Relaxation process of Fe(CuNb)SiB amorphous alloys investigated by dynamical calorimetry

Differential scanning calorimetry and dynamic differential scanning calorimetry were used to analyze the relaxation process of Fe (cid:126) CuNb (cid:33) SiB amorphous alloys. The Curie temperature ( T C ) evolution of the amorphous phase during relaxation as a function of heating rate, time and pre-annealing temperature were measured. Two distinct relaxation processes are observed, consequent with topological and chemical short range order changes. © 1997 American Institute of Physics. (cid:64) S0003-6951 (cid:126) 97 (cid:33) 04313-1 (cid:35)

͑Received 28 October 1996; accepted for publication 29 January 1997͒ Differential scanning calorimetry and dynamic differential scanning calorimetry were used to analyze the relaxation process of Fe͑CuNb͒SiB amorphous alloys. The Curie temperature (T C ) evolution of the amorphous phase during relaxation as a function of heating rate, time and pre-annealing temperature were measured. Two distinct relaxation processes are observed, consequent with topological and chemical short range order changes. © 1997 American Institute of Physics. ͓S0003-6951͑97͒04313-1͔ Since the discovery of the excellent soft magnetic properties of nanocrystallized Fe͑CuNb͒SiB amorphous alloys ͑FINEMET͒, 1 much effort has been devoted to investigating the nanocrystalline structure, the magnetic properties, the addition effect, etc. The kinetics of the nanocrystallization process is well established, but the relaxation phenomena before the crystallization as well as their action to induce nucleation have not been analyzed enough. During relaxation, short range order ͑SRO͒ changes occur in the amorphous phase. For instance, both the magnetic permeability and T C change their values. 2,3 In this letter, a systematic analysis of the amorphous phase T C evolution during relaxation and its physical interpretation in terms of SRO is attempted, as resulting from differential scanning calorimetry ͑DSC͒ and dynamic differential scanning calorimetry ͑DDSC͒ measurements.
Amorphous Fe 73.5 Cu 1 Nb 3 Si 22.5Ϫx B x ͑xϭ5, 8, and 12͒ ribbons with 15 mm width and about 20 m thickness were produced by the planar flow casting method. The experiments were carried out on a Perkin-Elmer DSC-7. The DDSC measurements are provided as a facility in the same equipment. The common temperature program is superimposed with a dynamic temperature change. 4 The specific dynamic regime was as follows: repeatedly heat up 2°C at a rate 8°C/min and cool down 1°C at a rate Ϫ4°C/min, in the temperature range from 300 to 350°C. By this technique the complex heat capacity was measured, C P ϭC P Ј ϩiC P Љ . The real part, C P Ј , named ''storage heat capacity,'' is the response of the material to store and release energy ͑reversible change͒; and the imaginary part, C P Љ , named ''loss heat capacity,'' is the tendency of the material to dissipate energy ͑nonreversible change͒. 5 Therefore, the quantity C P Ј gives the change of heat capacity during relaxation without any dissipative response. On the other hand, C P Ј is more sensitive to the structural changes during the relaxation process than the overall heat capacity measured by standard DSC. The DSC heating curves of the amorphous samples show clearly a relaxation process in a temperature range between T C and the first crystallization peak. The amplified heat flow versus temperature for sample A ͑xϭ5͒ is presented in Fig.  1; several DSC curves are included to show the influence of heating rate in the measurement. The relaxation process appears as a very wide exothermic peak which shifts to higher temperatures with increasing heating rate, as expected in an activated process. The average activation energy of the relaxation process was estimated by the Kissinger method ͑shown in Fig. 2͒. The values of the amorphous samples analyzed are in the range of 1.6-1.9 eV, rather insensitive to the chemical composition. These values are lower than the nanocrystallization activation energy of Fe͑Si͒ ͑3.8-4.4͒. 6 Because the relaxation process occurs in a large temperature range, the peak is flat to some extent, and therefore there is a large inaccuracy in the activation energy values measured.
The T C value of the amorphous phase increases as it relaxes and depends on heating rate ͑see Fig. 1͒ because relaxation also occurs below T C . For sample B ͑xϭ8͒ the results of T C versus heating rate are plotted in Fig. 3. The T C values exponentially decrease ͑consider the error͒ with the heating rate, since relaxation already occurred when the sample was heated up. With slow heating rate, more relaxation takes place during the heating, therefore the high heating rate asymptotical value may be considered as the T C of the initial amorphous without relaxation.
