Direct Evidence for the Source of Reported Magnetic Behavior in"CoTe"

In order to unambiguously identify the source of magnetism reported in recent studies of the Co-Te system, two sets of high-quality, epitaxial CoTe$_x$ films (thickness $\simeq$ 300 nm) were prepared by pulse laser deposition (PLD). X-ray diffraction (XRD) shows that all of the films are epitaxial along the [001] direction and have the hexagonal NiAs structure. There is no indication of any second phase metallic Co peaks (either $fcc$ or $hcp$) in the XRD patterns. The two sets of CoTe$_x$ films were grown on various substrates with PLD targets having Co:Te in the atomic ratio of 50:50 and 35:65. From the measured lattice parameters $c = 5.396 \AA$ for the former and $c = 5.402\AA$ for the latter, the compositions CoTe$_{1.71}$ (63.1% Te) and CoTe$_{1.76}$ (63.8% Te), respectively, are assigned to the principal phase. Although XRD shows no trace of metallic Co second phase, the magnetic measurements do show a ferromagnetic contribution for both sets of films with the saturation magnetization values for the CoTe$_{1.71}$ films being approximately four times the values for the CoTe$_{1.76}$ films. $^{59}$Co spin-echo nuclear magnetic resonance (NMR) clearly shows the existence of metallic Co inclusions in the films. The source of weak ferromagnetism reported in several recent studies is due to the presence of metallic Co, since the stoichiometric composition"CoTe"does not exist.


INTRODUCTION
Metal tellurides have been the focus of considerable research activity recently due to their unique properties and vast potential for practical applications. In particular, the Co-Te system, expressed here as CoTex, is being investigated for use as a non-precious metal electrocatalyst as well as for other specialized materials [1]. CoTex has been synthesized with a variety of nanostructured morphologies [1][2][3][4][5] and there are reports in the literature that they are magnetic semiconductors with distinctive electrical transport properties [2,4]. In spite of this effort, there are still differences of opinion concerning the nature of the magnetic behavior for CoTex. In a very early study, Uchida [6] reported that CoTex, with 1.00 ≤ x ≤ 1.20 was ferrimagnetic with the "stoichiometric CoTe" compound having a saturation magnetization of 7.52 emu/g (0.25 μB /Co) and ordering temperature of 1003 °C. Furthermore, when x = 1.20, the compound was no longer magnetic and became weakly paramagnetic with no temperature dependence. A short time later, Uchida [7] made detailed magnetic measurements on CoTex samples with 0 ≤ x ≤ 1.00, and concluded that the magnetic behavior observed earlier could be explained by assuming a eutectic mixture of metallic cobalt and the nonmagnetic compound CoTe1. 20. In a later Mössbauer study, Fano and Ortalli [8] concluded that the stoichiometric composition CoTe does not exist. However, there are still several recent reports in the literature of Co-Te nanostructured materials having the CoTe stoichiometry and being magnetic [9][10][11][12][13]. One such study reports CoTe nanowires which exhibit ferrimagnetic behavior well above room temperature with a saturation magnetization of 0.2 μB/Co [10]. Another report concerning CoTe nanotubes shows hysteresis loops with coercivity indicating ferromagnetic behavior [11].
A magnetic transition from paramagnetism to ferrimagnetism at approximately 40 K has been seen in Co-rich CoTe0.79 nanostructures [12]. Finally, a very recent study reports weak ferromagnetism in 100 nm nanorods [13]. As discussed below, an explanation of the magnetic behavior for CoTex follows from an understanding of the phase diagram and the nuclear magnetic resonance (NMR) results reported here. Figure 1 shows the equilibrium phase diagram for the binary Co-Te system based on a review of available data assembled by K. Ishida and T. Nishizawa [14].  [14,15]. As the Te content increases further, the system passes through a narrow two phase region and evolves into the orthorhombic γ(CoTe2 ) phase.
In this report, the structural and magnetic properties of two sets of CoTex epitaxial 300 nm films grown by pulse laser deposition (PLD) on different substrates are presented. The principal motivation for growing relatively thick films was to create a pathway for exploring the properties for high-quality bulk CoTex materials, not "thin" film properties. Epitaxial film growth is a way of obtaining single-crystal-like materials.

