Silicon nitride substrate

Abstract

To obtain a silicon nitride substrate having high uniformity, from which a piece having high quality stability in thermal cycle resistance can be obtained.SOLUTION: The present invention relates to a silicon nitride substrate in which a ratio αe / αc of an average linear expansion coefficient αc at a central part to an average linear expansion coefficient αe at an end part is 0.90 to 1.10.SELECTED DRAWING: None

Description

[Technical Field]

[0001] The present invention relates to silicon nitride substrates. [Background technology]

[0002] Silicon nitride sintered bodies have excellent mechanical strength, thermal conductivity, and electrical insulation properties, and are therefore used as substrates, particularly in applications requiring high voltages and currents, such as power modules (see, for example, Patent Documents 1 and 2).

[0003] In the case of silicon nitride substrates, from the viewpoint of manufacturing efficiency, a method is sometimes used in which a large silicon nitride substrate is manufactured and then cut into multiple individual pieces. In this case, if there is variation in the quality of the silicon nitride substrate before cutting, there will be variation in the quality among the individual pieces obtained by cutting. Therefore, in order to obtain individual pieces with high quality stability, high uniformity of the silicon nitride substrate is required. However, when attempting to manufacture large-sized silicon nitride substrates, quality variation is likely to occur depending on the part of the substrate.

[0004] In response to this, Patent Document 1 proposes a silicon nitride substrate with improved uniformity in the magnesium content, etc., from the perspective of preventing variations in warpage in large-sized substrates. Patent Document 2 conducts a study focusing on the void fraction in the center and the void fraction in the edge of large-sized substrates. [Prior art documents] [Patent documents]

[0005] [Patent Document 1] WO2020 / 203787 issue [Patent Document 2] Japanese Patent Application Publication No. 2019-059639 Summary of the Invention [Problem to be solved by the invention]

[0006] Silicon nitride substrates, taking advantage of their high thermal conductivity, are used in areas where heat dissipation performance is important, so it is important that they do not peel off from metal circuit boards even when subjected to repeated thermal cycles. Previously, studies focusing on uniformity, which affects warping and voids, as in Patent Documents 1 and 2, have been conducted, but no studies focusing on uniformity in thermal cycle resistance have been conducted.

[0007] The present invention has been made in view of the above circumstances, and an object of the present invention is to obtain a highly uniform silicon nitride substrate from which individual pieces having high quality stability in thermal cycle resistance can be obtained. [Means for solving the problem]

[0008] In order to solve the above problems, the present inventors have conducted extensive research and have succeeded in obtaining a silicon nitride substrate that has extremely high quality stability in terms of thermal cycle resistance.

[0009] The present invention provides an average linear expansion coefficient α C The silicon nitride substrate has a ratio αe / αc of the average linear expansion coefficient αe of the central portion to the average linear expansion coefficient αe of the edge portion, which is 0.90 to 1.10. -6 / K or more. Furthermore, the silicon nitride substrate of the present invention preferably has a main surface having a size of 140 mm×140 mm or more. DETAILED DESCRIPTION OF THE INVENTION

[0010] The silicon nitride substrate of the present invention has a ratio αe / αc of the average linear expansion coefficient αc at the center to the average linear expansion coefficient αe at the edge, of 0.90 to 1.10. This means that the average linear expansion coefficient is uniform between the center and edge of the silicon nitride substrate. The average linear expansion coefficient in the present invention is measured using a push rod type dilatometer (JIS R 1618:2002 "Method for measuring thermal expansion of fine ceramics by thermomechanical analysis").

[0011] In the present invention, the central portion refers to the vicinity of the intersection of two diagonal lines drawn from the four corners of the main surface of the silicon nitride substrate. Specifically, to measure the average linear expansion coefficient αc of the central portion, a test piece cut to a size of 3 mm × 15 mm and centered on the intersection of the diagonal lines is used.

