Copper dusting effects on perpendicular magnetic anisotropy in Pt/Co/Pt tri-layers

The effect of Cu dusting on perpendicular magnetic anisotropy of sputter grown Pt/Co/Pt stack in which the Cu layer is in proximity with that of Co is investigated in this work. We used magneto optic Kerr effect microscopy measurements to study the variation in the reversal mechanisms in films with Co thicknesses below 0.8nm by systematically varying their perpendicular magnetic anisotropy using controlled Cu dusting. Cu dusting was done separately above and below the cobalt layer in order to understand the role of bottom and top Pt layers in magnetization reversal mechanisms of sputtered Pt/Co/Pt stack. The introduction of even 0.3nm thick Cu layer below the cobalt layer drastically affected the perpendicular magnetic anisotropy as evident from the nucleation behavior. On the contrary, even a 4nm thick top Cu layer had little effect on the reversal mechanism. These observations along with magnetization data was used to estimate the role of top and bottom Pt in the origin of perpendicular magnetic anisotropy as well as magnetization switching mechanism in Pt/Co/Pt thin films. Also, with an increase in the bottom Cu dusting from 0.2 to 0.4nm there was an increase in the number of nucleation sites resulting in the transformation of domain wall patterns from a smooth interface type to a finger like one and finally to maze type.


I. INTRODUCTION
Perpendicular magnetic anisotropy (PMA) in ultra-thin films of Co has been an active area of research since 1980s due to its potential applications in magnetic recording. 1High values of PMA are preferred since they offer better retentivity of data and at the same time allow the reduction of the physical dimension of the bits without yielding to the thermal fluctuations. 2Systems with PMA are also preferred in magnetic tunnel junctions due to their lower switching current densities. 3here has been a recent surge in research in these systems due to predictions and observations of interesting new physics of spin orbit torque (SOT) 4,5 when the ferromagnetic layers are in proximity with heavy metals like Pt, Pd, Ta etc.Most of these important studies deal with systems having PMA because this ensures an effective spin orbital coupling from the heavy metals.This results in an anti-symmetric exchange like Dzyaloshinskii-Moriya interactions 6 which can stabilize new phases like skyrmions. 7xtensive research work has been carried out in recent times in the investigation of physical properties such as magnetization, 8 anisotropic constant 2 and the effect of introducing ultra-thin films of Cu, Ta etc. 9 (known as dusting) in proximity with Co layer in Pt/Co/Pt thin films.Here, we look at the effect of Cu dusting on the magnetization reversal mechanism in such films.This will determine the dynamics of the switching which are important for high speed data write and read processes.The role of top Cu layer in preventing the Pt sputter induced damage 9  the role of top Pt in the evolution of PMA is understood.Most importantly, we demonstrate that that by introducing Cu below Co, we can tune the anisotropy in such a way as to explore different parts of the magnetic domain phase diagram 10 (for systems with PMA) without making stacked multilayers 11 of Co/Pt.This study is important in the understanding of different domains structures and magnetization reversal in Co/Pt systems.

II. EXPERIMENTAL
Pt/Co/Pt thin films were grown on thermally oxidized silicon substrates at room temperature by dc magnetron sputtering in argon atmosphere at a working pressure of 5.6x10 -3 mbar in a high vacuum chamber maintained at a base vacuum of less than 3x10 -7 mbar.The deposition was monitored by quartz crystal monitor, which was carefully by measuring the thickness of deposited films using atomic force microscopy.A 3nm thick Ta film was grown as buffer layer on SiO 2 which promotes (111) textured growth in the bottom Pt layer in the Pt/Co/Pt stack.The deposition rates of different elements were Co = 0.2A/sec, Pt = 0.6A/sec, Cu = 0.3A/sec and Ta = 0.3 A/sec.Samples were grown with Cu dusting below and above the Co layer.In the case of bottom Cu dusting, two sets of samples with Co thicknesses of 0.52nm and 0.70nm were grown which will be here after referred as set A and B respectively.Here the Cu thickness was varied from 0.2nm to 0.6nm.In case of samples with top Cu the dusting thickness was chosen as 4nm, which was thick enough to separate the top Pt from seeing the Co layer.The magnetization curves were obtained using laser based polar MOKE setup.Magnetization reversal studies were carried out by domain imaging (in polar configuration) using Kerr Microscope from Evico Magnetics.The domain images shown in this article were taken while the magnetization reverses from up to down state.In our convention the darker and lighter contrast corresponds to down and up magnetization directions respectively.The saturation magnetization (M s ) and effective anisotropy (K eff ) of the samples are estimated by using SQUID magnetometer (Quantum Design) by performing room temperature out-of-plane and in-plane magnetization measurements.For the PMA samples, the in-plane (hard axis) saturation field is taken as the anisotropic field (H k ) and the effective anisotropy is estimated by using the equation K eff = H k M s /2.

