Resin composition and semiconductor encapsulant

Abstract

To achieve low viscosity in a resin composition that comprises silica particles and is low in thermal expansion.SOLUTION: A resin composition comprises resin and inorganic filler, the inorganic filler comprising: first silica particles with a median diameter of 2 μm or more and 5 μm or less; second silica particles, which are wet silica, with a median diameter of 0.2 μm or more and less than 0.5 μm; and third silica particles, which are wet silica, with a median diameter of 0.1 μm or more and less than 0.2 μm.SELECTED DRAWING: None

Description

[Technical Field]

[0001] One aspect of the present invention relates to a resin composition, a semiconductor encapsulant, and a prediction device. [Background technology]

[0002] With the rapid progress in miniaturization, thinning, and high-density packaging of electronic devices, the gap between the element and the substrate is becoming narrower, and semiconductor encapsulation materials are required to have even lower thermal expansion and higher moldability.

[0003] Silica particles are blended as a filler in resin compositions used as semiconductor encapsulants. Blending silica particles into a resin composition can reduce the thermal expansion coefficient of the resin composition. However, blending a large amount of silica particles into a resin composition increases the viscosity of the resin composition, resulting in reduced moldability.

[0004] As an example of a resin composition containing silica particles, Patent Document 1 describes an epoxy resin composition containing 260 parts by mass or more of a filler per 100 parts by mass of solid content. Patent Document 1 also describes that by using, as fillers, first silica particles that are spherical silica particles having an average particle diameter of 0.1 μm or more and less than 1.0 μm, second silica particles that are spherical silica particles having an average particle diameter of 1.0 μm or more and 5.0 μm or less, and third silica particles that are spherical silica nanoparticles having an average particle diameter of 20 nm or more and 200 nm or less, an epoxy resin composition that is easy to mold can be obtained even when a large amount of filler is blended. [Prior art documents] [Patent documents]

[0005] [Patent Document 1] Japanese Patent Application Laid-Open No. 2014-88509 Summary of the Invention [Problem to be solved by the invention]

[0006] However, the epoxy resin composition described in Patent Document 1 is a composition in which multiple silica particles of a specific particle size are combined and blended to obtain a composition with desired properties, and it is difficult to obtain a composition with similar properties when silica particles of other particle sizes are used. Because there are a huge number of options for silica particle sizes, their combinations, and blending ratios, it is not easy to obtain a resin composition by selecting and blending silica particles to obtain the desired properties based on the description in Patent Document 1. In particular, there is a trade-off between achieving low thermal expansion by blending a high amount of silica particles and achieving a low viscosity resin composition, and it is not easy to achieve both.

[0007] An object of one aspect of the present invention is to achieve a low viscosity in a low thermal expansion resin composition containing silica particles. [Means for solving the problem]

[0008] In order to solve the above problems, a resin composition according to one aspect of the present invention is a resin composition including a resin and an inorganic filler, wherein the inorganic filler contains: First silica particles having a median diameter of 2 μm or more and 5 μm or less; second silica particles, which are wet silica having a median diameter of 0.2 μm or more and less than 0.5 μm; third silica particles, which are wet silica having a median diameter of 0.1 μm or more and less than 0.2 μm; Includes:

[0009] A semiconductor encapsulant according to an aspect of the present invention includes the resin composition according to an aspect of the present invention.

[0010] A prediction device according to one aspect of the present invention is a resin composition containing a resin and an inorganic filler, which satisfies required properties of the resin composition, and the inorganic filler includes: (I) a blending ratio of first silica particles having a median diameter of 2 μm or more and 5 μm or less; (II) a blending ratio of second silica particles which are wet-processed silica having a median diameter of 0.2 μm or more and less than 0.5 μm; (III) a blending ratio of third silica particles which are wet-processed silica having a median diameter of 0.1 μm or more and less than 0.2 μm; and (IV) a blending ratio of fourth silica particles which are wet-processed silica having a median diameter of 0.5 μm or more and less than 1.5 μm. The prediction device for predicting the required properties of a resin composition includes a data acquisition unit that acquires resin composition required property data indicating the required properties, and a recommended data derivation unit that derives recommended inorganic filler blending data indicating each of (I) the blending ratio of the first silica particles, (II) the blending ratio of the second silica particles, (III) the blending ratio of the third silica particles, and (IV) the blending ratio of the fourth silica particles in the inorganic filler, which satisfy the resin composition required property data by inputting the resin composition required property data into a prediction model. [Effects of the Invention]

[0011] According to one aspect of the present invention, a low viscosity can be achieved in a low thermal expansion resin composition containing silica particles. [Brief explanation of the drawings]

[0012] [Figure 1] 1 is a block diagram showing a configuration of a main part of a resin composition production system including a prediction device according to an embodiment of the present invention. [Figure 2] FIG. 2 is a diagram illustrating an overview of a first prediction model. [Figure 3] FIG. 10 is a diagram illustrating an overview of a second prediction model. [Figure 4] FIG. 10 is a diagram illustrating an overview of a recommended data derivation unit. [Figure 5] FIG. 2 is a diagram showing an example of inorganic filler blend input data and second input data. [Figure 6] FIG. 10 is a diagram showing an example of derivation of explanatory variables in the first algorithm executing unit. [Figure 7]FIG. 10 is a diagram illustrating an example of generation of a first prediction model in a second algorithm executing unit. [Figure 8] FIG. 2 is a diagram illustrating an example of input and output in a first prediction model. [Figure 9] FIG. 10 is a diagram illustrating an example of calculation in a second prediction model generated by a third algorithm executing unit. DETAILED DESCRIPTION OF THE INVENTION

[0013] The present inventors have conducted extensive research to find a way to achieve even lower viscosity in a low-thermal-expansion resin composition containing silica particles, and have found that it is possible to achieve a resin composition that satisfies both low thermal expansion and low viscosity by inputting characteristic data required for the resin composition (resin composition required characteristic data) and using a prediction model that derives recommended inorganic filler blending data that represents the blending ratio of each silica particle in an inorganic filler that satisfies the resin composition required characteristic data.

[0014] Below, we first describe the composition of a resin composition that satisfies both low thermal expansion and low viscosity, and then we describe a resin composition production system that generates a prediction model used to realize the resin composition and predicts the blending ratio of inorganic fillers using the generated prediction model.

[0015] [Resin composition] A resin composition according to one embodiment of the present invention includes a resin and an inorganic filler, the inorganic filler including first silica particles having a median diameter of 2 μm to 5 μm, second silica particles being wet-process silica having a median diameter of 0.2 μm to less than 0.5 μm, and third silica particles being wet-process silica having a median diameter of 0.1 μm to less than 0.2 μm. The resin composition may further include fourth silica particles being wet-process silica having a median diameter of 0.5 μm to less than 1.5 μm as an inorganic filler.

[0016] The resin composition according to one embodiment of the present invention contains silica particles as an inorganic filler, and therefore has low thermal expansion. Resin compositions used as semiconductor encapsulants preferably have low thermal expansion and low viscosity to improve moldability. The resin composition according to one embodiment of the present invention has low viscosity, and the viscosity at room temperature (20 to 30°C) is preferably 250 Pa·s or less, and more preferably 200 Pa·s or less. A semiconductor encapsulant containing the resin composition according to one embodiment of the present invention is also included in the scope of the present invention.