[0012] In the present invention, the term "edge" refers to the vicinity of the four corners of the main surface. Specifically, to measure the average linear expansion coefficient αe of the edge, a test piece cut to a size of 3 mm x 15 mm and centered at a point 30 mm diagonally from the corner of the main surface is used. The average linear expansion coefficient αe of the edge is measured at all four points, and the point with the largest difference from αc, i.e., the point at which αe / αc is the value farthest from 1.00, is used.

[0013] Generally, silicon nitride substrates are most susceptible to differences in properties between the center and edges due to differences in the sintering environment. Therefore, uniformity between the center and edges can be considered to be uniform throughout. The delamination between a silicon nitride substrate and a metal circuit plate during repeated thermal cycles is primarily caused by stress generated by differences in the expansion and contraction of the silicon nitride substrate and the metal circuit plate. Therefore, differences in the average linear expansion coefficient between individual pieces lead to differences in thermal cycle resistance. However, the silicon nitride substrate of the present invention has a uniform average linear expansion coefficient, and is therefore considered to have equivalent thermal cycle resistance. Based on the above, individual pieces cut from a silicon nitride substrate having an αe / αc ratio in the range of 0.90 to 1.10 all have equivalent thermal cycle resistance and are therefore silicon nitride substrates with high quality stability. The silicon nitride substrate of the present invention preferably has an αe / αc ratio in the range of 0.95 to 1.05, more preferably in the range of 0.99 to 1.01.

[0014] The average linear expansion coefficient of the silicon nitride substrate of the present invention is αc of 2.8 × 10 -6 / K or more, and 3.0 × 10 -6 / K or more. When αc is in the above range, it becomes easy to improve the thermal cycle resistance. Similarly, αe is 2.8×10 -6 / K or more, and 3.0 × 10 -6 / K or more. The average linear expansion coefficient of the silicon nitride substrate is generally higher than that of copper (17.7 × 10 -6 / K), it is preferable because it is easier to improve the thermal cycle resistance. However, it is generally not possible to improve the average linear expansion coefficient to the same level as copper, and αc and αe are usually 3.3 × 10 -6 / K or less.

[0015] In the silicon nitride substrate of the present invention, the ratio Oe / Oc of the amount of dissolved oxygen in the center Oc to the amount of dissolved oxygen in the edge Oe is preferably in the range of 0.95 to 1.05, more preferably in the range of 0.97 to 1.03, and even more preferably in the range of 0.99 to 1.01, which means that the amount of dissolved oxygen is uniform between the center and the edge of the silicon nitride substrate.

[0016] The main factor that determines the average linear expansion coefficient of a silicon nitride substrate is the behavior of the silicon nitride particles, which are the main component. The amount of dissolved oxygen affects the state of the silicon nitride particles, and by setting the Oe / Oc ratio within the above range, it is possible to achieve a uniform amount of dissolved oxygen and therefore a uniform average linear expansion coefficient.

[0017] Furthermore, since oxygen is important for promoting the reaction between silicon nitride particles during sintering of silicon nitride substrates, if Oe / Oc is within the above range, it can be said that the sintering reaction is proceeding uniformly. Here, in a silicon nitride substrate, if the state of the silicon nitride particles, which are the main component, is uniform, the uniformity of the glass phase, which is the second most abundant component, affects the uniformity of the average linear expansion coefficient. Although it is difficult to directly analyze the uniformity of the glass phase state, since the glass phase is derived from a sintering aid and is formed in conjunction with the sintering reaction of silicon nitride particles, if the state of the silicon nitride particles is uniform, it means that the glass phase is formed uniformly. From this perspective, it is believed that it will be easy to set αe / αc within the above range.

