Method for searching for ligand

Abstract

One embodiment of the present invention is a method for searching for a ligand coordinating to cobalt. In the method, a processor carries out a step for searching for a ligand, from among ligands each represented by any of general formulae (I) to (VI), in which the logarithm of a predicted curing time [T] [sec] calculated by formula (1) is equal to or less than a prescribed value. Formula (1): log10 [T] = α0 + α1x1 + α2x2 (In the formula, x1 is the NBO charge of a cobalt atom of a complex dormant species in which a ligand is coordinated to the cobalt atom; x2 is the occupancy of a ligand present within 3 Å from a cobalt atom in the complex dormant species in which a ligand is coordinated to the cobalt atom; and a0, a1 and a2 are coefficients.

Description

Ligand search method

[0001] The present invention relates to a method for searching for ligands.

[0002] Conventionally, when radically polymerizable resins such as unsaturated polyester resins and vinyl ester resins are cured at room temperature, redox initiators containing a curing agent and a curing accelerator are generally used. Known combinations of the curing agent and curing accelerator (curing agent / curing accelerator) contained in the redox initiator include, for example, a combination of ketone peroxide and an organic acid salt of cobalt (ketone peroxide / organic acid salt of cobalt).

[0003] Adding a ligand to a cobalt organic acid salt changes the curing performance of the cobalt organic acid salt used as a curing accelerator, which in turn changes the curing performance of the redox initiator. Therefore, in order to cure radical-polymerizable resins at room temperature in a short time, methods are being investigated to find ligands contained in curing accelerators that can shorten the curing time of radical-polymerizable resins.

[0004] As a method for searching for a ligand to be added to a curing accelerator, for example, a method for searching for a ligand such as a substituted aniline, which has a curing time calculated by a linear regression equation that expresses the relationship between the LUMO level and HOMO level of a dormant species of a complex to which the ligand is coordinated, and an oxygen atom present within 2 [Å] from the cobalt atom, which is equal to or shorter than a predetermined time, has been disclosed (see, for example, Patent Document 1).

[0005] Japanese Patent Application Publication No. 2021-147350

[0006] Here, for example, in order to cure a radical polymerizable resin at room temperature in a shorter time using a redox initiator, it is important to improve the accuracy of predicting the curing time of the radical polymerizable resin when a different curing accelerator is used for the redox initiator. In the ligand search method of Patent Document 1, values ​​related to the electrical properties of the molecules constituting the curing accelerator are used as characteristic values ​​for predicting the curing time of the radical polymerizable resin, and three-dimensional properties are not taken into consideration.

[0007] An object of one aspect of the present invention is to provide a method for searching for a ligand that searches with high accuracy for a ligand of a curing accelerator that can shorten the curing time when curing a radically polymerizable resin at room temperature.

[0008] The present invention has the following features: [1] A method for searching for a ligand that coordinates to cobalt, comprising: a processor detecting a ligand represented by general formula (I): (In the formula, R １ , R ２ are each independently a hydrogen atom or an alkyl group having 1 to 10 carbon atoms, and R ３ ~R ７ are each independently a hydrogen atom, an alkyl group having 1 to 10 carbon atoms, an alkoxy group having 1 to 10 carbon atoms, an alkoxycarbonyl group having 2 to 11 carbon atoms, an aryl group having 6 to 12 carbon atoms, a hydroxy group, an amino group, a carboxy group, a cyano group, a nitro group, a formyl group, or a halogen atom), a ligand represented by general formula (II): (In the formula, R １１ ~R １３ are each independently a hydrogen atom, an alkyl group having 1 to 10 carbon atoms, an alkoxy group having 1 to 10 carbon atoms, an aryl group having 6 to 12 carbon atoms, an amino group, a carboxy group, a cyano group, or a halogen atom; R １２ and R １３ may be joined together to form a ring.) A ligand represented by general formula (III): (In the formula, R ２１ ~R ２３ are each independently a hydrogen atom, an alkyl group having 1 to 10 carbon atoms, an alkoxy group having 1 to 10 carbon atoms, an alkoxycarbonyl group having 2 to 11 carbon atoms, an aryl group having 6 to 12 carbon atoms, a hydroxy group, an amino group, a carboxy group, a cyano group, a nitro group, a formyl group, or a halogen atom), and a ligand represented by general formula (IV): (In the formula, R ３１ ~R ３５and each independently represent a hydrogen atom, an alkyl group having 1 to 10 carbon atoms, an alkoxy group having 1 to 10 carbon atoms, an alkoxycarbonyl group having 2 to 11 carbon atoms, an aryl group having 6 to 12 carbon atoms, a hydroxy group, an amino group, a carboxy group, a cyano group, a nitro group, a formyl group, a halogen atom, or a pyridine ring. １０ [T] = α ０ +α １ x １ +α ２ x ２ ...(1) (in the formula, x １ is the NBO charge of the cobalt atom in the dormant species of the complex in which the ligand is coordinated to the cobalt atom, and x ２ is the occupancy of the ligands present within 3 Å from the cobalt atom in the dormant species of the complex in which the ligands are coordinated to the cobalt atom, and a ０ , a １ and a ２ is a coefficient.) [2] The above formula (1) can be expressed as formula (2): log １０ [T](x ｉ )=5.20950x １ -0.50360x ２+47.001 ... (2) wherein the predetermined value is 5.0. [3] The ligand searching method according to [1] or [2], wherein the dormant species of the complex in which the ligand represented by general formula (I), (III) or (IV) is coordinated to cobalt is a dormant species of cobalt(III) in which an OH radical or a Me radical is coordinated in place of one molecule of the ligand represented by general formula (I), (III) or (IV) in a Werner complex of cobalt(II) in which two molecules of a carboxylate ligand, which is a bidentate ligand, and two molecules of a monodentate ligand, which is represented by general formula (I), (III) or (IV), are coordinated. [4] The dormant species of the complex in which the ligand represented by the general formula (II) is coordinated to cobalt is a dormant species of cobalt(III) in which an OH radical or an Me radical is coordinated in place of one of the coordinate positions of one molecule of the ligand represented by the general formula (II) in a Werner complex of cobalt(II) in which three molecules of the ligand represented by the general formula (II), which is a bidentate ligand, are coordinated. The method for searching for a ligand according to any one of [1] to [3].

[0009] According to one aspect of the present invention, it is possible to provide a method for searching for a ligand that searches with high accuracy for a ligand of a curing accelerator that can shorten the curing time when curing a radical polymerizable resin at room temperature.

[0010] 1 is a schematic diagram showing an example of a ligand searching system. FIG. 1 is a block diagram showing the functional configuration of a regression model generating device. FIG. 2 is a diagram showing an example of the correlation between measured curing time and predicted curing time. FIG. 3 is a block diagram showing the functional configuration of a ligand searching device. FIG. 4 is a block diagram showing the hardware configurations of a regression model generating device and a ligand searching device. FIG. 5 is a flowchart explaining a ligand searching method according to the present embodiment. FIG. 6 is a flowchart showing a regression model generating process. FIG. 7 is a flowchart showing a ligand searching process. FIG. 8 is a diagram showing the correlation between the logarithm of the predicted curing time and the logarithm of the measured curing time when the ligands of Reference Examples 1 to 13 are used. FIG. 9 is a diagram showing the plotted positions of the correlation between the logarithm of the predicted curing time and the logarithm of the measured curing time when the ligand of Reference Example 5 is used as test data. FIG. 10 is a diagram showing the plotted positions of the correlation between the logarithm of the predicted curing time and the logarithm of the measured curing time when the ligand of Reference Example 7 is used as test data. FIG. 11 is a diagram showing the plotted positions of the correlation between the logarithm of the predicted curing time and the logarithm of the measured curing time when the ligand of Reference Example 8 is used as test data.

[0011] Hereinafter, embodiments of the present invention will be described in detail. In this specification, unless otherwise specified, the term "to" indicating a range of values ​​means that the range includes the values ​​before and after it as the lower and upper limits.

