=Paper=
{{Paper
|id=Vol-2523/paper35
|storemode=property
|title=
Databases on the Properties of Substances and Computer-Assisted Design of Inorganic Compounds
|pdfUrl=https://ceur-ws.org/Vol-2523/paper35.pdf
|volume=Vol-2523
|authors=Nadezhda Kiselyova,Victor Dudarev,Andrey Stolyarenko
|dblpUrl=https://dblp.org/rec/conf/rcdl/KiselyovaDS19
}}
==
Databases on the Properties of Substances and Computer-Assisted Design of Inorganic Compounds
==
https://ceur-ws.org/Vol-2523/paper35.pdf
Databases on the Properties of Substances and
Computer-Assisted Design of Inorganic Compounds
Nadezhda Kiselyova1[0000-0001-7243-9096], Victor Dudarev2[0000-0002-3583-0704]
and Andrey Stolyarenko1
1 A.A. Baikov Institute of Metallurgy and Materials Science of RAS (IMET RAS), Moscow,
119334, Russia
2 National Research University Higher School of Economics, Moscow, 109028, Russia
kis@imet.ac.ru
Abstract. The virtually integrated distributed system of databases on the proper-
ties of inorganic substances and materials of the A.A. Baikov Institute of Metal-
lurgy and Materials Science, Russian Academy of Sciences is considered. The
information-analytical system for automation of process of new inorganic com-
pounds computer-assisted design based on machine learning methods usage for
search for regularities in information of the databases on inorganic substances
and materials properties is discussed. The results this system application for com-
pound design that have not yet been synthesized are presented.
Keywords: Database, Inorganic Substance and Material, Machine Learning.
1 Introduction
Modern information technologies have made it possible to systematize and make avail-
able a huge array of data accumulated by chemistry over the centuries. Chemists and
materials scientists make extensive use of the rich capabilities provided by numerous
databases (DB), including the database on the properties of inorganic substances and
materials (DBs PISM) [1], containing not only publications [2-5], but also data on the
properties of substances [1,6-11]. More detailed information on the information re-
sources of inorganic chemistry is given in the IRIC database developed by us [12].
Information service does not limit the capabilities of the developed databases. One
of the ways to make rational use of information on substances is the search for regular-
ities that connect the properties of substances with the parameters of components. The
objective existence of such regularities is a consequence of the Periodic Law. However,
numerous attempts to present the desired complex regularities in an analytical form, as
a rule, were unsuccessful, especially in the case of multicomponent substances. The
methods for finding such complicated regularities in the data, based on the ideas of
machine learning, were developed. In the mid-sixties, the idea of using machine learn-
ing to find regularities, that relate the properties of inorganic compounds to the param-
eters of components, was first proposed in our Institute of Metallurgy and Materials
Science (IMET) [13]. Already the first calculations allowed us to find the relationship
Copyright © 2019 for this paper by its authors. Use permitted under Creative
Commons License Attribution 4.0 International (CC BY 4.0).
358
between the properties of binary compounds and the parameters of chemical elements,
as well as to use the found regularities to predict compounds not yet obtained with an
accuracy of about 90% [14]. Our further research in this area was associated with the
use of more advanced machine learning programs [15–17] and complication of the
composition of the compounds being predicted [18–19].
2 Integrated database system of IMET RAS on the properties
of inorganic substances and materials
The source of information for the use of machine learning methods is the DBs PISM.
In contrast to the databases usually used for these purposes, the information systems
developed by the Institute of Metallurgy and Materials Science, Russian Academy of
Sciences [1, 6, 11], by their functional structure, are focused on the selection of infor-
mation for machine learning, which significantly reduces the time for preparation and
analyzing the necessary data.
One of the most important problems in the application of machine learning to inor-
ganic chemistry is the inconsistency of data obtained by different researchers. In this
regard, the selection of information for machine learning is carried-ot by qualified ex-
perts in this subject area. This procedure is facilitated by providing the experts with the
full texts of publications contained in our DBs PISM, from which examples are selected
for machine learning, as well as through special programs for detecting sharply distin-
guished objects (outliers).
