Noncollinear magnetic order in two-dimensional NiBr2 films grown on Au(111)

Noncollinear Magnetic Order in TwoDimensional NiBr2Films Grown on Au(111) Djuro Bikaljevic,*Carmen González-Orellana, Marina Pena-Díaz, Dominik Steiner, Jan Dreiser, Pierluigi Gargiani, Michael Foerster, Miguel Angel Nino, Lucía Aballe, Sandra Ruiz-Gomez, Niklas Friedrich, Jeremy Hieulle, Li Jingcheng, Maxim Ilyn,*Celia Rogero, and JoséIgnacio Pascual* Cite This: ACS Nano 2021, 15, 14985−14995 Read Online ACCESS Metrics & More Article Recommendations * sıSupporting Information ABSTRACT: Metal halides are a class of layered materials with promising electronic and magnetic properties persisting down to the two-dimensional limit. While most recent studies focused on the trihalide components of this family, the rather unexplored metal dihalides are also van der Waals layered systems with distinctive magnetic properties. Here we show that the dihalide NiBr2grows epitaxially on a Au(111) substrate and exhibits semiconducting and magnetic behavior starting from a single layer. Through a combination of a low-temperature scanningtunneling microscopy, low-energy electron diffraction, X-ray photoelectron spectroscopy, and photoemission electron microscopy, we identify two competing layer structures of NiBr2coexisting at the interface and a stoichiometrically pure layer-bylayer growth beyond. Interestingly, X-ray absorption spectroscopy measurements revealed a magnetically ordered state below 27 K with in-plane magnetic anisotropy and zero-remanence in the single layer of NiBr2/Au(111), which we attribute to a noncollinear magnetic structure. The combination of such two-dimensional magnetic order with the semiconducting behavior down to the 2D limit offers the attractive perspective of using these films as ultrathin crystalline barriers in tunneling junctions and low-dimensional devices. KEYWORDS: 2D metal dihalide, molecular beam epitaxy, van der Waals material, semiconductor, 2D magnetism Since the discovery of graphene in 2004, research on ultrathin 2D films has become a subject of tremendous interest. 1,2 The number of 2D materials is growing rapidly, including metals, semimetals, semiconductors, insulators, and superconductors. 3−7 By decreasing the dimensionality of the bulk material down to the 2D limit, fascinating physical and chemical properties appear, which offer great possibilities for low-dimensional applications, 8−12 but bear as well huge potential for the exploration of exotic physical phenomena by means of constructing van der Waals heterostructures. 4,13 Furthermore, it is foreseen that 2D materials can substitute silicon in the complementary metal− oxide−semiconductor switches, in order to provide lower power consumption, thus delivering a higher performance. 14 The prospect of using 2D materials in electronics requires investigation of their properties both on the microscopic and mesoscopic level, reaching the length scale relevant for possible low-dimensional applications. Magnetic 2D materials are of special interest for fabricating superconductor−ferromagnetic van der Waals interfaces, 7,15 but also for ultrathin barriers in tunneling devices. 16,17 However, two-dimensional intrinsic ferromagnetic (FM) materials are still in their emergence. 2 In order to realize a ferromagnetic 2D system, one has to overcome the constraints of the Mermin−Wagner theorem, 18 where long-range magnetic order is suppressed due to thermal fluctuations. 19,20 This can be achieved by, for example, applying a magnetic field or mechanical stress onto the system but alsomore interestinglyby intrinsic magnetic properties of the material as a consequence of strong spin−orbit coupling (SOC). 18,19 There are plenty of theoretical studies predicting magnetic monolayer materials, 21−28 but only a few intrinsic 2D ferromagnets were experimentally found. 7,20,29−37 The most explored magnetic 2D materials are trivalent halides of the type MX3(M = metal, X = halogen) . 7,29−31,37 Bulk divalent halides (MX2) are known to be magnetic, but they still have not been Received: June 18, 2021 Published: September 7, 2021 Article www.acsnano.org © 2021 The Authors. Published by American Chemical Society 14985 https://doi.org/10.1021/acsnano.1c05221 ACS Nano 2021, 15, 14985−14995 Downloaded via CSIC on January 26, 2022 at 10:03:33 (UTC). See https://pubs.acs.org/sharingguidelines for options on how to legitimately share published articles. studied thoroughly in the limit of a single layer. 38,39 A promising candidate that belongs to this family and was predicted to possess intrinsic ferromagnetism down to the 2D limit is NiBr2. 