[0018] The dissolved oxygen content in this invention can be measured by inert gas fusion-infrared absorption spectroscopy using an oxygen / nitrogen analyzer, and refers to the amount of oxygen dissolved in the crystals of silicon nitride particles in the substrate. Specifically, the silicon nitride substrate was crushed and then acid-treated to remove the additive phase. 25 mg of the sample was sealed in a tin capsule and placed in a graphite crucible. The crucible was heated at 5.5 kW for 20 seconds to degas the adsorbed gases, then heated at 0.8 kW for 10 seconds, and then heated from 0.8 kW to 4 kW over 350 seconds. The amount of carbon monoxide generated during this process, converted to carbon dioxide in the oxidizer, was measured, and the total oxygen content was calculated by converting the amount of carbon monoxide into carbon dioxide. During the 350-second heating period, the oxygen generated early corresponds to oxygen derived from oxides present on the surface of the silicon nitride particles (external oxygen), while the oxygen generated later corresponds to dissolved oxygen (internal oxygen) dissolved in the silicon nitride crystals. Therefore, a perpendicular line is drawn from the valley of these two measurement peaks, after subtracting the previously measured background, to separate the two peaks. The amounts of dissolved oxygen (internal oxygen) and external oxygen are calculated by proportionally allocating the respective peak areas. Note that, in addition to the two peaks, small peaks derived from impurities or the measurement environment may be detected. However, these small peaks are ignored and the largest two peaks are used to calculate the amount of dissolved oxygen. The valley of the two measurement peaks is the point between the two measurement peaks where the amount of oxygen is lowest.

[0019] The dissolved oxygen content Oc at the center is measured by first collecting a test piece for measuring the average linear expansion coefficient of the center, and then collecting a required amount of sample from around the test piece. The dissolved oxygen content Oe at the edge is measured by first collecting a test piece for measuring the average linear expansion coefficient of the edge, and then collecting a required amount of sample from around the test piece. The dissolved oxygen content Oe at the edge is measured at all four points, and the value with the largest difference from Oc is used, that is, the point at which Oe / Oc is the value farthest from 1.00 among the four points.

[0020] The amount of dissolved oxygen in the silicon nitride substrate of the present invention, Oc, is preferably 0.055% by mass or less, more preferably 0.050% by mass or less. By keeping Oc within this range, it becomes easier to achieve high thermal conductivity for the silicon nitride substrate. Similarly, Oe is preferably 0.055% by mass or less, more preferably 0.050% by mass or less. There are no particular lower limits for Oe and Oc, and they may be, for example, 0.030% by mass or more.

[0021] In the silicon nitride substrate of the present invention, the ratio De / Dc of the density Dc at the center to the density De at the edge is preferably in the range of 0.97 to 1.03, more preferably in the range of 0.98 to 1.02, and particularly preferably in the range of 0.99 to 1.01. When silicon nitride is sintered, minute voids are inevitably generated, which affect the average linear expansion coefficient. Since the generation of minute voids reduces the density, if De / Dc is in the above range, the degree of void generation is uniform, making it easy to set αe / αc in the above range. The values ​​of De and Dc are not particularly limited, and may be, for example, 3.0 g / cm 3 More than 3.5g / cm 3 Below, especially 3.1g / cm 3 More than 3.3g / cm 3 The following may be used.

[0022] The density in the present invention is measured by the Archimedes method using the sample for measuring the average linear expansion coefficient.

[0023] The silicon nitride substrate of the present invention preferably has a three-point bending strength of 600 MPa or more and a fracture toughness of 6.0 MPa m 1 / 2 It is preferable that the thermal conductivity is 80 (W / (m·K)) or more.

[0024] The silicon nitride substrate of the present invention is preferably plate-shaped. The size of the main surface is preferably 140 mm × 140 mm or more. As the size of the main surface increases, the number of individual pieces obtained by cutting it increases, thereby improving manufacturing efficiency. However, the environmental difference between the center and edges during firing generally becomes greater, making it difficult to improve quality stability, and therefore the effects of the present invention are significant. The upper limit of the size of the main surface is not particularly limited, but is generally 200 mm × 200 mm or less. The ratio of the long side to the short side of the main surface is not particularly limited. The thickness of the silicon nitride substrate of the present invention is not particularly limited, but is generally 0.1 mm to 2.0 mm.

[0025] As described above, the silicon nitride substrate of the present invention comprises silicon nitride particles and a glass phase derived from a sintering aid. The glass phase preferably contains yttrium and magnesium.

