Novel materials in the Materials Cloud 2D database Davide CampiyNicolas MounetyMarco GibertiniyGiovanni Pizziyand Nicola

2025-05-02 0 0 1.55MB 25 页 10玖币

侵权投诉

Novel materials in the Materials Cloud 2D

database

Davide Campi,∗,†Nicolas Mounet,†Marco Gibertini,†Giovanni Pizzi,†and Nicola

Marzari∗,†

†Theory and Simulation of Materials (THEOS), and National Centre for Computational

Design and Discovery of Novel Materials (MARVEL), École Polytechnique Fédérale de

Lausanne, CH-1015 Lausanne, Switzerland

‡Dipartimento di Scienza dei Materiali, University of Milano-Bicocca, Via R.Cozzi 55,

Milano, Italy

¶Dipartimento di Scienze Fisiche, Informatiche e Matematiche, University of Modena and

Reggio Emilia, I-41125 Modena, Italy

§Centro S3, Istituto di Nanoscienze - CNR, I-41125, Modena, Italy

kLaboratory for Materials Simulations (LMS), Paul Scherrer Institut, CH-5232 Villigen

PSI, Switzerland

E-mail: davide.campi@unimib.it; nicola.marzari@epﬂ.ch

Abstract

Two-dimensional (2D) materials are among the most promising candidates for beyond-

silicon electronic, optoelectronic and quantum computing applications. Recently, their

recognized importance sparked a push to discover and characterize novel 2D materials.

Within a few years, the number of experimentally exfoliated or synthesized 2D materials

went from a couple of dozens to more than a hundred, with the number of theoretically

predicted compounds reaching a few thousands. In 2018 we ﬁrst contributed to this

arXiv:2210.11301v1 [cond-mat.mtrl-sci] 20 Oct 2022

eﬀort with the identiﬁcation of 1825 compounds that are either easily (1036) or poten-

tially (789) exfoliable from experimentally known 3D compounds. Here, we report on a

major expansion of this 2D portfolio thanks to the extension of the screening protocol

to an additional experimental database (MPDS) as well as to the updated versions of

the two databases (ICSD and COD) used in our previous work. This expansion has led

to the discovery of additional 1252 unique monolayers, bringing the total to 3077 com-

pounds and, notably, almost doubling the number of easily exfoliable materials (2004).

Moreover, we optimized the structural properties of all these monolayers and explored

their electronic structure with a particular emphasis on those rare large-bandgap 2D

materials that could be precious to isolate 2D ﬁeld eﬀect transistors channels. Finally,

for each material containing up to 6 atoms per unit cell, we identiﬁed the best candi-

dates to form commensurate heterostructures, balancing requirements on the supercells

size and minimal strain.

Introduction

Two-dimensional (2D) materials represent a vast and broadly unexplored region of the ma-

terials space. Thanks to their extreme thinness they are regarded as ideal platforms for

electronic and optoelectronic applications in the beyond-silicon era1–6 as well as precious

candidates for electro- and photo-catalysis7,8 and for the realization of exotic states of mat-

ter.9–11 Moreover, their properties proved to be much easier to tune with strain, electric

ﬁelds, and doping with respect to their bulk counterparts, and they can be combined in

virtually endless van-der-Waals (vdW) heterostructures12 to engineer novel functionalities.

Although until few years ago only a few dozens of 2D materials had been actively studied,

in recent years major progress has taken place in the experimental synthesis or exfoliation

of a variety of more than a hundred 2D materials.13,14 Even more impressively, the number

of 2D materials theoretically predicted using high-throughput computational methods15 has

grown from less than two hundred16,17 to the range of the thousands.18–23 In our ﬁrst contri-

bution to this eﬀort21 we screened the Inorganic Crystal Structure Database24 (ICSD) and

the Crystallography Open Database25 (COD) for layered materials, identifying a total of

1825 2D materials that, on the basis of their computed binding energies, could be exfoliated

from their layered parent structures with mechanical26 or liquid-phase27 exfoliation meth-

ods. In this work we follow a protocol very similar to the one used in Ref.21 (starting from

a geometric and bonding criteria to identify layered materials, followed by the calculation

of binding energies using ﬁrst-principles vdW density-functional theory (DFT) simulations),

but we add to the sources a third database, the Pauling File28 (MPDS), and we repeat the

screening on ICSD and COD using their most up-to-date versions, as well as allowing the

inclusion of larger structures and using slightly less stringent thresholds on the geometrical

selection. The latter condition results in a more inclusive selection, at the price of a slightly

larger rate of false positives that are later ruled out by DFT calculations. This extended

screening allows us to identify 1252 additional 2D materials bringing the total to 3077 and,

notably, doubling to 2004 the number of compounds that should be most easily exfoliable.