[0012] A ligand searching method according to this embodiment will be described. The ligand searching method according to this embodiment is a method for searching for a ligand of a curing accelerator contained in a curing agent. Before describing the ligand searching method according to this embodiment, a ligand searching system used in the ligand searching method according to this embodiment will be described.

[0013] <Ligand Search System> An outline of a ligand search system used in the ligand search method according to this embodiment will be described. FIG. 1 is a schematic diagram showing an example of a ligand search system. As shown in FIG. 1, the ligand search system 1 includes a regression model generation device 100 and a ligand search device 200. In the ligand search system 1, the regression model generation device 100 acquires material data related to a curing accelerator (e.g., the structure, physical property values, and logarithm of the curing accelerator's ligands) from a material database 300, and generates a regression model that predicts the logarithm of the curing time from the physical property values ​​of the curing accelerator. In the ligand search system 1, the ligand search device 200 uses the generated regression model to calculate the logarithm (log １０ We search for a ligand of the curing accelerator that makes the logarithm of the curing accelerator (log [T]) equal to or less than a predetermined value. １０ [T] = α ０ +α １ x １ +α ２ x ２ ...(1) (in the formula, x １ is the NBO (natural bond orbital) charge of the cobalt atom in the dormant species of the complex in which the ligand is coordinated to the cobalt atom, and x ２ is the occupancy of the ligands present within 3 Å from the cobalt atom in the dormant species of the complex in which the ligands are coordinated to the cobalt atom, and a ０ , a １ and a ２ is the coefficient.)

[0014] (In the formula, R １ , R ２ are each independently a hydrogen atom or an alkyl group having 1 to 10 carbon atoms, and R ３ ~R ７are each independently a hydrogen atom, an alkyl group having 1 to 10 carbon atoms, an alkoxy group having 1 to 10 carbon atoms, an alkoxycarbonyl group having 2 to 11 carbon atoms, an aryl group having 6 to 12 carbon atoms, a hydroxy group, an amino group, a carboxy group, a cyano group, a nitro group, a formyl group, or a halogen atom.

[0015] (In the formula, R １１ ~R １３ are each independently a hydrogen atom, an alkyl group having 1 to 10 carbon atoms, an alkoxy group having 1 to 10 carbon atoms, an aryl group having 6 to 12 carbon atoms, an amino group, a carboxy group, a cyano group, or a halogen atom; R １２ and R １３ may be joined together to form a ring.)

[0016] (In the formula, R ２１ ~R ２３ are each independently a hydrogen atom, an alkyl group having 1 to 10 carbon atoms, an alkoxy group having 1 to 10 carbon atoms, an alkoxycarbonyl group having 2 to 11 carbon atoms, an aryl group having 6 to 12 carbon atoms, a hydroxy group, an amino group, a carboxy group, a cyano group, a nitro group, a formyl group, or a halogen atom.

[0017] (In the formula, R ３１ ~R ３５ are each independently a hydrogen atom, an alkyl group having 1 to 10 carbon atoms, an alkoxy group having 1 to 10 carbon atoms, an alkoxycarbonyl group having 2 to 11 carbon atoms, an aryl group having 6 to 12 carbon atoms, a hydroxy group, an amino group, a carboxy group, a cyano group, a nitro group, a formyl group, or a halogen atom.

[0018] The amino group in the above general formulas (I) to (IV) is a primary amino group (—NH ２ ), secondary amino groups (-NHR) and tertiary amino groups (-NR ２ ) is included.

[0019] In the above general formulas (I) to (IV), R is preferably an alkyl group having 1 to 10 carbon atoms.

[0020] When the amino group in the above general formulas (I) to (IV) is a tertiary amino group, the two R in the tertiary amino group may be the same or different, or may join together to form a ring.

[0021] Examples of the halogen atom in each R in the general formulae (I) to (IV) include a chlorine atom, a bromine atom, and an iodine atom.

[0022] The NBO charge is the charge calculated by the natural bond orbital (NBO) of each atom. ＊ The natural bond orbital (NBO) analysis of the electronic state calculated by the natural bond orbital (NBO) method can be used to obtain the natural electron density of each atom. The sum of the natural electron density and the atomic number corresponding to the positive charge density of each atom is the NBO charge. The NBO charge is an index that indicates the electrical bias of each atom.

[0023] The dormant species of the complex in which the above-mentioned ligand is coordinated to cobalt is generated when the active species (P.) coordinates to the cobalt that constitutes the Werner complex when a radically polymerizable resin is cured at room temperature using a redox initiator consisting of a ketone peroxide / organic acid salt of cobalt.

[0024] Examples of the active species (P·) include radicals generated by decomposition of ketone peroxide, and radicals generated by the reaction of radicals generated by decomposition of ketone peroxide with a radical polymerizable resin.

[0025] A Werner complex is an octahedral complex in which a ligand having an unshared electron pair is coordinated to a central metal ion.

[0026] In this embodiment, as the Werner complex, a cobalt(II) complex in which two molecules of a carboxylate ligand, which is a bidentate ligand, and two molecules of a monodentate ligand represented by general formula (I), (III), or (IV), or a cobalt(II) complex in which three molecules of a bidentate ligand represented by general formula (II), are coordinated is used.

[0027] The dormant species of the complex in which the above-mentioned ligand is coordinated to cobalt may be coordinated with an OH radical (.OH) or an Me radical (.Me) instead of the active species (P.) contained in the dormant species. Note that the Me radical may be a polymer chain grown by extending the carbon to, for example, Me, Et, etc.

[0028] That is, the dormant species of a complex in which a ligand represented by general formula (I), (III) or (IV) is coordinated to cobalt may be a structurally optimized dormant species of cobalt(III) in which, in a Werner complex of cobalt(II) in which two molecules of a carboxylate ligand, which is a bidentate ligand, and two molecules of a monodentate ligand, which is represented by general formula (I), (III) or (IV), are coordinated, one molecule of an OH radical, Me radical or the like is coordinated in place of one molecule of the ligand represented by general formula (I), (III) or (IV).

[0029] The dormant species of the complex in which the ligand represented by general formula (II) is coordinated to cobalt may be a structurally optimized dormant species of cobalt(III) in which one molecule of an OH radical, Me radical, or the like is coordinated in place of one of the coordinate positions of one molecule of the ligand represented by general formula (II) in a Werner complex of cobalt(III) in which three molecules of the ligand represented by general formula (II), which is a bidentate ligand, are coordinated.

[0030] Examples of the carboxylate ligand include an acetato ligand.

[0031] Here, cobalt complexes such as the complexes in which the above-mentioned ligands are coordinated to cobalt can generally cure radical polymerizable resins at room temperature by an exchange chain mechanism. For example, a Werner complex of cobalt (II) in which two molecules of a bidentate carboxylate ligand and two molecules of a monodentate ligand represented by general formula (I), (III), or (IV) are coordinated contains an active species (P １ A dormant species is formed by coordinating one molecule of cobalt (III) with an active species (P １・) to other active species (P ２ ・) and activated species (P １ ・) and releases active species (P １ ) reacts with the unsaturated units of the polymer chain of the radical polymerizable resin to form active species, and the resulting active species reacts with the cobalt complex to form dormant species, from which the active species (P ２ ・) is released.

[0032]

[0033] In addition, the Werner complex of cobalt (II) in which three molecules of the ligand represented by the general formula (II), which is a bidentate ligand, are coordinated contains an active species (P １ A dormant species is formed by coordinating one molecule of cobalt (III) with an active species (P １ ・) to other active species (P ２ ・) and activated species (P １ ・) and releases active species (P １ ) reacts with the unsaturated units of the polymer chain of the radical polymerizable resin to form active species, and the resulting active species reacts with the cobalt complex to form dormant species, from which the active species (P ２ ・) is released.

[0034]

[0035] In such an exchange chain mechanism in which activation and deactivation reactions occur simultaneously, the overall reaction rate is equal to the exchange reaction rate, and the higher the exchange reaction rate, the more the radical polymerizable resin hardens.