[0017] The resin composition may further contain a curing agent, an elastomer, an ion trapping agent, a pigment, a dye, an antifoaming agent, a stress relaxation agent, a pH adjuster, an accelerator, a surfactant, a coupling agent, and the like.

[0018] The resin composition can be obtained by mixing a resin with an inorganic filler. The inorganic filler to be mixed with the resin may be in the form of a dispersion in a solvent.

[0019] (resin) Examples of the resin contained in the resin composition include epoxy resin, acrylic resin, polyurethane resin, etc., and an example is epoxy resin. Examples of epoxy resins that can be used include bisphenol-type epoxy resins such as bisphenol A-type epoxy resins, bisphenol F-type epoxy resins, and bisphenol S-type epoxy resins; novolac-type epoxy resins such as phenol novolac-type epoxy resins and cresol novolac-type epoxy resins; biphenyl-type epoxy resins, xylylene-type epoxy resins, phenol aralkyl-type epoxy resins, biphenyl aralkyl-type epoxy resins, biphenyl dimethylene-type epoxy resins, trisphenolmethane novolac-type epoxy resins; glycidyl ethers; trifunctional or tetrafunctional glycidylamines; aryl alkylene-type epoxy resins such as tetramethylbiphenyl-type epoxy resins; naphthalene-skeleton-modified epoxy resins such as naphthalene-skeleton-modified cresol novolac-type epoxy resins, methoxynaphthalene-modified cresol novolac-type epoxy resins, and methoxynaphthalene dimethylene-type epoxy resins; anthracene-type epoxy resins; dicyclopentadiene-type epoxy resins; norbornene-type epoxy resins; fluorene-type epoxy resins; and flame-retardant epoxy resins obtained by halogenating the above-mentioned epoxy resins. One of these epoxy resins can be used alone, or two or more of them having different weight average molecular weights can be used in combination.

[0020] The resin may be contained in an amount of 10% by mass or more and 40% by mass or less, more preferably 10% by mass or more and 35% by mass or less, and even more preferably 15% by mass or more and 30% by mass or less, based on the entire resin composition.

[0021] (Inorganic filler) The inorganic filler includes first, second, and third silica particles. The inorganic filler may further include fourth silica particles. The median diameters of the first to fourth silica particles are determined by the particle diameter D at which the cumulative frequency reaches 50% in the particle size distribution determined by a laser diffraction / scattering method. 50The first to fourth silica particles may be spherical. The spherical silica particles can be produced by the methods for producing the respective silica particles described below.

[0022] The blending ratio of the inorganic filler in the resin composition is 60% by mass or more and 95% by mass or less, preferably 65% ​​by mass or more and 90% by mass or less, and more preferably 70% by mass or more and 85% by mass or less, based on the entire resin composition.

[0023] <First silica particles> The first silica particles are silica particles having a median diameter of 2 μm or more and 5 μm or less. The method for producing the first silica particles is not particularly limited, but the first silica particles can be produced by classifying a coarse silica powder obtained by a specific method using a specific classifier.

[0024] The method for producing the coarse silica powder used to produce the first silica particles can be, for example, a melting method, a wet method, a dry method, etc., and one example is the melting method. As the melting method, a known melting method can be used, and one example is the method described in JP 2015-86120 A.

[0025] In the melting method, a gaseous or liquid fuel is mixed with a combustion-supporting gas such as oxygen or air, and the resulting mixture is burned in a burner. The raw material, fumed silica, is fed into the flame using an ejector, a screw feeder, or a fluidized bed. The fed fumed silica melts to produce molten spherical silica, which is then introduced into a cyclone for solid-gas separation, yielding a coarse silica powder. The silica particles in the coarse silica powder thus obtained may be spherical.

[0026] The raw material fumed silica can be a silicon compound, particularly a silicon halide, generally a silicon chloride, usually produced by burning purified silicon tetrachloride in an oxyhydrogen flame, and can be obtained by a known method.

[0027] The fumed silica is preferably hydrophobized. The use of hydrophobized fumed silica improves dispersibility in flames, resulting in a spherical coarse silica powder with higher sphericity and a uniform particle size. The spherical coarse silica powder thus produced contains fewer fine silica particles with an extremely large specific surface area. The hydrophobization of the fumed silica is not particularly limited, and can be carried out by a conventionally known hydrophobization treatment. Regarding the hydrophobizing agent used, known treatment agents such as silylating agents can be used without any restrictions.

[0028] One example of a method for classifying the coarse silica powder obtained as described above is a method using a mechanism that classifies the powder by balancing the centrifugal force acting on the powder due to the rotation of the classification rotor with the drag force acting on the powder due to the airflow passing through the classification rotor in the axial direction. In this method, a dispersed airflow of coarse silica powder is supplied to a powder classifier with a resin-coated upper surface of the classification rotor, and coarse particles larger than 45 μm contained in the coarse silica powder are scattered and separated from above the classification rotor. The classification rotor may also be disk-shaped with an internal cavity that extends from the periphery to below the axial center, and suction is applied from the internal cavity, selectively drawing in fine particles smaller than 1 μm into the internal cavity to separate them.

[0029] <Second silica particles> The second silica particles are wet-process silica having a median diameter of 0.2 μm or more and less than 0.5 μm. The second silica particles can be produced by a known wet process. One example of the wet process is the sol-gel process described in republished patent WO2018 / 096876. The sol-gel process involves hydrolyzing and polycondensing silicon alkoxide in a reaction medium consisting of water and an organic solvent containing a catalyst to produce a silica sol, which is then gelled. The resulting solid is then isolated and dried to produce a silica powder. The silica powder produced by this method may be further calcined, if necessary.

[0030] In the sol-gel method described above, a reaction vessel is charged with water, a polar solvent other than water (organic solvent), and a basic catalyst, and then a silicon alkoxide (or an organic solvent solution of silicon alkoxide) and an aqueous solution of the basic catalyst are simultaneously added to the reaction vessel to cause a reaction. By appropriately setting the addition time of the silicon alkoxide and the basic catalyst and the reaction temperature, silica particles with a desired particle size distribution can be produced. Alternatively, the dispersion of silica particles produced by the sol-gel method may be wet-filtered to remove coarse particles.

[0031] Examples of silicon alkoxides include methyltrimethoxysilane, methyltriethoxysilane, tetramethoxysilane, tetraethoxysilane, tetraisopropoxysilane, and tetrabutoxysilane. Examples of basic catalysts include amine compounds and alkali metal hydroxides. Examples of organic solvents that are polar solvents other than water include alcohols such as methanol, ethanol, isopropyl alcohol, and butanol; ethers such as tetrahydrofuran and dioxane; and amide compounds such as dimethylformamide, dimethylacetamide, and N-methylpyrrolidone.

[0032] <Third Silica Particles> The third silica particles are wet-process silica having a median diameter of 0.1 μm or more and less than 0.2 μm. The third silica particles can be produced by a known wet process, similar to the second silica particles.

[0033] <Fourth Silica Particles> The fourth silica particles are wet-process silica having a median diameter of 0.5 μm or more and less than 1.5 μm. The fourth silica particles can be produced by a known wet process, similar to the second silica particles.

[0034] <Other silica particles> The inorganic filler may further contain silica particles other than the first to fourth silica particles. For example, the other silica particles contained in the inorganic filler may be silica particles having a median diameter of 1.5 μm or more and less than 2 μm. For example, the other silica particles contained in the inorganic filler may be silica particles having a median diameter of more than 5 μm.

