Voltage controlled Néel vector rotation in zero magnetic field

Multi-functional thin films of boron (B) doped Cr2O3 exhibit voltage-controlled and nonvolatile Néel vector reorientation in the absence of an applied magnetic field, H. Toggling of antiferromagnetic states is demonstrated in prototype device structures at CMOS compatible temperatures between 300 and 400 K. The boundary magnetization associated with the Néel vector orientation serves as state variable which is read via magnetoresistive detection in a Pt Hall bar adjacent to the B:Cr2O3 film. Switching of the Hall voltage between zero and non-zero values implies Néel vector rotation by 90 degrees. Combined magnetometry, spin resolved inverse photoemission, electric transport and scanning probe microscopy measurements reveal B-dependent TN and resistivity enhancement, spin-canting, anisotropy reduction, dynamic polarization hysteresis and gate voltage dependent orientation of boundary magnetization. The combined effect enables H = 0, voltage controlled, nonvolatile Néel vector rotation at high-temperature. Theoretical modeling estimates switching speeds of about 100 ps making B:Cr2O3 a promising multifunctional single-phase material for energy efficient nonvolatile CMOS compatible memory applications.


Supplementary Note 1. Magneto-Optical Kerr Measurements (MOKE)
While bulk chromia is an exceedingly good dielectric [2], the B:Cr2O3 film ( Figure S2, black line) exhibits significant broadening of the intensity versus energy.
This broadening of the spectral intensity is indicative of surface charging, leading to a decrease of the effective incident electron kinetic energy. This is characteristic of increased sample resistivity. The inverse photoemission spectra from the undoped Cr2O3 single crystal ( Figure S2, blue line), is very similar to the spectra taken for undoped Cr2O3 thin films [3], but is here seen to be characteristic of an n-type material (electron doped), a signature of oxygen vacancies.

Supplementary Note 3. Transport measurements on Hall bar devices
As mentioned in the main text, Fig

Supplementary Note 4: Kelvin Probe Force Microscopy (KPFM)
To rule out that the MFM contrast variations in Fig.   4 originate from the electrostatic long-range forces, we performed additional Kelvin probe microscopy measurements. Figure    This agrees well with the inverse photoemission data shown in Figure S2.

Supplementary Note 7. Estimation of the effective d33,eff coefficient
The PFM amplitude signal, A, can be estimated as A=d33,eff V Q, where d33,eff is the effective piezoelectric coefficient, V is the driving voltage and Q is the quality factor associated with the cantilever dynamics. An order of magnitude estimate of the d33,eff coefficient can be obtained by comparing the PFM amplitude signal of the material under investigation with the electromechanical response of a reference material with the well-known piezoelectric coefficients measured by independent methods. In our studies, lithium niobate (LNO) thin film crystals were used as the reference sample. PFM testing of LNO and B:Cr2O3 has been carried out using the same cantilevers. The d33,eff in LNO has been reported to be 8.4 pm/V [4] and by comparing the PFM amplitude signal in LNO [5], a calibration factor for similar cantilevers has been obtained and then used to estimate the d33,eff value in B:Cr2O3.

Pulsed Laser Deposition of Sesquioxides V2O3 and B-doped Cr2O3
The V2O3 films were grown in vacuum by pulsed laser deposition. The Al2O3 substrate was cleaned according to a modified RCA protocol which was developed specifically to achieve atomically clean Al2O3 substrates [6]. The clean Al2O3 substrate was then placed in the growth chamber and the temperature is raised to 750 °C. The system was equipped with a 248 nm KrF excimer laser which is focused on a 2-inch diameter target of V2O3 (99.9%). For deposition, the laser was set to operate at 150 mJ energy per pulse and at a repeat rate of 6 Hz. The laser ablates the rotating target (30 Hz) creating a plume of material which is subsequently deposited on the target. The laser is rastered across the rotating target for even wear of the target. The sample itself rotates at 3 Hz to ensure even deposition. The deposition rate is determined by growing several samples of various thicknesses and subsequently measuring the thickness via X-ray reflectivity. After the deposition of the V2O3, the doped chromia is grown on top.
The growth of the B-doped chromia is accomplished through the same procedure as above with the following exceptions; a chromia target (99.8%) is used, the temperature of the sample is reduced to 700 °C, the repetition rate of the laser is set to 10 Hz, and the deposition is done in the presence of a partial pressure of decaborane (B10H14) gas.
Decaborane is solid at room temperature, but highly volatile. A decaborane partial pressure was accomplished by storing a small amount of decaborane in a canister attached to the vacuum chamber with a precision leak valve in between. The decaborane canister was evacuated beforehand (to remove volatile impurities) and then heated to 53 °C prior to deposition, creating a decaborane vapor. The temperature of the decaborane was maintained throughout the entire deposition. The precision leak valve is then used to control the decaborane pressure in the vacuum chamber. The base pressure of the vacuum chamber was 9.1 × 10 −8 Torr, while during chromia deposition the pressure was maintained between 1.7 × 10 −6 Torr and 1.9 × 10 −6 Torr by adjusting the leak valve.

