = 1 Total tracks =1501 sp2=0,5V; Offset = 56 Encoder multiplier = 16x PMT circumf. =400V Total resolution :radial = 50,000µm, Auto PMT offsetcir.= 0 Total resolution :angular= 0 - 28,762 µm Preset GainRange

[0099] Table 3 summarizes further parameters for characterizing the GaAs wafers to be measured, relating to the wafer geometry, the wafer thickness (via the data set stored under “Focus”), as well as the general edge exclusion and a special edge exclusion for the notch. Table 3 Wafer Focus Imageangle Analysis area Image rotation angle round L_6inGaAs675 µm As scanned Start at 68000 µm radius 0 150mm Notch exclusion:1 = 1500µm, w=3000µm, Center =0 und 270° notch

[0100] As a result of the candela measurement, a data file is stored which, in addition to basic information about the sample and the measurement conditions, contains the measurement data together with the associated spatial coordinates in a binary packed format. 2. Subsurface remediation

[0101] To further improve the sensitivity of the method, it is useful to correct or clean up the twofold intensity fluctuations of the background caused by anisotropic reflection. For the spirally or circularly recorded data tracks, the twofold behavior is modeled according to the following equation using a discrete complex Fourier transform with corresponding frequencies and the Levenberg-Marquard algorithm (see JJ Moré): f(φ)=a0+a2 cos(4πφ+α), Here, φ is the function variable, α is an angular offset, a0 is the absolute offset, and a z the amplitude of the model function (see also Fig. 3) The background-cleaned measurement signal of a data track is in Fig. 4 shown. The background-cleaned mappings no longer show twofold symmetry (see Fig. 5), which could interfere with the characterization of deviations in the homogeneity of surface properties. 3. Statistical evaluation of the signal variations

[0102] To determine the homogeneity of a substrate surface, all signal variations from the ideally homogeneous surface (residues of the background-cleaned mapping) can be mapped in a histogram (see Fig. 6) Since this distribution typically does not correspond to a Gaussian normal distribution, the commonly used standard deviation cannot be used as a measure of homogeneity. Instead, suitable percentiles must be defined and calculated. To assess the homogeneity of the GaAs surfaces, the percentiles P1 and Q were used. 0,01 (See, for example, F. Schoonjans, D. De Bacquer, P. Schmid, “Estimation of population percentiles”, Epidemiology, 22, 2011, pp. 750-751) for comparisons. Due to possible fluctuations in excitation intensity, the percentiles are normalized to the wafer mean of the phase shift signal. In Fig. Figure 7 presents the data in a comparative format. Box-and-whisker plots are used. Conventional final cleaning technology produces wafers with surface homogeneities that exhibit the 1% percentiles defined above of less than -0.0065. Wafers produced according to the invention show values ​​greater than -0.0065, preferably greater than -0.0060, more preferably greater than -0.0055, even more preferably greater than -0.0050, and more preferably greater than -0.0045. Values ​​greater than -0.0040, preferably greater than -0.0030, more preferably greater than -0.0020, and particularly greater than 0.0010 and even down to 0.0000 can be considered. 4. Classification and counting of defects

[0103] The background-cleaned mappings can also be characterized using the defect classification of the Candela CS20's evaluation software. The mappings show light and dark areas as well as stripe structures that can be assigned to inhomogeneities or defects. The intensity and number of deviations from the background characterize the homogeneity of a surface. The intensity differences are classified and counted according to their magnitude and area. Specific parameters are selected to define the signals and defects to be evaluated. For the definitions used in this case, see Table 1 and Tables 2-3. Negative and positive intensity fluctuations (deviations) are considered above a threshold of 0.05%. The kernel length describes the averaging area, and the kernel type specifies the type of averaging.Provided that the positive and negative exceedances of the limit values ​​meet the spacing criteria defined under "Radial stitching" and "Circular stitching," they are counted as individual defects and summed across the wafer. The stitching parameters specify the minimum distance between measurement points that positive or negative exceedances must have to be counted as separate defects. These defects are only counted if they are larger than 1000 µm. 2 The sum of such defects is a measure of the homogeneity of a wafer (see Fig. 8) For wafers manufactured using conventional technology, based on a wafer or substrate diameter of 150 mm as a reference, defect counts of > 6000 and defect areas of > 2 cm² result. 2Wafers produced according to the invention show defect counts of < 6000, preferably < 5000, more preferably < 4000, even more preferably < 3000, even more preferably < 2000, even more preferably < 1000, even more preferably < 500, even more preferably < 300, even more preferably < 250, even more preferably < 200, even more preferably < 150 and particularly preferably < 100 and / or defect areas of < 2 cm² 2 , preferably smaller than 1 cm 2 , preferably smaller than 0.5 cm 2 , even more preferably smaller than 0.1 cm 2 , even more preferably smaller than 0.05 cm 2 , even more preferably smaller than 0.01 cm 2 , even more preferably smaller than 0.005 cm 2 and especially preferred to be smaller than 0.0035 cm 2 (see Fig. 9). 5. Roughness measurement