[0035] (Surface treatment) The second silica particles, the third silica particles, and the fourth silica particles may be treated with a surface treatment agent. As the surface treatment agent, known silane coupling agents, siloxanes, silazanes, etc. can be used without any particular limitation. For example, the silane coupling agent can be methyltrimethoxysilane, dimethyldimethoxysilane, phenyltrimethoxysilane, hexyltrimethoxysilane, octyltrimethoxysilane, decyltrimethoxysilane, vinyltrimethoxysilane, 3-glycidoxypropyltrimethoxysilane, 3-mercaptopropyltrimethoxysilane, 3-methacryloxypropyltrimethoxysilane, 3-acryloxypropyltrimethoxysilane, 3-aminopropyltrimethoxysilane, etc. Examples of siloxanes that can be used include disiloxane, hexamethyldisiloxane, hexamethylcyclotrisiloxane, octamethylcyclotetrasiloxane, and decamethylcyclopentasiloxane, as well as polysiloxanes such as polydimethylsiloxane, and examples of silazanes that can be used include hexamethyldisilazane, etc. The surface treatment agent is preferably a silane coupling agent.

[0036] The method for treating the second, third, and fourth silica particles with a surface treatment agent is not particularly limited, but includes a method in which the silica particles and the surface treatment agent are stirred and mixed in a mixer and then heat-treated. Examples of the method for treating silica particles with a surface treatment agent include the methods described in Republished Patent Publication WO2019 / 044929 and JP2014-201454A.

[0037] The surface treatment with the surface treatment agent is preferably carried out after obtaining a dispersion of the second silica particles, the third silica particles, and the fourth silica particles by a sol-gel method and before wet filtration. This allows for the removal of aggregates and residual surface treatment agent that may result from the surface treatment. The method for adding the surface treatment agent to the mixer is not particularly limited. However, if the surface treatment agent is a low-viscosity liquid at room temperature and normal pressure, it may be added as is. If the surface treatment agent is a high-viscosity liquid or solid at room temperature and normal pressure, it may be added to an appropriate organic solvent to form a solution or dispersion, which may then be added to the mixer. The amount of surface treatment agent used to surface treat the silica particles may be appropriately determined depending on the application, etc.

[0038] (mixing ratio) The mixing ratio a of the first silica particles in the inorganic filler is 50% by mass or more and 80% by mass or less, preferably 55% by mass or more and 80% by mass or less, and more preferably 60% by mass or more and 75% by mass or less, relative to the total amount of the inorganic filler.

[0039] The blending ratio c of the second silica particles in the inorganic filler is 5% by mass or more and 20% by mass or less, and preferably 5% by mass or more and 15% by mass or less, based on the total amount of the inorganic filler.

[0040] The blending ratio d of the third silica particles in the inorganic filler is 5% by mass or more and 30% by mass or less, and preferably 10% by mass or more and 25% by mass or less, based on the total amount of the inorganic filler.

[0041] The blending ratio b of the fourth silica particles in the inorganic filler is 0% by mass or more and 15% by mass or less, and preferably 0% by mass or more and 10% by mass or less, based on the total amount of the inorganic filler.

[0042] The blending ratio of each silica particle in the inorganic filler may be selected so as to satisfy the properties required for the resin composition. For example, the blending ratio of each silica particle in the inorganic filler is selected so as to satisfy the viscosity required for the resin composition. Preferably, the blending ratio of each silica particle in the inorganic filler is selected so that the viscosity of the resin composition becomes 250 Ps·s or less.

[0043] The blending ratio of each silica particle in the inorganic filler that makes the viscosity of the resin composition 250 Ps s or less is determined by the following formula (1), where the blending ratio of (i) the second silica particles in the inorganic filler is c, the blending ratio of (ii) the third silica particles is d, and the blending ratio of (iii) the fourth silica particles is b. -0.078×b -0.105×c -0.011×d -0.0159×b 2 -0.0217×c 2 -0.0130×d 2 -0.0393×b×c-0.0257×c×d+0.0218×d×b+25.498≧0 (1) Examples 1 to 5 described below are resin compositions in which silica particles are blended in a blending ratio that satisfies formula (1).

[0044] In addition, when the blending ratio of (i) the second silica particles in the inorganic filler is c, the blending ratio of (ii) the third silica particles is d, and the blending ratio of (iii) the fourth silica particles is b, the following formula (2) can be obtained: -0.205×b+0.125×c+0.244×d-0.0151×b 2 -0.0102×c 2 -0.0079×d 2 -0.0154×b×c-0.0104×c×d+0.0252×d×b+0.964≧0 (2) By using a blending ratio that satisfies formula (2), it is possible to obtain a resin composition with a viscosity of 250 Ps·s or less with a higher probability than with a blending ratio that satisfies formula (1). Examples 3 to 5 described below are resin compositions in which each silica particle is blended at a blending ratio that satisfies formula (2).

[0045] The mixing ratio of each silica particle in the inorganic filler may be calculated based on the above formula (1) or (2), and then the mixing ratio may be adjusted depending on the type and content of the resin contained in the resin composition.

[0046] [Prediction device for predicting the blending ratio of each silica particle in an inorganic filler] A prediction device according to one embodiment of the present invention is a resin composition containing a resin and an inorganic filler, which satisfies the required properties of the resin composition, and the inorganic filler contains: (I) a blending ratio of first silica particles having a median diameter of 2 μm or more and 5 μm or less; (II) a blending ratio of second silica particles which are wet-processed silica having a median diameter of 0.2 μm or more and less than 0.5 μm; (III) a blending ratio of third silica particles which are wet-processed silica having a median diameter of 0.1 μm or more and less than 0.2 μm; and (IV) a blending ratio of fourth silica particles which are wet-processed silica having a median diameter of 0.5 μm or more and less than 1.5 μm. The prediction device predicts each of the blending ratios, and includes a data acquisition unit that acquires resin composition required property data indicating required properties, and a recommended data derivation unit that derives recommended inorganic filler blending data indicating each of the blending ratios of (I) first silica particles, (II) second silica particles, (III) third silica particles, and (IV) fourth silica particles in the inorganic filler that satisfy the resin composition required property data by inputting the resin composition required property data into a prediction model.

[0047] (Resin composition manufacturing system 1) A resin composition production system 1 equipped with a prediction device according to one embodiment of the present invention will be described below. For the sake of brevity, descriptions of matters similar to those of known techniques will be omitted as appropriate. Please note that the configurations and values ​​described in this specification are merely examples unless otherwise specified.

[0048] FIG. 1 is a block diagram showing the configuration of a main part of a resin composition production system 1. The resin composition production system 1 includes a model generation device 100 and a prediction device 300. As described below, the model generation device 100 generates a prediction model for predicting required property fulfillment conditions for a resin composition containing one or more inorganic fillers and one or more resins. Specifically, the model generation device 100 generates a second prediction model (MODEL2) described below as the prediction model. Then, the prediction device 300 predicts the required property fulfillment conditions using MODEL2.

