Volume and bond length dependences of the electronic structure of 6-fold and 8-fold coordinated Co2+ in pressure transformed CoF2

This work reports an optical absorption study on CoF2 under high-pressure conditions in the 0-80 GPa range. A recent structural study1 performed on the pressure-induced phase-transition sequence of CoF2 reveals that this compound undergoes a structural transformation associated with a change of Co2+ coordination from 6 (rutile type) to 8 (fluorite type) at 15 GPa, what implies a profound change in the Co2+ electronic ground state associated with the inversion of d orbitals from (t2g5eg2) to (eg4t2g3). This allows us to explore the Co2+ electronic structure, and hence 10Dq, as a function of the density and the Co-F bond length in both CoF6 (rutile phase) and CoF8 (fluorite phase) coordinations. The results are compared with those obtained in KCoF3 (perovskite) in a wide range of Co-F distances (from 2.04 to 1.83 Å).


Introduction
There is a lack of experimental studies devoted to investigate the volume/density and bond length (R) dependencies of the crystal-field d-orbital splitting e g -t 2g (10Dq) in transition metal ions (TM) occupying high-symmetry crystallographic sites. The situation is even harder if different coordination geometries are involved. Although experimental studies on 10Dq(R) have been usually performed at ambient pressure on series of TM compounds with different crystal structure and composition [1][2][3][4], rest-of-lattice and compositional effects can significantly mask the actual bond-length dependence of the TM electronic structure [5]. High-pressure spectroscopy solves out this problem since the electronic structure of a given compound can be measured as a function of pressure and eventually of V and/or R provided that its structure as a function of pressure is known.
Although pioneering works by Drickamer et al. [6] dealt with that problem in a lot of TM oxides and halides, correlations between 10Dq and V were performed on TM-doped Al 2  Cr, V) since the equation of state was known only for few materials at that time [6,7]. However impurity systems pose a scale problem of whether the TM-ligand bond length and its variation with pressure are the same for the host (Al-O) and impurity (Cr-O) as for example in ruby (Al 2 O 3 : Cr 3+ ) [4,[8][9][10]. Recent high-pressure studies by optical spectroscopy provided such correlations in pure TMion compounds: NiO [11] and BiFeO 3 [11,12]. These studies gave evidence of 10Dq(V), and thus of 10Dq(R) dependences close to V -5/3 and R -5 , respectively, given that in these compounds R(P) scales as V 1/3 (P), and therefore both dependences obey the same power law, in agreement with crystal-field theory (CFT) predictions. This result is quite surprising since CFT in the point charge approximation is very simple and does not take explicitly into account the TM-ligand bonding [13][14][15]. Although efforts to justify the R -5 law have been done on the basis of ligand field theory [16,17] and by means of LCAO-type calculations, there is a lack of experimental studies aiming to validate whether CFT predictions on 10Dq apply for 1) different TM coordinations in the same crystal, i.e. 6-fold and 8-fold cubic coordinations in CoF 2 ; 2) the same coordination in different crystals, i.e. 6-fold coordination in CoF 2 and KCoF 3 ; and 3) a coordination change, i.e. 10Dq(6-fold)/10Dq(8-fold) for the same R value. This work reports an optical absorption study on CoF 2 under high-pressure conditions in the 0-80 GPa range. A recent structural study [18] performed on the pressure-induced phase-transition sequence of CoF 2 reveals that this compound undergoes a structural transformation associated with a change of Co 2+ coordination from 6 (rutile type) to 8 (fluorite type) at 15 GPa, what implies a profound change in the Co 2+ electronic ground state associated with inversion of d orbitals splitting from (t 2g 5 e g 2 ) to (e g 4 t 2g 3 ). This different ground state electronic configuration allows us to explore clearly the Co 2+ electronic structure, and hence 10Dq, as a function of the density and R Co-F in the two coordinations: CoF 6 (rutile phase) and CoF 8 (fluorite phase), taking the structural data reported elsewhere [18]. The results will be compared with those obtained in KCoF 3 where the 6-fold coordination of Co 2+ in the perovskite structure is stable up to 80 GPa, thus providing the widest 10Dq(R) range ever measured in TM ions.