A. Effect of Copper under layers
The Fig. 1(a) and 2(a) shows the polar hysteresis of the samples with Co thicknesses of 0.52nm (set A) and 0.70nm (set B) respectively.The samples without Cu dusting from both set A and B exhibit a square hysteresis loop indicating fewer nucleation sites and fast domain wall motion, as evident from the domain image in Fig. 1(b) and 2(b) respectively.These images were captured during the transit of the domain wall in the field of view of Kerr microscope.The domain walls in these cases are found to be relatively smooth and flat, with occasional pinning sites.Now for set A we can see that the introduction of even a 0.2 nm thick Cu film below the Co layer leads to a reduction in the Kerr rotation of the sample; which is a result of the high reflectivity from non-magnetic Cu layer [Fig.1(c)].In this sample, during reversal we can see finger like short stripes of un-reversed domains.At a Cu thickness of 0.3nm the domain patterns during the switching process shows [Fig.1(d)] longer stripes of un-reversed domains separating the larger reversed domains when compared to samples with 0.2nm Cu dusting.These observations can be attributed to striped phases coexisting with the up and down domains as discussed by Y.L. Iunin 11 et.al.As the thickness of Cu is increased to 0.4nm [Fig.1(e)] the remanence suddenly drops to zero in the reversal curve.Kerr microscope imaging for this sample shows a disordered maze like multi-domain state, with domain widths lesser than 2microns [as shown in inset of Fig. 1(e)].The imaging was done at the resolving limits of the microscope.We will need imaging techniques with higher resolution to probe these states, such as scanning electron microscopy with polarization analysis, 12 magnetic force microscopy 13 etc. which is beyond the scope of the current work.It can also be seen that this sample gets saturated at a very low out of plane field (<50 Oe), still indicating the presence PMA but is

056122-4
Parakkat, Ganesh, and Anil Kumar AIP Advances 6, 056122 (2016) be smaller than the resolution limit of the optical microscope.Finally, for set A sample with 0.6nm Cu dusting, there was no signature of saturation up to a field of 600 Oe.The decrease in PMA (K eff ) with increase in Cu dusting thickness is evident from the Fig. 1(g), which also correlates well with the observed domain images.For thicknesses above 0.4nm of Cu, the samples show identical saturation behavior in the in-plane and out-of-plane directions, which indicates absence of PMA.(Supplemetary 14 data on magnetization measurements are provided).
In the case of set B samples (with 0.70nm Co), we see a systematic evolution of domain pattern as the Cu dusting layer thickness is increased, but with some observable differences as compared to those of set A. At a Cu thickness of 0.2nm the domain wall seems to be smoother as in the case of similar sample from set A, but has comparatively very less pinning sites.When the Cu dusting thickness was increased to 0.3nm, we can still observe square hysteresis with 100% remanence but during reversal there are large numbers of un-reversed narrow stripe domains.These stripes shrink in size and disappear upon increasing the field.There is also an increase in the number of nucleation sites in this sample.The sample with Cu thickness of 0.35nm indicated a random maze pattern of up and down domains.The hysteresis loop shows that the remanence was very low but non zero as unreversed domains were bigger in size than reversed ones.For a dusting thickness of 0.4nm we again observe a similar trend as in the corresponding sample in set A, showing very low remanence.In this case also we expect very narrow and densely intertwined domains of up and down magnetization at remanence.Kerr imaging was not possible in this case due the extremely small size of domains.For samples with 0.5 and 0.6 nm Cu dusting, no sign of saturation was observed till 600 Oe of applied magnetic field.This is in contrast to the corresponding set A samples where a 0.5nm Cu dusting gave a hysteresis loop which saturated at higher field (∼200 Oe).This can be attributed to fact that set B sample having a higher Co thickness of 0.7nm will have a higher demagnetization field in out of plane direction, compared to set A. The variation in PMA of set B samples is similar to that of set A [Fig 2(f)].It is found that the set A having lower Co thickness has comparatively higher PMA than set B.
All these experiments clearly indicated that Cu dusting as an under layer has a significant effect on the magnetization reversal mechanism of the Co layer and also its PMA.This also indicates that the PMA predominantly originates from the interface of Co with the bottom Pt layer [Fig.1(g)].This may be due to strained growth of Co on Pt or due orbital overlaps of Co and Pt giving rise to spin orbit coupling. 15The copper being immiscible with Co removes the possibility of any intermixing or alloying. 16The introduction of Cu might affect the growth of Co and it microstructure, which might be leading to reduction in PMA.The Cu layer at low thickness of 0.2 nm may not be continuous and hence leads to the formation of localized regions of weaker PMA.These regions acts as nucleation sites for reversal at lower fields whereas the regions of higher anisotropy acts as strong pinning sites which in turn gives rise to stripe domains during reversal.These stripes continue to exist due to dipolar repulsion between the big reversed domains until the applied field reverses them by narrowing them down.It was observed that as the Cu thickness becomes 0.3nm the length of these stripes becomes longer and is less susceptible to reversal.This may be attributed to the weakening of PMA [Fig.1(g)] which in turn can lead to the formation of larger areas/lengths of domain walls.At Cu thickness of 0.4nm the PMA is very weak and there is probably a delicate balance of PMA and the demagnetization energy, which finally leads to a disordered multidomain maze pattern with zero net magnetization.