21,22,40,41 Bulk NiBr2is a van der Waals semiconducting material with an interlayer spacing of 3.24 Å. 21 This large separation implies a weak interlayer coupling and hints at the possibility of exfoliation into single-layer sheets. A single, pristine layer of NiBr2consists of a Ni plane embedded between two Br planes with an in-plane lattice constant of 3.7 Å. 21,22 The nickel ion is in a +2 oxidation state after using two of their valence electrons for ionic bonding with bromine. The octahedral coordination of the nickel ion splits the electronic states of the d-shell into two groups of levels with t2g and egsymmetry. The first group of levels is full with six electrons, while the second is half-filled, endowing the Ni2+ ion with a nominal magnetic moment of 2 μB. 38 The ordered magnetic state observed in bulk NiBr2below 52 K comprises ferromagnetic slabs stacked antiferromagnetically along the crystallographic c-axis. 38 Calculations on bidimensional NiBr2suggest a ferromagnetic ground state with a net magnetic moment between 1.57 and 1.88 μB. 21,22 The predicted value of the Curie temperature (TC) for a monolayer of NiBr2is 136−140 K, which is rather high as compared to 2D trivalent halides (TC< 45 K). 21,22 In this work we show that NiBr2grows epitaxially on a Au(111) substrate and exhibits semiconducting character and magnetic ordering behavior already from the first monolayer. To determine the growth modes and their chemical stoichiometry, we combined diverse surface analytical methods such as low-temperature scanning-tunneling microscopy (STM) and spectroscopy (STS), low-energy electron diffraction (LEED) and microscopy (LEEM), X-ray photoemission (XPS), and photoemission electron microscopy (PEEM). We find that the layer-by-layer growth of NiBr2 proceeds in the form of weakly coupled layers, with the exception of partly decomposed NiBr2domainsatthe interface, probably caused by the catalytic effect of the gold substrate. X-ray magnetic circular dichroism (XMCD) measurements unveiled a magnetically ordered state below 27 K with in-plane magnetic anisotropy and zero remanence in the single NiBr2layer that we interpret as a noncollinear magnetic structure. RESULTS AND DISCUSSION Growth and Structure of NiBr2on Au(111). To investigate the growth mode of NiBr2, we compare in Figure 1low-temperature STM images of the substrate with different NiBr2coverages. High-resolution STM images of every different domain are shown in Figure 1d−g to identify their layer structure and interpret their origin. For a nominal coverage of ∼0.1 ML, highly ordered two-dimensional islands appear coexisting with atomic chains distributed along the Au(111) surface (Figure 1a). We attribute these to residual Br atoms forming a mesh on the gold surface. 42 According to thermochemical investigations, sublimation of NiBr2produces a gas of monomers and dimers of stoichiometric NiBr2, while a thermal decomposition of the precursor is negligible. 43 However, the catalytic activity of the Au(111) surface in the dehalogenation reaction leads to decreasing energy barriers for dissociation of halogen atoms. 44 Therefore, the presence of residual Br atoms is most probably a result of the dissociative adsorption of some part of the NiBr2molecules. 45 Further, the Au(111) reconstruction seems to be slightly modified, suggesting a chemical interaction of the NiBrxspecies with the metal substrate. The internal layer structure of the NiBrx island, depicted in Figure 1d, appears with a complex chiral structure, with atomic periodicity of ∼3.6 Å, and with an additional superstructure of ∼11.9 Å (these can be extracted from the fast Fourier transformation (FFT) analysis shown in the Supporting Information (SI), in Figure S1). This layer Figure 1. STM images of different coverages of NiBr2on Au(111) and the corresponding layer structure. (a) Submonolayer coverage with islands of NiBrx(labeled with a green rectangle) and distributed Br atoms on the surface. I= 60 pA, U= 1 V. (b) Coexistence of three different first layers (green, yellow, red rectangles) and second-layer islands (black rectangle). I= 100 pA, U= 1 V. (c) NiBr2multilayer. I= 30 pA, U= 1 V. (d) Atomic layer structure of NiBrx(green rectangle). I= 20 pA, U= 1 V, scale bar: 5 nm. (e−g) Atomic resolution of the first (red rectangle), second (black rectangle), and third (violet rectangle) pristine layer of NiBr2, showing the same lattice parameters: (e) I = 4.7 nA, U= 0.01 V, scale bar: 2 nm; (f) I= 1.8 nA, U= 0.05 V, scale bar: 2 nm; (g) I= 100 pA, U=−1.6 V, scale bar: 1 nm. ACS Nano www.acsnano.org Article https://doi.org/10.1021/acsnano.1c05221 ACS Nano 2021, 15, 14985−14995 14986 structure is further confirmed by its distinctive LEED pattern, displayed in Figure 2b. As indicated by the red and blue circles, the unit cell of the NiBrxoverlayer is a factor 4/3 larger than the Au(111) surface, resulting in a 4 ×4 superstructure with respect to the underlying Au(111) (note that the lattice constant of the NiBrxlayer, 3.6 Å, is close to a factor 4/3 of the Au(111) lattice parameter, 2.87 Å). Thus, we suggest that the NiBrxphase is composed of a Au-commensurate Ni-plane with chemically bound Br atoms on top. Stoichiometic NiBr2layers can also be observed directly on the Au(111) surface, especially when increasing the coverage. Deposition of a ∼1 ML of NiBr2leads to the appearance of additional hexagonal regions (red square in Figure 1b) with a lattice constant of 3.8 Å (Figure 1e). We attribute this layer to intact NiBr2domains, since their structure and lattice periodicity coincide with that of the nominal value for a free two-dimensional NiBr2layer (with lattice constant 3.7 Å, according to DFT simulations 21,22 ). It is noteworthy that the first NiBr2layer appears with some contrast formed by brighter