Powder and single-crystal neutron-scattering experiments found that the magnetic ordering pat- tern in Na2IrO3 developing below TN = 15 K is of zig-zag type [41, 42]. Neutron-scattering
experiments on Na2IrO3are, however, a challenge because of the high absorption cross-section
of iridium and the limited size of available single crystals. In collaboration with Dmytro Inosov of TU Dresden, a low-temperature neutron diffraction experiment was performed on a stack of several co-aligned epitaxial Na2IrO3 thin films deposited on 1.5” sapphire wafers. For typical
6.2 Magnetic order - neutron diffraction on 1.5” samples
Figure 6.2 A typical Na2IrO3film deposited on a 1.5” a-sapphire substrate at 0.016 mbar, 550 °C (450 W) and
with 50,000 laser pulses. A stack of seven equivalent samples was used for neutron diffraction (cf. Sec. 6.2). (a) XRD 2θ -ω-scan. (b) An XRD φ -scan illustrates the [010]Na2IrO3k[00.1]Al2O3in-plane epitaxial relationship
on a-plane sapphire. Peak widths are about 10°. For comparison, a film prepared with intermediate c-ZnO layer (gray) has an average half-width of only 2° (cf. Sec. 6.3). (c) XRD ω-scan of the (131) structural Bragg reflex. Gauss fitting gives a peak half-width of about 7°. [Samples: E3884, E4029.]
structural properties, obtained by means of XRD, see Fig. 6.2. The benefits of such a sam- ple design are a large surface-to-volume-ratio and an equivalent volume comparable to that of single-crystals. The aim of the experiment was to determine the magnetic ordering wave vector, which previously was determined to be q = (0,1,0.5) [41, 42]. Additionally, this experiment represented a feasibility test for a future systematic investigations of the (Na1-xLix)2IrO3com-
positional series. Within this series, magnetic order could be suppressed down to 1.2 K at an intermediate value of x ≈ 0.7 [77].
For the experiment, seven (001)-oriented Na2IrO3films were deposited on 1.5” a-plane sapphire
wafers with an in-plane epitaxial relationship of [010]Na2IrO3k[00.1]Al2O3 (cf. Sec. 6.1). The
peak widths of about 10° observed in XRD φ -scans of Na2IrO3(131) suggest a relatively large
in-plane mosaic spread, see Fig. 6.2. In direct comparison, films deposited on a c-ZnO buffer layer had significantly lower peak widths of about 2°. For protection and to prevent degradation of Na2IrO3, all films were passivated with a 110-nm layer of SiNxusing PECVD. Subsequently,
Figure 6.3 Illustration of the sample holder employed for mounting the Na2IrO3 film stack within the 49 mm
cryostat bore. The stack is glued onto a c-shaped piece of 0.5 mm-thick aluminum that can be fixed and tilted at an arbitrary angle. [Drawing by Pavlo Portnichenko, TU Dresden]
the samples were stacked and glued together on an aluminum holder, of which a schematic is shown in Fig. 6.3. Within the stack, the samples were co-aligned with respect to the wafer’s primary flat indicating the [00.1] direction of sapphire. With 50,000 laser pulses, the individual film thickness, determined by spectroscopic ellipsometry, is approximately 500 (± 50) nm. Thus, the stack of seven 1.5” wafers yields a total Na2IrO3 volume of approximately 4 mm3
which is comparable to some of the larger Na2IrO3single crystals [41].
Neutron diffraction was performed with the thermal neutron two-axis diffractometer D23 at ILL (Sec. 4.2) and a wavelength of 2.38 Å. As a starting point, the structural C2/m unit cell of Na2IrO3 refined by Ye et al. [41] was used for the calculation of sample alignment and
scattering geometry. Initially, measurements were intended to be performed within the (h, 2k + h, k − h) scattering plane that is spanned by the (0, 2, 1) and (1, 1, −1) allowed structural Bragg reflexes used for sample alignment1. Within (h, 2k + h, k − h), the (0, 1, 0.5) and (0.5, 0.5, −0.5) magnetic Bragg reflexes are nearly orthogonal and indicative of antiferromagnetic zig-zag order in Na2IrO3. Due to the monoclinic crystal structure, (0, 1, 0.5) and (0.5, 0.5, −0.5) are not
equivalent and only the existence of the former reflex has been reported so far in measurements within the (0, k, l) scattering plane [41]. Unfortunately, measurements within (h, 2k + h, k − h)
6.2 Magnetic order - neutron diffraction on 1.5” samples
were unsuccessful because neither the (0, 2, 1) and (1, 1, −1) structural reflexes nor any of the magnetic reflexes were found at or near their expected q-values.