B. Effect of Copper capping layers
In contrast to underlayer dusting with Cu, the top Cu (above Co) has very little effect on the domain structure of the sample.Figure 3(a) shows out of plane hysteresis loops of samples with 0.52nm thick Co with and without Cu capping indicating square hysteresis.There were fewer nucleation sites and the reversal predominantly happened by domain wall motion, even in the case of the sample with Cu capping thickness as high as 4nm [Fig.3(b)].The K eff of samples with and without top Cu [K eff = (9.16± 1.07)×10 6 erg/cm 3 and K eff = (8.36 ± 0.627)×10 6 erg/cm 3 respectively] were found to be similar.This confirms the fact that the top Pt had minor role on enhancing the PMA of Pt/Co/Pt films.An improvement in coercivity is observed for Cu capped sample.This may be attributed to fewer nucleation sites, as result of less miscible Cu capping which in turn reduces sputtering damage from the top Pt.

IV. CONCLUSION
In this paper we demonstrate the effect of top and bottom Cu dusting on the magnetization reversal of Pt/Co/Pt trilayers with PMA.We show that by controlled introduction of Cu as an underlayer, different phases of domain configurations can be achieved.We were able to demonstrate that by varying the Cu dusting thickness a fine control could be achieved over the transition from a PMA to a dipolar dominated system.The domain imaging revealed interesting reversal modes which otherwise was only possible with more complicated multilayer stacks.Bottom Cu dusting resulted in the evolution of domains with smooth edges which further transformed to dendritic formation and finally ending up in a disordered maze like pattern.The introduction of Cu as capping material to Co maintained the PMA which indicated minimal role of the capping Pt to PMA of Pt/Co/Pt film.This study sheds light on the possibility of using non-magnetic dusting layers to control the anisotropy and reversal mechanisms of systems with PMA.
FIG. 1. a) Polar magneto-optic Kerr effect data of Ta 3nm/Pt 3nm/Cu (0-0.6nm)/Co0.52nm/Pt 2nm samples.b-f) Domain images of samples with Cu under layer thicknesses as indicated.The corresponding scale bars are shown with the images.g) Variation of K eff with underlayer copper thicknesses.
FIG. 3. a) Polar MOKE data showing the effect of Cu capping layer.The coercivity of Pt/Co/Cu is significantly higher that Pt/Co/Pt sample.b) Domain image of Ta3nm/Pt3nm/Co 0.52nm/Cu 4nm/Pt2nm films while magnetization reversal.
of Co as well as