and darker patches, probably caused by interaction with the underlying Au substrate. The STM images for this coverage also show some second-layer nucleation on top of the first NiBr2layer, with the same bulk-like NiBr2periodicity (black rectangle, see Figure 1b,f). The second layer does not show any prominent bright−dark superstructure. Figure 2c represents a LEED pattern for the ∼1.5 ML sample. It is similar to the pattern in Figure 2b. However, the spots indicated by the inner blue circle are more intense, suggesting a coexistence of NiBrxand NiBr2, with increasing proportion of the latter. STM images of thicker NiBr2films, with an estimated coverage above 2 ML, show the nucleation of a third NiBr2 layer on top of a complete second one (Figure 1c), also with the bulk lattice structure (Figure 1g), in agreement with a layer-by-layer growth mode. The LEED pattern of the multilayer film comprises only the diffuse hexagonal pattern and has no traces of the bare Au(111) pattern. The dimension of the hexagonal cell in Figure 2d (indicated by a blue circle) fits well with the unit cell of the bulk NiBr2(3.8 Å). The absence of additional spots characteristic of the 4 ×4 superstructure emphasizes the growth of exclusively stoichiometric NiBr2layers for higher coverages. The chemical composition of the epitaxial NiBr2sheets has been probed using X-ray photoemission spectroscopy. Figure 3a shows the Ni 2p core level measured for coverages ranging from submonolayer to ∼2.5 ML. The two spin−orbit components (Ni 2p3/2 and Ni 2p1/2), separated by Δ≈17.3 eV, exhibit a complex multiple-peak structure due to different core−hole final states and plasmon resonances. 46 The spectra measured for 2.5 ML, blue in Figure 3a, reproduce the eight peak shape spectra reported for bulk NiBr2, 47 indicating that Figure 2. LEED pattern for different coverages of NiBr2on Au(111). (a) Hexagonal pattern for the clean Au(111) surface. (b) LEED pattern representing the sub-monolayer regime with the majority of NiBrx. (c) LEED image showing ∼1.5 ML NiBr2. Spots characteristic of NiBr2 and indicated by the blue circle are more pronounced than in (b). (d) Hexagonal pattern showing the crystal structure of pristine NiBr2. All LEED images have been collected at 137 eV. Figure 3. (a) XPS spectra showing the characteristic Ni 2p core level peaks for different coverages of NiBr2on Au(111). Gray dashed lines demonstrate the position of the peaks. (b−d) Detailed analysis of the Ni 2p3/2 signal within the range of 850−865 eV, showing the proportional composition of the bulk (NiBr2) and the interface (NiBrx) phases. ACS Nano www.acsnano.org Article https://doi.org/10.1021/acsnano.1c05221 ACS Nano 2021, 15, 14985−14995 14987 the NiBr2stoichiometric phase is indeed preserved after sublimation. For lower coverage, both the shape and position of the peaks change, revealing the presence of NiBrxwith a different Ni chemical environment. Figure 3b−d compare in detail the Ni 2p3/2 spectra for the three measured coverages together with their decomposition in components. Following the model of ref 46 we used three peaks (main peak located at ∼855.2 eV plus two satellites at higher B.E.) to fit the Ni 2p3/2 spectra of NiBr2(red peaks in Figure 3b−d). We found out that a good description of the experimental spectra can be obtained if we keep their positions constant and assign another three peaks shifted ∼2.5 eV toward the lower binding energy to represent a contribution of the NiBrxphase (green peaks in Figure 3b−d). All spectra were aligned and normalized using the Au 4d peak (not shown). Figure 3b−d clearly demonstrate quick attenuation of the NiBrxsignal with nucleation of the second and third layers of NiBr2in the 1.5 and 2.5 ML samples, respectively. It corroborates that NiBrxexists only in the Au interface. Furthermore, Figure 3b shows substantial contribution of the NiBr2signal in the experimental spectrum of 0.5 ML sample, proving that the first layer comprises both phases. Thus, XPS data agree with the STM and LEED results. Layer-Dependent Electronic Band Gap. The electronic properties of ultrathin NiBr2films on Au(111) were measured by means of dI/dVspectra at various surface positions. Figure 4a shows an STM image with a nominal coverage of 1 ML NiBr2on the Au(111) surface. We study first the unoccupied states of the different halide phases and layers. Constant-current dI/dVspectra measured at the positions marked as colored circles in the STM image are displayed in Figure 4b. The spectrum on the NiBrxinterface phase (blue in Figure 4b) does not show any distinct features in the dI/dV signal. It is also noteworthy that the Au(111) Shockley surface state at ca.