Temperature driven structural-memory-effects in carbon nanotubes filled with Fe3C nano crystals

We report the observation of novel temperature-driven structural-memory-effects in carbon nanotubes (CNTs) filled with Fe3C nano-crystals. These structural-transitions were measured by means of temperature (T) dependent x-ray diffraction (XRD) in the T-range from 298 K to 12 K. A clear reversible 2θ-shift in the 002-peak of the graphitic-CNTs-walls is found with the decrease of the temperature. As determined by Rietveld refinement, such 2θ-shift translates in a not previously reported decrease in the value of the CNT graphitic c-axis with the decrease of the temperature (from 298 K to 12 K). Also, a clear reversible 2θ-shift in the 031 and 131 diffraction-peaks of Fe3C is observed within the same T-range. Rietveld refinements confirm the existence of such memory-effect and also reveal a gradual decrease of the 010-axis of Fe3C with the decrease of the temperature. These observations imply that the observed structural-memory-effect is a characteristic of CNTs when Fe3C is the encapsulated ferromagnet. The generality of such memory-effects was further confirmed by additional measurements performed on other types of CNTs characterized by continuous Fe3C-filling. XRD measurements in the T-range from 298 K to 673 K revealed also an unusual reversible decrease of the Fe3C-peak intensities with the increase of the temperature. These observations can have important implications on the magnetic data recording applications of these nanostructures by helping in better understanding the unusual temperature-dependent magnetic instabilities of iron-based nano-crystals which have been recently reported in literature.

These nanostructures are generally grown in the form of vertically aligned films by chemical vapour deposition (CVD) methods involving the use of single or mixed organometallic compounds as synthesis precursors. The obtained CNTs-structures generally exhibit partial filling rates [12][13][14][15][16][17]. In the attempt to Original content from this work may be used under the terms of the Creative Commons Attribution 3.0 licence.
Any further distribution of this work must maintain attribution to the author(s) and the title of the work, journal citation and DOI. control and tune the nano-crystal filling-rate and therefore the magnetization characteristics, the addition of (1) Cl-containing precursors [19][20][21]23] or (2) local-perturbations has been proposed [18]. It has been shown that the metallocene-pyrolysis leads to the formation of numerous molecular species which include: Fe+H 2 +CH 4 +C 5 H 6 +K [17]. The CNTs obtained from these syntheses methods have been reported to encapsulate mixed-or single-iron-based phases. Specifically, the presence of mixed Fe 3 C, α-Fe and γ-Fe phases has been reported for experiments involving the pyrolysis of the only ferrocene precursor [12][13][14][15][16][17]. Instead, when additional Cl-containing precursors are added to ferrocene, the presence of large quantities of Fe 3 C has been shown [19][20][21]23]. In the attempt to enhance the magnetization properties of these structures, the use of annealing methods has been considered for the conversion of γ-Fe into α-Fe [24][25][26]. Particularly, it has been shown that annealing of Fe-filled CNTs in vacuum can allow an efficient conversion of γ-Fe into α-Fe at relatively low annealing-temperatures [27].
However, despite the large number of reports on this topic, very little is known about the temperaturedependent structural arrangement of these structures. Indeed previous literature works have mainly focused their attention on the room-temperature structural properties, without analysing the effect of cooling or heating on the structure of these important materials. In this paper we address this missing point and we report the observation of not previously reported temperature-driven structural-memory-effects in CNTs filled with Fe 3 C nano-crystals.
Such transitions were measured by the means of temperature dependent x-ray diffraction in the temperature range from 298 K to 12 K. A clear reversible 2θ-shift in the 002 peak of the graphitic CNTs walls is found with the decrease of the temperature. As determined by Rietveld refinement methods, such 2θ-shift translates in a not previously reported decrease in the value of the CNT graphitic c-axis with the decrease of the temperature from 298 K to 12 K. Additionally, a clear reversible 2θ-shift in the 031 and 131 peaks of the encapsulated Fe 3 C crystals within the XRD patterns is found in the same temperature range. Such memory effect in the encapsulated Fe 3 C is further confirmed by Rietveld refinements analyses, which show a clear decrease in the value of the encapsulated Fe 3 C-010 crystal-axis with the decrease of the temperature. These observations clearly prove the existence of a not previously observed temperature-dependent structural memory effect in both the encapsulated Fe 3 C crystals and the CNTs. The generality of such memory-effects was further confirmed by additional measurements performed on other types of CNTs produced with different synthesis methods and characterized by a continuous Fe 3 C filling. These observations imply that the observed structural transition (structuralmemory-effect) is a characteristic of both partially-filled and continuously filled CNTs when Fe 3 C is the encapsulated ferromagnet.