Now the integrated system of the DBs PISM includes the information systems de-
veloped in the IMET [1, 6, 11]: on the phase diagrams of semiconductor systems (Dia-
gram), the properties of the acoustooptical, electro-optical, and nonlinear optical sub-
stances (Crystal), the band gap of inorganic substances (Bandgap), the properties of
inorganic compounds (Phases), and the properties of chemical elements (Elements), the
AtomWork database on the properties of inorganic substances, developed at the Na-
tional Institute for Materials Science (NIMS, Japan) [8], and the TKV on substances
thermal constants, developed in the Joint Institute for High Temperatures of RAS and
Lomonosov Moscow State University cooperation.
The Phases database on the properties of inorganic compounds currently contains
information on the properties of approximately 54000 ternary compounds and more
than 34000 quaternary compounds, collected using more than 36000 publications. It
includes brief information about the most common properties of inorganic compounds:
crystal chemical (the type of crystal structure with indication of the temperature and
pressure above which this structure is implemented, the crystal system, the space group,
the number of formula units in the unit cell, and the lattice parameters) and thermo-
physical (the melting type and temperature, the temperature of decomposition of the
compound in solid or gaseous phases, and the boiling point at atmospheric pressure)
data. In addition, the database contains information on the superconducting properties
of compounds. This database is available on the Internet for registered users [11].
The Elements database includes information about 90 of the most common proper-
ties of chemical elements: the thermal (the melting and boiling points at 1 atm and the
359
standard values of thermal conductivity, molar heat capacity, enthalpy of atomization,
entropy, etc.), size (the ionic, covalent, metal, and pseudopotential radii, the atomic
volume, etc.), and other physical properties (the magnetic susceptibility, electrical con-
ductivity, hardness, density, etc.); etc. The database is available on the Internet [11].
The Diagram database contains data on phase P,T,x-diagrams of binary and ternary
semiconductor systems and the physicochemical properties of phases formed in them,
collected and evaluated by highly qualified experts. The Diagram database is available
on the Internet for registered users [11].
The Bandgap database includes information about the band gap of more than 3600
inorganic substances and is available on the Internet [11]. It has English version only.
The Crystal database includes information about the piezoelectric (piezoelectric co-
efficients, elastic constants, etc.), nonlinear optical (nonlinear optical coefficients, the
Miller tensor components, etc.), crystal chemical (the type of the crystal structure, crys-
tal system, space and point groups, the number of formula units per unit cell, and the
crystal lattice parameters), optical (refractive indices, the transparency band, etc.), and
thermal (melting point, specific heat, thermal conductivity, etc.) properties of more than
140 acousto-optical, electro-optical, and nonlinear optical materials, collected and eval-
uated by highly qualified experts in the subject area. It has Russian and English versions
available for registered users on the Internet [11].
The AtomWork Inorganic Material Database (NIMS, Japan) contains information
about more than 82000 crystal structures, 55000 values of the properties of materials,
and 15000 phase diagrams; it is also available on the Internet [8].
The TKV DB on substances thermal constants contains information, available online
from the Internet, on about 27 thousand substances formed by all chemical elements.
The complex integration approach that combines integration at data and user inter-
faces level is applied to these database integration [20]. The special single entry point
allows a search for the all data on certain substance from different DBs.
3 Inorganic Compounds Computer-Assisted Design System
Machine learning procedure involves several stages:
1. The objects selection for machine learning.
2. The attribute description formation (including the most informative attributes selec-
tion and filling attribute values gaps also).
3. The best ML algorithms selection.
4. Machine learning including application of algorithms ensembles and collective so-
lution synthesis in a case of several algorithms usage.
5. ML quality estimation.
6. New objects prediction and results interpretation.
The special information-analytical system (IAS), which, in addition to the information
service for professionals, is designed to search for regularities in big chemical data and
computer design of inorganic compounds was developed in IMET [21]. It includes (Fig.
1), along with the integrated system of DBs PISM, a subsystem of information analysis
360
and predictions, bringing together a set of programs of machine learning, a base of
found regularities (the knowledge base), a base of predictions of the possibility of form-
ing and properties of inorganic compounds that have not been yet synthesized, and a
management subsystem.
Fig. 1. The information-analytical system structure for inorganic compounds design.
3.1 Subsystems for searching for classifying regularities and predictions
In the development of this subsystem, the most important task was the selection of the
most appropriate mathematical methods for searching regularities in chemical data.