−0.5 V does not survive upon deposition of NiBrx (see Figure S2 in the SI). A rather metallic character can be observed that agrees with a partial debrominated phase and with the larger hybridization of Ni ions with the metal substrate. In contrast, spectra on NiBr2layers show sharp resonances for the unoccupied density of states, whose positions depend on the number of layers: the first NiBr2 layer (black spectrum in Figure 4b) shows a single, broad resonance at ∼0.8 V (peak 1); the spectrum on the second layer (red) shows an additional resonance peak at ∼1.3 V (peak 2), while the third layer (gray) shows two additional resonances, one attributed to peak 2 shifted to ∼1.4 V and an additional peak 3, at ∼1.7 V. Band structure calculations for a free NiBr2layer found that the material is a semiconductor with a band gap of 4 eV 21 and a characteristic unoccupied flat band above EF, attributed to the largely localized Ni d orbitals. The sharp dI/dVresonances obtained for the different layers in Figure 4b agree in energy alignment and shape with these frontier flat bands. Interestingly, we observe an increasing number of resonances with the number of layers. We interpret these as different resonant tunneling processes, each through the Ni-d flat band of the respective stacked NiBr2layers. Due to the weak van der Waals interaction between each layer, the Ni-d bands are highly localized at the respective Ni position and very weakly coupled to neighboring layers, so they can also be aligned with different energy. We expect that the increasing separation of every layer from the gold substrate leads to a decreasing screening effect by the metal substrate. This causes the flat bands of higher layers to appear shifted to higher energy in the spectrum, thus leading to a larger effective band gap. In addition, we note that for such layered system the applied potential difference between tip and sample during the STS measurement may cause electrostatic shifts in each isolated Nid band (i.e., local gating) depending on their position in the junction (a schematic model of the tunneling process is depicted in the inset of Figure 5). To obtain the values of the effective band gaps for every layer, we compare in Figure 5 constant-current dI/dVspectra on the first, second, and third slab of the pristine NiBr2, including spectra of the occupied states (an equivalent constant-height dI/dVspectrum, showing the fully depleted DOS within the energy gap, can be found in the SI in Figure S3). All layers show an increase in the dI/dVsignal at ca.−2.6 V that we ascribe to the onset of the valence band (VB) of the NiBr2film. In contrast to the strongly localized flat Ni-d states for the unoccupied states, the weak and rather broad resonance peak at ca.−2.6 V is attributed to the existence of overlapping s, p, and d bands with a major contribution from the Br-4p orbitals. 21,22 Attributing the different resonance peaks at positive sample biases to the respective conduction band (CB) onsets, the data reveal that the semiconducting behavior persists for each layer. The separation between the CB and VB peak(s) of the first layer of NiBr2amounts to EG= 3.4 (±0.2) eV and increases to EG= 3.9 (±0.2) eV and EG= 4.5 (±0.2) eV for the second and third layer, suggesting an increasing Figure 4. STS spectra measured on different surface layer structures, revealing corresponding resonances for positive sample bias (unoccupied states). (a) STM image with spectroscopy positions marked in different colors; I= 100 pA, U= 3 V. (b) Constant-current dI/ dVspectra measured on the different points indicated in (a) and corresponding to different layer structures (current set point I= 100 pA). ACS Nano www.acsnano.org Article https://doi.org/10.1021/acsnano.1c05221 ACS Nano 2021, 15, 14985−14995 14988 effective gap of the system. This is probably due to both a gradually smaller screening effect of the metal substrate and a small local gating of the unoccupied bands during the STS measurement. Nevertheless, these values lie close to the gap obtained from first-principle calculations of 4.0 eV. 22 Magnetic Properties. To resolve the possible survival of a magnetic state in two-dimensional NiBr2layers, magnetic measurements were performed by means of the spatially averaging XMCD technique. XAS spectra acquired for three samples with different dihalide coverage are shown in Figure 6 together with the resulting XMCD spectra. Peaks of the XAS absorption at the Ni L3edge have two components that give rise to two distinctive maxima in the XMCD spectra close to 850 eV. Comparing these data with XPS spectra from Figure 3, we can attribute the component with an absorption peak below 850 eV with a NiBrxphase that has a lower binding energy of the respective Ni 2p3/2 state and the second component, with an absorption peak above 850 eV, with the NiBr2phase. Evolution of the XAS spectra with coverage follows the trend observed with XPS: a 0.3 ML sample contains more NiBrxthan NiBr2, but in the 0.6 ML sample the stoichiometric NiBr2 already prevails over the NiBrxphase. Eventually, in the XAS spectra acquired for the 2 ML sample the contribution of the NiBrxphase is negligible in comparison to the strong signal corresponding to the NiBr2(see Figure 6c). As mentioned before, the eight d electrons of the octahedrally coordinated divalent Ni-ion in bulk NiBr2occupy the six available states in the t2g orbitals completely and half fill the egones, giving rise to the configuration with a full spin S= 1 and with zero orbital moment. 