II. EXPERIMENTAL
A.

Sample Preparation
The CoTex films were grown using PLD with a target consisting of elemental cobalt and tellurium powders. The powders were mixed together and shaken vigorously.
A thickness value of ≈ 300 nm was obtained from the cross section images of a representative film using scanning electron microscopy (SEM  Fig. 2(a). For all four of the epitaxial films, (001), (002), and (004) peaks were seen and assigned to the NiAs structure which characterizes the β(Co2Te3) phase [15]. As discussed below, confirmation that these reflections can be assigned to the NiAs structure was obtained from the Oxford 2D diffractometer XRD patterns. The 2θ-values for the peak positions were essentially independent of the substrate; however, there was a small difference between the CoTe(1)-MgO and CoTe(2)-MgO films as shown in Fig. 2 the (00ℓ) peaks where ℓ is odd are forbidden. However, with the removal of some Co in the middle plane (see Fig. 1, insert), these reflections will start to appear. This can explain the relatively weak appearance of the (001) peak in the patterns shown in Fig.   2(a); however, the (003) peak is still not seen. A possible explanation is that it is too weak to be detected in the XRD patterns [17].  [18]. Figure   3(b) shows the azimuthal crystallization pattern for the CoTe(1)-Al2O3 film. The azimuthal scan was taken as the film was rotated about the c-axis. Since the NiAs structure is six-fold symmetric, the appearance of the (103) peak every 30° indicates that there are two azimuthal stacking orientations in the film. The insert shows a θ/2θ-scan with the azimuthal angle set at the peak maximum in the azimuthal scan.

C. Energy-Dispersive X-Ray Spectroscopy
In order to obtain a quantitative estimate of the elemental Te/Co content, energydispersive x-ray spectroscopy (EDXS) was carried out using a JEOL JSM-6335F field emission scanning electron microscope (SEM). erroneously enhance the Co Lα-edge spectrum [19].

D. Magnetization
Measurements of the dc magnetization were carried out for magnetic fields −50 kOe ≤ H ≤ +50 kOe over the temperature range 5.0 K ≤ T ≤ 300 K using a superconducting quantum interference device (SQUID) MPMS-5 magnetometer from Quantum Design. The hysteresis loops were obtained with the magnetic field applied in the plane (parallel to the ab-plane) of the film; the films were mounted on a quartz rod for the parallel field measurements. In order to check for spurious background contributions, corresponding magnetic measurements were also made on clean substrates. As is the case with this work, great care must be taken to avoid contamination when making measurements involving nanoscale magnetism (e.g., moment values < 10 −4 emu) [20].
Furthermore, the SQUID response curves are somewhat distorted as the samples are in the form of finite planes and not ideal dipoles [21]. Finally, magnetic artifacts can occur if the sample chamber contains even small amounts of residual oxygen [22]. These issues, all of which are important for the work reported here, are discussed in detail in [20][21][22]. These corrections were then subtracted from the curves in Fig. 5(a) to obtain the curves shown in Fig. 5(b). For comparison, consistent correction values were also obtained by directly measuring the magnetic field dependence and temperature dependence of the magnetization for the various clean substrates. In order to make a quantitative comparison between the two films, the film ordered moment obtained after the subtraction was divided by the effective volume of the film. However, since the films are two phase, "arbitrary units" are used in Fig. 5(b). From Fig. 5(b), it can be seen that the ferromagnetic component for the CoTe(1)-MgO film is approximately four times that for the CoTe(2)-MgO film. As described below, NMR clearly identifies the ferromagnetic component as metallic Co. For the CoTe(1)-MgO film, the Co ferromagnetic moment value is 0.0005 emu as seen in Fig. 5(a). Using the saturation magnetization for bulk Co (≈ 160 emu/g), an estimate of the mass density for β(Co2Te3) (≈ 8 g/cm 3  relaxation time T2 were made at selected frequencies across the spectrum by varying the pulse separation time from τ = 15 μs to 45 μs. As discussed below, the frequency dependence of T2 can result in a significant correction to the NMR spectrum [23].
Operation at liquid He temperature was carried out using a conventional glass double dewar system.
As mentioned above, the motivation for the NMR experiments was to confirm the existence and identify the structure of the Co metal phase in the CoTex films. The Co metal phase was indicated by the appearance of a ferromagnetic contribution in the SQUID magnetic measurements. A serious challenge existed in the NMR experiments due to the extremely small NMR coil filling factor. In addition, the Co metal occurs as a "trace amount" or "second phase" in the films. The observation of the 59 Co signal is only possible due to the existence of the NMR enhancement factor which occurs in ferromagnetically-and ferrimagnetically-ordered materials [24]. In these experiments, the four-turn NMR coil had a rectangular cross-section 3.0 mm × 6.0 mm and was 6.0 mm in length. Five 5.0 mm × 5.0 mm CoTe(1) films on substrates were stacked tightly inside the coil. MHz, and 225 MHz, respectively. Since T2 is relatively long compared to the rf pulse separation for spectrum acquisition (20 µs), the correction is minimal.