[0049] The prediction device 300 includes a third input data acquisition unit (data acquisition unit) 33, a recommended data derivation unit 34, and an output unit 35. The prediction device 300 derives recommended inorganic filler blending data indicating the blending ratios of the first to fourth silica particles in the inorganic filler that satisfy the resin composition required characteristics that indicate the required characteristics of the resin composition. Here, the required resin composition characteristics are, for example, the viscosity of the resin composition. That is, the prediction device 300 can derive the blending ratios of the first to fourth silica particles in the inorganic filler to obtain a resin composition with a desired viscosity. When a desired low viscosity is set as the resin composition required characteristic data, the prediction device 300 derives the blending ratio that achieves this using a prediction model. Then, by blending the first to fourth silica particles in the derived blending ratio to produce a resin composition, an even lower viscosity can be achieved in a low-thermal-expansion resin composition containing silica particles. Details of the prediction device 300 will be described later.

[0050] In the following description, one or more silica particles will be collectively referred to as inorganic filler A, and one or more resins will be collectively referred to as resin B. A resin composition containing inorganic filler A and resin B will be collectively referred to as resin composition C.

[0051] It should be noted that the "required property satisfying conditions" in this specification are not limited to "conditions that completely satisfy the required properties of resin composition C." The "required property satisfying conditions" in this specification also include, for example, "required properties that are closest to the complete required properties of resin composition C" and "required properties that are somewhat close to the complete required properties of resin composition C." Therefore, it is sufficient for the resin composition production system 1 to predict, as the required property satisfying conditions, conditions that generally satisfy the complete required properties of resin C. For this reason, for example, as described below, the resin composition production system 1 may predict the required property satisfying conditions based on a probabilistic mathematical model.

[0052] The processing of the resin composition production system 1 is roughly divided into a learning phase (processing in the model generating device 100) and a prediction phase (processing in the prediction device 300). The prediction phase is also called an inference phase. First, the learning phase will be described.

[0053] (Model generation device 100) The model generation device 100 generates a prediction model to be used in the prediction device 300. The model generation device 100 includes a first input data acquisition unit 11, a second input data acquisition unit 12, a first machine learning unit 21, and a second machine learning unit 22.

[0054] The first input data acquiring unit 11 acquires first input data 110. The first input data 110 is input data including inorganic filler blend input data 111. The inorganic filler blend input data 111 is data relating to the blending ratio of inorganic filler A in resin composition C in which multiple types of inorganic filler A are blended with resin B. As an example, the inorganic filler blend input data 111 is data indicating the blending ratio of inorganic filler A.

[0055] The first input data acquiring unit 11 may be any data acquisition interface. As an example, the first input data acquiring unit 11 may be an input unit that accepts a user's input operation. In this case, the first input data acquiring unit 11 acquires first input data 110 input by the user. As another example, the first input data acquiring unit 11 may acquire first input data 110 pre-stored in a storage unit (not shown) within the model generation device 100. As yet another example, the first input data acquiring unit 11 may communicate with an external device (not shown) of the model generation device 100 and acquire the first input data 110 from the external device. Examples of the external device include a storage server and a measurement device. The description of the first input data acquiring unit 11 also applies to the second input data acquiring unit 12 described below and the third input data acquiring unit 33 described below.

[0056] The second input data acquisition unit 12 acquires second input data 120. Specifically, the second input data acquisition unit 12 acquires resin composition characteristic input data as the second input data 120. The resin composition characteristic input data indicates the characteristics of the resin composition C. From this, it can be said that the second input data 120 is data that forms a pair with the first input data 110.

[0057] The first machine learning unit 21 acquires first input data 110 from the first input data acquisition unit 11 and second input data 120 from the second input data acquisition unit 12. The first machine learning unit 21 generates a first prediction model (hereinafter referred to as MODEL1) based on the first input data 110 and the second input data 120.

[0058] Fig. 2 is a diagram illustrating an overview of MODEL 1. As shown in Fig. 2, MODEL 1 is a model (mathematical model) for predicting unknown resin composition property data 1200 from arbitrary inorganic filler blend data 1110. In this specification, a data set including arbitrary inorganic filler blend data 1110 is referred to as an arbitrary input data set 1100.

[0059] In the example of FIG. 2, an arbitrary input data set 1100 is an example of an explanatory variable (X). The unknown resin composition property data 1200 is an example of a response variable (y). The explanatory variable is also called an independent variable. In contrast, the response variable is also called a dependent variable or an explained variable. In the example of FIG. 2, MODEL1 can be expressed as a function f that indicates the relationship y=f(X). In this way, MODEL1 is a model for solving a forward problem (a model for deriving y from X).

[0060] In this specification, it is assumed that the type of data included in the arbitrary input dataset 1100 matches the type of training data used to generate MODEL 1. That is, it is assumed that the data structure of the arbitrary input dataset 1100 matches the data structure of the training data used to generate MODEL 1. Therefore, for example, the data structure of the arbitrary input dataset 1100 matches the data structure of the first input data 110.

[0061] The second machine learning unit 22 acquires MODEL1 from the first machine learning unit 21. The second machine learning unit 22 generates a second prediction model (hereinafter referred to as MODEL2) based on MODEL1.

[0062] FIG. 3 is a diagram illustrating an overview of MODEL2. As shown in FIG. 3, MODEL2 is a model for predicting predicted inorganic filler blend data 2310 that satisfies arbitrary resin composition characteristic data 2200. In this specification, the phrase "satisfying arbitrary resin composition characteristic data 2200" means "satisfying the required properties of resin composition C indicated by the arbitrary resin composition characteristic data 2200." In this specification, a dataset including predicted inorganic filler blend data 10 is referred to as a predicted dataset 2300.

[0063] In the example of FIG. 3, arbitrary resin composition property data 2200 is an example of a response variable (y). And, prediction data set 2300 is an example of an explanatory variable (X). In the example of FIG. 3, MODEL2 can be expressed as a function g that indicates the relationship X=g(y). Note that g≒f -1 In other words, function g is an approximate inverse function of function f. In this way, MODEL2 is a model for solving inverse problems (a model for deriving X from y). As described above, MODEL2 is a model that pairs with MODEL1.

[0064] In this specification, the type of the arbitrary resin composition property data 2200 is assumed to match the type of the training data used to generate MODEL 2. That is, the data structure of the arbitrary resin composition property data 2200 is assumed to match the data structure of the training data used to generate MODEL 2. Therefore, for example, the data structure of the arbitrary resin composition property data 2200 is assumed to match the data structure of the second input data 120.

[0065] 1, the first machine learning unit 21 has a first algorithm execution unit 211 and a second algorithm execution unit 212. The first algorithm execution unit 211 acquires first input data 110 from the first input data acquisition unit 11. The first algorithm execution unit 211 executes a first algorithm that derives an explanatory variable (X) corresponding to data indicating the characteristics of the resin composition C obtained from the second input data 120. An example of the first algorithm will be described later.

[0066] The second algorithm execution unit 212 acquires the explanatory variable (X) from the first algorithm execution unit 211, and acquires the second input data 120 from the second input data acquisition unit 12. The second algorithm execution unit 212 executes a second algorithm that generates MODEL1 based on X and the second input data 120.

[0067] The second algorithm may be any algorithm capable of generating MODEL1 based on X and the second input data 120. In other words, the second algorithm may be any algorithm capable of generating a model for solving a forward problem. As an example, the second algorithm may be Gaussian process regression, Support Vector Machines, Linear regression, Decision trees, Random forests, Neural networks, and Gradient boosted trees, At least one of the following is true:

[0068] 1, the second machine learning unit 22 has a third algorithm execution unit 223. The third algorithm execution unit 223 executes a third algorithm that generates MODEL2 based on MODEL1.