[0036] The ligand search system 1 calculates log １０ Using the value of [T] as a standard, a search is made for a ligand of a curing accelerator used when curing a radical polymerizable resin at room temperature. The regression model is based on the NBO charge x of the cobalt atom in the dormant species of a complex in which a ligand is coordinated to the cobalt atom, as shown in the above formula (1). １and the occupancy rate x of the ligands present within 3 Å from the cobalt atom in the dormant species of the complex in which the ligands are coordinated to the cobalt atom. ２ Two physical properties x ｉ (i=1, 2). The logarithm calculated by the linear regression equation (1) is １０ A decrease in [T] corresponds to a reduction in the curing time T actually measured using a curing accelerator containing an organic acid salt of cobalt and a ligand (hereinafter referred to as the actual curing time T). Therefore, the ligand searching system 1 can search with higher accuracy for a ligand of a curing accelerator that can cure a radical polymerizable resin at room temperature in a shorter time.

[0037] [Regression Model Generating Device] The regression model generating device 100 will be described. FIG. 2 is a block diagram showing the functional configuration of the regression model generating device 100. As shown in FIG. 2, the regression model generating device 100 has a physical property value acquiring unit 110 and a regression model generating unit 120. The regression model generating device 100 acquires material data (e.g., curing time) related to a curing accelerator from a material database 300, and generates a regression model that predicts the curing time from the physical property values ​​of the curing accelerator. The regression model generating device 100 provides the generated regression model to the ligand searching device 200.

[0038] (Physical property value acquisition unit) The physical property value acquisition unit 110 acquires material data to be used for generating a regression model from the material database 300, and performs predetermined processing such as quantum chemical calculations on the acquired material data to acquire the physical property values ​​of the material.

[0039] The physical property value acquisition unit 110 uses, for example, quantum chemistry calculation software to determine the NBO charge of the cobalt atom in a dormant species of a complex in which a ligand is coordinated to the cobalt atom, and the occupancy of the ligands present within 3 Å of the cobalt atom in the dormant species of the complex in which a ligand is coordinated to the cobalt atom.

[0040] To simplify the calculation, an Me radical (.Me) or an OH radical (.OH) may be used in place of the active species (P.) coordinated to the dormant species.

[0041] As quantum chemistry calculation software, for example, Gaussian09 manufactured by Gaussian Inc. may be used. Gaussian09 is based on density functional theory, and is spin-unrestricted, with RωB97XD as the functional and 6-31 as the basis function. ＋ G(d) may also be used. A frequency analysis is performed on all calculations to confirm that the number of imaginary frequencies is 0, and it is determined that the potential surface converges to the bottom.

[0042] The physical property value acquisition unit 110 may acquire a non-optimized structure or an optimized structure from the substance database 300 .

[0043] When searching for ligands that coordinate to cobalt, a Werner complex of cobalt(II) is used, in which two molecules of a bidentate carboxylate ligand are coordinated with one molecule of a ligand (L) that is either a monodentate ligand or a bidentate ligand other than a carboxylate ligand. The ligands can be divided into two types: amine-based systems in which the partial structure of the ligand is replaced by an amine-based structure, and diketone-based systems in which the partial structure of the ligand is replaced by a diketone-based structure.

[0044] When the ligand (L) is an amine-based ligand with an amine-based moiety substituted therein, the ligand (L) is represented by the following chemical formula (A): Chemical formula (A) corresponds to the above general formula (I), (III), or (IV) which is a monodentate ligand.

[0045] That is, when the ligand (L) is a monodentate ligand represented by the above general formula (I), (III), or (IV), a Werner complex of cobalt(II) is used in which two molecules of a bidentate carboxylate ligand and a monodentate ligand represented by the above general formula (I), (III), or (IV) are coordinated. In this Werner complex, a dormant species of cobalt(III) in which one molecule of the active species (P.) is coordinated in place of one molecule of the ligand (L) represented by the above general formula (I), (III), or (IV) is used as the initial structure of the dormant species in which the ligand represented by the above general formula (I), (III), or (IV) is coordinated to an organic acid salt of cobalt. To simplify the calculation, an Me radical (·Me) is used instead of the active species (P·), and the carboxylate ligand is approximated by an acetate ligand with a short alkyl chain, so that the initial structure of the resulting dormant species can be a structure represented by the following chemical formula (A):

[0046]

[0047] When the ligand (L) is a diketone-based ligand in which a diketone-based partial structure is substituted, the ligand (L) is represented by the following chemical formula (B): Chemical formula (B) corresponds to the above general formula (II) which is a bidentate ligand.

[0048] That is, when the ligand (L) is a bidentate ligand represented by the general formula (II) above, and three molecules of the ligand coordinated to cobalt are used as the Werner complex when searching for a ligand to coordinate to cobalt, a cobalt(II) Werner complex in which three molecules of the bidentate ligand represented by the general formula (II) above are coordinated is used. In this Werner complex, a dormant species of cobalt(III) in which one molecule of the ligand represented by the general formula (II) above is coordinated with one molecule of the active species (P.) is used as the initial structure of the dormant species in which the ligand represented by the general formula (II) above is coordinated to an organic acid salt of cobalt. Then, to simplify the calculation, a dormant species in which an Me radical (.Me) is coordinated instead of the active species (P.) is used, and the initial structure of the resulting dormant species may be a structure represented by the following chemical formula (B).

[0049]

[0050] The physical property value acquisition unit 110 preferably performs structural optimization of the initial structure of the dormant species. The regression model to be estimated is a regression model of two physical property values ​​x ｉ (i is 0, 1, or 2), and a linear regression equation expressed by a linear combination including the following equation (1a) is performed. ｉ (where i is 0, 1, or 2). １０ [T] = α ０ +α １ x １ +α ２ x ２ ...(1a) (in the formula, x １ is the NBO charge of the cobalt atom in the dormant species of the complex in which the ligand is coordinated to the cobalt atom, and x ２ is the occupancy of the ligands present within 3 Å from the cobalt atom in the dormant species of the complex in which the ligands are coordinated to the cobalt atom, and a ０ , a １ and a ２ is the coefficient.)

[0051] Physical property value x ｉ is a physical property value obtained by first-principles calculation of the dormant species, and the coefficient a ｉis a rational number with two decimal places. ｉ The dormant species with optimized structure may be calculated by first-principles calculation.

[0052] x in formula (1a) １ As described above, is the NBO charge of the cobalt atom in the dormant species of the complex in which the ligand is coordinated to the cobalt atom. In this case, when the initial structure of the dormant species coordinated to the ligand in the Werner complex of cobalt (II) is a monodentate ligand represented by the general formula (I), (III) or (IV) as in the above chemical formula (A), x in formula (1a) １ is the NBO charge of the cobalt atom indicated by the arrow in the following formula (α1).

[0053]

[0054] When the initial structure of the dormant species coordinated to the ligand coordinated to the Werner complex of cobalt (II) is a bidentate ligand represented by the general formula (II) as shown in the chemical formula (B), x in formula (1a) １ is the NBO charge of the cobalt atom indicated by the arrow in formula (α2) below.

[0055]

[0056] x in formula (1a) ２ is the occupancy of the ligands present within 3 Å from the cobalt atom in the dormant species of the complex in which the ligands are coordinated to the cobalt atom, as described above.

[0057] When the initial structure of the dormant species coordinated to the ligand coordinated to the Werner complex of cobalt (II) is a monodentate ligand represented by the general formula (I), (III) or (IV) as in the chemical formula (A), x in formula (1a) ２ As shown in the following formula (β1), is the occupancy rate of the ligands present inside a sphere with a radius of 3 Å centered on the cobalt atom.

[0058]

[0059] When the initial structure of the dormant species coordinated to the ligand coordinated to the Werner complex of cobalt (II) is three molecules of the ligand represented by the general formula (II) above, which is a bidentate ligand, as in the chemical formula (B) above, x in formula (1a) ２ As shown in the following formula (β2), is the occupancy rate of the ligands present inside a sphere with a radius of 3 Å centered on the cobalt atom.

[0060]

[0061] The physical property value acquisition unit 110 acquires the measured hardening time T of the dormant species of the complex to which the ligand is coordinated and the logarithm (log １０ T) is calculated.