Typically, this task is performed by the trial-and-error method. In the selection of ma-
chine learning methods for analysis of chemical information, many years’ experience
in the application of these methods to inorganic compounds design was taken into ac-
count [18]. The following methods and programs have been selected:
─ a wide range of algorithms of the Recognition multifunctional system, developed at
the Computing Center of the Russian Academy of Sciences [16] and bringing to-
gether, in addition to well-known techniques, the algorithms of pattern-recognition
(based on calculation of estimates), voting algorithms based on deadlock tests, vot-
ing algorithms based on logical regularities, weighted statistical voting algorithms,
etc.;
361
─ a ConFor computer system for training a computer in the procedure for concept for-
mation [15], which is based on an original organization of data in the computer
memory in the form of growing pyramidal networks.
As a rule, it is not possible to specify in advance which algorithm would be the most
efficient for solving a particular problem. Therefore, it seems promising to apply the
methods of prediction by algorithms ensembles. In a collective decision creation, the
possible prediction errors of individual algorithms can be compensated in many cases
by correct results of other algorithms. Based on this, we included programs that imple-
ment different strategies for collective decision-making, for example, the Bayesian
method, methods using clustering and selection, decision templates, logical correction,
the method of a convex stabilizer, the Woods dynamic method, committee methods,
etc., [16] in the developed IAS [21].
3.2 Subsystem for searching the classifying properties of components
For the selection of informative properties of the chemical compound’s components,
we included programs based on algorithms [22-24] in the IAS. The selection of the
properties of the components, the most informative for the classification of substances,
has a double meaning. On the one hand, it drastically decreases the volume of the in-
formation analyzed, which for multicomponent substances comprise hundreds of prop-
erties of elements and simpler compounds, as well as functions of these properties. On
the other hand, the selection of properties of the components most important for the
classification of chemical substances, enables the physical interpretation of the classi-
fying regularities, which enhances the credibility of the predictions obtained and find-
ing substantial causal links between the parameters of the objects and the development
of the physical and chemical models of phenomena.
3.3 Visualization subsystem
This subsystem facilitates the results interpretation, which constructs the projections of
the points corresponding to the compounds in two-dimensional spaces of the properties
of components, including not only the initial parameters but also user-specified alge-
braic functions of these parameters.
3.4 Knowledge and prediction bases
The knowledge base contains the obtained classifying regularities. The prediction base
contains the results of previous computer experiments, as well as links to operation
information stored in the knowledge base. Using the prediction base helped to improve
the functionality of the databases on the properties of inorganic substances and materi-
als, developed at the IMET, by providing the user with not only known data about al-
ready studied substances but also predictions for inorganic compounds not yet synthe-
sized and evaluations of their properties.
362
3.5 Management subsystem
The management subsystem organizes the computing process, ensures interaction be-
tween the functional subsystems of the IAS, and provides access to the system on the
Internet. In addition, the management subsystem provides the expert with software for
data preparation for analysis, outputting reports, and implementation of other service
functions. In particular, we developed a special subsystem to retrieve information from
the database, which, after evaluation of the expert, is used to learn the computer, and to
prepare it for further analysis. It gives the expert the capability to edit the found infor-
mation and to form training samples for analysis. In the latter case, the expert marks
only the selected properties of the components in a special table (menu), and the sub-
system for the sample preparation for analysis retrieves the selected property values
from the Elements database. If needed, the algebraic functions of the initial properties
are formed in the subsystem and the description of the compounds is assembled in the
form of an Excel table, which is then input to the prediction subsystem. The subsystem
of result delivery is intended to make predictions in a tabular form conventional to
chemists and materials scientists.
4 Use the IAS for predicting new compounds and evaluation of
their properties
The machine learning application allowed a search for inorganic compounds formation
regularities, a prediction of thousands not yet synthesized substances and some their
properties evaluation using obtained regularities. This approach efficiency to inorganic
compounds design can be illustrated by comparison of the predictions results with
newer experimental data obtained after publication of our predictions.
4.1 Prediction of the TiNiSi crystal structure type for compounds with the
composition ABAl
The equiatomic aluminides are of interest for the search for new magnetic materials.