[0026] The silicon nitride substrate of the present invention can be produced, for example, by mixing silicon nitride powder, a sintering aid, and water for 20 hours or more, adding a binder and mixing for 20 hours or more to obtain a raw material slurry, and then molding the slurry to a density of 1.95 g / cm. 3 The above-described green body can be obtained by further firing the green body.

[0027] While the reason why the silicon nitride substrate of the present invention can be obtained by the above-described manufacturing method is unclear, the inventors believe it to be as follows. In a typical silicon nitride substrate manufacturing process, the surface of silicon nitride powder gradually oxidizes during processes such as raw material mixing, molding, and firing. Since a small amount of oxygen is required for sintering silicon nitride particles, if the oxide film on the surface of the silicon nitride powder is not uniform during the sintering reaction, the sintering reaction may proceed unevenly. For example, the sintering reaction proceeds relatively quickly in areas with sufficient oxygen, but relatively slowly in areas with insufficient oxygen. If the sintering reaction proceeds unevenly, the state of the silicon nitride particles will also be uneven. Furthermore, if the sintering reaction of silicon nitride proceeds unevenly, the reaction of the surrounding sintering aid will also be uneven, resulting in an uneven state of the glass phase. In silicon nitride substrates, the thermal expansion and contraction behavior is strongly influenced by the state of the silicon nitride particles, which are the main component, particularly the state of the Si-N skeleton, and the state of the glass phase surrounding the silicon nitride particles. Therefore, if the state of the silicon nitride particles and the glass phase is nonuniform, the average linear expansion coefficient will also be nonuniform. On the other hand, by mixing silicon nitride powder, a sintering aid, and water in a ball mill for 20 hours or more, and then adding a binder and mixing in a water-containing state for 20 hours or more, as described above, the oxidation of the silicon nitride powder surface is accelerated by the action of water, resulting in the formation of a sufficient oxide film over almost the entire surface of the silicon nitride powder in the green body. In particular, the binder is essential for molding silicon nitride powder, and in the present invention, it is particularly important to add the binder and continue mixing for another 20 hours or more. Mixing after adding the binder prevents the silicon nitride powder from being crushed by absorbing impact. If the silicon nitride powder is crushed, new surfaces will be generated, resulting in variations in the oxidation state of the silicon nitride powder surface. However, by adding the binder and mixing for 20 hours or more while preventing crushing, the silicon nitride powder surface can be sufficiently oxidized, resulting in uniform oxidation.Once a sufficient oxide film has formed on the surface of the silicon nitride powder, no further oxidation reaction occurs. Therefore, the progress of the oxidation reaction in subsequent steps becomes uniform within the silicon nitride substrate, resulting in a uniform glass phase state, and it is presumed that a silicon nitride substrate with an extremely uniform average linear expansion coefficient can be obtained.

[0028] In addition, the voids that inevitably occur in silicon nitride substrates affect the behavior of thermal expansion and contraction, causing the average linear expansion coefficient to become non-uniform. If a large amount of voids are generated during sintering, it is difficult to control the uniformity of the voids. However, in the above manufacturing method, the density of the green body is kept at 1.95 g / cm. 3 By setting the density of the green body to 2.00 g / cm or more, it is presumed that the generation of voids can be suppressed, the uniformity of the degree of void generation in the silicon nitride substrate can be increased, and a silicon nitride substrate with an extremely high uniformity of the average linear expansion coefficient can be obtained. 3 The upper limit of the density of the green body is not particularly limited, but is generally 2.50 g / cm 3 The density of the green body can be adjusted appropriately by known methods, for example, by adjusting the particle size and particle size distribution of the silicon nitride powder and sintering aid powder by pulverization or classification, thereby improving the packing property and thereby adjusting the density of the green body.