[0069] The third algorithm may be any algorithm capable of generating MODEL2 based on MODEL1. In other words, the third algorithm may be any algorithm capable of generating a model for solving an inverse problem. As an example, the third algorithm may be: Genetic algorithms, · Gradient descent method, Grid search, and Bayesian optimization, At least one of the following is true:

[0070] In one embodiment of the present invention, the resin composition characteristic input data (second input data 120) may be data indicating any characteristic of the resin composition C. As an example, the resin composition characteristic input data may be data indicating any characteristic of the resin composition C, Viscosity, fluidity, moldability, adhesiveness, transparency, color tone, strength, water absorption, linear expansion coefficient, elastic modulus, yield stress, tensile strength, fracture toughness, electrical conductivity, dielectric constant, dielectric loss tangent, thermal conductivity, and stability. At least one of the above is shown as a characteristic of the resin composition C. The resin composition characteristic input data may be the viscosity of the resin composition C. Note that the explanation regarding the resin composition characteristic input data also applies to each data corresponding to the resin composition characteristic input data.

[0071] (Prediction device 300) Next, the prediction phase will be described. The prediction device 300 includes a third input data acquisition unit (data acquisition unit) 33, a recommended data derivation unit 34, and an output unit 35.

[0072] The third input data acquisition unit 33 acquires third input data 330. Specifically, the third input data acquisition unit 33 acquires resin composition required property data as the third input data 330. The resin composition required property data indicates required properties of the resin composition C.

[0073] The recommended data derivation unit 34 acquires resin composition required property data (third input data 330) from the third input data acquisition unit 33, and also acquires MODEL2 from the model generation device 100 (more specifically, the second machine learning unit 22).

[0074] Fig. 4 is a diagram illustrating an overview of the recommended data derivation unit 34. As shown in Fig. 4, the recommended data derivation unit 34 derives recommended data 340 by inputting resin composition required property data into MODEL 2. The recommended data 340 includes recommended inorganic filler blend data 341.

[0075] The recommended inorganic filler blending data 341 indicates a blending of inorganic filler A that satisfies the resin composition required property data. In this specification, the phrase "satisfies the resin composition required property data" means "satisfies the required properties of resin composition C indicated by the resin composition required property data." A specific example of the process of deriving the recommended data 340 by the recommended data derivation unit 34 will be described later.

[0076] As described above, the "conditions for satisfying required properties" in this specification are not limited to "conditions for completely satisfying the required properties of resin composition C." Naturally, therefore, the recommended data 340 in this specification are not limited to "data that completely satisfy the required properties data of the resin composition." The recommended data 340 in this specification also includes "data that generally satisfy the required properties data of the resin composition."

[0077] Therefore, the recommended inorganic filler blending data 341 in this specification may be any blending of inorganic filler A that generally satisfies the resin composition required property data. Note that these explanations regarding the recommended inorganic filler blending data 341 also apply to the above-mentioned predicted inorganic filler blending data 2310.

[0078] The output unit 35 acquires the recommended data 340 from the recommended data derivation unit 34. Then, the output unit 35 outputs the recommended data 340. The output unit 35 may be any output interface. As an example, the output unit 35 may be a display (display device). In this case, the output unit 35 can visually present the recommended data 340 to the user by displaying the recommended data 340. In this way, the output unit 35 may output the recommended data 340 in a visual form. As another example, the output unit 35 may transfer the recommended data 340 to a storage unit within the model generation device 100. As yet another example, the output unit 35 may transfer the recommended data 340 to a device external to the model generation device 100.

[0079] In one embodiment of the present invention, the resin composition required property data (third input data 330) may be data indicating any required property of the resin composition C. As is clear from the above description of the resin composition property input data, for example, the resin composition required property data may be data indicating the desired property of the resin composition C, Viscosity, fluidity, moldability, adhesiveness, transparency, color tone, strength, water absorption, linear expansion coefficient, elastic modulus, yield stress, tensile strength, fracture toughness, electrical conductivity, dielectric constant, dielectric loss tangent, thermal conductivity, and stability. At least one of the above is shown as the required property of the resin composition C. As an example, the required property data of the resin composition is the viscosity of the resin composition C.

[0080] (Example of processing in resin composition production system 1) The following describes an example of processing in the resin composition production system 1. In the following example, a case will be described in which there are five types of inorganic filler A and one type of resin B. In the following example, it is assumed that the properties of resin B and the blending ratio of resin B are both set to certain fixed values ​​(fixed conditions).

[0081] Therefore, in the following examples, the properties and blending ratio of resin B are not considered as explanatory variables. Therefore, in the following explanation, the resin property input data 112 and the resin blend input data 114 are not mentioned. For this reason, in the following, the blending ratio of inorganic filler A will also be referred to simply as blending ratio. In the following examples, the viscosity (unit: Pa s) of resin composition C will be exemplified as a property of resin composition C. In addition, the viscosity of resin composition C will also be referred to simply as viscosity.

[0082] (Examples of the first input data 110 and the second input data 120) FIG. 5 is a diagram showing an example of inorganic filler blending input data 111 and second input data 120. The inorganic filler blending input data 111 in the example of FIG. 5 is data showing the blending ratio of five types of inorganic fillers in weight percent (wt%). In the following description, the five types of inorganic fillers will be referred to as inorganic filler 0 to inorganic filler 4, respectively. The blending ratio of inorganic filler i will be referred to as x0i, where i is an integer satisfying 0≦i≦4. For example, x01 represents the blending ratio of inorganic filler 1.

[0083] Here, as will be apparent to those skilled in the art, x00+x01+x02+x03+x04=100 Therefore, x00 is x00=100-x01-x02-x03-x04 As shown above, it is uniquely determined according to preset x01 to x04. Therefore, in the inorganic filler blend input data 111 in the example of Fig. 5, only x01 to x04 are set in order to reduce the number of dimensions of the explanatory variables.

[0084] In the inorganic filler blend input data 111, a plurality of blend ratio patterns (combination patterns of x01 to x04) are set. id in FIG. 5 is the identification number of the blend ratio pattern. For example, id=1 indicates the first blend ratio pattern (hereinafter also referred to as the first blend ratio pattern). In the example of FIG. 5, the first blend ratio pattern is "x00=60, x01=10, x02=30, x03=0, x04=0". In the following, for example, id=1 will be abbreviated as id1 as appropriate.

[0085] The second input data 120 in the example of FIG. 5 indicates the viscosity (y01) of resin composition C corresponding to each blending ratio pattern. Specifically, the viscosity value actually measured for each blending ratio pattern is recorded as y01 in the second input data 120. Prior to the computer simulation in this example, the inventors of the present application (hereinafter simply referred to as "the inventors") produced resin composition C according to the first blending ratio pattern described above. Then, when the inventors actually measured the viscosity of resin composition C, a value of 455 was obtained. Therefore, in the second input data 120 in the example of FIG. 5, a value of y01=455 is set for id1.

[0086] In the example of FIG. 5, a data set showing the correspondence between each compounding ratio pattern and y01 is created for each id. In the example of FIG. 5, the data set corresponding to the jth id (idj) is referred to as DATASET_idj. As an example, DATASET_id1 is a data set showing the correspondence between the first compounding ratio pattern and y01. Hereinafter, the jth compounding ratio pattern will also be referred to as the jth compounding ratio pattern.