Experimental
The synthesis experiments were carried out by using a CVD system composed of a quartz tube of length 1.5 m, one zone electric furnace and an Ar flow rate of 10-12 ml min −1 . The reactor dimensions used for the production of Fe 3 C-filled CNTs were as follows: for the partially-filled CNTs a quartz tube with an inner diameter of 44 mm and a wall thickness of 3 mm was used. The temperature of pyrolysis was set to that of 990°C. The precursors (approximately 1 g of ferrocene and 0.15-0.65 ml of dichlorobenzene) were evaporated with a preheater at a temperature of approximately 200°C and 70°C. Instead, for the continuously-filled CNTs, a different type of CVD reactor consisting of a quartz tube of 22 mm outer diameter, wall thickness of 2.5 mm and length of 1.5 m was used together with lower quantities of ferrocene and dichlorobenzene (60 mg of ferrocene and 0.05 ml of dichlorobenzene). In both cases the samples were cooled down until the temperature of 25°C by removing the furnace along a rail-system.
Variable low-temperature XRD measurements were performed with a Panalytical Empyrean powder x-ray diffractometer (Cu K-α, λ=0.154 nm) equipped with a primary Johansson monochromator, an Oxford Cryosystems PhoeniX cryostat operating under vacuum below 10 −2 Pa, and a X'celerator linear detector. All the high temperature experiments were performed with a Rigaku Smartlab powder x-ray diffractometer (Cu K-α, λ=0.154 nm) under vacuum values below 7 Pa in the temperature range from 139 K to 673 K. A 200 kV American FEI Tecnai G2F20 was employed to obtain transmission electron microscopy (TEM) and high resolution TEM (HRTEM) images. XPS measurements were performed with a Escalab 250Xi. Magnetic measurements were performed with a Quantum Design MPMS XL-7.