[0029] The time for mixing the silicon nitride powder, sintering aid, and water in the ball mill before adding the binder is not particularly limited as long as it is 20 hours or more, but a longer mixing time reduces production efficiency, so it is preferably 50 hours or less, and particularly 30 hours or less. The mixing time after adding the binder is also not particularly limited as long as it is 20 hours or more, but a longer mixing time reduces production efficiency, so it is preferably 50 hours or less, and particularly 30 hours or less.

[0030] The rotation speed of the ball mill during ball mill mixing is 10 rpm to 50 rpm, preferably 20 rpm to 30 rpm, which allows for sufficient oxidation while preventing excessive pulverization of the silicon nitride powder, which would reduce the uniformity of the surface oxidation state.

[0031] The silicon nitride powder may be composed of α-type silicon nitride, β-type silicon nitride, or a mixture thereof. However, β-type silicon nitride is preferred because β-type powder is less susceptible to grain growth during sintering than α-type powder and is easier to control uniformity. The β-phase ratio of the silicon nitride powder is preferably 80% or more, more preferably 90% or more, and even more preferably 99% or more. The β-phase ratio of the silicon nitride powder refers to the peak intensity ratio of the β-phase to the total of the α-phase and β-phase in the silicon nitride powder [100 × (peak intensity of β-phase) / (peak intensity of α-phase + peak intensity of β-phase)], and is determined by powder X-ray diffraction (XRD) measurement using CuKα radiation. More specifically, it can be determined by calculating the mass ratio of the α-phase and β-phase in the silicon nitride powder using the method described in C.P. Gazzara and D.R. Messier: Ceram. Bull., 56 (1977), 777-780.

[0032] The particle size of the silicon nitride powder is not particularly limited, but the average particle size D50 is preferably 0.4 to 1.2 μm, and more preferably 0.6 to 1.0 μm. The average particle size D50 can be measured using a laser diffraction / scattering particle size distribution analyzer. The specific surface area of ​​the silicon nitride powder is preferably 12 to 30 m. 2 / g, and 18 to 24m 2 The specific surface area means the BET specific surface area measured by the BET single-point method using nitrogen gas adsorption.

[0033] The sintering aid may be any known sintering aid without any particular limitation, and examples thereof include oxides such as yttria, magnesia, ceria, and calcia, as well as oxygen-free compounds such as carbonitride compounds and nitride compounds. Examples of carbonitride compounds include Y2Si4N6C, Yb2Si4N6C, Ce2Si4N6C, and MgSi4N6C. Examples of nitride compounds include MgSiN2. These sintering aids may be used alone or in combination of two or more. The amount of the sintering aid is not particularly limited, but is preferably 5 to 20 parts by mass, and more preferably 7 to 10 parts by mass, per 100 parts by mass of the silicon nitride powder.

[0034] The amount of water is not particularly limited, but from the viewpoint of thoroughly mixing and forming a sufficient oxide film on the surface of the silicon nitride powder, it is preferable to use an amount that results in a solids concentration in the slurry (slurry concentration) of 50 to 70 mass %.

[0035] The binder can be any known binder without any particular limitation, and examples thereof include polyvinyl alcohol, polyvinyl acetal, polyvinylpyrrolidone, acrylic resins, polyacrylamide, urethane resins, polyester, polyether, melamine, epoxy resins, cellulose resins, and starch. These binders may be used alone or in combination of two or more. The content of the binder is preferably 1 to 30 parts by mass per 100 parts by mass of the silicon nitride powder.

[0036] The raw material slurry may contain other components in addition to the silicon nitride powder, sintering aid, binder, and water. Examples of other components include dispersants, plasticizers, and antifoaming agents. The other components may be added before adding the binder, at the same time as adding the binder, or at any time after adding the binder.