Crystal structure and x-ray diffraction
Single crystals of CoF 2 and KCoF 3 grown by the Bridgman method according to methods described elsewhere [18]. At ambient conditions CoF 2 crystallizes in the tetragonal space group P4 2 /mnm (rutile phase) [18] and KCoF 3 in the cubic Pm3m (perovskite phase) [19,20]. The evolution of the crystal structure of KCoF 3 with pressure was studied by angle dispersive x-ray diffraction (AXRD) on powdered samples using the 12.2.2 beamline at the Advance Light Source (ALS). Pressure was applied by means of a symmetric piston cylinder Diamond Anvil Cell (DAC). High pressure experiments on polycrystalline CoF 2 were performed in the Materials Science and Powder Diffraction beamline (BL04) at ALBA synchrotron using a Boehler-Almax DAC. In all experiments samples were loaded with several Ruby spheres (10 μm diameter) as pressure gauge [21] using methanol-ethanol-water and silicone oil as pressure transmitting media.

Optical Absorption and Raman Spectroscopy
Optical absorption and Raman experiments were performed on single-crystal plates of CoF 2 (90x80x14 μm 3 ) and KCoF 3 (100x80x20 μm 3 ) were used for high-pressure optical absorption experiments. The optical spectroscopy experiments were carried out in membrane and Almax-Boehler DACs. 200-μm-thick Inconel gaskets were preindented to 40 μm. 200-μm-diameter holes were perforated with a BETSA motorized electrical discharge machine. The DAC was loaded with a single crystal plate of CoF 2 or KCoF 3 and ruby microspheres (10 μm diameter) as pressure probes [21] using silicon oil as pressure-transmitting medium.
Optical absorption under high-pressure conditions was performed on a prototype fiber-optics microscope equipped with two 25× reflecting objectives mounted on two independent x -y -z translational stages for the microfocus beam and the collector objective, and a third independent x -y -z translational stage for the DAC holder. Optical absorption data and images were obtained simultaneously with the same device [22]. Spectra in the UV-VIS and NIR were recorded with Ocean Optics USB 2000 and NIRQUEST 512 monochromators using Si-and InGaAs-CCD detectors, respectively. Unpolarized micro-Raman scattering measurements were performed with a triple monochromator Horiba-Jobin-Yvon T64000 spectrometer in subtractive mode backscattering configuration, equipped with an Horiba Symphony liquid-nitrogen-cooled CCD detector. The 514.5-nm and 647-nm lines of an Coherent Innova 70 Ar + -Kr + laser were focused on the sample with a 20× objective for micro-Raman, and the laser power was kept below 4 mW in order to avoid heating effects. The laser spot was 20 μm in diameter and the spectral resolution was better than 1 cm −1 . The Raman technique was used in combination with X-ray diffraction to check the sample structure through the characteristic first-order modes in CoF 2 [18] and absence of first-order modes in KCoF 3 as well as to determine structural phase-transition pressures. The Raman high-pressure experiments were performed on the same CoF 2 and KCoF 3 single crystals employed in the optical absorption measurements.  Figure 1 shows the electronic absorption spectrum of CoF 2 and KCoF 3 at ambient pressure, as well as the Tanabe-Sugano and corresponding B and 10Dq parameters, which account for the Co 2+ transition energies. Both compounds exhibit a similar spectrum as it mainly corresponds to the CoF 6 coordination unit with similar Co-F bond distances: R = 2.042 and 2.035 Å [23,24]; and associated CF parameters: B = 0.107 and 0.105 eV; and 10Dq = 0.962 and 0.950 eV, for CoF 2 and KCoF 3 , respectively. Interestingly, the similarity of spectra strengthen the relevance of CoF 6 to account for the CF electronic structure of Co 2+ beyond other ionic shells of the crystal, the influence of which seems to be much weaker than the first Fshell in spite of the distinct crystal structure: rutile (CoF 2 ) and perovskite (KCoF 3 ). However, the slightly bigger 10Dq of CoF 2 with respect to KCoF 3 contrasts with the shorter Co-F distance in KCoF 3 must be ascribed to rest-of-the-lattice CF effects beyond CoF 6 , which is estimated to be less than 4% of 10Dq.

