Rapid magnetic hardening by rapid thermal annealing in NdFeB-based nanocomposites

Transcription

1 INSTITUTE OF PHYSICS PUBLISHING JOURNAL OF PHYSICS D: APPLIED PHYSICS J. Phys. D: Appl. Phys. 38 (5) 9 1 doi:1.188/-377/38// Rapid magnetic hardening by rapid thermal annealing in NdFeB-based nanocomposites Kung-Te Chu, Z Q Jin, Vamsi M Chakka and JPLiu Department of Physics, University of Texas at Arlington, Arlington, TX 719, USA pliu@uta.edu Received August 5, in final form 5 September 5 Published 7 November 5 Online at stacks.iop.org/jphysd/38/9 Abstract A systematic study of heat treatments and magnetic hardening of NdFeB-based melt-spun nanocomposite ribbons have been carried out. Comparison was made between samples treated by rapid thermal annealing and by conventional furnace annealing. Heating rates up to K s 1 were adopted in the rapid thermal processing. It was observed that magnetic hardening can be realized in an annealing time as short as 1 s. Coercivity of 1. koe in the nanocomposites has been obtained by rapid thermal annealing for 1 s, and prolonged annealing did not give any increase in coercivity. Detailed results on the effects of annealing time, temperature and heating rate have been obtained. The dependence of magnetic properties on the annealing parameters has been investigated. Structural characterization revealed that there is a close correlation between magnetic hardening and nanostructured morphology. The coercivity mechanism was also studied by analysing the magnetization minor loops. (Some figures in this article are in colour only in the electronic version) 1. Introduction Magnetic composites with an exchange interaction at nanoscale between magnetically hard phase with high anisotropy and soft phase with high magnetization have attracted great attention since first initiated by Coehoorn et al in 1989 [1] due to their potential to have extremely high maximum energy products. Various techniques, including melt spinning, mechanical milling, sputtering and chemical synthesis, have been explored to prepare nanocomposite magnets. Since grain coarsening will result in the occurrence of exchange decoupling and significant deterioration of the magnetic properties, the key issue in the fabrication of these nanocomposite materials is the deliberate control of nanostructured morphology [ 7]. Recently, it was reported that rapid thermal processing (RTP) with heating rate up to several hundreds of degrees per second and also with high cooling rate can be utilized to prepare ultrafine nanocrystalline magnetic materials, due to its significant effect on the dynamic crystallization behaviour and nucleation rate [8 1]. One of the heating modes involved in the RTP technique is isothermal annealing which designates constant temperature around the samples and gives a uniform lateral and transverse heating profile throughout a wide area. Controlled annealing has resulted in high coercivity values in nanostructured NdFeB [8] SmCo [9] and FePt thin films [11]. High heating rate results in reduced grain size, which accounts for the improved hard magnetic properties. It is also noteworthy that the transition temperature of FePt nanoparticles from a chemically disordered structure to an ordered structure, which leads to magnetic hardening, can also be reduced with the adoption of RTP as compared with the conventional furnace annealing [1]. The remanence ratio was increased to higher than.5 in partially ordered nanoparticle assemblies after the RTP treatment. Different coercivity mechanisms were observed upon processing at different cooling rates [13]. A primary question remaining in the area is how fast the magnetic hardening can be realized in a given magnetic material. In this study, the magnetic properties of NdFeB nanocomposite systems developed by conventional furnace annealing and rapid thermal annealing have been compared -377/5/9+$3. 5 IOP Publishing Ltd Printed in the UK 9