[0087] (Example of Derivation of Explanatory Variables in the First Algorithm Executing Unit 211) 6 is a diagram showing an example of derivation of explanatory variables in the first algorithm executing unit 211. The first algorithm executing unit 211 derives an explanatory variable (X) corresponding to data indicating the characteristics of the resin composition C obtained from the first input data 110. In this example, the first algorithm executing unit 211 derives an explanatory variable corresponding to data indicating the viscosity of the resin composition C obtained from the inorganic filler blending input data 111. In this way, in this example, the explanatory variable for explaining the viscosity, which is the objective variable, is derived as X.

[0088] Specifically, the first algorithm executing unit 211 executes the first algorithm to derive X based on the inorganic filler blend input data 111. In the example of Fig. 6, the first machine learning unit 21 executes weighted average calculation and principal component analysis as the first algorithm.

[0089] In the example of FIG. 6, the first machine learning unit 21 derives a particle size distribution-related vector (a vector having xx01 to xx05 in FIG. 6 as components) as an explanatory variable by performing principal component analysis on the inorganic filler blend input data 111. The first machine learning unit 21 may derive the explanatory variable by, for example, weighting the inorganic filler blend input data 111 based on the characteristic data of the inorganic filler and performing principal component analysis on the inorganic filler blend input data after the weighted average. One example of the characteristic data of the inorganic filler may be the particle size distribution of the inorganic filler. The weight values ​​used in the weighted average calculation may be set by a known method based on, for example, the characteristic data of the inorganic filler. The number of dimensions reduced by the principal component analysis may be set arbitrarily. The first machine learning unit 21 calculates a particle size distribution-related vector for each ID. Therefore, for example, as shown in FIG. 7, the explanatory variables derived by the first machine learning unit 21 include the first particle size distribution-related vector (particle size distribution-related vector corresponding to ID1) described below.

[0090] In this specification, the particle size distribution-related vector corresponding to the jth blending ratio pattern (in other words, the particle size distribution-related vector corresponding to idj) is referred to as the jth particle size distribution-related vector. The jth particle size distribution-related vector is calculated from the particle size distribution of inorganic filler 0 to inorganic filler 4 in resin composition C when the jth blending ratio pattern is applied. Therefore, for example, the first particle size distribution-related vector is calculated from the particle size distribution of inorganic filler 0 to inorganic filler 4 in resin composition C when the above-mentioned first blending ratio pattern is applied.

[0091] (Example of generation of MODEL1 in the second algorithm execution unit 212) 7 is a diagram showing an example of generation of MODEL1 in the second algorithm executing unit 212. The second algorithm executing unit 212 generates MODEL1 based on (i) the explanatory variable (X) derived by the first algorithm executing unit 211 and (ii) the second input data 120. Specifically, the second algorithm executing unit 212 generates MODEL1 based on X and the second input data 120 by executing the second algorithm.

[0092] In the example of FIG. 7, the second algorithm executing unit 212 executes a neural network as the second algorithm. Specifically, for each id, the second algorithm executing unit 212 acquires an objective variable corresponding to X from the second input data 120. For example, the second algorithm executing unit 212 acquires the viscosity (y01) shown in DATASET_id1 as the correct answer data for the objective variable (y) corresponding to X for id1. Then, the second algorithm executing unit 212 executes the neural network using the correct answer data to derive a function f that satisfies the relationship between X and y for each id. In this way, the second algorithm executing unit 212 generates MODEL1 as a function f that indicates the relationship y=f(X).

[0093] The following describes an example in which MODEL1 is generated by the second algorithm execution unit 212 performing ensemble learning using the bagging method. Therefore, MODEL1 in this example includes multiple neural networks (weak learners) with different hyperparameters. In this way, MODEL1 in this example is generated as a strong learner that integrates multiple weak learners.

[0094] (Example of input and output for MODEL1) FIG. 8 is a diagram showing an example of input and output in MODEL1. In FIG. 8, the above-mentioned first particle-size distribution-attributed vector (particle-size distribution-attributed vector corresponding to id1) is exemplified as X. As shown in FIG. 8, by inputting X to MODEL1, a histogram (y_Hist) showing the distribution of y can be obtained. Specifically, by inputting X to each of a plurality of weak learners in MODEL1, a plurality of y's output from the plurality of weak learners can be integrated to obtain y_Hist.

[0095] In this example, preprocessing of the correct data is performed prior to generating MODEL1 using the second algorithm described above. Specifically, in this example, logarithmic transformation of the correct data is performed prior to generating MODEL1. Therefore, strictly speaking, MODEL1 is generated as a model that outputs log(y). Therefore, the horizontal axis of y_Hist in the example of FIG. 8 is log(y). However, for simplicity, this specification will be described assuming that MODEL1 is a model that outputs y. Furthermore, y_Hist will be described as a histogram showing the distribution of y.

[0096] Then, MODEL1 in this example determines predetermined data based on y_Hist as a final predicted value (a predicted value as a strong learner) and outputs the final predicted value. Specifically, MODEL1 in this example outputs the average value (μ) of y_Hist as the final predicted value (y). μ is also called an expected value.

[0097] 2, MODEL1 generated as a strong learner can acquire an arbitrary input data set 1100 as an explanatory variable (X) and output unknown resin composition property data 1200 as a response variable (y). For example, by inputting the first blending ratio pattern described above into MODEL1 as X, a predicted value of viscosity corresponding to the first blending ratio pattern can be output as y.

[0098] In addition, MODEL1 uses the variance of y_Hist (σ 2 ) may be output together with μ. The variance is an example of an index of uncertainty in the predicted value of MODEL1. Alternatively, MODEL1 may output the standard deviation (σ) of y_Hist together with μ. The standard deviation is another example of an index of uncertainty in the predicted value of MODEL1.

[0099] (Example of calculation in MODEL2 generated by the third algorithm execution unit 223) 9 is a diagram showing an example of calculation in MODEL2 generated by the third algorithm executing unit 223. The third algorithm executing unit 223 generates MODEL2 based on MODEL1 by executing the third algorithm. In other words, the third algorithm executing unit 223 determines the above-mentioned function g (an approximate inverse function of the function f) based on the function f determined in advance by the second algorithm executing unit 212 (see also FIG. 3 above).

[0100] 9, the third algorithm executing unit 223 uses grid search as the third algorithm to generate MODEL 2. Within the generated MODEL 2, calculations in two stages (first and second stages) described below are executed.

[0101] First, in the first stage, MODEL2 uses MODEL1 to calculate a predicted viscosity value (μ) for each of a plurality of possible blending ratio patterns (combinations of x01 to x04). Specifically, MODEL2 inputs explanatory variables (X) corresponding to each blending ratio pattern into MODEL1, causing MODEL1 to output μ as a response variable. In this example, MODEL1 outputs σ in addition to μ for each blending ratio pattern.

[0102] Next, in the second stage, MODEL2 outputs a blending ratio pattern (X) that maximizes the probability of obtaining any viscosity data input to MODEL2 based on μ and σ calculated in the first stage (see also Figure 3 above).

[0103] 9, MODEL2 predicts a predicted data set 2300 that has the highest probability of obtaining any resin composition characteristic data 2200. The example of FIG. 9 illustrates a case where the any resin composition characteristic data 2200 is data indicating a viscosity of "μ=290 to 310."