[0062] The actual curing time T may be obtained from the substance database 300 or may be actually measured. The actual curing time T may be measured in accordance with, for example, JIS K 6901 ("Test methods for liquid unsaturated polyester resins," "Room temperature curing characteristics (exothermic method)").

[0063] The physical property value acquisition unit 110 acquires the physical property value x ｉ and log １０ T is sent to the regression model generation unit 120.

[0064] (Regression Model Generation Unit) The regression model generation unit 120 generates a regression model using the acquired physical property value x ｉ and log １０ A multiple regression analysis is performed on T, and the coefficient a of the linear regression equation (1a) is used as the regression model. ０ , a １ , a ２ Specifically, the regression model generation unit 120 estimates the ligand L ｎ (n is an integer) ｉ Substituting into the linear regression equation to be estimated, [T]_L ｎ That is, [T]_L ｎ is the ligand L ｎ There are n ligands in total, the same number as the number of ligands.

[0065] The regression model generation unit 120 ｎ The coefficient a is set to reproduce the actual curing time T of eachｉ (i = 0 to 2) is determined. For example, the coefficient of determination R ２ By log １０ The reproducibility of T may be evaluated. The coefficient of determination R ２ It is known that the closer to 1 the correlation between two values ​​is, and generally, if the value is 0.6 or more, it can be said that there is a correlation. ２ When the reproducibility is 0.6 or more, it can be evaluated as being reproducible.

[0066] The regression model generation unit 120 calculates log １０ T and log １０ Coefficient of determination R of [T] ２ The coefficient a is adjusted using, for example, an evolutionary algorithm so that ｉ (i=0 to 2) may be determined.

[0067] For example, the estimation of regression coefficients using an evolutionary algorithm can be realized by using the evolutionary algorithm in the solver included in Excel (registered trademark).

[0068] As calculation conditions, for example, convergence, mutation rate, population size, random seed, and maximum time until improvement is not observed may be appropriately set to predetermined values, and the regression coefficients may be estimated by the solver. Also, upper and lower limits of variables are not required, and the a output as a calculation result may be ｉ The value of does not need to be limited to positive or negative.

[0069] The regression model generation unit 120 calculates the ligand L ｎ The NBO charge x of the cobalt atom in the dormant species of the complex in which the ligand is coordinated to the cobalt atom is obtained for １ , the occupancy rate x of the ligands present within 3 Å from the cobalt atom in the dormant species of the complex in which the ligands are coordinated to the cobalt atom ２ Each physical property value x ｉ With respect to a predetermined coefficient a ｉ A linear regression equation with (i=0 to 2) can be derived.

[0070] The regression model generation unit 120 uses the derived linear regression equation to calculate the physical property value x of the dormant species of the complex in which the ligand is coordinated to the organic acid salt of cobalt. ｉ For log １０ Find [T].

[0071] log １０ [T] and log １０ An example of the relationship between log １０ To match the T, log １０ A constant multiplication of [T] and a parallel translation of the intercept may be performed. １０ [T] and log １０ The linear regression equation can be obtained by adjusting the slope of the linear approximation equation showing the relationship with T so that it becomes 1.0. The coefficient of determination R ２ is a predetermined value (for example, 0.6), there is sufficient correlation and log １０ It can be said that log [T] is expected. １０ The operation of moving [T] is log １０ [T] is log １０ It is not essential as long as it is adjusted to fit the scale of T.

[0072] The regression model generation unit 120 sends the regression model represented by the linear regression equation estimated using the above equation (1a) to the ligand screening apparatus 200.

[0073] <Ligand Searching Device> The configuration of the ligand searching device 200 will be described. The ligand searching device 200 uses the regression model acquired from the regression model generating device 100 to search for the log １０ Predict [T] and calculate log １０ A ligand of the curing accelerator is searched for, which makes [T] equal to or less than a predetermined value.

[0074] 4 is a block diagram showing the functional configuration of the ligand searching apparatus 200. As shown in FIG. 4, the ligand searching apparatus 200 includes a physical property value acquiring unit 210, a curing time calculating unit 220, and a ligand searching unit 230.

[0075] The physical property value acquisition unit 210 acquires the physical property values ​​of the curing accelerator for each ligand. Specifically, similar to the physical property value acquisition process described with respect to the regression model generation device 100, the physical property value acquisition unit 210 performs structural optimization on the material data of the material to be processed acquired from the material database 300, and calculates the NBO charge x of the cobalt atom in the dormant species of the complex in which the ligand is coordinated to the cobalt atom, as described above, by first-principles calculation of the structurally optimized dormant species. １ , the occupancy rate x of the ligands present within 3 Å from the cobalt atom in the dormant species of the complex in which the ligands are coordinated to the cobalt atom ２ Each physical property value x ｉ Ask for.

[0076] The hardening time calculation unit 220 calculates the obtained physical property value x ｉ is substituted into the linear regression equation, and the logarithm of the predicted curing time [T] is obtained. １０ Calculate [T] and calculate the logarithm of the calculated １０ [T] is provided to the ligand search unit 230 .

[0077] The ligand searching unit 230 calculates the log １０ Determine whether [T] is equal to or less than a predetermined value, and １０ A curing accelerator having a ligand for which [T] is equal to or less than a predetermined value is determined as the curing accelerator to be searched for.

[0078] Log calculated by the linear regression equation of the above equation (1) １０ The smaller [T] is, the shorter the actual curing time T of the curing accelerator containing the organic acid salt of cobalt and the ligand is. １０ If [T] is equal to or less than a predetermined value, it can be said that the curing time of the curing accelerator having a ligand is short.

[0079] The predetermined value may be, for example, 5.0. １０ If [T] exceeds a predetermined value, the ligand searching unit 230 determines that the curing accelerator having the ligand is not a search target, and searches for a curing accelerator having the next ligand.

[0080] (Hardware Configuration) The regression model generation device 100 and the ligand screening device 200 may have a hardware configuration as shown in Fig. 5. That is, the regression model generation device 100 and the ligand screening device 200 have a drive device 101, an auxiliary storage device 102, a memory device 103, a CPU (Central Processing Unit) 104, an input / output device 105, and a communication device 106, which are interconnected via a bus B.

[0081] Various computer programs, including programs for implementing various functions and processes described below in the regression model generating device 100 and the ligand searching device 200, may be provided on a recording medium 107 such as a CD-ROM (Compact Disk-Read Only Memory). When the recording medium 107 storing the programs is inserted into the drive device 101, the programs are installed from the recording medium 107 to the auxiliary storage device 102 via the drive device 101. However, the programs do not necessarily have to be installed using the recording medium 107, but may be downloaded from any external device via a network or the like. The auxiliary storage device 102 stores the installed programs as well as necessary files, data, etc. The memory device 103 reads and stores the programs and data from the auxiliary storage device 102 when a program startup instruction is received. The CPU 104, functioning as a processor, executes various functions and processes of the regression model generation device 100 and the ligand searching device 200, as described below, in accordance with programs stored in the memory device 103 and various data such as parameters required to execute the programs. The input / output device 105 provides an interface between the regression model generation device 100 and the ligand searching device 200 and a user, and may be composed of input devices such as a keyboard, mouse, and touch screen, and output devices such as a display and speaker. The communication device 106 executes various communication processes for communicating with external devices.

[0082] However, the regression model generating device 100 and the ligand searching device 200 are not limited to the above-mentioned hardware configuration, and may be realized by any other appropriate hardware configuration and configuration via a network.

[0083] <Ligand Search Method> A ligand search method according to this embodiment will be described. In the ligand search method according to this embodiment, a ligand search system 1 having a configuration as shown in FIG. 1 acquires substance data (e.g., curing time) related to a curing accelerator from a substance database 300, generates a regression model that predicts the curing time from the physical property values ​​of the curing accelerator, and a ligand search device 200 searches for a ligand for the curing accelerator using the generated regression model. The ligand search method according to this embodiment is executed, for example, by the above-described ligand search system 1, and specifically, may be realized by the regression model generation device 100 and the processor of the ligand search device 200 that constitute the ligand search system 1. Because the substance search method according to this embodiment can be performed using the above-described ligand search system 1, some of the details already described for the above-described ligand search system 1 will be omitted in each step.

