Direct MoB MBene domain formation in magnetron sputtered MoAlB thin ﬁ lms †

Two-dimensional (2D) inorganic transition metal boride nanosheets are emerging as promising post-graphene materials in energy research due to their unique properties. State-of-the-art processing strategies are based on chemical etching of bulk material synthesized via solid-state reaction at temperatures above 1000 °C. Here, we report the direct formation of MoB MBene domains in a MoAlB thin ﬁ lm by Al deintercalation from MoAlB in the vicinity of AlO x regions. Hence, based on these results a straightforward processing pathway for the direct formation of MoB MBene-AlO x heterostructures without employing chemical etching is proposed here.

Bulk MAB phases are usually synthesized at temperatures above 1000°C (ref. 10-12, 14, 15, 20 and 21) and used as precursor for MBenes. 15,20 The latter are then obtained by removing Al from the bulk phase via chemical etching. 15,20 Alameda et al. studied the local formation of MoB MBenes from a MoAlB single crystal through Al deintercalation using NaOH at RT and 700°C, respectively. 15,20 Upon Al deintercalation, initially areas with a high density of stacking faults are formed. 20 Further Al removal results in the formation of cavities where inter-grown Mo 2 AlB 2 compositional defects are present. These metastable Mo 2 AlB 2 defects are proposed to act as MBene precursors, since isolated MBenes occur within the cavity only directly adjacent to the former. 20 In another work, Alameda et al. synthesized Mo 2 AlB 2 -AlO x nanolaminates via etching and subsequent annealing procedures as future MBene precursors. 21 Kim et al. obtained Mo 2 AlB 2 as a potential MBene precursor by etching of MoAlB using LiF/HCl. 22 In the present work, we report the first direct synthesis of 2D MoB MBene via physical vapor deposition (PVD), thus omitting the etching process. Scanning transmission electron microscopy (STEM) proves the formation of MBene adjacent to AlO x regions, resulting in MoAlB-MoB-AlO x heterostructures. Density functional theory (DFT) calculations confirm the observed projected bond length values.

Thin film synthesis
MoAlB thin films were deposited by direct current magnetron sputtering (DCMS) with similar parameters as described in detail in our previous work. 23  was approximately 10 cm. A 2-inch Al 2 O 3 (0001) single crystal wafer, polished on one side, was used as a substrate. Prior to deposition, a pure Mo layer approximately 300 nm in thickness was deposited on the backside of the wafer to prevent heat radiation transparency from the heater positioned below the substrate. The growth temperature was set to 700°C and an Ar (99.999% purity) pressure of 0.4 Pa was used inside the sputtering chamber. The obtained thickness of the film was 3 μm as measured by cross-sectional TEM.

Transmission electron microscopy
High-resolution imaging work in STEM mode was performed on a ThermoFisher probe-corrected Titan Themis at an accelerating voltage of 300 kV. The instrument was equipped with a Bruker SuperX detector for energy-dispersive X-ray spectroscopy (EDX) analysis. TEM cross-sectional samples were prepared and thinned down to <35 nm by conventional mechanical polishing and Ar ion milling in a Gatan PIPS. For the imaging condition in STEM, a convergence angle of 23.8 mrad was chosen. Collection angle ranges of 78-200 and 10-16 mrad for high-angle annular dark field (HAADF) and annular bright field (ABF) STEM imaging, respectively, were used. Multivariate statistical analyses 36 on STEM-EDX spectrum imaging datasets to separate and quantify the different phases are available in the ESI. †

Density functional theory (DFT)
Density functional theory (DFT) calculations were performed using the Vienna ab initio Simulation Package (VASP, University of Vienna), 24-26 employing projector-augmented waves 27 for basis set representation. The established Perdew, Burke and Ernzerhof (PBE) parametrization of the generalized gradient approximation 28 was utilized to account for electron exchange and correlation. Brillouin zone integration was performed using the method of Monkhorst and Pack, 29 and a k-mesh of sufficient density was chosen to ensure energetic convergence. The initial structural model for MoAlB was taken from literature 30 and fully optimized with respect to lattice parameters and atomic positions. To obtain MoB, the Al atoms were removed from the same starting model and the resulting binary system also underwent full optimization.