[0104] The above numerical range of μ = 290 to 310 is an example of a numerical range set assuming that resin composition C is used as an adhesive. For example, if the viscosity of the adhesive is too high, it is difficult to shape the adhesive into a desired shape for use. On the other hand, if the viscosity of the adhesive is too low, the adhesive is likely to drip. Therefore, it is believed that there is a suitable numerical range for the viscosity of the adhesive. The above numerical range of μ = 290 to 310 is an example of this suitable numerical range.

[0105] In the example of FIG. 9, MODEL2 selects the mixture ratio pattern with the highest "probability that μ falls within the target range of 290 to 310" (hereinafter referred to as "probability for the target range") as the optimal mixture ratio pattern. MODEL2 then outputs this optimal mixture ratio pattern as the prediction result (i.e., prediction data set 2300). In the example of FIG. 9, MODEL2 calculates the probability for the target range for all combinations of multiple possible mixture ratio patterns. Specifically, in the example of FIG. 9, MODEL2 calculates the probability for the target range under the assumption that μ follows a normal distribution.

[0106] In the example of Fig. 9, a data set showing the correspondence between "each blending ratio pattern," "μ and σ," and "probability relative to the target range" is created for each id. In the example of Fig. 9, the data set corresponding to the jth id (idj) is called DATASET2_idj.

[0107] In the example of Figure 9, MODEL2 searches for a dataset with the maximum probability for the target range for all j (hereinafter referred to as the maximum probability dataset). In the example of Figure 9, as a result of the search by MODEL2, DATASET2_id1 was found to be the maximum probability dataset. Note that the first blending ratio pattern in the example of Figure 9 is a blending ratio pattern of "x00=65, x01=5, x02=20, x03=0, x04=10".

[0108] MODEL2 selects the maximum probability dataset as the optimal dataset (DATASET2_OPT). Then, MODEL2 determines the combination ratio pattern corresponding to the optimal dataset as the optimal combination ratio pattern. In the example of FIG. 9, MODEL2 selects DATASET2_id1 as the optimal dataset. Then, the third algorithm execution unit 223 determines the above-mentioned first combination ratio pattern corresponding to DATASET2_id1 as the optimal combination ratio pattern. Thus, in the example of FIG. 9, the first combination ratio pattern is output as the prediction result.

[0109] As described above, the third algorithm executing unit 223 generates MODEL2 (a model that predicts a prediction data set 2300 that satisfies any resin composition characteristic data 2200) by using the third algorithm (e.g., grid search). However, as will be clear to those skilled in the art, the method for determining the optimal blending ratio pattern is not limited to the above example.

[0110] For example, MODEL2 may search for a blending ratio pattern corresponding to a data set (hereinafter referred to as a most recent predicted value data set) that has a predicted value (μ) closest to 300 (the median value of the above-mentioned viscosity range) among all blending ratio patterns in the example of Fig. 9. Then, MODEL2 may determine the blending ratio pattern corresponding to the most recent predicted value data set as the optimal blending ratio pattern.

[0111] (Example of derivation of recommended data by the recommended data derivation unit 34) By using MODEL2 generated as described above, the recommended data derivation unit 34 can derive recommended data 340 based on the third input data 330 (resin composition required property data) (see also FIG. 4 above).

[0112] As an example, the third input data 330 may be data indicating a viscosity of "μ=350-370." In this case, the recommended data derivation unit 34 can input the third input data 330 to MODEL2, thereby deriving recommended data 340 corresponding to the third input data 330. In this example, the recommended data 340 is, for example, an optimal blending ratio pattern corresponding to a viscosity of μ=350-370.

[0113] (Specific examples of recommended data derived by the recommended data derivation unit 34) By using MODEL2 generated as described above, the recommended data derivation unit 34 can derive the blending ratio of (I) the first silica particles, (II) the second silica particles, (III) the third silica particles, and (IV) the fourth silica particles in the inorganic filler A as recommended inorganic filler blending data that satisfies the required properties for the resin composition C.

[0114] As an example, the recommended data derivation unit 34 derives inorganic filler blending data that has a probability of 25% or more of satisfying the required characteristics of the resin composition C as a required characteristic satisfying necessary condition. Also, as an example, the recommended data derivation unit 34 derives inorganic filler blending data that has a probability of 90% or more of satisfying the required characteristics of the resin composition C as a required characteristic satisfying sufficient condition.

[0115] As an example, the third input data 330 is data indicating a viscosity of "μ = 250 Pa s or less." In this case, the recommended data derivation unit 34 inputs the third input data 330 into MODEL2, thereby deriving recommended data 340 corresponding to the third input data 330. In this example, the recommended data 340 is, for example, data indicating the blending ratio of inorganic fillers that can achieve a viscosity of μ = 250 Pa s or less with a predetermined probability.

[0116] In this example, the prediction device 300 first acquires data indicating a viscosity of "μ = 250 Pa s or less" as third input data 330 using the data acquisition unit 33. Then, the recommended data derivation unit 34 inputs the data indicating a viscosity of "μ = 250 Pa s or less" into MODEL2, thereby deriving recommended inorganic filler blending data for resin composition C to satisfy the viscosity of "μ = 250 Pa s or less."

[0117] The numerical range of μ = 250 Pa·s or less is an example of a numerical range set assuming that resin composition C is used as a semiconductor encapsulant. For example, if the viscosity of the semiconductor encapsulant is too high, it is difficult to shape the semiconductor encapsulant into the desired shape for use. Therefore, it is believed that there is a suitable numerical range for the viscosity of the semiconductor encapsulant. The numerical range of μ = 250 Pa·s or less is an example of this suitable numerical range.

[0118] When data indicating a numerical range of viscosity μ = 250 Pa·s or less is used as the third input data 330, the required condition for satisfying the required characteristics, which is derived in the recommended data derivation unit 34 so that the viscosity of the resin composition C is lower than μ, is expressed by the following formula (1), where the blending ratio of the (II) second silica particles in the inorganic filler A is c, the blending ratio of the (III) third silica particles is d, and the blending ratio of the (IV) fourth silica particles is b: -0.078×b -0.105×c -0.011×d -0.0159×b 2 -0.0217×c 2 -0.0130×d 2 -0.0393×b×c-0.0257×c×d+0.0218×d×b+25.498≧0 (1) The data is represented by:

[0119] When data indicating a numerical range of viscosity μ = 250 Pa·s or less is used as the third input data 330, the required characteristic sufficient condition for the viscosity of the resin composition C to be lower than μ, which is derived in the recommended data derivation unit 34, is expressed by the following formula (2): -0.205×b+0.125×c+0.244×d-0.0151×b 2 -0.0102×c 2 -0.0079×d 2 -0.0154×b×c-0.0104×c×d+0.0252×d×b+0.964≧0 (2) The data is represented by:

[0120] (effect) The prediction device 300 can predict the required property fulfillment conditions for a resin composition using MODEL2, which was generated in advance by the model generation device 100. Specifically, the prediction device 300 can derive recommended data 340 by inputting resin composition required property data into MODEL2. As is clear from the above explanations, the recommended data 340 is a prediction result for the resin composition required property data.