III. DISCUSSION AND CONCLUSIONS
Concerning previous work on the CoTex system, two issues need to be addressed. The first issue is whether or not the stoichiometric CoTe composition actually occurs; i.e.
does the system form the "perfect" NiAs structure with all of the Co sites filled in the middle planes between the Te atoms (see Fig. 1). As discussed above, it has been suggested that CoTex for 0 ≤ x ≤ 1.20 might be a eutectic mixture of metallic Co and the nonmagnetic compound CoTe1.20. In support of this picture, Uchida [7] has made detailed magnetic measurements which are compelling. On the other hand, there are reports in the literature of materials with the CoTe stoichiometry, also having the NiAs structure with somewhat smaller a and c lattice parameters than those for CoTe1.20 [3,15,18]. The second issue concerns the source of the magnetism observed in recent studies on nanostructured CoTex materials [9][10][11][12][13]. Wang et al. [1] carried out high-resolution x-ray photoelectron spectroscopy (XPS) on cobalt telluride branched nanostructured samples.
They observed weak satellite peaks in their Co 2p3/2/2p1/2 spectra which they attribute to elemental Co [1]. Wu et al. [5] observed second phase fcc Co peaks in their XRD patterns from carbon supported CoTe1.20 nanoparticles; the second phase peaks were reduced for higher heat treatment temperatures.
This report presents a structural and magnetic study of two sets of epitaxial CoTex films (x = 1.71 and 1.76) prepared by PLD using two targets with different Co:Te atomic ratios. While XRD did not see the cobalt phase in the films, 59 Co NMR clearly identifies metallic Co as the magnetic second phase (see Fig. 6). Although the existence of some amount of cobalt oxide phases(s) cannot be completely ruled out, the oxides tend to be antiferromagnetic with a completely different NMR signature. Furthermore, the calculation above with the magnetic data indicates that essentially all of the Co is accounted for in the films. Figure 7 shows a reduction in the NMR signal intensity with the application of a magnetic field. Based on a consideration of the (high) rf power level, the magnetic field behavior suggests that a significant portion of the NMR signal arises from domains, as would be the case for nanoparticles, and not domain walls, which characterize multidomain or bulk-like particles [24]. If this is the case, the Co inclusions in the films would be smaller than 76 nm, which is the approximate domain size in metallic Co [25]. This would also provide a possible explanation for the absence of metallic Co lines in the XRD scans. While both the fcc and hcp phases can exist at room temperature, the fcc structure is preferred above 450 °C with the hcp structure preferred below 450 °C [26]. However, the fcc structure has appeared for nanoscale particles at room temperature and below [26].
In order to identify crystallographic phase(s) from the 59 Co NMR spectra, an assignment of the spectral features must include both structural and magnetic considerations. Recently, a comprehensive discussion of the various spectral features for both fcc and hcp Co has been provided by Andreev et al. [27]. For reference, the NMR Co are each characterized by only a single main peak. Due to local field anisotropy, the situation for the hcp domain case is more complex [27]. However, based on experimental results for hcp nanoparticles, an assignment of 225.5 MHz is reasonable [29]. As can be seen in Fig. 6, the 59 Co NMR spectrum for the CoTe(1) films is characterized by a broad peak with apparently two structural components; the components clearly do not match peaks associated with domain walls in the fcc or hcp phases. A more likely description is that the broad peak is characteristic of a combination of fcc and hcp domain peaks. The application of an external magnetic field results in a downward shift and narrowing of the component peaks; however, the separation remains consistent with the fcc and hcp domain frequencies. It is noteworthy that the epsilon cobalt (ε-Co) phase, which has been identified in nanoparticles as well as single crystals ≤ 0.3 μm, has two distinct Co sites [26]. There is one report of 59 Co NMR in ε-Co; however, the spectra were obtained from single-domain 6.5 nm particles and, therefore, extremely broad, particularly on the low frequency side [23]. Although the ε-Co NMR spectra showed considerable intensity over the frequency range 220 MHz to 225 MHz, the association of the spectral features shown in Fig. 6 to ε-Co is not possible. In any case, 59 Co NMR clearly identifies the existence of metallic cobalt as a second phase and a source of magnetism in the CoTex films. On the other hand, XRD shows that the principal phase is hexagonal β(Co2Te3) which has the NiAs structure. This result suggests that recent reports of CoTe being a magnetic semiconductor should be revisited.      field. (220 MHz is at a relatively flat part of the spectrum and hence, the reduction in echo amplitude is not due to the shift.) The magnetic field behavior is consistent with a reduction of the NMR enhancement factor most likely within domains.