Exploring the Single Atom Spin State by Electron Spectroscopy

To control the spin state of an individual atom is an ultimate goal for spintronics. A single atom magnet, which may lead to a supercapacity memory device if realized, requires the high-spin state of an isolated individual atom. Here, we demonstrate the realization of well isolated transition metal (TM) atoms fixed at atomic defects sparsely dispersed in graphene. Core-level electron spectroscopy clearly reveals the high-spin state of the individual TM atoms at the divacancy or edge of the graphene layer. We also show for the first time that the spin state of single TM atoms systematically varies with the coordination of neighboring nitrogen or oxygen atoms. These structures can be thus regarded as the smallest components of spintronic devices with controlled magnetic behavior.

Figure 1

Imaging and spectroscopy of the isolated high-spin TM atom in graphene. (a) A schematic concept of high-spin TM atoms present at the edge, SV or DV, or defect with the presence of O or N dopants in graphene vacancies. The TM atom is represented by the purple ball, while carbon and dopant atoms are represented by gray and orange balls. The spin orientation is randomly displayed by arrows. (b) A typical experimental ADF image showing the isolated TM atoms in brighter contrast at a graphene layer. The distance between the adjacent TM atoms is typically 1–20 nm. (c) A diagram of $L_{23}$ fine structure edges including two dipole transitions of $2p$ electron to $3d$ empty states. The atomic magnetic moment (high spin or low spin) can be classified on the basis of the $L_3/L_2$ branching ratio. (d),(e) Experimental ADF images and atomic structures of the Cr atom at the graphene edge (Cr@edge) and the Fe atom at DV (Fe@DV), respectively. (f) The EEL spectra of single TM atoms, Cr@edge, and Fe@DV. These three single TM atoms all exhibit a high-spin state based on the analysis of the $L_3/L_2$ branching ratio. Note that the typical $L_3/L_2$ values in the literature for high and low spin states are ($2.4/1.2$) for Cr and ($4.1/3.0$) for Fe [6, 10, 11]. The error values were derived from the Lorentzian fitting errors of each single EELS spectrum assuming the major factor comes from the CCD readout noise. Scale bar, 2 Å.

Figure 2

Various spin states of TM atoms at different defect sites. (a),(b) ADF images and simulated atomic models of Fe@DV and $\mathrm{Fe}+4\mathrm{N}$. The Fe atom is represented by a blue ball while the N atom is represented by an orange ball. (c) The corresponding EEL spectra of Fe@DV (blue line) and $\mathrm{Fe}+4\mathrm{N}$ (black dashed line). The Fe $L$ edge shows a different branching ratio ($L_3/L_2$) for the Fe@DV (high-spin state) and for the $\mathrm{Fe}+4\mathrm{N}$ (low-spin state). EEL spectra are taken from these individual Fe atoms. For the case of $\mathrm{Fe}+4\mathrm{N}$, $2\times 3$ spectra are extracted from an EELS map (Fig. S5) [12] and summed up. The acquisition time is in total $6\times 0.2\sec$. (d) The schematic of the Fe atom showing the high spin with no charge transfer in DV. (e) The schematic of charge transfer from multiple pyridinic-Ns to the Fe atom. (f),(g) The ADF image and simulated atomic models of an $\mathrm{Cr}+\mathrm{O}$ and Cr metallic nanocluster. The Cr atom is represented by a red ball while the oxygen atom is represented by a green ball. (h) The corresponding EEL spectra of $\mathrm{Cr}+\mathrm{O}$ (red line) and Cr nanocluster (black dashed line). A trace of oxygen $K$ edge is also seen in the spectra from $\mathrm{Cr}+\mathrm{O}$. (i) The schematic of charge transfer from the Cr ion to the O atom. (j) The schematic of metallic bonding in the Cr nanocluster. Scale bar, 2 Å.

Figure 3

The Cr $L_{23}$ absorption edges of (a) a single Cr atom bonded to an oxygen atom, (b) a single Cr atom bonded to graphene edge, (c) a single Cr atom anchored at DV, (d)–(g) single Cr atoms bonded with 1, 2, 3, and 4 N atoms, and (h) an reference for metallic Cr from a metallic nanocluster. All the atomic configurations are shown in Fig. S6 [12].

Figure 4

The diagram of the $L_3/L_2$ branching ratio of the single Cr atom with the relation to the N coordination number. Several measurements of EELS have been performed for each measurement. The averaged branching ratio and its statistic error have been obtained from 7, 4, 5, 3, and 3 distinct EELS measurements for each atomic configuration, Cr@DV, $\mathrm{Cr}+1\mathrm{N}$, $\mathrm{Cr}+2\mathrm{N}$, $\mathrm{Cr}+3\mathrm{N}$, and $\mathrm{Cr}+4\mathrm{N}$, respectively. The branching ratio drastically decreases (21.3%–30.3%) when Cr atoms are bonded to the N atom.