[0084] 6 is a flowchart illustrating the ligand search method according to this embodiment. As shown in FIG. 6, the ligand search method according to this embodiment includes a regression model generation process S10 and a ligand search process S20.

[0085] [Regression Model Generation Process] The regression model generation process S10 will be described. The regression model generation process may be executed by, for example, the regression model generation device 100 described above, and specifically, may be realized by the processor of the regression model generation device 100.

[0086] 7 is a flowchart showing an example of the regression model generation process S10. As shown in Fig. 7, in the regression model generation process S10, the regression model generation device 100 performs structural optimization of dormant species for the substance data acquired from the substance database 300 (structural optimization step: step S101).

[0087] Next, the regression model generating device 100 calculates the NBO charge x of the cobalt atom in the dormant species of the complex in which the ligand is coordinated to the cobalt atom from the structurally optimized dormant species. １ , the occupancy rate x of the ligands present within 3 Å from the cobalt atom in the dormant species of the complex in which the ligands are coordinated to the cobalt atom ２ Each physical property value x ｉ (Acquisition process of physical property values: step S102).

[0088] Next, the regression model generating device 100 calculates the obtained physical property value x ｉ A multiple regression analysis is performed using the above as explanatory variables, and a linear regression equation is estimated as regression model i (i is an integer from 1 to 8). The coefficient of determination of the linear regression equation is R2 ｉ is estimated (first regression model and coefficient of determination estimation step: step S103).

[0089] For example, the regression model generating device 100 ｉ is substituted into the following equation (10), which is the linear regression equation to be estimated, to estimate the regression model i.

[0090]

[0091] In this case, the physical property value x when i is 1 to 8 in formula (10) ｉ is as follows: x １ :SOMO x ２ :LUMO x ３ : Distance x between Me and Co ４ : Co NBO charge x ５ : NBO charge of carbon of Me x ６ : average value of NBO charge of double bond oxygen x ７ : average distance between Co and oxygen atoms x ８ : Ligand occupancy within 3 Å sphere from Co

[0092] The regression model generating device 100 calculates the coefficients a of the linear regression equation so as to reproduce the actually measured curing time T. ｉ Determine the coefficient of determination R ２ A regression model i is estimated based on

[0093] At this time, the regression model generating device 100 calculates the coefficient of determination R ２If the value is 0.6 or more, it is assumed that there is reproducibility.

[0094] The regression model generating device 100 uses, for example, log １０ [T] and log １０ Coefficient of determination of T R ２ The coefficient a is calculated using an evolutionary algorithm so that approaches 1. ｉ When the multiple regression analysis is performed using the evolutionary algorithm in the solver included in Excel (registered trademark), the calculation conditions, such as convergence, mutation rate, population size, random seed, and maximum time for which no improvement is observed, may be appropriately set to predetermined values, and the regression coefficients may be estimated by the solver.

[0095] Next, the regression model generating device 100 evaluates the contribution of each term of the linear regression equation, which is the regression model i (step of evaluating the contribution of each term of the regression model: step S104).

[0096] The contribution of each term in the regression model can be evaluated by implementing the following two types of formulas (11) and (12) and adopting the one with the higher contribution among the contributions obtained by formulas (11) and (12). Note that the training data for formulas (11) and (12) is the physical property value x ｉ Contribution rate = |coefficient a ｉ × average value of training data | ... (11) Contribution rate = | coefficient a ｉ × standard deviation of training data | ... (12)

[0097] Next, the regression model generating device 100 calculates the physical property value x of the term with the lowest contribution among the contributions of the terms of the linear regression equation expressed by equation (10) that is the evaluated regression model i. ｉ One of the items is excluded (exclusion step: step S105).

[0098] Next, the regression model generating device 100 estimates the acquired physical property value x ｉ A multiple regression analysis is performed using as an explanatory variable, a linear regression equation is estimated as a regression model (i + 1), and the coefficient of determination of the linear regression equation is R2 ｉ＋１is estimated (step of estimating the second regression model and the coefficient of determination: step S106).

[0099] The linear regression equation, which is the regression model (i+1), and its coefficient of determination R2 ｉ＋１ The estimation method of is the same as the estimation step of the first regression model and the coefficient of determination (step S103), so details are omitted.

[0100] Next, the regression model generation device 100 estimates the first regression model and the coefficient of determination R2 ｉ and the coefficient of determination R2 estimated in the second regression model and coefficient of determination estimation step (step S106) ｉ＋１ Compared with the coefficient of determination R2 ｉ Coefficient of determination R2 ｉ＋１ It is determined whether the rate of decrease exceeds 10% (first determination step: step S107).

[0101] Coefficient of determination R ｉ Coefficient of determination R2 ｉ＋１ The rate of decrease is expressed as follows: ((R2 ｉ＋１ -R2 ｉ ) / R2 ｉ ) × 100 < −10 [%] ... (13)

[0102] The regression model generating device 100 calculates the coefficient of determination R2 ｉ Coefficient of determination R2 ｉ＋１ The rate of decrease (R2 ｉ＋１ / R2 ｉ ) × 100) exceeds 10% (step S107: YES), it is determined that the regression model i is more accurate than the regression model (i+1), and the regression model i is adopted (selected) (first adoption step: step S108).

[0103] On the other hand, in the first determination step (step S107), the regression model generation device 100 determines the coefficient of determination R ｉ Coefficient of determination R2 ｉ＋１If it is determined that the rate of decrease is 10% or less (step S107: NO), it is determined that the regression model (i+1) is more accurate than the regression model i, and the regression model (i+1) is adopted (third adoption step: step S111).

[0104] Then, the regression model generating device 100 calculates the physical property value x ｉ It is determined whether the number of training data is less than one-fourth of the number of training data (number of training data / 4) (third determination step: step S112).

[0105] The number of explanatory variables is not limited to less than one-fourth the number of training data, and it is sufficient that the number of explanatory variables is sufficiently small relative to the number of training data to function as an indicator for avoiding overfitting. The number of explanatory variables may be set to an arbitrary value less than the number of training data, for example, less than one-half.

[0106] If the regression model generation device 100 determines that the number of explanatory variables is less than one-fourth of the number of training data (step S112: YES), it determines that the explanatory variables have avoided overfitting with respect to the number of training data, and proceeds to the second determination step (step S109).

[0107] On the other hand, if the regression model generation device 100 determines that the number of explanatory variables is equal to or greater than one-fourth the number of training data (step S112: NO), it determines that overfitting of the explanatory variables relative to the number of training data cannot be avoided, and proceeds to the process of estimating the first regression model and the coefficient of determination (step S103).

[0108] In this way, the regression model generation device 100 repeats the steps of estimating the first regression model and the coefficient of determination (step S103) to the first adoption step (step S108), the third adoption step (step S111), and the third judgment step (step S112), and calculates the coefficient a, which is an arbitrary variable of the linear regression equation shown in Equation (10), which is the regression model i. ｉ The reduction is repeated one by one, and finally the coefficient a ｉ This is repeated until there are two. In other words, the regression model i obtained by the first estimation is the explanatory variable, the physical property value xｉ There are eight (x １ ~x ８ ), physical property value x ｉ The number of ｉ is reduced to only two terms (i=1, 2). This gives the following equation (1): １０ [T] = a ０ +a １ x １ +a ２ x ２ ...(1)

[0109] In addition, x in the above formula (1) １ and x ２ are respectively the x in the above formula (10). １ and x ２ Unlike the eight physical properties x ｉ Of these, the two remaining physical properties x ｉ a in the above formula (1) means １ and a ２ Also, a in the above formula (10) １ and a ２ Unlike the two remaining physical properties x ｉ It may be determined appropriately depending on the situation.