Results and discussion
Two impurity phases were identified by STEM EDX in the MoAlB (space group Cmcm) thin film, similar to the previously investigated one where we had used two sputter targets. 23  This observation provides evidence that the integral film com-position is close-to-stoichiometric and within the measurement uncertainties identical to the film reported in. 23 The local chemical composition was analyzed by EDX in STEM mode, shown in Fig. 1 Fig. S4. † Based on all high-resolution investigations carried out on this sample ( Fig. 2 and Fig. S7 † as well as material not shown here) we estimate the areal fraction of MoB MBene domains compared to all other phases identified to be <1%.

(a)-(d). A comparison of EDX spectra and Mo and Al intensities in the MoAlB and MoB regions are available in the ESI
The formation of orthorhombic MoAlB is evident from the HAADF image shown in Fig. 1(a). This result is consistent with X-ray diffraction and TEM data from the MoAlB film deposited with MoB and Al targets as reported earlier. 23 A low magnification HAADF STEM image is available in the Fig. S5 (ESI †) to reveal an average statistical analysis of the formation of a MoAlB domain in terms of average size and distribution. A few representative grain size measurements are shown in Fig. S5(a). † The average range in length and width is observed around 600-1650 nm and 100-400 nm, respectively. It is also worth to mention here that most of the grains are columnar and elongated slightly inclined to the substrate normal. Some seed crystals show a size of 20-30 nm and are overgrown by columnar grains. Furthermore, defects coexisting with pristine MoAlB are identified. These are tilt grain boundaries and compositional defects where Al layers within the MoAlB units are missing. Due to size constraints, this topic will be addressed in detail in another forthcoming report; here, we shall focus on the formation of the MoB MBene. Clearly, the dark region visible in the upper right-hand corner of Fig. 1(a) is Modeficient (see the elemental distribution in Fig. 1(b)), Al-rich ( Fig. 1(c)) as well as O-rich ( Fig. 1(d)). The relative elemental intensity maps shown in Fig. 1(e-g) corroborate these findings for the Mo, Al and O distributions. Thus, the dark region can clearly be identified as an aluminum oxide phase, AlO x . Additional EDX maps as well as area fraction analysis are given in the ESI (Fig. S1 and S2 †). This impurity phase is likely formed due to incorporation of residual gases during thin film growth 31 and is consistent with the identification of Al-and O-rich regions in a film synthesized using two sputter targets. 23 In Fig. 2(a) and (b), identical HAADF STEM images are shown, where in the latter Mo, Al and B atomic columns are marked for easier identification of pristine MoAlB and Mo 2 B 2 subunits. In these images, the intensity is roughly proportional to Z 2 , 32 with Z being the atomic number, so that the brightest atomic columns are Mo columns and the dark grey ones are Al columns. The decrease in intensity with respect to particular atomic columns can help to identify missing atoms/ vacancies. Due to the low scattering amplitude B is invisible and accordingly, the B position is inferred from the ABF images given in Fig. 3. From Fig. 2(a) and (b), it is evident that the Mo 2 B 2 subunits are bent in proximity to the AlO x region identified via EDX (see Fig. 1(c) and (g)) and exhibit a variation in distance to the nearest-neighbor Mo 2 B 2 subunit. Both observations are consistent with the formation of 2D MBenes. 20 To measure the projected in-plane distance between the Mo 2 B 2 subunits quantitatively, we performed a Gauss fitting procedure 33 to determine the center of mass of each atomic column in the experimental STEM images (Fig. S6 †). While in pristine MoAlB the projected in-plane distance between the Mo 2 B 2 subunits is 0.26 ± 0.01 nm (double-headed black arrows in Fig. 2a), the distances in the bent region vary between 0.21 ± 0.01 nm (white) and 0.29 ± 0.01 nm (green), resulting in relative differences of approx. +11% to −19% to the distance measured in pristine MoAlB. This variation in interlayer spacing is related to the MoB layer bending (Fig. 2(a) and (b)). In Fig. 2(a), red and blue arrows are drawn parallel to the Al layers and the corresponding intensity line profiles are shown in Fig. 2(c) and (d). These indicate a significant decrease in Al concentration as the AlO x region is approached. All three observations, namely the bent Mo 2 B 2 subunits, the observed distance variations between them and the decreasing Al intensity obtained via line profile scans, are consistent with the formation of MoB MBenes. Hence, it is reasonable to assume that MBene is formed by deintercalation of Al from MoAlB and that AlO x accommodates the deintercalated Al. Another example where we identified MBene adjacent to an AlO x region, also comprising the characteristically bent Mo 2 B 2 subunits, is shown in the ESI (Fig. S7 †).