[0104] White light interferometry was used for roughness measurements within the scope of this invention. In white light interferometry, interference images are captured with a camera. These images result from the superposition of light from the object being measured with light reflected from a reference mirror. For topography measurements, the z-position of the lens is adjusted in small increments, and an interference image is captured at each position. This produces a stack of images from which the height data is calculated. By using a white light source with a short coherence length, surfaces can be measured with the very high height resolution known for interferometric measurement methods. The Zygo NewView 5022S device was used for the roughness measurements within the scope of this invention. The measurements were performed with a device equipped with a 20x magnification lens. The measurement field size was 180 x 130 µm. The specified roughness Ra is the difference between the maximum and minimum height values ​​for the specified measurement field size. 6. Determination of the etching loss

[0105] To determine the etching depth, a chemically resistant adhesive tape was applied to the front face of a GaAs wafer between dry oxidation and wet cleaning according to the various embodiments of the present invention. After the cleaning process, this special adhesive tape was removed without leaving any residue, and the height of the resulting step was measured at five points using a white light interferometer according to the method described above. Comparative example 1

[0106] After the final polishing step, a GaAs wafer undergoes basic cleaning with a 0.5% NH4OH solution and acidic cleaning with a 5% HF solution to remove metallic contaminants. Subsequently, particles are removed from the wafer surface by a brush scrubbing process. The GaAs wafer cleaning procedure is completed by rinsing with deionized water and drying via spin drying.Following this conventional process for surface cleaning of the GaAs wafer, when measuring surface homogeneity with the Candela CS20 measuring device using the above-described recipes for measurement and defect evaluation, a variation in the laterally resolved background-corrected measurement signal is found, the distribution percentile of which normalized to the wafer mean of the phase shift signal is less than -0.0065, and, based on a 150 mm GaAs wafer as a reference size, a defect count of greater than 6000 and a total defect area of ​​greater than 2 cm² are found. 2 on the surface (see Fig. ). Comparative example 2

[0107] Following conventional cleaning and drying, as described in the first comparative example, a GaAs wafer undergoes an oxidation process. This oxidation is achieved by irradiating the entire wafer surface with short-wave UV light (wavelength 220–480 nm, power 20–40 mW / cm²). 2The wafer is slowly rotated for, for example, one minute. It is then subjected to basic cleaning in a 0.5% NH3 solution under megasonic influence in a process tray, subsequently rinsed in the overflow, and then removed from the tray and dried by spin drying at 2500 rpm. Measuring surface defects on such cleaned substrates with the Candela CS20 instrument reveals a variation in the laterally resolved, background-corrected measurement signal, whose 1% distribution percentile normalized to the wafer mean of the phase shift signal is less than -0.0065, more than 6000 individual defects, and a total defect area greater than 2 cm². 2 on the surface of the 150 mm wafer as a reference (see Fig. ). Example 1

[0108] Following the conventional cleaning procedure as in Comparative Example 1, a GaAs wafer undergoes the further steps of dry oxidation, NH4OH cleaning, and DI water rinsing as in Comparative Example 2. However, unlike Comparative Example 2, the wafer is dried using the Marangoni process instead of spin drying.

