Magnetisation reversal in Co-Cr thin films

This thesis contains the results of the characterisation of Co-Cr films for perpendicular recording and multilayer films for magnetooptica! recording. The research is carried out in the scope of the continuous demand for higher bit densities in data storage systems. Higher bit densities also imply more transitions of the magnetisation per unit of lengtIl, making the knowledge about the reversa! mechanism very important to maintain a goOO signa! to noise ratio. Until now the switching of magnetic materia! can be Studied macroscopically with conventiona! methOOs such as V SM or sQum, but a microscopica! Study of the switching of individua! bits is not possible. Another approach to this problem are computer simulations of the reversa! process. At present the assumptions and the limited number of ce1ls make them not yet realistic enough, but with increasing computer power and more sophisticated programs they become of more and more interest. The anoma!ous Hall effect can be used as a very sensitive tool for thin films and small structures and can operate in the area between the conventiona! macroscopic measurement techniques and micromagnetic simulations. For this purpose an anoma!ous Hall effect measurement set-up has been built. It can perform computer controlled measurements on samples with different si zes and shape. Much attention has ~n given to the sensitivity and the noise in the system. The sensitivity of the AHE set-up was large enough to detect the reversa! of individua! columns in a Co-Cr film. With this set-up Co-Cr films and multilayer fIlms have been measured, we start here with the latter. In chapter 6 the results of the anoma!ous Hall measurements which have been carried out on Co/Pt and Co/Pd multilayers to determine the anisotropy of these films were shown. The methOO proposed by Hajjar [ref. 7.1: Hajjar] was applied and two series of multilayers were used with different Co-thicknesses. The results were compared with torque measurements. From the results it appeared that there are severa! drawbacks to this method as the magnetisation in the sample must be perpendicular (therefore MLs with a thick Co layers having an in-plane anisotropy are not suitable) and the applied field must be high enough to saturate the film completely. Only samples with a perpendicular anisotropy and which do not have a very high anisotropy were therefore in good quantitative agreement with the torque values. Por the Co/Pd a very large contribution of the planar Hall effect could also be observed in the measurements and care had to be taken to adjust the angle y (see figure 6.4) very accurately to zero before starting the measurement. The main attention in this thesis is however focussed on Co-Cr films for perpendicular recording in which submicron Hall-structures have been prepared. These Hall crosses have only a limited number of columns and are used to investigate the reversal mechanism. A procedure to prepare submicron samples by e-beam lithography and ion-beam etching in Co-Cr was developed and was described in detail in chapter 3. The final results were weIl defined Hall crosses with smallest dimensions of 0.3 x 0.3 Jlm2. The damage due to the etch process was less than 50 to 100 nm on the edges of the contact paths and the etch angle was about 60° observed on cross section SEM images. First the effect of the smaller Hall crosses was studied by measuring the AHE hysteresis curves of several samples with different sizes of the cross trom the 'Twente' sample. The expected increase in the slope of the loop is consistent with the decrease in demagnetising field inside the Hall cross due to the fact that a lot of Co-Cr has been removed in the vicinity of the cross. The decrease in N z has also been calculated for blocks and strips and the outcome is qualitatively in good agreement with the measurements. More important is that besides the slope the shape of the curve is not altered, this may indicate that the reversal behaviour and the domain pattem in the cross is not influenced too much by the reduction of the Hall cross size. Prom measuring detailed hysteresis curves (with 4000 points) of the 'Twente' sample it was clear that there were several steps in the Hall curve which were larger than the noise, but only a few correspond with the reversal of one column. The largest steps in the 'Twente' sample correspond to the reversal of a magnetic volume of one average column, however there are not many steps with this size. To get a good idea of the noise in the measurement, statistical analyses were carried out and the original hysteresis loops were divided into 5 loops with steps of different sizes. This resulted in Stepsize-distinguished Partlal Hysteresis loops (SPH-loops) with only large steps (Class 1), small steps (Class 2), very sma11 steps (Class 3) and two loops with reverse steps (Class 3 and 4). ft is interesting to see that the coercivity is smaller for SPH-loops with smaller steps (volumes) and that for the SPH-loops with larger re verse steps the coercivity increa$es again with stepsize. It was also ohserved that even for SPH-Ioops with very small steps the loop is not closed indicating still a magnetic contrihution (reversihle rotation). To study the reversal mechanism in more detail three samples with completely different magnetic properties were investigated. The already mentioned 'Twente' sample was sputtered at RT and had a coercivity of 88 kNm. The other two samples were a low and a high T s (= suhstrate temperature during deposition) 'Maeda' sample which were sputtered at 40°C and 250°C and which had a coercivity of 12 and 176 kNm respectively. The low Ts sample had a large contrihution of the normal Hall effect and the curves were also difficult to measure, therefore we concentrare on the samples with the moderate coercivity ('Twente') and the high coercivity (high T s) sample. The main difference was that the 'Twente' samples show steps in the hysteresis curves with meta-stahle states (plateaus) between these steps as the low and high T s samples do not show steps corresponding with the switching of significant parts of columns. From the detailed hysteresis loops it was deduced that the largest steps occur around the coercivity as expected and they are about equally distrihuted for the field values higher and lower than Hc. The cumulative reverse switching volume is very large for the high T s sample ( -60% ) and there are no steps in the hysteresis curve observed. From literature it is known that steps have also been ohserved in the hysteresis curve due to the switching oftwo columns [ref. 7.2: Wehh]. The film