[0121] The prediction device 300 can derive inorganic filler blending data that has the highest probability of satisfying the required properties of the resin composition C. The prediction device 300 can also derive inorganic filler blending data that satisfies the required properties of the resin composition C with a desired probability, such as inorganic filler blending data that has a probability of satisfying the required properties of the resin composition C of 25% or more (required property satisfying necessary condition) or inorganic filler blending data that has such a probability of satisfying the required properties of the resin composition C of 90% or more (required property satisfying sufficient condition). [Example]

[0122] A resin composition was produced using spherical silica particles A (corresponding to the first silica particles), spherical silica particles B (corresponding to the fourth silica particles), spherical silica particles C (corresponding to the second silica particles), and spherical silica particles D (corresponding to the third silica particles) shown in Table 1. [Table 1]

[0123] The spherical silica particles (A) were Silica Excelica UF-320 (manufactured by Tokuyama), the spherical silica particles (B) were Silica Sunseal SP-10P (manufactured by Tokuyama), the spherical silica particles (C) were Silica Sunseal SP-03P (manufactured by Tokuyama), and the spherical silica particles (D) were Silica Sunseal SP-01P (manufactured by Tokuyama). The BET specific surface area, volume-based particle size distribution, and surface carbon content of each spherical silica particle were measured as follows.

[0124] (BET specific surface area) Using a specific surface area measuring device SA-1000 manufactured by Shibata Rikagaku Co., Ltd., the BET specific surface area S (m 2 / g) was measured.

[0125] (Volume-based particle size distribution by laser diffraction and scattering method) Approximately 0.1 g of spherical silica particles were weighed out using an electronic balance and placed in a 50 mL glass bottle. Approximately 40 mL of ethanol was added, and the particles were dispersed using an ultrasonic homogenizer (BRANSON, Sonifier 250) at 40 W for 10 minutes. The volume-based 50% diameter (D50) (μm) and coefficient of variation (CV) of the spherical silica particles were then measured using a laser diffraction / scattering particle size distribution analyzer (Beckman Coulter, LS 13 320).

[0126] (surface carbon amount measurement) The carbon content (mass%) of the spherical silica particles was measured by a combustion oxidation method (Horiba, Ltd., EMIA-511). Specifically, the spherical silica particle sample was heated to 1350°C in an oxygen atmosphere, and the resulting carbon content was calculated by converting it to a value per unit mass. The spherical silica particles used for the measurement were pre-treated by heating at 80°C and reducing the pressure in the system to remove moisture adsorbed in the air, before being used for the carbon content measurement.

[0127] 24.8 g of an inorganic filler containing spherical silica particles (A) to (D) in the proportions shown in Examples 1 to 5 and Comparative Examples 1 and 2 in Table 2 was added to 7 g of bisphenol A+F epoxy resin (ZX-1059, manufactured by Nippon Steel Chemical & Material Co., Ltd.) and mixed by hand. In Table 2, the amount of inorganic filler indicates the proportion (wt%) of the inorganic filler relative to the entire resin composition, and a to d indicate the blending proportion (%) of spherical silica particles (A) to (D) relative to the entire inorganic filler.

[0128] The hand-kneaded resin composition was pre-kneaded using a planetary centrifugal mixer (THINKY, Awatori Rentaro AR-500) (kneading: 1000 rpm, 8 minutes; degassing: 2000 rpm, 2 minutes). The pre-kneaded resin composition was stored in a 25°C thermostatic water bath and then kneaded using a three-roll mill (IMEX, BR-150HCV, roll diameter φ63.5). The kneading conditions were a kneading temperature of 25°C, a roll distance of 20 μm, and five kneading passes. The resulting resin composition was degassed for 30 minutes under reduced pressure using a vacuum pump (Sato Vacuum, TSW-150).

[0129] The viscosity of the kneaded resin composition was measured using a rheometer (HAAKE MARS40, manufactured by Thermo Fisher Scientific) at a temperature of 25°C and using a C35 / 1 sensor (cone-plate type, diameter 35 mm, angle 1°, made of titanium). [Table 2] Examples 1 and 2 satisfy the above formula (1). Examples 3 to 5 satisfy the above formulas (1) and (2). Comparative Examples 1 and 2 do not satisfy the above formulas (1) and (2). It has been shown that by satisfying at least one of the above formulas (1) and (2), it is possible to produce the following resin composition with a viscosity of 250 Pa s.

[0130] [Software implementation example] The functions of the resin composition manufacturing system 1 (hereinafter referred to as the "system") can be realized by a program that causes a computer to function as the system, and a program that causes a computer to function as each control block of the system (particularly, each part included in the model generation device 100 and the prediction device 300).

[0131] In this case, the system includes a computer having at least one control device (e.g., a processor) and at least one storage device (e.g., a memory) as hardware for executing the program. The control device and storage device execute the program, thereby realizing the functions described in each of the above embodiments.

[0132] The program may be non-transitory and may be recorded on one or more computer-readable recording media. The recording media may or may not be included in the device. In the latter case, the program may be supplied to the device via any wired or wireless transmission medium.

[0133] Furthermore, some or all of the functions of the control blocks can be realized by logic circuits. For example, an integrated circuit in which a logic circuit that functions as each of the control blocks is formed is also included in the scope of the present invention. In addition, the functions of the control blocks can also be realized by, for example, a quantum computer.

[0134] Furthermore, each process described in each of the above embodiments may be executed by AI (Artificial Intelligence). In this case, the AI ​​may run on the control device or on another device (for example, an edge computer or a cloud server).

[0135] [Additional Notes] The present invention is not limited to the above-described embodiments, and various modifications are possible within the scope of the claims. Embodiments obtained by appropriately combining the technical means disclosed in different embodiments are also included in the technical scope of the present invention. [Industrial Applicability]

[0136] One aspect of the present invention can be used, for example, in the field of manufacturing electronic devices. [Explanation of symbols]

[0137] 1. Resin composition manufacturing system 100 Model generation device 300 Prediction Device 33 Third input data acquisition unit (data acquisition unit) 34 Recommended Data Derivation Section

Claims

1. A resin composition comprising a resin and an inorganic filler, The inorganic filler includes: First silica particles having a median diameter of 2 μm or more and 5 μm or less; second silica particles, which are wet silica particles having a median diameter of 0.2 μm or more and less than 0.5 μm; third silica particles, which are wet silica particles having a median diameter of 0.1 μm or more and less than 0.2 μm; and fourth silica particles that are wet silica having a median diameter of 0.5 μm or more and less than 1.5 μm, In the inorganic filler (i) the blending ratio of the second silica particles is c (mass%), (ii) the blending ratio of the third silica particles is d (mass%), (iii) When the blending ratio of the fourth silica particles is b (mass%), the following formula (1) is satisfied: -0.078×b-0.105×c-0.011×d-0.0159×b ２ -0.0217×c ２ -0.0130×d ２ -0.0393×b×c-0.0257×c×d+0.0218×d×b+25.498≧0・・・・(1) A resin composition that satisfies the above requirements.

2. The blending ratio of the inorganic filler in the resin composition is 60% by mass or more and 95% by mass or less, The resin composition according to claim 1 , wherein a mixing ratio a of the first silica particles in the inorganic filler is 50% by mass or more and 80% by mass or less.

3. The resin composition according to claim 1 , wherein the second silica particles, the third silica particles, and the fourth silica particles are treated with a surface treatment agent.

4. The resin composition according to claim 3 , wherein the surface treatment agent is a silane coupling agent.

5. The resin composition according to claim 1 , wherein the resin is an epoxy resin.

6. A semiconductor encapsulant comprising the resin composition according to claim 1 .

Images