Results and discussion
The morphological properties of the as grown filled CNTs were first revealed by TEM analyses performed in transmission and scanning TEM mode. Typical examples of the as grown CNTs with the two synthesis approaches described above are shown in figures supp. 1, 2,is available online at stacks.iop.org/MRX/5/025010/mmedia while typical HRTEM images are shown in figure 1. The observed lattice spacings characterized by a preferred direction with respect to the CNT-axis (see figure 1) were associated to the 010 and 100 reflections of Fe 3 C with space group Pnma (see also figure supp. 13 for XPS analyses).
The evaluation of the temperature-dependent (T-dependent) structural properties of the as grown CNTs was carried out by XRD analysis under vacuum at variable temperatures. As a first step, the attention was focused on the structural properties of CNTs partially filled with Fe 3 C in the temperature range from 298 K down to 12 K. The presence of a multi-walled arrangement of the CNTs-walls was revealed by the XRD patterns in the 2θ range from 23 to 30 degrees (see figure 2). A clear peak at approximately 26 degrees 2θ was found and associated to the 002 reflection, which identifies the graphitic arrangement of the CNT-walls. Also, as shown in figure supp. 3, XRD measurements in the 2θ range from 35 to 55 degrees revealed the presence of a large quantity of Fe 3 C crystals with space group Pnma (as confirmed by Rietveld refinement in figure supp. 3) and minor quantities of γ-Fe with space group Fm-3m and α-Fe with space group Im-3m phases. Interestingly, as shown in figure 2, the T-dependent measurements revealed the presence of a shift in the position of the 002 peak toward higher values of 2θ degrees with the decrease of the temperature. Such shift could be associated to a decrease in the value of the c-axis of the graphitic-unit cell of the CNT with the decrease of the temperature.
Such reversible transition can be observed in figure 3(A), where the variation of the graphitic-CNT-unit-cell c-axis is plotted against the temperature. Curiously, as shown in figure 3(B) a decrease in the value of the 010 In addition, Rietveld refinement of the data (shown in figure supp. 3) provides evidence that the observed shift in the 131 peak-position may be associated to a slight increase of the 100 and 001 axis-values of Fe 3 C with the decrease of the temperature (note however that such slight change could not be well quantified due to the accuracy limit of the instrument, as shown in figures supp. 5, 6). These observations imply that Fe 3 C-filled CNTs are characterized by unusual structural-memory characteristics as confirmed by the observations in figure 4(B).
In the attempt to verify the generality of this property, further measurements were carried out on a different type of CNTs filled with Fe 3 C (CNTs filled with continuous Fe 3 C crystals) produced with different synthesis conditions (60 mg of ferrocene, 0.05 ml of dichlorobenzene, see experimental section). As shown in figures supp. 7, 8, also in this case, a similar trend is found. A shift in the position of the 002 graphitic-peak toward higher values of 2θ degrees is found with the decrease of the temperature. Furthermore, a similar shift in the 031 peak position is observed also in this type of CNTs.
These findings suggest that the observed structural-memory-effect is a general characteristic of multiwall CNTs filled with Fe 3 C crystals (independent from the used synthesis method, see also figures supp. 14, 15 for additional measurements performed in a third type of Fe 3 C-filled CNTs). It is important to mention that previous literature works have shown that the magnetic moment of Fe 3 C can be strongly influenced by the values of the crystal unit cell axes [30][31][32][33][34]. Therefore our observations may have important implications from a magnetic perspective, by helping in better understanding the unusual magnetic properties of iron-based crystals which have been recently reported in literature by zero field cooled (ZFC) and field cooled (FC) studies of the magnetization [30][31][32][33][34]. Particularly it is interesting to notice that Karmakar et al [33] have reported an unusual variation of the magnetization with temperature and applied field in CNTs films containing large quantities of α-Fe and Fe 3 C and very small quantities of γ-Fe (below 1%). These observations are in agreement with our magnetization analyses (shown in figures supp. 16 and 17) where ZFC measurements exhibited a similar unusual variation in the magnetization values with temperature and applied field. Therefore a possible influence of the structural-memory effect reported in this work on such unusual variation of the magnetization characteristics cannot be excluded. Future neutron diffraction experiments may be helpful to fully elucidate the origin of such magnetic effects.
In the attempt to further investigate the T-dependent structural properties of Fe 3 C filled CNTs, the attention was then turned on possible high temperature structural transitions in the range from approximately 298 K to 673 K. The result of the XRD measurements performed in such temperature range is shown in figures supp. 9-11 and in figure 5. Interestingly a reversible shift of the 002 peak-position toward lower values of 2θ degrees   was found with the increase of the temperature. Rietveld refinement analyses confirmed (see figure supp. 9), that an increase in the value of the graphitic c-axis of the CNT-unit cell (graphitic-unit-cell) is present with the increase of the temperature (figure supp. 10).
This last observation is in agreement with that recently reported in the case of annealing-experiments of CNTs-buckypapers filled with α-Fe/γ-Fe crystals at high temperatures by T-dependent XRD in vacuum [27]. However, the observed shift is different with respect to that reported by F Y Wu et al [28] in the case of hollow multiwall CNTs, where significant shifts (in the same high temperature range from 298 K to 673 K) in the 002 peak were observed only after high temperature treatment of as grown CNTs (due to the improved structural arrangement) and not in the as grown samples [28]. A complementary interpretation was also given by the work of Y Maniwa et al [29], where the presence of disorder and defects in the graphitic layers of carbon nanotubes was classified in different CNTs structural-arrangements, namely: (1) Russian Doll, (2) Jelly Roll, (3) mixed Russian Doll and Jelly Roll and (4) polyhedral graphite with defects at the ridges [29]. In the specific case of our measurements, it is important to notice that at high temperatures no significant changes in the values of the 031 Fe 3 C peak position are found (see supp. figure 11). Such difference with respect to the low-temperature characteristics could be associated to the increase of the thermal agitation which may also affect the structural arrangement of the Fe 3 C crystals. Indeed an unusual reversible change in the observed peak-intensities and shape is found with the increase of the temperature and can possibly be associated to such increase in the thermal agitation. Note that the observed diffraction peak-position and intensities were found to come-back to the original position after cooling-down the samples back to room temperature, as shown in figure supp. 12. In this context, future works with additional techniques will be considered to extract deeper information on the origin of such unusual reversible peak-intensities variations.

Conclusion
In conclusion we reported the observation of temperature-dependent structural-memory-effects in CNTs filled with Fe 3 C nano-crystals. These transitions were measured by the means of temperature dependent XRD in the temperature range from 298 K to 12 K.
A reversible decrease in the values of (1) the graphitic CNTs-c-axis and (2) the 010 Fe 3 C crystal-axis (with the decrease of the temperature) was extracted by using Rietveld refinement methods. The presence and the generality of such memory-effects were further confirmed by additional XRD measurements performed on other types of CNTs characterized by continuous Fe 3 C filling. In addition XRD measurements were also performed at high temperatures in the range from 298 K to 673 K where an unusual reversible decrease of Fe 3 C peak intensities was found with the increase of the temperature.