38 Nevertheless, for the samples with a sub-monolayer amount of NiBr2on Au(111) the orbital magnetic moment calculated via sum rules reaches 0.34−0.41 μBper Ni atom and its spin magnetic moment ascends only to 1.27−1.49 μB, which is substantially lower than 2μB, which would be the nominal value for the system with Figure 5. Determination of the band gap for different numbers of NiBr2layers: (a) first slab NiBr2, (b) second slab NiBr2, (c) third slab NiBr2. The spectrum on the third layer (gray) was measured on a different preparation, as in Figure 1c. Inset: Schematics of the tunneling process with respect to the unoccupied states through the layers of NiBr2. Figure 6. XAS and respective XMCD spectra taken with circular right (black) and circular left (red) polarization in a field of 6.5 T and at a temperature of 3 K (a, b) and of 6 T at 3.5 K (c). All data were collected in grazing incidence geometry (70°). It is worth mentioning that positions of the peaks in XAS and XPS spectroscopy do not coincide because Ni L3,2 absorption edges correspond to the transition from Ni 2p to the Ni 3d state, while XPS emission from the Ni 2p state measures transitions to the continuum. ACS Nano www.acsnano.org Article https://doi.org/10.1021/acsnano.1c05221 ACS Nano 2021, 15, 14985−14995 14989 spin S= 1 (see Table S1) and discussion therein about the assumptions that were made to perform the analysis of the multicomponent system). Since the XAS spectra measured over Ni L3,2 absorption edges contain the contributions from both NiBrxand NiBr2, analysis of the respective XMCD spectra via sum rules yields the values of the moments that are the weighted average of both components. Therefore, results for the samples with 0.3−1.0 ML coverage that have a notable amount of NiBrxshow that the magnetic properties of the decomposed phase are also rather different from stoichiometric NiBr2, in line with the striking difference demonstrated above for the electronic properties. In contrast, the 2 ML sample with a higher relative amount of NiBr2possesses a lower average orbital magnetic moment of 0.11 μBand spin magnetic moment of 1.47 μB, revealing that the magnetic properties of the first and second layers of the NiBr2on Au(111) are approaching the properties of bulk NiBr2. 38 XMCD magnetization loops were measured tracking the variation of the highest peak of the XMCD signal at the L3Ni edge as a function of the applied magnetic field. Figure 7a,b show two of these curves acquired for the 0.3 ML sample at 3 K in grazing (GI) and normal incidence (NI), respectively. Measurements of this kind are not capable of assessing the magnetic moment close to zero field, yielding the artifacts within ±0.1 T, and therefore do not capture the hysteresis with the coercive force falling within this range. Nevertheless, they reveal that in the NI geometry magnetization remains almost constant. Further analysis provided in the SI proves the presence of nonzero remanence that is not concerned with magnetic anisotropy but originates from the magnetic order. Such a low-temperature ferromagnetic phase of NiBrxis Figure 7. Comparison of magnetization loops measured in the grazing (70°) and normal incidence at the Ni L3edge for the sample having a 0.3 ML coverage (majority of NiBrxphase) (a, b) and 0.6 ML coverage (majority of NiBr2phase) (c, d) at 3 K. Black and red points correspond to descending and ascending branches, respectively. NI and GI stands for normal incidence and grazing incidence, respectively. Figure 8. (a) XMCD spectra measured over the Ni L3edge for the 0.6 ML sample at GI in the field of 6.5 T at different temperatures. The inset shows the in-plane saturation magnetization obtained via sum rules analysis and its behavior expected for the paramagnetic sample obeying the Brillouin function. (b) Arrott plot made for the 0.6 ML sample, having a majority of monolayer-thick NiBr2phase and (c) respective intercept values. ACS Nano www.acsnano.org Article https://doi.org/10.1021/acsnano.1c05221 ACS Nano 2021, 15, 14985−14995 14990 confirmed by inspecting the corresponding Arrott plots shown in the SI. Furthermore, since the low-field magnetization is almost equal to the saturation magnetization in NI geometry, such ferromagnetic phase has an out-of-plane easy magnetization direction (OOP anisotropy). Magnetization loops measured for the 0.6 ML sample (Figure 7c,d) show that the magnetization approaches zero in low magnetic fields for both orientations. Nonhysteretic magnetization reversal and a lack of remanent magnetization allow us to discard collinear ferromagnetic order down to 3 K in the monolayer-thick NiBr2, which is the majority phase in the 0.6 ML sample. Nevertheless, these curves do not show the Curie−Weiss behavior expected for a simple paramagnet. Indeed, a saturation field for NI is almost twice as high as for the GI direction. The loops have no characteristic S-shape, but, instead, the magnetization increases almost linearly until saturation (a small nonlinear contribution close to zero field can be attributed to the minor amount of a NiBrxphase present in this sample). Furthermore, sum rules analysis of the XMCD spectra collected at different temperatures (Figure 8a) allowed us to identify the evolution of the in-plane saturation magnetization. The experimental curve shown in the inset of Figure 8a displays only a slightly descending tendency in contrast with a quick decay predicted by the Brillouin function for the paramagnetic system. 