Neutron Diffraction Analysis

A third diffraction technique that has a good chance of being able to observe the presence or absence of the reflections associated with chemical ordering, is neutron powder diffraction. The scattering lengths for Bi, Fe, Cr and O are now dissimilar enough that weak features associated with the ordering of the B-site can be distinguished without domination from the Bi signal or the background. Neutrons can interact with magnetic spins giving rise to magnetic scattering from electrons in addition to nuclear scattering. This method requires a large amount of sample which has not yet been achieved in other literature reports, but due to the new synthesis method detailed above, the nature of the chemical ordering and magnetic ordering (if any) in bulk BFCO samples can be determined.

Figure 3.8 shows a preliminary dataset collected at room temperature using the Echidna diffractometer at the Australian Centre for Neutron Scattering, showing several important features. Firstly, with the change in scattering factors an enhanced signal can now be seen from the impurity phases which were not immediately obvious in the XRPD data, and they are now positively identified as Cr2O3 and Bi metal. This does mean that the main phase of BFCO in this sample more than likely deviated from the nominal stoichiometry and is in fact iron rich. Secondly, and most importantly, a peak can be seen at d = ~ 4.6 Å which is where a peak corresponding to chemical ordering would be expected (and cannot be matched to other additional phases). However, this is also the same location one would expect to see a peak from G-type antiferromagnetic ordering which is reported for in the majority of BFO related compounds and is also predicted in the initial computational studies43, 58_.

Figure 3.8. A neutron powder diffraction pattern of BFCO collected with the ‘Echidna’ diffractometer. A basic profile fitting (no magnetism, via Jana2006) is shown in red, for phases BFCO (R3c, D-phase), Bi (R-3m) and Cr2O3 (R-3c). An extra peak is noted at d = ~ 4.6 Å.

G-type magnetic ordering is described by (111)P planes of B-site ions which are ferromagnetically coupled within the plane (have spins pointing in the same directions) but antiferromagnetically coupled between planes (neighbouring planes along [111]p will have antiparallel spin alignment). A basic pictorial representation of this type of magnetic ordering for both the O-phase and D-phase variants is shown back in Figure 3.5. This means, scattering from the magnetic structure will also produce a diffraction peak of the type <½ ½ ½>P* which describes the magnetic propagation vector. It is important to note that both the chemically ordered (O-phase) and chemically disordered (D-phase) unit cells can potentially exhibit antiferromagnetic coupling like this, but the net magnetization will have different values. The existence of this extra peak in the neutron pattern gives three possible options here: (1) the structure is D-phase with antiferromagnetic ordering only, (2) the structure is O-phase with no magnetic ordering or (3) the structure is O-phase and also possesses antiferromagnetic ordering. This experiment was repeated by collecting further neutron diffraction data on the Polaris neutron spectrometer at the ISIS neutron and muon source in the UK. When the data were refined considering Model (1) the additional peak could be described as being contributed to be B-site ions with a moment of ~ 0.21 μB and an overall Rwp factor of 1.67%. This moment is below what that expect from the Hund’s rule predictions, but is not outside the realm of possibility for a highly frustrated system. The refinement result using this model is shown in Figure 3.9 (a). When applying Model (2) to the data, a fit with a comparable R factor, also 1.67% can be obtained (shown in Figure 3.9 (b)), which does not allow separation of these two options based on statistics. Refining both the chemical and magnetic ordering as is needed for Model (3) was not meaningful due to the resolution of the detector bank in this region, and attempts to try this resulted in an overestimation of the peak which does suggest that this model is unlikely. Given the comparable statistics for Model (1) and (2), a further experiment was required.

Figure 3.9. In (a) a neutron powder diffraction pattern obtained from the Polaris instrument is shown (room temperature) refined in GSAS using Model 1 (chemically disordered, magnetically ordered) and (b) shows the same data refined with Model 2 (chemically ordered, no magnetism). The fit is in black, the raw data is red and the difference plot is blue. Figure adapted from ref. [328_{] with permission.}

3.2.3.2 High Temperature Study

Foreseeably if this low angle neutron diffraction peak is contributed to in any way by magnetic ordering, then heating the sample above its TN, should remove that contribution. From magnetization experiments (to be presented later in 3.3.1.1) it was suspected that the TN temperature was around 400 K. Neutron diffraction data collected at 500 K presented in Figure 3.10 show that the peak at d = ~ 4.6 Å has vanished, leaving no signal behind that could correspond to chemical ordering eliminating model (2) and (3) as options. This quite strongly suggests that the material has the D-phase structure and possesses antiferromagnetic order at room temperature.

Figure 3.10. Neutron powder diffraction data obtained from the Polaris instrument at 500 K, refined with chemical disorder and no magnetic ordering (having passed through TN), Rwp = 1.74 %. An absence of underlying chemical ordering is indicated. Note: Bi peaks have disappeared, and having not recrystallized after the heating cycle. Figure adapted from ref. [328_{] with permission.}

3.2.4 Structural Simulations