2 K-T Chu et al and investigated in detail to try to resolve this issue. RTP with a heating time as short as 1 s has been adopted in this work, and very fast magnetic hardening has been realized in the RTP processed nanocomposite samples. The coercivity mechanism of these materials has been discussed following the magnetic analysis.. Experimental procedure Starting materials used for this investigation are amorphous ribbons prepared by melt-spinning with approximately µm thickness and 3 mm width. The selected compositions are Nd 9 Co 1 Ti B 1.5 Fe 8.5 (referred to as NFe9), Nd Fe 1 B/Fe 3 B (15 vol % Fe 3 B) (NFe15), Nd Fe 1 B/α-Fe ( vol % α-fe) (NFe). RTP experiments were performed by using Modular Process Technology s isothermal type Rapid Thermal Processor RTP-S. The RTP-S used high intensity visible/infrared radiations from tungsten halogen lamps to heat a single wafer, on which the ribbon samples were placed, for a short time at precisely controlled temperatures. The tungsten halogen lamps and quartz chamber allowed fast wafer heating rates. Purging N and water-cooling system were used to control the temperature of chamber and the cooling rate after a heating cycle. It could reach heating rates in the range of C s 1, which is user controllable, and cooling rates up to 15 C s 1. The steady state temperature stability was about ± C and had a uniform temperature within the chamber. Annealing temperatures (5 8 C), annealing times (1 s), heating rates ( C s 1 ) and annealing times of RTP in an Ar environment were investigated in comparison with conventional furnace annealing (ramp rate of.5 1 C s 1 ) at 5 8 C for 1 3 min. Siemens D5 powder x-ray diffractometer (Philip- Cu K α radiation) and JEOL-1 transmission electron microscope (TEM) were used for examining the crystal structure and the grain morphology. The crystallization and the phase transition temperature of the amorphous NdFeB samples were analysed using a Perkin-Elmer DTA7 differential thermal analyzer (DTA) in an Ar atmosphere from 5 to 8 C at different heating rates. The magnetic properties of crystallized samples were measured using either alternating gradient magnetometer (AGM) (Princeton Measurements Corporation Model 9 MicroMag ) at an applied field up to 1. T or superconducting quantum interference devices (SQUID) with a maximum applied field of 7 T. After several measurements, we found that the ribbons prepared in the same melt-spinning process with same composition exhibited minor differences in their magnetic properties which may be due to non-uniformity in their composition. Therefore, to ensure the reproducibility of the measurements, each measurement was based on an average of several samples processed at the same condition and for each sample in two directions which were perpendicular to each other. It is well known that coercivity has a strong dependence on grain size. The prevailing coercivity mechanisms in hard magnetic materials, including domain wall pinning, reverse domain nucleation and coherent rotation, depend heavily on the morphology of these materials which in turn depend on the processing techniques used. In order to investigate the coercivity mechanisms in the nanocomposites, Intensity (a.u.) Nd Fe 1 B α Fe FA ( o C, 1 min) RTP ( o C, 1s) As-spun 3 5 θ (degree) Figure 1. XRD patterns of the NFe9 sample in as-spun status and the samples subjected to rapid thermal annealing at C for 1 s or furnace annealing at C for 1 min. the minor magnetization loops measurements were performed by using AGM. 3. Results and discussions The analysis of the crystal structures using XRD indicated that the amorphous structure of as-spun samples crystallized into a nanocomposite structure consisting of soft magnetic α-fe and hard magnetic Nd Fe 1 B tetragonal phase. Figure 1 shows the XRD patterns of as-spun sample NFe9 and those annealed using furnace or RTP at C. The intensity distribution among the Nd Fe 1 B peaks indicates no texture in the samples. It can be seen from the peak broadening and intensity ratio between Nd Fe 1 B and α-fe that the peaks of sample in rapid thermal annealing for 1 s is much lower in intensity and broader than those in furnace annealing for 1 min, revealing that the grain size in the former is smaller than that in the latter situation. It was assumed that there is a possibility of the presence of Fe 3 B, but the Fe 3 B peaks are not clearly visible in the XRD pattern due to their immersion in background noise. It is known that high volume fraction of soft α-fe with a large grain size would be fatal to the coercivity of traditional sintered NdFeB magnets, while the α-fe with ultrafine grain size is necessary to achieve an effective exchange coupling and enhanced high magnetic properties in the nanocomposite magnets. The magnetic measurements showed that the as-spun ribbons of amorphous structure had a nearly zero coercivity. After heat treatment at elevated temperatures, coercivity was developed in the nanocomposites. The annealing temperature dependence of coercivity of the samples for both furnace annealing and rapid thermal annealing are plotted in figure, where the annealing time for furnace annealing has been chosen to be the one which gave the highest coercivity. Apparently, no hard magnetic Nd Fe 1 B phase was formed for annealing temperature below 5 C for NFe9 and NFe15 because it is below their crystallization temperatures around C (as examined by differential thermal analysis) [1]. The formation of the hard phase is necessary for magnetic hardening. It is also noticed that both the furnace annealing and RTP above C resulted in magnetic hardening although the processing time of RTP was as short as 1 s. The coercivity values of the RTP samples are slightly higher than the values of 1