[0110] In equation (1), the regression model generating device 100 uses, for example, log １０ [T] and log １０ Coefficient of determination of T R ２ The coefficient a is calculated using an evolutionary algorithm so that approaches 1. １ and a ２ When the multiple regression analysis is performed using the evolutionary algorithm in the solver included in Excel (registered trademark), the calculation conditions, such as convergence, mutation rate, population size, random seed, and maximum time for which no improvement is observed, may be appropriately set to predetermined values, and the regression coefficients may be estimated by the solver. １０ [T] and log １０ To match the absolute value of T, coefficient a ０ Determine.

[0111] Next, the regression model generating device 100 calculates the physical property value x ｉ All regression models i' are estimated excluding one, and the coefficient of determination R2 of the linear regression equation, which is the regression model i', is calculated. ｉ ' are estimated and compared, and the regression model j with the largest coefficient of determination R2i' is found. The coefficient of determination R2i of regression model j is ｊ is the coefficient of determination R2 of regression model i ｉ It is determined whether or not it is equal to or greater than this (second determination step: step S109).

[0112] Physical property value x ｉ The regression model i' excluding one term is a linear regression equation excluding any one of the terms of the linear regression equation of equation (10), which is the regression model i. That is, when the linear regression equation is ０ +a １ x １ +a ２ x ２ +a ３ x ３ +a ４ x ４ +a ５ x ５ +a ６ x ６ +a ７ x ７ +a ８ x ８ If a １ x １ , a ２ x ２ , a ３ x ３ , a ４ x ４ , a ５ x ５ , a ６ x ６ , a ７ x ７ and a ８ x ８ These are eight patterns of linear regression equations, each excluding

[0113] Coefficient of determination of the linear regression equation R2 ｉ ' is the coefficient of determination R2 of all linear regression equations when any one of the terms of the linear regression equation in equation (10) is excluded. ｉThat is, the linear regression equation is a ０ +a １ x １ +a ２ x ２ +a ３ x ３ +a ４ x ４ +a ５ x ５ +a ６ x ６ +a ７ x ７ +a ８ x ８ If a １ x １ , a ２ x ２ , a ３ x ３ , a ４ x ４ , a ５ x ５ , a ６ x ６ , a ７ x ７ and a ８ x ８ The coefficient of determination R2 of the 8 patterns of linear regression equations excluding １ ~R2 ８ is.

[0114] In the step of evaluating the contribution of each term of the regression model (step S104), the contribution rate of each term in the obtained linear regression equation is very small, so it is difficult to compare only the contribution rate. ｉ The coefficient of determination R2 of the linear regression equation, which is the regression model i', excluding one ｉ ' and estimate the coefficient of determination R2 of regression model j. ｊ By comparing it with the above, it is easy to compare and determine whether it is better to keep the terms in the linear regression equation.

[0115] The regression model generating device 100 calculates the coefficient of determination R2 ｊ is the coefficient of determination R2 ｉ If it is determined that this is the case (step S109: YES), it is determined that regression model j is more accurate than regression model i or regression model (i+1), and regression model j is adopted (second adoption process: step S110).

[0116] In the second determination step (step S109), the regression model generating device 100 determines the coefficient of determination R ｉ is the coefficient of determination R2 ｉ If it is determined that the difference is less than 1 (step S109: NO), it is determined that regression model i is more accurate than regression model j or regression model (i+1), and regression model i is adopted (fourth adoption step: step S113).

[0117] As described above, after adopting regression model i, regression model (i+1), or regression model j, the regression model generation device 100 terminates the regression model generation process and provides the finally obtained linear regression equation to the ligand searching device 200 as the generated regression model.

[0118] [Ligand Search Process] The ligand search process S20 will be described. The ligand search process S20 is executed by the ligand search device 200 described above, and may be realized by the processor of the ligand search device 200, for example.

[0119] 8 is a flowchart showing an example of the ligand search process S20. As shown in FIG. 8, in the ligand search process S20, the ligand search device 200 searches for a physical property value x ｉ (Acquisition process of physical properties: step S201).

[0120] Next, the ligand searching device 200 uses a regression model to find the acquired physical property value x ｉ From log １０ [T] is calculated (calculation process of the logarithm of the predicted curing time: step S202).

[0121] Next, the ligand searching apparatus 200 calculates the calculated log １０ Determine whether [T] is equal to or less than a predetermined value (for example, 5.0), and calculate log １０ A search is made for a curing accelerator having a ligand whose [T] is equal to or less than a predetermined value (searching step: step S203).

[0122] Calculated log １０When a curing accelerator having a ligand for which [T] is equal to or less than a predetermined value (for example, 5.0) is detected, the ligand searching device 200 determines the detected curing accelerator as the curing accelerator to be searched for and outputs it.

[0123] According to the ligand search method of this embodiment, the log １０ By searching for a ligand based on the value of [T], it is possible to search for a ligand of a curing accelerator that can shorten the curing time when curing a radical polymerizable resin at room temperature. Therefore, the method for searching for a ligand according to this embodiment can be used to select a curing accelerator that can shorten the curing time when curing a radical polymerizable resin at room temperature.

[0124] As described above, the ligand searching system 1 includes a regression model generating device 100 and a ligand searching device 200. The regression model generating device 100 generates a regression model that predicts the logarithm of the curing time from the physical property values ​​of the curing accelerator. The ligand searching device 200 uses the generated regression model to predict the logarithm of the curing time from the physical property values ​​x of the curing accelerator having various ligands. ｉ Log predicted from １０ Based on [T], the ligand searching system 1 can search for a ligand of a curing accelerator that can shorten the curing time when curing a radical polymerizable resin at room temperature.

[0125] The ligand searching system 1 can search for the ligand of the curing accelerator with higher accuracy, and therefore can select a curing accelerator that is effective for curing the radical polymerizable resin at room temperature in a shorter time.

[0126] <Curing Accelerator> The curing accelerator searched for by the ligand searching method according to this embodiment will be described. The curing accelerator contains an organic acid salt of cobalt and a ligand, which can accelerate the curing of the radical polymerizable resin.

[0127] Examples of radical polymerizable resins include unsaturated polyester resins and vinyl ester resins.

[0128] Examples of organic acid salts of cobalt include cobalt octylate, cobalt naphthenate, cobalt stearate, and cobalt neodecanoate.

[0129] The ligand contains an amine structure or a β-diketone structure, and examples of the ligand include those represented by the above general formulae (i) to (xiv).

[0130] The ligand may be synthesized using a generally known synthesis method, such as those described in JP 2004-26765 A, German Patent No. 1089760, JP 2013-522343 A, WO 95 / 21525, Calter et al., Journal of Organic Chemistry 2004, vol. 69 (4), pp. 1270-1275, and Hilgenkamp et al., Tetrahedron 2001, vol. 57, #42, pp. 8793-8800.

[0131] The physical property values ​​of the curing accelerator are the average distance between oxygen atoms and cobalt atoms in dormant species of a complex in which a ligand is coordinated to a cobalt atom, and the occupancy of ligands present within 3 Å of a cobalt atom in dormant species of a complex in which a ligand is coordinated to a cobalt atom.

[0132] The curing time is, for example, the time required to cure the radical polymerizable resin at room temperature (for example, 25° C.) using a ketone peroxide as a curing agent and the above-mentioned curing accelerator.

[0133] When the radical polymerizable resin is, for example, an unsaturated polyester resin, the curing time of the unsaturated polyester resin can be measured in accordance with, for example, JIS K 6901 ("Testing methods for liquid unsaturated polyester resins," "Room temperature curing characteristics (exothermic method)").

[0134] The liquid unsaturated polyester resin is a solution in which the unsaturated polyester resin is dissolved in a monomer (for example, styrene) that can be polymerized with the unsaturated polyester resin.

[0135] Although the embodiments have been described above, they are presented as examples and the present invention is not limited to the above embodiments. The above embodiments can be implemented in various other forms, and various combinations, omissions, substitutions, modifications, etc. can be made without departing from the spirit of the invention. These embodiments and their modifications are included within the scope and spirit of the invention, and are also included in the inventions and their equivalents as set forth in the claims.