Thirty years ago, the prediction of new compounds of this type was carried out by us
[25]. The algorithm based on the growing pyramidal networks learning (GPNL) [15]
was used in the search for the criteria of this crystal structure type formation at ambient
conditions. The learning set contained 39 examples of the compounds ABAl (hereinaf-
ter, A and B are various chemical elements) with the TiNiSi crystal structure type and
57 examples of the compounds with the structures different from TiNiSi. The following
properties of elements A and B (attributes) were chosen for description of intermetal-
lics: the distribution of electrons in the energy levels of isolated atoms of the chemical
elements, the first three ionization potentials, the metal radii by Bokii and Belov, the
standard entropies of individual substances, the melting points, the number of complete
electronic shells, the number of electrons in incomplete s-, p-, d- or f-electronic shells
for the atoms of elements.
363
Table 1. Part of a table illustrating the prediction of the crystal structure type TiNiSi for com-
pounds with the composition ABAl [25].
A La Ce Pr Nd Pm Sm Eu Gd Tb Dy Ho Er Tm Yb Lu
B
Ru - O - - - + - + + + + + + - -
Rh O O - + - © + + © + © -
Os + + + + + + + + + + + + + + +
Ir © © © © + © + © © © © © © + ©
Pt © © © © + + ©
Table 1 is a result of comparing the predictions for each sets of properties and con-
tains the comparison results of our predictions with newer experimental data. The fol-
lowing notations are used: + - prediction of the TiNiSi crystal structure type; - - predic-
tion of the absence of the TiNiSi crystal structure type; - a compound with composi-
tion ABAl has the TiNiSi crystal structure type and this fact was used for machine
learning; © - prediction of the TiNiSi crystal structure type was confirmed experimen-
tally; O - prediction of the crystal structure type different from TiNiSi was confirmed
experimentally; - prediction of the crystal structure type different from TiNiSi was
not confirmed experimentally; here and in other Tables the blank spaces correspond to
the disagreement of the predictions with the use of different sets of the component’s
properties; all data and predictions are given for the substances under ambient condi-
tions. A comparison of our predictions with newer experimental data has shown that
the prediction error is lower than 12 %.
4.2 Design of compounds with composition ABX2 (X – S, Se, or Te)
The chalcogenides with composition ABX2 are a class of compounds that is promising
for the search for new semiconducting and nonlinear optical materials. Taking into ac-
count the perspective of these compounds practical use the design of their not yet syn-
thesized analogues was made [26]. Previously we predicted new compounds of this
composition [27] also.
The task solution was subdivided into to stages: (1) prediction of the formation of
compounds with composition ABX2; and (2) prediction of the crystal structure type of
these compounds under ambient conditions.
Prediction of the formation of compounds with composition ABX 2. Data on 667
examples of the formation of ABX2 (X = S, Se, or Te) and 504 examples of the absence
of this composition compounds in the systems A2X–B2X3 and AX–BX under ambient
conditions were used for machine learning. The data were taken from the DB Phases.
84 properties of the elements A, B, and X, whose values were taken from the DB Ele-
ments, were used for the compounds representation in the computer memory.
For the data analysis, several machine learning algorithms that are included into IAS
were used. The learning quality was estimated on the basis of examination recognition
in the mode of cross-validation. The analysis of the results using various algorithms has
364
shown that the best predictions under cross-validation have been obtained for the deci-
sion tree method (DT) [16] (accuracy of prediction being 72%), the logical regularities
voting algorithm (LoReg) [16] (accuracy of prediction being 67.3%), and the deadlock
test algorithm [16] (accuracy of prediction being 67.6%). These algorithms have been
used for collective decisions using the committee method, in which the resulting pre-
diction is calculated as an average arithmetic value of predictions obtained using dif-
ferent algorithms [16]. Using this procedure, the compound’s formation predictions
were obtained.
Prediction of the crystal structure type of compounds ABX 2. Data on 158 exam-
ples of the formation of ABX2 with the crystal structure under ambient conditions α-
NaFeO2, 44 compounds with NaCl structure, 47 compounds with chalcopyrite struc-
ture, and 24 compounds with TlSe structure were used for machine learning. The same
properties were used for the compound’s representation.
The problem was solved in two ways. In the first case, multi-class learning and pre-
diction, where the cumulative information on the four above-mentioned crystal phases
has been used, was applied. In the second case, four problems of the dichotomy were
solved - division into two classes, e.g., class 1, compounds with chalcopyrite crystal
structure, and class 2, compounds with another structure. The results of predictions
were compared, and a decision was made if the predictions obtained by multi-class
prediction and dichotomies did not contradict each other. The results are summarized
in Table 2.