thickness of this film was more than I micron, with better developed columns and with an average column diameter of 100 nm. The reason for this multiple-column switching could be that there is more exchange between columns as they grow larger. From bis experiment Wehh concluded that 85% ofthe reversal took place hy discrete steps. The coercivity of the three samples was also measured with the anomalous Hall effect for several samples and it was surprising that the coercivity of the 'Twente' sample increases with decreasing Hall cross size as the H c of the high T s sample was rather unaffected hy the size reduction. A possihle explanation for this discrepancy is that a large part of the reversal in the 'Twente' sample takes place hy domain wan motion and the edges of the Hall cross can act as pinning points which increases the coercivity. ff the high coercivity sample has a more particulate like reversal behaviour (rotation of the magnetisation) than the edges do not influence the coercivity . From the MFM-images it appeared that the magnetic structures in the high T s film are larger (0247 nm) than for the 'Twente' sample (0 150 nm). Both results indicates that there are many columns in one domain (in the remanent state). It is likely that these differences between the 'Twente' sample and the high T s sample can be related to the differences in deposition parameters which result in a different structure and chemica1 composition. The structura1 parameters have been ana1ysed with SEM and AFM and it was shown that the colurnn diameter was a1most the same for both fi1ms (0 75 and 70 nm) and the high T s fi1ms show a large surface roughness. Another important parameter which influences the reversa1 is the distribution of the Cr in the Co-Cr fi1ms. Co-Cr is believed to have a colurnnar structure with Cr segregation on the colurnn boundaries. To verify this assumption EDX (Energy Dispersive X-ray) measurements were performed on an as-sputtered sample (Sl) and on an annea1ed sample of Co-Cr (S2) with a thickness of 60 nm. The substrate was loca1ly removed by chemica1 etching until a window appeared in the substrate to a1low planar view TEM observations. The annea1ed sample (300°C for 30 rnin) has a higher coercivity (16%) and a higher squareness (0.38 instead of 0.33). A third sample (S3) with an origina1 thickness of 200 nm was ion-rnilled on both sides first to reduce the thickness local1y to less than 100 nm to a1low TEM observations (actua1ly a hole appeared in the membrane ). This sample is a Co-Cr sample which is a1so used for the Hal1 measurements on submicron structures. The composition of the Co-Cr was investigated for these three fi1ms by EDX and the results for al1 three samples show that there is a -1.8 % difference in at% Cr concentration between the boundaries and the inside of a Co-Cr colurnn (with more Cr on the boundaries). The influence of the annealing on the compositiona1 separation was not so apparent but the standard deviation for the Cr- concentration on the boundaries was significant1y larger. The third sample has the same differences in Cr concentration between the boundaries of a colurnn and inside, but a strange phenomena occurred in the grains at the edge of the hole in the window as the Cr-concentration was considerably lower here (14 instead of 22 at% Cr), this can be caused by an etching effect as the Cr etChes more easi1y than the Co during the ion-rnilling. In 1987 Maeda proposed a chrysanthemum pattem like structure (CP- structure) of the Cr distributed in the Co-Cr colurnns [ref. 7.3: Maeda]. He observed this pattem for Co-Cr fi1ms after selective chemica1 etching of the Cr by TEM and by SEM images. The fi1ms were sputtered on substrates with a high temperature during deposition and these fi1ms had a1so very high coercivity. One of the these samples with an assumed CP-structure, the high T s sample, together with a low T s sample and the lTwente' sample were ana1ysed by NMR and by SEM after selective etching. The NMR spectra were taken and there was a shift of the main peak (M) and from the appearance of satellite peaks it has been shown that the high coercivity filrns contain highly Co-rich regions with 90 at% Co and 97 at% Co for the 'Twente' and the high T s sample respectively. Selective etching revea1ed stripes in the high T s sample and faint stripes in the 'Twente' sample. in the low T s sample no stripes were observed. The stripes indicate a CP structure and segregation of Cr inside the colurnns. On the other hand the EDX-measurements seem to indicate that a Co-rich column is surrounded by a Cr-rich boundary. but the resolution of the EDX is much less than the resolution of the NMR which has a resolution on the atornic sca1e. Recent1y more evidence was found for the CP-structure by atom probe field ion rnicroscopy (APFIM) [ref. 7.4: Hono]. From the in-plane VSM hysteresis curve it was deduced that the 'Twente' sample has a large initial layer and that this can a1so influence the reversa1 behaviour. Also from the initial curves measured with a VSM and the HJHk va1ues different conclusions could be drawn for the magnetic switching of the materia1. According to the theory of Kronmtiller the Co-Cr columns behave like multidomain particles [ref. 7.5: Kronmtiller] and from the HJHk va1ues of the samples a stripedomain pattem for the low T s sample and a more dot1ike domain pattem for the 'Twente' and the high T s sample is expected [ref. 7.6: Lodder]. The computer simulations showed the results of three Co-Cr samples with 25 columns and different volumes of in-plane components of the magnetisation (0. 2 and 8 % ). The results show some interesting sirnilarities with the measurements such as the lower Hc for the SPH-loops of sma11er volumes and higher Hc for loops with only larger volumes. Also a11 three samples showed the 'plateaus' between the steps. A larger in-plane volume lowers the coercivity and facilitates the rotation of the magnetisation and the separate reversa1 of columns. It can be concluded that the reversa1 mechanism of Co-Cr is sti11 not revea1ed completely. it is very strongly influenced by many different parameters such as temperature during deposition. substrate type. target composition. pressure etc. More measurements with the anoma1ous Ha11 effect on subrnicron Ha11 structures (as function of temperature and time) can be useful to study the interaction between the switching volumes. Also domain observation during the reversa1 inside the Ha11 cross is very interesting. to estimate the disturbance of the reversa1 process caused by the sma11er dimensions and correlate the change m magnetisation to a domain structure.