[0136] The present embodiment will be described in more detail below with reference to examples, but the present embodiment is not limited to these examples.

[0137] Reference Examples 1 to 13 [Preparation of Ligand Search System] A ligand search system 1 having a regression model generation device 100 and a ligand search device 200 shown in FIG. 1 was prepared.

[0138] [Generation of Regression Model] (Preparation of Ligand) As substance data used to generate the regression model, the ligands L of Reference Examples 1 to 13 below were used. １ ~L １３ was obtained.

[0139] ((Ligands of Reference Examples 1 to 13)) Ligand of Reference Example 1 (L １ ):

[0140] The ligand of Reference Example 2 (L ２ ):

[0141] The ligand of Reference Example 3 (L ３ ):

[0142] The ligand of Reference Example 4 (L ４ ):

[0143] The ligand of Reference Example 5 (L ５ ):

[0144] The ligand of Reference Example 6 (L ６ ):

[0145] The ligand of Reference Example 7 (L ７ ):

[0146] The ligand of Reference Example 8 (L ８ ):

[0147] The ligand of Reference Example 9 (L ９ ):

[0148] The ligand of Reference Example 10 (L １０ ):

[0149] The ligand of Reference Example 11 (L １１ ):

[0150] The ligand of Reference Example 12 (L １２ ):

[0151] The ligand of Reference Example 13 (L １３ ):

[0152] The ligands (L １ ~L １３ ) are specific examples of ligands represented by any one of the above general formulas (I) to (IV). Ligands of Reference Examples 2 and 11 to 13: specific examples of ligands represented by general formula (III) Ligands of Reference Examples 3 to 6: specific examples of ligands represented by general formula (I) Ligands of Reference Examples 7, 8, and 10: specific examples of ligands represented by general formula (II) Ligands of Reference Examples 1 and 9: specific examples of ligands represented by general formula (IV)

[0153] (Creation of Dormant Species) Using the regression model generation device 100, the ligands L of Reference Examples 1 to 13 were １ ~L １３ The dormant species to which the ligand L of Reference Examples 1 to 13 is coordinated is structurally optimized. １ ~L １３ We prepared a structurally optimized dormant species in which

[0154] (Calculation of Physical Property Values) Using the regression model generation device 100, the ligands L of Reference Examples 1 to 13 were calculated. １ ~L １３Quantum chemical calculations were performed on the structure-optimized dormant species coordinated with the following physical property values ​​x using quantum chemical calculation software (Gaussian09 manufactured by Gaussian Corporation). ｉ In Gaussian09, based on the density functional theory, the spin is unrestricted and the functional is RωB97XD, and the basis function is 6-31. ＋ G(d) was used. Frequency analysis was performed in all calculations, and it was confirmed that the number of imaginary vibrations was 0, and it was determined that the potential surface converged to the bottom. ((Physical property value x ｉ )) Physical property value x １ : NBO charge of the cobalt atom in the dormant species of the complex in which the ligand is coordinated to the cobalt atom. Physical property value x ２ : Occupancy of ligands present within 3 Å from the cobalt atom in dormant species of a complex in which a ligand is coordinated to the cobalt atom

[0155] (Calculation of the measured curing time T and the logarithm of the measured curing time T) The measured curing time T and the logarithm (log １０ The actual curing time T was obtained from the material database mentioned above.

[0156] (Estimation of coefficients of linear regression equation (1a)) The regression model generation unit 120 is used to estimate the coefficients of the physical property value x acquired from the physical property value acquisition unit 110. ｉ and log １０ A multiple regression analysis is performed on T, and the coefficient a of the linear regression equation (1a) is used as the regression model. ０ , a １ , a ２ Specifically, the ligands L of Reference Examples 1 to 13 were estimated as follows: ｊ The physical property value x obtained for (j = 1 to 13) ｉ is substituted into the linear regression equation of the estimation target, and the ligand (L ｊ ) is the formula [T]_L ｊ (j = 1 to 13) was obtained. [T]_L ｊ Coefficient a ｉ (i = 0 to 2) is [T]_L ｊ The coefficient of determination R, which is the square of the correlation coefficient, was determined so as to reproduce the actual hardening time T of each of the above. ２By log １０ The reproducibility of T was evaluated. The coefficient of determination R ２ The closer to 1, the more correlation there is between the two values. ２ The reproducibility was evaluated when the logarithm was 0.6 or more. １０ T and log １０ Coefficient of determination R of [T] ２ Using an evolutionary algorithm, the coefficient a of the linear regression equation (1a) is adjusted so that ｉ (i = 1 and 2) was determined. At that time, the coefficient a １ and a ２ The reduction was done one by one. ０ is log １０ [T] and log １０ The absolute value of T was determined to be the same. Estimation of the regression coefficients using the evolutionary algorithm was performed using the evolutionary algorithm in the solver included in Excel (registered trademark). The calculation conditions were: convergence 0.0001, mutation rate 0.075, population size 100, random seed 100, and maximum time without improvement 300. The regression coefficients were estimated using the solver.

[0157] (Derivation of Linear Regression Equation (2a)) Ligand L of Reference Examples 1 to 13 ｉ The NBO charge x of the cobalt atom in the dormant species of the complex in which the ligand is coordinated to the cobalt atom is obtained for １ , the occupancy rate x of the ligands present within 3 Å from the cobalt atom in the dormant species of the complex in which the ligands are coordinated to the cobalt atom ２ Each physical property value x ｉ A linear regression equation was derived for

[0158] Ligands L of Reference Examples 1 to 13 ｉ The NBO charge x of the cobalt atom in the dormant species of the complex in which the ligand is coordinated to the cobalt atom is obtained for １ , the occupancy rate x of the ligands present within 3 Å from the cobalt atom in the dormant species of the complex in which the ligands are coordinated to the cobalt atom ２ Each physical property value x ｉFor log, the following linear regression equation (2a) was derived: １０ [T](x ｉ )=5.20950x １ -0.50360x ２ +47.001...(2a)

[0159] (logarithm of predicted cure time [T]) １０ Using the derived formula (2a) above, the physical property value x of the dormant species of the complexes in which the ligands of Reference Examples 1 to 13 are coordinated to the organic acid salt of cobalt is calculated. ｉ For log １０ [T] was calculated. １０ T and log １０ The relationship between [T] and [T] is shown in Figure 9. In Figure 9, the logarithm of Reference Examples 1 to 13 １０ To match the T, log １０ Multiply [T] by a constant, move the intercept parallel, and １０ T and log １０ The slope of the linear approximation equation showing the relationship with [T] was adjusted to be 1.0. As shown in FIG. 9, the coefficient of determination R ２ was approximately 0.86, the linear approximation of equation (2a) is log １０ Logarithm with sufficient correlation to T １０ It can be said that [T] was predicted.

[0160] The physical property values ​​x of the dormant species of the complexes in which the ligands of Reference Examples 1 to 13 are coordinated １ , x ２ , log １０ [T], measured curing time T and log １０ T is shown in Table 1.

[0161]

[0162] (Generation of Regression Model) The linear regression equation expressed by the above equation (2a) was used as the regression model.

[0163] [Ligand Search] Using a specific example of a ligand represented by the following general formula (II) (ligand of evaluation example h1) as test data, the ligand searching device 200 was used to search for the ligand of evaluation example h1 using the log１０ It was confirmed whether the ligand had a value of [T] of 5.0 or less. Specifically, in the same manner as in the ligands of Reference Examples 1 to 13, the physical property value x ｉ After calculating, substitute into equation (2a) to obtain log １０ [T] was obtained.

[0164] Ligand of Evaluation Example h1:

[0165] The ligand searching device 200 calculates the logarithm of the ligand search result when the above-mentioned ligand is used in the ligand searching section 230. １０ Calculate [T] and calculate the calculated log １０ The search results were obtained based on the value of [T]. １０ When the value of [T] is 5.0 or less, the ligand is effective in shortening the curing time (determined as "A" in Table 2), and the log １０ When the value of [T] exceeded 5.0, the ligand was judged to be ineffective (determined as "B" in Table 2). １０ An example of [T] and the search results is shown in Table 2.