Table 2. Part of a table illustrating the prediction of the crystal structure type for compounds
with the composition ABX2 (X – S, Se, or Te) [26].
X S Se Te
A Li Na K Cu Rb Ag Cs Tl Li Na K Cu Rb Ag Cs Tl Li Na K Cu Rb Ag Cs Tl
B
B #5 #5 #5 ©3 #5 3 #5 #5 1 1 #5 1 3 1 4 1 1 4 1 1 4
Al #5 #3 #3 ©4 #5 #4 #5 #3 #3 #5 #4 #4 #3 #3 4 #4
Sc #1 #1 ©1 #5 ©1 #1 1 1 1 1 #5 1 1 1 1 1
Cr #5 #1 #1 #5 #1 #5 #5 $2 #1 ©1 #5 #1 #5 5 5 3 5 #5 5 #5
Fe #5 #5 #3 #5 #3 #5 1 #5 #3 #5 #3 #5 1 #3 1 #3 1 ©5
Ga #5 #5 #5 #3 #5 #3 ©5 #5 ©5 4 #5 #3 4 #3 #5 #3 #4 #5 #3 #3 #4
As #5 #5 #5 5 #5 #5 #5 #5 #5 #2 #5 ©5 2 4 #6 $1 #2 5
Y #2 #1 #1 ©5 ©1 #5 5 #1 #1 #1 1 1 #5 1 #1 1 1 #5 1 #5 1 #1
In #5 #1 #5 #3 #5 #3 #5 #4 #5 #1 ©5 #3 #5 #3 #4 #3 #4 #4 #3 #3 $4 #4
Sb ©5 #5 #5 #5 $2 ©5 ©2 ©2 ©5 #5 #5 #5 #2 ©5 #2 #2 4 #5 #5 #2 #5 #1
La #2 #1 #1 #1 #6 #5 #1 ©1 #5 #1 1 #6 2 ©1 1 1 6
Ce #2 #1 #5 #1 #1 #6 #5 #1 1 #5 #1 #6 1 #6 ©1 ©1 1 #6
Pr #2 #1 #1 #5 #1 #5 1 #1 1 #5 #1 #6 1 #1 2 ©1 #5 1 1 #1
Nd #2 #1 #1 #5 #1 #5 #1 1 #1 1 #1 1 1 ©1 #5 ©1 ©1 #1
365
X S Se Te
A Li Na K Cu Rb Ag Cs Tl Li Na K Cu Rb Ag Cs Tl Li Na K Cu Rb Ag Cs Tl
B
Pm 1 1 5 1 5 1 1 1 1 1 1 1 1 1 1 1
Sm #2 #1 #1 #5 #1 #5 #5 #1 #1 ©1 #5 #1 #6 1 ©1 ©1 ©1 ©1 1 #1
Gd #2 #1 #1 #5 #1 #5 #5 #1 #1 #1 1 #5 #1 #5 1 #1 1 1 ©1 1 #5 1 #1
Tb #2 #1 #1 #5 #1 #5 #5 #1 #1 #1 1 #5 #1 1 #1 1 1 1 ©5 1 #1
Ho #1 #1 #1 #5 #1 #5 #5 #1 #1 #1 1 #1 #5 1 #1 1 1 1 #5 1 #1
Er #1 #1 #1 #5 #1 ©5 #5 #1 #1 #1 1 #1 #5 1 #1 1 1 #1 1 #5 1 #1
Tm #1 #1 #1 #5 #1 #5 #5 #1 1 1 1 1 #5 1 #1 1 1 1 #5 1 #1
Yb #1 #1 #1 #5 #1 #5 #5 #1 #1 #1 1 #5 #5 ©1 2 1 1 #5 1 1
Lu #1 #1 #1 #5 #1 #2 #5 #1 1 1 1 #1 #5 1 #1 1 1 1 #5 1 #1
Bi #2 #2 #2 #5 #1 #1 #1 #2 #2 #2 #2 #5 #5 #2 #2 #2 #5 $1 #1
In Table 2, the following notations were used: 1 ⎯ prediction of the structure of the
α-NaFeO2 type; 2 ⎯ prediction of the structure of the NaCl type; 3 ⎯ prediction of the
structure of the chalcopyrite type; 4 ⎯ prediction of the structure of the TlSe type; 5 ⎯
prediction of the structure different from the ones mentioned above; 6 ⎯ prediction of
the absence of ABX2; the symbol # is used for objects for the machine learning; © -
predictions was confirmed experimentally; $ - predictions was not confirmed experi-
mentally.
