Emerging opportunities for voltage-driven magneto-ionic control in ferroic heterostructures

With the advent of the Big-Data era,¹ cloud computing, Internet of Things, 5G communication, artificial intelligence, and other technologies continue to rise and infiltrate every corner of people’s daily life, leading to geometric growth of today’s data volume. The storage and readout of massive data needs convenient, fast, safe, efficient, and green devices with low energy consumption. Therefore, the requirements of data storage in the emerging information industry are developing toward the direction of low energy consumption, non-volatile nature, high density, high speed, and high integration.

In the past few decades, a great deal of effort has been made to develop energy-efficient materials, processes, and devices in the field of digital communication and information storage. Spintronic devices, enabled by effectively controlling and manipulating the spin degrees of freedom in solid-state systems through electric means, have been widely used in the semiconductor information industry.^2–6 Generally, electric current-induced magnetization switching has been achieved by spin-polarized current-induced spin transfer torques or spin orbit torques.^7–9 However, Joule heating will inevitably occur in the electric current-based mechanisms, which causes an undesirable energy dissipation problem. An electric field (electric voltage) is easier to localize than magnetic fields and is non-power-dissipating in contrast to electric current. Therefore, using electric voltage to manipulate magnetism usually enables the development of spintronic devices with a combination of advantages including low power, on-chip design, non-volatile nature, reversibility, high speed, and good compatibility with the conventional semiconductor industry.^4,5 Thus, voltage control of magnetism (VCM) caters for a variety of potential applications such as data storage, non-von Neumann computing, quantum transport, logics, and energy conversion.

In 2000, Ohno et al.¹⁰ first reported the phenomenon of electric-field control of ferromagnetism in the diluted magnetic semiconductor (In, Mn) As and attributed the mechanism to the carrier concentration modulation by the electric field. Inspired by studies on magnetic semiconductor, the material systems of VCM were then rapidly extended to magnetic metals and oxides by researchers. A large number of experimental and theoretical investigations on the modulation and switching of magnetism have emerged in recent years. Accordingly, different mechanisms are discussed to explain the observed VCM phenomena in different material systems, mainly including the carrier modulation, strain effect, exchange coupling, orbital reconstruction, and electrochemistry effect (also known as ion migration or magneto-ionic effects).^4,5,11–14 From the symmetric beauty of physics, the one-to-one correspondence between the physical nature of the first four mechanisms and the four degrees of freedom of charge, lattice, spin, and orbital in strongly correlated electronic systems has been established.⁴ However, with the in-depth development of this field, how to obtain a greater electric-field control effect has become a major problem in this field. Fortunately, the emergence of new types of dielectric gating materials, such as ionic liquids (ILs) and high oxygen ion mobility conductors (GdO_x,^15,16 ZrO₂:Y₂O₃,¹⁷ SrCoO_2.5,¹⁸ etc.), provides a solution to this problem, which is accompanied by the emergence of magneto-ionic mechanism in VCM.^{4,11,12,14–18} The magneto-ionic effect,^14–21 understood as the change in magnetic properties of materials due to the voltage-driven ion transport, acquires a leading role among other VCM mechanisms in the field.¹⁹ This is triggered by its capability to remarkably modulate magnetic properties in a permanent and energy-efficient way.^11,16,19,22 Nowadays, ionic evolution is considered as a new degree of freedom for interacting with electronics, and it has already established a new concept termed “iontronics.”²³ Figure 1 presents an overview diagram of the relevant integration of the performances, functions, and potential applications brought by ion-controlled electronics. We can see that besides the modulation of magnetic properties by magneto-ionics, ionic evolution also controls a wide range of other properties, covering electrical, optical, thermal, and electrochemical properties.

Usually, magneto-ionic systems consist of layered heterostructures in which ferromagnetic (FM) target materials, such as Co,^15–18 Fe,²² (La, Sr)MnO₃,^24–26 SrCoO_x,^27,28 and SrRuO₃,^29,30 are placed/grown adjacent to typical gate electrolytes (e.g., ionic liquid^4,5 or GdO_x^15,16). Depending on the voltage polarity, these electrolytes accept or donate magneto-ions, acting as ion reservoirs.^11,16,19 In this way, in recent years, the magnetic properties of ferromagnetic layers, including saturation magnetization,^15,16 remanence,^15–17 coercivity,^16,30–32 Dzyaloshinskii–Moriya (DM) interaction,³⁰ magnetic anisotropy,^15,16 Curie temperature,^27,31 exchange bias,^33–35 anomalous Hall effect,³⁰ magnetoresistance effect,^29,30 domain wall,^36,37 etc., have been effectively controlled (Fig. 1). Figure 2 summarizes the several typical gate electrolytes and architectures, as well as several typical transferable ions in voltage-driven magneto-ionic control devices. Electrolytes, known for the separation of cations and anions to opposite electrodes under an electrical voltage, have attracted a great deal of attention for applications in magneto-ionics, mainly including ILs with a distinctive electric double layer (EDL) and solid-state electrolytes or ionic conductors with high oxygen mobility.^4,5,12,14 By using the combination of IL and other dielectric gate oxides (e.g., MgO,³⁸ Al₂O₃,³⁹ HfO₂,^31,32,39 and ZrO₂^4,39) with high permittivity and low oxygen mobility also provides an additional method to achieve the magneto-ionic control effect. In addition, the utilization of aqueous electrolytes (e.g., 1M KOH) in electrochemical cells^21,40 and use of resistive switching devices^{18,20,41–44} to achieve magnetic control have also made important contributions to the field of magneto-ionics in recent years. With these liquid-state and solid-state gate electrolytes, the insertion/removal of ions, such as O²⁻,^15–18 H⁺,^27,30,45,46 Li⁺,^44,47–49 F⁻,^21,50 and N²,¹⁹ into/from target magnets enable an effective control of magnetism by electric fields via an electrochemical process, providing a fine perspective on energy-efficient, non-volatile, and high-density data storage in spintronics.