3 Coercivity (koe) 1 8 NFe9 NFe15 NFe Coercivity (koe) 8 Rapid magnetic hardening in NdFeB-based nanocomposites o C 7 o C 8 o C NFe9 NFe Annealing time (minutes) Figure 3. The coercivity as a function of annealing time for NFe9 and NFe samples after furnace annealing at different temperatures. Furnace Annealing RTP-Ramp Rate ( o C/s) Annealing Temperature ( O C) Figure. The coercivity as a function of annealing temperature for samples of three different compositions after furnace annealing for optimized time or RTP treatments for 1 s. the conventionally furnace-annealed samples. They are 1 koe for NFe9, 5 koe for NFe15 and 5. koe for NFe (which is close to the furnace-annealed samples), respectively. High coercivity values of all the samples were usually achieved at 7 C. Higher annealing temperatures led to a decrease of coercivity for most samples due to grain coarsening. It is also noticed that the optimal annealing temperature of C for NFe9 is lower than that of 7 C for NFe15 and NFe samples. The addition of Co in NFe9 apparently reduced the crystallization temperature of Nd Fe 1 B phase, which is in agreement with a previous report [15]. On comparing the coercivity of all the annealed samples, it was also found that the NFe9 samples have higher coercivity than others, which may be due to different compositions, i.e. the addition of Co and Ti and the higher B content which either promoted the formation of Nd Fe 1 B or resulted in the grain refinement and homogenization of crystallized microstructure in the samples. Such crystallization acceleration and grain growth inhibition effects have also been widely reported in other references [15 19]. Moreover, the magnetic hardening for NFe has been observed for furnace annealing at 5 C, while no detectable coercivity appears for RTP sample. This difference may be from the well-known phenomenon that the lower heating rate would lead to lower phase transition temperature. Low crystallization temperature has also been reported in other NdFeB nanocomposites with high Fe and low B components [,1]. The increase of coercivity with the annealing temperature in the range of 5 7 C reflected the gradual crystallization of the amorphous phase. This was also evidenced by TEM analysis which showed the presence of the amorphous phase along the grain boundaries. The amorphous phase can also act as an interfacial layer as observed in other cases [1]. Coercivity (koe) 1 8 NFe9 NFe NFe15 Solid line-low ramp rate Dash line-high ramp rate Annealing time (minutes) Figure. The coercivity as a function of annealing time for the samples after RTP treatments at C with low ( C s 1 ) or high ( C s 1 ) heating rates. Since the annealing time is a great issue to be considered for the optimization of heat treatment process and the improvement of magnetic properties, the dependences of coercivity on annealing time for both furnace annealed samples and RTP samples were also investigated and depicted in figures 3 and, respectively. Herewith, the annealing temperature in the case of furnace processing (figure 3) has been optimized at 8 C to display the highest coercivity values obtained, since temperatures below C are not enough to promote full crystallization and temperatures above 8 C lead to excessive grain growth. It can be seen that prolonged annealing time leads to a continuous increase of coercivity up to 9 koe for C annealed NFe9 samples. A reduction in coercivity took place after 1 min of annealing at 8 C. There was almost no change with annealing time in coercivity for samples annealed at 7 C after 1 min. In order to understand more about the increase of coercivity with increasing annealing time, the NFe9 samples were annealed for 1 h at C. The coercivity was measured as 9.5 koe. It is confirmed that the coercivity increases with the annealing time. However, this only applies to the samples annealed at C which may be related to the fact that at relatively low temperature the grain growth was restricted. Similar 11