[0166]

[0167] From Table 2, it can be seen that the redox initiator composed of ketone peroxide / cobalt organic acid salt obtained using the ligand of Evaluation Example h1 has a log １０ It was confirmed that [T] could be reduced to 3.0 or less.

[0168] The ligand of Evaluation Example h1 is a ligand included in general formula (II) among the general formulas (I) to (IV) above, and has a logarithm that approximately corresponds to the linear regression equation represented by the above formula (2a). １０ [T] was obtained. The linear regression equation represented by the above formula (2a) is a formula calculated based on the ligands contained in the general formula (II) and the ligands contained in the above general formulas (I), (III) and (IV). Therefore, for the ligands contained in the general formula (II) other than Evaluation Example h1 and the ligands contained in the above general formulas (I), (III) and (IV), a logarithm that approximately corresponds to the linear regression equation represented by the above formula (2a) was also obtained. １０ It can be said that [T] is obtained.

[0169] Therefore, when the linear regression equation represented by the above formula (2a) is used as a regression model, if a redox initiator consisting of ketone peroxide / cobalt organic acid salt is used among the ligands represented by the above general formulas (I) to (IV), the log １０ It can be said that it is possible to search for a ligand that can reduce the curing time by making the value of [T] equal to or less than 5.0. Therefore, it can be said that the method for searching for a ligand according to this embodiment can be used to search for a ligand of a curing accelerator that can reduce the curing time when curing a radical polymerizable resin at room temperature, among the ligands of the general formulas (I) to (IV).

[0170] [Evaluation of extrapolation ability of generated regression model] (Derivation of linear regression formula) The ligand of Evaluation Example h1 was used as training data, and the ligand of Reference Example 5 (L ５ ), the ligand of Reference Example 7 (L ７ ) or the ligand of Reference Example 8 (L ８ ) is used as test data, and the logarithm of the predicted curing time [T] (log １０ A linear regression equation was derived in the same manner as in the calculation of the ligand (L ５ ), the ligand of Reference Example 7 (L ７ ) and the ligand of Reference Example 8 (L ８ ) Each acquired physical property value x ｉ (In the dormant species of the complex in which the ligand is coordinated to the cobalt atom, the NBO charge x of the cobalt atom １ , the occupancy rate x of the ligands present within 3 Å from the cobalt atom in the dormant species of the complex in which the ligands are coordinated to the cobalt atom ２ ), the following linear regression equations (2a-1) to (2a-3) were derived. ５ ) and the equation (2a-2) is a linear regression equation when the ligand (L ７ ) and the equation (2a-3) is a linear regression equation when the ligand (L ８ ) is the linear regression equation when log １０ [T](x ｉ )=3.41227x １ -0.449646x２ +43.239 ... (2a-1) log １０ [T](x ｉ )=3.777831x １ -0.470965x ２ +44.97...(2a-2) log １０ [T](x ｉ )=2.81583x １ -0.448084x ２ +43.53...(2a-3)

[0171] (logarithm of predicted cure time [T]) １０ Calculation of [T]) Using the derived formulas (2a-1) to (2a-3), the ligand (L ５ ), the ligand of Reference Example 7 (L ７ ) and the ligand of Reference Example 8 (L ８ ) １０ [T] was calculated. ５ , L ７ , L ８ When using log as test data １０ [T] and log １０ The relationship between the ligand (L ５ ), (L ７ ) and (L ８ ) as test data are shown by white circles. ５ ), (L ７ ) and (L ８ ) are expressed by the linear regression formula (2a-1) to (2a-3) １０ It was confirmed that the ligand [T] was 5.0 or less. ５ ), (L ７ ) and (L ８ ) for ligands with short curing times, the prediction accuracy is sufficient, and the extrapolation capability of the generated regression model is high.

[0172] This application claims priority based on Japanese Patent Application No. 2024-8833 filed with the Japan Patent Office on January 24, 2024, and Japanese Patent Application No. 2024-45903 filed with the Japan Patent Office on March 22, 2024. All contents of these prior applications are incorporated by reference.

[0173] REFERENCE SIGNS LIST 1 Ligand search system 100 Regression model generation device 110, 210 Physical property value acquisition unit 120 Regression model generation unit 200 Ligand search device 220 Curing time calculation unit 230 Ligand search unit 300 Substance database

Claims

1. A method for searching for ligands that coordinate to cobalt, wherein a processor uses the general formula (I): (In the formula, R １ , R ２ are each independently a hydrogen atom or an alkyl group having 1 to 10 carbon atoms, and R ３ to R ７ are each independently a hydrogen atom, an alkyl group having 1 to 10 carbon atoms, an alkoxy group having 1 to 10 carbon atoms, an alkoxycarbonyl group having 2 to 11 carbon atoms, an aryl group having 6 to 12 carbon atoms, a hydroxy group, an amino group, a carboxy group, a cyano group, a nitro group, a formyl group or a halogen atom.) A ligand represented by the general formula (II): (In the formula, R １１ to R １３ are each independently a hydrogen atom, an alkyl group having 1 to 10 carbon atoms, an alkoxy group having 1 to 10 carbon atoms, an aryl group having 6 to 12 carbon atoms, an amino group, a carboxy group, a cyano group or a halogen atom, and R １２ and R １３ may jointly form a ring.) A ligand represented by the general formula (III): (In the formula, R ２１ to R ２３ are each independently a hydrogen atom, an alkyl group having 1 to 10 carbon atoms, an alkoxy group having 1 to 10 carbon atoms, an alkoxycarbonyl group having 2 to 11 carbon atoms, an aryl group having 6 to 12 carbon atoms, a hydroxy group, an amino group, a carboxy group, a cyano group, a nitro group, a formyl group or a halogen atom.) A ligand represented by the general formula (IV): (In the formula, R ３１ to R ３５ are each independently a hydrogen atom, an alkyl group having 1 to 10 carbon atoms, an alkoxy group having 1 to 10 carbon atoms, an alkoxycarbonyl group having 2 to 11 carbon atoms, an aryl group having 6 to 12 carbon atoms, a hydroxy group, an amino group, a carboxy group, a cyano group, a nitro group, a formyl group, a halogen atom or a pyridine ring.) In a ligand represented by any of the ligands, a step of searching for a ligand in which the logarithm of the predicted curing time [T] [seconds] calculated by the following formula (1) is equal to or less than a predetermined value is performed. log １０ [T]=α ０ +α １ x １ +α ２ x ２ ・・・(1) (In the formula, x １ is the NBO charge of the cobalt atom in the dormant species of the complex in which the ligand coordinates to the cobalt atom, and x ２ is the occupancy of the ligand existing within 3 Å from the cobalt atom in the dormant species of the complex in which the ligand coordinates to the cobalt atom, and a ０ , a １ and a ２ are coefficients.) 2. The formula (1) is the formula (2): log １０ [T](x ｉ ) = 5.20950x １ −0.50360x ２ + 47.001 ··· (2), and the predetermined value is 5.

0. The method for searching for a ligand according to claim 1.

3. The dominant species of the complex in which the ligand represented by the general formula (I), (III) or (IV) is coordinated to cobalt is a Werner complex of cobalt (II) in which two molecules of the carboxylate ligand, which is a bidentate ligand, and two molecules of the ligand represented by the general formula (I), (III) or (IV), which is a monodentate ligand, are coordinated, and is the dominant species of cobalt (III) in which an OH radical or a Me radical is coordinated instead of one molecule of the ligand represented by the general formula (I), (III) or (IV). The method for searching for a ligand according to claim 1 or 2.

4. The dominant species of the complex in which the ligand represented by the general formula (II) is coordinated to cobalt is a Werner complex of cobalt (III) in which three molecules of the ligand represented by the general formula (II), which is a bidentate ligand, are coordinated, and is the dominant species of cobalt (III) in which an OH radical or a Me radical is coordinated instead of one coordination of one molecule of the ligand represented by the general formula (II). The method for searching for a ligand according to any one of claims 1 to 3.

Images