In recent years, some reviews on VCM have more or less involved with the magneto-ionic effect.^{4–6,11,12,14,20,21,51,52} Nevertheless, a more comprehensive introduction to the magneto-ionics in different material systems is still lacking. In particular, this field is developing rapidly, and some new phenomena and new applications of the magneto-ionic effects are constantly being refreshed. Therefore, we believe that it is necessary to summarize the present voltage-driven magneto-ionic control effect in order to guide and stimulate the field to a better and deeper direction. So, this research update aims at giving a timely review on the recent advance and breakthrough in voltage-driven magneto-ionic control of ferroic thin films and heterostructures. After this brief introduction in Sec. I, we will first introduce the electric-field-controlled ionic liquid gating (ILG) technology, and then, we mainly discuss the voltage-driven magneto-ionic control in ferroic oxide heterostructures in Sec. II, including (ferro)magnetic oxide, multiferroic [ferromagnetic/ferroelectric (FE)] oxide, and ferroelectric oxide heterostructures. Subsequently, in Sec. III, magneto-ionic control effects in magnetic metal-based heterostructures will be presented. Finally, we conclude the review and put forward the future challenges and prospective for the burgeoning field of magnto-ionics in Sec. IV.

Oxide materials are ubiquitous in nature, and their impressive functional properties also make them ubiquitous in scientific and technological applications. In particular, the voltage-driven magneto-ionic effect is more general and plentiful in ferroic oxide thin films and heterostructures. This is mainly due to the following reasons: (i) The strong coupling between lattice, charge, spin, and orbital freedom exists in complex oxides, providing a broad platform for the magneto-ionic effect. (ii) Due to the semiconductor or even insulator properties, oxide films usually display a much larger screening length, thus guaranteeing a strong electric-field effect for the manipulation of magnetic properties.⁴ (iii) Transition metal oxides, especially ABO₃ perovskites, are also an important class of functional materials with a wide range of physical, chemical, and electrochemical properties based on oxygen vacancies (oxygen ions),⁵³ providing a good opportunity for voltage-driven ion migration. So, in this section, we will give a brief overview on the magneto-ionic studies of ferroic oxide heterostructures and devices. To summarize and compare the voltage-driven magneto-ionic control research in ferroic oxide-based heterostructures, the different types of gate electrolytes/gate architectures together with the main experimental observations on various types of voltage-controlled material systems are listed in Table I.

Before introducing magneto-ionic effects to specific oxide materials, it is necessary to describe the electric-field-controlled ionic liquid gating (ILG) first. A typical example of a VCM effect is a multiferroic oxide bilayer based on ferromagnetic/ferroelectric coupling.^4,5,11 However, in common ferroelectric/ferromagnetic oxide bilayers, the field effect of the ferroelectric material has limited ability to control the carrier concentration, and the charged ion displacement in ferroelectrics is only confined to the interfacial modulation of neighboring oxides. Therefore, obtaining a strong VCM effect in oxide materials has become a new challenge in the field.^4,14 Researchers are always looking for alternative ways to meet these challenges. Around the late 2000s, however, an alternative, high-impact, and widely applicable method was introduced into the field, that is, electric-double-layer transistors (EDLTs),^4,5,14,23 as shown in Fig. 3. The essential concept is to replace the gate dielectrics in the field effect device with ILs or electrolytes [Fig. 3(a)].^4,5,14,23 In EDLTs, two mechanisms are often accompanied, namely, electrostatic mechanism and electrochemical mechanism, as shown in Figs. 3(b) and 3(c), respectively.^4,5,14,23,54 As illustrated in Fig. 3(b), when the typical negative gate voltage is applied, cations and anions in ILs are electrostatically attached to the external gate electrodes and IL–oxide channel interface. Subsequently, EDLs, which are pairs of negative and positive charge sheets consisting of the ions in the IL and the electrostatically induced charges on the surface of the oxide channel, are formed.^4,5,14,23 The beauty of this ILG is that the EDL is essentially a nanogap capacitor with an extremely large specific capacitance, up to 100 µF cm⁻². It is equivalent to giant induced charge carrier densities up to 10¹⁵ cm⁻² under a just few volts, thereby realizing the densities needed for oxide channel materials.^4,14 More generally, in the context of EDLTs, the term “electrochemical effect (ion migration) mechanism” has been used to describe all mechanisms other than the electrostatic case.¹⁴ As shown in Fig. 3(c), the applied typical negative voltage drives the migration of ions from the electrolyte into the oxide channel material and produces changes in structures and properties in the gated channel material. Thus, the electrostatic mechanism involves only ion movement in the IL electrolyte, while the electrochemical mechanism involves additional ion migration and chemical reaction in the gating material and across the interface.^4,14 It should be noted that in EDLTs, the mechanisms of electrostatic doping and electrochemical (ion migration) are not opposite and often coexist in the same system depending on the magnitude and polarity of voltage. In many cases of ILG control, such as the magnetoelectric supercapacitors in La_0.74Sr_0.26MnO₃,^55,56 magnetic modulation of La_0.5Sr_0.5CoO_3-δ,⁵⁷ electrical modulation of interface magnetic states in metal Co films,⁵⁸ voltage modulation of Ruderman–Kittel–Kasuya–Yosida (RKKY) interaction via ILG in synthetic antiferromagnetic multilayers of FeCoB/Ru/FeCoB and (Pt/Co)₂/Ru/(Co/Pt)₂,⁵⁹ etc., it has been proved that the two mechanisms coexist.

Recently, voltage-driven ionic motion in magnetic oxide single films has attracted a huge interest since it is not limited to the electric-field screening length and, therefore, can be relevant in relatively thick films.^4,14,51 At present, the magneto-ionic effect has been reported in many complex oxides, with the system studies including La_1−xSr_xMnO_3−δ (LSMO),^{24–26,55,56} L_1−xSr_xCoO_3−δ (LSCO),^57,60 L_1−xSr_xFeO_3−δ (LSFO),⁶¹ SrCoO_3−δ (SCO),^27,28 SrFeO_3−δ (SFO),⁶² Sr_1−xCo_1−xFeO_3−δ (SCFO),⁶³ SrRuO₃ (SRO),^29,30 and NiCo₂O₄.⁶⁴ In the scope of magnetic oxides, electrolyte-gated LSMO devices showed a substantial formation and annihilation of oxygen vacancies affecting magnetic bulk properties (20 nm)²⁴ and providing switching capabilities for large sample areas with excellent spatial homogeneity and high switching speed.^56,65 In addition, the non-volatile electric-field effect of electrolyte gating based on the migration of oxygen ions (oxygen vacancy) ensures that many electric-field-controlled ex-situ tests are possible.^4,25 A classic example is that Cui et al.²⁵ used ILG to electrically manipulate the orbital occupancy and magnetic anisotropy in LSMO in 2015, as shown in Figs. 4(a) and 4(b). Ferromagnetic SrRuO₃ is another model system exhibiting the magneto-ionic effect. Especially, SrRuO₃ is also a rare example of 4d band metal with itinerant ferromagnetism, which has fascinating properties originating from the strong spin–orbit coupling.^66,67 In 2020, Li et al.³⁰ demonstrated an effective control of magnetism and topological Hall effect in a thick SrRuO₃ system (∼90 nm) through the electric-field-induced proton (H⁺) evolution with ILG, as displayed in Fig. 4(d). The ILG-induced large proton concentration gradient in protonated H_xSrRuO₃ was proposed to result in a large Dzyaloshinskii–Moriya (DM) interaction with the emergent topological Hall effect.³⁰ The origin of the topological Hall effect in thick H_xSrRuO₃ is different from the case in the ultrathin SrRuO₃ system where the origin is attributed to inequivalent interfaces in studied ultrathin SrRuO₃ films.^68–72 In short, ILG-induced proton (H⁺) evolution has emerged as a novel strategy to manipulate magnetism in complex oxides.

Most research on VCM effects based on ion migration has focused on the evolution of individual ions (e.g., oxygen, hydrogen, or lithium). In 2017, Lu et al.²⁷ first experimentally realized a reversible non-volatile electric-field control of tri-state phase transformation based on the migration of dual ions (oxygen and hydrogen ions) in SCO films [Fig. 4(c)]. As the selected model material, antiferromagnetic SrCoO_2.5 with a brownmillerite structure has good ordered oxygen vacancy channels and multivalent cobalt ions, which provide favorable conditions for its structure and magnetic phase transition.^27,73 As shown in Fig. 4(c), they found that oxygen ions in ionic liquids can migrate into the film to form a ferromagnetic metal SrCoO_3–δ (SCO) phase with a perovskite structure under negative voltage, while hydrogen ions can migrate into the film under positive voltage to form a weakly ferromagnetic insulator HSrCoO_2.5 phase with the new structure. In addition, the voltage-driven dual-ion control approach via ILG was also utilized to obtain a bi-directionally tuned thermal conductivity across one order of magnitude in SCO at room temperature in 2020.⁷⁴ 

Beyond ionic electrolyte gating, solid–solid devices also exhibit promising magneto-ionic effects. It has been shown that using strongly reductive Gd films as a capping layer on LSMO (40 nm) induces significant oxygen migration and enables the control of magnetic properties even without applying voltage.⁷⁵ Similarly, the ionic distributions and magnetization in LSCO films (36 nm) could also be controlled in the solid state by depositing a Gd capping layer [Fig. 4(e)].⁶⁰ Coupling these works of achieving interfacial oxygen migration coated with the solid Gd capping layer with recent advances in electric-field control of ion distributions opens a new pathway toward non-volatile control of ionic materials at the nanoscale.

Voltage-driven ion motion is an effective method for manipulating the performances of oxide materials, which has been extensively studied in various single oxide layers in recent years. However, the research on magneto-ionic control in magnetic oxide bilayer or multilayer films has been largely ignored, although the multilayer film structure is more important in many practical devices or electric appliances, e.g., solid-oxide fuel cells and multiferroic tunnel junctions.^76–78 On one hand, the magnetic oxide multilayer films tend to contain abundant interface coupling effects, which adds complexity to the magneto-ionic control to a certain extent. On the other hand, understanding ion migration behaviors across the interface under the electric field is a challenging task owing to complex ion transport between different layers. In 2017, Cui et al.⁷⁸ reported a circulation of electronic phase transition via ILG in the SrCoO_3–x/La_0.45Sr_0.55MnO_3–y (SCO/LSMO) heterostructure by sweeping an external voltage of several volts, as shown in Fig. 5(c). They demonstrated that the priority of electric-field-induced oxygen vacancy formation/annihilation in the SCO/LSMO bilayer system is mainly determined by the oxygen vacancies’ formation energies and Gibbs free energy differences. These results not only realize the reversible manipulation of magnetic phase transition in the SCO/LSMO oxide bilayer by electrical means, but also provide a novel understanding of the oxygen vacancy formation and annihilation in oxide heterostructures. For the same oxide bilayer SCO/LSMO (the bilayer here is SrCoO_x/La_0.67Sr_0.33MnO₃), Song et al.⁷⁹ achieved a large reversible modulation of the magnetic anisotropy of the bottom LSMO ferromagnetic layer in 2019 by controlling the dual-ion (oxygen and hydrogen ions) migration into the upper SCO thin film with positive and negative voltages via ILG. Different crystal structures at the interfaces of tri-phase SrCoO_3−δ, SrCoO_2.5, and HSrCoO_2.5 with LSMO lead to different deformation of the MnO₆ octahedral structure at the interface of the LSMO layer, thus effectively modulating the magnetic anisotropy.

A key feature of oxide heterostructures is the emergent interlayer coupling, such as exchange bias, charge transfer, magnetoelectric coupling, and orbital reconstruction.^{4,11,13,62,80} However, efficient tuning of the interfacial coupling by external stimuli and the resulting effects were less investigated. In 2020, Song et al.⁸¹ obtained simultaneous electric tuning of the magnetic anisotropy and exchange bias in La_0.8Sr_0.2CoO₃/La_0.67Sr_0.33MnO₃ (LSCO/LSMO) bilayers by using ILG to control dual-ion migration [Fig. 5(a)]. The magnetic easy axis of the bottom LSMO layer is repeatedly switched between the out-of-plane and the in-plane directions, and the interfacial exchange coupling displays a corresponding modulation [Fig. 5(b)]. Furthermore, x-ray absorption spectroscopy and x-ray-linear-dichroism analysis indicated that orbital reconstruction at the LSCO/LSMO interface resulting from Mn-to-Co charge transfer is responsible for the variation of the magnetic properties.⁸¹ Similarly, interfacial exchange bias in SrFeO_3−δ/La_0.7Sr_0.3MnO₃(SFO/LSMO) bilayers was also modulated by Saleem et al.⁶² and Zhang et al.⁸² via ILG in 2019. However, by optimizing the voltage and time of ILG, the voltage-driven oxygen ion motion only occurs at the top of the SFO layer and does not enter the bottom of the LSMO layer. Under the electric field of the ionic liquid, SFO switches between brownmillerite SrFeO_2.5 and perovskite SrFeO_3−δ phases, which changes its antiferromagnetic configuration, and then affects the exchange bias in SFO/LSMO bilayers.

When designing different complex oxides with multiple functions such as heterojunctions, multilayer films, and superlattices, the chemical and physical properties of the system can be changed to make it significantly different from the bulk form, creating many novel magnetic and electrical properties, thus providing more freedom in material design.⁸³ This point is clearly reflected in recent work done by Yi et al.⁸⁴ of the emergent electric-field control of phase transformation in oxide superlattices. As shown in Fig. 5(d), in superlattices comprised of alternating one-unit cell of SrIrO₃ and La_0.2Sr_0.8MnO₃, they found a reversible phase transformation with a 7% lattice change [Fig. 5(e)] and a dramatic modulation in chemical, electronic, magnetic, and optical properties, mediated by the reversible transfer of oxygen and hydrogen ions via ILG at room temperature. By contrast, these phenomena are not otherwise observed in the individual constituent oxides, solid solutions, or larger period superlattices.⁸⁴ Therefore, from the perspective of the structure of thin film devices, it is expected that there will be more exploration of voltage-driven ion migration based on complex bilayer and multilayer film devices in the future.

In this section, we present magneto-ionic controlled multiferroic oxide heterostructures, mainly referring to the ferroelectric/ferromagnetic (FE/FM) bilayers driven by electric-field-controlled ILG.¹² Conventionally, most of the studied multiferroic heterostructures used for VCM have not actually added additional ionic liquid electrolytes to their systems. For example, the mostly studied multiferroic bilayer systems encompass LSMO or SRO combined with BaTiO₃ (BTO) or BiFeO₃ (BFO).^4,13,72,85 The underlying VCM mechanisms in these traditionally multiferroic heterostructures only comprise carrier modulation, strain effect, exchange coupling, and orbital reconstruction at the FE/FM interface.^{4,5,11–14,85} For this reason, electrolyte-assisted control of magnetism in multiferroic structures is still considered to be rather undeveloped. In fact, ILG provides an effective approach to polarize and switch the large-area FE films in a controlled manner,^86–88 which lays a foundation for further realization of magneto-ionic control in multiferroic heterostructures.

In 2016, Cui et al.⁸⁹ first demonstrated the effective magneto-ionic control of bulk magnetism in a La_0.5Sr_0.5MnO₃ (LSMO) (20 nm) and BTO (3.2–5.2 nm) multiferroic heterostructure through ILG. Interestingly, they found that an “orbital switch” can be formed at the interface between BTO and La_0.5Sr_0.5MnO₃ (LSMO) to manipulate the electric-field effect in the LSMO/BTO heterostructure. The orbital switch is based on the connection or breakdown of the interfacial Ti–O–Mn bond due to the FE displacement under an external electric field [Figs. 6(a) and 6(b)].⁸⁹ The insertion of a FE-BTO layer between the IL and LSMO creates EDL-like-1 and EDL-like-2 via ILG. The two EDL-like structures are superimposed to produce a remarkable enhancement of the electric field at the interface between BTO and LSMO, as well as in the following LSMO film.⁸⁹ Such a strong electric field could promote the oxygen migration in the whole bulk of the film with the electron injection or extraction through the interfacial Ti–O–Mn covalent bond. In conjunction with the enhanced field effect produced by FE polarization, positive gate voltage with an orbital switch in the “ON” state promotes the increase in Curie temperature (T_C) in LSMO [Fig. 6(a)]. However, when the negative gate voltage is applied, the orbital switch is in the “OFF” state, and there is no electric-field modulation effect [Fig. 6(b)]. Additionally, it is important to note that Herklotz and co-workers achieved the reversible control of interfacial magnetism with the aid of polarized neutron reflectometry in a La_0.5Sr_0.5MnO₃ (LSMO)/PbZr_0.2Ti_0.8O₃ (PZT) multiferroic heterostructure through similar IL-assisted FE switching in 2017.⁸⁶ Distinctively, the authors concluded that the origin of the enhancement/reduction in the interfacial magnetization in the PZT/LSMO system relied on the electrostatic hole accumulation/depletion produced by the FE polarization rather than the voltage-driven ion migration mechanism. Similarly, in 2018, Nishino et al.⁹⁰ also used ILs to achieve reversible and electrostatic switching of the FE polarization in PZT (25 nm) aiming at a FE control of transport properties in SRO. Herein, they observed a large T_C change from 115 to 90 K in ultrathin SRO (4 nm).

At present, the research issue of controlling the oxygen migration across the proximal FE layer under an applied electric field through IL has been proven to remain topical. In 2019, Gu et al.⁹¹ further reported that via IL and FE dual-gating, the remote control of oxygen vacancies and magnetic phase transition can be achieved in bulk SCO films (18 nm) capped with an ultrathin ferroelectric BTO layer (3.2–4 nm) at room temperature. Furthermore, this process is associated with fascinating magnetoelectric coupling [Fig. 6(c)]. In particular, they found that the ultrathin BTO layer acts as an atomic “oxygen valve” and is semitransparent to oxygen ion transport due to the competing interaction between vertical electron tunneling and FE polarization plus surface electrochemical changes in itself, thus resulting in the striking emergence of new SrCoO_x mixed phase.⁹¹ The lateral coexistence of brownmillerite phase SrCoO_2.5 and perovskite phase SrCoO_3−δ was directly observed by scanning transmission electron microscopy (TEM) [Fig. 6(d)]. The ability to control the flow of oxygen ions across the heterointerface by the so-called “oxygen valve” provides a new approach at the atomic scale for designing multi-state memories, sensors, and solid-oxide fuel cells.

Currently, most of the investigations on electric-field control effect based on ILs are mainly focused on ferromagnetic (FM) material, while the studies on FE materials (especially perovskite-type ferroelectric oxide materials) are just beginning. As a class of important functional materials, FE materials have important application prospects in the field of non-volatile information storage.^76,92,93 Furthermore, we can see from Sec. II C that FE materials play an essential supporting role in the VCM of multiferroic heterostructures, but the response mechanism of FE materials to ILG is sometimes ambiguous. So, it is necessary to understand and explore the response of FE materials to the ILG, which can lay the foundation for a deeper magneto-ionic effect based on multiferroic heterostructure devices in the future. Therefore, this section is devoted to extend the ILG to FE oxide thin films even though the aim of this section is not the control of magnetism.

Several research teams have reported the ILG on FE heterostructures in recent years. A research team from Oak Ridge National Laboratory has been investigating the possibility of ILs to realize the electrostatic and reversible reversal of FE polarization in a large area of ultrathin FE films since 2017. One of the works was done by Herklotz et al.⁸⁶ for IL/PZT (∼8 nm)/LSMO, already mentioned in Sec. II C. Importantly, they pointed out that the use of ILs for switching the FE polarization provides following advantages:⁸⁶ (i) It has the ability to switch a large area of FE films that otherwise cannot be accomplished by conventional metallic top electrodes due to shorting via pinholes or other defects. (ii) It possesses feasibility to thinner FE films since the leakage currents are smaller through the ILG. (iii) ILG allows a reversible in situ switching of the FE polarization on the same sample while avoiding the ambiguities arising from the self-poling of differently prepared samples. Furthermore, in 2019, Sharma et al.⁸⁷ demonstrated that ILs can be used to induce reversible large-area polarization switching in ultrathin and highly defective FE-PZT (PbZr_0.2Ti_0.8O₃) films. They draw the conclusions that long range electrostatic charge control is induced by modifying the EDL at an IL–PZT interface with electrostatic and electrochemical control of polarization orientation in the FE layer.⁸⁷ As shown in Fig. 7(a), 16 nm PZT films are grown on very rough 7 nm-thick SRO conducting bottom electrodes, which results in the existence of defects and partially conductive channels in PZT. By partially covering the film with the IL gate, the boundary between the pristine film and +4 V IL-gated areas can be studied by using a pizoresponse force microscope (PFM). The PFM images in Fig. 7(a) indicate that the IL biasing switches the high-quality regions of the FE film even in close proximity to metallic conducting channels, highlighting the IL’s ability to effectively ignore the defect structures and electronic short channel. In addition, ILG of other oxide systems is known to induce oxygen migration, so the authors also examined the possible oxygen migration (electrochemical mechanism) into and out of the lattice during FE switching through time-of-flight secondary ion mass spectroscopy (TOF-SIMS) in isotopic oxygen-18 (¹⁸O) environments for PZT (43 nm)/LSMO (7 nm)/STO devices.⁸⁷ As displayed in Fig. 7(b), they observed that the ¹⁸O ration decreases the natural abundance level around 4 nm into the PZT layer, demonstrating that the IL biasing provides a sufficient electric field at the PZT interface to induce electrochemical changes. Interestingly, these results are consistent with the experimental results of Cui et al.⁸⁹ and Gu et al.⁹¹ They both proved that oxygen ions can penetrate the ultrathin ferroelectric BTO film of ∼4 nm in the magneto-ionic control experiments. In a word, these results suggest that apart from ionic field-driven electrostatic control of polarization switching, the electromigration of oxygen ions can also affect the FE properties of the films.

In addition to finding an effective means to switch the FE polarization state of ultrathin FE films in a large area, it is more important to reliably analyze, measure, and control the macroscopic FE properties of ultrathin ferroelectric films, which is of great significance for practical device applications. Another representative research group in the field of ILG of FE heterojunctions is Kawasaki et al. from the University of Tokyo. The team used EDLTs to act on BTO⁹⁴ and PTO (PbTiO₃)⁹⁵ films in 2014 and 2018, respectively. They found that EDLTs can induce conductivity on the FE surface, but the surface conductivity of FE BTO and PTO films is slightly different.

In recent years, although the ferroelectricity of ultrathin films has been studied by scanning probe technology (such as piezoelectric force microscopy (PFM)⁹⁶ or structural analysis with x-ray diffraction⁹⁷ and transmission electron microscopy⁹⁸), there are few reports on measuring the macroscopic FE properties of the capacitor structure with ultrathin FE materials by traditional electrical measurement methods due to the leakage current in the ultrathin FE films. In terms of ILG of ultrathin FE films, the Kawasaki team truly demonstrated the measurement, modulation, and improvement of the macroscopic FE properties of low-dimensional PTO (2.6 nm) thin films by using ILs in 2020 [Fig. 7(c)].⁹⁹ Surprisingly, by using EDLTs to suppress the leakage current in ultrathin FE films, they clearly observed the macroscopic distinct remnant polarization (∼20 µC cm⁻²) down to a 2.6 nm-thick film [Fig. 7(d)]. Finally, they used in situ x-ray diffraction measurements under an external electric field with EDLTs to monitor the c-axis lattice deformation [Fig. 7(e)], revealing that the reduced tetragonality in ultrathin films is mostly recovered by canceling out the depolarization field.⁹⁹ This discovery further reveals that the EDLT technique is an excellent method to explore the macroscopic physical properties of ultrathin FE films.

Besides the obvious voltage-driven magneto-ionic phenomena observed in oxide thin films and heterostructures, there have been many reports recently on the magneto-ionic effect in a variety of magnetic metal-based heterostructures and devices. In 2014, voltage-induced oxygen migration was presented as a method to control magnetism and magnetic anisotropy in a new field termed “magneto-ionics.”^15,16,100 Although it was previously thought that it was difficult to realize a large enough electric-field effect in metals due to the short screening length,^4,101,102 the scalable manipulation of magnetism with an electric field has been observed in magnetic metals due to the magneto-ionic effect. This is mainly due to the emergence of new gate medium materials, such as ionic liquids and solid ionic conductors (GdO_x,^15,16,100 ZrO₂:Y₂O₃,¹⁷ and SrCoO_2.5¹⁸). Finally, in this section, we briefly give a description on the magneto-ionic effect on magnetic metal-based heterostructures. For comparison, the research on voltage-driven magneto-ionic control in representative magnetic metal-based heterostructures is summarized with the corresponding data in Table II.

Usually, when the oxygen mobility of the oxide is high enough, a more controllable and extensive VCM effect can be observed.⁴ A representative example is the GdO_x/magnetic metal heterostructure where the dielectric layer GdO_x is known as a good solid-state ionic conductor.^15,16,100 In the GdO_x-based heterostructure, relatively small voltage-induced oxygen ion (O²⁻) migration to and from the GdO_x–metal interface can lead to a periodic reoxidation of the ferromagnetic metal, thus modulating the magnetic properties of the ferromagnetic metal.^{4,6,14–16,100} Specifically, as shown in Fig. 8(a), Bi et al.¹⁵ patterned the sample comprised of the Si/SiO₂/Pt(4 nm)/Co(0.7 nm)/Gd₂O₃(80 nm)/Ta(5 nm)/Ru(100 nm) multilayer into the Hall bar structure for transport measurements in 2014. They found that both the saturation magnetization and anisotropy field of ultrathin Co films adjacent to Gd₂O₃ gate oxides could be significantly modulated by electric fields in a non-volatile fashion, resulting a record high change in magnetic anisotropy energy up to 0.73 erg/cm² with gate voltages of only a few volts [Fig. 8(b)].¹⁵ Through a combination of structural, magnetic, transport, and spectroscopic studies, they demonstrated that this giant VCM effect is achieved by voltage-induced reversible oxidation of the Co layer, which can be understood by a large interfacial electric field and the high O²⁻ ion mobility in Gd₂O₃. Almost at the same time, Bauer et al.¹⁶ similarly showed that the easy axis of a thin Co film (0.9 nm) could be tuned with voltage when Co is grown in contact with GdO_x. What advanced is that they directly observed in situ voltage-driven O²⁻ migration in a Co/GdO_x bilayer by using cross sectional transmission electron microscopy (TEM) and high resolution electron energy-loss spectroscopy. Moreover, the response speed of such a device can be increased by ∼6 orders of magnitude by changing the temperature and gate voltage. By reducing the thickness of O²⁻ diffusion barrier or further optimizing the morphology of gate oxides and electrodes, the time scale of magneto-ionic switching can be significantly shortened. In 2017, Gilbert et al.¹⁰⁰ further performed structural and magnetic depth-profile analyses with polarized neutron reflectometry to understand how the depth-dependent oxygen diffusion affects the magnetic properties of relatively thick AlO_x/GdO_x/Co(15 nm) films. The depth profiles showed that the voltage drives oxygen deep into the Co film (>10 nm), reducing the magnetization by >80% at the interface and 38% in the bulk. After reversing the polarity of gate voltage, the process was reversible up to 92%. Importantly, the authors confirmed the changes in the Co oxidation state, but not in the Gd, revealing that the GdO_x transmits the oxygen rather than surrendering it.

Having established that by making use of voltage-gated O²⁻ transport (oxygen-based magneto-ionics) could enable magnetism modulation in magnetic metals by controlling interfacial or bulk oxidation states. However, the problems such as irreversibility and device degradation exist in oxygen-based magneto-ionic devices because of the chemical and structural changes in target ferromagnets. In addition, elevated temperatures are required for these types of devices. In 2019, Tan and colleagues developed the reversible and non-destructive toggling of magnetic anisotropy at room temperature using a small gate voltage through H⁺ pumping in all-solid-state Au/GdO_x/Co/Pt heterostructures.⁴⁵ Importantly, in these heterostructures, a 90° magnetization switching can be achieved without requiring a redox reaction in the target ferromagnetic Co and without relying on oxygen migration in the GdO_x. In that work, when voltage is on, H₂O molecules were adsorbed by the Au layer and H⁺ was generated via electrolysis (Au serves as the catalyst). The H⁺ was then pumped (by the electric feld) into the GdO_x and transported to the GdO_x/Co interface [Fig. 8(c)].^45,103 The injected hydrogen can either bind directly with the CoO or react with the oxide to form hydroxides or interstitial water. The accumulation of hydrogen at the GdO_x/Co interface then switches the perpendicular magnetic anisotropy (PMA) of ultrathin Co into in-plane anisotropy. When the reverse voltage is applied, interstitial water and hydroxides in the Gd₂O₃ release oxygen, which reacts with the Co, and hydrogen is ejected [Fig. 8(d)].^45,103 Harnessing this discovery, the authors developed a new class of devices in which the magnetic properties are controllable by electric-field-moderated hydrogen migration. The small size, low atomic weight, and weaker binding energy of hydrogen, as compared to oxygen, mean that these devices might operate much faster (100 ms) than their oxide counterparts, where the switching time would be of the order of minutes to tens of minutes even at elevated temperature.

Generally, the magneto-ionic effect is slow because it involves the thermally activated process.^11,19 While Tan et al. demonstrate a record-breaking (for magneto-ionics) switching time of 100 ms, it may not be appropriate for fast and real-time device applications yet. In terms of improving the response speed of magneto-ionics, Lee et al.¹⁷ further achieved ∼1 ms reliable (>10³ cycles) magnetization toggle switching using the yttria-stabilized zirconia [ZrO₂:Y₂O₃ (YSZ)] gate oxide as a solid-oxide proton electrolyte in 2020. This work established that gate oxide material engineering may be a key pathway for achieving fast and reliable switching in magneto-ionic devices, highlighting the significant potential toward future spintronic applications. It is worthwhile to note that the shortest switching time reported so far for such magneto-ionic switching is ∼0.2 ms in Co/SrCoO_2.5 system accompanied by bipolar resistance switching,¹⁸ where a 40 nm-thick brownmillerite SrCoO_2.5 oxide has superior oxygen ion mobility due to its natural-ordered oxygen vacancy channels. However, the magnetization switching between the high-resistance state and the low-resistance state in Co/SrCoO_2.5 is not significant, where the magnetic hysteresis loop was only slightly altered (the coercivity was tuned by a few oersted). It suggests that the interfacial redox reaction barely occurs or is at least incomplete.