[0037] For example, the dispersant is used to improve the dispersibility of the silicon nitride powder and sintering aid powder in the slurry, and generally, a surfactant can be suitably used. Known surfactants can be used without any restrictions. Specific examples of surfactants that can be suitably used in the present invention include carboxylated trioxyethylene tridecyl ether, diglycerol monooleate, diglycerol monostearate, carboxylated heptaoxyethylene tridecyl ether, tetraglycerol monooleate, hexaglycerol monooleate, sorbitan laurate, sorbitan oleate, sorbitan trioleate, polyoxyethylene sorbitan monolaurate, polyoxyethylene sorbitan monooleate, and polyoxyethylene sorbitan trioleate. These surfactants may be used alone or in combination of two or more. The amount of the dispersant can be appropriately selected, and can typically be selected from the range of 0.1 to 5 parts by mass per 100 parts by mass of the silicon nitride powder and the sintering aid powder combined. Within this range, the upper limit of the amount of dispersant is preferably 3 parts by mass or less, more preferably 2 parts by mass or less, and even more preferably 1 part by mass or less.

[0038] The raw material slurry can be molded into a desired shape to obtain a green body. A doctor blade method can be used as a molding method for easily obtaining a plate-shaped green body from the raw material slurry.

[0039] Density 1.95g / cm 3 To obtain the above-described green body, the raw material slurry is molded and then dried to remove the water contained in the molded body. Since water has a lower density than silicon nitride and sintering aids, the density of the green body cannot be improved if it contains water. If the green body is fired while it still contains water, the water evaporates during firing, creating voids, making it difficult to improve the uniformity of the void generation. The drying conditions are not particularly limited, but the molded body can be dried by heating it to approximately 30°C to 150°C.

[0040] In addition, degreasing is performed after drying to remove organic components (binders, dispersants, plasticizers, antifoaming agents, etc.) contained in the green body. Because organic components have a lower density than silicon nitride or sintering aids, the density of the green body cannot be improved if the organic components are present. If the green body is fired while still containing organic components, voids will be generated due to the evaporation or decomposition of the organic components during firing, making it difficult to improve the uniformity of the void generation rate. The degreasing conditions are not particularly limited, but may be performed, for example, by heating the green body to 450 to 650°C in air or an inert atmosphere such as nitrogen or argon.

[0041] The density thus obtained was 1.95 g / cm 3 The above green body can be fired to obtain a silicon nitride substrate. The firing conditions are not particularly limited, but the firing temperature is preferably 1700°C to 2000°C, and particularly preferably 1800°C to 1950°C, and the firing time is preferably 1 to 30 hours, and particularly preferably 3 to 20 hours. The firing pressure may be normal pressure or may be increased.

[0042] The silicon nitride substrate of the present invention can be cut into a plurality of pieces. Because the silicon nitride substrate of the present invention has a highly uniform average linear expansion coefficient, when cut into a plurality of pieces, it is easy to obtain pieces with the same thermal cycle resistance, and it can be said that the quality is stable. In other words, by using the silicon nitride substrate of the present invention, it is possible to efficiently produce pieces with excellent quality stability.

[0043] There are no particular limitations on the method for cutting the silicon nitride substrate, and it can be done by a slicer, laser scribing, etc. The number of pieces obtained by cutting one silicon nitride substrate is not particularly limited, but it is preferable to cut it into six or more pieces, and more preferably nine or more pieces, in which the pieces are different in size in the central region and the edge regions. [Example]

[0044] Examples will be described below to specifically explain the present invention, but the present invention is not limited to these examples. The measurements of the various items in the examples and comparative examples were performed by the following methods.

[0045] (1) Average linear expansion coefficient of silicon nitride substrate The average linear expansion coefficient of the silicon nitride substrate was measured using a push rod type dilatometer (JIS R 1618:2002 "Method for measuring thermal expansion by thermomechanical analysis of fine ceramics"). Specifically, a standard sample with a known expansion coefficient and an unknown sample to be measured were measured simultaneously to detect the difference in expansion (ΔL) between them when the temperature was raised from 25°C to 800°C. This was then divided by the product of the initial length of the unknown sample and the temperature rise width, and finally the expansion coefficient of the standard sample was added to calculate the expansion coefficient of the unknown sample, thereby determining the average linear expansion coefficient.