Hence, the sample was reoriented inside the cryostat to allow measurements within the (0, k, l) scattering plane, with (0, 1, 0.5) again being the magnetic Bragg reflex of interest. For sample alignment in this plane, only the Na2IrO3 (0, 0, ±1) structural Bragg reflexes could be clearly
measured, as shown in Fig. 6.4(a). In addition, multiple reflexes were measured close to the expected q-value of (0, ±6, 0) and two of these are displayed in Fig. 6.4(b). Due to the 10° in-plane mosaicity, the multiple (0, ±6, 0) reflexes likely stem from different grains within the sample, preventing a precise sample alignment within (0, k, l). Gaussian fits of (0, 0, ±1) and (0, ±6, 0) yield lattice parameters c = 5.492(9) Å and b = 9.45(2) Å, respectively2. The c lattice parameter is reliable and differs from the assumed unit cell dimensions by only -0.8 %. It coincides well with previous XRD measurements on the individual films. Note, that the b lattice parameter is unreliable due to the difficulties in sample alignment and given for completion’s sake, only.
A subsequent measurement of (0, 1, 0.5) within (0, k, l) at 1.5 K, i.e. within the magnetically ordered phase, did again not reveal any measurable signal. A repeat measurement at 20 K, i.e., above the Neél temperature, was performed for the purpose of temperature subtraction. However, no magnetic reflex was uncovered by this indirect method, as shown in Fig. 6.5(a).
Figure 6.4 γ -scans of structural Bragg reflexes of Na2IrO3. (a) (0, 0, ±1) and (b) (0, ±6, 0) measured at 8 and
1.5 K, respectively. All data were fitted by Gaussians (dashed lines) to determine the c and b lattice parameters.
An unexpected finding related to the in-plane epitaxial relationship is that the q-vectors belong-
2Calculated via 1 d0kl2 = k2 b2+ l 2 c2sin2 β, where dhkl= λ 2 sin(γ/2)and β = 108.67°.
ing to the Al2O3(0, 1, 2) and Na2IrO3(0, 2, 1) structural Bragg reflexes are very similar in both
size and direction, as shown in Fig6.5(b). This was not considered prior to the experiment but may have had direct consequences for measurements of (0, 1, 0.5) at only half these q-vectors. Since the experiment was performed using only one monochromator, it is possible that the second order (0, 1, 2) substrate reflex was still very intense and masked the nearby (0, 1, 0.5) magnetic reflex.
Figure 6.5 (a) Top: Measurement of the (0, 1, 0.5) magnetic Bragg reflex expected at γ = 19.80° (dashed line) at 1.5 and 20 K ,i.e., below and above the Néel temperature of 15 K. Bottom: Temperature subtraction gives no indication of a magnetic reflex. (b) Rocking curves of Al2O3(0, 1, 2) and Na2IrO3(0, 2, 1) structural Bragg
reflexes measured at 8 K and at angles ν and γ as indicated. Their corresponding q-vectors are similar in size and direction.
In summary, magnetic Bragg peaks were not measurable directly or indirectly using temperature subtraction in a stack of seven Na2IrO3thin films. Sample alignment on both the (h, 2k + h, k −
h) and (0, k, l) planes was attempted. While some structural Bragg reflexes of both film and substrate were observed within these scattering planes, the (0, 1, 0.5) magnetic Bragg reflex was not. The lack of signal at the expected magnetic q-vectors is predominantly due to the significant in-plane mosaicity present in the individual films and, in addition, prevented a precise sample alignment. For future attempts, this obstacle has to be tackled primarily. As is shown in Sec. 6.3, deposition of an intermediate buffer layer on either a- or c-plane sapphire leads to a significant and reproducible reduction of in-plane mosaicity, and might justify a renewed neutron-diffraction experiment on Na2IrO3thin films.