Figure 2.
Pressure dependence of the optical absorption spectrum of CoF 2 . Spectra correspond to rutile-type structure (blue), orthorhombic (green) with 6-fold coordinated Co 2+ , and to fluorite-type phase (red) with 8-fold coordinated Co 2+ [18]. The bands correspond to crystal-field transitions from the electronic ground state 4 T 1 (F) in CoF 6 and 4 A 2 (F) in CoF 8 to different excited states Γ i according to the assignment given in the corresponding Tanabe-Sugano diagrams for d 7 and d 3 , respectively. The energy of the Γ i state is represented as a function of the crystal-field splitting and are both given in units of the Racah parameter B [24]. Colour representing Γ i state has been maintained in both figures. The two sets of experimental points in the Tanabe-Sugano diagram denote the low pressure and high-pressure limits in both 6-fold and 8-fold coordinations. The relative crystal-field, Δ10Dq / 10Dq (or δΔ/Δ), and corresponding δR/R variations are given right side.
The variation of the absorption spectra with pressure for the two compounds is shown in figures 2 and 3. The extreme pressure points are represented in the Tanabe-Sugano diagram to illustrate the 10Dq/B range swept by pressure in each coordination: CoF 6 (d 7 -like) or CoF 8 (d 3 -like). The variations of 10Dq and R in the corresponding pressure range are also indicated. Interestingly, the CoF 6 (rutile) to CoF 8 (fluorite) change of coordination taking place in CoF 2 at 15 GPa induces a drastic change in the absorption spectrum which is associated with the corresponding change of electronic configuration (from d 7 -like to d 3 -like) affecting the ground and excited states, and hence the absorption spectra. However such an abrupt change is not observed in KCoF 3 since the CoF 6 octahedral coordination is stable in the 0 -80 GPa range as indicated by AXRD experiments whose V(P) data are represented in figure 3 (right). Consistently, the absorption spectrum of KCoF 3 in the whole pressure range is concomitant with an electronic configuration d 7 and thus the experiment provides the widest 10Dq(R) range explored ever in Co 2+ . In general, pressure-induced phase transitions in both CoF 2 and KCoF 3 affect the electronic spectra as evidenced through their 10Dq(P) behaviour ( figure 4). The different slope of 10Dq(P) exhibited by the perovskite KCoF 3 unravels the effect of a cubic (Pm-3m) to orthorhombic (Pnma) second-order phase transition at 24 GPa. Such a phase transition, which is associated with rotations of the CoF 6 octahedra, keeps the 6-fold octahedral coordination of Co 2+ in the high-pressure orthorhombic phase, and thus 10Dq varies continuously at the phase transition, although its pressure dependence is  different in each phase. This pressure behaviour contrasts with that exhibited by CoF 2 , whose 10Dq(P) values show a peculiar pressure dependence through which the corresponding 10Dq(R) have been derived using the available XRD and XAS structural data reported elsewhere [18]. In CoF 2 pressure induces an increase of 10Dq from 0.96 to 1.06 eV in the 0-6 GPa (rutile phase), whereas at 15 GPa, it abruptly decreases by -0.3 eV along with the 6-to 8-fold coordination change. Above 15 GPa, 10Dq increases at a rate of 9 meV GPa -1 in the fluorite phase, and at 4.4 meV GPa -1 beyond 44 GPa in the cotunnite phase [18].

Concluding remarks
The knowledge of EOS and R(P) in the two investigated compounds from XRD and XAS ( Figure 3) [18,20] allows us to derive 10Dq(R) from the corresponding 10Dq(P). A noteworthy result concerns the regular variation of 10Dq(R) in CoF 6 and CoF 8 with respect to variations of 10Dq(P) that show anomalies at the phase transitions in both CoF 2 and KCoF 3 . Such anomalies are concurrent with similar anomalies in R(P) thus highlighting the suitability of R (and neither V nor density) to properly describe pressure-induced variations of 10Dq. Furthermore 10Dq(R) obeys a potential law as R -n with n close to 5 in both compounds. In particular, we get an exponent n = 5.1±0.3 in KCoF 3 in an R range of 12% whereas in CoF 2 n = 5.0±0.1 (rutile phase) and n = 6.0±0.2 (fluorite phase) as deduced from their relative changes of R and 10Dq in the explored R-ranges (figures 2 and 3). The reliability of CFT to fairly account for R-dependence of 10Dq is also confirmed through the relative 10Dq values obtained for CoF 6 and CoF 8 in CoF 2 at the same Co-F distance, R = 2.04 Å, as 10Dq(oct)/10Dq(cub) = 1.10. Its similitude with the theoretical value of 9/8 supports CFT predictions to accurately describe experimental 10Dq(R).
Finally, this study constitutes a reference system to validate ab initio methods aiming to calculate electronic structure (including excited states) of TM ions besides of well-suited DFT methods to calculate crystallographic and vibrational structures.