40 compositions have been experimentally tested and only in five cases the predic-
tions turned out to be incorrect, i.e., the prediction error was about 12.5 %. Beyond that
the melting point and bandgap were evaluated for compounds with the chalcopyrite
crystal structure type [28].
From ternary to quaternary compounds. Prediction of the crystal structure
type of compounds A2BCHal6. Searching for and studying halide compounds having
the composition A2BCHal6 (Hal = F, Cl, Br, or I) with the elpasolite crystal structure
type is related to the development of new luminescent, laser, and magnetic materials.
The set for computer-assisted analysis included information about 289 (A ≠ C) com-
pounds having the elpasolite structure; 20 compounds with Cs2NaCrF6 type of crystal
structure; 57 compounds with crystal structures another than the ones given above un-
der ambient conditions; and 81 AHal–BHal3–CHal systems where compounds are not
formed [19]. The 134 properties of chemical elements A, B, C, and Hal were included
in the initial set of component parameters.
The problem of predicting new halo-elpasolites included solving three intermediate
tasks. Formation of compounds with composition A 2BCHal6 was predicted in the first
of them (task 1). The next task included searching for regularities and predicting the
formation of compounds with given composition and the most common types of crystal
structures (elpasolite or Cs2NaCrF6). The latter task was divided into two smaller ones.
When solving the first of them, the multi-class prediction of belonging to four classes
(elpasolites, compounds with the Cs2NaCrF6 structure, compounds with the structure
366
different from those shown above, and the systems containing no compounds with com-
position A2BCHal6 (task 2)) was performed. Next, halide systems were consecutively
divided into three classes: the target class, e.g., 1 - elpasolites; class 2 - compounds with
non-elpasolite structure; and class 3 - the AHal–BHal3–CHal systems containing no
compounds with composition A2BCHal6 (task 3). The final decision regarding the class
that a compound being predicted belongs to, was made by comparing the predictions
obtained when solving all three tasks. If the results were inconsistent, the prediction
was regarded to be uncertain and the prediction table cell was left empty.
The algorithms LoReg, artificial neural network learning (ANN), K-nearest neigh-
bor (KNN), and support vector machine (SVM) ensure the best accuracy of prediction
of compound formation (task 1) in the cross-validation mode and the collective deci-
sion-making software based on the algorithm of generalized polynomial corrector [16]
provided the best estimate for prediction accuracy, namely 95%.
When solving task 2 of multi-class prediction, the set of algorithms including DT,
KNN, SVM, ANN, learning a multilayer perceptron, and the algorithm of the convex
stabilizer [16] for collective decision-making, ensured the best accuracy of examination
prediction: 89%. When forming the regularity that allows one to demarcate elpasolites
from compounds with differing crystal structures and from systems where no
A2BCHal6 compounds are formed (task 3), the best accuracy (80%) was provided by
the set of algorithms that included the algorithms LoReg, ANN, KNN, SVM, and the
Bayesian method of collective decision-making [16].
Some results of comparing the predictions found by solving all three classification
tasks are summarized in Table 3. The following notations are used: 1 - prediction of
compounds with the elpasolite crystal structure; 2 - prediction of compounds with the
Cs2NaCrF6 structure type; 3 - prediction of compounds having crystal structure another
than the abovementioned ones; and 4 - prediction of the absence of an compound in the
ACl–BCl3–CCl system; the # symbol is used to denote previously studied compounds;
the information about them was used for machine learning.
Table 3. Part of a table illustrating the prediction of the crystal structure type for compounds
with the composition A2BCCl6 [19].
