Isothermal tuning of exchange bias using pulsed fields

Exchange bias, HE, and coercivity, HC, of antiferromagnetic (AFM)/ferromagnetic bilayers can be adjusted, after deposition, at temperatures below the Neel temperature of the AFM by subjecting the samples to large pulsed fields (in excess of HPulse=550 kOe). The efficiency of the process depends on the AFM system and the direction of the applied field with respect of the unidirectional anisotropy direction. Textured (111) Fe19Ni81/Fe50Mn50 bilayers show an HE reduction and a HC increase when the pulse field is applied antiparallel to the unidirectional anisotropy, while they only exhibit a reduction in HC when the pulse is applied parallel to their unidirectional anisotropy. On the other hand, textured (111) NiO/Co bilayers exhibit a change of the angular dependence of HE when the pulse is applied away from the unidirectional anisotropy. The effects could be caused by field induced changes in the domain structure of the AFM or transitions in the AFM (spin–flop or AFM–paramagnetic).

Exchange bias, i.e., the shift of the hysteresis loop along the field axis, 1 resulting from the exchange coupling at the interface between ferromagnetic ͑FM͒ and antiferromagnetic ͑AFM͒ materials, plays a fundamental role in magnetoelectronic devices. 2 To induce exchange bias, AFM-FM systems are usually either ͑i͒ field cooled through the Ne ´el temperature, T N , of the AFM or ͑ii͒ deposited in the presence of a field. 1 At a fixed temperature, the loop shift, H E , and the coercivity, H C , of the system are controlled by intrinsic parameters such as AFM-FM coupling at the interface, FM and AFM thicknesses, interface roughness, or grain size. 1 Hence, H E and H C should remain fixed after deposition.
4][5][6][7][8][9] Irreversible approaches, such as high temperature annealing 3 or ion irradiation 4 induce structural changes in the bilayers.Hence, H E and H C can only be adjusted a limited number of times.Among the reversible techniques, cooling through T N in large fields, [5][6][7] cooling in combinations of continuous and alternating fields, 8 or cooling in zero field from different magnetization states, 9 all require warming the bilayer above T N .Another approach to tune H E and H C is to go to large negative fields to saturate the FM and to wait a certain time and subsequently carry out the remaining part of the loop. 10H E and H C depend on the waiting time.This procedure can be slow, thus often to enhance its efficiency the samples are warmed close to T N .Warming the bilayers with AFMs with T N ӷ300 K could induce unwanted structural changes ͑e.g., interdiffusion͒ in the system, which could deteriorate the performance of the device. 3Moreover, in other AFM materials, such as LaFeO 3 ͑Ref.11͒ or ␣-Fe 2 O 3 , 12 the samples cannot be warmed close or above T N , either because the AFM decomposes ͑e.g., LaFeO 3 ) or due to the exceedingly large T N .
In this letter, we present a reversible process to tune H E and H C in exchange biased bilayers, with T N Ͼ300 K, without the need to warm the samples above T N .The procedure involves applying large field pulses at room temperature ͑RT͒.The field induced changes depend on the direction of the applied pulse and the type of AFM material.
Textured ͑111͒ Fe 19 Ni 81 /Fe 50 Mn 50 and textured ͑111͒ NiO/Co were studied.The Fe 19 Ni 81 (10 nm)/ Fe 50 Mn 50 (15 nm) ͑FeNi/FeMn͒ bilayer was sputtered at RT onto Corning glass.Nonmagnetic (Ni 81 Fe 19 ) 50 Cr 50 was used as buffer and capping layers.A field of Hϭ400 Oe was applied during growth to induce a unidirectional anisotropy.The NiO ͑20 nm͒/Co ͑15 nm͒ samples were evaporated onto oxidized Si͑100͒ substrates.The NiO layer was e-beam evaporated from a Ni ingot in an O 2 atmosphere while the substrate was at T S ϭ200 °C.Before the Co deposition the O 2 was evacuated to Ͻ2ϫ10 Ϫ7 Torr.The Co layer was e-beam evaporated at T S ϭ200 °C.The samples were capped by an Al layer.
The in-plane dependence of the hysteresis loops was measured at RT using a magneto-optic Kerr effect apparatus, before and after the application of field pulses.The pulsed field was always applied in the sample plane and along different directions with respect to the unidirectional anisotropy direction ͑UAD͒.The FeNi /FeMn bilayer was cut in several pieces and pulses were applied at 0°, 45°, 90°, and 180°of the direction of the field applied during growth ͑i.e., the UAD͒.For the NiO/Co samples, the pulse was applied at 45°o f the direction of maximum H E .The rise and fall times of the pulse are about 22 and 150 ms, respectively, while the maximum achieved field is Hϭ556 kOe.
Shown in Fig. 1͑a͒ are the hysteresis loops of the FeNi /FeMn bilayer before and after applying a field pulse antiparallel to the UAD.There is a clear reduction of H E and an increase of H C , after the 180°field pulse.However, applying a pulse along 45°͓Fig.1͑b͔͒ only results in the reduction of H C .Moreover, the hysteresis loops become more asymmetric after the field pulses.When comparing the angular dependence of H E and H C for the FeNi /FeMn bilayer before and after field pulses applied along different directions ͑Fig.2͒, one can observe that the shape of the angular dependence does not change significantly.However, H E becomes smaller as the pulse direction moves away from the UAD.Remarkably, the coercivity decreases, along the unidirectional axis, for pulses applied at 0°͑not shown͒, 45°, and 90°, while it increases considerably for pulses applied along 180°.