[0109] Measurement of surface defects with the Candela CS20 instrument reveals a variation in the laterally resolved, background-corrected measurement signal, whose 1% distribution percentile, normalized to the wafer mean of the phase shift signal, is greater than -0.0065. Furthermore, fewer than 100 individual defects and a total defect area of ​​less than 2 cm² were detected. 2on the GaAs surface of the 150 mm wafer, thus demonstrating the superiority of this cleaning method over conventional cleaning in comparison example 1 and even over improved cleaning but with conventional drying in comparison example 2 (see Fig. ). Example 2

[0110] The cleaning of a GaAs wafer is initially carried out in a conventional manner, as in Comparative Example 2. This is followed by dry oxidation and alkaline cleaning with subsequent DI water rinsing, as in Comparative Example 2 or Example 1. Following the DI water rinsing, this embodiment of the present invention includes a further acidic cleaning step in combination with ozone dissolved in the liquid. During the DI water rinsing step following the acidic cleaning step, an acid or base is also added to advantageously prevent the formation of an inhomogeneous oxide layer during the rinsing process.When using 0.2% HCl and 50 ppm ozone in the acidic cleaning step for a process time of 3 minutes, followed by a DI water rinse with the addition of a small amount of 25% NH3 solution, the surface homogeneity measurement performed after Marangoni drying with the Candela CS20 instrument shows a variation in the laterally resolved background-corrected measurement signal, whose 1% distribution percentile, normalized to the wafer mean of the phase shift signal, is greater than -0.0065. Furthermore, fewer than 100 defects and a total defect area of ​​less than 2 cm² are found. 2 on the surface of the 150 mm GaAs wafer (see Fig. ).

[0111] If the final DI water rinse is performed without the addition of a pH-modifying substance, the defect level measured after the cleaning process can rise to over 2000 defects on a 150 mm GaAs wafer. Example 3

[0112] If hydrogen peroxide is used as the oxidizing agent in the acidic cleaning step instead of ozone in the treatment according to Example 2, the Candela measurement after Marangoni drying yields over 4000 defects.

Claims

[1] Method for producing a surface-treated gallium arsenide substrate comprising the steps: a) Providing a gallium arsenide substrate; b) Oxidation treatment of at least one surface of the gallium arsenide substrate in the dry state using UV radiation and / or ozone gas; c) Contacting at least one surface of the gallium arsenide substrate with at least one liquid medium; comprising the following sub-steps: ci) Contacting at least one surface of the gallium arsenide substrate with alkaline aqueous solution, c-ii) subsequent contact of at least one surface of the gallium arsenide substrate with water, c-iii) subsequent contact of at least one surface of the gallium arsenide substrate with acidic aqueous solution, c-iv) subsequent further contact of at least one surface of the gallium arsenide substrate with water. d) Marangoni drying of the gallium arsenide substrate. [2] Method according to claim 1, wherein step ci) is carried out using megasound. [3] Method according to claim 1 or 2, wherein in step ci) an alkaline aqueous solution is used which is a solution of NH3 or organic amine in water, preferably of NH3. [4] Method according to any one of claims 1 to 3, wherein step c-iii) is carried out in the presence of an oxidizing agent; and in step c-iv) the water contains at least an initial pH-modifying additive. [5] Method according to claim 4, wherein the acidic aqueous solution is a solution of HCl or HF in water and the optional oxidizing agent in the acidic aqueous solution is ozone or H2O2. [6] Method according to claim 4 or 5, wherein in step c) following step iv) further steps according to steps i) and ii) are carried out. [7] Method for producing a plurality of surface-treated gallium arsenide substrates, wherein simultaneously a plurality of gallium arsenide substrates are subjected to the respective steps b)-d) in the method according to one of the preceding claims. [8] Gallium arsenide substrate which has at least one surface which, when ellipsometrically lateral substrate mapping with an optical surface analyzer, with reference to a substrate diameter of 150 mm, shows a defect count of < 6000 and / or a total defect area of ​​less than 2 cm² 2 exhibits a defect defined as a contiguous area larger than 1000 µm 2 defined by a deviation from the mean measurement signal in ellipsometric lateral substrate mapping with an optical surface analyzer of at least ± 0.05%. [9] Gallium arsenide substrate according to claim 8, wherein the diameter is at least 100 mm, preferably at least 150 mm and further preferably at least 200 mm. [10] Gallium arsenide substrate according to one of claims 8 or 9, wherein the at least one surface exhibits a non-deteriorating variation of the laterally resolved background-corrected measurement signal within 6 months after production when using ellipsometric lateral substrate mapping with an optical surface analyzer. [11] Use of the gallium arsenide substrate according to one of claims 8 or 9 for epitaxial crystal growth, optionally after storage and preferably without pretreatment after providing the gallium arsenide substrate and prior to epitaxial crystal growth.

Images