C Li Na K Rb
A Na K Rb Cs Tl Li K Rb Cs Tl Li Na Rb Cs Li Na K Cs
B
Al 4 #4 4 4 1 4 #4 4 #4 1 #4 #4 4 4 4 4 4 4
Sc #3 #1 #2 #1 #1 #1 #3 4 1 #1 4 4 #4
Ti 3 3 3 1 1 #1 1 4 1 4 4 3 1
V 3 #3 #3 #1 #1 1 1 4 4 1 4 4 1
Cr 3 #3 #1 #1 #1 4 #3 #1 4 #4 3 3
Fe 4 #4 3 4 #4 #1 #4 #4 4 4 4 #4
Y #1 #1 #4 #3 #1 3 4 #4 1 #1 4 4 1 1
In #1 #3 3 1 #1 #3 3 3 #3 4 3
La #4 #4 #1 #4 4 #1 #4 4 1 1 4 4 1 1
Ce 4 #4 1 #1 4 4 #1 #4 4 1 1 4 4 1 1
Pr #4 #4 #4 #1 #4 #4 #4 #1 #4 #4 1 #1 #4 #4 1 1
367
C Li Na K Rb
A Na K Rb Cs Tl Li K Rb Cs Tl Li Na Rb Cs Li Na K Cs
B
Nd 4 4 4 #1 4 #4 4 #1 4 #4 1 #1 4 4 1 1
Pm 3 1 3 1 4 4 1 1 4 1 1
Sm 4 #3 #1 4 #4 #1 #4 1 #1 4 4 1
Eu 3 #3 #1 3 #3 #1 3 4 1 #1 4 1 1
Gd 3 #3 #1 3 #3 #1 3 4 1 #1 4 4 1
Tb 3 #1 #1 3 3 #1 3 1 #1 4 1 1
Dy 3 #1 #1 3 #4 #3 #1 3 #4 1 #1 4 1 1
Ho 3 #1 #1 3 #3 #1 3 1 #1 4 1 1
Er 3 #1 #1 3 #3 #1 3 1 #1 4 1 1
Tm #3 #1 #1 #3 #3 #1 #3 1 #1 4 1 1
Yb 3 #1 #1 #3 3 #3 #1 3 1 #1 4 1 1
Lu #1 #3 #3 3 #3 #1 3 1 1 4 4 1 1
Tl 1 1 1 1 #1 4 1 1 4 4 1 1
U #4 #4 4 #1 #4 #3 #1 #4 4 1 4 4 1
Pu 4 4 1 4 4 3 #1 4 4 1 1 4 4 1
5 Conclusions
During half of the century the predictions of thousands of inorganic compounds in bi-
nary, ternary and more complicated chemical systems were obtained and some their
properties (melting point, critical temperature of superconductivity, band gap energy,
etc.) were estimated in IMET. The obtained predictions usage allows an essential pro-
gress provision in a search for new magnetic, semiconductor, superconductor, nonlinear
optical, electro-optical, acousto-optical and other materials. Hundreds of predicted
compounds were synthesized and our results experimental verification shows that the
average prediction accuracy is higher than 80%. Machine learning methods application
to search for regularities in big chemical data of DB PISM gives an opportunity for
theoretic design of new inorganic compounds that allows substantially reduce the costs
for search for new materials with predefined properties, replacing them by computa-
tions. It is important to note that only information on components properties (chemical
elements or more simple compounds) is used in prediction process.
This work was partially supported by the Russian Foundation for Basic Research
(project nos. 17-07-01362 and 18-07-00080) and the State task № 075-00746-19-00.
We are grateful to V.V. Ryazanov, O.V. Sen’ko, and A.A. Dokukin for long-term help
and collaboration.
368
References
1. Kiselyova, N.N., Dudarev, V.A., and Zemskov, V.S.: Computer information resources in
inorganic chemistry and materials science. Russ. Chem. Rev., 79(2), 145-166 (2010).
2. ACS Publications Homepage, https://pubs.acs.org/, last accessed 2019/04/22.
3. ScienceDirect Homepage, https://www.sciencedirect.com/, last accessed 2019/04/22.
4. Springer Nature Homepage, https://link.springer.com/, last accessed 2019/04/22.
5. Wiley Online Library Homepage, https://onlinelibrary.wiley.com/, last accessed
2019/04/22.
6. Kiselyova, N.N., Dudarev, V.A., Stolyarenko, A.V.: Integrated system of databases on the
properties of inorganic substances and materials. High Temperature, 54(2), 215-222 (2016).
7. NIST Data Gateway, https://www.nist.gov/srd, last accessed 2019/04/22.
8. NIMS Materials Database (MatNavi) Homepage, http://mits.nims.go.jp/index_en.html,
last accessed 2019/04/22.
