Article pubs.acs.org/JACS
Chemically Functionalized Phosphorene: Two-Dimensional Multiferroics with Vertical Polarization and Mobile Magnetism Qing Yang,† Wei Xiong,‡ Lin Zhu,† Guoying Gao,† and Menghao Wu*,† †
School of Physics and Wuhan National High Magnetic Field Center, Huazhong University of Science and Technology, Wuhan, Hubei 430074, China ‡ Wuhan National Laboratory for Optoelectronics, Huazhong University of Science and Technology, Wuhan, Hubei 430074, China S Supporting Information *
ABSTRACT: In future nanocircuits based on two-dimensional (2D) materials, the ideal nonvolatile memories (NVMs) would be based on 2D multiferroic materials that can combine both efficient ferroelectric writing and ferromagnetic reading, which remain hitherto unreported. Here we show first-principles evidence that a halogen-intercalated phosphorene bilayer can be multiferroic with most long-sought advantages: its “mobile” magnetism can be controlled by ferroelectric switching upon application of an external electric field, exhibiting either an “on” state with spin-selective and highly p-doped channels, or an “off” state, insulating against both spin and electron transport, which renders efficient electrical writing and magnetic reading. Vertical polarization can be maintained against a depolarizing field, rendering high-density data storage possible. Moreover, all those functions in the halogenated regions can be directly integrated into a 2D phosphorene wafer, similar to n/p channels formed by doping in a silicon wafer. Such formation of multiferroics with vertical polarization robust against a depolarizing field can be attributed to the unique properties of covalently bonded ferroelectrics, distinct from ionic-bonded ferroelectrics, which may be extended to other van der Waals bilayers for the design of NVM in future 2D wafers. Every intercalated adatom can be used to store one bit of data: “0” when binding to the upper layer and “1” when binding to the down layer, giving rise to a possible approach of realizing single atom memory for high-density data storage. experimentally.12−17 In this paper, it will be revealed that halogen anions may transport on the surface of phosphorene at ambient conditions, potentially creating magnetoelectric devices since those covalently bonded anions carry both magnetic moments and charges. It is known that ferromagnetic (FM) materials with switchable magnetizations, and ferroelectric (FE) materials with exhibiting switchable electrical polarizations, both have potential applications as nonvolatile memories (NVMs). For commercial random access memories (RAMs), data writing in FM RAMs is energy-consuming, while the data reading operation in FE RAMs is destructive; therefore, multiferroic materials18 that are both FM and FE19 and may combine efficient writing by an electric field and energy-saving reading by magnetism are highly desirable. Their rarity in nature may be due to the conflict that most FM materials are metallic while FE materials must be insulating. Almost all the multiferroic materials known to date are antiferromagnetic or ferrimagnetic, except for EuTiO3, which becomes FM under a large strain.20 Additionally, they lack vertical polarization in thin films for high-density data storage, due to the effect of a surface
1. INTRODUCTION Two-dimensional (2D) materials with atomic thickness and high mobility, like graphene1 and transition-metal dichalcogenide (TMDC),2 can be promising candidates for replacing silicon in the future as basic materials for nanoelectronics. Problems such as quantum tunneling and gate leakage in traditional transistor at nanoscalewhich has already led to the end of Moore’s laware likely to be solved in 2D materials. Phosphorene, a monolayer of black phosphorus, has been a new member of the family of 2D materials since 2014.3,4 It is known that graphene possesses ultrahigh mobility but with a zero bandgap, while MoS2 has a suitable bandgap for semiconductors but only a moderate mobility. In comparison, phosphorene possesses a moderate direct bandgap around 0.6− 1.5 eV, depending on the number of layers, that can be tuned via strain.5 Moreover, it can also exhibit a high mobility up to 1000 cm2 V−1 S−1 and a high on/off ratio up to 104 when applied as a field-effect transistor, which is highly desirable in nanoelectronic devices.3 Interesting properties in strain engineering have been predicted theoretically: superior mechanical flexibility,6 strain-tunable bandgap,5,7,8 phase transition upon application of strain,9,10 ferroelasticity,11 etc. Ionic (Li, Na) transport and covalent functionalization in phosphorene have also been studied both theoretically and © 2017 American Chemical Society