48 This behavior is characteristic of some ordered magnetic state; therefore we have built an Arrott plot (Figure 8b) that helps to prove the presence of the magnetic order and find the temperature of the phase transition. This is a well-established technique that requires a set of magnetization loops measured at different temperatures, 48−50 and recently it was successfully used with XMCD magnetization curves as well. 51−53 Mean field theory predicts that in the paramagnetic state a linear (high-field) part of the M2vs H/M plot, extrapolated to the low-field region, intercepts the positive part of the H/Maxis, yielding the value of the inverse magnetic susceptibility. For the ferromagnetic state the crossing point at the H/Maxis will be negative, while the isotherm, corresponding to the phase transition, passes through the origin (see also the discussion in the SI). 48−50 Figure 8c shows positions of the interception H/MM=0 at different temperatures, obtained from the Arrott plot for the 0.6 ML sample. The negative intercept values found for all explored temperatures corroborate the existence of a magnetic order for the monolayer-thick NiBr2, which is the majority phase in this sample. The linear fitinFigure 8c results in a phase transition temperature of 27 K. Figure 9. (a) Bright-field LEEM image of ∼1.5 ML NiBr2grown on Au(111) at 400 K. (b) Corresponding LEED pattern, where the red circle highlights the (00) spot used for the bright-field images and the yellow circle indicates the spot characteristic of the 4 ×4 superstructure employed for the dark-field imaging. Dashed red circles show (10) and (01) spots that belong to the pattern of the NiBr2hexagonal crystal structure. (c) Dark-field LEEM image. (d) XPEEM image showing the contrast in the absorption of the X-rays with the energy of the beam tuned to the peak of the L3edge of Ni. All images were taken in the same area. Distinctive features repeated in all images are marked with the blue dotted rectangle. ACS Nano www.acsnano.org Article https://doi.org/10.1021/acsnano.1c05221 ACS Nano 2021, 15, 14985−14995 14991 The existence of a noncollinear state agrees with theoretical calculations, which predict that a free-standing monolayer of NiBr2is ordered ferromagnetically, but the competing interactions with the nearest and next nearest neighbors create a long-range helical magnetic texture. 41 Experimentally, a helical magnetic structure with a periodicity of ∼40 unit cells was observed in the bulk NiBr2at temperatures below 23 K. 38 This noncollinear structure has zero magnetic moment if it is averaged over the region comparable in size to its periodicity. That would explain the absence of remanence in the monolayer-thick NiBr2, as observed in the 0.6 ML sample (Figure 7c,d). Nevertheless, we cannot exclude other types of noncollinear order, as proposed for 2D materials with in-plane magnetic anisotropy. 51,54 Mesoscopic Scale. Since the combined STS/XMCD measurements reveal a magnetic semiconducting behavior of NiBr2down to the 2D limit, this material has an interesting potential to be used as a barrier in tunneling devices. It is thus important to study the growth modes and prove that film thickness is uniform within the length scale relevant for applications. For this aim, a sample of NiBr2with ∼1.5 ML coverage was grown in situ in the preparation chamber of the CIRCE beamline (ALBA) and studied by means of LEEM and PEEM. A LEED pattern closely resembling the pattern shown in Figure 2c (that was measured in our own chamber for a sample with the same nominal coverage) demonstrated the presence of both NiBrxand NiBr2phases. Nevertheless, the LEEM image (not shown) acquired with a field of view of 10 μm showed no clear contrast, revealing the relatively small size of the crystalline domains, comparable to the resolution of the instrument. This observation corroborates the STM results, yielding as well a maximum island extension of 50−100 nm (see Figure 1). A more uniform growth was observed when the substrate was warmed up during deposition. Figure 9a shows a bright-field image, measured using the central (00) spot of the LEED pattern (Figure 9b) of a film with the same nominal amount of material but grown at 400 K. Micrometer-scale domains are clearly visible and were identified by dark-field LEEM (Figure 9c). In this mode a spot of the reconstruction pattern characteristic of the NiBrx(Figure 1d) is used to build the image; therefore the domains of NiBrxappear brighter than the domains of NiBr2. Further, inspection with XPEEM shows a contrast in the Xray absorption mode with a photon energy tuned to the peak of the Ni L3edge (Figure 9d). Brighter zones correspond to the regions with stronger absorption (higher amount of Ni) and coincide with the domains of the NiBr2phase. There are no macroscopic regions with distinctively different levels of absorption except these two, implying that a slab of NiBr2 uniformly grows on a continuous NiBrxlayer. Since a third layer of NiBr2tends to cover the surface completely before the next layer starts to nucleate (see Figure 1c), we conclude that at least the first few layers of NiBr2grow in a layer-by-layer mode, following the formation of a 1 ML thick wetting layer that contains NiBrxand NiBr2in a proportion depending on the substrate temperature. CONCLUSIONS To summarize, we have demonstrated that the 2D NiBr2 compound grows epitaxially on Au(111) by means of sublimation of the stoichiometric precursor from a Knudsen cell. A combined LT-STM, LEED, and XPS study has unveiled that ultrathin 2D NiBr2sheets on Au(111) tend to grow in a layer-by-layer mode. The first layer comprises two different phases: a stoichiometric NiBr2layer and a NiBrxcompound with a commensurate crystalline structure and a distinctive reconstruction. The lower binding energy of Ni 2p states for the reconstructed phase seen with XPS together with a detailed analysis of the crystalline structure by LT-STM suggests that the NiBrxphase corresponds to a partially dehalogenated NiBr2. Additionally, constant-current STS revealed a metallic behavior of this debrominated layer. The different layers of NiBr2show layer-dependent characteristic dI/dVconstantcurrent STS spectra, with a distinctive band gap for each layer and revealing strongly localized Ni-d states as predicted theoretically. XMCD experiments have revealed noncollinear magnetic order with in-plane anisotropy in the NiBr2down to the 2D limit. The critical temperature of the magnetic phase transition was found to be 27 K. At the same time, a monolayer-thick NiBrxpossesses simple ferromagnetic order with out-of-plane magnetic anisotropy below 40 K. A mesoscopic scale characterization performed by means of LEEM and XPEEM confirmed a coexistence of small domains of both phases in the first monolayer if grown at room temperature. Warming the substrate to 400 K promotes a nucleation of NiBrx, leading to a continuous layer of this compound, while the next layers grow exclusively as stoichiometric NiBr2. Taking into account the easy preparation recipe with moderate technological requirements, high mechanical and electrical stability, and the large band gap down to the two-dimensional limit, we can foresee using NiBr2 for thin magnetic crystalline barriers in tunneling junctions and other electronic devices. EXPERIMENTAL AND METHODS The Au(111) single crystal was cleaned in ultrahigh vacuum (UHV) by several sputtering−annealing cycles with an annealing temperature around 460 °C and kept at this temperature for 10 min. NiBr2powder was sublimated from a Knudsen cell heated to 440 °C onto the clean Au(111) kept at room temperature, except one of the samples was prepared at 400 K for PEEM/LEEM characterization. The evaporation process of NiBr2was monitored by a microbalance. The low-temperature STM experiments were carried out with an LTSTM (Createc GmbH) operated at 6 K under UHV conditions with a base pressure of 5 ×10−11 mbar. XPS and LEED experiments were performed in a separate UHV chamber (base pressure of 1 ×10−10 mbar). The XPS measurements were performed with a Phoibos 100 photoelectron spectrometer, using a non-monochromatic Al KαX-ray source. XMCD measurements were taken at two different beamlines at two Synchrotron Light Facilities: BOREAS beamline at ALBA and XTreme beamline at SLS. 55,56 Absorption spectra at the Ni L2,3 edge for both normal (θ=0°, out of plane) and grazing (θ=70°, almost in plane) incidence geometries were studied at different temperatures in a magnetic fields up to 6.5 T. For BOREAS experiments two samples were grown in our home laboratory with coverage close to 1 and 2 ML of NiBr2and characterized by LEED and XPS prior to being transferred to the measurement chamber of the Boreas beamline, in a vacuum suitcase under a pressure below 1 ×10−9mbar. In this way we can directly relate the XMCD signal with the sample structure and composition. The two sub-monolayer samples measured at the XTreme beamline were grown in situ in the preparation chamber of the X-Treme beamline, which is equipped with an STM that allows establishing the direct relationship between the coverage and the integrated intensity of the polarization-averaged spectra (white line) listed in Table S1. Mesoscopic-scale characterization was performed by means of LEEM/PEEM microscopy at the CIRCE beamline (ALBA). 57 The ACS Nano www.acsnano.org Article https://doi.org/10.1021/acsnano.1c05221 ACS Nano 2021, 15, 14985−14995 14992 samples were grown in situ in the preparation chamber. LEEM was employed to obtain brightand dark-field images as well as microspot LEED patterns. In addition, XPEEM imaging at the Ni L3edge was done. ASSOCIATED CONTENT * sıSupporting Information The Supporting Information is available free of charge at https://pubs.acs.org/doi/10.1021/acsnano.1c05221. FFT analysis of the NiBrxphase; constant-height dI/dV spectra for bare Au(111) and NiBrxand for the first and second layer of NiBr2; calculation of magnetic moment via sum rules analysis; Arrott plots for NiBrx; roomtemperature STM and LEED data acquired at the Swiss Light Source (PDF) AUTHOR INFORMATION Corresponding Authors Djuro Bikaljevic −CIC nanoGUNE-BRTA, 20018 DonostiaSan Sebastián, Spain; Institute of Physical Chemistry, University of Innsbruck, A-6020 Innsbruck, Austria; orcid.org/0000-0003-3409-7042; Email: [email protected] Maxim Ilyn −Centro