Magnetron sputtering of Pt layer and lithography of Hall bar devices
A 5 nm thick Pt film was deposited on Cr2O3 film via DC magnetron sputtering in a vacuum chamber with base pressure of 1 × 10 -8 Torr (1.33 × 10 -8 mbar) and a process pressure of 5 mTorr (6.67 × 10 −3 mbar) with an applied power of 30W. The heterostructure thus consists of a Pt thin film on top of 200 nm B-doped Chromia film grown on a back gate V2O3 film (20 nm). A schematic of the heterostructure is shown in Fig. S3 (c). A two-step E-beam lithography procedure was used to fabricate the Hall bar devices. First, a negative tone resist was patterned in the form of Hall bars of sizes 1.5 x 8 μm and 1.0 x 8 μm and subsequently developed. Ar ion etching was used to etch away the exposed Pt layer while protecting the Hall bars. Next, a 30 nm SiO2 dielectric spacer layer was deposited via RF sputtering onto the patterned sample, followed by lift-off. In the second lithography step, a positive tone resist was patterned to fabricate wires and contact pads. Finally, Ti/Au (3 nm/50 nm) electrodes were deposited using e-beam assisted evaporation. This step was followed by final lift-off in acetone to reveal the finished devices. It is important to note that the leakage profile of 30 nm SiO2 film was individually determined by I-V measurements on films deposited on sapphire substrates. Upon applying 10 V on the top and bottom contact pads across a 30 nm SiO2 film, a current of 20 nA was measured, thus denoting its ability to withstand sizeable voltage levels.

Molecular beam Epitaxy of CoPt multilayers with perpendicular and in-plane anisotropy
The CoPd films were grown using molecular beam epitaxy. The thin film ferromagnet was grown under ultra-high vacuum, with a background pressure of ~5 × 10 −9 mbar during growth. After introduction to the vacuum the sample was heated to 300 °C. The growth rate of the Co and Pd has been previously gauged by growing a series of films of various thicknesses and characterizing the thicknesses by X-ray reflectivity. The resulting linear growth rate is extrapolated down for films of single nanometer proportions. In addition, the growth rate was monitored in-situ, by a quartz crystal thin-film thickness monitor which has been previously calibrated. Both methods provided the same measurement for film thickness.

Vibrating Sample Magnetometry
The Quantum Design VersaLab vibrating sample magnetometer is designed for magnetic characterization over a large temperature range. When placed in the sample holder, the temperature of the sample can be controlled from 50 K to 400 K. The vibrating sample magnetometer operates by moving a magnetic sample through stationary pickup coils. By measuring the induced current in the coils the moment of the magnetic sample can be calculated. Using the integrated lock-in measurement technique the magnetometer has a noise floor of < 6.0 × 10 −7 emu. For the measurements shown in Figure 2, the temperature of the sample was set between 66 K and 400 K in intervals of 4 K. The sample temperature was allowed to stabilize before the measurement was made. By sweeping the magnetic field, the hysteresis loop of the sample was measured. The coercive field values were taken as the field values at which the measured magnetization crossed zero, the exchange bias was then calculated as the average of the two coercive field values.

Polar Kerr Magnetometry
The data shown in Fig. 1  These differences were detected by a photodiode which is used as an input to the lock-in amplifier. An in-depth description of the Kerr magnetometry used here can be found in Ref [7]. By cycling the magnetic field, a hysteresis loop can be observed in the signal, corresponding to the magnetization of the sample. A linear background is also present corresponding to Faraday rotation of the light as it passes through the quartz window in the presence of a magnetic field. The linear background was subtracted and the hysteresis loops are normalized between 1 and -1. After normalization, the coercive field values of the hysteresis loop are taken to be the zeroes of the data.

Spin Resolved Inverse Photoemission Spectroscopy
The spin-polarized inverse photoemission (SPIPES) spectra were taken at various different temperature. The magnetic and electric field were both applied to achieve field cooling from 500 K to 300 K with • < 0. After field cooling, the spectra were taken at ~300 K and sample was gradually heated to targeted temperature (~315 K, ~325 K) and corresponding spectra were taken when the thermal equilibrium was achieved. Our spinpolarized inverse photoemission experiments utilize a transversely polarized spin electron gun based upon the Ciccacci design [8,9]. As described elsewhere [9], the spin electron gun was used in combination with an iodine-based Geiger-Müller isochromat photon detector with a SrF2 window. As is typical of such instruments, the electron gun has 28% spin polarization, and the data have been corrected for this incident gun polarization. The direction of electron polarization is in the plane of the sample. The magnetoelectric cooling was accomplished in an axial magnetic field of > 40mT and a voltage of 3 kV applied across the film thickness. The Fermi level was established from tantalum and gold foils in electrical contact with the sample. We find that the in-plane spin polarization increases with increasing temperature over the range 300 to 325 K in spin-polarized inverse. Since the spin asymmetry is increasing in a region where the boundary magnetization is decreasing, this generally indicates that the in-plane polarization increases faster than the boundary magnetization declines. Independent validation of the spin-polarized inverse photoemission, measuring the in-plane polarization Fig. S9 (c), comes from in-plane X-ray magnetic circular dichroism Fig. S9 (a), which provides a small, but nonetheless non-zero signal at the Cr L3 (2p3/2) edge indicative of a small in-plane Cr 3+ moment Fig S9 (b) at 320 K. X-ray absorption (XAS) and X-ray magnetic circular dichroism (XMCD) studies were carried out at Canadian Light Source. Spatial resolution of the Elmitec GmbH PEEM microscope is better than 30 nm for an ideal flat sample. The spectra were taken in thermal equilibrium conditions estimated about ~320 K based on the reading of a thermocouple.
Both "up" and "down" domains were observed in XMCD and the spectra closely resemble those taken of chromia and reported elsewhere [10]. Note that the XMCD spectra showing net boundary moments and the SPIPES spectra with the largest spin asymmetry were taken in the temperature region of ~320 K which is higher than the Néel temperature of pure Cr2O3 indicating an enhanced magnetic ordering temperature with boron doping.

Hall measurements
Prior to taking Hall measurements, the electrical resistance of Pt films of various thicknesses deposited on Al2O3 substrates was determined. The resistivity of 5 nm Pt film

Scanning Probe Microscopy
Scanning probe microscopy was performed to characterize the morphology and leakage properties of the heterostructure. As shown in the supplementary Fig S4. (a) a 2 x 2 µm Atomic Force Microscopy (AFM) scan of the Cr2O3 surface is recorded. The root mean square (RMS) roughness of a 1 x 1 µm area (shown with a box) on the film is determined to be 0.17 nm. Together the RMS value and the peak to peak height difference ensure that the Pt film deposited on top is continuous and without any voids. An AFM conductivity mapping (C-AFM) recorded simultaneously with topography image is also shown (supplementary Fig 3 (b)), where the image is taken with DC bias of +0.5V maintained between the bottom V2O3 film and the metallic tip. The instrument used for these measurements is the Bruker AFM in PeakForce Tunneling AFM (TUNA) in the tapping mode. In this mode of operation, the tip is intermittently made to contact the sample surface, thus avoiding the lateral forces during imaging. The conductivity image shows no leakage through the heterostructure, a condition essential to maintain sizeable electric field across the device. Since the AFM tip is never in ohmic contact with the surface, the situation is different than actual devices where the Pt hall bars are in contact with a larger area on the Cr2O3 film and hence a minimal leakage is present. Nevertheless, the C-AFM is an excellent technique in determining any structural defects on the surface that are responsible for enhanced leakage rendering the device unsuitable for application of essential electric field, as has been reported in our earlier study [11].
Magnetic force microscopy (MFM) utilizes the long-range force, originating from interaction between a magnetized tip and the gradient of the magnetic stray field of the sample surface. By maintaining a tip lift-height of 30 nm the van der Waals interaction becomes negligible while the magnetic signal dominates. ASYMMFMHM tips with high magnetic moment CoCr coating were used for MFM imaging.