The NiO/Co bilayer exhibits a different behavior ͑Fig.3͒.Although, H E decreases and H C increases after a 45°field pulse, the angular dependence of H E changes considerably.Namely, a new UAD, i.e., the direction with maximum H E , is created in a direction closer to the pulsed field.
From the results certain trends can be extracted: ͑i͒ the field induced changes appear to be maximum for pulses applied antiparallel to the UAD; ͑ii͒ the kind of tuning can be controlled by the pulse angle; and ͑iii͒ different AFMs with similar microstructure respond differently to the applied pulses.
The described field induced changes could have different origins, such as changes in the AFM domain structure or field induced transitions in the AFM.Several theories outline the importance AFM domains may have in exchange bias. 13uch AFM domains could be in metastable states.For example, changes in the AFM domain configuration have been claimed to be responsible for time dependence effects 10 or training effects. 14Thus, such domains could probably be altered by the large pulsed fields.Certainly, in NiO single crys-FIG.1. Hysteresis loops measured along the unideirectional anisotropy direction for a Fe 19 Ni 81 /Fe 50 Mn 50 bilayer ͑a͒ before ͑᭹͒ and after a field pulse at 180°away from the unidirectional anistropy ͑᭺͒ and ͑b͒ before ͑᭹͒ and after a field pulse at 45°away from the unidirectional anistropy ͑ᮀ͒.tals AFM domains have been found to reorient after being subjected to large fields, 15 when in contact with a FM ͑Ref.16͒ or when subjected to strains. 17Thus, if the AFM domains are modified by the pulse field, this effect should, in turn, influence H E and H C or even their angular dependence.Moreover, due to magnetostriction the pulsed fields could generate large strains at the interface, which could affect the AFM domain configuration.Note that since the AFM anisotropy for NiO is much smaller than for FeMn , and in the latter case the unidirectional anisotropy is better established, one would expect larger domain based effects in the NiO case.Although the absolute change in H E is larger in FeMn , the fact that the angular dependence in textured NiO actually changes after the field pulse could be regarded as an indication of stronger reaction to the field pulse.
AFM materials have, below T N , a transition from AFM to paramagnetic ͑PM͒ at high enough applied fields. 18Thus, in principle, in order to induce exchange bias it should be analogous to field cooling through T N or to apply large enough fields ͑larger than the AFM-PM phase transition͒ at a given temperature below T N .In both cases, the AFM reaches a PM state and returns to the AFM state in the presence of a field.Since the samples are textured, i.e., without a clear AFM anisotropy axis in plane, pulses applied away from the original UAD should reset it to the pulse direction.However, the critical field for the AFM-PM transition, H AFM-PM , is proportional to the exchange field of the AFM, which can be rather large in FeMn and NiO due to their large T N . 18For example, in bulk NiO, H AFM-PM Ͼ4000 kOe at RT. 19 Therefore, due to the limited strength of the field pulse, AFM-PM transitions would appear as an unlikely cause for the observed effects.However, AFM/FM bilayers typically exhibit rather large distributions of blocking temperatures, ⌬T B , [20][21][22] with a sizable percentage of the AFM particles having T B close to RT.Moreover, considering that H AFM-PM is drastically reduced close to T N ͑or T B ) ͑Ref.18͒ this would lead to some of the AFM particles to actually being sensitive to the field pulse.Consequently, H E and H C could be in practice partially reset.Due to the larger ⌬T B usually observed for NiO, 20,21 this AFM would be expected to be more sensitive to the field pulses.Moreover, ⌬T B will lead to a ''distribution of responses,'' which should result in an asymmetry of the hysteresis loop. 23inally, it is well known that moderate fields applied along the easy axis of the AFM material induce a spin-flop transition 18 ͑e.g., for the NiO the spin-flop field can be estimated to be ϳ90 kOe at RT͒. 19 This transition is known to affect exchange bias for highly crystalline AFM materials. 6ince the spin-flop transition is rather sensitive to the angle between the AFM easy axis and the applied field, 24 this effect should be expected to be small on textured AFMs ͑i.e., with a random orientation of AFM easy axes͒.However, due to random in-plane distribution of crystallites some of them would actually have their easy axes aligned within a few degrees of the pulse direction.Hence, a percentage of particles could contribute to the changes observed in the exchange bias properties and the loop asymmeties.
In conclusion, we have demonstrated that H E and H C can be controlled at room temperature ͑i.e., below T N ), after sample growth, by applying pulsed fields.The effects observed in textured Fe 50 Mn 50 based bilayers are a decrease in H E , while H C can either increase or decrease depending on the direction of the pulsed field.Similar effects are observed in textured NiO based bilayers, although in this case there is a reconfiguration of the unidirectional anisotropy axis towards the pulse direction.Hence, NiO appears to be more sensitive to the pulse, probably due to its smaller anisotropy or its larger T B distribution.These effects are probably due to field induced transitions in the AFM ͑AFM-PM or spin-flop͒ or changes in the AFM domain structure.

FIG. 2 .
FIG. 2. Angular dependence of the exchange bias, H E ͑a͒, and coercivity, H C ͑b͒, for a Fe 19 Ni 81 /Fe 50 Mn 50 bilayer before ͑᭹͒ and after the application of pulses along 45°͑ϫ͒, 90°͑᭺͒, and 180°͑᭡͒ with respect to the unidirectional anisotropy.Lines are guides for the eye.