Alameda et al. previously indicated the formation of a twocomponent heterostructure consisting of Mo 2 AlB 2 -AlO x nanosheets via chemical etching and subsequent annealing, and the possibility of subsequent further Al deintercalation resulting in 2D MoB formation was suggested. 21 In contrast, it is important to realize that in the present study, local MBene formation occurs during magnetron sputtering and is thus obtained in a direct synthesis route, without chemical etching. We observe a three-component heterostructure consisting of MoAlB, MoB and amorphous AlO x , thereby confirming the prior prediction of Alameda et al. 21 via thin film growth at moderate growth temperature (700°C). We reiterate that the evidence points towards direct 2D MoB MBene formation without chemical etching 20 and propose to explore this by intentional incorporation of AlO x regions during sputtering resulting in the formation of multilayered MoAlB/AlOx heterostructures. However, the exact role of AlO x during MBene formation observed here and in literature 21 needs to be investigated further by future in situ and ex situ annealing experiments. Also, the role of stacking faults or intergrown intermediate precursor phases as described in literature 20,21 needs to be considered further. As alluded to before, we do find, e.g., compositional defects adjacent to the MBenes (see Fig. 2(a), where three instead of two Mo atomic columns are clearly present within one subunit), but cannot yet conclusively determine whether they are required for MBene formation or not.
In ABF STEM images the intensity is proportional to Z 1/3 , 34 thus, ABF micrographs can be used for detection of light atoms. The corresponding ABF STEM image, shown in Fig. 3(a), of the same area as examined before (Fig. 2(a)) identifies, besides the amorphous AlO x in the upper right-hand corner, deintercalated Al (red arrows) in the MoB MBene region and the Al double layer (blue arrows) in the MoAlB MAB phase. Enlarged ABF micrograph excerpts depict the atomic arrangements of each element in the MoB MBene ( Fig. 3(b)) and the MoAlB MAB (Fig. 3(c)) phases. The ABF line profiles (dotted red and blue lines in Fig. 3(d)) along the Mo-B-Mo unit (direction marked via red and blue arrows in Fig. 3 Fig. 3(d)). The quantitative values and the error bar for the Mo-Mo bond length were extracted over several areas using the peak fitting procedure mentioned above.
To validate the experimentally obtained bond lengths, Mo-Mo distances were obtained from DFT calculations. To allow for a direct comparison with the line scan measurements, only the projected component in the ab plane, and not the true 3D distance, is given. For Mo-Mo, this projected distance, corresponding to the diagonal measurement in Fig. 3(d), is 4.07 Å in MoAlB and 4.00 Å in MoB. It is therefore in excellent agreement with the experimental data which were obtained from  center of mass analysis of the STEM images. Both values lie within the error margin of the aforementioned HAADF measurements. Furthermore, the here observed formation of MoB MBene during magnetron sputtering can be rationalized based on an energy requirement of approx. 0.115 eV per atom for the formation of MoB with respect to pristine MoAlB, as obtained from the energies of formation in the ground state for both systems. The magnitude of this energy barrier is readily overcome during magnetron sputtering, as proven by the synthesis of metastable fcc-(Ti,Al)N, for which energetic barriers between 0.123 and 0.181 eV per atom had been predicted. 35

Conclusions
In summary, the concurrent formation of pristine MoAlB and MoB MBene in the vicinity of AlO x was observed in magnetronsputtered MoAlB thin films. The characteristically bent Mo 2 B 2 subunits, the distance variations between neighboring Mo 2 B 2 subunits as well as spatially resolved composition data clearly indicate both Al deintercalation and 2D MoB MBene formation in regions adjacent to AlO x . Hence, the reported synthesis pathway for MBene formation by magnetron sputtering is direct and does not require chemical etching. Overall, the data presented here underlines the untapped potential of magnetron sputtering for the direct formation of MBene in MoAlB/ AlO x multilayered heterostructures.

Conflicts of interest
The authors declare no competing financial interests.