Furthermore, for each of these 3077 monolayers we optimize the cell and the internal geome-

try treating it as an isolated 2D system and we compute its electronic band structure. On the

basis of their optimized geometry we suggest, for each material up to 6 atoms per unit cell,

ideal candidates to build simple, commensurate vertical heterostructures or lattice-matched

lateral heterostructures with minimal strain. Finally, we study a handful of materials with

exceptionally large bandgaps that could serve as insulating layers for 2D Field Eﬀect Tran-

sistor (FET) channels with superior performance with respect to the widely used BN.29 The

full reproducibility of the study is ensured by the AiiDA30,31 materials’ informatics infras-

tructure, which keeps track of the provenance of each calculation; therefore the results are

openly available together with their entire provenance through the Materials Cloud.32,33

Results and discussion

Identiﬁcation of layered materials and optimization of bulk com-

pounds

Following the recipes and tools detailed in Ref.21 we start the computational exfoliation

protocol by extracting the bulk 3D crystal structures, in the form of CIF ﬁles,34 from three

experimental repositories: ICSD,24 COD25 and the Pauling File (MPDS).28 We exclude

structures with partial occupations together with CIF ﬁles that do not provide the explicit

positions of one or several atoms, cannot be parsed, or are obviously wrong. Theoretically

predicted structures are also discarded when signaled. This results in a total of 147731

structural entries for ICSD, 279885 for COD and 355016 for MPDS. In this work we focus

on entries containing at most 6diﬀerent species and 100 atoms or less in the primitive unit

cell, reducing the number of entries to 140586,91161 and 262010 for ICSD, COD and MPDS,

respectively. The CIF ﬁles are extracted and then converted into AiiDA structures using

pymatgen.35 All these 3D structures are separately analyzed to ﬁnd possible candidates

for exfoliation using the same geometrical screening procedure originally described in Ref.,21

building a connection between two atoms when their distance is smaller than the sum of their

respective VdW radii at least by ∆, which is a parameter in the protocol. In the current

screening we assume slightly larger uncertainties in the VdW radii,36 thus allowing ∆to range

between 1.0Å and 1.5Å (1.1Å - 1.5Å was used in the original screening). The geometrical

selection identiﬁes thus a total of 8963 layered materials in ICSD, 6794 in COD, and 11530

in MPDS. These selected structures are then processed with the spglib software37 to ﬁnd the

primitive cells, and ﬁltered for uniqueness (separately for each source) using the pymatgen

structure matcher.38 Finally a second cutoﬀ on the number of atoms ( ≤40 atoms/unit cell

independently of the number of atomic species) has been applied, leaving respectively 6933,

6283 and 5907 structures for the three databases. These structures coming from the three

diﬀerent databases have been combined and ﬁltered for uniqueness a second time (this time

across the diﬀerent databases) giving a total of 9306 layered candidates, 3689 of which were

not included in our previous screening.21 In Ref.21 the structures of the layered materials

obtained from the geometrical selection were optimized using two diﬀerent non-local vdW-

compliant functionals: the vdW-DF2 functional39 with C09 exchange40,41 (DF2-C09) and

the revised Vydrov-Van Voorhis42–44 (rVV10) functional. Subsequently the binding energy

(the diﬀerence per unit area between the total energy of optimized 3D bulk structure and

the sum of the energies of each isolated substructure of any dimensionality45) was computed

with both functionals. The two resulting binding energies turned out to be rather similar

and very rarely changed the classiﬁcation of a material. For this reason we abandon here this

redundancy and employ only the vdW DF2-c09 functional for the structural optimization

and subsequently the calculation of the binding energies for the new 3689 entries.

Binding energies

Before calculating the binding energies, all the structures, optimized with vdW DF2-c09, are

further screened with the geometrical selection algorithm to assess whether they maintain

their layered nature after relaxation. This further selection, together with some unavoidable

convergence failures, resulted in a total of 2251 binding energies successfully computed, to

be added to the 3210 computed in our previous work.21 We report in Fig. 1 the distribution

of the binding energies for the new combined database compared with the distribution of the

3210 binding energies obtained in Ref.21 The colors reﬂect the classiﬁcation of the materials

developed in Ref.21 into three classes according to their binding energies: easily exfoliable

(with binding energies below 30 meV/Å2, which is close to that of materials routinely ex-

foliated with standard techniques like graphene or MoS2), potentially exfoliable (binding

energies between 30 and 120 meV/Å2) and non-exfoliable (larger than 120 meV/Å2).

Similar to Ref.,21 the majority of the materials exhibit binding energies below 30 meV/Å2.

In the current work the predominance of materials with a low binding energy is even more

prominent, with a sharper main peak positioned at lower energies and a relatively faster

文档加载中……请稍候！
如果长时间未打开，您也可以点击刷新试试。

下载文档到电脑，查找使用更方便

10 玖币 0人已下载

立即下载

摘要：

NovelmaterialsintheMaterialsCloud2DdatabaseDavideCampi,,yNicolasMounet,yMarcoGibertini,yGiovanniPizzi,yandNicolaMarzari,yyTheoryandSimulationofMaterials(THEOS),andNationalCentreforComputationalDesignandDiscoveryofNovelMaterials(MARVEL),ÉcolePolytechniqueFédéraledeLausanne,CH-1015Lausanne,Switzerla...

展开>> 收起<<

Novel materials in the Materials Cloud 2D database Davide CampiyNicolas MounetyMarco GibertiniyGiovanni Pizziyand Nicola.pdf

共25页,预览5页

还剩页未读，继续阅读

声明：本站为文档C2C交易模式，即用户上传的文档直接被用户下载，本站只是中间服务平台，本站所有文档下载所得的收益归上传人(含作者)所有。玖贝云文库仅提供信息存储空间，仅对用户上传内容的表现方式做保护处理，对上载内容本身不做任何修改或编辑。若文档所含内容侵犯了您的版权或隐私，请立即通知玖贝云文库，我们立即给予删除！

Novel materials in the Materials Cloud 2D database Davide CampiyNicolas MounetyMarco GibertiniyGiovanni Pizziyand Nicola

相关推荐

开通VIP享超值会员特权

作者详情

相关内容

热门标签

举报选择: