Voltage-controlled ON switching and manipulation of magnetization via the redox transformation of β-FeOOH nanoplatelets

Redox-based metal/metal oxide transformations achieved via electrolytic gating recently emerged as a novel, magneto-ionic route for voltage control of magnetism. So far, mainly metal or oxide thin films and nanoporous metal alloy structures are used as starting materials. The present study demonstrates a magneto-ionic transformation starting from a stable electrodeposited FeOOH nanoplatelet structure. The application of a low voltage in a Li-based electrolyte results in the reduction of the virtually non-magnetic FeOOH into ferromagnetic Fe, yielding an ON switching of magnetization. The magnetization can be tuned in a large range by the time of voltage application and remains stable after voltage-switch off. A reversible magneto-ionic change of magnetization of up to 15% is achieved in the resulting iron films with a thickness of about 30 nm. This large magneto-ionic effect is attributed to the enhanced roughness of the iron films obtained from the nanoplatelet structure. The robust, voltage-controlled, and non-volatile ON switching of magnetism starting from a stable oxide structure is promising for the development of energy-efficient magnetic switches, magnetic actuation and may offer new avenues in magnetoelectronic devices.

Redoxbased metal/metal oxide transformations achieved via electrolytic gating recently emerged as a novel, magnetoionic route for voltage control of magnetism. So far, mainly metal or oxide thin films and nanoporous metal alloy structures are used as starting materials. The present study demonstrates a magnetoionic transformation starting from a stable electrodeposited FeOOH nanoplatelet structure. The application of a low voltage in a Libased electrolyte results in the reduction of the virtually nonmagnetic FeOOH into ferromagnetic Fe, yielding an ON switching of magnetization. The magnetization can be tuned in a large range by the time of voltage application and remains stable after voltageswitch off. A reversible magnetoionic change of magnetization of up to 15% is achieved in the resulting iron films with a thickness of about 30 nm. This large magnetoionic effect is attributed to the enhanced roughness of the iron films obtained from the nanoplatelet structure. The robust, voltagecontrolled, and nonvolatile ON switching of magnetism starting from a stable oxide structure is promising for the development of energyefficient magnetic switches, magnetic actuation and may offer new avenues in magnetoelectronic devices.
Keywords: magnetoelectric actuation, magnetoionic effects, voltage control of magnetism, iron oxyhydroxide, iron films (Some figures may appear in colour only in the online journal) Original content from this work may be used under the terms of the Creative Commons Attribution 3.0 licence. Any further distribution of this work must maintain attribution to the author(s) and the title of the work, journal citation and DOI.
anisotropy [2,8,9], coercivity [10,11], domain wall pinning [3] and exchange bias [12]. In contrast to capacitive electronic charging and most other magnetoelectric approaches that are volatile and require the constant application of voltage, magnetoionic effects can persist after the voltage has been switched off [12][13][14]. Due to this nonvolatility, the magnetic state in magnetoionic materials can ideally be set at will by the external voltage, requiring a small current only for the charge transfer as the setting. This provides a huge potential for improving the energy efficiency of magnetic technologies in computing, spintronics, actuation, and lab on chip devices [4,15,16].
To date, such redoxbased voltage control of magnetism is mainly studied for ultrathin films with thickness of or below few nanometers [9,10,13,14,17,18]. Since the magneto ionic mechanism is based on interface reactions, an increase of the surface/volume ratio and the use of porous structures are promising routes to the magnetoionic control of larger structures [5,6]. Liquid electrolytes are beneficial here, since they can easily infiltrate the structures and they exhibit high ionic mobility. In addition, the formation of the electrochem ical double layer in liquid electrolytes makes magnetoionic effects possible at much lower voltage than in solid electro lytes [16,19].
The extension of electrochemically controlled magnetism towards bulk has already been achieved via liquid electrolyte gating and Li intercalation for metal organic frameworks [20] and various porous or nanosized metal oxides [21][22][23]. These materials, however, cannot compete with ferromagnetic metals in terms of magnetization values. Therefore, the combination of porous magnetoionic materials involving a transforma tion to a ferromagnetic metal seems attractive. For example, a transformation from nonmagnetic Fe 2 O 3 nanoparticles to ferromagnetic Fe has been demonstrated using a Libased nonaqueous liquid electrolyte [24]. In several porous metal alloy systems, voltageinduced magnetization changes are achieved via oxidation/reduction reactions [6,25]. In these cases, the porous metal oxide/metal alloy structure, which is used as a starting material [6,25] already contains a ferro magnetic component. This impedes an ON switching of metal ferromagnetism from a complete OFF state, as, e.g. desired for magnetic switches and actuation.
In the present study, longterm stable βFeOOH (aka genéite) nanoplatelets in combination with simple alkaline aqueous solution are utilized as virtually nonmagnetic and high surface area starting material for magnetoionic manip ulation. So far, βFeOOH is mainly studied with regard to catalysis and as battery anode material [26,27]. A high sur face/volume ratio is beneficial to achieve a highly electro chemically active FeOOH material, despite the low intrinsic conductivity [28]. We use electrodeposition as an efficient method to achieve nanosized βFeOOH at room temperature and ambient pressure [26,28]. The nanoplatelets are then polarized in different alkaline solutions, aiming at a voltage induced change in magnetism. The associated voltageinduced structural and morphological changes are probed in detail and correlated to the measured magnetic changes. This approach, starting from paramagnetic FeOOH, enables complete and nonvolatile ON switching of ferromagnetic layers and large reversible magnetoionic effects.

Methods
Electrodeposition of βFeOOH is carried out according to the the routine described by Zou et al [26] from unstirred aqueous solution of 0.1 M FeCl 2 · 4 H 2 O and 0.05 M NaNO 3 . A Si wafer with a thermal oxide of 1 µm and sputtered Au (10 nm)/ Cr (4 nm) buffer layers is used as substrate and working elec trode. The electrochemical cell is a two electrode cylindrical teflon cell with Pt counter electrode. Galvanostatic deposition is performed at 0.2 mA cm −2 on a circular area of 38.5 mm 2 . Deposition times t dep are between 180 s and 600 s.
For the redoxbased magnetoionic transformation starting from the FeOOH deposit, three liquid alkaline electrolytes, 1 M NaOH, 1 M KOH, and 1 M LiOH in aqueous solution, are tested. Cyclic voltammetry (CV) and the potentiostatic elec troreduction process are carried out in 3electrode arrange ment with a saturated calomel electrode (SCE) as reference electrode and a Pt wire as the counter electrode. The voltage quoted thus represents the potential E of the working elec trode versus SCE.
Highresolution transmission electron microscopy (TEM) is conducted on a doubleaberrationcorrected FEI Titan 3 80300 microscope operated at 300 kV to investigate the nanostructure and morphology of the FeOOH nanoplatelets. The phase transformation of the deposits is observed using θ-2θ xray diffraction (XRD, Bruker D8 Advance, CoKα radiation). The surface and cross section of the as deposited and transformed nanostructures are characterized via scan ning electron microscopy (SEM, LEO 15030 Gemini from Zeiss). Atomic force microscopy (AFM, Dimension ICON) in tapping mode is carried out to characterize the surface morph ology and roughness.
The magnetic properties are characterized prior and quickly after the voltageinduced redoxtransformation via anomalous Hall effect (AHE) measurement in a physical property meas urement system (PPMS 6100, Quantum design). A magnetic field µ 0 H of up to ± 3 T is applied perpendicular to the film plane with a field rate of 20 mT s −1 . Simultaneously, a current of 10 mA is applied in the inplane direction, and the Hall resistance is measured in the orthogonal inplane direction. The current is expected to flow through the metallic but non magnetic Au layer, and, after electroreduction, through the metallic ferromagnetic Fe. In ferromagnets, the Hall resistance R H exhibits two contributions: is the anomalous Hall resistance R AHE . In the present case, the R H (H) curves are corrected for the normal Hall effect. The remaining R AHE curves thus scale with M ⊥ . Changes in R S are assumed negligible, since the measurements are performed at constant temperature. Therefore, even though absolute values of magnetization cannot be derived, the changes of R AHE are used as a measure for the relative changes in magnetization.
For the reversible magnetoionic manipulation of the trans formed iron layers, in situ AHE measurements are performed in a custombuilt twoelectrode electrochemical cell with a Pt wire as counter electrode. Further in situ cell details are described in a previous work [30]. The whole deposit area is exposed to the electrolytic gating, since the in situ cell has the same opening dimension for the working electrode as the electrodeposition cell. In some cases, a drift of the AHE curves occurred, which is possibly due to ongoing temperature stabi lization and/or reduction/oxidation processes. This drift is less pronounced after magnetic field reversal. Therefore, both for ex situ and in situ measurements, the branches of the AHE curves from −3 T to +3 T are evaluated. The voltages quoted for the in situ measurements represent the applied cell voltage.

Electric ON switching of magnetism via FeOOH nanoplatelet transformation
The electrodeposited FeOOH nanostructure, as shown in figures 1(a), (b) and (d), consists of polycrystalline nanoplate lets with a thickness in the range of few to tens of nanometers and lateral platelet extensions of several hundreds of nano meters. The electron diffraction patterns (figure 1(c)) and the calculated diffractograms from high resolution TEM (figure 1(d)) are consistent with the tetragonal crystal structure of β FeOOH. The random distribution of the platelets gives rise to the porous and high surface area morphology which is typ ical for electrodeposited βFeOOH [26]. The thickness of the porous FeOOH deposit can be tuned by t dep .
The change of the magnetic state by voltage is antici pated by transforming the paramagnetic βFeOOH [31] to ferromagnetic Fe during electrolytic gating. To achieve the required voltagetriggered reduction reaction, three alkaline aqueous electrolytes, 1 M NaOH, 1 M KOH and 1 M LiOH, are tested. In these electrolytes, the FeOOH to Fe transforma tion is expected to proceed via the following reactions [32]: To identify the reactions and chose suitable potentials for the voltageinduced reduction/oxidation, CV is performed for the βFeOOH samples in the three electrolytes (figure 2). The CV curves start at the open circuit potential (−0.16 V for the NaOH and KOH electrolytes, −0.15 V for the LiOH electro lyte) in cathodic (negative) direction. In all three electrolytes, a cathodic current density (j ) peak at −1.00 V appears, which shows the reduction of Fe 3+ ions present in βFeOOH to Fe 2+ ions in Fe(OH) 2 (reaction (1)). For more negative potentials, a strong increase in cathodic j is measured, which is due to the hydrogen ion reduction reaction. In the case of LiOH, an additional shoulder peak at −1.35 V (see inset in figure 2) is observed, which indicates that the desired reduction from Fe 2+ ions to Fe 0 (reaction (2)) readily occurs. This obser vation is in line with reports on an easy reducibility of iron oxides in Libased electrolytes [33]. In the NaOH and KOH electrolytes, a feature for the Fe 2+ to Fe 0 reduction reaction is not resolved in the CV; but it may be hidden due to a super position with the strong hydrogen evolution at more negative potentials. Indeed, macroscopically, the reduction of rustyred FeOOH to metallic Fe could be observed in all three electro lytes when polarizing at −1.27 V for 180 min. In the CV, after potential reversal, two smaller (positive) j peaks appear in the anodic scan, which can be ascribed to the reoxidation of Fe to (FeOH) 2 (reverse reaction (1)) and FeOOH (reverse reaction (2)) [32,34,35]. The facile electrochemical reduction of nanoplatelet βFeOOH to Fe at room temperature presents an interesting result on its own. So far, studies on the reduction of iron oxide phases to Fe focus on oxides which naturally occur in iron ores. For example, the electroreduction of Fe 2 O 3 to Fe in alkaline electrolytes is described for 100 °C [36]. The room temper ature electroreduction of Fe 3 O 4 films to iron films in KOH solution has only recently been proposed for the design of layered metal/oxide heterostructures [37]. The present results reveal that the reduction of nanoplatelet βFeOOH is another viable room temperature route to the fabrication of magnetic iron layers.
For the further experiments targeting the magnetoionic transformation we chose E = −1.27 V applied via the LiOH electrolyte. At this E, the reduction reaction to Fe in LiOH just starts, and the hydrogen evolution reaction is still negligible.
The voltageinduced phase transformation is proven by XRD ( figure 3) for an FeOOH sample obtained with t dep = 600 s. For the as deposited state, only the substrate peaks are observed. The FeOOH peaks are not resolved, which is connected to the nanocrystalline and porous nature of the FeOOH deposit. After application of −1.27 V for 180 min, a strong peak at exactly the position of the bcc Fe(1 1 0) reflex is osberved. Other bcc Fe peaks are not resolved. As the measurement is carried out in BraggBrentano geometry, this shows a preferred orientation of the Fe {1 1 0} planes parallel to the surface. Such a (1 1 0) fiber texture is typical for elec trolytic iron films and connected to the low surface energy of the close packed (1 1 0) plane in bcc Fe [38]. The electrolytic transformation can also be followed macroscopically, which is depicted in the inset photographs in figure 3. The color of the deposit changes from the rustyred FeOOH to metallic grey iron after voltage application.
The FeOOH to Fe phase transformation goes along with a strong morphological transformation, which becomes evident in the cross section SEM images in figure 4. The starting point is the nanoplatelet structure of as deposited FeOOH ( figure  4(a)). Upon voltage application, for increasing reduction times (t red ), the platelet structure gradually transforms into a film with granular morphology. After 10 min ( figure 4(b)), remains of the platelets are still evident and iron is present in the form of nanosized islands. For longer times (60 min and 180 min, figures 4(c) and (d), respectively), more compact and thicker iron films evolve. An iron film thickness of ca. 28 nm is found after 180 min ( figure 4(d)).
This morphology change suggests that the reduction reac tion proceeds via a dissolution/redeposition process and involves the whole FeOOH structure. Such a mechanism is already under debate for the reduction of Fe 3 O 4 to Fe [37]. Indeed, for several iron (oxyhydr)oxides the dissolution and associated release of Fe 2+ ions is known upon the application of a reduction potential [39]. We could confirm that the disso lution step does not proceed without voltage application. The FeOOH nanoplatelet structure is stable in the electrolyte in the absence of an external voltage and it thus acts as iron ion reser voir, which can be activated to form a metallic iron layer upon voltage application. Since the starting material is porous, the electrolyte penetrates the whole structure and a large amount of material can be transformed. This may be an advantage over thin film oxides as starting material, since for these the trans formed volume is restricted by the reaction layer thickness of few nanometers or below. The AFM image (figure 4(e)) reveals a roughness Rq = 7.5 nm for the resulting iron layer. This Rq value presents a lower limit, as the AFM tip is not infinitely sharp and cannot probe all recesses. The roughness is high in comparison to most metal films used for magneto ionics so far, which are prepared by physical methods [9,13]. Thus, despite the collapse of the nanoplatelets, a high surface area morphology is still present.
The voltageinduced change in magnetic properties associ ated with the FeOOH to Fe redox transformation is presented in figure 5. In as deposited FeOOH state, no AHE effect is observed, which is in line with the paramagnetic nature of FeOOH [31]. Upon voltage application, clear AHE signals are   figure 5(a). The AHE curves correlate to the out ofplane magnetization curves of the formed iron layer [5]. As expected for an iron layer, for which the film plane is the magn etic easy plane, the AHE curve shapes signify a continuous rotation of M upon the application of the perpend icular magn etic field. The saturated anomalous Hall resist ance (R AHE,S ) increases with prolonged t red ( figure 5(b)). This increase in R AHE,S is a measure for the increase in satur ation magnetiza tion which accompanies the FeOOH to Fe transformation.
The presented AHE curves in figure 5(a) are measured after voltage switch off and electrolyte removal. This signifies that different magnetic layers, from few to almost 30 nm thick iron films, can be set by voltage in a nonvolatile manner. The porosity of the initial FeOOH nanostructure seems crucial for this large magnetoionic ON switching effect. The reason is that the electrolyte can penetrate into the nanoporous FeOOH and thereby enable a large reactive interface area, where the FeOOH to Fe reduction reaction can proceed. In consequence, the whole FeOOH structure can be transformed into an iron layer with thicknesses of several tens of nanometers. In prin ciple, even larger Fe layer thickness should become possible, when the thickness of the initial FeOOH deposit is increased further. Thus, the FeOOH-Fe transformation enables the voltagetriggered ON switching of a magnetic layer beyond the ultrathin limit.

Reversible voltage-induced magnetization changes in electrolytic iron films
The full ON switching of magnetism from the FeOOH plate lets is a onetime process due to the collapse of the FeOOH structure. The iron films produced by this reduction process, due to their large roughness, are interesting to study reversible magnetoionic magnetization changes. Figure 6 depicts the AHE curves measured for iron films with a thickness of 28 nm (see figure 4(d)) when subjected to voltageinduced reduction and oxidation in 1 M LiOH electrolyte. The cell voltages are chosen according to a previous study on magnetoionic iron films in LiOH solution [12]. A clear difference in R AHE,S as a measure for M S is observed. In the reduced state (at −1.84 V), R AHE,S is increased in comparison to the state in the oxidation regime (at −0.12 V). This increased M S at −1.84 V is related to the reduction of the iron oxide surface layer to ferromagn etic iron and in line with previous studies [5,9]. During re oxidation at −0.12 V, this surface layer is transformed into iron oxide, associated with a lower M S . This behaviour is repeatable, as depicted in figure 6(b). The voltageinduced M S changes lie around 15%. In previous studies, similar and larger magnetization changes up to ON/OFF switching of magnetism are achieved in Co films below 2 nm and in nano sized iron islands, but the magnetic changes quickly decay when increasing the thickness [5,14]. In the present study, the significant magnetic change is achieved for a much larger iron film thickness of about 28 nm. This large magnetoionic effect can be explained by the large surface area in the very rough films (see figure 4(e) and corresponding discussion). The present measurements show that despite the high surface area the films are stable in the alkaline electrolyte for at least 13 measurement cycles, equivalent to a time of 3 h. The drift  towards larger relative R AHE,S values with increasing measure ment steps has been observed in similar systems before and may be related to a stepwise change of the nature of the passive layer during the repeated electrooxidation steps [9]. The drop of R AHE,S at the last oxidation step may also be connected to a change of the passive layer properties, e.g. a training effect as reported for iron nanoislands in alkaline solution [5]. However, in order to ensure that these effects are systematic also in the present system, and to understand them in detail, further sta tistics and in situ structural characterization will be required.
The stability of iron oxide/iron in alkaline solution, irre spective of the morphology, is as expected from the EpH diagram for the present potentials and pH [35,40]. A similar stability has been experimentally found in previous studies on smooth iron films [9], iron nanoislands [5], and iron nano particles [41]. For the latter, the stability has been confirmed for at least several hundred oxidation/reduction cycles [41]. In comparison to magnetoionic changes in porous Co-Pt microdiscs, the observed magnetization changes in the pre sent study are smaller [25]. However, this is compensated by a better reversibility and a higher overall saturation magnetiza tion of Fe in comparison to Co-Pt.
While voltageinduced full magnetic ON switching starting from FeOOH is readily achieved in the present study, only partial OFF switching of magnetization is observed after reoxidation. The reason is that during the reoxidation of the iron films, only a surface layer of few nanometers is affected [9]. A full magnetic OFF switching would become possible when the whole ferromagnetic iron film is transformed back into FeOOH. Indeed, from an electrochemical point of view, the reoxidation of metal Fe to FeOOH should proceed when applying a suitable oxidation potential (see reactions (1) and (2)). However, the collapse of the porous morphology and formation of a compact iron film impedes the possibility of full reoxidation, because the electrolyte cannot access all of the buried iron metal. For a fully reversible ON/OFF switching of magnetism in the present case, the porous morph ology must be retained. This could be possible by com bining FeOOH nanostructures with a 3D carrier structure. An embedding of βFeOOH in porous carbon, as proposed for battery materials [27], might be a promising avenue. Indeed, in similar FeOOH/carbon nanocomposites, a fully reversible reduction to Fe and reoxidation to FeOOH nanostructures is already demonstrated [34]. For the implementation of the solid/liquid architectures, in analogy to battery systems, cap suled architectures containing FeOOH and electrolyte could be designed which can then be magnetically activated by voltage application.

Conclusion
We demonstrated the voltagecontrolled ON switching of fer romagnetic iron layers starting from paramagnetic FeOOH nanoplatelets via a redox transformation triggered by liquid electrolyte gating. The transformation proceeds via voltage induced dissolution/redeposition, with the electrodeposited FeOOH structure as iron ion reservoir. The timedependency of the reduction reaction enables the nonvolatile setting of magnetic layers with different total magnetic moments. In the resulting iron layers, with thickness beyond the ultrathin limit, large reversible magnetoionic changes of magnetization (up to 15%) are achieved. The approach is promising for novel costeffective routes towards energyefficient magnetic actua tion and magnetolectronic devices.