de Física de Materiales (CSIC/UPVEHU), 20018 Donostia-San Sebastián, Spain; Email: [email protected] José Ignacio Pascual −CIC nanoGUNE-BRTA, 20018 Donostia-San Sebastián, Spain; Ikerbasque, Basque Foundation for Science, 48013 Bilbao, Spain; orcid.org/ 0000-0002-7152-4747; Email: [email protected] Authors Carmen González-Orellana −Centro de Física de Materiales (CSIC/UPV-EHU), 20018 Donostia-San Sebastián, Spain Marina Pena-Díaz −Centro de Física de Materiales (CSIC/ UPV-EHU), 20018 Donostia-San Sebastián, Spain Dominik Steiner −Institute of Physical Chemistry, University of Innsbruck, A-6020 Innsbruck, Austria Jan Dreiser −Paul Scherrer Institut, CH-5232 Villigen, PSI, Switzerland; orcid.org/0000-0001-7480-1271 Pierluigi Gargiani −ALBA Synchrotron Light Source, 08290 Barcelona, Spain; orcid.org/0000-0002-6649-0538 Michael Foerster −ALBA Synchrotron Light Source, 08290 Barcelona, Spain Miguel A  ngel Nino−ALBA Synchrotron Light Source, 08290 Barcelona, Spain; orcid.org/0000-0003-3692-147X Lucía Aballe −ALBA Synchrotron Light Source, 08290 Barcelona, Spain Sandra Ruiz-Gomez −ALBA Synchrotron Light Source, 08290 Barcelona, Spain Niklas Friedrich −CIC nanoGUNE-BRTA, 20018 DonostiaSan Sebastián, Spain Jeremy Hieulle −CIC nanoGUNE-BRTA, 20018 DonostiaSan Sebastián, Spain; orcid.org/0000-0003-4891-4007 Li Jingcheng −CIC nanoGUNE-BRTA, 20018 Donostia-San Sebastián, Spain Celia Rogero −Centro de Física de Materiales (CSIC/UPVEHU), 20018 Donostia-San Sebastián, Spain; Donostia International Physics Center DIPC, 20018 Donostia-San Sebastián, Spain; orcid.org/0000-0002-2812-8853 Complete contact information is available at: https://pubs.acs.org/10.1021/acsnano.1c05221 Notes The authors declare no competing financial interest. ACKNOWLEDGMENTS D.B. acknowledges funding from the Austrian Science Fund (FWF) under the Erwin Schrödinger fellowship agreement (project number: J4395-N). C.G.-O. and M.P.-D. acknowledge funding of the Ph.D. fellowship from the MPC Foundation. We gratefully acknowledge financial support from Spanish AEI (Grant Nos. PID2019-107338RB-C6, RTI-2018-095303-C53, and the Maria de Maeztu Units of Excellence Programme MDM-2016-0618) and from the European Union (EU) through Horizon 2020 (SUPERTED Grant No. 800923), from Interred POCTEFA V-A Spain/France/Andorra Program (EFA 194/16/TNSI), the Basque Government (GV/EJ) under grants IT-1255-19, and the European Regional Development Fund. P.G. acknowledges funding from MINECO Grant No. FIS2016-78591-C3-2-R and FLAG-ERA Grant No. PCI2019-111908-2. REFERENCES (1) Novoselov, K. S.; Geim, A. K.; Morozov, S. V.; Jiang, D.; Zhang, Y.; Dubonos, S. V.; Grigorieva, I. V.; Firsov, A. A. Electric Field Effect in Atomically Thin Carbon Films. Science (Washington, DC, U. S.) 2004,306, 666−669. (2) Samarth, N. Condensed-Matter Physics: Magnetism in Flatland. Nature 2017,546, 216−218. (3) Novoselov, K. S.; Mishchenko, A.; Carvalho, A.; Castro Neto, A. H. 2D Materials and van der Waals Heterostructures. Science (Washington, DC, U. S.) 2016,353, aac9439. (4) Geim, A. K.; Grigorieva, I. V. van der Waals Heterostructures. Nature 2013,499, 419−425. (5) Koós, A. A.; Vancsó, P.; Magda, G. Z.; Osváth, Z.; Kertész, K.; Dobrik, G.; Hwang, C.; Tapasztó, L.; Biró, L. P. STM Study of the MoS2 Flakes Grown on Graphite: A Model System for Atomically Clean 2D Heterostructure Interfaces. Carbon 2016,105, 408−415. (6) Trainer, D. J.; Wang, B.; Bobba, F.; Samuelson, N.; Xi, X.; Zasadzinski, J.; Nieminen, J.; Bansil, A.; Iavarone, M. ProximityInduced Superconductivity in Monolayer MoS2. ACS Nano 2020,14, 2718−2728. (7) Kezilebieke, S.; Huda, M. N.; Vano, V.; Aapro, M.; Ganguli, S. C.; Silveira, O. J.; Głodzik, S.; Foster, A. S.; Ojanen, T.; Liljeroth, P. Topological Superconductivity in a van der Waals Heterostructure. Nature 2020,588, 424−428. (8) Radisavljevic, B.; Radenovic, A.; Brivio, J.; Giacometti, V.; Kis, A. Single-Layer MoS2 Transistors. Nat. Nanotechnol. 2011,6, 147−150. (9) Mak, K. F.; McGill, K. L.; Park, J.; McEuen, P. L. Valleytronics. The Valley Hall Effect in MoS2Transistors. Science (Washington, DC, U. S.) 2014,344, 1489−1492. (10) Xia, F.; Mueller, T.; Lin, Y.-M.; Valdes-Garcia, A.; Avouris, P. Ultrafast Graphene Photodetector. Nat. Nanotechnol. 2009,4, 839− 843. (11) Glavin, N. R.; Rao, R.; Varshney, V.; Bianco, E.; Apte, A.; Roy, A.; Ringe, E.; Ajayan, P. M. Emerging Applications of Elemental 2D Materials. Adv. Mater. (Weinheim, Ger.) 2020,32, No. e1904302. (12) Qiao, H.; Huang, Z.; Ren, X.; Liu, S.; Zhang, Y.; Qi, X.; Zhang, H. Self−Powered Photodetectors Based on 2D Materials. Adv. Opt. Mater. 2020,8, 1900765. (13) Liu, Y.; Weiss, N. O.; Duan, X.; Cheng, H.-C.; Huang, Y.; Duan, X. van der Waals Heterostructures and Devices. Nat. Rev. Mater. 2016,1,DOI: 10.1038/natrevmats.2016.42. (14) Fiori, G.; Bonaccorso, F.; Iannaccone, G.; Palacios, T.; Neumaier, D.; Seabaugh, A.; Banerjee, S. K.; Colombo, L. Electronics Based on Two-Dimensional Materials. Nat. Nanotechnol. 2014,9, 768−779. (15) Burch, K. S.; Mandrus, D.; Park, J.-G. Magnetism in TwoDimensional van der Waals Materials. Nature 2018,563,47−52. ACS Nano www.acsnano.org Article https://doi.org/10.1021/acsnano.1c05221 ACS Nano 2021, 15, 14985−14995 14993