Besides the ionic conductors with high oxygen mobility, the combination of ionic liquids and other dielectric oxides (e.g., MgO,³⁸ Al₂O₃,³⁹ HfO₂,^31,32,39 and ZrO₂^4,39) with high permittivity k and low oxygen mobility provides an optional method to achieve the magneto-ionic control effect. As an example, Jiang et al.¹⁰⁴ achieved the electrochemical manipulation of the phase transition from the antiferromagnetic to the ferromagnetic state of 5 nm-thick FeRh with a combination of IL and HfO₂ oxide gating. The authors attributed the magnetic phase modulation to the large electric effect of EDLs associated with the extraction and injection of oxygen ions for positive and negative gate voltages, respectively [Fig. 8(e)].¹⁰⁴ Similarly, the effective magneto-ionic effect has already been reported in Pt/Co/Ni/HfO₂/IL^31,32 and Pt/Co/HfO₂/IL¹⁰⁵ devices. Specially, for example, the inert oxide MgO, as a static supporting material, is beneficial to the perpendicular magnetic anisotropy (PMA) in typical trilayers, such as Pt/Co(Fe)/MgO and Ta/CoFeB/MgO sandwiches.^106–108 The study of magneto-ionic effect with these fundamental heterostructures is of great significance to the emerging magnetic random access memory (MRAM) technologies.

In this perspective, we have summarized the latest progress in the nascent area of magneto-ionics, which is a powerful tool to effectively manipulate both the interface and bulk magnetism in ferroic heterostructures and has great potential for energy-efficient, high density, and non-volatile data storage. Of course, it is still very hard to cover all aspects of magneto-ionics. From the perspective of materials, two kinds of materials (ferroic oxides and magnetic metals) and two representative dielectric gating materials (ILs and ionic conductors) were mainly discussed, associated with their special performance on magneto-ionics. Especially in the framework of the ferroic oxide heterostructure, we have also extended the magneto-ionic control to FE materials, elucidating that the FE properties can also be electrostatically and electrochemically tuned with applied gate potentials via ILG. In short, the topic of magneto-ionics forms a multidisciplinary cross covering four main disciplines: spintronics, materials science, electrochemistry, and condensed matter physics, as shown in Fig. 9. At present, the VCM mechanism based on voltage-driven ion transport has been widely used in spintronics, interface science, high-performance power transistors, phase transition, quantum transport, etc.

Looking ahead, the research on magneto-ionic control in ferroic heterostructures and devices, however, is still in its infancy, and thus, there remain many open questions, such as:

1. It is established that both electrostatic and electrochemical (magneto-ionics) mechanisms operate in the electrolyte gating of materials such as oxides. The cooperation and competition between two mechanisms are often affected by many factors, such as the voltage magnitude, voltage polarity, and voltage action time, as well as target magnet type and thickness. Sometimes, it is important but challengeable to sort out the dominant mechanism by cutting-edge microscopic analysis.

2. The obvious weakness of magneto-ionics is that its room temperature switching speed is normally limited by the redox reaction rate or ion migration speed. In terms of response speed, magneto-ionic devices need to be further improved in the future. Some new approaches such as engineering the gate oxide materials (using new types of ionic conductors and ionic liquids),¹⁷ using an electrochemical capacitor configuration,¹⁰⁹ and combining with resistive switching¹¹⁰ could give the perspective to further boost the response time.

3. Usually, there is an inherent voltage trade-off between magnetism modulation, speed, and cyclability in magneto-ionic devices. Specifically, the degree of magnetism modulation increases with voltage, whereas cyclability degrades for exceedingly high voltages due to irreversible losses of mobile ions. It requires researchers to balance all aspects of the modulation of a single device to find an optimal solution.

4. Controlling ion migration is the core of magneto-ionics. There is additionally much to learn from other fields such as batteries, noble metal catalysis, and supercapacitors, where a weather of knowledge and ideas can be tapped to develop new types of magneto-ionic devices.

5. Associated with the emergent research directions in the magnetism community, such as voltage-controlled behaviors, e.g., Skyrmions,^111,112 Dzyaloshinskii–Moriya interaction,^70,113 and topological Hall effect,^69–71 as well as the VCM in atomically thin van der Waals magnets (e.g., Fe₃GeTe₂, CrI₃, and Cr₂Ge₂Te₆),^114–116 magneto-ionics will have a more broad platform to show its flexibility, bringing about a wealth of intriguing fundamental science discoveries and extensive application potential.

6. Since the human bodies are driven by electrical signals of ions, development of the magneto-ionic technology is expected to lead to more applications in the future, such as healthcare, neuromorphic, memristive, and synaptic devices.

The authors acknowledge the support from the Beamline BL08U1A in the Shanghai Synchrotron Radiation Facility (SSRF). This work was financially supported by the National Natural Science Foundation of China (Grant Nos. 52031014 and 51871130). Y.G. acknowledges the support from the Young Talents Project of Shenyang National Laboratory for Materials Science and Innovation Fund from the Institute of Metals Research.

The data that support the findings of this study are available from the corresponding author upon reasonable request.