4 K-T Chu et al Figure 5. TEM images of NFe9 samples annealed at C (a) for 1 h in furnace annealing and (b) for 1 s in rapid thermal annealing. Figure 7. Minor loops of NFe15 samples with RTP treatment at (a) C and (b) 7 C for 1 s and (c) is the field dependence of coercivity for samples shown in (a) and (b). Figure. Minor loops of NFe9 samples with RTP treatment at (a) C and (b) 8 C for 1 s and (c) is the field dependence of coercivity for samples shown in (a) and (b). The initial magnetization curves were marked using circles. annealing time dependence of coercivity was observed for NFe15. For NFe, the highest coercivity values appeared for the annealing temperature of 7 C. The higher coercivity of NFe9 compared with those of NFe originated from the grain refinement by the addition of Co and Ti as discussed above. This has also been verified in RTP samples as shown in figure. The most striking observation in figure is that magnetic hardening could be developed for an annealing time as short as 1 s with both low and high ramp rates. There is no regular variation in the relationship between coercivity and annealing time for low ( C s 1 ) and high ramp rate ( C s 1 ). Many factors, such as ribbon thickness, composition, thermal conductivity and annealing temperature, can affect this relationship. Similar results were observed for annealing temperatures of 7 and 8 C except for the slightly lower coercivity than those for the annealing at C. This means that the crystallization can be achieved at C even for a time as short as 1 s. This is consistent with the results in RTP treated Fe-based amorphous soft magnetic ribbons, where crystallization took place for a period of several seconds and after s a significant grain coarsening was observed []. The highest coercivity (1. koe) at this point is higher than the highest value (9.5 koe) of samples annealed in a conventional furnace. The microstructure of the investigated materials was thus examined by TEM images as given in figure 5, from which it can be seen that the RTP samples have finer and more uniform grain size than the sample by furnace annealing. The calculated mean grain size was found to be around nm for rapid thermal annealing and nm for furnace annealing. The magnetic measurement verified that the finer grain size led to an enhanced exchange coupling between the hard magnetic Nd Fe 1 B and soft Fe phases and therefore an increased remanence ratio (from.75 for furnace 1

5 Rapid magnetic hardening in NdFeB-based nanocomposites Table 1. The coercivity mechanism of samples RTP treated and furnace annealed at selected temperatures for different times. (P denotes pinning mechanism, N denotes nucleation mechanism.) RTP treatment Methods Samples Temp. Low ramp rate High ramp rate Furnace annealing ( C) 1 s 7 s s 1 s 7 s s 1 min min 3 min NFe9 P P P P P P P P P 7 P N + P P 8 P N + P P NFe15 N + P N N N N N N + P N + P N + P 7 N P N 8 N N N NFe P P N + P P P P P P P 7 P P N + P 8 N + P N + P N + P annealing to.78 in RTP samples). After examining the fluctuation of the experimental curves of coercivity versus annealing temperature and time, it can be assumed that grain growth is more sensitive to temperature. Diverse approaches of increasing or reducing the coercivity of magnetic materials involve the controlling of magnetic domains within the materials. The increased difficulty for domain wall movement and nucleation of reversal domain results in an increase of coercivity. In order to investigate the coercivity mechanism, the magnetization minor loop measurements were performed and the analyses were conducted based on the measurements. Figures (a) and (b) show the minor loops of NFe9 samples after RTP treatments at and 8 C for 1 s with high ramp rate. The minor hysteresis loops also reflect the trace of initial magnetization curves. It is accepted that the steep behaviour of initial magnetization curve (large susceptibility) at low field represents nucleation mechanism, while the slow increase of magnetization for the field smaller than H c followed by a rapid increase denotes the pinning mechanism. Apparently, the type of the initial magnetization curve in figure (a) reflects a typical pinning mechanism and the curve in figure (b) is more like a nucleation mechanism. This can also be clearly verified further from figure (c), which shows the field dependence of coercivity H c on the maximum applied field H max obtained from the minor loops. In the case of the pinning mechanism, the coercivity curve has a big jump with increasing applied fields, while in the case of the nucleation mechanism the curve looks more flattened. A similar situation can be found in NFe15 where Fe 3 B acted as the soft phase. Different coercivity mechanism was observed upon RTP at C for a high ramp rate ( C s 1 ). From figure 7(a), it can be seen that the magnetization on the trace of initial magnetization curve monotonically increases, which is different from the results of 7 C (figure 7(b)). With reference to the field dependence of coercivity for C samples (figure 7(c)) where the coercivity linearly increases with the applied field strength, it can be determined that the low processing temperature led to the occurrence of the nucleation type. As the temperature increases, the coercivity mechanism was changed from nucleation to the pinning type. It should be pointed out that the coercivity mechanisms for NFe15 were observed as the nucleation type for all annealing time and temperature at low ramp rates ( C s 1 ). Table 1 summarizes the coercivity mechanisms of the samples treated with RTP and furnace annealing at selected times and temperatures. The complexity in the variation of coercivity mechanism revealed the complicated microstructure in magnetic nanocomposite system. This is related not only to the grain size, size distribution, crystallite defect, grain boundary, but also to the composition, the volume fraction of each magnetic phase etc. Generally, the pinning mechanism leads to higher coercivity. We are also studying the correlation between the energy products of the ribbons and the RTP parameters; the results will be reported elsewhere.. Summary Our studies clearly depicted the effect of RTP on the magnetic hardening in the NdFeB-based melt-spun nanocomposite ribbons. It has been revealed that the annealing time as short as 1 s promotes a full crystallization transition from the amorphous phase to magnetic nanocomposites, resulting in high coercivity. The fine tuning of heat treatment parameters (annealing time, temperature and heating rate) revealed a close correlation between the magnetic hardening and the nanostructured morphology in the nanocomposite ribbons. The high heating and cooling rates of RTP technique led to much finer grain morphology in comparison with conventional furnace annealing. The analysis of minor loop measurements reflected a complexity of coercivity mechanisms, including pinning mechanism and nucleation mechanism and their mixed state, depending on the composition, grain size and processing methods. The higher coercivity is usually connected with the pinning mechanism in the studied samples. It can be concluded that the rapid thermal annealing leads to better magnetic hardening compared with conventional furnace annealing because high heating rate and short annealing time in RTP result in better controlled nanostructures of the materials. It is believed that this technique can be successfully extended to the processing of other nanocomposites where grain refinement is required for an improved performance. 13

6 K-T Chu et al Acknowledgments This work was supported by US DoD/DARPA through ARO under grant DAAD and by DoD/MURI program under grant N We also acknowledge generous support by Dr S F Cheng at the Naval Research Laboratory for providing nanocomposite samples and Dr B Z Cui and Dr K Han for the TEM observations. References [1] Coehoorn R, de Mooij D B and DeWaard C 1989 J. Magn. Magn. Mater [] Zeng H, Li J, Liu J P, Wang Z L and Sun S H Nature [3] LiSD,GuBX,BiH,TianZJ,XieGZ,ZhuYJandDuYW J. Appl. Phys [] Skomski R and Coey JMD1993 Phys. Rev. B [5] Jin Z Q, Okumura H, Zhang Y, Wang H L, MuñozJS, Hadjipanayis G C J. Magn. Magn. Mater 8 1 [] Neu V and Schultz L 1 J. Appl. Phys [7] Liu W, Zhang Z D, Liu J P, Cui B Z, Sun X K, Zhou J and Sellmyer D J 3 J. Appl. Phys [8] Yu M, Liu Y, Liou S H and Sellmyer D J 1998 J. Appl. Phys [9] Liu J P, Liu Y, Skomski R and Sellmyer D J 1999 J. Appl. Phys [1] Zhang J, Zhang S Y, Zhang H W and Shen B G 1 J. Appl. Phys [11] Shao Y, Yan M L and Sellmyer D J 3 J. Appl. Phys [1] Zeng H, Sun S H, Sandstrom R L and Murray C B 3 J. Magn. Magn. Mater. 7 [13] Kuo C M, Kuo P C, Wu H C, Yao Y D and Lin C H 1999 J. Appl. Phys [1] Chu K T 199 MSc Thesis University of Texas, Arlington [15] You C Y, Sun X K, Liu W, Cui B Z, Zhao X G, Geng DYand Zhang Z D J. Phys. D: Appl. Phys [1] Zhang W Y, Chang H W, Chiu C H, Han JZandWangWC J. Alloys Compounds [17] Chin T S, Lin C H, Huang S H, Yau J M, ChuTYandWuCD 199 Japan. J. Appl. Phys [18] Yang C J, Park E B, Hwang Y S and Kim E C J. Magn. Magn. Mater [19] Shield J E, Kappes B B, Branagan D J and Bentley J J. Magn. Magn. Mater [] Hadjipanayis G C and Withanawasam L 1995 IEEE Trans. Magn [1] Zhang X Y, Guan Y, Zhang J W, Sprengel W, Reichle K J, Blaurock K, Reimann K and Schaefer H E Phys. Rev. B 113 [] Trudeau M L, Boily S and Schultz R 199 Mater. Sci. Forum