9. SpringerMaterials Homepage, https://materials.springer.com/, last accessed 2019/04/22.
10. Blokhin, E., Villars, P.: The PAULING FILE Project and Materials Platform for Data Sci-
ence: From Big Data Toward Materials Genome. In: Andreoni, W., Yip, S. (eds.) Handbook
of Materials Modeling, pp.1-26. Springer, Heidelberg (2018).
11. IMET RAS DBs Homepage, http://www.imet-db.ru/, last accessed 2019/04/22.
12. DB IRIC (Information Resources on Inorganic Chemistry) Homepage, http://iric.imet-
db.ru/, last accessed 2019/04/22.
13. Savitskii, E.M., Devingtal’, Yu.V., and Gribulya, V.B.: Prediction of metallic compounds
with composition A3B using computer. Doklady Physical Chemistry, 183(5), 1110-1112
(1968).
14. Savitskii, E.M., Gribulya, V.B.: Application of computer techniques in the prediction of
inorganic compounds. Oxonian Press Pvt., Ltd., New Delhi-Calcutta (1985).
15. Gladun V.P.: Processes of formation of new knowledge. SD "Pedagog 6”, Sofia (1995).
16. Zhuravlev, Yu.I., Ryazanov, V.V., and Sen’ko, O.V.: RECOGNITION. Mathematical meth-
ods. Software system. Practical solutions. Phasis, Moscow (2006).
17. Pedregosa, F., Varoquaux, G., Gramfort, A., et al.: Scikit-learn: Machine learning in Python.
J. Machine Learning Research 12(Oct.), 2825-2830 (2011).
18. Kiselyova, N.N.: Komp’yuternoe konstruirovanie neorganicheskikh soedinenii.
Ispol’zovanie baz dannykh i metodov iskusstvennogo intellekta (Computer Design of Inor-
ganic Compounds: Use of Databases and Artificial Intelligence Methods). Nauka, Moscow
(2005).
19. Kiselyova, N.N., Stolyarenko, A.V., Ryazanov, V.V., et al.: Prediction of New Halo-Elpaso-
lites. Russ. J. Inorg. Chem. 61(5), 604-609 (2016).
20. Dudarev, V.A.: Information systems on inorganic chemistry and materials science integra-
tion. Krasand, Moscow (2016).
21. Kiselyova, N.N., Stolyarenko, A.V., Ryazanov, V.V., et al.: A system for computer-assisted
design of inorganic compounds based on computer training. Pattern Recognition and Image
Analysis, 21(1), 88-94 (2011).
22. Senko, O.V.: An Optimal Ensemble of Predictors in Convex Correcting Procedures. Pattern
Recognition and Image Analysis. 19(3), 465-468 (2009.
23. Yuan, G.-X., Ho, C.-H., Lin, C.-J.: An Improved GLMNET for L1-regularized Logistic Re-
gression. J. Machine Learning Research. 13, 1999-2030 (2012).
24. Yang, Y., Zou, H.: A Coordinate Majorization Descent Algorithm for L1 Penalized Learn-
ing. J. Statistical Computation & Simulation. 84(1), 1-12 (2014).
369
25. Kiseleva, N.N., Burkhanov, G.S.: Search for new ternary phases with Al, Ga, and In using
an information-prediction system. Russian Metallurgy, 1, 223-226 (1989).
26. Kiselyova, N.N., Podbel’skii, V.V., Ryazanov, V.V., Stolyarenko, A.V.: Computer-aided
design of new inorganic compounds with composition ABX2 (X = S, Se or Te). Inorg. Ma-
ter.: Applied Researches. 1(1), 9-16 (2010).
27. Savitskii, E.M., Kiseleva, N.N.: Cybernetic prediction of the existence of ABX 2 phases. In-
org. Mater. 15(6), 866-868 (1979).
28. Kiselyova, N.N., Stolyarenko, A.V., Gu, T., et al.: Computer-aided design of new inorganic
compounds promising for search for electronic materials. In: Proc. The Sixth Int.Conf. on
Computer-Aided Design of Discrete Devices (CAD DD 07). v ol. 1, pp. 23 6 -2 4 . 2 UIPI
NASB, Minsk (2007).
370