To analyze the evolution of the relaxation process, preannealing of the amorphous samples at some selected tem-a͒ Electronic mail: IFFIO@cc.uab.es peratures, T a , for 1 h were performed. Then, to check to which extent the relaxation process occurred during that preannealing, the T C was measured again under continuous heating. The sequential DSC curves of the pre-annealed amorphous samples present a significant change of the profile extending in the whole temperature range where relaxation occurs, as shown in Fig. 4͑a͒. The comparison of the DSC curve of amorphous sample with the DSC curves after pre-annealing indicates that at T a Ͻ450°C the relaxation is still operating after pre-annealing. However, when T a ϭ450°C the relaxation was completed during the preannealing, but if T a ϭ470°C the nucleation and growth of Fe͑Si͒ is even promoted at a lower temperature than that observed without pre-annealing. Another important phenomenon presented in Fig. 4͑a͒ is that the T C of the pre-annealed amorphous phase increases with T a , as plotted in Fig. 4͑b͒. When T a ϭ470°C the T C shows a large increase resulting from some Fe͑Si͒ nuclei formed during annealing. There are only a few Fe͑Si͒ nuclei in the sample, even though the effect is clearly detected by the T C value.
As is well known, the relaxation process is related to some atom rearrangement ͑chemical or topological SRO changes͒ indirectly observed as energy is released in DSC curves. 2,3 Owing to the fact that T C depends on the average distance between magnetic atoms, the T C increase can probably be explained as SRO changes where the Fe-Fe bond distance decreases, the Cu, Nb, and/or B atoms diffusing into the surroundings. The local structure in the Fe neighborhood has the tendency to transform towards ordered arrangement like bcc or DO 3 , becoming embryos of the Fe͑Si͒ phase as corroborated by Mössbauer analysis. 7,8 To investigate the time dependence of relaxation in situ measurements of the T C evolution by annealing in the temperature range of 300-350°C were carried out by DDSC. The real value of T C appears in these measurements as the peak of heat capacity C P Ј without any effect of dissipative C P Љ . In Figs. 5͑a͒ and 5͑b͒ we present by C P Ј of samples B and C ͑xϭ12͒, respectively. It is clear that T C increases with annealing time. As shown in Fig. 6, in the first few hours the relaxation process proceeds rapidly, then it becomes slow. The shape of T C versus annealing time cannot be fitted by only one exponential function. Two relaxation times, 1 and 2 are necessary for each sample. That is where the T 0 is the T C value of the initial amorphous without relaxation, ⌬T i is the T C change due to i relaxation process, and i is its characteristic decay time. In our analysis, both ⌬T i and i are assumed constant in the temperature range explored.
Best fitted values are: T 0 ϭ318.1°C ͑B͒ and 313.0°C ͑C͒, which are in agreement with the DSC continuous heating results; these data show that the T 0 of amorphous depends on the initial chemical composition. The characteristic decay time 1 for both samples B and C is close to 0.45Ϯ0.03 h, the characteristic decay time 2 value is 5.5 h ͑B͒, and 6.6 h ͑C͒. The values of ⌬T 1 are 11.2°C ͑B͒ and 11.9°C ͑C͒, and are very similar. The quantity ⌬T 2 has nearly the same  value, 5.16Ϯ0.01°C, for both samples. It means that independent of the composition, two different relaxation processes influence the T C value.
The rapid relaxation process ͑named 1͒ is probably the response to the topological rearrangement in some local regions, acting relatively fast at the measured temperature. The slow relaxation process ͑named 2͒ could account for the chemical SRO tendency which needs enhanced atomic mobility not available at these relatively low temperatures, but promoting nuclei formation at higher temperatures. The T C is more sensitive to the topological SRO changes than to the chemical SRO ones, since ⌬T 1 is about twice of ⌬T 2 . That is, shortening the Fe-Fe bond distance significantly affects T C . However, once some embryos appear, large changes of T C will come out ͓see Fig. 4͑b͔͒.
From the systematic analysis of the relaxation on some FINEMET amorphous alloys, as resulting from the heat treatment monitored by DSC and DDSC experiments, it is shown that most of the relaxation heat is released in the temperature interval between T C and the onset of nanocrystallization, with a mean activation energy of 1.75Ϯ0.15 eV.
From the storage heat capacity, the in situ measurement by DDSC in the temperature range of 300-350°C shows the T C enhancement with the annealing time. It was found that there are two distinct relaxation processes with characteristic decay times 1 ϭ0.45Ϯ0.03 h and 2 ϭ6Ϯ0.5 h, which are suggested to be caused by topological and chemical SRO changes, respectively. Financial support from CIRIT by Project No. 1995SGR-00514 is acknowledged.