Received: May 5, 2017 Published: July 26, 2017 11506
DOI: 10.1021/jacs.7b04422 J. Am. Chem. Soc. 2017, 139, 11506−11512
Article
Journal of the American Chemical Society depolarization field. Currently, neither FM nor FE RAMs have taken the place of flash memory as the prevailing commercial NVM devices, even though flash memory cannot offer arbitrary random-access rewrite or erase operation, in addition to memory wear and read disturb.21 The major reason is that silicon-based flash memory can be directly integrated in silicon wafers utilizing a mature silicon process. In comparison, the combination of FE or FM RAMs into silicon is technically challenged by many problems. Similar to n/p doping channels in silicon wafers, ferroics induced by dopants will be desirable for formation of NVM regions. However, despite intensive research devoted to trying to incorporate ferromagnetism into diluted magnetic semiconductors, their Curie temperature (TC) barely reaches room temperature.22 Efforts toward combining ferroelectricity with semiconductors have been scarcely reported, as ferroelectricity cannot be induced simply by doping 3d metal ions. With the downscaling of integrated circuits size to nanoscale, however, flash memories as well as FM and FE RAMs need to be redesigned to be incorporated into circuits based on 2D materials with atomic thickness. The issue of quantum tunneling in flash memories due to non-degenerate “0” and “1” states will be difficult to solve at the nanoscale, which may provide a chance for FM or FE RAMs with two equivalent states. However, doping 3d magnetic ions in 2D materials like graphene will be much more difficult than in traditional semiconductors. Meanwhile, the number of reports on 2D FE materials is increasing but still limited11,23−28 since the first related prediction in 2013,29 especially for those with the vertical polarization desirable for high-density data storage.30 Here we show first-principles evidence that halogen-decorated phosphorene systems can meet all those challenges, forming 2D multiferroic metal-free materials that combine multiple advantages simultaneously: efficient FE writing + FM reading, vertical polarization for high-density data storage, and directly integration into 2D wafers. It is also revealed that vertical polarization in two dimensions will be more likely to survive in covalent-bonded ferroelectrics compared with ionic-bonded ferroelectrics, which may be extended to other 2D materials to be employed as NVMs.
Figure 1. (a) Geometric structure and spin density distribution (in blue) for phosphorene (P atoms denoted by pink spheres) bonded with a Cl adatom (denoted by green sphere). (b) Illustrations of migration directions for halogen adatoms. (c) Computed diffusion pathway for Cl adatom along the armchair and zigzag directions. (d) Comparison of diffusion pathway for F, Cl, and Br adatoms along the zigzag direction.
Table 1. Binding Energies Compared with X2 (Defined as Eb = E(phosphorene-X) − E(phosphorene) − E(X2)/2, and Negative Values Indicate Favorable in Energy), Hirshfeld Charge Analysis, and Diffusion Barrier of F, Cl, and Br Adatoms in Phosphorene
F Cl Br
Eb (eV/ atom)
charge (e)
monolayer diffusion barrier (eV)
bilayer diffusion barrier (eV)
−2.0 −0.20 −0.022
−0.212 −0.222 −0.224
0.72 0.39 0.29
0.59 0.19 0.074
charges on halogen adatoms reveal that they will be able to diffuse when driven by an electric field if the diffusion barriers are within a moderate range. Take X = Cl as an example. In Figure 1c we calculated the diffusion barrier for a Cl adatom along the zigzag direction (1 to 2) and the armchair direction (1 to 3) using the NEB method. It turns out that diffusion is more efficient along the zigzag direction, with a relatively low barrier of 0.39 eV, but much more blocked along the armchair direction, with a high barrier of 0.72 eV. At ambient conditions, the diffusion will be feasible along the zigzag direction but almost forbidden along the armchair direction. The diffusion paths for X = F, Cl, Br along the zigzag direction are displayed in Figure 1d. It is shown that F has the largest diffusion barrier (>0.7 eV), which is obstructive for its diffusion, while Br has a much lower diffusion barrier (