[0046] (2) Amount of dissolved oxygen in silicon nitride substrate The amount of dissolved oxygen in the silicon nitride substrate was measured by inert gas fusion-infrared absorption spectroscopy using an oxygen / nitrogen analyzer (HORIBA: EMGA-920). A 25 mg measurement sample was obtained from the silicon nitride substrate obtained in each Example and Comparative Example. The measurement sample was enclosed in a tin capsule (LECO Tin Cupsule was used for the tin capsule) and introduced into a graphite crucible. The sample was heated at 5.5 kW for 20 seconds to degas the adsorbed gas, then heated at 0.8 kW for 10 seconds. The temperature was then increased from 0.8 kW to 4 kW over 350 seconds. The amount of carbon monoxide generated during this period was measured and converted to carbon dioxide in the oxidizer. This was then converted to oxygen content. During the 350-second temperature increase, the oxygen generated initially corresponded to oxygen derived from oxides present on the surface of the silicon nitride particles (external oxygen), while the oxygen generated later corresponded to dissolved oxygen (internal oxygen) solid-dissolved in the silicon nitride crystals. Therefore, a perpendicular line was drawn from the valley of these two measurement peaks, after subtracting the previously measured background, to separate the two peaks. The amount of dissolved oxygen (internal oxygen) and the amount of external oxygen were calculated by proportionally allocating the respective peak areas.

[0047] (3) Density of silicon nitride substrate Using a test piece for measuring the average linear expansion coefficient, the average linear expansion coefficient was measured by the Archimedes method before measurement.

[0048] The silicon nitride substrate was manufactured using the following raw materials containing silicon nitride powder and a sintering aid. <Silicon nitride powder> S1: Silicon nitride powder (β-conversion rate: 99%, average particle size D50: 0.9 μm) ·S2: Silicon nitride powder (β conversion rate: 99%, average particle size D50: 1.8 μm) <Sintering aid> Yttria (manufactured by Shin-Etsu Chemical Co., Ltd.) Magnesia (manufactured by Ube Materials Co., Ltd.) <Binders and dispersants> Binder resin: Acrylic resin (manufactured by Fujikura Kasei Co., Ltd.) Dispersant: Cerna D735 (manufactured by Chukyo Yushi Co., Ltd.)

[0049] Example 1 100 parts by mass of silicon nitride powder S1, 5 parts by mass of yttria, 3 parts by mass of magnesia, and 0.5 parts by mass of dispersant were weighed out and mixed in a ball mill using water as a solvent, a resin pot, and silicon nitride balls at 25 rpm for 24 hours. Water was pre-weighed so that the slurry concentration was 60% by mass and added to the resin pot. 22 parts by mass of binder was then added, and mixing was continued for another 24 hours at 25 rpm. Next, degassing and viscosity adjustment were performed using a vacuum degassing machine (manufactured by Sayama Riken Co., Ltd.) to prepare a raw material slurry. The raw material slurry was then molded into a sheet with a width of 750 mm and a thickness of 420 μm using a doctor blade method. The obtained molded body was dried in air at 100°C to evaporate the solvent, and then cut to a size of 200 mm x 268 mm. The main surface was then degreased in dry air at 550°C to obtain a green body (green sheet). The density of the obtained green sheet was 2.01 g / cm 3The green sheet obtained as described above was placed in a firing vessel and fired at 1880°C for 9 hours in a nitrogen atmosphere under a pressure of 0.9 MPaG to obtain a silicon nitride substrate with a main surface measuring 150 mm x 200 mm. The evaluation results are shown in Table 1.

[0050] <Example 2> 100 parts by mass of silicon nitride powder S1, 5 parts by mass of yttria, 3 parts by mass of magnesia, and 0.5 parts by mass of dispersant were weighed out and mixed in a ball mill using water as a solvent, a resin pot, and silicon nitride balls at 25 rpm for 20 hours. Water was pre-weighed so that the slurry concentration was 60% by mass and added to the resin pot. 22 parts by mass of binder was then added, and the mixture was further mixed at 25 rpm for 24 hours. Next, degassing and viscosity adjustment were performed using a vacuum degassing machine (manufactured by Sayama Riken Co., Ltd.) to prepare a raw material slurry. The raw material slurry was then molded into a sheet with a width of 750 mm and a thickness of 420 μm using a doctor blade method. The obtained molded body was dried in air at 100°C to evaporate the solvent, and then cut to a size of 200 mm x 268 mm. The main surface was then degreased in dry air at 550°C to obtain a green body (green sheet). The density of the obtained green sheet was 2.05 g / cm 3 The green sheet obtained as described above was placed in a firing vessel and fired at 1880°C for 9 hours in a nitrogen atmosphere under a pressure of 0.9 MPaG to obtain a silicon nitride substrate with a main surface measuring 150 mm x 200 mm. The evaluation results are shown in Table 1.

[0051] <Comparative Example 1> 100 parts by mass of silicon nitride powder S1, 5 parts by mass of yttria, 3 parts by mass of magnesia, and 0.5 parts by mass of dispersant were weighed out and mixed in a ball mill using water as a solvent, a resin pot, and silicon nitride balls at 25 rpm for 24 hours. Water was pre-weighed so that the slurry concentration was 60% by mass and added to the resin pot. 22 parts by mass of binder was then added, and mixing was continued for another 12 hours at 25 rpm. Next, degassing and viscosity adjustment were performed using a vacuum degassing machine (manufactured by Sayama Riken Co., Ltd.) to prepare a raw material slurry. The raw material slurry was then molded into a sheet with a width of 750 mm and a thickness of 420 μm using a doctor blade method. The obtained molded body was dried in air at 100°C to evaporate the solvent, and then cut to a size of 200 mm x 268 mm. The main surface was then degreased in dry air at 550°C to obtain a green body (green sheet). The density of the obtained green sheet was 2.00 g / cm 3 The green sheet obtained as described above was placed in a firing vessel and fired at 1880°C for 9 hours in a nitrogen atmosphere under a pressure of 0.9 MPaG to obtain a silicon nitride substrate with a main surface measuring 150 mm x 200 mm. The evaluation results are shown in Table 1.

[0052] <Comparative Example 2> 100 parts by mass of silicon nitride powder S2, 5 parts by mass of yttria, 3 parts by mass of magnesia, and 0.5 parts by mass of dispersant were weighed out and mixed in a ball mill using water as a solvent, a resin pot, and silicon nitride balls at 25 rpm for 24 hours. Water was pre-weighed so that the slurry concentration was 60% by mass and added to the resin pot. 22 parts by mass of binder was then added, and mixing was continued for another 24 hours at 25 rpm. Next, degassing and viscosity adjustment were performed using a vacuum degassing machine (manufactured by Sayama Riken Co., Ltd.) to prepare a raw material slurry. The raw material slurry was then molded into a sheet with a width of 750 mm and a thickness of 420 μm using a doctor blade method. The obtained molded body was dried in air at 100°C to evaporate the solvent, and then cut to a size of 200 mm x 268 mm. The main surface was then degreased in dry air at 550°C to obtain a green body (green sheet). The density of the obtained green sheet was 1.85 g / cm 3The green sheet obtained as described above was placed in a firing vessel and fired at 1880°C for 9 hours in a nitrogen atmosphere under a pressure of 0.9 MPaG to obtain a silicon nitride substrate with a main surface measuring 150 mm x 200 mm. The evaluation results are shown in Table 1.

[0053] [Table 1]

Claims

1. A silicon nitride substrate, wherein the ratio αe / αc of the average linear expansion coefficient αc of the central portion to the average linear expansion coefficient αe of the edge portion is 0.90 to 1.

10.

2. The average linear expansion coefficient αc of the central portion is 2.8×10 －６ 2. The silicon nitride substrate of claim 1, wherein the surface roughness is 0.1 / K or more.

3. 3. The silicon nitride substrate according to claim 1, wherein the size of the main surface is 140 mm x 140 mm or more.

Images